User login
Spectral gradient acoustic reflectometry aids diagnosis of acute otitis media and otitis media with effusion
Spectral gradient acoustic reflectometer (SGAR) is a technology to assist in the detection of middle ear fluid occurring in the context of diagnosing acute otitis media (AOM) and otitis media with effusion (OME). The technology involves sending a harmless, inaudible sonar-like sound wave from the emitter that goes through the tympanic membrane, hits the posterior wall of the middle ear space, and bounces back to the sound detector in the device. If there is only air in the middle ear space, the sound wave bounces back quickly, and you get a high reading. If the sound wave bounces back more slowly, there is middle ear effusion. The thicker the effusion, the more likely it is pus and an AOM or a chronic OME (depending on the clinical situation), causing the sound wave to bounce back more slowly and giving a low reading.
The specificity of a high reading is remarkable at around 95%, so a high reading is a big reassurance that middle ear effusion is absent. A lower reading suggests effusion and the lower it is, the greater the sensitivity. When I get an unexpected higher or lower reading, I go back and reexamine the patient.
I asked our nurses to compare the handheld tympanometer to the SGAR. They actually perform the testing, and I interpret it. The nurses said:
• The SGAR is easier to use because of how quickly a readout is obtained.
• If a child is crying or moving, they can still get a readout.
• You don’t have to change the tip of the SGAR for the size of the external ear canal.
• The SGAR is easier to read than the tympanometer.
• The SGAR is easier to interpret for the parents.
• You don’t have to get a seal with the ear canal with SGAR, as you do with a tympanometer.
• The SGAR uses a disposable tip.
I asked our office manager to look up our use of the SGAR and tympanometer during our everyday practice. We found that SGAR or tympanometry was used in 12% of patient encounters in which the diagnosis of AOM or OME was part of the chief complaint. The ratio of use was 3:1, favoring SGAR. The most frequent use was in 30% of patient encounters tied to the diagnosis of "otalgia" (388.70) because with that diagnosis, we are stating to parents and patients that there is no middle ear pathology seen on exam, and it is confirmed by a test using sonar waves with the SGAR device. Our nurse practitioners and physician assistants particularly find the use of the SGAR beneficial in helping to reassure the parents and patients that they have not missed an AOM or OME.
The billing code is the same for SGAR and tympanometry (92567), so the fee payment is the same for both tests. Our second most common use is in association with possible AOM (382.9) at 12% of visits. Third is OME (381.02) used in a follow-up visit to determine the presence and thickness of persisting effusion.
About one-quarter of children seen in our practice with a chief complaint of "earache" receive the diagnosis of otalgia, often confirmed by SGAR, and do not receive an antibiotic. Thus, they are offsetting the charge for the procedure by saving on the costs of antibiotics and the accumulation of excessive diagnoses of AOM and OME leading to ear tube surgeries and tonsillectomy/adenoidectomy. The diagnosis of AOM and OME requires a middle ear effusion to be accurate, and only SGAR measures detection of middle ear effusion. SGAR is a must own device for clinicians who exam ears. SGAR can help in conjunction with otoscopy for a difficult diagnosis of AOM. If I am having troubleremoving wax, or if the external ear canal is particularly curved, or if I’m on the fence or the parent seems to need further evidence of my diagnosis, I turn to the SGAR. If I can get a reading, then it can really help, and my nurses are successful in getting a reading about 90% of the time. The main issue is ear canal wax, because occlusion by wax of more than 50% of the external ear canal opening causes invalid readings.
We should prescribe antibiotics for AOM in my opinion, but not for otalgia and not if the diagnosis is uncertain. The SGAR device when properly used can help to reduce unnecessary use of antibiotics and their complications. In prior "ID Consult" columns, I have discussed improving the diagnostic accuracy of AOM and OME. Performing a good otoscopic exam with the best tools available and combining that exam with SGAR or tympanometry, in selected cases, is the best practice in my opinion, and what I do in my own practice.
Dr. Pichichero, a specialist in pediatric infectious diseases, is director of the Research Institute, Rochester General Hospital, N.Y. He is also a pediatrician at Legacy Pediatrics in Rochester. E-mail him at [email protected]. Innovia Medical, the company that is bringing the SGAR EarCheck Pro back to market in 2014 after improvement and the addition of a USB port to allow the import of the data readout into the electronic medical record, asked Dr. Pichichero to assess the SGAR device.
Spectral gradient acoustic reflectometer (SGAR) is a technology to assist in the detection of middle ear fluid occurring in the context of diagnosing acute otitis media (AOM) and otitis media with effusion (OME). The technology involves sending a harmless, inaudible sonar-like sound wave from the emitter that goes through the tympanic membrane, hits the posterior wall of the middle ear space, and bounces back to the sound detector in the device. If there is only air in the middle ear space, the sound wave bounces back quickly, and you get a high reading. If the sound wave bounces back more slowly, there is middle ear effusion. The thicker the effusion, the more likely it is pus and an AOM or a chronic OME (depending on the clinical situation), causing the sound wave to bounce back more slowly and giving a low reading.
The specificity of a high reading is remarkable at around 95%, so a high reading is a big reassurance that middle ear effusion is absent. A lower reading suggests effusion and the lower it is, the greater the sensitivity. When I get an unexpected higher or lower reading, I go back and reexamine the patient.
I asked our nurses to compare the handheld tympanometer to the SGAR. They actually perform the testing, and I interpret it. The nurses said:
• The SGAR is easier to use because of how quickly a readout is obtained.
• If a child is crying or moving, they can still get a readout.
• You don’t have to change the tip of the SGAR for the size of the external ear canal.
• The SGAR is easier to read than the tympanometer.
• The SGAR is easier to interpret for the parents.
• You don’t have to get a seal with the ear canal with SGAR, as you do with a tympanometer.
• The SGAR uses a disposable tip.
I asked our office manager to look up our use of the SGAR and tympanometer during our everyday practice. We found that SGAR or tympanometry was used in 12% of patient encounters in which the diagnosis of AOM or OME was part of the chief complaint. The ratio of use was 3:1, favoring SGAR. The most frequent use was in 30% of patient encounters tied to the diagnosis of "otalgia" (388.70) because with that diagnosis, we are stating to parents and patients that there is no middle ear pathology seen on exam, and it is confirmed by a test using sonar waves with the SGAR device. Our nurse practitioners and physician assistants particularly find the use of the SGAR beneficial in helping to reassure the parents and patients that they have not missed an AOM or OME.
The billing code is the same for SGAR and tympanometry (92567), so the fee payment is the same for both tests. Our second most common use is in association with possible AOM (382.9) at 12% of visits. Third is OME (381.02) used in a follow-up visit to determine the presence and thickness of persisting effusion.
About one-quarter of children seen in our practice with a chief complaint of "earache" receive the diagnosis of otalgia, often confirmed by SGAR, and do not receive an antibiotic. Thus, they are offsetting the charge for the procedure by saving on the costs of antibiotics and the accumulation of excessive diagnoses of AOM and OME leading to ear tube surgeries and tonsillectomy/adenoidectomy. The diagnosis of AOM and OME requires a middle ear effusion to be accurate, and only SGAR measures detection of middle ear effusion. SGAR is a must own device for clinicians who exam ears. SGAR can help in conjunction with otoscopy for a difficult diagnosis of AOM. If I am having troubleremoving wax, or if the external ear canal is particularly curved, or if I’m on the fence or the parent seems to need further evidence of my diagnosis, I turn to the SGAR. If I can get a reading, then it can really help, and my nurses are successful in getting a reading about 90% of the time. The main issue is ear canal wax, because occlusion by wax of more than 50% of the external ear canal opening causes invalid readings.
We should prescribe antibiotics for AOM in my opinion, but not for otalgia and not if the diagnosis is uncertain. The SGAR device when properly used can help to reduce unnecessary use of antibiotics and their complications. In prior "ID Consult" columns, I have discussed improving the diagnostic accuracy of AOM and OME. Performing a good otoscopic exam with the best tools available and combining that exam with SGAR or tympanometry, in selected cases, is the best practice in my opinion, and what I do in my own practice.
Dr. Pichichero, a specialist in pediatric infectious diseases, is director of the Research Institute, Rochester General Hospital, N.Y. He is also a pediatrician at Legacy Pediatrics in Rochester. E-mail him at [email protected]. Innovia Medical, the company that is bringing the SGAR EarCheck Pro back to market in 2014 after improvement and the addition of a USB port to allow the import of the data readout into the electronic medical record, asked Dr. Pichichero to assess the SGAR device.
Spectral gradient acoustic reflectometer (SGAR) is a technology to assist in the detection of middle ear fluid occurring in the context of diagnosing acute otitis media (AOM) and otitis media with effusion (OME). The technology involves sending a harmless, inaudible sonar-like sound wave from the emitter that goes through the tympanic membrane, hits the posterior wall of the middle ear space, and bounces back to the sound detector in the device. If there is only air in the middle ear space, the sound wave bounces back quickly, and you get a high reading. If the sound wave bounces back more slowly, there is middle ear effusion. The thicker the effusion, the more likely it is pus and an AOM or a chronic OME (depending on the clinical situation), causing the sound wave to bounce back more slowly and giving a low reading.
The specificity of a high reading is remarkable at around 95%, so a high reading is a big reassurance that middle ear effusion is absent. A lower reading suggests effusion and the lower it is, the greater the sensitivity. When I get an unexpected higher or lower reading, I go back and reexamine the patient.
I asked our nurses to compare the handheld tympanometer to the SGAR. They actually perform the testing, and I interpret it. The nurses said:
• The SGAR is easier to use because of how quickly a readout is obtained.
• If a child is crying or moving, they can still get a readout.
• You don’t have to change the tip of the SGAR for the size of the external ear canal.
• The SGAR is easier to read than the tympanometer.
• The SGAR is easier to interpret for the parents.
• You don’t have to get a seal with the ear canal with SGAR, as you do with a tympanometer.
• The SGAR uses a disposable tip.
I asked our office manager to look up our use of the SGAR and tympanometer during our everyday practice. We found that SGAR or tympanometry was used in 12% of patient encounters in which the diagnosis of AOM or OME was part of the chief complaint. The ratio of use was 3:1, favoring SGAR. The most frequent use was in 30% of patient encounters tied to the diagnosis of "otalgia" (388.70) because with that diagnosis, we are stating to parents and patients that there is no middle ear pathology seen on exam, and it is confirmed by a test using sonar waves with the SGAR device. Our nurse practitioners and physician assistants particularly find the use of the SGAR beneficial in helping to reassure the parents and patients that they have not missed an AOM or OME.
The billing code is the same for SGAR and tympanometry (92567), so the fee payment is the same for both tests. Our second most common use is in association with possible AOM (382.9) at 12% of visits. Third is OME (381.02) used in a follow-up visit to determine the presence and thickness of persisting effusion.
About one-quarter of children seen in our practice with a chief complaint of "earache" receive the diagnosis of otalgia, often confirmed by SGAR, and do not receive an antibiotic. Thus, they are offsetting the charge for the procedure by saving on the costs of antibiotics and the accumulation of excessive diagnoses of AOM and OME leading to ear tube surgeries and tonsillectomy/adenoidectomy. The diagnosis of AOM and OME requires a middle ear effusion to be accurate, and only SGAR measures detection of middle ear effusion. SGAR is a must own device for clinicians who exam ears. SGAR can help in conjunction with otoscopy for a difficult diagnosis of AOM. If I am having troubleremoving wax, or if the external ear canal is particularly curved, or if I’m on the fence or the parent seems to need further evidence of my diagnosis, I turn to the SGAR. If I can get a reading, then it can really help, and my nurses are successful in getting a reading about 90% of the time. The main issue is ear canal wax, because occlusion by wax of more than 50% of the external ear canal opening causes invalid readings.
We should prescribe antibiotics for AOM in my opinion, but not for otalgia and not if the diagnosis is uncertain. The SGAR device when properly used can help to reduce unnecessary use of antibiotics and their complications. In prior "ID Consult" columns, I have discussed improving the diagnostic accuracy of AOM and OME. Performing a good otoscopic exam with the best tools available and combining that exam with SGAR or tympanometry, in selected cases, is the best practice in my opinion, and what I do in my own practice.
Dr. Pichichero, a specialist in pediatric infectious diseases, is director of the Research Institute, Rochester General Hospital, N.Y. He is also a pediatrician at Legacy Pediatrics in Rochester. E-mail him at [email protected]. Innovia Medical, the company that is bringing the SGAR EarCheck Pro back to market in 2014 after improvement and the addition of a USB port to allow the import of the data readout into the electronic medical record, asked Dr. Pichichero to assess the SGAR device.
Low serum uric acid levels protect against progressions of renal disease
SAN DIEGO – Patients who achieve a serum uric acid level of less than 6 mg/dL based on current gout management guidelines demonstrated a 37% reduction in progression of renal disease, a large retrospective study showed.
"There are numerous studies showing that people with renal disease can develop hyperuricemia," Dr. Gerald D. Levy said during a press briefing at the annual meeting of the American College of Rheumatology. "Some of them will also develop gout. There are a few small studies showing that in humans, you can reverse hyperuricemia with urate lowering therapy and make an impact in renal disease. We wanted to see if this is true."
Dr. Levy of the division of rheumatology in the department of internal medicine at Kaiser Permanente Medical Group, Downey, Calif., was the lead investigators in a study of 111,992 Kaiser Permanente Southern California patients with a serum uric acid (SUA) level of 7 mg/dL or greater from Jan. 1, 2002, to Dec. 31, 2010. Patients with at least 12 months of health plan membership, including drug benefit prior to the index date, were studied. All patients had at least one SUA and glomerular filtration rate (GFR) level measurement in the 6-month period prior to the index date and at least one SUA and one GFR in the follow-up period following the index date. Primary outcome events were at least a 30% worsening of renal function, initiation of dialysis, having a GFR of less than 15 mL/min, and undergoing a kidney transplant.
Patients with a new diagnosis of cancer were excluded from the analysis, as were those with HIV, glomerulonephritis, and/or organ transplant other than a kidney transplant.
Dr. Levy reported results from 16,186 patients who were divided into three groups: those who were never treated with urate-lowering therapy (ULT; 11,192); those who were treated with ULT less than 80% of the time from the index date to the end of follow-up period (3,902); and those who were treated with ULT 80% of the time or more from the index date to the end of the follow-up period (1,092). Of the three treatment groups, those who were treated with ULT 80% of the time or more during the study tended to be older and have more comorbid conditions, compared with the other two groups. They also began their ULT therapy earlier.
Among all patients combined, factors significantly associated with renal disease progression included having diabetes (hazard ratio, 1.96), hypertension (HR, 1.50), heart failure (HR, 1.39), previous hospitalizations (HR, 1.33), and being female (HR, 1.49) and older (HR, 1.03). The researchers found that time on ULT was not significantly associated with a reduction in renal disease progression outcome events (HR, 1.27, among those on ULT less than 80% of the time during the study vs. HR, 1.08, among those on ULT 80% of the time or more during the study). However, patients who achieved an SUA level below 6 mg/dL – a treatment goal in the 2012 ACR guidelines for management of gout – demonstrated a 37% reduction in renal disease progression (HR, 0.63; P less than .0001).
Dr. Levy acknowledged certain limitations of the study, including its retrospective design and the fact that patients with stage 4 and 5 chronic kidney disease were not included. "This is an important area, because if we can delay the worsening of renal disease in these folks, perhaps we’re abetting dialysis, which is growing by leaps and bounds in this country," he said. "Each dialysis patient now costs about $80,000 per year to take care of. If we could push that back even for a few years it would have a tremendous impact."
Dr. Levy had no relevant financial conflicts to disclose.
SAN DIEGO – Patients who achieve a serum uric acid level of less than 6 mg/dL based on current gout management guidelines demonstrated a 37% reduction in progression of renal disease, a large retrospective study showed.
"There are numerous studies showing that people with renal disease can develop hyperuricemia," Dr. Gerald D. Levy said during a press briefing at the annual meeting of the American College of Rheumatology. "Some of them will also develop gout. There are a few small studies showing that in humans, you can reverse hyperuricemia with urate lowering therapy and make an impact in renal disease. We wanted to see if this is true."
Dr. Levy of the division of rheumatology in the department of internal medicine at Kaiser Permanente Medical Group, Downey, Calif., was the lead investigators in a study of 111,992 Kaiser Permanente Southern California patients with a serum uric acid (SUA) level of 7 mg/dL or greater from Jan. 1, 2002, to Dec. 31, 2010. Patients with at least 12 months of health plan membership, including drug benefit prior to the index date, were studied. All patients had at least one SUA and glomerular filtration rate (GFR) level measurement in the 6-month period prior to the index date and at least one SUA and one GFR in the follow-up period following the index date. Primary outcome events were at least a 30% worsening of renal function, initiation of dialysis, having a GFR of less than 15 mL/min, and undergoing a kidney transplant.
Patients with a new diagnosis of cancer were excluded from the analysis, as were those with HIV, glomerulonephritis, and/or organ transplant other than a kidney transplant.
Dr. Levy reported results from 16,186 patients who were divided into three groups: those who were never treated with urate-lowering therapy (ULT; 11,192); those who were treated with ULT less than 80% of the time from the index date to the end of follow-up period (3,902); and those who were treated with ULT 80% of the time or more from the index date to the end of the follow-up period (1,092). Of the three treatment groups, those who were treated with ULT 80% of the time or more during the study tended to be older and have more comorbid conditions, compared with the other two groups. They also began their ULT therapy earlier.
Among all patients combined, factors significantly associated with renal disease progression included having diabetes (hazard ratio, 1.96), hypertension (HR, 1.50), heart failure (HR, 1.39), previous hospitalizations (HR, 1.33), and being female (HR, 1.49) and older (HR, 1.03). The researchers found that time on ULT was not significantly associated with a reduction in renal disease progression outcome events (HR, 1.27, among those on ULT less than 80% of the time during the study vs. HR, 1.08, among those on ULT 80% of the time or more during the study). However, patients who achieved an SUA level below 6 mg/dL – a treatment goal in the 2012 ACR guidelines for management of gout – demonstrated a 37% reduction in renal disease progression (HR, 0.63; P less than .0001).
Dr. Levy acknowledged certain limitations of the study, including its retrospective design and the fact that patients with stage 4 and 5 chronic kidney disease were not included. "This is an important area, because if we can delay the worsening of renal disease in these folks, perhaps we’re abetting dialysis, which is growing by leaps and bounds in this country," he said. "Each dialysis patient now costs about $80,000 per year to take care of. If we could push that back even for a few years it would have a tremendous impact."
Dr. Levy had no relevant financial conflicts to disclose.
SAN DIEGO – Patients who achieve a serum uric acid level of less than 6 mg/dL based on current gout management guidelines demonstrated a 37% reduction in progression of renal disease, a large retrospective study showed.
"There are numerous studies showing that people with renal disease can develop hyperuricemia," Dr. Gerald D. Levy said during a press briefing at the annual meeting of the American College of Rheumatology. "Some of them will also develop gout. There are a few small studies showing that in humans, you can reverse hyperuricemia with urate lowering therapy and make an impact in renal disease. We wanted to see if this is true."
Dr. Levy of the division of rheumatology in the department of internal medicine at Kaiser Permanente Medical Group, Downey, Calif., was the lead investigators in a study of 111,992 Kaiser Permanente Southern California patients with a serum uric acid (SUA) level of 7 mg/dL or greater from Jan. 1, 2002, to Dec. 31, 2010. Patients with at least 12 months of health plan membership, including drug benefit prior to the index date, were studied. All patients had at least one SUA and glomerular filtration rate (GFR) level measurement in the 6-month period prior to the index date and at least one SUA and one GFR in the follow-up period following the index date. Primary outcome events were at least a 30% worsening of renal function, initiation of dialysis, having a GFR of less than 15 mL/min, and undergoing a kidney transplant.
Patients with a new diagnosis of cancer were excluded from the analysis, as were those with HIV, glomerulonephritis, and/or organ transplant other than a kidney transplant.
Dr. Levy reported results from 16,186 patients who were divided into three groups: those who were never treated with urate-lowering therapy (ULT; 11,192); those who were treated with ULT less than 80% of the time from the index date to the end of follow-up period (3,902); and those who were treated with ULT 80% of the time or more from the index date to the end of the follow-up period (1,092). Of the three treatment groups, those who were treated with ULT 80% of the time or more during the study tended to be older and have more comorbid conditions, compared with the other two groups. They also began their ULT therapy earlier.
Among all patients combined, factors significantly associated with renal disease progression included having diabetes (hazard ratio, 1.96), hypertension (HR, 1.50), heart failure (HR, 1.39), previous hospitalizations (HR, 1.33), and being female (HR, 1.49) and older (HR, 1.03). The researchers found that time on ULT was not significantly associated with a reduction in renal disease progression outcome events (HR, 1.27, among those on ULT less than 80% of the time during the study vs. HR, 1.08, among those on ULT 80% of the time or more during the study). However, patients who achieved an SUA level below 6 mg/dL – a treatment goal in the 2012 ACR guidelines for management of gout – demonstrated a 37% reduction in renal disease progression (HR, 0.63; P less than .0001).
Dr. Levy acknowledged certain limitations of the study, including its retrospective design and the fact that patients with stage 4 and 5 chronic kidney disease were not included. "This is an important area, because if we can delay the worsening of renal disease in these folks, perhaps we’re abetting dialysis, which is growing by leaps and bounds in this country," he said. "Each dialysis patient now costs about $80,000 per year to take care of. If we could push that back even for a few years it would have a tremendous impact."
Dr. Levy had no relevant financial conflicts to disclose.
AT THE ACR ANNUAL MEETING
Major finding: Patients who achieved a serum uric acid level below 6 mg/dL – a treatment goal in the 2012 ACR guidelines for management of gout – demonstrated a 37% reduction in renal disease progression (HR, 0.63; P less than .0001).
Data source: A study of 16,186 patients who were divided into three groups: those who were never treated with urate-lowering therapy (ULT; 11,192), those who were treated with ULT less than 80% of the time from the index date to the end of follow-up period (3,902), and those who were treated with ULT 80% of the time or more from the index date to the end of the follow-up period (1,092).
Disclosures: Dr. Levy said that he had no relevant financial conflicts to disclose.
Factor can prevent bleeding in hemophilia A
Results of a phase 3 trial suggest a recombinant factor VIII Fc fusion protein (rFVIIIFc/efmoroctocog alfa, Eloctate/Elocta) can be used to prevent or reduce bleeding episodes in patients with severe hemophilia A.
Researchers found that prophylaxis with rFVIIIFc resulted in low annualized bleeding rates, and patients did not develop neutralizing antibodies.
Furthermore, the product was generally well-tolerated and had a prolonged half-life when compared with recombinant factor VIII.
Data from this study, called A-LONG, have been published in Blood.
Researchers tested rFVIIIFc in 165 male patients with severe hemophilia A who were 12 years of age and older.
Patients were divided into 3 treatment arms. Arm 1 received individualized prophylaxis, or 25 to 65 IU/kg every 3 to 5 days (n=118). Patients in arm 2 received a weekly prophylactic dose of 65 IU/kg (n=24). And patients in arm 3 received episodic treatment at doses of 10 to 50 IU/kg (n=23).
A total of 153 patients completed the study, and 757 bleeding episodes were treated with rFVIIIFc. Across the treatment arms, 87.3% of bleeding episodes were resolved with 1 injection of rFVIIIFc.
The annualized bleeding rate was significantly reduced with prophylaxis—by 92% for patients in arm 1 and 76% for those in arm 2—when compared with episodic treatment.
This was based on annualized bleeding rate estimates from a negative binomial regression model—2.91 for arm 1, 8.92 for arm 2, and 37.25 for arm 3.
The median annualized bleeding rates were 1.6 in arm 1, 3.6 in arm 2, and 33.6 in arm 3.
In arm 3, there were 9 patients who received rFVIIIFc to control bleeding during major surgery. In these cases, physicians rated the hemostatic response as “excellent” (n=8) or “good” (n=1).
There were no serious adverse events related to rFVIIIFc, and none of the patients developed neutralizing antibodies.
The most common adverse events (with an incidence of 5% or more) that occurred outside the perioperative period included nasopharyngitis, arthralgia, headache, and upper respiratory infection.
The researchers also compared the pharmacokinetics of rFVIIIFc and recombinant factor VIII. And they found the terminal half-life of rFVIIIFc was extended 1.5-fold compared to recombinant factor VIII—19.0 hours and 12.4 hours, respectively.
rFVIIIFc was developed using Fc fusion technology, which takes advantage of a naturally occurring pathway that delays the breakdown of IgG1 protein in the body by recycling it back into the bloodstream. This technology prolongs the time rFVIIIFc circulates in the body.
“There is an unmet medical need in the hemophilia community for longer intervals between prophylactic infusions while maintaining good control of bleeding episodes,” said study author Johnny Mahlangu, MD, director of the Haemophilia Comprehensive Care Centre at the University of the Witwatersrand and National Health Laboratory Service in Johannesburg, South Africa.
“A-LONG is the first clinical study to show that effective control over breakthrough bleeding may be achieved with once- or twice-weekly prophylactic infusions in people with severe hemophilia A.”
This study was funded by Biogen Idec, makers of rFVIIIFc. 
Results of a phase 3 trial suggest a recombinant factor VIII Fc fusion protein (rFVIIIFc/efmoroctocog alfa, Eloctate/Elocta) can be used to prevent or reduce bleeding episodes in patients with severe hemophilia A.
Researchers found that prophylaxis with rFVIIIFc resulted in low annualized bleeding rates, and patients did not develop neutralizing antibodies.
Furthermore, the product was generally well-tolerated and had a prolonged half-life when compared with recombinant factor VIII.
Data from this study, called A-LONG, have been published in Blood.
Researchers tested rFVIIIFc in 165 male patients with severe hemophilia A who were 12 years of age and older.
Patients were divided into 3 treatment arms. Arm 1 received individualized prophylaxis, or 25 to 65 IU/kg every 3 to 5 days (n=118). Patients in arm 2 received a weekly prophylactic dose of 65 IU/kg (n=24). And patients in arm 3 received episodic treatment at doses of 10 to 50 IU/kg (n=23).
A total of 153 patients completed the study, and 757 bleeding episodes were treated with rFVIIIFc. Across the treatment arms, 87.3% of bleeding episodes were resolved with 1 injection of rFVIIIFc.
The annualized bleeding rate was significantly reduced with prophylaxis—by 92% for patients in arm 1 and 76% for those in arm 2—when compared with episodic treatment.
This was based on annualized bleeding rate estimates from a negative binomial regression model—2.91 for arm 1, 8.92 for arm 2, and 37.25 for arm 3.
The median annualized bleeding rates were 1.6 in arm 1, 3.6 in arm 2, and 33.6 in arm 3.
In arm 3, there were 9 patients who received rFVIIIFc to control bleeding during major surgery. In these cases, physicians rated the hemostatic response as “excellent” (n=8) or “good” (n=1).
There were no serious adverse events related to rFVIIIFc, and none of the patients developed neutralizing antibodies.
The most common adverse events (with an incidence of 5% or more) that occurred outside the perioperative period included nasopharyngitis, arthralgia, headache, and upper respiratory infection.
The researchers also compared the pharmacokinetics of rFVIIIFc and recombinant factor VIII. And they found the terminal half-life of rFVIIIFc was extended 1.5-fold compared to recombinant factor VIII—19.0 hours and 12.4 hours, respectively.
rFVIIIFc was developed using Fc fusion technology, which takes advantage of a naturally occurring pathway that delays the breakdown of IgG1 protein in the body by recycling it back into the bloodstream. This technology prolongs the time rFVIIIFc circulates in the body.
“There is an unmet medical need in the hemophilia community for longer intervals between prophylactic infusions while maintaining good control of bleeding episodes,” said study author Johnny Mahlangu, MD, director of the Haemophilia Comprehensive Care Centre at the University of the Witwatersrand and National Health Laboratory Service in Johannesburg, South Africa.
“A-LONG is the first clinical study to show that effective control over breakthrough bleeding may be achieved with once- or twice-weekly prophylactic infusions in people with severe hemophilia A.”
This study was funded by Biogen Idec, makers of rFVIIIFc.
Results of a phase 3 trial suggest a recombinant factor VIII Fc fusion protein (rFVIIIFc/efmoroctocog alfa, Eloctate/Elocta) can be used to prevent or reduce bleeding episodes in patients with severe hemophilia A.
Researchers found that prophylaxis with rFVIIIFc resulted in low annualized bleeding rates, and patients did not develop neutralizing antibodies.
Furthermore, the product was generally well-tolerated and had a prolonged half-life when compared with recombinant factor VIII.
Data from this study, called A-LONG, have been published in Blood.
Researchers tested rFVIIIFc in 165 male patients with severe hemophilia A who were 12 years of age and older.
Patients were divided into 3 treatment arms. Arm 1 received individualized prophylaxis, or 25 to 65 IU/kg every 3 to 5 days (n=118). Patients in arm 2 received a weekly prophylactic dose of 65 IU/kg (n=24). And patients in arm 3 received episodic treatment at doses of 10 to 50 IU/kg (n=23).
A total of 153 patients completed the study, and 757 bleeding episodes were treated with rFVIIIFc. Across the treatment arms, 87.3% of bleeding episodes were resolved with 1 injection of rFVIIIFc.
The annualized bleeding rate was significantly reduced with prophylaxis—by 92% for patients in arm 1 and 76% for those in arm 2—when compared with episodic treatment.
This was based on annualized bleeding rate estimates from a negative binomial regression model—2.91 for arm 1, 8.92 for arm 2, and 37.25 for arm 3.
The median annualized bleeding rates were 1.6 in arm 1, 3.6 in arm 2, and 33.6 in arm 3.
In arm 3, there were 9 patients who received rFVIIIFc to control bleeding during major surgery. In these cases, physicians rated the hemostatic response as “excellent” (n=8) or “good” (n=1).
There were no serious adverse events related to rFVIIIFc, and none of the patients developed neutralizing antibodies.
The most common adverse events (with an incidence of 5% or more) that occurred outside the perioperative period included nasopharyngitis, arthralgia, headache, and upper respiratory infection.
The researchers also compared the pharmacokinetics of rFVIIIFc and recombinant factor VIII. And they found the terminal half-life of rFVIIIFc was extended 1.5-fold compared to recombinant factor VIII—19.0 hours and 12.4 hours, respectively.
rFVIIIFc was developed using Fc fusion technology, which takes advantage of a naturally occurring pathway that delays the breakdown of IgG1 protein in the body by recycling it back into the bloodstream. This technology prolongs the time rFVIIIFc circulates in the body.
“There is an unmet medical need in the hemophilia community for longer intervals between prophylactic infusions while maintaining good control of bleeding episodes,” said study author Johnny Mahlangu, MD, director of the Haemophilia Comprehensive Care Centre at the University of the Witwatersrand and National Health Laboratory Service in Johannesburg, South Africa.
“A-LONG is the first clinical study to show that effective control over breakthrough bleeding may be achieved with once- or twice-weekly prophylactic infusions in people with severe hemophilia A.”
This study was funded by Biogen Idec, makers of rFVIIIFc.
FDA approves ibrutinib for previously treated MCL
The US Food and Drug Administration (FDA) has has granted accelerated approval for
ibrutinib (Imbruvica) to treat patients with mantle cell lymphoma (MCL) who have received at least 1 prior therapy.
Ibrutinib works by inhibiting the function of Bruton’s tyrosine kinase, a molecule that plays an important role in the survival of malignant B cells.
The drug showed promising results in the phase 2 PCYC-1104 trial, which was presented at ASH 2012 and published in NEJM in August.
The FDA granted ibrutinib breakthrough therapy designation because of these results and the life-threatening nature of MCL. Ibrutinib is the second drug with breakthrough therapy designation to receive FDA approval.
The FDA granted ibrutinib accelerated approval, rather than traditional approval, because the drug has not yet shown a clinical benefit. Accelerated approval of a drug is based
on a surrogate or intermediate endpoint—in this case, overall response rate—that is reasonably likely to
predict clinical benefit.
PCYC-1104 trial
The data published in NEJM included 111 patients who received ibrutinib at 560 mg daily in continuous, 28-day cycles until disease progression.
The
overall response rate was 68%, with a complete response rate of 21% and
a partial response rate of 47%. With an estimated median follow-up of
15.3 months, the estimated median response duration was 17.5 months.
The
estimated progression-free survival was 13.9 months, and the overall
survival was not reached. The estimated rate of overall survival was 58%
at 18 months.
Common nonhematologic adverse events included
diarrhea (50%), fatigue (41%), nausea (31%), peripheral edema (28%),
dyspnea (27%), constipation (25%), upper respiratory tract infection
(23%), vomiting (23%), and decreased appetite (21%). The most common
grade 3, 4, or 5 infection was pneumonia (6%).
Grade 3 and 4
hematologic adverse events included neutropenia (16%), thrombocytopenia
(11%), and anemia (10%). Grade 3 bleeding events occurred in 5 patients.
The “Warnings and Precautions” section of ibrutinib’s prescribing information notes that patients taking ibrutinib have experienced hemorrhage, fatal and non-fatal infections, myelosuppression, renal toxicity, second primary malignancies, and embryo-fetal toxicity.
For the full prescribing information, visit http://www.imbruvica.com/downloads/Prescribing_Information.pdf.
Ibrutinib is now commercially available. It is co-marketed by Pharmacyclics (based in Sunnyvale, California) and Janssen Biotech, Inc. (based in Raritan, New Jersey).
The US Food and Drug Administration (FDA) has has granted accelerated approval for
ibrutinib (Imbruvica) to treat patients with mantle cell lymphoma (MCL) who have received at least 1 prior therapy.
Ibrutinib works by inhibiting the function of Bruton’s tyrosine kinase, a molecule that plays an important role in the survival of malignant B cells.
The drug showed promising results in the phase 2 PCYC-1104 trial, which was presented at ASH 2012 and published in NEJM in August.
The FDA granted ibrutinib breakthrough therapy designation because of these results and the life-threatening nature of MCL. Ibrutinib is the second drug with breakthrough therapy designation to receive FDA approval.
The FDA granted ibrutinib accelerated approval, rather than traditional approval, because the drug has not yet shown a clinical benefit. Accelerated approval of a drug is based
on a surrogate or intermediate endpoint—in this case, overall response rate—that is reasonably likely to
predict clinical benefit.
PCYC-1104 trial
The data published in NEJM included 111 patients who received ibrutinib at 560 mg daily in continuous, 28-day cycles until disease progression.
The
overall response rate was 68%, with a complete response rate of 21% and
a partial response rate of 47%. With an estimated median follow-up of
15.3 months, the estimated median response duration was 17.5 months.
The
estimated progression-free survival was 13.9 months, and the overall
survival was not reached. The estimated rate of overall survival was 58%
at 18 months.
Common nonhematologic adverse events included
diarrhea (50%), fatigue (41%), nausea (31%), peripheral edema (28%),
dyspnea (27%), constipation (25%), upper respiratory tract infection
(23%), vomiting (23%), and decreased appetite (21%). The most common
grade 3, 4, or 5 infection was pneumonia (6%).
Grade 3 and 4
hematologic adverse events included neutropenia (16%), thrombocytopenia
(11%), and anemia (10%). Grade 3 bleeding events occurred in 5 patients.
The “Warnings and Precautions” section of ibrutinib’s prescribing information notes that patients taking ibrutinib have experienced hemorrhage, fatal and non-fatal infections, myelosuppression, renal toxicity, second primary malignancies, and embryo-fetal toxicity.
For the full prescribing information, visit http://www.imbruvica.com/downloads/Prescribing_Information.pdf.
Ibrutinib is now commercially available. It is co-marketed by Pharmacyclics (based in Sunnyvale, California) and Janssen Biotech, Inc. (based in Raritan, New Jersey).
The US Food and Drug Administration (FDA) has has granted accelerated approval for
ibrutinib (Imbruvica) to treat patients with mantle cell lymphoma (MCL) who have received at least 1 prior therapy.
Ibrutinib works by inhibiting the function of Bruton’s tyrosine kinase, a molecule that plays an important role in the survival of malignant B cells.
The drug showed promising results in the phase 2 PCYC-1104 trial, which was presented at ASH 2012 and published in NEJM in August.
The FDA granted ibrutinib breakthrough therapy designation because of these results and the life-threatening nature of MCL. Ibrutinib is the second drug with breakthrough therapy designation to receive FDA approval.
The FDA granted ibrutinib accelerated approval, rather than traditional approval, because the drug has not yet shown a clinical benefit. Accelerated approval of a drug is based
on a surrogate or intermediate endpoint—in this case, overall response rate—that is reasonably likely to
predict clinical benefit.
PCYC-1104 trial
The data published in NEJM included 111 patients who received ibrutinib at 560 mg daily in continuous, 28-day cycles until disease progression.
The
overall response rate was 68%, with a complete response rate of 21% and
a partial response rate of 47%. With an estimated median follow-up of
15.3 months, the estimated median response duration was 17.5 months.
The
estimated progression-free survival was 13.9 months, and the overall
survival was not reached. The estimated rate of overall survival was 58%
at 18 months.
Common nonhematologic adverse events included
diarrhea (50%), fatigue (41%), nausea (31%), peripheral edema (28%),
dyspnea (27%), constipation (25%), upper respiratory tract infection
(23%), vomiting (23%), and decreased appetite (21%). The most common
grade 3, 4, or 5 infection was pneumonia (6%).
Grade 3 and 4
hematologic adverse events included neutropenia (16%), thrombocytopenia
(11%), and anemia (10%). Grade 3 bleeding events occurred in 5 patients.
The “Warnings and Precautions” section of ibrutinib’s prescribing information notes that patients taking ibrutinib have experienced hemorrhage, fatal and non-fatal infections, myelosuppression, renal toxicity, second primary malignancies, and embryo-fetal toxicity.
For the full prescribing information, visit http://www.imbruvica.com/downloads/Prescribing_Information.pdf.
Ibrutinib is now commercially available. It is co-marketed by Pharmacyclics (based in Sunnyvale, California) and Janssen Biotech, Inc. (based in Raritan, New Jersey).
Ibrutinib approved for mantle cell lymphoma
Ibrutinib is now approved for the treatment of patients with mantle cell lymphoma who have received at least one prior therapy, the Food and Drug Administration announced Nov. 13.
The once-daily, oral therapy, marketed as Imbruvica, is the second drug to receive FDA approval under the breakthrough therapy designation established to speed the development and review of treatments for serious or life-threatening diseases.
"Imbruvica’s approval demonstrates the FDA’s commitment to making treatments available to patients with rare diseases," Dr. Richard Pazdur, director of hematology and oncology products in the FDA’s Center for Drug Evaluation and Research, said in a statement.
Mantle cell lymphoma is an orphan disease, with only about 2,900 new cases of MCL diagnosed each year. MCL comprises only about 6% of all non-Hodgkin’s lymphoma cases in the United States.
Ibrutinib’s approval comes a little more than 4 months after the new drug application was filed in June 2013 and is based on a phase II study reporting an investigator-assessed overall response rate of 66% at a daily dose of 560 mg ibrutinib in 111 patients with relapsed or refractory MCL after a median of three prior therapies. The median duration of response was 17.5 months. An improvement in survival and disease-related symptoms has not been established.
Ibrutinib works by blocking Bruton’s tyrosine kinase, a mediator of the B-cell receptor signaling pathway that has been shown in nonclinical studies to inhibit malignant B-cell survival.
The FDA also granted ibrutinib priority review and orphan-product designation, because the drug demonstrated "the potential to be a significant improvement in safety or effectiveness in the treatment of a serious condition and is intended to treat a rare disease," according to the FDA statement.
Ibrutinib is the third drug approved to treat MCL. In June 2013, the FDA approved the oral thalidomide analogue lenalidomide (Revlimid) for the treatment of MCL that had relapsed or progressed after two prior therapies including bortezomib (Velcade), a subcutaneous therapy that has been available for MCL since 2006.
"It is gratifying to see an early example of the new breakthrough therapy designation pathway meeting its intention – getting promising treatments to patients who are waiting for new options," Dr. Ellen V. Sigal, chairperson and founder of the Washington-based Friends of Cancer Research advocacy organization, said in a statement issued by Janssen Biotech, which is comarketing the drug with Pharmacyclics.
The two companies are expected to continue with phase III studies of ibrutinib and have also submitted the drug to the FDA for the treatment of previously treated chronic lymphocytic leukemia/small lymphocytic lymphoma.
In the pivotal MCL trial, the most common treatment-related adverse events with single-agent ibrutinib were mild or moderate diarrhea, fatigue, and nausea (N. Engl. J. Med. 2013;369:507-16). Grade 3 or higher hematologic events were neutropenia (16%), thrombocytopenia (11%), and anemia (10%).
Ibrutinib is now approved for the treatment of patients with mantle cell lymphoma who have received at least one prior therapy, the Food and Drug Administration announced Nov. 13.
The once-daily, oral therapy, marketed as Imbruvica, is the second drug to receive FDA approval under the breakthrough therapy designation established to speed the development and review of treatments for serious or life-threatening diseases.
"Imbruvica’s approval demonstrates the FDA’s commitment to making treatments available to patients with rare diseases," Dr. Richard Pazdur, director of hematology and oncology products in the FDA’s Center for Drug Evaluation and Research, said in a statement.
Mantle cell lymphoma is an orphan disease, with only about 2,900 new cases of MCL diagnosed each year. MCL comprises only about 6% of all non-Hodgkin’s lymphoma cases in the United States.
Ibrutinib’s approval comes a little more than 4 months after the new drug application was filed in June 2013 and is based on a phase II study reporting an investigator-assessed overall response rate of 66% at a daily dose of 560 mg ibrutinib in 111 patients with relapsed or refractory MCL after a median of three prior therapies. The median duration of response was 17.5 months. An improvement in survival and disease-related symptoms has not been established.
Ibrutinib works by blocking Bruton’s tyrosine kinase, a mediator of the B-cell receptor signaling pathway that has been shown in nonclinical studies to inhibit malignant B-cell survival.
The FDA also granted ibrutinib priority review and orphan-product designation, because the drug demonstrated "the potential to be a significant improvement in safety or effectiveness in the treatment of a serious condition and is intended to treat a rare disease," according to the FDA statement.
Ibrutinib is the third drug approved to treat MCL. In June 2013, the FDA approved the oral thalidomide analogue lenalidomide (Revlimid) for the treatment of MCL that had relapsed or progressed after two prior therapies including bortezomib (Velcade), a subcutaneous therapy that has been available for MCL since 2006.
"It is gratifying to see an early example of the new breakthrough therapy designation pathway meeting its intention – getting promising treatments to patients who are waiting for new options," Dr. Ellen V. Sigal, chairperson and founder of the Washington-based Friends of Cancer Research advocacy organization, said in a statement issued by Janssen Biotech, which is comarketing the drug with Pharmacyclics.
The two companies are expected to continue with phase III studies of ibrutinib and have also submitted the drug to the FDA for the treatment of previously treated chronic lymphocytic leukemia/small lymphocytic lymphoma.
In the pivotal MCL trial, the most common treatment-related adverse events with single-agent ibrutinib were mild or moderate diarrhea, fatigue, and nausea (N. Engl. J. Med. 2013;369:507-16). Grade 3 or higher hematologic events were neutropenia (16%), thrombocytopenia (11%), and anemia (10%).
Ibrutinib is now approved for the treatment of patients with mantle cell lymphoma who have received at least one prior therapy, the Food and Drug Administration announced Nov. 13.
The once-daily, oral therapy, marketed as Imbruvica, is the second drug to receive FDA approval under the breakthrough therapy designation established to speed the development and review of treatments for serious or life-threatening diseases.
"Imbruvica’s approval demonstrates the FDA’s commitment to making treatments available to patients with rare diseases," Dr. Richard Pazdur, director of hematology and oncology products in the FDA’s Center for Drug Evaluation and Research, said in a statement.
Mantle cell lymphoma is an orphan disease, with only about 2,900 new cases of MCL diagnosed each year. MCL comprises only about 6% of all non-Hodgkin’s lymphoma cases in the United States.
Ibrutinib’s approval comes a little more than 4 months after the new drug application was filed in June 2013 and is based on a phase II study reporting an investigator-assessed overall response rate of 66% at a daily dose of 560 mg ibrutinib in 111 patients with relapsed or refractory MCL after a median of three prior therapies. The median duration of response was 17.5 months. An improvement in survival and disease-related symptoms has not been established.
Ibrutinib works by blocking Bruton’s tyrosine kinase, a mediator of the B-cell receptor signaling pathway that has been shown in nonclinical studies to inhibit malignant B-cell survival.
The FDA also granted ibrutinib priority review and orphan-product designation, because the drug demonstrated "the potential to be a significant improvement in safety or effectiveness in the treatment of a serious condition and is intended to treat a rare disease," according to the FDA statement.
Ibrutinib is the third drug approved to treat MCL. In June 2013, the FDA approved the oral thalidomide analogue lenalidomide (Revlimid) for the treatment of MCL that had relapsed or progressed after two prior therapies including bortezomib (Velcade), a subcutaneous therapy that has been available for MCL since 2006.
"It is gratifying to see an early example of the new breakthrough therapy designation pathway meeting its intention – getting promising treatments to patients who are waiting for new options," Dr. Ellen V. Sigal, chairperson and founder of the Washington-based Friends of Cancer Research advocacy organization, said in a statement issued by Janssen Biotech, which is comarketing the drug with Pharmacyclics.
The two companies are expected to continue with phase III studies of ibrutinib and have also submitted the drug to the FDA for the treatment of previously treated chronic lymphocytic leukemia/small lymphocytic lymphoma.
In the pivotal MCL trial, the most common treatment-related adverse events with single-agent ibrutinib were mild or moderate diarrhea, fatigue, and nausea (N. Engl. J. Med. 2013;369:507-16). Grade 3 or higher hematologic events were neutropenia (16%), thrombocytopenia (11%), and anemia (10%).
Opioids and Opioid‐Related Adverse Events
Recent reports have drawn attention to the high and increasing rates of opioid prescribing and overdose‐related deaths in the United States.[1, 2, 3, 4, 5, 6, 7, 8, 9] These studies have focused on community‐based and emergency department prescribing, leaving prescribing practices in the inpatient setting unexamined. Given that pain is a frequent complaint in hospitalized patients, and that the Joint Commission mandates assessing pain as a vital sign, hospitalization is potentially a time of heightened use of such medications and could significantly contribute to nosocomial complications and subsequent outpatient use.[10] Variation in prescribing practices, unrelated to patient characteristics, could be a marker of inappropriate prescribing practices and poor quality of care.
Using a large, nationally representative cohort of admissions from July 2009 to June 2010, we sought to determine patterns and predictors of opioid utilization in nonsurgical admissions to US medical centers, hospital variation in use, and the association between hospital‐level use and the risk of opioid‐related adverse events. We hypothesized that hospitals with higher rates of opioid use would have an increased risk of an opioid‐related adverse event per patient exposed.
METHODS
Setting and Patients
We conducted a retrospective cohort study using data from 286 US nonfederal acute‐care facilities contributing to the database maintained by Premier (Premier Healthcare Solutions, Inc., Charlotte, NC). This database, created to measure healthcare utilization and quality of care, is drawn from voluntarily participating hospitals and contains data on approximately 1 in every 4 discharges nationwide.[11] Participating hospitals are similar in geographic distribution and metropolitan (urban/rural) status to hospitals nationwide, although large, nonteaching hospitals are slightly overrepresented in Premier. The database contains patient demographics, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes, hospital demographics, and a date‐stamped log of all charges during the course of each hospitalization, including diagnostic tests, therapeutic treatments, and medications with dose and route of administration. The study was approved by the institutional review board at Beth Israel Deaconess Medical Center and granted a waiver of informed consent.
We studied a cohort of all adult nonsurgical admissions to participating hospitals from July 1, 2009, through June 30, 2010. We chose to study nonsurgical admissions, as patients undergoing surgical procedures have a clear indication for, and almost always receive, opioid pain medications. We defined a nonsurgical admission as an admission in which there were no charges for operating‐room procedures (including labor and delivery) and the attending of record was not a surgeon. We excluded admissions with unknown gender, since this is a key demographic variable, and admissions with a length of stay greater than 365 days, as these admissions are not representative of the typical admission to an acute‐care hospital. At the hospital level, we excluded hospitals contributing <100 admissions owing to resultant lack of precision in corresponding hospital prescribing rates, and hospitals that did not prescribe the full range of opioid medications (these hospitals had charges for codeine only), as these facilities seemed likely to have unusual limitations on prescribing or incomplete data capture.
Opioid Exposure
We defined opioid exposure as presence of 1 charge for an opioid medication during the admission. Opioid medications included morphine, hydrocodone, hydromorphone, oxycodone, fentanyl, meperidine, methadone, codeine, tramadol, buprenorphine, levorphanol, oxymorphone, pentazocine, propoxyphene, tapentadol, butorphanol, dezocine, and nalbuphine. We grouped the last 9 into an other category owing to infrequent use and/or differing characteristics from the main opioid drug types, such as synthetic, semisynthetic, and partial agonist qualities.
Severe Opioid‐Related Adverse Events
We defined severe opioid‐related adverse events as either naloxone exposure or an opioid‐related adverse drug event diagnosis code. Naloxone use in an adult patient exposed to opioids is one of the Institute for Healthcare Improvement's Trigger Tools for identifying adverse drug events[12] and previously has been demonstrated to have high positive predictive value for a confirmed adverse drug event.[13] We defined naloxone exposure as presence of 1 charge for naloxone. We excluded charges on hospital day 1 to focus on nosocomial events. We defined opioid‐related adverse drug events using ICD‐9‐CM diagnosis codes for poisoning by opioids (overdose, wrong substance given, or taken in error; ICD‐9‐CM 965.02, 965.09, E850.1, E850.2) and drugs causing adverse effects in therapeutic use (ICD‐9‐CM E935.1, E935.2), as specified in prior analyses by the Agency for Healthcare Research and Quality (AHRQ).[14, 15] To avoid capturing adverse events associated with outpatient use, we required the ICD‐9‐CM code to be qualified as not present on admission using the present on admission indicator required by the Centers for Medicare and Medicaid Services for all discharge diagnosis codes since 2008.[16]
Covariates of Interest
We were interested in the relationship between both patient and hospital characteristics and opioid exposure. Patient characteristics of interest included (1) demographic variables, such as age, sex, race (self‐reported by patients at the time of admission), marital status, and payer; (2) whether or not the patient spent any time in the intensive care unit (ICU); (3) comorbidities, identified via ICD‐9‐CM secondary diagnosis codes and diagnosis‐related groups using Healthcare Cost and Utilization Project Comorbidity Software, version 3.7, based on the work of Elixhauser et al[17, 18]; (4) primary ICD‐9‐CM discharge diagnosis groupings, selected based on hypothesized associations with receipt of opioids, and based on the Clinical Classifications Software (CCS)a diagnosis and procedure categorization scheme maintained by the AHRQ, and defined in the Appendix[19]; (5) and nonoperating‐room‐based procedures potentially necessitating opioids during the admission, selected from the 50 most common ICD‐9‐CM procedure codes in our cohort and grouped as cardiovascular procedures (catheterization and insertion of vascular stents), gastrointestinal procedures (upper and lower endoscopy), and mechanical ventilation, further defined in the Appendix. Hospital characteristics of interest included number of beds, population served (urban vs rural), teaching status, and US census region (Northeast, Midwest, South, West).
Statistical Analysis
We calculated the percent of admissions with exposure to any opioid and the percent exposed to each opioid, along with the total number of different opioid medications used during each admission. We also calculated the percent of admissions with parenteral administration and the percent of admissions with oral administration, among those exposed to the individual categories, and in aggregate. Because medications after discharge were unavailable in Premier's dataset, we report the percent of patients with a charge for opioids on the day of discharge.
We determined the daily dose of an opioid by taking the sum of the doses for that opioid charged on a given day. The average daily dose of an opioid was determined by taking the sum of the daily doses and dividing by the number of days on which 1 dose was charged. To facilitate comparison, all opioids, with the exception of those for which standard equivalences are unavailable (tramadol, other opioid category, oral fentanyl, epidural route for all), were converted to oral morphine equivalents using a standard equivalence conversion table.[20, 21] We excluded from our dosage calculations those charges for which standard morphine equivalence was unavailable, or for which dosage was missing. We also excluded from our dosage calculations any dose that was >3 standard deviations (SD) above the mean dose for that opioid, as such extreme values seemed physiologically implausible and more likely to be a data entry error that could lead to significant overestimation of the mean for that opioid.
All multivariable models used a generalized estimating equation (GEE) via the genmod procedure in SAS, with a Poisson distribution error term and a log link, controlling for repeated patient admissions with an autoregressive correlation structure.
To identify independent predictors of opioid receipt, we used a GEE model of opioid receipt where all patient and hospital characteristics listed in Table 1 were included as independent variables.
| Patient characteristics, N=1,139,419 | N | % |
|---|---|---|
| ||
| Age group, y | ||
| 1824 | 37,464 | 3 |
| 2534 | 66,541 | 6 |
| 3544 | 102,701 | 9 |
| 4554 | 174,830 | 15 |
| 5564 | 192,570 | 17 |
| 6574 | 196,407 | 17 |
| 75+ | 368,906 | 32 |
| Sex | ||
| Male | 527,062 | 46 |
| Female | 612,357 | 54 |
| Race | ||
| White | 711,993 | 62 |
| Black | 176,993 | 16 |
| Hispanic | 54,406 | 5 |
| Other | 196,027 | 17 |
| Marital status | ||
| Married | 427,648 | 38 |
| Single | 586,343 | 51 |
| Unknown/other | 125,428 | 11 |
| Primary insurance | ||
| Private/commercial | 269,725 | 24 |
| Medicare traditional | 502,301 | 44 |
| Medicare managed care | 126,344 | 11 |
| Medicaid | 125,025 | 11 |
| Self‐pay/other | 116,024 | 10 |
| ICU care | ||
| No | 1,023,027 | 90 |
| Yes | 116,392 | 10 |
| Comorbidities | ||
| AIDS | 5724 | 1 |
| Alcohol abuse | 79,633 | 7 |
| Deficiency anemias | 213,437 | 19 |
| RA/collagen vascular disease | 35,210 | 3 |
| Chronic blood‐loss anemia | 10,860 | 1 |
| CHF | 190,085 | 17 |
| Chronic pulmonary disease | 285,954 | 25 |
| Coagulopathy | 48,513 | 4 |
| Depression | 145,553 | 13 |
| DM without chronic complications | 270,087 | 24 |
| DM with chronic complications | 70,732 | 6 |
| Drug abuse | 66,886 | 6 |
| Hypertension | 696,299 | 61 |
| Hypothyroidism | 146,136 | 13 |
| Liver disease | 38,130 | 3 |
| Lymphoma | 14,032 | 1 |
| Fluid and electrolyte disorders | 326,576 | 29 |
| Metastatic cancer | 33,435 | 3 |
| Other neurological disorders | 124,195 | 11 |
| Obesity | 118,915 | 10 |
| Paralysis | 38,584 | 3 |
| PVD | 77,334 | 7 |
| Psychoses | 101,856 | 9 |
| Pulmonary circulation disease | 52,106 | 5 |
| Renal failure | 175,398 | 15 |
| Solid tumor without metastasis | 29,594 | 3 |
| Peptic ulcer disease excluding bleeding | 536 | 0 |
| Valvular disease | 86,616 | 8 |
| Weight loss | 45,132 | 4 |
| Primary discharge diagnoses | ||
| Cancer | 19,168 | 2 |
| Musculoskeletal injuries | 16,798 | 1 |
| Pain‐related diagnosesb | 101,533 | 9 |
| Alcohol‐related disorders | 16,777 | 1 |
| Substance‐related disorders | 13,697 | 1 |
| Psychiatric disorders | 41,153 | 4 |
| Mood disorders | 28,761 | 3 |
| Schizophrenia and other psychotic disorders | 12,392 | 1 |
| Procedures | ||
| Cardiovascular procedures | 59,901 | 5 |
| GI procedures | 31,224 | 3 |
| Mechanical ventilation | 7853 | 1 |
| Hospital characteristics, N=286 | ||
| Number of beds | ||
| <200 | 103 | 36 |
| 201300 | 63 | 22 |
| 301500 | 81 | 28 |
| >500 | 39 | 14 |
| Population served | ||
| Urban | 225 | 79 |
| Rural | 61 | 21 |
| Teaching status | ||
| Nonteaching | 207 | 72 |
| Teaching | 79 | 28 |
| US Census region | ||
| Northeast | 47 | 16 |
| Midwest | 63 | 22 |
| South | 115 | 40 |
| West | 61 | 21 |
To assess hospital variation in opioid prescribing after adjusting for patient characteristics, we used a GEE model of opioid receipt, controlling for all patient characteristics listed in Table 1. We then took the mean of the predicted probabilities of opioid receipt for the patients within each hospital in our cohort to derive the hospital prescribing rate adjusted for patient characteristics. We report the mean, SD, and range of the prescribing rates for the hospitals in our cohort before and after adjustment for patient characteristics.
To assess whether patients admitted to hospitals with higher rates of opioid prescribing have higher relative risk of severe opioid‐related adverse events, we stratified hospitals into opioid‐prescribing rate quartiles and compared the rates of opioid‐related adverse eventsboth overall and among opioid exposedbetween quartiles. To adjust for patient characteristics, we used a GEE model in which severe opioid‐related adverse event (yes/no) was the dependent variable and hospital‐prescribing rate quartile and all patient characteristics in Table 1 were independent variables. We also performed a sensitivity analysis in which we assessed the association between hospital prescribing‐rate quartile and the individual components of our composite outcome. Our results were qualitatively unchanged using this approach, and only the results of our main analysis are presented.
All analyses were carried out using SAS software, version 9.2 (SAS Institute Inc., Cary, NC).
RESULTS
Patient Admission Characteristics
There were 3,190,934 adult admissions to 300 acute‐care hospitals during our study period. After excluding admissions with a length of stay >365 days (n=25), missing patient sex (n=17), and charges for operating‐room procedures or a surgical attending of record (n=2,018,553), 1,172,339 admissions were available for analysis. There were 12 hospitals with incomplete opioid‐prescribing data (n=32,794) and 2 hospitals that contributed <100 admissions each (n=126), leaving 1,139,419 admissions in 286 hospitals in our analytic cohort. The median age of the cohort was 64 years (interquartile range, 4979 years), and 527,062 (46%) were men. Table 1 shows the characteristics of the admissions in the cohort.
Rate, Route, and Dose of Opioid Exposures
Overall, there were 576,373 (51%) admissions with charges for opioid medications. Among those exposed, 244,760 (43%) had charges for multiple opioids during the admission; 172,090 (30%) had charges for 2 different opioids; and 72,670 (13%) had charges for 3 different opioids. Table 2 shows the percent exposed to each opioid, the percent of exposed with parenteral and oral routes of administration, and the mean daily dose received in oral morphine equivalents.
| Exposed | Parenteral Administration | Oral Administration | Dose Received, in Oral Morphine Equivalents | |||||
|---|---|---|---|---|---|---|---|---|
| N | %a | N | %b | N | %b | Mean | SDc | |
| ||||||||
| All opioids | 576,373 | 51 | 378,771 | 66 | 371,796 | 65 | 68 | 185 |
| Morphine | 224,811 | 20 | 209,040 | 93 | 21,645 | 10 | 40 | 121 |
| Hydrocodone | 162,558 | 14 | 0 | 0 | 160,941 | 99 | 14 | 12 |
| Hydromorphone | 146,236 | 13 | 137,936 | 94 | 16,052 | 11 | 113 | 274 |
| Oxycodone | 126,733 | 11 | 0 | 0 | 125,033 | 99 | 26 | 37 |
| Fentanyl | 105,052 | 9 | 103,113 | 98 | 641 | 1 | 64 | 75 |
| Tramadol | 35,570 | 3 | 0 | 0 | 35,570 | 100 | ||
| Meperidine | 24,850 | 2 | 24,398 | 98 | 515 | 2 | 36 | 34 |
| Methadone | 15,302 | 1 | 370 | 2 | 14,781 | 97 | 337 | 384 |
| Codeine | 22,818 | 2 | 178 | 1 | 22,183 | 97 | 9 | 15 |
| Other | 45,469 | 4 | 5821 | 13 | 39,618 | 87 | ||
Among the medications/routes for which conversion to morphine equivalents was possible, dosage was missing in 39,728 out of 2,294,673 opioid charges (2%). The average daily dose received in oral morphine equivalents was 68 mg. A total dose of 50 mg per day was received in 39% of exposed, and a total dose of 100 mg per day was received in 23% of exposed. Among those exposed, 52% (26% of overall admissions) had charges for opioids on the day of discharge.
Rates of Opioid Use by Patient and Hospital Characteristics
Table 3 reports the association between admission characteristics and opioid use. Use was highest in patients between the ages of 25 and 54 years. Although use declined with age, 44% of admissions age 65 years had charges for opioid medication. After adjustment for patient demographics, comorbidities, and hospital characteristics, opioid use was more common in females than males, those age 2554 years compared with those older and younger, those of Caucasian race compared with non‐Caucasian race, and those with Medicare or Medicaid primary insurance. Among the primary discharge diagnoses, patients with musculoskeletal injuries, various specific and nonspecific pain‐related diagnoses, and cancer were significantly more likely to receive opioids than patients without these diagnoses, whereas patients with alcohol‐related disorders and psychiatric disorders were significantly less likely to receive opioids than patients without these diagnoses. Patients admitted to hospitals in the Midwest, South, and West were significantly more likely to receive opioid medications than patients in the Northeast.
| Exposed, N=576,373 | Unexposed, N=563,046 | % Exposed | Adjusted RRa | 95% CI | |
|---|---|---|---|---|---|
| |||||
| Patient characteristics | |||||
| Age group, y | |||||
| 1824 | 17,360 | 20,104 | 46 | (ref) | |
| 2534 | 37,793 | 28,748 | 57 | 1.17 | 1.16‐1.19 |
| 3544 | 60,712 | 41,989 | 59 | 1.16 | 1.15‐1.17 |
| 4554 | 103,798 | 71,032 | 59 | 1.11 | 1.09‐1.12 |
| 5564 | 108,256 | 84,314 | 56 | 1.00 | 0.98‐1.01 |
| 6574 | 98,110 | 98,297 | 50 | 0.84 | 0.83‐0.85 |
| 75+ | 150,344 | 218,562 | 41 | 0.71 | 0.70‐0.72 |
| Sex | |||||
| Male | 255,315 | 271,747 | 48 | (ref) | |
| Female | 321,058 | 291,299 | 52 | 1.11 | 1.10‐1.11 |
| Race | |||||
| White | 365,107 | 346,886 | 51 | (ref) | |
| Black | 92,013 | 84,980 | 52 | 0.93 | 0.92‐0.93 |
| Hispanic | 27,592 | 26,814 | 51 | 0.94 | 0.93‐0.94 |
| Other | 91,661 | 104,366 | 47 | 0.93 | 0.92‐0.93 |
| Marital status | |||||
| Married | 222,912 | 204,736 | 52 | (ref) | |
| Single | 297,742 | 288,601 | 51 | 1.00 | 0.99‐1.01 |
| Unknown/other | 55,719 | 69,709 | 44 | 0.94 | 0.93‐0.95 |
| Primary insurance | |||||
| Private/commercial | 143,954 | 125,771 | 53 | (ref) | |
| Medicare traditional | 236,114 | 266,187 | 47 | 1.10 | 1.09‐1.10 |
| Medicare managed care | 59,104 | 67,240 | 47 | 1.11 | 1.11‐1.12 |
| Medicaid | 73,583 | 51,442 | 59 | 1.13 | 1.12‐1.13 |
| Self‐pay/other | 63,618 | 52,406 | 55 | 1.03 | 1.02‐1.04 |
| ICU care | |||||
| No | 510,654 | 512,373 | 50 | (ref) | |
| Yes | 65,719 | 50,673 | 56 | 1.02 | 1.01‐1.03 |
| Comorbiditiesb | |||||
| AIDS | 3655 | 2069 | 64 | 1.09 | 1.07‐1.12 |
| Alcohol abuse | 35,112 | 44,521 | 44 | 0.92 | 0.91‐0.93 |
| Deficiency anemias | 115,842 | 97,595 | 54 | 1.08 | 1.08‐1.09 |
| RA/collagen vascular disease | 22,519 | 12,691 | 64 | 1.22 | 1.21‐1.23 |
| Chronic blood‐loss anemia | 6444 | 4416 | 59 | 1.04 | 1.02‐1.05 |
| CHF | 88,895 | 101,190 | 47 | 0.99 | 0.98‐0.99 |
| Chronic pulmonary disease | 153,667 | 132,287 | 54 | 1.08 | 1.08‐1.08 |
| Coagulopathy | 25,802 | 22,711 | 53 | 1.03 | 1.02‐1.04 |
| Depression | 83,051 | 62,502 | 57 | 1.08 | 1.08‐1.09 |
| DM without chronic complications | 136,184 | 133,903 | 50 | 0.99 | 0.99‐0.99 |
| DM with chronic complications | 38,696 | 32,036 | 55 | 1.04 | 1.03‐1.05 |
| Drug abuse | 37,202 | 29,684 | 56 | 1.14 | 1.13‐1.15 |
| Hypertension | 344,718 | 351,581 | 50 | 0.98 | 0.97‐0.98 |
| Hypothyroidism | 70,786 | 75,350 | 48 | 0.99 | 0.99‐0.99 |
| Liver disease | 24,067 | 14,063 | 63 | 1.15 | 1.14‐1.16 |
| Lymphoma | 7727 | 6305 | 55 | 1.16 | 1.14‐1.17 |
| Fluid and electrolyte disorders | 168,814 | 157,762 | 52 | 1.04 | 1.03‐1.04 |
| Metastatic cancer | 23,920 | 9515 | 72 | 1.40 | 1.39‐1.42 |
| Other neurological disorders | 51,091 | 73,104 | 41 | 0.87 | 0.86‐0.87 |
| Obesity | 69,584 | 49,331 | 59 | 1.05 | 1.04‐1.05 |
| Paralysis | 17,497 | 21,087 | 45 | 0.97 | 0.96‐0.98 |
| PVD | 42,176 | 35,158 | 55 | 1.11 | 1.11‐1.12 |
| Psychoses | 38,638 | 63,218 | 38 | 0.91 | 0.90‐0.92 |
| Pulmonary circulation disease | 26,656 | 25,450 | 51 | 1.05 | 1.04‐1.06 |
| Renal failure | 86,565 | 88,833 | 49 | 1.01 | 1.01‐1.02 |
| Solid tumor without metastasis | 16,258 | 13,336 | 55 | 1.14 | 1.13‐1.15 |
| Peptic ulcer disease excluding bleeding | 376 | 160 | 70 | 1.12 | 1.07‐1.18 |
| Valvular disease | 38,396 | 48,220 | 44 | 0.93 | 0.92‐0.94 |
| Weight loss | 25,724 | 19,408 | 57 | 1.09 | 1.08‐1.10 |
| Primary discharge diagnosesb | |||||
| Cancer | 13,986 | 5182 | 73 | 1.20 | 1.19‐1.21 |
| Musculoskeletal injuries | 14,638 | 2160 | 87 | 2.02 | 2.002.04 |
| Pain‐related diagnosesc | 64,656 | 36,877 | 64 | 1.20 | 1.20‐1.21 |
| Alcohol‐related disorders | 3425 | 13,352 | 20 | 0.46 | 0.44‐0.47 |
| Substance‐related disorders | 8680 | 5017 | 63 | 1.03 | 1.01‐1.04 |
| Psychiatric disorders | 7253 | 33,900 | 18 | 0.37 | 0.36‐0.38 |
| Mood disorders | 5943 | 22,818 | 21 | ||
| Schizophrenia and other psychotic disorders | 1310 | 11,082 | 11 | ||
| Proceduresb | |||||
| Cardiovascular procedures | 50,997 | 8904 | 85 | 1.80 | 1.79‐1.81 |
| GI procedures | 27,206 | 4018 | 87 | 1.70 | 1.69‐1.71 |
| Mechanical ventilation | 5341 | 2512 | 68 | 1.37 | 1.34‐1.39 |
| Hospital characteristics | |||||
| Number of beds | |||||
| <200 | 100,900 | 88,439 | 53 | (ref) | |
| 201300 | 104,213 | 99,995 | 51 | 0.95 | 0.95‐0.96 |
| 301500 | 215,340 | 209,104 | 51 | 0.94 | 0.94‐0.95 |
| >500 | 155,920 | 165,508 | 49 | 0.96 | 0.95‐0.96 |
| Population served | |||||
| Urban | 511,727 | 506,803 | 50 | (ref) | |
| Rural | 64,646 | 56,243 | 53 | 0.98 | 0.97‐0.99 |
| Teaching status | |||||
| Nonteaching | 366,623 | 343,581 | 52 | (ref) | |
| Teaching | 209,750 | 219,465 | 49 | 1.00 | 0.99‐1.01 |
| US Census region | |||||
| Northeast | 99,377 | 149,446 | 40 | (ref) | |
| Midwest | 123,194 | 120,322 | 51 | 1.26 | (1.25‐1.27) |
| South | 251,624 | 213,029 | 54 | 1.33 | (1.33‐1.34) |
| West | 102,178 | 80,249 | 56 | 1.37 | (1.36‐1.38) |
Variation in Opioid Prescribing
Figure 1 shows the histograms of hospital opioid prescribing rate for the 286 hospitals in our cohort (A) before and (B) after adjustment for patient characteristics. The observed rates ranged from 5% in the lowest‐prescribing hospital to 72% in the highest‐prescribing hospital, with a mean (SD) of 51% (10%). After adjusting for patient characteristics, the adjusted opioid‐prescribing rates ranged from 33% to 64%, with a mean (SD) of 50% (4%).
Severe Opioid‐Related Adverse Events
Among admissions with opioid exposure (n=576,373), naloxone use occurred in 2345 (0.41%) and opioid‐related adverse drug events in 1174 (0.20%), for a total of 3441 (0.60%) severe opioid‐related adverse events (some patients experienced both). Table 4 reports the opioid exposure and severe opioid‐related adverse‐event rates within hospital opioid‐prescribing rate quartiles, along with the adjusted association between the hospital opioid‐prescribing rate quartile and severe opioid‐related adverse events. After adjusting for patient characteristics, the relative risk of a severe opioid‐related adverse event was significantly greater in hospitals with higher opioid‐prescribing rates, both overall and among opioid exposed.
| Quartile | No. of Patients | Opioid Exposed, n (%) | Opioid‐Related Adverse Events, n (%) | Adjusted RR in All Patients, RR (95% CI), N=1,139,419a | Adjusted RR in Opioid Exposed, RR (95% CI), N=576,373a |
|---|---|---|---|---|---|
| |||||
| 1 | 349,747 | 132,824 (38) | 719 (0.21) | (ref) | (ref) |
| 2 | 266,652 | 134,590 (50) | 729 (0.27) | 1.31 (1.17‐1.45) | 1.07 (0.96‐1.18) |
| 3 | 251,042 | 139,770 (56) | 922 (0.37) | 1.72 (1.56‐1.90) | 1.31 (1.19‐1.44) |
| 4 | 271,978 | 169,189 (62) | 1071 (0.39) | 1.73 (1.57‐1.90) | 1.23 (1.12‐1.35) |
DISCUSSION
In this analysis of a large cohort of hospitalized nonsurgical patients, we found that more than half of all patients received opioids, with 43% of those exposed receiving multiple opioids during their admission and 52% receiving opioids on the day of discharge. Considerable hospital variation in opioid use was evident, and this was not fully explained by patient characteristics. Severe opioid‐related adverse events occurred more frequently at hospitals with higher opioid‐prescribing rates, and the relative risk of a severe adverse event per patient prescribed opioids was also higher in these hospitals. To our knowledge, this is the first study to describe the scope of opioid utilization and the relationship between utilization and severe opioid‐related adverse events in a sample of nonsurgical patients in US acute‐care facilities.
Our use of naloxone charges and opioid‐specific ICD‐9‐CM coding to define an opioid‐related adverse event was intended to capture only the most severe opioid‐related adverse events. We chose to focus on these events in our analysis to maximize the specificity of our outcome definition and thereby minimize confounding in our observed associations. The rate of less‐severe opioid‐related adverse events, such as nausea, constipation, and pruritis, is likely much higher and not captured in our outcome definition. Prior analyses have found variable rates of opioid‐related adverse events of approximately 1.8% to 13.6% of exposed patients.[22, 23, 24] However, these analyses focused on surgical patients and included less‐severe events. To our knowledge, ours is the first analysis of severe opioid‐related adverse events in nonsurgical patients.
Our finding that severe opioid‐related adverse events increase as opioid prescribing increases is consistent with that which has been demonstrated in the community setting, where rates of opioid‐related adverse events and mortality are higher in communities with higher levels of opioid prescribing.[2, 8, 25] This finding is expected, as greater use of a class of medications with known side effects would be expected to result in a higher overall rate of adverse events. More concerning, however, is the fact that this relationship persists when focusing exclusively on opioid‐exposed patients. Among similar patients receiving opioids at different hospitals, those hospitalized in facilities with higher opioid‐prescribing rates have higher rates of severe opioid‐related adverse events. This suggests that hospitals that use opioids more frequently do not do so more safely. Rather, the increased overall prescribing rates are associated with heightened risk for a serious adverse event per patient exposed and may reflect unsafe prescribing practices.
Furthermore, our results demonstrate both regional and hospital variation in use of opioids not fully explained by patient characteristics, similar to that which has been demonstrated for other drugs and heathcare services.[26, 27, 28, 29, 30] The implications of these findings are limited by our lack of information on pain severity or prior outpatient treatment, and our resultant inability to evaluate the appropriateness of opioid use in this analysis. Additionally, although we controlled for a large number of patient and hospital characteristics, there could be other significant predictors of use not accounted for in our analysis. However, it seems unlikely that differential pain severity or patient characteristics between patients in different regions of the country could fully explain a 37% relative difference in prescribing between the lowest‐ and highest‐prescribing regions, after accounting for the 44 patient‐level variables in our models. Whereas variation in use unrelated to patient factors could represent inappropriate prescribing practices, it could also be a marker of uncertainty regarding what constitutes appropriate prescribing and high‐quality care in this realm. Although guidelines advocate for standard pain assessments and a step‐up approach to treatment,[31, 32, 33] the lack of objective measures of pain severity and lack of evidence‐based recommendations on the use of opioids for noncancer pain[34] will almost certainly lead to persistent variation in opioid prescribing despite guideline‐driven care.
Nonetheless, our findings suggest that opportunities exist to make opioid prescribing safer in hospitalized patients. Studies aimed at elucidating the source of regional and hospital variation are necessary. Additionally, efforts should focus on identifying patient and prescribing characteristics associated with heightened risk of opioid‐related adverse events. Prior studies have demonstrated that the risks of opioid medications increase with increasing age of the patient.[35, 36] Although opioid use in our cohort declined with age, 44% of admissions age 65 years had charges for opioid medications. Studies in outpatients have also demonstrated that the risks of opioid overdose and overdose‐related death increase with dose.[5, 7] One study demonstrated a 3.7‐fold increased risk of overdose at doses of 5099 mg/day in oral morphine equivalents, and an 8.9‐fold increased risk at doses of 100 mg/day, compared with doses of 20 mg/day.[7] The prevalence of high dose exposure observed in our cohort, coupled with the older age of hospitalized patients, suggests potential targets for promoting safer use in hospitalized patients through interventions such as computerized decision support and enhanced monitoring in those at highest risk.
Because medications after discharge were unavailable in our dataset, the percentage of patients given a prescription for opioid medication on discharge is unknown. However, given that opioids are often tapered rather than abruptly discontinued, our finding that 26% of hospitalized nonsurgical patients received opioids on the day of discharge suggests that a substantial proportion of patients may be discharged with a prescription for opioid medication. Given the possibility of coexistent outpatient opioid prescriptions, these findings draw attention to the importance of assuring development and streamlined accessibility of data from state prescription drug monitoring programs and suggest that increased attention should be paid to the role that inpatient opioid prescribing plays in the increased rates of chronic opioid use and overdose‐related deaths in the United States.
There are additional limitations to our analysis. First, although the database used for this analysis captures a large proportion of admissions to US acute‐care facilities and is similar in composition, it is possible that participating medical centers differ from nonparticipating medical centers in ways that could be associated with opioid prescribing. Additionally, although Premier performs extensive validation and correction processes to assure the quality of their data, there is still likely to be a small amount of random error in the database, which could particularly impact dosage calculations. The lack of preadmission medications in our database precluded identification of the proportion of patients newly started on opioid medications. Lastly, it is possible that the hospital prescribing‐rate quartile is associated with patient characteristics unaccounted for in our analysis, and, therefore, the possibility of residual confounding still exists.
In conclusion, the majority of hospitalized nonsurgical patients are exposed to opioid medications during the course of their hospitalizations, often at high doses. More than half of those exposed are still receiving these medications on the day of discharge. We found hospital and regional variation in opioid use that was not fully explained by patient characteristics, and higher levels of hospital use were associated with higher risk of severe opioid‐related adverse events in opioid‐exposed patients. Further research is necessary to investigate the appropriateness of opioid use in this patient population, the sources of variation in use, and the predictors of opioid‐related adverse events in hospitalized patients to allow development of interventions to make hospital use safer.
Disclosures
Disclosures: Dr. Herzig had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Herzig was funded by grant no. K23AG042459 from the National Institute on Aging. Dr. Marcantonio was funded by grant nos. P01AG031720, R01AG030618, R03AG028189, and K24AG035075 from the National Institute on Aging. The funding organization had no involvement in any aspect of the study, including design, conduct, and reporting of the study. None of the authors have any conflicts of interest to disclose.
- . A flood of opioids, a rising tide of deaths. N Engl J Med. 2010;363(21):1981–1985.
- , , . Increasing deaths from opioid analgesics in the United States. Pharmacoepidemiol Drug Saf. 2006;15(9):618–627.
- , , , . Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70–78.
- , , , . Trends in medical use and abuse of opioid analgesics. JAMA. 2000;283(13):1710–1714.
- , , , et al. Association between opioid prescribing patterns and opioid overdose‐related deaths. JAMA. 2011;305(13):1315–1321.
- , , , et al. Prescription opioid mortality trends in New York City, 1990–2006: examining the emergence of an epidemic. Drug Alcohol Depend. 2013;132(1‐2):53–62.
- , , , et al. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med. 2010;152(2):85–92.
- , , , et al. Relationship of opioid prescription sales and overdoses, North Carolina. Drug Alcohol Depend. 2013;132(1‐2):81–86.
- Deaths from prescription opioids soar in New York. BMJ. 2013;346:f921.
- , , . Mortality in elderly dementia patients treated with risperidone. J Clin Psychopharmacol. 2006;26(6):566–570.
- , , , et al. Medication reconciliation: a practical tool to reduce the risk of medication errors. J Crit Care. 2003;18(4):201–205.
- , , . Adverse drug event trigger tool: a practical methodology for measuring medication related harm. Qual Saf Health Care. 2003;12(3):194–200.
- , , , , . Improvement in the detection of adverse drug events by the use of electronic health and prescription records: an evaluation of two trigger tools. Eur J Clin Pharmacol. 2013;69(2):255–259.
- , . Adverse Drug Events in U.S. Hospitals, 2004. Rockville, MD: Agency for Healthcare Research and Quality; April 2007. Statistical Brief 29.
- , , . Medication‐Related Adverse Outcomes in U.S. Hospitals and Emergency Departments, 2008. Rockville, MD: Agency for Healthcare Research and Quality; April 2011. Statistical Brief 109.
- US Centers for Medicare and Medicaid Services. Hospital‐Acquired Conditions (Present on Admission Indicator). Available at: http://www.cms.hhs.gov/HospitalAcqCond/. Accessed August 16, 2011. Updated September 20, 2012.
- , , . Fracture risk associated with the use of morphine and opiates. J Intern Med. 2006;260(1):76–87.
- , , , . Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.
- Clinical Classifications Software, Healthcare Cost and Utilization Project (HCUP). Rockville, MD; Agency for Healthcare Research and Quality; December 2009.
- , , , , . Clinical application of opioid equianalgesic data. Clin J Pain. 2003;19(5)286–297.
- , , , , , . Choosing the unit of measurement counts: the use of oral morphine equivalents in studies of opioid consumption is a useful addition to defined daily doses. Palliat Med. 2011;25(7):725–732.
- , , , . Cost and quality implications of opioid‐based postsurgical pain control using administrative claims data from a large health system: opioid‐related adverse events and their impact on clinical and economic outcomes. Pharmacotherapy. 2013;33(4):383–391.
- , , , et al. Opioid‐related adverse drug events in surgical hospitalizations: impact on costs and length of stay. Ann Pharmacother. 2007;41(3):400–406.
- , , , et al. Cost of opioid‐related adverse drug events in surgical patients. J Pain Symptom Manage. 2003;25(3):276–283.
- , . Opioid analgesics and rates of fatal drug poisoning in the United States. Am J Prev Med. 2006;31(6):506–511.
- , , , et al. Geographic variation in the treatment of acute myocardial infarction: the Cooperative Cardiovascular Project. JAMA. 1999;281(7):627–633.
- , , , et al; GUSTO‐1 Investigators. Regional variation across the United States in the management of acute myocardial infarction. Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries. N Engl J Med. 1995;333(9):565–572.
- , , . Predictors of broad‐spectrum antibiotic prescribing for acute respiratory tract infections in adult primary care. JAMA. 2003;289(6):719–725.
- , , . Geographic variation in Medicare drug spending. N Engl J Med. 2010;363(5):405–409.
- , , . Geographic variation in outpatient antibiotic prescribing among older adults. Arch Intern Med. 2012;172(19):1465–1471.
- , , , et al. Clinical policy: critical issues in the prescribing of opioids for adult patients in the emergency department. Ann Emerg Med. 2012;60(4):499–525.
- The Joint Commission. Facts About Pain Management. Available at: http://www.jointcommission.org/pain_management/. Accessed July 23, 2012.
- The Joint Commission. Sentinel Event Alert: Safe Use of Opioids in Hospitals. Published August 8, 2012. Available at: http://www.jointcommission.org/assets/1/18/SEA_49_opioids_8_2_12_final.pdf. Accessed March 4, 2013.
- , , , , . Research gaps on use of opioids for chronic noncancer pain: findings from a review of the evidence for an American Pain Society and American Academy of Pain Medicine clinical practice guideline. J Pain. 2009;10(2):147–159.
- , , , , , . Side effects of opioids during short‐term administration: effect of age, gender, and race. Clin Pharmacol Ther. 2003;74(2):102–112.
- , , , . Postoperative day one: a high‐risk period for respiratory events. Am J Surg. 2005;190(5):752–756.
Recent reports have drawn attention to the high and increasing rates of opioid prescribing and overdose‐related deaths in the United States.[1, 2, 3, 4, 5, 6, 7, 8, 9] These studies have focused on community‐based and emergency department prescribing, leaving prescribing practices in the inpatient setting unexamined. Given that pain is a frequent complaint in hospitalized patients, and that the Joint Commission mandates assessing pain as a vital sign, hospitalization is potentially a time of heightened use of such medications and could significantly contribute to nosocomial complications and subsequent outpatient use.[10] Variation in prescribing practices, unrelated to patient characteristics, could be a marker of inappropriate prescribing practices and poor quality of care.
Using a large, nationally representative cohort of admissions from July 2009 to June 2010, we sought to determine patterns and predictors of opioid utilization in nonsurgical admissions to US medical centers, hospital variation in use, and the association between hospital‐level use and the risk of opioid‐related adverse events. We hypothesized that hospitals with higher rates of opioid use would have an increased risk of an opioid‐related adverse event per patient exposed.
METHODS
Setting and Patients
We conducted a retrospective cohort study using data from 286 US nonfederal acute‐care facilities contributing to the database maintained by Premier (Premier Healthcare Solutions, Inc., Charlotte, NC). This database, created to measure healthcare utilization and quality of care, is drawn from voluntarily participating hospitals and contains data on approximately 1 in every 4 discharges nationwide.[11] Participating hospitals are similar in geographic distribution and metropolitan (urban/rural) status to hospitals nationwide, although large, nonteaching hospitals are slightly overrepresented in Premier. The database contains patient demographics, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes, hospital demographics, and a date‐stamped log of all charges during the course of each hospitalization, including diagnostic tests, therapeutic treatments, and medications with dose and route of administration. The study was approved by the institutional review board at Beth Israel Deaconess Medical Center and granted a waiver of informed consent.
We studied a cohort of all adult nonsurgical admissions to participating hospitals from July 1, 2009, through June 30, 2010. We chose to study nonsurgical admissions, as patients undergoing surgical procedures have a clear indication for, and almost always receive, opioid pain medications. We defined a nonsurgical admission as an admission in which there were no charges for operating‐room procedures (including labor and delivery) and the attending of record was not a surgeon. We excluded admissions with unknown gender, since this is a key demographic variable, and admissions with a length of stay greater than 365 days, as these admissions are not representative of the typical admission to an acute‐care hospital. At the hospital level, we excluded hospitals contributing <100 admissions owing to resultant lack of precision in corresponding hospital prescribing rates, and hospitals that did not prescribe the full range of opioid medications (these hospitals had charges for codeine only), as these facilities seemed likely to have unusual limitations on prescribing or incomplete data capture.
Opioid Exposure
We defined opioid exposure as presence of 1 charge for an opioid medication during the admission. Opioid medications included morphine, hydrocodone, hydromorphone, oxycodone, fentanyl, meperidine, methadone, codeine, tramadol, buprenorphine, levorphanol, oxymorphone, pentazocine, propoxyphene, tapentadol, butorphanol, dezocine, and nalbuphine. We grouped the last 9 into an other category owing to infrequent use and/or differing characteristics from the main opioid drug types, such as synthetic, semisynthetic, and partial agonist qualities.
Severe Opioid‐Related Adverse Events
We defined severe opioid‐related adverse events as either naloxone exposure or an opioid‐related adverse drug event diagnosis code. Naloxone use in an adult patient exposed to opioids is one of the Institute for Healthcare Improvement's Trigger Tools for identifying adverse drug events[12] and previously has been demonstrated to have high positive predictive value for a confirmed adverse drug event.[13] We defined naloxone exposure as presence of 1 charge for naloxone. We excluded charges on hospital day 1 to focus on nosocomial events. We defined opioid‐related adverse drug events using ICD‐9‐CM diagnosis codes for poisoning by opioids (overdose, wrong substance given, or taken in error; ICD‐9‐CM 965.02, 965.09, E850.1, E850.2) and drugs causing adverse effects in therapeutic use (ICD‐9‐CM E935.1, E935.2), as specified in prior analyses by the Agency for Healthcare Research and Quality (AHRQ).[14, 15] To avoid capturing adverse events associated with outpatient use, we required the ICD‐9‐CM code to be qualified as not present on admission using the present on admission indicator required by the Centers for Medicare and Medicaid Services for all discharge diagnosis codes since 2008.[16]
Covariates of Interest
We were interested in the relationship between both patient and hospital characteristics and opioid exposure. Patient characteristics of interest included (1) demographic variables, such as age, sex, race (self‐reported by patients at the time of admission), marital status, and payer; (2) whether or not the patient spent any time in the intensive care unit (ICU); (3) comorbidities, identified via ICD‐9‐CM secondary diagnosis codes and diagnosis‐related groups using Healthcare Cost and Utilization Project Comorbidity Software, version 3.7, based on the work of Elixhauser et al[17, 18]; (4) primary ICD‐9‐CM discharge diagnosis groupings, selected based on hypothesized associations with receipt of opioids, and based on the Clinical Classifications Software (CCS)a diagnosis and procedure categorization scheme maintained by the AHRQ, and defined in the Appendix[19]; (5) and nonoperating‐room‐based procedures potentially necessitating opioids during the admission, selected from the 50 most common ICD‐9‐CM procedure codes in our cohort and grouped as cardiovascular procedures (catheterization and insertion of vascular stents), gastrointestinal procedures (upper and lower endoscopy), and mechanical ventilation, further defined in the Appendix. Hospital characteristics of interest included number of beds, population served (urban vs rural), teaching status, and US census region (Northeast, Midwest, South, West).
Statistical Analysis
We calculated the percent of admissions with exposure to any opioid and the percent exposed to each opioid, along with the total number of different opioid medications used during each admission. We also calculated the percent of admissions with parenteral administration and the percent of admissions with oral administration, among those exposed to the individual categories, and in aggregate. Because medications after discharge were unavailable in Premier's dataset, we report the percent of patients with a charge for opioids on the day of discharge.
We determined the daily dose of an opioid by taking the sum of the doses for that opioid charged on a given day. The average daily dose of an opioid was determined by taking the sum of the daily doses and dividing by the number of days on which 1 dose was charged. To facilitate comparison, all opioids, with the exception of those for which standard equivalences are unavailable (tramadol, other opioid category, oral fentanyl, epidural route for all), were converted to oral morphine equivalents using a standard equivalence conversion table.[20, 21] We excluded from our dosage calculations those charges for which standard morphine equivalence was unavailable, or for which dosage was missing. We also excluded from our dosage calculations any dose that was >3 standard deviations (SD) above the mean dose for that opioid, as such extreme values seemed physiologically implausible and more likely to be a data entry error that could lead to significant overestimation of the mean for that opioid.
All multivariable models used a generalized estimating equation (GEE) via the genmod procedure in SAS, with a Poisson distribution error term and a log link, controlling for repeated patient admissions with an autoregressive correlation structure.
To identify independent predictors of opioid receipt, we used a GEE model of opioid receipt where all patient and hospital characteristics listed in Table 1 were included as independent variables.
| Patient characteristics, N=1,139,419 | N | % |
|---|---|---|
| ||
| Age group, y | ||
| 1824 | 37,464 | 3 |
| 2534 | 66,541 | 6 |
| 3544 | 102,701 | 9 |
| 4554 | 174,830 | 15 |
| 5564 | 192,570 | 17 |
| 6574 | 196,407 | 17 |
| 75+ | 368,906 | 32 |
| Sex | ||
| Male | 527,062 | 46 |
| Female | 612,357 | 54 |
| Race | ||
| White | 711,993 | 62 |
| Black | 176,993 | 16 |
| Hispanic | 54,406 | 5 |
| Other | 196,027 | 17 |
| Marital status | ||
| Married | 427,648 | 38 |
| Single | 586,343 | 51 |
| Unknown/other | 125,428 | 11 |
| Primary insurance | ||
| Private/commercial | 269,725 | 24 |
| Medicare traditional | 502,301 | 44 |
| Medicare managed care | 126,344 | 11 |
| Medicaid | 125,025 | 11 |
| Self‐pay/other | 116,024 | 10 |
| ICU care | ||
| No | 1,023,027 | 90 |
| Yes | 116,392 | 10 |
| Comorbidities | ||
| AIDS | 5724 | 1 |
| Alcohol abuse | 79,633 | 7 |
| Deficiency anemias | 213,437 | 19 |
| RA/collagen vascular disease | 35,210 | 3 |
| Chronic blood‐loss anemia | 10,860 | 1 |
| CHF | 190,085 | 17 |
| Chronic pulmonary disease | 285,954 | 25 |
| Coagulopathy | 48,513 | 4 |
| Depression | 145,553 | 13 |
| DM without chronic complications | 270,087 | 24 |
| DM with chronic complications | 70,732 | 6 |
| Drug abuse | 66,886 | 6 |
| Hypertension | 696,299 | 61 |
| Hypothyroidism | 146,136 | 13 |
| Liver disease | 38,130 | 3 |
| Lymphoma | 14,032 | 1 |
| Fluid and electrolyte disorders | 326,576 | 29 |
| Metastatic cancer | 33,435 | 3 |
| Other neurological disorders | 124,195 | 11 |
| Obesity | 118,915 | 10 |
| Paralysis | 38,584 | 3 |
| PVD | 77,334 | 7 |
| Psychoses | 101,856 | 9 |
| Pulmonary circulation disease | 52,106 | 5 |
| Renal failure | 175,398 | 15 |
| Solid tumor without metastasis | 29,594 | 3 |
| Peptic ulcer disease excluding bleeding | 536 | 0 |
| Valvular disease | 86,616 | 8 |
| Weight loss | 45,132 | 4 |
| Primary discharge diagnoses | ||
| Cancer | 19,168 | 2 |
| Musculoskeletal injuries | 16,798 | 1 |
| Pain‐related diagnosesb | 101,533 | 9 |
| Alcohol‐related disorders | 16,777 | 1 |
| Substance‐related disorders | 13,697 | 1 |
| Psychiatric disorders | 41,153 | 4 |
| Mood disorders | 28,761 | 3 |
| Schizophrenia and other psychotic disorders | 12,392 | 1 |
| Procedures | ||
| Cardiovascular procedures | 59,901 | 5 |
| GI procedures | 31,224 | 3 |
| Mechanical ventilation | 7853 | 1 |
| Hospital characteristics, N=286 | ||
| Number of beds | ||
| <200 | 103 | 36 |
| 201300 | 63 | 22 |
| 301500 | 81 | 28 |
| >500 | 39 | 14 |
| Population served | ||
| Urban | 225 | 79 |
| Rural | 61 | 21 |
| Teaching status | ||
| Nonteaching | 207 | 72 |
| Teaching | 79 | 28 |
| US Census region | ||
| Northeast | 47 | 16 |
| Midwest | 63 | 22 |
| South | 115 | 40 |
| West | 61 | 21 |
To assess hospital variation in opioid prescribing after adjusting for patient characteristics, we used a GEE model of opioid receipt, controlling for all patient characteristics listed in Table 1. We then took the mean of the predicted probabilities of opioid receipt for the patients within each hospital in our cohort to derive the hospital prescribing rate adjusted for patient characteristics. We report the mean, SD, and range of the prescribing rates for the hospitals in our cohort before and after adjustment for patient characteristics.
To assess whether patients admitted to hospitals with higher rates of opioid prescribing have higher relative risk of severe opioid‐related adverse events, we stratified hospitals into opioid‐prescribing rate quartiles and compared the rates of opioid‐related adverse eventsboth overall and among opioid exposedbetween quartiles. To adjust for patient characteristics, we used a GEE model in which severe opioid‐related adverse event (yes/no) was the dependent variable and hospital‐prescribing rate quartile and all patient characteristics in Table 1 were independent variables. We also performed a sensitivity analysis in which we assessed the association between hospital prescribing‐rate quartile and the individual components of our composite outcome. Our results were qualitatively unchanged using this approach, and only the results of our main analysis are presented.
All analyses were carried out using SAS software, version 9.2 (SAS Institute Inc., Cary, NC).
RESULTS
Patient Admission Characteristics
There were 3,190,934 adult admissions to 300 acute‐care hospitals during our study period. After excluding admissions with a length of stay >365 days (n=25), missing patient sex (n=17), and charges for operating‐room procedures or a surgical attending of record (n=2,018,553), 1,172,339 admissions were available for analysis. There were 12 hospitals with incomplete opioid‐prescribing data (n=32,794) and 2 hospitals that contributed <100 admissions each (n=126), leaving 1,139,419 admissions in 286 hospitals in our analytic cohort. The median age of the cohort was 64 years (interquartile range, 4979 years), and 527,062 (46%) were men. Table 1 shows the characteristics of the admissions in the cohort.
Rate, Route, and Dose of Opioid Exposures
Overall, there were 576,373 (51%) admissions with charges for opioid medications. Among those exposed, 244,760 (43%) had charges for multiple opioids during the admission; 172,090 (30%) had charges for 2 different opioids; and 72,670 (13%) had charges for 3 different opioids. Table 2 shows the percent exposed to each opioid, the percent of exposed with parenteral and oral routes of administration, and the mean daily dose received in oral morphine equivalents.
| Exposed | Parenteral Administration | Oral Administration | Dose Received, in Oral Morphine Equivalents | |||||
|---|---|---|---|---|---|---|---|---|
| N | %a | N | %b | N | %b | Mean | SDc | |
| ||||||||
| All opioids | 576,373 | 51 | 378,771 | 66 | 371,796 | 65 | 68 | 185 |
| Morphine | 224,811 | 20 | 209,040 | 93 | 21,645 | 10 | 40 | 121 |
| Hydrocodone | 162,558 | 14 | 0 | 0 | 160,941 | 99 | 14 | 12 |
| Hydromorphone | 146,236 | 13 | 137,936 | 94 | 16,052 | 11 | 113 | 274 |
| Oxycodone | 126,733 | 11 | 0 | 0 | 125,033 | 99 | 26 | 37 |
| Fentanyl | 105,052 | 9 | 103,113 | 98 | 641 | 1 | 64 | 75 |
| Tramadol | 35,570 | 3 | 0 | 0 | 35,570 | 100 | ||
| Meperidine | 24,850 | 2 | 24,398 | 98 | 515 | 2 | 36 | 34 |
| Methadone | 15,302 | 1 | 370 | 2 | 14,781 | 97 | 337 | 384 |
| Codeine | 22,818 | 2 | 178 | 1 | 22,183 | 97 | 9 | 15 |
| Other | 45,469 | 4 | 5821 | 13 | 39,618 | 87 | ||
Among the medications/routes for which conversion to morphine equivalents was possible, dosage was missing in 39,728 out of 2,294,673 opioid charges (2%). The average daily dose received in oral morphine equivalents was 68 mg. A total dose of 50 mg per day was received in 39% of exposed, and a total dose of 100 mg per day was received in 23% of exposed. Among those exposed, 52% (26% of overall admissions) had charges for opioids on the day of discharge.
Rates of Opioid Use by Patient and Hospital Characteristics
Table 3 reports the association between admission characteristics and opioid use. Use was highest in patients between the ages of 25 and 54 years. Although use declined with age, 44% of admissions age 65 years had charges for opioid medication. After adjustment for patient demographics, comorbidities, and hospital characteristics, opioid use was more common in females than males, those age 2554 years compared with those older and younger, those of Caucasian race compared with non‐Caucasian race, and those with Medicare or Medicaid primary insurance. Among the primary discharge diagnoses, patients with musculoskeletal injuries, various specific and nonspecific pain‐related diagnoses, and cancer were significantly more likely to receive opioids than patients without these diagnoses, whereas patients with alcohol‐related disorders and psychiatric disorders were significantly less likely to receive opioids than patients without these diagnoses. Patients admitted to hospitals in the Midwest, South, and West were significantly more likely to receive opioid medications than patients in the Northeast.
| Exposed, N=576,373 | Unexposed, N=563,046 | % Exposed | Adjusted RRa | 95% CI | |
|---|---|---|---|---|---|
| |||||
| Patient characteristics | |||||
| Age group, y | |||||
| 1824 | 17,360 | 20,104 | 46 | (ref) | |
| 2534 | 37,793 | 28,748 | 57 | 1.17 | 1.16‐1.19 |
| 3544 | 60,712 | 41,989 | 59 | 1.16 | 1.15‐1.17 |
| 4554 | 103,798 | 71,032 | 59 | 1.11 | 1.09‐1.12 |
| 5564 | 108,256 | 84,314 | 56 | 1.00 | 0.98‐1.01 |
| 6574 | 98,110 | 98,297 | 50 | 0.84 | 0.83‐0.85 |
| 75+ | 150,344 | 218,562 | 41 | 0.71 | 0.70‐0.72 |
| Sex | |||||
| Male | 255,315 | 271,747 | 48 | (ref) | |
| Female | 321,058 | 291,299 | 52 | 1.11 | 1.10‐1.11 |
| Race | |||||
| White | 365,107 | 346,886 | 51 | (ref) | |
| Black | 92,013 | 84,980 | 52 | 0.93 | 0.92‐0.93 |
| Hispanic | 27,592 | 26,814 | 51 | 0.94 | 0.93‐0.94 |
| Other | 91,661 | 104,366 | 47 | 0.93 | 0.92‐0.93 |
| Marital status | |||||
| Married | 222,912 | 204,736 | 52 | (ref) | |
| Single | 297,742 | 288,601 | 51 | 1.00 | 0.99‐1.01 |
| Unknown/other | 55,719 | 69,709 | 44 | 0.94 | 0.93‐0.95 |
| Primary insurance | |||||
| Private/commercial | 143,954 | 125,771 | 53 | (ref) | |
| Medicare traditional | 236,114 | 266,187 | 47 | 1.10 | 1.09‐1.10 |
| Medicare managed care | 59,104 | 67,240 | 47 | 1.11 | 1.11‐1.12 |
| Medicaid | 73,583 | 51,442 | 59 | 1.13 | 1.12‐1.13 |
| Self‐pay/other | 63,618 | 52,406 | 55 | 1.03 | 1.02‐1.04 |
| ICU care | |||||
| No | 510,654 | 512,373 | 50 | (ref) | |
| Yes | 65,719 | 50,673 | 56 | 1.02 | 1.01‐1.03 |
| Comorbiditiesb | |||||
| AIDS | 3655 | 2069 | 64 | 1.09 | 1.07‐1.12 |
| Alcohol abuse | 35,112 | 44,521 | 44 | 0.92 | 0.91‐0.93 |
| Deficiency anemias | 115,842 | 97,595 | 54 | 1.08 | 1.08‐1.09 |
| RA/collagen vascular disease | 22,519 | 12,691 | 64 | 1.22 | 1.21‐1.23 |
| Chronic blood‐loss anemia | 6444 | 4416 | 59 | 1.04 | 1.02‐1.05 |
| CHF | 88,895 | 101,190 | 47 | 0.99 | 0.98‐0.99 |
| Chronic pulmonary disease | 153,667 | 132,287 | 54 | 1.08 | 1.08‐1.08 |
| Coagulopathy | 25,802 | 22,711 | 53 | 1.03 | 1.02‐1.04 |
| Depression | 83,051 | 62,502 | 57 | 1.08 | 1.08‐1.09 |
| DM without chronic complications | 136,184 | 133,903 | 50 | 0.99 | 0.99‐0.99 |
| DM with chronic complications | 38,696 | 32,036 | 55 | 1.04 | 1.03‐1.05 |
| Drug abuse | 37,202 | 29,684 | 56 | 1.14 | 1.13‐1.15 |
| Hypertension | 344,718 | 351,581 | 50 | 0.98 | 0.97‐0.98 |
| Hypothyroidism | 70,786 | 75,350 | 48 | 0.99 | 0.99‐0.99 |
| Liver disease | 24,067 | 14,063 | 63 | 1.15 | 1.14‐1.16 |
| Lymphoma | 7727 | 6305 | 55 | 1.16 | 1.14‐1.17 |
| Fluid and electrolyte disorders | 168,814 | 157,762 | 52 | 1.04 | 1.03‐1.04 |
| Metastatic cancer | 23,920 | 9515 | 72 | 1.40 | 1.39‐1.42 |
| Other neurological disorders | 51,091 | 73,104 | 41 | 0.87 | 0.86‐0.87 |
| Obesity | 69,584 | 49,331 | 59 | 1.05 | 1.04‐1.05 |
| Paralysis | 17,497 | 21,087 | 45 | 0.97 | 0.96‐0.98 |
| PVD | 42,176 | 35,158 | 55 | 1.11 | 1.11‐1.12 |
| Psychoses | 38,638 | 63,218 | 38 | 0.91 | 0.90‐0.92 |
| Pulmonary circulation disease | 26,656 | 25,450 | 51 | 1.05 | 1.04‐1.06 |
| Renal failure | 86,565 | 88,833 | 49 | 1.01 | 1.01‐1.02 |
| Solid tumor without metastasis | 16,258 | 13,336 | 55 | 1.14 | 1.13‐1.15 |
| Peptic ulcer disease excluding bleeding | 376 | 160 | 70 | 1.12 | 1.07‐1.18 |
| Valvular disease | 38,396 | 48,220 | 44 | 0.93 | 0.92‐0.94 |
| Weight loss | 25,724 | 19,408 | 57 | 1.09 | 1.08‐1.10 |
| Primary discharge diagnosesb | |||||
| Cancer | 13,986 | 5182 | 73 | 1.20 | 1.19‐1.21 |
| Musculoskeletal injuries | 14,638 | 2160 | 87 | 2.02 | 2.002.04 |
| Pain‐related diagnosesc | 64,656 | 36,877 | 64 | 1.20 | 1.20‐1.21 |
| Alcohol‐related disorders | 3425 | 13,352 | 20 | 0.46 | 0.44‐0.47 |
| Substance‐related disorders | 8680 | 5017 | 63 | 1.03 | 1.01‐1.04 |
| Psychiatric disorders | 7253 | 33,900 | 18 | 0.37 | 0.36‐0.38 |
| Mood disorders | 5943 | 22,818 | 21 | ||
| Schizophrenia and other psychotic disorders | 1310 | 11,082 | 11 | ||
| Proceduresb | |||||
| Cardiovascular procedures | 50,997 | 8904 | 85 | 1.80 | 1.79‐1.81 |
| GI procedures | 27,206 | 4018 | 87 | 1.70 | 1.69‐1.71 |
| Mechanical ventilation | 5341 | 2512 | 68 | 1.37 | 1.34‐1.39 |
| Hospital characteristics | |||||
| Number of beds | |||||
| <200 | 100,900 | 88,439 | 53 | (ref) | |
| 201300 | 104,213 | 99,995 | 51 | 0.95 | 0.95‐0.96 |
| 301500 | 215,340 | 209,104 | 51 | 0.94 | 0.94‐0.95 |
| >500 | 155,920 | 165,508 | 49 | 0.96 | 0.95‐0.96 |
| Population served | |||||
| Urban | 511,727 | 506,803 | 50 | (ref) | |
| Rural | 64,646 | 56,243 | 53 | 0.98 | 0.97‐0.99 |
| Teaching status | |||||
| Nonteaching | 366,623 | 343,581 | 52 | (ref) | |
| Teaching | 209,750 | 219,465 | 49 | 1.00 | 0.99‐1.01 |
| US Census region | |||||
| Northeast | 99,377 | 149,446 | 40 | (ref) | |
| Midwest | 123,194 | 120,322 | 51 | 1.26 | (1.25‐1.27) |
| South | 251,624 | 213,029 | 54 | 1.33 | (1.33‐1.34) |
| West | 102,178 | 80,249 | 56 | 1.37 | (1.36‐1.38) |
Variation in Opioid Prescribing
Figure 1 shows the histograms of hospital opioid prescribing rate for the 286 hospitals in our cohort (A) before and (B) after adjustment for patient characteristics. The observed rates ranged from 5% in the lowest‐prescribing hospital to 72% in the highest‐prescribing hospital, with a mean (SD) of 51% (10%). After adjusting for patient characteristics, the adjusted opioid‐prescribing rates ranged from 33% to 64%, with a mean (SD) of 50% (4%).
Severe Opioid‐Related Adverse Events
Among admissions with opioid exposure (n=576,373), naloxone use occurred in 2345 (0.41%) and opioid‐related adverse drug events in 1174 (0.20%), for a total of 3441 (0.60%) severe opioid‐related adverse events (some patients experienced both). Table 4 reports the opioid exposure and severe opioid‐related adverse‐event rates within hospital opioid‐prescribing rate quartiles, along with the adjusted association between the hospital opioid‐prescribing rate quartile and severe opioid‐related adverse events. After adjusting for patient characteristics, the relative risk of a severe opioid‐related adverse event was significantly greater in hospitals with higher opioid‐prescribing rates, both overall and among opioid exposed.
| Quartile | No. of Patients | Opioid Exposed, n (%) | Opioid‐Related Adverse Events, n (%) | Adjusted RR in All Patients, RR (95% CI), N=1,139,419a | Adjusted RR in Opioid Exposed, RR (95% CI), N=576,373a |
|---|---|---|---|---|---|
| |||||
| 1 | 349,747 | 132,824 (38) | 719 (0.21) | (ref) | (ref) |
| 2 | 266,652 | 134,590 (50) | 729 (0.27) | 1.31 (1.17‐1.45) | 1.07 (0.96‐1.18) |
| 3 | 251,042 | 139,770 (56) | 922 (0.37) | 1.72 (1.56‐1.90) | 1.31 (1.19‐1.44) |
| 4 | 271,978 | 169,189 (62) | 1071 (0.39) | 1.73 (1.57‐1.90) | 1.23 (1.12‐1.35) |
DISCUSSION
In this analysis of a large cohort of hospitalized nonsurgical patients, we found that more than half of all patients received opioids, with 43% of those exposed receiving multiple opioids during their admission and 52% receiving opioids on the day of discharge. Considerable hospital variation in opioid use was evident, and this was not fully explained by patient characteristics. Severe opioid‐related adverse events occurred more frequently at hospitals with higher opioid‐prescribing rates, and the relative risk of a severe adverse event per patient prescribed opioids was also higher in these hospitals. To our knowledge, this is the first study to describe the scope of opioid utilization and the relationship between utilization and severe opioid‐related adverse events in a sample of nonsurgical patients in US acute‐care facilities.
Our use of naloxone charges and opioid‐specific ICD‐9‐CM coding to define an opioid‐related adverse event was intended to capture only the most severe opioid‐related adverse events. We chose to focus on these events in our analysis to maximize the specificity of our outcome definition and thereby minimize confounding in our observed associations. The rate of less‐severe opioid‐related adverse events, such as nausea, constipation, and pruritis, is likely much higher and not captured in our outcome definition. Prior analyses have found variable rates of opioid‐related adverse events of approximately 1.8% to 13.6% of exposed patients.[22, 23, 24] However, these analyses focused on surgical patients and included less‐severe events. To our knowledge, ours is the first analysis of severe opioid‐related adverse events in nonsurgical patients.
Our finding that severe opioid‐related adverse events increase as opioid prescribing increases is consistent with that which has been demonstrated in the community setting, where rates of opioid‐related adverse events and mortality are higher in communities with higher levels of opioid prescribing.[2, 8, 25] This finding is expected, as greater use of a class of medications with known side effects would be expected to result in a higher overall rate of adverse events. More concerning, however, is the fact that this relationship persists when focusing exclusively on opioid‐exposed patients. Among similar patients receiving opioids at different hospitals, those hospitalized in facilities with higher opioid‐prescribing rates have higher rates of severe opioid‐related adverse events. This suggests that hospitals that use opioids more frequently do not do so more safely. Rather, the increased overall prescribing rates are associated with heightened risk for a serious adverse event per patient exposed and may reflect unsafe prescribing practices.
Furthermore, our results demonstrate both regional and hospital variation in use of opioids not fully explained by patient characteristics, similar to that which has been demonstrated for other drugs and heathcare services.[26, 27, 28, 29, 30] The implications of these findings are limited by our lack of information on pain severity or prior outpatient treatment, and our resultant inability to evaluate the appropriateness of opioid use in this analysis. Additionally, although we controlled for a large number of patient and hospital characteristics, there could be other significant predictors of use not accounted for in our analysis. However, it seems unlikely that differential pain severity or patient characteristics between patients in different regions of the country could fully explain a 37% relative difference in prescribing between the lowest‐ and highest‐prescribing regions, after accounting for the 44 patient‐level variables in our models. Whereas variation in use unrelated to patient factors could represent inappropriate prescribing practices, it could also be a marker of uncertainty regarding what constitutes appropriate prescribing and high‐quality care in this realm. Although guidelines advocate for standard pain assessments and a step‐up approach to treatment,[31, 32, 33] the lack of objective measures of pain severity and lack of evidence‐based recommendations on the use of opioids for noncancer pain[34] will almost certainly lead to persistent variation in opioid prescribing despite guideline‐driven care.
Nonetheless, our findings suggest that opportunities exist to make opioid prescribing safer in hospitalized patients. Studies aimed at elucidating the source of regional and hospital variation are necessary. Additionally, efforts should focus on identifying patient and prescribing characteristics associated with heightened risk of opioid‐related adverse events. Prior studies have demonstrated that the risks of opioid medications increase with increasing age of the patient.[35, 36] Although opioid use in our cohort declined with age, 44% of admissions age 65 years had charges for opioid medications. Studies in outpatients have also demonstrated that the risks of opioid overdose and overdose‐related death increase with dose.[5, 7] One study demonstrated a 3.7‐fold increased risk of overdose at doses of 5099 mg/day in oral morphine equivalents, and an 8.9‐fold increased risk at doses of 100 mg/day, compared with doses of 20 mg/day.[7] The prevalence of high dose exposure observed in our cohort, coupled with the older age of hospitalized patients, suggests potential targets for promoting safer use in hospitalized patients through interventions such as computerized decision support and enhanced monitoring in those at highest risk.
Because medications after discharge were unavailable in our dataset, the percentage of patients given a prescription for opioid medication on discharge is unknown. However, given that opioids are often tapered rather than abruptly discontinued, our finding that 26% of hospitalized nonsurgical patients received opioids on the day of discharge suggests that a substantial proportion of patients may be discharged with a prescription for opioid medication. Given the possibility of coexistent outpatient opioid prescriptions, these findings draw attention to the importance of assuring development and streamlined accessibility of data from state prescription drug monitoring programs and suggest that increased attention should be paid to the role that inpatient opioid prescribing plays in the increased rates of chronic opioid use and overdose‐related deaths in the United States.
There are additional limitations to our analysis. First, although the database used for this analysis captures a large proportion of admissions to US acute‐care facilities and is similar in composition, it is possible that participating medical centers differ from nonparticipating medical centers in ways that could be associated with opioid prescribing. Additionally, although Premier performs extensive validation and correction processes to assure the quality of their data, there is still likely to be a small amount of random error in the database, which could particularly impact dosage calculations. The lack of preadmission medications in our database precluded identification of the proportion of patients newly started on opioid medications. Lastly, it is possible that the hospital prescribing‐rate quartile is associated with patient characteristics unaccounted for in our analysis, and, therefore, the possibility of residual confounding still exists.
In conclusion, the majority of hospitalized nonsurgical patients are exposed to opioid medications during the course of their hospitalizations, often at high doses. More than half of those exposed are still receiving these medications on the day of discharge. We found hospital and regional variation in opioid use that was not fully explained by patient characteristics, and higher levels of hospital use were associated with higher risk of severe opioid‐related adverse events in opioid‐exposed patients. Further research is necessary to investigate the appropriateness of opioid use in this patient population, the sources of variation in use, and the predictors of opioid‐related adverse events in hospitalized patients to allow development of interventions to make hospital use safer.
Disclosures
Disclosures: Dr. Herzig had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Herzig was funded by grant no. K23AG042459 from the National Institute on Aging. Dr. Marcantonio was funded by grant nos. P01AG031720, R01AG030618, R03AG028189, and K24AG035075 from the National Institute on Aging. The funding organization had no involvement in any aspect of the study, including design, conduct, and reporting of the study. None of the authors have any conflicts of interest to disclose.
Recent reports have drawn attention to the high and increasing rates of opioid prescribing and overdose‐related deaths in the United States.[1, 2, 3, 4, 5, 6, 7, 8, 9] These studies have focused on community‐based and emergency department prescribing, leaving prescribing practices in the inpatient setting unexamined. Given that pain is a frequent complaint in hospitalized patients, and that the Joint Commission mandates assessing pain as a vital sign, hospitalization is potentially a time of heightened use of such medications and could significantly contribute to nosocomial complications and subsequent outpatient use.[10] Variation in prescribing practices, unrelated to patient characteristics, could be a marker of inappropriate prescribing practices and poor quality of care.
Using a large, nationally representative cohort of admissions from July 2009 to June 2010, we sought to determine patterns and predictors of opioid utilization in nonsurgical admissions to US medical centers, hospital variation in use, and the association between hospital‐level use and the risk of opioid‐related adverse events. We hypothesized that hospitals with higher rates of opioid use would have an increased risk of an opioid‐related adverse event per patient exposed.
METHODS
Setting and Patients
We conducted a retrospective cohort study using data from 286 US nonfederal acute‐care facilities contributing to the database maintained by Premier (Premier Healthcare Solutions, Inc., Charlotte, NC). This database, created to measure healthcare utilization and quality of care, is drawn from voluntarily participating hospitals and contains data on approximately 1 in every 4 discharges nationwide.[11] Participating hospitals are similar in geographic distribution and metropolitan (urban/rural) status to hospitals nationwide, although large, nonteaching hospitals are slightly overrepresented in Premier. The database contains patient demographics, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) codes, hospital demographics, and a date‐stamped log of all charges during the course of each hospitalization, including diagnostic tests, therapeutic treatments, and medications with dose and route of administration. The study was approved by the institutional review board at Beth Israel Deaconess Medical Center and granted a waiver of informed consent.
We studied a cohort of all adult nonsurgical admissions to participating hospitals from July 1, 2009, through June 30, 2010. We chose to study nonsurgical admissions, as patients undergoing surgical procedures have a clear indication for, and almost always receive, opioid pain medications. We defined a nonsurgical admission as an admission in which there were no charges for operating‐room procedures (including labor and delivery) and the attending of record was not a surgeon. We excluded admissions with unknown gender, since this is a key demographic variable, and admissions with a length of stay greater than 365 days, as these admissions are not representative of the typical admission to an acute‐care hospital. At the hospital level, we excluded hospitals contributing <100 admissions owing to resultant lack of precision in corresponding hospital prescribing rates, and hospitals that did not prescribe the full range of opioid medications (these hospitals had charges for codeine only), as these facilities seemed likely to have unusual limitations on prescribing or incomplete data capture.
Opioid Exposure
We defined opioid exposure as presence of 1 charge for an opioid medication during the admission. Opioid medications included morphine, hydrocodone, hydromorphone, oxycodone, fentanyl, meperidine, methadone, codeine, tramadol, buprenorphine, levorphanol, oxymorphone, pentazocine, propoxyphene, tapentadol, butorphanol, dezocine, and nalbuphine. We grouped the last 9 into an other category owing to infrequent use and/or differing characteristics from the main opioid drug types, such as synthetic, semisynthetic, and partial agonist qualities.
Severe Opioid‐Related Adverse Events
We defined severe opioid‐related adverse events as either naloxone exposure or an opioid‐related adverse drug event diagnosis code. Naloxone use in an adult patient exposed to opioids is one of the Institute for Healthcare Improvement's Trigger Tools for identifying adverse drug events[12] and previously has been demonstrated to have high positive predictive value for a confirmed adverse drug event.[13] We defined naloxone exposure as presence of 1 charge for naloxone. We excluded charges on hospital day 1 to focus on nosocomial events. We defined opioid‐related adverse drug events using ICD‐9‐CM diagnosis codes for poisoning by opioids (overdose, wrong substance given, or taken in error; ICD‐9‐CM 965.02, 965.09, E850.1, E850.2) and drugs causing adverse effects in therapeutic use (ICD‐9‐CM E935.1, E935.2), as specified in prior analyses by the Agency for Healthcare Research and Quality (AHRQ).[14, 15] To avoid capturing adverse events associated with outpatient use, we required the ICD‐9‐CM code to be qualified as not present on admission using the present on admission indicator required by the Centers for Medicare and Medicaid Services for all discharge diagnosis codes since 2008.[16]
Covariates of Interest
We were interested in the relationship between both patient and hospital characteristics and opioid exposure. Patient characteristics of interest included (1) demographic variables, such as age, sex, race (self‐reported by patients at the time of admission), marital status, and payer; (2) whether or not the patient spent any time in the intensive care unit (ICU); (3) comorbidities, identified via ICD‐9‐CM secondary diagnosis codes and diagnosis‐related groups using Healthcare Cost and Utilization Project Comorbidity Software, version 3.7, based on the work of Elixhauser et al[17, 18]; (4) primary ICD‐9‐CM discharge diagnosis groupings, selected based on hypothesized associations with receipt of opioids, and based on the Clinical Classifications Software (CCS)a diagnosis and procedure categorization scheme maintained by the AHRQ, and defined in the Appendix[19]; (5) and nonoperating‐room‐based procedures potentially necessitating opioids during the admission, selected from the 50 most common ICD‐9‐CM procedure codes in our cohort and grouped as cardiovascular procedures (catheterization and insertion of vascular stents), gastrointestinal procedures (upper and lower endoscopy), and mechanical ventilation, further defined in the Appendix. Hospital characteristics of interest included number of beds, population served (urban vs rural), teaching status, and US census region (Northeast, Midwest, South, West).
Statistical Analysis
We calculated the percent of admissions with exposure to any opioid and the percent exposed to each opioid, along with the total number of different opioid medications used during each admission. We also calculated the percent of admissions with parenteral administration and the percent of admissions with oral administration, among those exposed to the individual categories, and in aggregate. Because medications after discharge were unavailable in Premier's dataset, we report the percent of patients with a charge for opioids on the day of discharge.
We determined the daily dose of an opioid by taking the sum of the doses for that opioid charged on a given day. The average daily dose of an opioid was determined by taking the sum of the daily doses and dividing by the number of days on which 1 dose was charged. To facilitate comparison, all opioids, with the exception of those for which standard equivalences are unavailable (tramadol, other opioid category, oral fentanyl, epidural route for all), were converted to oral morphine equivalents using a standard equivalence conversion table.[20, 21] We excluded from our dosage calculations those charges for which standard morphine equivalence was unavailable, or for which dosage was missing. We also excluded from our dosage calculations any dose that was >3 standard deviations (SD) above the mean dose for that opioid, as such extreme values seemed physiologically implausible and more likely to be a data entry error that could lead to significant overestimation of the mean for that opioid.
All multivariable models used a generalized estimating equation (GEE) via the genmod procedure in SAS, with a Poisson distribution error term and a log link, controlling for repeated patient admissions with an autoregressive correlation structure.
To identify independent predictors of opioid receipt, we used a GEE model of opioid receipt where all patient and hospital characteristics listed in Table 1 were included as independent variables.
| Patient characteristics, N=1,139,419 | N | % |
|---|---|---|
| ||
| Age group, y | ||
| 1824 | 37,464 | 3 |
| 2534 | 66,541 | 6 |
| 3544 | 102,701 | 9 |
| 4554 | 174,830 | 15 |
| 5564 | 192,570 | 17 |
| 6574 | 196,407 | 17 |
| 75+ | 368,906 | 32 |
| Sex | ||
| Male | 527,062 | 46 |
| Female | 612,357 | 54 |
| Race | ||
| White | 711,993 | 62 |
| Black | 176,993 | 16 |
| Hispanic | 54,406 | 5 |
| Other | 196,027 | 17 |
| Marital status | ||
| Married | 427,648 | 38 |
| Single | 586,343 | 51 |
| Unknown/other | 125,428 | 11 |
| Primary insurance | ||
| Private/commercial | 269,725 | 24 |
| Medicare traditional | 502,301 | 44 |
| Medicare managed care | 126,344 | 11 |
| Medicaid | 125,025 | 11 |
| Self‐pay/other | 116,024 | 10 |
| ICU care | ||
| No | 1,023,027 | 90 |
| Yes | 116,392 | 10 |
| Comorbidities | ||
| AIDS | 5724 | 1 |
| Alcohol abuse | 79,633 | 7 |
| Deficiency anemias | 213,437 | 19 |
| RA/collagen vascular disease | 35,210 | 3 |
| Chronic blood‐loss anemia | 10,860 | 1 |
| CHF | 190,085 | 17 |
| Chronic pulmonary disease | 285,954 | 25 |
| Coagulopathy | 48,513 | 4 |
| Depression | 145,553 | 13 |
| DM without chronic complications | 270,087 | 24 |
| DM with chronic complications | 70,732 | 6 |
| Drug abuse | 66,886 | 6 |
| Hypertension | 696,299 | 61 |
| Hypothyroidism | 146,136 | 13 |
| Liver disease | 38,130 | 3 |
| Lymphoma | 14,032 | 1 |
| Fluid and electrolyte disorders | 326,576 | 29 |
| Metastatic cancer | 33,435 | 3 |
| Other neurological disorders | 124,195 | 11 |
| Obesity | 118,915 | 10 |
| Paralysis | 38,584 | 3 |
| PVD | 77,334 | 7 |
| Psychoses | 101,856 | 9 |
| Pulmonary circulation disease | 52,106 | 5 |
| Renal failure | 175,398 | 15 |
| Solid tumor without metastasis | 29,594 | 3 |
| Peptic ulcer disease excluding bleeding | 536 | 0 |
| Valvular disease | 86,616 | 8 |
| Weight loss | 45,132 | 4 |
| Primary discharge diagnoses | ||
| Cancer | 19,168 | 2 |
| Musculoskeletal injuries | 16,798 | 1 |
| Pain‐related diagnosesb | 101,533 | 9 |
| Alcohol‐related disorders | 16,777 | 1 |
| Substance‐related disorders | 13,697 | 1 |
| Psychiatric disorders | 41,153 | 4 |
| Mood disorders | 28,761 | 3 |
| Schizophrenia and other psychotic disorders | 12,392 | 1 |
| Procedures | ||
| Cardiovascular procedures | 59,901 | 5 |
| GI procedures | 31,224 | 3 |
| Mechanical ventilation | 7853 | 1 |
| Hospital characteristics, N=286 | ||
| Number of beds | ||
| <200 | 103 | 36 |
| 201300 | 63 | 22 |
| 301500 | 81 | 28 |
| >500 | 39 | 14 |
| Population served | ||
| Urban | 225 | 79 |
| Rural | 61 | 21 |
| Teaching status | ||
| Nonteaching | 207 | 72 |
| Teaching | 79 | 28 |
| US Census region | ||
| Northeast | 47 | 16 |
| Midwest | 63 | 22 |
| South | 115 | 40 |
| West | 61 | 21 |
To assess hospital variation in opioid prescribing after adjusting for patient characteristics, we used a GEE model of opioid receipt, controlling for all patient characteristics listed in Table 1. We then took the mean of the predicted probabilities of opioid receipt for the patients within each hospital in our cohort to derive the hospital prescribing rate adjusted for patient characteristics. We report the mean, SD, and range of the prescribing rates for the hospitals in our cohort before and after adjustment for patient characteristics.
To assess whether patients admitted to hospitals with higher rates of opioid prescribing have higher relative risk of severe opioid‐related adverse events, we stratified hospitals into opioid‐prescribing rate quartiles and compared the rates of opioid‐related adverse eventsboth overall and among opioid exposedbetween quartiles. To adjust for patient characteristics, we used a GEE model in which severe opioid‐related adverse event (yes/no) was the dependent variable and hospital‐prescribing rate quartile and all patient characteristics in Table 1 were independent variables. We also performed a sensitivity analysis in which we assessed the association between hospital prescribing‐rate quartile and the individual components of our composite outcome. Our results were qualitatively unchanged using this approach, and only the results of our main analysis are presented.
All analyses were carried out using SAS software, version 9.2 (SAS Institute Inc., Cary, NC).
RESULTS
Patient Admission Characteristics
There were 3,190,934 adult admissions to 300 acute‐care hospitals during our study period. After excluding admissions with a length of stay >365 days (n=25), missing patient sex (n=17), and charges for operating‐room procedures or a surgical attending of record (n=2,018,553), 1,172,339 admissions were available for analysis. There were 12 hospitals with incomplete opioid‐prescribing data (n=32,794) and 2 hospitals that contributed <100 admissions each (n=126), leaving 1,139,419 admissions in 286 hospitals in our analytic cohort. The median age of the cohort was 64 years (interquartile range, 4979 years), and 527,062 (46%) were men. Table 1 shows the characteristics of the admissions in the cohort.
Rate, Route, and Dose of Opioid Exposures
Overall, there were 576,373 (51%) admissions with charges for opioid medications. Among those exposed, 244,760 (43%) had charges for multiple opioids during the admission; 172,090 (30%) had charges for 2 different opioids; and 72,670 (13%) had charges for 3 different opioids. Table 2 shows the percent exposed to each opioid, the percent of exposed with parenteral and oral routes of administration, and the mean daily dose received in oral morphine equivalents.
| Exposed | Parenteral Administration | Oral Administration | Dose Received, in Oral Morphine Equivalents | |||||
|---|---|---|---|---|---|---|---|---|
| N | %a | N | %b | N | %b | Mean | SDc | |
| ||||||||
| All opioids | 576,373 | 51 | 378,771 | 66 | 371,796 | 65 | 68 | 185 |
| Morphine | 224,811 | 20 | 209,040 | 93 | 21,645 | 10 | 40 | 121 |
| Hydrocodone | 162,558 | 14 | 0 | 0 | 160,941 | 99 | 14 | 12 |
| Hydromorphone | 146,236 | 13 | 137,936 | 94 | 16,052 | 11 | 113 | 274 |
| Oxycodone | 126,733 | 11 | 0 | 0 | 125,033 | 99 | 26 | 37 |
| Fentanyl | 105,052 | 9 | 103,113 | 98 | 641 | 1 | 64 | 75 |
| Tramadol | 35,570 | 3 | 0 | 0 | 35,570 | 100 | ||
| Meperidine | 24,850 | 2 | 24,398 | 98 | 515 | 2 | 36 | 34 |
| Methadone | 15,302 | 1 | 370 | 2 | 14,781 | 97 | 337 | 384 |
| Codeine | 22,818 | 2 | 178 | 1 | 22,183 | 97 | 9 | 15 |
| Other | 45,469 | 4 | 5821 | 13 | 39,618 | 87 | ||
Among the medications/routes for which conversion to morphine equivalents was possible, dosage was missing in 39,728 out of 2,294,673 opioid charges (2%). The average daily dose received in oral morphine equivalents was 68 mg. A total dose of 50 mg per day was received in 39% of exposed, and a total dose of 100 mg per day was received in 23% of exposed. Among those exposed, 52% (26% of overall admissions) had charges for opioids on the day of discharge.
Rates of Opioid Use by Patient and Hospital Characteristics
Table 3 reports the association between admission characteristics and opioid use. Use was highest in patients between the ages of 25 and 54 years. Although use declined with age, 44% of admissions age 65 years had charges for opioid medication. After adjustment for patient demographics, comorbidities, and hospital characteristics, opioid use was more common in females than males, those age 2554 years compared with those older and younger, those of Caucasian race compared with non‐Caucasian race, and those with Medicare or Medicaid primary insurance. Among the primary discharge diagnoses, patients with musculoskeletal injuries, various specific and nonspecific pain‐related diagnoses, and cancer were significantly more likely to receive opioids than patients without these diagnoses, whereas patients with alcohol‐related disorders and psychiatric disorders were significantly less likely to receive opioids than patients without these diagnoses. Patients admitted to hospitals in the Midwest, South, and West were significantly more likely to receive opioid medications than patients in the Northeast.
| Exposed, N=576,373 | Unexposed, N=563,046 | % Exposed | Adjusted RRa | 95% CI | |
|---|---|---|---|---|---|
| |||||
| Patient characteristics | |||||
| Age group, y | |||||
| 1824 | 17,360 | 20,104 | 46 | (ref) | |
| 2534 | 37,793 | 28,748 | 57 | 1.17 | 1.16‐1.19 |
| 3544 | 60,712 | 41,989 | 59 | 1.16 | 1.15‐1.17 |
| 4554 | 103,798 | 71,032 | 59 | 1.11 | 1.09‐1.12 |
| 5564 | 108,256 | 84,314 | 56 | 1.00 | 0.98‐1.01 |
| 6574 | 98,110 | 98,297 | 50 | 0.84 | 0.83‐0.85 |
| 75+ | 150,344 | 218,562 | 41 | 0.71 | 0.70‐0.72 |
| Sex | |||||
| Male | 255,315 | 271,747 | 48 | (ref) | |
| Female | 321,058 | 291,299 | 52 | 1.11 | 1.10‐1.11 |
| Race | |||||
| White | 365,107 | 346,886 | 51 | (ref) | |
| Black | 92,013 | 84,980 | 52 | 0.93 | 0.92‐0.93 |
| Hispanic | 27,592 | 26,814 | 51 | 0.94 | 0.93‐0.94 |
| Other | 91,661 | 104,366 | 47 | 0.93 | 0.92‐0.93 |
| Marital status | |||||
| Married | 222,912 | 204,736 | 52 | (ref) | |
| Single | 297,742 | 288,601 | 51 | 1.00 | 0.99‐1.01 |
| Unknown/other | 55,719 | 69,709 | 44 | 0.94 | 0.93‐0.95 |
| Primary insurance | |||||
| Private/commercial | 143,954 | 125,771 | 53 | (ref) | |
| Medicare traditional | 236,114 | 266,187 | 47 | 1.10 | 1.09‐1.10 |
| Medicare managed care | 59,104 | 67,240 | 47 | 1.11 | 1.11‐1.12 |
| Medicaid | 73,583 | 51,442 | 59 | 1.13 | 1.12‐1.13 |
| Self‐pay/other | 63,618 | 52,406 | 55 | 1.03 | 1.02‐1.04 |
| ICU care | |||||
| No | 510,654 | 512,373 | 50 | (ref) | |
| Yes | 65,719 | 50,673 | 56 | 1.02 | 1.01‐1.03 |
| Comorbiditiesb | |||||
| AIDS | 3655 | 2069 | 64 | 1.09 | 1.07‐1.12 |
| Alcohol abuse | 35,112 | 44,521 | 44 | 0.92 | 0.91‐0.93 |
| Deficiency anemias | 115,842 | 97,595 | 54 | 1.08 | 1.08‐1.09 |
| RA/collagen vascular disease | 22,519 | 12,691 | 64 | 1.22 | 1.21‐1.23 |
| Chronic blood‐loss anemia | 6444 | 4416 | 59 | 1.04 | 1.02‐1.05 |
| CHF | 88,895 | 101,190 | 47 | 0.99 | 0.98‐0.99 |
| Chronic pulmonary disease | 153,667 | 132,287 | 54 | 1.08 | 1.08‐1.08 |
| Coagulopathy | 25,802 | 22,711 | 53 | 1.03 | 1.02‐1.04 |
| Depression | 83,051 | 62,502 | 57 | 1.08 | 1.08‐1.09 |
| DM without chronic complications | 136,184 | 133,903 | 50 | 0.99 | 0.99‐0.99 |
| DM with chronic complications | 38,696 | 32,036 | 55 | 1.04 | 1.03‐1.05 |
| Drug abuse | 37,202 | 29,684 | 56 | 1.14 | 1.13‐1.15 |
| Hypertension | 344,718 | 351,581 | 50 | 0.98 | 0.97‐0.98 |
| Hypothyroidism | 70,786 | 75,350 | 48 | 0.99 | 0.99‐0.99 |
| Liver disease | 24,067 | 14,063 | 63 | 1.15 | 1.14‐1.16 |
| Lymphoma | 7727 | 6305 | 55 | 1.16 | 1.14‐1.17 |
| Fluid and electrolyte disorders | 168,814 | 157,762 | 52 | 1.04 | 1.03‐1.04 |
| Metastatic cancer | 23,920 | 9515 | 72 | 1.40 | 1.39‐1.42 |
| Other neurological disorders | 51,091 | 73,104 | 41 | 0.87 | 0.86‐0.87 |
| Obesity | 69,584 | 49,331 | 59 | 1.05 | 1.04‐1.05 |
| Paralysis | 17,497 | 21,087 | 45 | 0.97 | 0.96‐0.98 |
| PVD | 42,176 | 35,158 | 55 | 1.11 | 1.11‐1.12 |
| Psychoses | 38,638 | 63,218 | 38 | 0.91 | 0.90‐0.92 |
| Pulmonary circulation disease | 26,656 | 25,450 | 51 | 1.05 | 1.04‐1.06 |
| Renal failure | 86,565 | 88,833 | 49 | 1.01 | 1.01‐1.02 |
| Solid tumor without metastasis | 16,258 | 13,336 | 55 | 1.14 | 1.13‐1.15 |
| Peptic ulcer disease excluding bleeding | 376 | 160 | 70 | 1.12 | 1.07‐1.18 |
| Valvular disease | 38,396 | 48,220 | 44 | 0.93 | 0.92‐0.94 |
| Weight loss | 25,724 | 19,408 | 57 | 1.09 | 1.08‐1.10 |
| Primary discharge diagnosesb | |||||
| Cancer | 13,986 | 5182 | 73 | 1.20 | 1.19‐1.21 |
| Musculoskeletal injuries | 14,638 | 2160 | 87 | 2.02 | 2.002.04 |
| Pain‐related diagnosesc | 64,656 | 36,877 | 64 | 1.20 | 1.20‐1.21 |
| Alcohol‐related disorders | 3425 | 13,352 | 20 | 0.46 | 0.44‐0.47 |
| Substance‐related disorders | 8680 | 5017 | 63 | 1.03 | 1.01‐1.04 |
| Psychiatric disorders | 7253 | 33,900 | 18 | 0.37 | 0.36‐0.38 |
| Mood disorders | 5943 | 22,818 | 21 | ||
| Schizophrenia and other psychotic disorders | 1310 | 11,082 | 11 | ||
| Proceduresb | |||||
| Cardiovascular procedures | 50,997 | 8904 | 85 | 1.80 | 1.79‐1.81 |
| GI procedures | 27,206 | 4018 | 87 | 1.70 | 1.69‐1.71 |
| Mechanical ventilation | 5341 | 2512 | 68 | 1.37 | 1.34‐1.39 |
| Hospital characteristics | |||||
| Number of beds | |||||
| <200 | 100,900 | 88,439 | 53 | (ref) | |
| 201300 | 104,213 | 99,995 | 51 | 0.95 | 0.95‐0.96 |
| 301500 | 215,340 | 209,104 | 51 | 0.94 | 0.94‐0.95 |
| >500 | 155,920 | 165,508 | 49 | 0.96 | 0.95‐0.96 |
| Population served | |||||
| Urban | 511,727 | 506,803 | 50 | (ref) | |
| Rural | 64,646 | 56,243 | 53 | 0.98 | 0.97‐0.99 |
| Teaching status | |||||
| Nonteaching | 366,623 | 343,581 | 52 | (ref) | |
| Teaching | 209,750 | 219,465 | 49 | 1.00 | 0.99‐1.01 |
| US Census region | |||||
| Northeast | 99,377 | 149,446 | 40 | (ref) | |
| Midwest | 123,194 | 120,322 | 51 | 1.26 | (1.25‐1.27) |
| South | 251,624 | 213,029 | 54 | 1.33 | (1.33‐1.34) |
| West | 102,178 | 80,249 | 56 | 1.37 | (1.36‐1.38) |
Variation in Opioid Prescribing
Figure 1 shows the histograms of hospital opioid prescribing rate for the 286 hospitals in our cohort (A) before and (B) after adjustment for patient characteristics. The observed rates ranged from 5% in the lowest‐prescribing hospital to 72% in the highest‐prescribing hospital, with a mean (SD) of 51% (10%). After adjusting for patient characteristics, the adjusted opioid‐prescribing rates ranged from 33% to 64%, with a mean (SD) of 50% (4%).
Severe Opioid‐Related Adverse Events
Among admissions with opioid exposure (n=576,373), naloxone use occurred in 2345 (0.41%) and opioid‐related adverse drug events in 1174 (0.20%), for a total of 3441 (0.60%) severe opioid‐related adverse events (some patients experienced both). Table 4 reports the opioid exposure and severe opioid‐related adverse‐event rates within hospital opioid‐prescribing rate quartiles, along with the adjusted association between the hospital opioid‐prescribing rate quartile and severe opioid‐related adverse events. After adjusting for patient characteristics, the relative risk of a severe opioid‐related adverse event was significantly greater in hospitals with higher opioid‐prescribing rates, both overall and among opioid exposed.
| Quartile | No. of Patients | Opioid Exposed, n (%) | Opioid‐Related Adverse Events, n (%) | Adjusted RR in All Patients, RR (95% CI), N=1,139,419a | Adjusted RR in Opioid Exposed, RR (95% CI), N=576,373a |
|---|---|---|---|---|---|
| |||||
| 1 | 349,747 | 132,824 (38) | 719 (0.21) | (ref) | (ref) |
| 2 | 266,652 | 134,590 (50) | 729 (0.27) | 1.31 (1.17‐1.45) | 1.07 (0.96‐1.18) |
| 3 | 251,042 | 139,770 (56) | 922 (0.37) | 1.72 (1.56‐1.90) | 1.31 (1.19‐1.44) |
| 4 | 271,978 | 169,189 (62) | 1071 (0.39) | 1.73 (1.57‐1.90) | 1.23 (1.12‐1.35) |
DISCUSSION
In this analysis of a large cohort of hospitalized nonsurgical patients, we found that more than half of all patients received opioids, with 43% of those exposed receiving multiple opioids during their admission and 52% receiving opioids on the day of discharge. Considerable hospital variation in opioid use was evident, and this was not fully explained by patient characteristics. Severe opioid‐related adverse events occurred more frequently at hospitals with higher opioid‐prescribing rates, and the relative risk of a severe adverse event per patient prescribed opioids was also higher in these hospitals. To our knowledge, this is the first study to describe the scope of opioid utilization and the relationship between utilization and severe opioid‐related adverse events in a sample of nonsurgical patients in US acute‐care facilities.
Our use of naloxone charges and opioid‐specific ICD‐9‐CM coding to define an opioid‐related adverse event was intended to capture only the most severe opioid‐related adverse events. We chose to focus on these events in our analysis to maximize the specificity of our outcome definition and thereby minimize confounding in our observed associations. The rate of less‐severe opioid‐related adverse events, such as nausea, constipation, and pruritis, is likely much higher and not captured in our outcome definition. Prior analyses have found variable rates of opioid‐related adverse events of approximately 1.8% to 13.6% of exposed patients.[22, 23, 24] However, these analyses focused on surgical patients and included less‐severe events. To our knowledge, ours is the first analysis of severe opioid‐related adverse events in nonsurgical patients.
Our finding that severe opioid‐related adverse events increase as opioid prescribing increases is consistent with that which has been demonstrated in the community setting, where rates of opioid‐related adverse events and mortality are higher in communities with higher levels of opioid prescribing.[2, 8, 25] This finding is expected, as greater use of a class of medications with known side effects would be expected to result in a higher overall rate of adverse events. More concerning, however, is the fact that this relationship persists when focusing exclusively on opioid‐exposed patients. Among similar patients receiving opioids at different hospitals, those hospitalized in facilities with higher opioid‐prescribing rates have higher rates of severe opioid‐related adverse events. This suggests that hospitals that use opioids more frequently do not do so more safely. Rather, the increased overall prescribing rates are associated with heightened risk for a serious adverse event per patient exposed and may reflect unsafe prescribing practices.
Furthermore, our results demonstrate both regional and hospital variation in use of opioids not fully explained by patient characteristics, similar to that which has been demonstrated for other drugs and heathcare services.[26, 27, 28, 29, 30] The implications of these findings are limited by our lack of information on pain severity or prior outpatient treatment, and our resultant inability to evaluate the appropriateness of opioid use in this analysis. Additionally, although we controlled for a large number of patient and hospital characteristics, there could be other significant predictors of use not accounted for in our analysis. However, it seems unlikely that differential pain severity or patient characteristics between patients in different regions of the country could fully explain a 37% relative difference in prescribing between the lowest‐ and highest‐prescribing regions, after accounting for the 44 patient‐level variables in our models. Whereas variation in use unrelated to patient factors could represent inappropriate prescribing practices, it could also be a marker of uncertainty regarding what constitutes appropriate prescribing and high‐quality care in this realm. Although guidelines advocate for standard pain assessments and a step‐up approach to treatment,[31, 32, 33] the lack of objective measures of pain severity and lack of evidence‐based recommendations on the use of opioids for noncancer pain[34] will almost certainly lead to persistent variation in opioid prescribing despite guideline‐driven care.
Nonetheless, our findings suggest that opportunities exist to make opioid prescribing safer in hospitalized patients. Studies aimed at elucidating the source of regional and hospital variation are necessary. Additionally, efforts should focus on identifying patient and prescribing characteristics associated with heightened risk of opioid‐related adverse events. Prior studies have demonstrated that the risks of opioid medications increase with increasing age of the patient.[35, 36] Although opioid use in our cohort declined with age, 44% of admissions age 65 years had charges for opioid medications. Studies in outpatients have also demonstrated that the risks of opioid overdose and overdose‐related death increase with dose.[5, 7] One study demonstrated a 3.7‐fold increased risk of overdose at doses of 5099 mg/day in oral morphine equivalents, and an 8.9‐fold increased risk at doses of 100 mg/day, compared with doses of 20 mg/day.[7] The prevalence of high dose exposure observed in our cohort, coupled with the older age of hospitalized patients, suggests potential targets for promoting safer use in hospitalized patients through interventions such as computerized decision support and enhanced monitoring in those at highest risk.
Because medications after discharge were unavailable in our dataset, the percentage of patients given a prescription for opioid medication on discharge is unknown. However, given that opioids are often tapered rather than abruptly discontinued, our finding that 26% of hospitalized nonsurgical patients received opioids on the day of discharge suggests that a substantial proportion of patients may be discharged with a prescription for opioid medication. Given the possibility of coexistent outpatient opioid prescriptions, these findings draw attention to the importance of assuring development and streamlined accessibility of data from state prescription drug monitoring programs and suggest that increased attention should be paid to the role that inpatient opioid prescribing plays in the increased rates of chronic opioid use and overdose‐related deaths in the United States.
There are additional limitations to our analysis. First, although the database used for this analysis captures a large proportion of admissions to US acute‐care facilities and is similar in composition, it is possible that participating medical centers differ from nonparticipating medical centers in ways that could be associated with opioid prescribing. Additionally, although Premier performs extensive validation and correction processes to assure the quality of their data, there is still likely to be a small amount of random error in the database, which could particularly impact dosage calculations. The lack of preadmission medications in our database precluded identification of the proportion of patients newly started on opioid medications. Lastly, it is possible that the hospital prescribing‐rate quartile is associated with patient characteristics unaccounted for in our analysis, and, therefore, the possibility of residual confounding still exists.
In conclusion, the majority of hospitalized nonsurgical patients are exposed to opioid medications during the course of their hospitalizations, often at high doses. More than half of those exposed are still receiving these medications on the day of discharge. We found hospital and regional variation in opioid use that was not fully explained by patient characteristics, and higher levels of hospital use were associated with higher risk of severe opioid‐related adverse events in opioid‐exposed patients. Further research is necessary to investigate the appropriateness of opioid use in this patient population, the sources of variation in use, and the predictors of opioid‐related adverse events in hospitalized patients to allow development of interventions to make hospital use safer.
Disclosures
Disclosures: Dr. Herzig had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Herzig was funded by grant no. K23AG042459 from the National Institute on Aging. Dr. Marcantonio was funded by grant nos. P01AG031720, R01AG030618, R03AG028189, and K24AG035075 from the National Institute on Aging. The funding organization had no involvement in any aspect of the study, including design, conduct, and reporting of the study. None of the authors have any conflicts of interest to disclose.
- . A flood of opioids, a rising tide of deaths. N Engl J Med. 2010;363(21):1981–1985.
- , , . Increasing deaths from opioid analgesics in the United States. Pharmacoepidemiol Drug Saf. 2006;15(9):618–627.
- , , , . Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70–78.
- , , , . Trends in medical use and abuse of opioid analgesics. JAMA. 2000;283(13):1710–1714.
- , , , et al. Association between opioid prescribing patterns and opioid overdose‐related deaths. JAMA. 2011;305(13):1315–1321.
- , , , et al. Prescription opioid mortality trends in New York City, 1990–2006: examining the emergence of an epidemic. Drug Alcohol Depend. 2013;132(1‐2):53–62.
- , , , et al. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med. 2010;152(2):85–92.
- , , , et al. Relationship of opioid prescription sales and overdoses, North Carolina. Drug Alcohol Depend. 2013;132(1‐2):81–86.
- Deaths from prescription opioids soar in New York. BMJ. 2013;346:f921.
- , , . Mortality in elderly dementia patients treated with risperidone. J Clin Psychopharmacol. 2006;26(6):566–570.
- , , , et al. Medication reconciliation: a practical tool to reduce the risk of medication errors. J Crit Care. 2003;18(4):201–205.
- , , . Adverse drug event trigger tool: a practical methodology for measuring medication related harm. Qual Saf Health Care. 2003;12(3):194–200.
- , , , , . Improvement in the detection of adverse drug events by the use of electronic health and prescription records: an evaluation of two trigger tools. Eur J Clin Pharmacol. 2013;69(2):255–259.
- , . Adverse Drug Events in U.S. Hospitals, 2004. Rockville, MD: Agency for Healthcare Research and Quality; April 2007. Statistical Brief 29.
- , , . Medication‐Related Adverse Outcomes in U.S. Hospitals and Emergency Departments, 2008. Rockville, MD: Agency for Healthcare Research and Quality; April 2011. Statistical Brief 109.
- US Centers for Medicare and Medicaid Services. Hospital‐Acquired Conditions (Present on Admission Indicator). Available at: http://www.cms.hhs.gov/HospitalAcqCond/. Accessed August 16, 2011. Updated September 20, 2012.
- , , . Fracture risk associated with the use of morphine and opiates. J Intern Med. 2006;260(1):76–87.
- , , , . Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.
- Clinical Classifications Software, Healthcare Cost and Utilization Project (HCUP). Rockville, MD; Agency for Healthcare Research and Quality; December 2009.
- , , , , . Clinical application of opioid equianalgesic data. Clin J Pain. 2003;19(5)286–297.
- , , , , , . Choosing the unit of measurement counts: the use of oral morphine equivalents in studies of opioid consumption is a useful addition to defined daily doses. Palliat Med. 2011;25(7):725–732.
- , , , . Cost and quality implications of opioid‐based postsurgical pain control using administrative claims data from a large health system: opioid‐related adverse events and their impact on clinical and economic outcomes. Pharmacotherapy. 2013;33(4):383–391.
- , , , et al. Opioid‐related adverse drug events in surgical hospitalizations: impact on costs and length of stay. Ann Pharmacother. 2007;41(3):400–406.
- , , , et al. Cost of opioid‐related adverse drug events in surgical patients. J Pain Symptom Manage. 2003;25(3):276–283.
- , . Opioid analgesics and rates of fatal drug poisoning in the United States. Am J Prev Med. 2006;31(6):506–511.
- , , , et al. Geographic variation in the treatment of acute myocardial infarction: the Cooperative Cardiovascular Project. JAMA. 1999;281(7):627–633.
- , , , et al; GUSTO‐1 Investigators. Regional variation across the United States in the management of acute myocardial infarction. Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries. N Engl J Med. 1995;333(9):565–572.
- , , . Predictors of broad‐spectrum antibiotic prescribing for acute respiratory tract infections in adult primary care. JAMA. 2003;289(6):719–725.
- , , . Geographic variation in Medicare drug spending. N Engl J Med. 2010;363(5):405–409.
- , , . Geographic variation in outpatient antibiotic prescribing among older adults. Arch Intern Med. 2012;172(19):1465–1471.
- , , , et al. Clinical policy: critical issues in the prescribing of opioids for adult patients in the emergency department. Ann Emerg Med. 2012;60(4):499–525.
- The Joint Commission. Facts About Pain Management. Available at: http://www.jointcommission.org/pain_management/. Accessed July 23, 2012.
- The Joint Commission. Sentinel Event Alert: Safe Use of Opioids in Hospitals. Published August 8, 2012. Available at: http://www.jointcommission.org/assets/1/18/SEA_49_opioids_8_2_12_final.pdf. Accessed March 4, 2013.
- , , , , . Research gaps on use of opioids for chronic noncancer pain: findings from a review of the evidence for an American Pain Society and American Academy of Pain Medicine clinical practice guideline. J Pain. 2009;10(2):147–159.
- , , , , , . Side effects of opioids during short‐term administration: effect of age, gender, and race. Clin Pharmacol Ther. 2003;74(2):102–112.
- , , , . Postoperative day one: a high‐risk period for respiratory events. Am J Surg. 2005;190(5):752–756.
- . A flood of opioids, a rising tide of deaths. N Engl J Med. 2010;363(21):1981–1985.
- , , . Increasing deaths from opioid analgesics in the United States. Pharmacoepidemiol Drug Saf. 2006;15(9):618–627.
- , , , . Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70–78.
- , , , . Trends in medical use and abuse of opioid analgesics. JAMA. 2000;283(13):1710–1714.
- , , , et al. Association between opioid prescribing patterns and opioid overdose‐related deaths. JAMA. 2011;305(13):1315–1321.
- , , , et al. Prescription opioid mortality trends in New York City, 1990–2006: examining the emergence of an epidemic. Drug Alcohol Depend. 2013;132(1‐2):53–62.
- , , , et al. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med. 2010;152(2):85–92.
- , , , et al. Relationship of opioid prescription sales and overdoses, North Carolina. Drug Alcohol Depend. 2013;132(1‐2):81–86.
- Deaths from prescription opioids soar in New York. BMJ. 2013;346:f921.
- , , . Mortality in elderly dementia patients treated with risperidone. J Clin Psychopharmacol. 2006;26(6):566–570.
- , , , et al. Medication reconciliation: a practical tool to reduce the risk of medication errors. J Crit Care. 2003;18(4):201–205.
- , , . Adverse drug event trigger tool: a practical methodology for measuring medication related harm. Qual Saf Health Care. 2003;12(3):194–200.
- , , , , . Improvement in the detection of adverse drug events by the use of electronic health and prescription records: an evaluation of two trigger tools. Eur J Clin Pharmacol. 2013;69(2):255–259.
- , . Adverse Drug Events in U.S. Hospitals, 2004. Rockville, MD: Agency for Healthcare Research and Quality; April 2007. Statistical Brief 29.
- , , . Medication‐Related Adverse Outcomes in U.S. Hospitals and Emergency Departments, 2008. Rockville, MD: Agency for Healthcare Research and Quality; April 2011. Statistical Brief 109.
- US Centers for Medicare and Medicaid Services. Hospital‐Acquired Conditions (Present on Admission Indicator). Available at: http://www.cms.hhs.gov/HospitalAcqCond/. Accessed August 16, 2011. Updated September 20, 2012.
- , , . Fracture risk associated with the use of morphine and opiates. J Intern Med. 2006;260(1):76–87.
- , , , . Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.
- Clinical Classifications Software, Healthcare Cost and Utilization Project (HCUP). Rockville, MD; Agency for Healthcare Research and Quality; December 2009.
- , , , , . Clinical application of opioid equianalgesic data. Clin J Pain. 2003;19(5)286–297.
- , , , , , . Choosing the unit of measurement counts: the use of oral morphine equivalents in studies of opioid consumption is a useful addition to defined daily doses. Palliat Med. 2011;25(7):725–732.
- , , , . Cost and quality implications of opioid‐based postsurgical pain control using administrative claims data from a large health system: opioid‐related adverse events and their impact on clinical and economic outcomes. Pharmacotherapy. 2013;33(4):383–391.
- , , , et al. Opioid‐related adverse drug events in surgical hospitalizations: impact on costs and length of stay. Ann Pharmacother. 2007;41(3):400–406.
- , , , et al. Cost of opioid‐related adverse drug events in surgical patients. J Pain Symptom Manage. 2003;25(3):276–283.
- , . Opioid analgesics and rates of fatal drug poisoning in the United States. Am J Prev Med. 2006;31(6):506–511.
- , , , et al. Geographic variation in the treatment of acute myocardial infarction: the Cooperative Cardiovascular Project. JAMA. 1999;281(7):627–633.
- , , , et al; GUSTO‐1 Investigators. Regional variation across the United States in the management of acute myocardial infarction. Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries. N Engl J Med. 1995;333(9):565–572.
- , , . Predictors of broad‐spectrum antibiotic prescribing for acute respiratory tract infections in adult primary care. JAMA. 2003;289(6):719–725.
- , , . Geographic variation in Medicare drug spending. N Engl J Med. 2010;363(5):405–409.
- , , . Geographic variation in outpatient antibiotic prescribing among older adults. Arch Intern Med. 2012;172(19):1465–1471.
- , , , et al. Clinical policy: critical issues in the prescribing of opioids for adult patients in the emergency department. Ann Emerg Med. 2012;60(4):499–525.
- The Joint Commission. Facts About Pain Management. Available at: http://www.jointcommission.org/pain_management/. Accessed July 23, 2012.
- The Joint Commission. Sentinel Event Alert: Safe Use of Opioids in Hospitals. Published August 8, 2012. Available at: http://www.jointcommission.org/assets/1/18/SEA_49_opioids_8_2_12_final.pdf. Accessed March 4, 2013.
- , , , , . Research gaps on use of opioids for chronic noncancer pain: findings from a review of the evidence for an American Pain Society and American Academy of Pain Medicine clinical practice guideline. J Pain. 2009;10(2):147–159.
- , , , , , . Side effects of opioids during short‐term administration: effect of age, gender, and race. Clin Pharmacol Ther. 2003;74(2):102–112.
- , , , . Postoperative day one: a high‐risk period for respiratory events. Am J Surg. 2005;190(5):752–756.
© 2013 Society of Hospital Medicine
Caring About Prognosis
Prognostication continues to be a challenge to the clinician despite over 100 prognostic indices that have been developed during the past few decades to inform clinical practice and medical decision making.[1] Physicians are not accurate in prognostication of patients' risk of death and tend to overestimate survival.[2, 3] In addition, many physicians do not feel comfortable offering a prognosis to patients, despite patients' wishes to be informed.[4, 5] Regardless of the prevalence in the literature and value in improving physicians' prognostic accuracy, prognostic indices of survival are not regularly utilized in the hospital setting. Prognostic tools available for providers are often complicated and may require data about patients that are not readily available.[6, 7, 8] Prognostic indices may be too specific to a patient population, too difficult to remember, or too time consuming to use. A simple, rapid, and practical prognostic index is important in the hospital setting to assist in identifying patients at high risk of death so that primary palliative interventions can be incorporated into the plan of care early in the hospital stay. Patient and family education, advance care planning, formulating the plan of care based on patientfamily goals, and improved resource utilization could be better executed by more accurate risk of death prediction on hospital admission.
The CARING criteria are the only prognostic index to our knowledge that evaluates a patient's risk of death in the next year, with information readily available at the time of hospital admission (Table 1).[9] The CARING criteria are a unique prognostic tool: (1) CARING is a mnemonic acronym, making it more user friendly to the clinician. (2) The 5 prognostic indicators are readily available from the patient's chart on admission; gathering further data by patient or caretaker interviews or by obtaining laboratory data is not needed. (3) The timing for application of the tool on admission to the hospital is an ideal opportunity to intervene and introduce palliative interventions early on the hospital stay. The CARING criteria were developed and validated in a Veteran's Administration hospital setting by Fischer et al.[9] We sought to validate the CARING criteria in a broader patient populationmedical and surgical patients from a tertiary referral university hospital setting and a safety‐net hospital setting.
METHODS
Study Design
This study was a retrospective observational cohort study. The study was approved by the Colorado Multiple Institutional Review Board and the University of Colorado Hospital Research Review Committee.
Study Purpose
To validate the CARING criteria in a tertiary referral university hospital (University of Colorado Hospital [UCH]) and safety‐net hospital (Denver Health and Hospitals [DHH]) setting using similar methodology to that employed by the original CARING criteria study.[9]
Study Setting/Population
All adults (18 years of age) admitted as inpatients to the medical and surgical services of internal medicine, hospitalist, pulmonary, cardiology, hematology/oncology, hepatology, surgery, intensive care unit, and intermediary care unit at UCH and DHH during the study period of July 2005 through August 2005. The only exclusion criteria were those patients who were prisoners or pregnant. Administrative admission data from July 2005 to August 2005 were used to identify names of all persons admitted to the medicine and surgical services of the study hospitals during the specified time period.
The 2 study hospitals, UCH and DHH, provide a range of patients who vary in ethnicity, socioeconomic status, and medical illness. This variability allows for greater generalizability of the results. Both hospitals are affiliated with the University of Colorado School of Medicine internal medicine residency training program and are located in Denver, Colorado.
At the time of the study, UCH was a licensed 550‐bed tertiary referral, academic hospital serving the Denver metropolitan area and the Rocky Mountain region as a specialty care and referral center. DHH was a 398‐bed, academic, safety‐net hospital serving primarily the Denver metropolitan area. DHH provides 42% of the care for the uninsured in Denver and 26% of the uninsured care for the state of Colorado.
Measures
The CARING criteria were developed and validated in a Veteran's Administration (VA) hospital setting by Fischer et al.[9] The purpose of the CARING criteria is to identify patients, at the time of hospital admission, who are at higher risk of death in the following year. The prognostic index uses 5 predictors that can be abstracted from the chart at time of admission. The CARING criteria were developed a priori, and patients were evaluated using only the medical data available at the time of admission. The criteria include items that are already part of the routine physician admission notes and do not require additional data collection or assessments. The criteria include: C=primary diagnosis of cancer, A=2 admissions to the hospital for a chronic illness within the last year; R=resident in a nursing home; I=intensive care unit (ICU) admission with multiorgan failure (MOF), NG=noncancer hospice guidelines (meeting 2 of the National Hospice and Palliative Care Organization's [NHPCO] guidelines).
Patients were identified using name, date of birth, social security number, address, and phone number. This identifying information was then used for tracing death records 1 year after hospital admission.
Mortality at 1 year following the index hospitalization was the primary end point. To minimize missing data and the number of subjects lost to follow‐up, 3 determinants of mortality were used. First, electronic medical records of the 2 participating hospitals and their outpatient clinics were reviewed to determine if a follow‐up appointment had occurred past the study's end point of 1 year (August 2006). For those without a confirmed follow‐up visit, death records from the Colorado Department of Public Health and Vital Records were obtained. For those patients residing outside of Colorado or whose mortality status was still unclear, the National Death Index was accessed.
Medical Record Review
Medical records for all study participants were reviewed by J.Y. (UCH) and B.C. (DHH). Data collection was completed using direct data entry into a Microsoft Access (Microsoft Corp., Redmond, WA) database utilizing a data entry form linked with the database table. This form utilized skip patterns and input masks to ensure quality of data entry and minimize missing or invalid data. Inter‐rater reliability was assessed by an independent rereview (S.F.) of 5% of the total charts. Demographic variables were collected using hospital administrative data. These included personal identifiers of the participants for purposes of mortality follow‐up. Clinical data including the 5 CARING variables and additional descriptive variables were abstracted from the paper hospital chart and the electronic record of the chart (together these constitute the medical record).
Death Follow‐up
A search of Colorado death records was conducted in February 2011 for all subjects. Death records were used to determine mortality and time to death from the index hospitalization. The National Death Index was then searched for any subjects without or record of death in Colorado.
Analysis
All analyses were conducted using the statistical application software SAS for Windows version 9.3 (SAS Institute, Cary, NC). Simple frequencies and means ( standard deviation) were used to describe the baseline characteristics. Multiple logistic regression models were used to model 1‐year mortality. The models were fitted using all of the CARING variables and age. As the aim of the study was to validate the CARING criteria, the variables for the models were selected a priori based on the original index. Two hospital cohorts (DHH and UCH) were modeled separately and as a combined sample. Kaplan‐Meier survival analysis was conducted to compare those subjects who met 1 of the CARING criteria with those who did not through the entire period of mortality follow‐up (20052011). Finally, using the probabilities from the logistic regression models, we again developed a scoring rule appropriate for a non‐VA setting to allow clinicians to easily identify patient risk for 1‐year mortality at the time of hospital admission.
RESULTS
There were a total of 1064 patients admitted to the medical and surgical services during the study period568 patients at DHH and 496 patients at UCH. Sample characteristics of each individual hospital cohort and the entire combined study cohort are detailed in Table 2. Overall, slightly over half the population were male, with a mean age of 50 years, and the ethnic breakdown roughly reflects the population in Denver. A total of 36.5% (n=388) of the study population met 1 of the CARING criteria, and 12.6% (n=134 among 1063 excluding 1 without an admit date) died within 1 year of the index hospitalization. These were younger and healthier patients compared to the VA sample used in developing the CARING criteria.
| |
| Renal | Dementia |
| Stop/decline dialysis | Unable to ambulate independently |
| Not candidate for transplant | Urinary or fecal incontinence |
| Urine output < 40cc/24 hours | Unable to speak with more than single words |
| Creatinine > 8.0 (>6.0 for diabetics) | Unable to bathe independently |
| Creatinine clearance 10cc/min | Unable to dress independently |
| Uremia | Co‐morbid conditions: |
| Persistent serum K + > 7.0 | Aspiration pneumonia |
| Co‐morbid conditions: | Pyelonephritis |
| Cancer CHF | Decubitus ulcer |
| Chronic lung disease AIDS/HIV | Difficulty swallowing or refusal to eat |
| Sepsis Cirrhosis | |
| Cardiac | Pulmonary |
| Ejection fraction < 20% | Dyspnea at rest |
| Symptomatic with diuretics and vasodilators | FEV1 < 30% |
| Not candidate for transplant | Frequent ER or hospital admits for pulmonary infections or respiratory distress |
| History of cardiac arrest | Cor pulmonale or right heart failure |
| History of syncope | 02 sat < 88% on 02 |
| Systolic BP < 120mmHG | PC02 > 50 |
| CVA cardiac origin | Resting tachycardia > 100/min |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| Liver | Stroke/CVA |
| End stage cirrhosis | Coma at onset |
| Not candidate for transplant | Coma >3 days |
| Protime > 5sec and albumin <2.5 | Limb paralysis |
| Ascites unresponsive to treatment | Urinary/fecal incontinence |
| Hepatorenal syndrome | Impaired sitting balance |
| Hepatic encephalopathy | Karnofsky < 50% |
| Spontaneous bacterial peritonitis | Recurrent aspiration |
| Recurrent variceal bleed | Age > 70 |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| HIV/AIDS | Neuromuscular |
| Persistent decline in function | Diminished respiratory function |
| Chronic diarrhea 1 year | Chosen not to receive BiPAP/vent |
| Decision to stop treatment | Difficulty swallowing |
| CNS lymphoma | Diminished functional status |
| MAC‐untreated | Incontinence |
| Systemic lymphoma | Co‐morbid conditions as listed in Renal |
| Dilated cardiomyopathy | |
| CD4 < 25 with disease progression | |
| Viral load > 100,000 | |
| Safety‐Net Hospital Cohort, N=568 | Academic Center Cohort, N=496 | Study Cohort,N=1064 | Original CARING Cohort, N=8739 | |
|---|---|---|---|---|
| ||||
| Mean age ( SD), y | 47.8 (16.5) | 54.4 (17.5) | 50.9 (17.3) | 63 (13) |
| Male gender | 59.5% (338) | 50.1% (248) | 55.1% (586) | 98% (856) |
| Ethnicity | ||||
| African American | 14.1% (80) | 13.5% (65) | 13.8% (145) | 13% (114) |
| Asian | 0.4% (2) | 1.5% (7) | 0.9% (9) | Not reported |
| Caucasian | 41.7% (237) | 66.3% (318) | 53.0 % (555) | 69% (602) |
| Latino | 41.9% (238) | 9.6% (46) | 27.1% (284) | 8% (70) |
| Native American | 0.5% (3) | 0.4% (2) | 0.5% (5) | Not reported |
| Other | 0.5% (3) | 0.6% (3) | 0.6% (6) | 10% (87) |
| Unknown | 0.9% (5) | 8.1% (39) | 4.2% (44) | Not reported |
| CARING criteria | ||||
| Cancer | 6.2% (35) | 19.4% (96) | 12.3% (131) | 23% (201) |
| Admissions to the hospital 2 in past year | 13.6% (77) | 42.7% (212) | 27.2% (289) | 36% (314) |
| Resident in a nursing home | 1.8% (10) | 3.4% (17) | 2.5% (27) | 3% (26) |
| ICU with MOF | 3.7% (21) | 1.2% (6) | 2.5% (27) | 2% (17) |
| NHPCO (2) noncancer guidelines | 1.6% (9) | 5.9% (29) | 3.6% (38) | 8% (70) |
Reliability testing demonstrated excellent inter‐rater reliability. Kappa for each criterion is as follows: (1) primary diagnosis of cancer=1.0, (2) 2 admissions to the hospital in the past year=0.91, (3) resident in a nursing home=1.0, (4) ICU admission with MOF=1.0, and (5) 2 noncancer hospice guidelines=0.78.
This study aimed to validate the CARING criteria9; therefore, all original individual CARING criterion were included in the validation logistic regression models. The 1 exception to this was in the university hospital study cohort, where the ICU criterion was excluded from the model due to small sample size and quasiseparation in the model. The model results are presented in Table 3 for the individual hospitals and combined study cohort.
| Safety Net Hospital Cohort, C Index=0.76 | Academic Center Cohort, C Index=0.76 | Combined Hospital Cohort, C Index=0.79 | ||||
|---|---|---|---|---|---|---|
| Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | |
| ||||||
| Cancer | 1.92 | 6.85 (2.83‐16.59)a | 1.85 | 6.36 (3.54‐11.41)a | 1.98 | 7.23 (4.45‐11.75)a |
| Admissions to the hospital 2 in past year | 0.55 | 1.74 (0.76‐3.97) | 0.14 | 0.87 (0.51‐1.49) | 0.20 | 1.22 (0.78‐1.91) |
| Resident in a nursing home | 0.49 | 0.61 (0.06‐6.56) | 0.27 | 1.31 (0.37‐4.66) | 0.09 | 1.09 (0.36‐3.32) |
| ICU with MOF | 1.85 | 6.34 (2.0219.90)a | 1.94 | 6.97 (2.75‐17.68)a | ||
| NHPCO (2) noncancer guidelines | 3.04 | 20.86 (4.25102.32)a | 2.62 | 13.73 (5.86‐32.15)a | 2.74 | 15.55 (7.2833.23)a |
| Ageb | 0.38 | 1.46 (1.05‐2.03)a | 0.45 | 1.56 (1.23‐1.98)a | 0.47 | 1.60 (1.32‐1.93)a |
In the safety‐net hospital, admission to the hospital with a primary diagnosis related to cancer, 2 noncancer hospice guidelines, ICU admission with MOF, and age by category all were significant predictors of 1‐year mortality. In the university hospital cohort, primary diagnosis of cancer, 2 noncancer hospice guidelines, and age by category were predictive of 1‐year mortality. Finally, in the entire study cohort, primary diagnosis of cancer, ICU with MOF, 2 noncancer hospice guidelines, and age were all predictive of 1‐year mortality. Parameter estimates were similar in 3 of the criteria compared to the VA setting. Differences in patient characteristics may have caused the differences in the estimates. Gender was additionally tested but not significant in any model. One‐year survival was significantly lower for those who met 1 of the CARING criteria versus those who did not (Figure 1).
Based on the framework from the original CARING criteria analysis, a scoring rule was developed using the regression results of this validation cohort. To predict a high probability of 1‐year mortality, sensitivity was set to 58% and specificity was set at 86% (error rate=17%). Medium to high probability was set with a sensitivity of 73% and specificity of 72% (error rate=28%). The coefficients from the regression model of the entire study cohort were converted to scores for each of the CARING criteria. The scores are as follows: 0.5 points for admission from a nursing home, 1 point for 2 hospital admissions in the past year for a chronic illness, 10 points for primary diagnosis of cancer, 10 points for ICU admission with MOF, and 14 points for 2 noncancer hospice guidelines. For every age category increase, 2 points are assigned so that 0 points for age <55 years, 2 points for ages 56 to 65 years, 4 points for ages 66 to 75 years, and 6 points for >75 years. Points for individual risk factors were proportional to s (ie, log odds) in the logistic regression model for death at 1 year. Although no linear transformation exists between s and probabilities (of death at 1 year), the aggregated points for combinations of risk factors shown in Table 4 follow the probabilities in an approximately linear fashion, so that different degrees of risk of death can be represented contiguously (as highlighted by differently shaded regions in the scoring matrix) (Table 4). The scoring matrix allows for quick identification for patients at high risk for 1‐year mortality. In this non‐VA setting with healthier patients, low risk is defined at a lower probability threshold (0.1) compared to the VA setting (0.175).
| CARING Criteria Components | |||||||
|---|---|---|---|---|---|---|---|
| None | Resident in a Nursing Home | Admitted to the Hospital 2 Times in the Past Year | Resident in a Nursing Home Admitted to the Hospital 2 Times in the Past Year | Primary Diagnosis of Cancer | ICU Admission With MOF | Noncancer Hospice Guidelines | |
| |||||||
| Age | |||||||
| 55 years | 0 | 0.5 | 1 | 1.5 | 10 | ||
| 5565 years | 2 | 2.5 | 3 | 3.5 | 10 | ||
| 6675 years | 4 | 4.5 | 5 | 5.5 | 10 | ||
| >75 years | 6 | 6.5 | 7 | 7.5 | 10 | ||
| Risk | |||||||
| Low | 3.5 | Probability<0.1 | |||||
| Medium | 46.5 | 0.1probability <0.175 | |||||
| High | 7 | Probability0.175 | |||||
DISCUSSION
The CARING criteria are a practical prognostic tool that can be easily and rapidly applied to patients admitted to the hospital to estimate risk of death in 1 year, with the goal of identifying patients who may benefit most from incorporating palliative interventions into their plan of care. This study validated the CARING criteria in a tertiary referral university hospital and safety‐net hospital setting, demonstrating applicability in a much broader population than the VA hospital of the original CARING criteria study. The population studied represented a younger population by over 10 years, a more equitable proportion of males to females, a broader ethnic diversity, and lower 1‐year deaths rates than the original study. Despite the broader representation of the population, the significance of each of the individual CARING criterion was maintained except for 2 hospital admissions in the past year for a chronic illness (admission from a nursing home did not meet significance in either study as a sole criterion). As with the original study, meeting 2 of the NHPCO noncancer hospice guidelines demonstrated the highest risk of 1‐year mortality following index hospitalization, followed by primary diagnosis of cancer and ICU admission with MOF. Advancing age, also similar to the original study, conferred increased risk across the criterion.
Hospitalists could be an effective target for utilizing the CARING criteria because they are frequently the first‐line providers in the hospital setting. With the national shortage of palliative care specialists, hospitalists need to be able to identify when a patient has a limited life expectancy so they will be better equipped to make clinical decisions that are aligned with their patients' values, preferences, and goals of care. With the realization that not addressing advance care planning and patient goals of care may be considered medical errors, primary palliative care skills become alarmingly more important as priorities for hospitalists to obtain and feel comfortable using in daily practice.
The CARING criteria are directly applicable to patients who are seen by hospitalists. Other prognostic indices have focused on select patient populations, such as the elderly,[10, 11, 12] require collection of data that are not readily available on admission or would not otherwise be obtained,[10, 13] or apply to patients post‐hospital discharge, thereby missing the opportunity to make an impact earlier in the disease trajectory and incorporate palliative care into the hospital plan of care when key discussions about goals of care and preferences should be encouraged.
Additionally, the CARING criteria could easily be incorporated as a trigger for palliative care consults on hospital admission. Palliative care consults tend to happen late in a hospital stay, limiting the effectiveness of the palliative care team. A trigger system for hospitalists and other primary providers on hospital admission would lend to more effective timing of palliative measures being incorporated into the plan of care. Palliative care consults would not only be initiated earlier, but could be targeted for the more complex and sick patients with the highest risk of death in the next year.
In the time‐pressured environment, the presence of any 1 of the CARING criteria can act as a trigger to begin incorporating primary palliative care measures into the plan of care. The admitting hospitalist provider (ie, physician, nurse practitioner, physician assistant) could access the CARING criteria through an electronic health record prompt when admitting patients. When a more detailed assessment of mortality risk is helpful, the hospitalist can use the scoring matrix, which combines age with the individual criterion to calculate patients at medium or high risk of death within 1 year. Limited resources can then be directed to the patients with the greatest need. Patients with a focused care need, such as advance care planning or hospice referral, can be directed to the social worker or case manager. More complicated patients may be referred to a specialty palliative care team.
Several limitations to this study are recognized, including the small sample size of patients meeting criterion for ICU with MOF in the academic center study cohort. The patient data were collected during a transition time when the university hospital moved to a new campus, resulting in an ICU at each campus that housed patients with differing levels of illness severity, which may have contributed to the lower acuity ICU patient observed. Although we advocate the simplicity of the CARING criteria, the NHPCO noncancer hospice guidelines are more complicated, as they incorporates 8 broad categories of chronic illness. The hospice guidelines may not be general knowledge to the hospitalist or other primary providers. ePrognosis (
CONCLUSION
The CARING criteria are a simple, practical prognostic tool predictive of death within 1 year that has been validated in a broad population of hospitalized patients. The criteria hold up in a younger, healthier population that is more diverse by age, gender, and ethnicity than the VA population. With ready access to critical prognostic information on hospital admission, clinicians will be better informed to make decisions that are aligned with their patients' values, preferences, and goals of care.
Disclosure
Nothing to report.
- , , . Predicting death: an empirical evaluation of predictive tools for mortality. Arch Intern Med. 2011;171:1721–1726.
- , . Extent and determinants of error in physicians' prognoses in terminally ill patients: prospective cohort study. West J Med. 2000;172:310–313.
- , , , et al. A systematic review of physicians' survival predictions in terminally ill cancer patients. BMJ. 2003;327:195–198.
- , . Attitude and self‐reported practice regarding prognostication in a national sample of internists. Arch Intern Med. 1998;158:2389–2395.
- , , , et al. Discussing prognosis: balancing hope and realism. Cancer J. 2010;16:461–466.
- , , , . Acute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today's critically ill patients. Crit Care Med. 2006;34:1297–1310.
- , , , , . SAPS 3 admission score: an external validation in a general intensive care population. Intensive Care Med. 2008;34:1873–1877.
- , , , , , . Prospective validation of the intensive care unit admission Mortality Probability Model (MPM0‐III). Crit Care Med. 2009;37:1619–1623.
- , , , , , . A practical tool to identify patients who may benefit from a palliative approach: the CARING criteria. J Pain Symptom Manage. 2006;31:285–292.
- , , , et al. Prediction of survival for older hospitalized patients: the HELP survival model. J Am Geriatr Soc. 2000;48:S16–S24.
- , , , et al. Development and validation of a multidimensional prognostic index for one‐year mortality from comprehensive geriatric assessment in hospitalized older patients. Rejuvenation Res. 2008;11:151–161.
- , , , et al. Burden of illness score for elderly persons: risk adjustment incorporating the cumulative impact of diseases, physiologic abnormalities, and functional impairments. Med Care. 2003;41:70–83.
- , , , et al. The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to understand prognoses and preferences for outcomes and risks of treatments. Ann Intern Med. 1995;122:191–203.
Prognostication continues to be a challenge to the clinician despite over 100 prognostic indices that have been developed during the past few decades to inform clinical practice and medical decision making.[1] Physicians are not accurate in prognostication of patients' risk of death and tend to overestimate survival.[2, 3] In addition, many physicians do not feel comfortable offering a prognosis to patients, despite patients' wishes to be informed.[4, 5] Regardless of the prevalence in the literature and value in improving physicians' prognostic accuracy, prognostic indices of survival are not regularly utilized in the hospital setting. Prognostic tools available for providers are often complicated and may require data about patients that are not readily available.[6, 7, 8] Prognostic indices may be too specific to a patient population, too difficult to remember, or too time consuming to use. A simple, rapid, and practical prognostic index is important in the hospital setting to assist in identifying patients at high risk of death so that primary palliative interventions can be incorporated into the plan of care early in the hospital stay. Patient and family education, advance care planning, formulating the plan of care based on patientfamily goals, and improved resource utilization could be better executed by more accurate risk of death prediction on hospital admission.
The CARING criteria are the only prognostic index to our knowledge that evaluates a patient's risk of death in the next year, with information readily available at the time of hospital admission (Table 1).[9] The CARING criteria are a unique prognostic tool: (1) CARING is a mnemonic acronym, making it more user friendly to the clinician. (2) The 5 prognostic indicators are readily available from the patient's chart on admission; gathering further data by patient or caretaker interviews or by obtaining laboratory data is not needed. (3) The timing for application of the tool on admission to the hospital is an ideal opportunity to intervene and introduce palliative interventions early on the hospital stay. The CARING criteria were developed and validated in a Veteran's Administration hospital setting by Fischer et al.[9] We sought to validate the CARING criteria in a broader patient populationmedical and surgical patients from a tertiary referral university hospital setting and a safety‐net hospital setting.
METHODS
Study Design
This study was a retrospective observational cohort study. The study was approved by the Colorado Multiple Institutional Review Board and the University of Colorado Hospital Research Review Committee.
Study Purpose
To validate the CARING criteria in a tertiary referral university hospital (University of Colorado Hospital [UCH]) and safety‐net hospital (Denver Health and Hospitals [DHH]) setting using similar methodology to that employed by the original CARING criteria study.[9]
Study Setting/Population
All adults (18 years of age) admitted as inpatients to the medical and surgical services of internal medicine, hospitalist, pulmonary, cardiology, hematology/oncology, hepatology, surgery, intensive care unit, and intermediary care unit at UCH and DHH during the study period of July 2005 through August 2005. The only exclusion criteria were those patients who were prisoners or pregnant. Administrative admission data from July 2005 to August 2005 were used to identify names of all persons admitted to the medicine and surgical services of the study hospitals during the specified time period.
The 2 study hospitals, UCH and DHH, provide a range of patients who vary in ethnicity, socioeconomic status, and medical illness. This variability allows for greater generalizability of the results. Both hospitals are affiliated with the University of Colorado School of Medicine internal medicine residency training program and are located in Denver, Colorado.
At the time of the study, UCH was a licensed 550‐bed tertiary referral, academic hospital serving the Denver metropolitan area and the Rocky Mountain region as a specialty care and referral center. DHH was a 398‐bed, academic, safety‐net hospital serving primarily the Denver metropolitan area. DHH provides 42% of the care for the uninsured in Denver and 26% of the uninsured care for the state of Colorado.
Measures
The CARING criteria were developed and validated in a Veteran's Administration (VA) hospital setting by Fischer et al.[9] The purpose of the CARING criteria is to identify patients, at the time of hospital admission, who are at higher risk of death in the following year. The prognostic index uses 5 predictors that can be abstracted from the chart at time of admission. The CARING criteria were developed a priori, and patients were evaluated using only the medical data available at the time of admission. The criteria include items that are already part of the routine physician admission notes and do not require additional data collection or assessments. The criteria include: C=primary diagnosis of cancer, A=2 admissions to the hospital for a chronic illness within the last year; R=resident in a nursing home; I=intensive care unit (ICU) admission with multiorgan failure (MOF), NG=noncancer hospice guidelines (meeting 2 of the National Hospice and Palliative Care Organization's [NHPCO] guidelines).
Patients were identified using name, date of birth, social security number, address, and phone number. This identifying information was then used for tracing death records 1 year after hospital admission.
Mortality at 1 year following the index hospitalization was the primary end point. To minimize missing data and the number of subjects lost to follow‐up, 3 determinants of mortality were used. First, electronic medical records of the 2 participating hospitals and their outpatient clinics were reviewed to determine if a follow‐up appointment had occurred past the study's end point of 1 year (August 2006). For those without a confirmed follow‐up visit, death records from the Colorado Department of Public Health and Vital Records were obtained. For those patients residing outside of Colorado or whose mortality status was still unclear, the National Death Index was accessed.
Medical Record Review
Medical records for all study participants were reviewed by J.Y. (UCH) and B.C. (DHH). Data collection was completed using direct data entry into a Microsoft Access (Microsoft Corp., Redmond, WA) database utilizing a data entry form linked with the database table. This form utilized skip patterns and input masks to ensure quality of data entry and minimize missing or invalid data. Inter‐rater reliability was assessed by an independent rereview (S.F.) of 5% of the total charts. Demographic variables were collected using hospital administrative data. These included personal identifiers of the participants for purposes of mortality follow‐up. Clinical data including the 5 CARING variables and additional descriptive variables were abstracted from the paper hospital chart and the electronic record of the chart (together these constitute the medical record).
Death Follow‐up
A search of Colorado death records was conducted in February 2011 for all subjects. Death records were used to determine mortality and time to death from the index hospitalization. The National Death Index was then searched for any subjects without or record of death in Colorado.
Analysis
All analyses were conducted using the statistical application software SAS for Windows version 9.3 (SAS Institute, Cary, NC). Simple frequencies and means ( standard deviation) were used to describe the baseline characteristics. Multiple logistic regression models were used to model 1‐year mortality. The models were fitted using all of the CARING variables and age. As the aim of the study was to validate the CARING criteria, the variables for the models were selected a priori based on the original index. Two hospital cohorts (DHH and UCH) were modeled separately and as a combined sample. Kaplan‐Meier survival analysis was conducted to compare those subjects who met 1 of the CARING criteria with those who did not through the entire period of mortality follow‐up (20052011). Finally, using the probabilities from the logistic regression models, we again developed a scoring rule appropriate for a non‐VA setting to allow clinicians to easily identify patient risk for 1‐year mortality at the time of hospital admission.
RESULTS
There were a total of 1064 patients admitted to the medical and surgical services during the study period568 patients at DHH and 496 patients at UCH. Sample characteristics of each individual hospital cohort and the entire combined study cohort are detailed in Table 2. Overall, slightly over half the population were male, with a mean age of 50 years, and the ethnic breakdown roughly reflects the population in Denver. A total of 36.5% (n=388) of the study population met 1 of the CARING criteria, and 12.6% (n=134 among 1063 excluding 1 without an admit date) died within 1 year of the index hospitalization. These were younger and healthier patients compared to the VA sample used in developing the CARING criteria.
| |
| Renal | Dementia |
| Stop/decline dialysis | Unable to ambulate independently |
| Not candidate for transplant | Urinary or fecal incontinence |
| Urine output < 40cc/24 hours | Unable to speak with more than single words |
| Creatinine > 8.0 (>6.0 for diabetics) | Unable to bathe independently |
| Creatinine clearance 10cc/min | Unable to dress independently |
| Uremia | Co‐morbid conditions: |
| Persistent serum K + > 7.0 | Aspiration pneumonia |
| Co‐morbid conditions: | Pyelonephritis |
| Cancer CHF | Decubitus ulcer |
| Chronic lung disease AIDS/HIV | Difficulty swallowing or refusal to eat |
| Sepsis Cirrhosis | |
| Cardiac | Pulmonary |
| Ejection fraction < 20% | Dyspnea at rest |
| Symptomatic with diuretics and vasodilators | FEV1 < 30% |
| Not candidate for transplant | Frequent ER or hospital admits for pulmonary infections or respiratory distress |
| History of cardiac arrest | Cor pulmonale or right heart failure |
| History of syncope | 02 sat < 88% on 02 |
| Systolic BP < 120mmHG | PC02 > 50 |
| CVA cardiac origin | Resting tachycardia > 100/min |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| Liver | Stroke/CVA |
| End stage cirrhosis | Coma at onset |
| Not candidate for transplant | Coma >3 days |
| Protime > 5sec and albumin <2.5 | Limb paralysis |
| Ascites unresponsive to treatment | Urinary/fecal incontinence |
| Hepatorenal syndrome | Impaired sitting balance |
| Hepatic encephalopathy | Karnofsky < 50% |
| Spontaneous bacterial peritonitis | Recurrent aspiration |
| Recurrent variceal bleed | Age > 70 |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| HIV/AIDS | Neuromuscular |
| Persistent decline in function | Diminished respiratory function |
| Chronic diarrhea 1 year | Chosen not to receive BiPAP/vent |
| Decision to stop treatment | Difficulty swallowing |
| CNS lymphoma | Diminished functional status |
| MAC‐untreated | Incontinence |
| Systemic lymphoma | Co‐morbid conditions as listed in Renal |
| Dilated cardiomyopathy | |
| CD4 < 25 with disease progression | |
| Viral load > 100,000 | |
| Safety‐Net Hospital Cohort, N=568 | Academic Center Cohort, N=496 | Study Cohort,N=1064 | Original CARING Cohort, N=8739 | |
|---|---|---|---|---|
| ||||
| Mean age ( SD), y | 47.8 (16.5) | 54.4 (17.5) | 50.9 (17.3) | 63 (13) |
| Male gender | 59.5% (338) | 50.1% (248) | 55.1% (586) | 98% (856) |
| Ethnicity | ||||
| African American | 14.1% (80) | 13.5% (65) | 13.8% (145) | 13% (114) |
| Asian | 0.4% (2) | 1.5% (7) | 0.9% (9) | Not reported |
| Caucasian | 41.7% (237) | 66.3% (318) | 53.0 % (555) | 69% (602) |
| Latino | 41.9% (238) | 9.6% (46) | 27.1% (284) | 8% (70) |
| Native American | 0.5% (3) | 0.4% (2) | 0.5% (5) | Not reported |
| Other | 0.5% (3) | 0.6% (3) | 0.6% (6) | 10% (87) |
| Unknown | 0.9% (5) | 8.1% (39) | 4.2% (44) | Not reported |
| CARING criteria | ||||
| Cancer | 6.2% (35) | 19.4% (96) | 12.3% (131) | 23% (201) |
| Admissions to the hospital 2 in past year | 13.6% (77) | 42.7% (212) | 27.2% (289) | 36% (314) |
| Resident in a nursing home | 1.8% (10) | 3.4% (17) | 2.5% (27) | 3% (26) |
| ICU with MOF | 3.7% (21) | 1.2% (6) | 2.5% (27) | 2% (17) |
| NHPCO (2) noncancer guidelines | 1.6% (9) | 5.9% (29) | 3.6% (38) | 8% (70) |
Reliability testing demonstrated excellent inter‐rater reliability. Kappa for each criterion is as follows: (1) primary diagnosis of cancer=1.0, (2) 2 admissions to the hospital in the past year=0.91, (3) resident in a nursing home=1.0, (4) ICU admission with MOF=1.0, and (5) 2 noncancer hospice guidelines=0.78.
This study aimed to validate the CARING criteria9; therefore, all original individual CARING criterion were included in the validation logistic regression models. The 1 exception to this was in the university hospital study cohort, where the ICU criterion was excluded from the model due to small sample size and quasiseparation in the model. The model results are presented in Table 3 for the individual hospitals and combined study cohort.
| Safety Net Hospital Cohort, C Index=0.76 | Academic Center Cohort, C Index=0.76 | Combined Hospital Cohort, C Index=0.79 | ||||
|---|---|---|---|---|---|---|
| Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | |
| ||||||
| Cancer | 1.92 | 6.85 (2.83‐16.59)a | 1.85 | 6.36 (3.54‐11.41)a | 1.98 | 7.23 (4.45‐11.75)a |
| Admissions to the hospital 2 in past year | 0.55 | 1.74 (0.76‐3.97) | 0.14 | 0.87 (0.51‐1.49) | 0.20 | 1.22 (0.78‐1.91) |
| Resident in a nursing home | 0.49 | 0.61 (0.06‐6.56) | 0.27 | 1.31 (0.37‐4.66) | 0.09 | 1.09 (0.36‐3.32) |
| ICU with MOF | 1.85 | 6.34 (2.0219.90)a | 1.94 | 6.97 (2.75‐17.68)a | ||
| NHPCO (2) noncancer guidelines | 3.04 | 20.86 (4.25102.32)a | 2.62 | 13.73 (5.86‐32.15)a | 2.74 | 15.55 (7.2833.23)a |
| Ageb | 0.38 | 1.46 (1.05‐2.03)a | 0.45 | 1.56 (1.23‐1.98)a | 0.47 | 1.60 (1.32‐1.93)a |
In the safety‐net hospital, admission to the hospital with a primary diagnosis related to cancer, 2 noncancer hospice guidelines, ICU admission with MOF, and age by category all were significant predictors of 1‐year mortality. In the university hospital cohort, primary diagnosis of cancer, 2 noncancer hospice guidelines, and age by category were predictive of 1‐year mortality. Finally, in the entire study cohort, primary diagnosis of cancer, ICU with MOF, 2 noncancer hospice guidelines, and age were all predictive of 1‐year mortality. Parameter estimates were similar in 3 of the criteria compared to the VA setting. Differences in patient characteristics may have caused the differences in the estimates. Gender was additionally tested but not significant in any model. One‐year survival was significantly lower for those who met 1 of the CARING criteria versus those who did not (Figure 1).
Based on the framework from the original CARING criteria analysis, a scoring rule was developed using the regression results of this validation cohort. To predict a high probability of 1‐year mortality, sensitivity was set to 58% and specificity was set at 86% (error rate=17%). Medium to high probability was set with a sensitivity of 73% and specificity of 72% (error rate=28%). The coefficients from the regression model of the entire study cohort were converted to scores for each of the CARING criteria. The scores are as follows: 0.5 points for admission from a nursing home, 1 point for 2 hospital admissions in the past year for a chronic illness, 10 points for primary diagnosis of cancer, 10 points for ICU admission with MOF, and 14 points for 2 noncancer hospice guidelines. For every age category increase, 2 points are assigned so that 0 points for age <55 years, 2 points for ages 56 to 65 years, 4 points for ages 66 to 75 years, and 6 points for >75 years. Points for individual risk factors were proportional to s (ie, log odds) in the logistic regression model for death at 1 year. Although no linear transformation exists between s and probabilities (of death at 1 year), the aggregated points for combinations of risk factors shown in Table 4 follow the probabilities in an approximately linear fashion, so that different degrees of risk of death can be represented contiguously (as highlighted by differently shaded regions in the scoring matrix) (Table 4). The scoring matrix allows for quick identification for patients at high risk for 1‐year mortality. In this non‐VA setting with healthier patients, low risk is defined at a lower probability threshold (0.1) compared to the VA setting (0.175).
| CARING Criteria Components | |||||||
|---|---|---|---|---|---|---|---|
| None | Resident in a Nursing Home | Admitted to the Hospital 2 Times in the Past Year | Resident in a Nursing Home Admitted to the Hospital 2 Times in the Past Year | Primary Diagnosis of Cancer | ICU Admission With MOF | Noncancer Hospice Guidelines | |
| |||||||
| Age | |||||||
| 55 years | 0 | 0.5 | 1 | 1.5 | 10 | ||
| 5565 years | 2 | 2.5 | 3 | 3.5 | 10 | ||
| 6675 years | 4 | 4.5 | 5 | 5.5 | 10 | ||
| >75 years | 6 | 6.5 | 7 | 7.5 | 10 | ||
| Risk | |||||||
| Low | 3.5 | Probability<0.1 | |||||
| Medium | 46.5 | 0.1probability <0.175 | |||||
| High | 7 | Probability0.175 | |||||
DISCUSSION
The CARING criteria are a practical prognostic tool that can be easily and rapidly applied to patients admitted to the hospital to estimate risk of death in 1 year, with the goal of identifying patients who may benefit most from incorporating palliative interventions into their plan of care. This study validated the CARING criteria in a tertiary referral university hospital and safety‐net hospital setting, demonstrating applicability in a much broader population than the VA hospital of the original CARING criteria study. The population studied represented a younger population by over 10 years, a more equitable proportion of males to females, a broader ethnic diversity, and lower 1‐year deaths rates than the original study. Despite the broader representation of the population, the significance of each of the individual CARING criterion was maintained except for 2 hospital admissions in the past year for a chronic illness (admission from a nursing home did not meet significance in either study as a sole criterion). As with the original study, meeting 2 of the NHPCO noncancer hospice guidelines demonstrated the highest risk of 1‐year mortality following index hospitalization, followed by primary diagnosis of cancer and ICU admission with MOF. Advancing age, also similar to the original study, conferred increased risk across the criterion.
Hospitalists could be an effective target for utilizing the CARING criteria because they are frequently the first‐line providers in the hospital setting. With the national shortage of palliative care specialists, hospitalists need to be able to identify when a patient has a limited life expectancy so they will be better equipped to make clinical decisions that are aligned with their patients' values, preferences, and goals of care. With the realization that not addressing advance care planning and patient goals of care may be considered medical errors, primary palliative care skills become alarmingly more important as priorities for hospitalists to obtain and feel comfortable using in daily practice.
The CARING criteria are directly applicable to patients who are seen by hospitalists. Other prognostic indices have focused on select patient populations, such as the elderly,[10, 11, 12] require collection of data that are not readily available on admission or would not otherwise be obtained,[10, 13] or apply to patients post‐hospital discharge, thereby missing the opportunity to make an impact earlier in the disease trajectory and incorporate palliative care into the hospital plan of care when key discussions about goals of care and preferences should be encouraged.
Additionally, the CARING criteria could easily be incorporated as a trigger for palliative care consults on hospital admission. Palliative care consults tend to happen late in a hospital stay, limiting the effectiveness of the palliative care team. A trigger system for hospitalists and other primary providers on hospital admission would lend to more effective timing of palliative measures being incorporated into the plan of care. Palliative care consults would not only be initiated earlier, but could be targeted for the more complex and sick patients with the highest risk of death in the next year.
In the time‐pressured environment, the presence of any 1 of the CARING criteria can act as a trigger to begin incorporating primary palliative care measures into the plan of care. The admitting hospitalist provider (ie, physician, nurse practitioner, physician assistant) could access the CARING criteria through an electronic health record prompt when admitting patients. When a more detailed assessment of mortality risk is helpful, the hospitalist can use the scoring matrix, which combines age with the individual criterion to calculate patients at medium or high risk of death within 1 year. Limited resources can then be directed to the patients with the greatest need. Patients with a focused care need, such as advance care planning or hospice referral, can be directed to the social worker or case manager. More complicated patients may be referred to a specialty palliative care team.
Several limitations to this study are recognized, including the small sample size of patients meeting criterion for ICU with MOF in the academic center study cohort. The patient data were collected during a transition time when the university hospital moved to a new campus, resulting in an ICU at each campus that housed patients with differing levels of illness severity, which may have contributed to the lower acuity ICU patient observed. Although we advocate the simplicity of the CARING criteria, the NHPCO noncancer hospice guidelines are more complicated, as they incorporates 8 broad categories of chronic illness. The hospice guidelines may not be general knowledge to the hospitalist or other primary providers. ePrognosis (
CONCLUSION
The CARING criteria are a simple, practical prognostic tool predictive of death within 1 year that has been validated in a broad population of hospitalized patients. The criteria hold up in a younger, healthier population that is more diverse by age, gender, and ethnicity than the VA population. With ready access to critical prognostic information on hospital admission, clinicians will be better informed to make decisions that are aligned with their patients' values, preferences, and goals of care.
Disclosure
Nothing to report.
Prognostication continues to be a challenge to the clinician despite over 100 prognostic indices that have been developed during the past few decades to inform clinical practice and medical decision making.[1] Physicians are not accurate in prognostication of patients' risk of death and tend to overestimate survival.[2, 3] In addition, many physicians do not feel comfortable offering a prognosis to patients, despite patients' wishes to be informed.[4, 5] Regardless of the prevalence in the literature and value in improving physicians' prognostic accuracy, prognostic indices of survival are not regularly utilized in the hospital setting. Prognostic tools available for providers are often complicated and may require data about patients that are not readily available.[6, 7, 8] Prognostic indices may be too specific to a patient population, too difficult to remember, or too time consuming to use. A simple, rapid, and practical prognostic index is important in the hospital setting to assist in identifying patients at high risk of death so that primary palliative interventions can be incorporated into the plan of care early in the hospital stay. Patient and family education, advance care planning, formulating the plan of care based on patientfamily goals, and improved resource utilization could be better executed by more accurate risk of death prediction on hospital admission.
The CARING criteria are the only prognostic index to our knowledge that evaluates a patient's risk of death in the next year, with information readily available at the time of hospital admission (Table 1).[9] The CARING criteria are a unique prognostic tool: (1) CARING is a mnemonic acronym, making it more user friendly to the clinician. (2) The 5 prognostic indicators are readily available from the patient's chart on admission; gathering further data by patient or caretaker interviews or by obtaining laboratory data is not needed. (3) The timing for application of the tool on admission to the hospital is an ideal opportunity to intervene and introduce palliative interventions early on the hospital stay. The CARING criteria were developed and validated in a Veteran's Administration hospital setting by Fischer et al.[9] We sought to validate the CARING criteria in a broader patient populationmedical and surgical patients from a tertiary referral university hospital setting and a safety‐net hospital setting.
METHODS
Study Design
This study was a retrospective observational cohort study. The study was approved by the Colorado Multiple Institutional Review Board and the University of Colorado Hospital Research Review Committee.
Study Purpose
To validate the CARING criteria in a tertiary referral university hospital (University of Colorado Hospital [UCH]) and safety‐net hospital (Denver Health and Hospitals [DHH]) setting using similar methodology to that employed by the original CARING criteria study.[9]
Study Setting/Population
All adults (18 years of age) admitted as inpatients to the medical and surgical services of internal medicine, hospitalist, pulmonary, cardiology, hematology/oncology, hepatology, surgery, intensive care unit, and intermediary care unit at UCH and DHH during the study period of July 2005 through August 2005. The only exclusion criteria were those patients who were prisoners or pregnant. Administrative admission data from July 2005 to August 2005 were used to identify names of all persons admitted to the medicine and surgical services of the study hospitals during the specified time period.
The 2 study hospitals, UCH and DHH, provide a range of patients who vary in ethnicity, socioeconomic status, and medical illness. This variability allows for greater generalizability of the results. Both hospitals are affiliated with the University of Colorado School of Medicine internal medicine residency training program and are located in Denver, Colorado.
At the time of the study, UCH was a licensed 550‐bed tertiary referral, academic hospital serving the Denver metropolitan area and the Rocky Mountain region as a specialty care and referral center. DHH was a 398‐bed, academic, safety‐net hospital serving primarily the Denver metropolitan area. DHH provides 42% of the care for the uninsured in Denver and 26% of the uninsured care for the state of Colorado.
Measures
The CARING criteria were developed and validated in a Veteran's Administration (VA) hospital setting by Fischer et al.[9] The purpose of the CARING criteria is to identify patients, at the time of hospital admission, who are at higher risk of death in the following year. The prognostic index uses 5 predictors that can be abstracted from the chart at time of admission. The CARING criteria were developed a priori, and patients were evaluated using only the medical data available at the time of admission. The criteria include items that are already part of the routine physician admission notes and do not require additional data collection or assessments. The criteria include: C=primary diagnosis of cancer, A=2 admissions to the hospital for a chronic illness within the last year; R=resident in a nursing home; I=intensive care unit (ICU) admission with multiorgan failure (MOF), NG=noncancer hospice guidelines (meeting 2 of the National Hospice and Palliative Care Organization's [NHPCO] guidelines).
Patients were identified using name, date of birth, social security number, address, and phone number. This identifying information was then used for tracing death records 1 year after hospital admission.
Mortality at 1 year following the index hospitalization was the primary end point. To minimize missing data and the number of subjects lost to follow‐up, 3 determinants of mortality were used. First, electronic medical records of the 2 participating hospitals and their outpatient clinics were reviewed to determine if a follow‐up appointment had occurred past the study's end point of 1 year (August 2006). For those without a confirmed follow‐up visit, death records from the Colorado Department of Public Health and Vital Records were obtained. For those patients residing outside of Colorado or whose mortality status was still unclear, the National Death Index was accessed.
Medical Record Review
Medical records for all study participants were reviewed by J.Y. (UCH) and B.C. (DHH). Data collection was completed using direct data entry into a Microsoft Access (Microsoft Corp., Redmond, WA) database utilizing a data entry form linked with the database table. This form utilized skip patterns and input masks to ensure quality of data entry and minimize missing or invalid data. Inter‐rater reliability was assessed by an independent rereview (S.F.) of 5% of the total charts. Demographic variables were collected using hospital administrative data. These included personal identifiers of the participants for purposes of mortality follow‐up. Clinical data including the 5 CARING variables and additional descriptive variables were abstracted from the paper hospital chart and the electronic record of the chart (together these constitute the medical record).
Death Follow‐up
A search of Colorado death records was conducted in February 2011 for all subjects. Death records were used to determine mortality and time to death from the index hospitalization. The National Death Index was then searched for any subjects without or record of death in Colorado.
Analysis
All analyses were conducted using the statistical application software SAS for Windows version 9.3 (SAS Institute, Cary, NC). Simple frequencies and means ( standard deviation) were used to describe the baseline characteristics. Multiple logistic regression models were used to model 1‐year mortality. The models were fitted using all of the CARING variables and age. As the aim of the study was to validate the CARING criteria, the variables for the models were selected a priori based on the original index. Two hospital cohorts (DHH and UCH) were modeled separately and as a combined sample. Kaplan‐Meier survival analysis was conducted to compare those subjects who met 1 of the CARING criteria with those who did not through the entire period of mortality follow‐up (20052011). Finally, using the probabilities from the logistic regression models, we again developed a scoring rule appropriate for a non‐VA setting to allow clinicians to easily identify patient risk for 1‐year mortality at the time of hospital admission.
RESULTS
There were a total of 1064 patients admitted to the medical and surgical services during the study period568 patients at DHH and 496 patients at UCH. Sample characteristics of each individual hospital cohort and the entire combined study cohort are detailed in Table 2. Overall, slightly over half the population were male, with a mean age of 50 years, and the ethnic breakdown roughly reflects the population in Denver. A total of 36.5% (n=388) of the study population met 1 of the CARING criteria, and 12.6% (n=134 among 1063 excluding 1 without an admit date) died within 1 year of the index hospitalization. These were younger and healthier patients compared to the VA sample used in developing the CARING criteria.
| |
| Renal | Dementia |
| Stop/decline dialysis | Unable to ambulate independently |
| Not candidate for transplant | Urinary or fecal incontinence |
| Urine output < 40cc/24 hours | Unable to speak with more than single words |
| Creatinine > 8.0 (>6.0 for diabetics) | Unable to bathe independently |
| Creatinine clearance 10cc/min | Unable to dress independently |
| Uremia | Co‐morbid conditions: |
| Persistent serum K + > 7.0 | Aspiration pneumonia |
| Co‐morbid conditions: | Pyelonephritis |
| Cancer CHF | Decubitus ulcer |
| Chronic lung disease AIDS/HIV | Difficulty swallowing or refusal to eat |
| Sepsis Cirrhosis | |
| Cardiac | Pulmonary |
| Ejection fraction < 20% | Dyspnea at rest |
| Symptomatic with diuretics and vasodilators | FEV1 < 30% |
| Not candidate for transplant | Frequent ER or hospital admits for pulmonary infections or respiratory distress |
| History of cardiac arrest | Cor pulmonale or right heart failure |
| History of syncope | 02 sat < 88% on 02 |
| Systolic BP < 120mmHG | PC02 > 50 |
| CVA cardiac origin | Resting tachycardia > 100/min |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| Liver | Stroke/CVA |
| End stage cirrhosis | Coma at onset |
| Not candidate for transplant | Coma >3 days |
| Protime > 5sec and albumin <2.5 | Limb paralysis |
| Ascites unresponsive to treatment | Urinary/fecal incontinence |
| Hepatorenal syndrome | Impaired sitting balance |
| Hepatic encephalopathy | Karnofsky < 50% |
| Spontaneous bacterial peritonitis | Recurrent aspiration |
| Recurrent variceal bleed | Age > 70 |
| Co‐morbid conditions as listed in Renal | Co‐morbid conditions as listed in Renal |
| HIV/AIDS | Neuromuscular |
| Persistent decline in function | Diminished respiratory function |
| Chronic diarrhea 1 year | Chosen not to receive BiPAP/vent |
| Decision to stop treatment | Difficulty swallowing |
| CNS lymphoma | Diminished functional status |
| MAC‐untreated | Incontinence |
| Systemic lymphoma | Co‐morbid conditions as listed in Renal |
| Dilated cardiomyopathy | |
| CD4 < 25 with disease progression | |
| Viral load > 100,000 | |
| Safety‐Net Hospital Cohort, N=568 | Academic Center Cohort, N=496 | Study Cohort,N=1064 | Original CARING Cohort, N=8739 | |
|---|---|---|---|---|
| ||||
| Mean age ( SD), y | 47.8 (16.5) | 54.4 (17.5) | 50.9 (17.3) | 63 (13) |
| Male gender | 59.5% (338) | 50.1% (248) | 55.1% (586) | 98% (856) |
| Ethnicity | ||||
| African American | 14.1% (80) | 13.5% (65) | 13.8% (145) | 13% (114) |
| Asian | 0.4% (2) | 1.5% (7) | 0.9% (9) | Not reported |
| Caucasian | 41.7% (237) | 66.3% (318) | 53.0 % (555) | 69% (602) |
| Latino | 41.9% (238) | 9.6% (46) | 27.1% (284) | 8% (70) |
| Native American | 0.5% (3) | 0.4% (2) | 0.5% (5) | Not reported |
| Other | 0.5% (3) | 0.6% (3) | 0.6% (6) | 10% (87) |
| Unknown | 0.9% (5) | 8.1% (39) | 4.2% (44) | Not reported |
| CARING criteria | ||||
| Cancer | 6.2% (35) | 19.4% (96) | 12.3% (131) | 23% (201) |
| Admissions to the hospital 2 in past year | 13.6% (77) | 42.7% (212) | 27.2% (289) | 36% (314) |
| Resident in a nursing home | 1.8% (10) | 3.4% (17) | 2.5% (27) | 3% (26) |
| ICU with MOF | 3.7% (21) | 1.2% (6) | 2.5% (27) | 2% (17) |
| NHPCO (2) noncancer guidelines | 1.6% (9) | 5.9% (29) | 3.6% (38) | 8% (70) |
Reliability testing demonstrated excellent inter‐rater reliability. Kappa for each criterion is as follows: (1) primary diagnosis of cancer=1.0, (2) 2 admissions to the hospital in the past year=0.91, (3) resident in a nursing home=1.0, (4) ICU admission with MOF=1.0, and (5) 2 noncancer hospice guidelines=0.78.
This study aimed to validate the CARING criteria9; therefore, all original individual CARING criterion were included in the validation logistic regression models. The 1 exception to this was in the university hospital study cohort, where the ICU criterion was excluded from the model due to small sample size and quasiseparation in the model. The model results are presented in Table 3 for the individual hospitals and combined study cohort.
| Safety Net Hospital Cohort, C Index=0.76 | Academic Center Cohort, C Index=0.76 | Combined Hospital Cohort, C Index=0.79 | ||||
|---|---|---|---|---|---|---|
| Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | Estimate | Odds Ratio (95% CI) | |
| ||||||
| Cancer | 1.92 | 6.85 (2.83‐16.59)a | 1.85 | 6.36 (3.54‐11.41)a | 1.98 | 7.23 (4.45‐11.75)a |
| Admissions to the hospital 2 in past year | 0.55 | 1.74 (0.76‐3.97) | 0.14 | 0.87 (0.51‐1.49) | 0.20 | 1.22 (0.78‐1.91) |
| Resident in a nursing home | 0.49 | 0.61 (0.06‐6.56) | 0.27 | 1.31 (0.37‐4.66) | 0.09 | 1.09 (0.36‐3.32) |
| ICU with MOF | 1.85 | 6.34 (2.0219.90)a | 1.94 | 6.97 (2.75‐17.68)a | ||
| NHPCO (2) noncancer guidelines | 3.04 | 20.86 (4.25102.32)a | 2.62 | 13.73 (5.86‐32.15)a | 2.74 | 15.55 (7.2833.23)a |
| Ageb | 0.38 | 1.46 (1.05‐2.03)a | 0.45 | 1.56 (1.23‐1.98)a | 0.47 | 1.60 (1.32‐1.93)a |
In the safety‐net hospital, admission to the hospital with a primary diagnosis related to cancer, 2 noncancer hospice guidelines, ICU admission with MOF, and age by category all were significant predictors of 1‐year mortality. In the university hospital cohort, primary diagnosis of cancer, 2 noncancer hospice guidelines, and age by category were predictive of 1‐year mortality. Finally, in the entire study cohort, primary diagnosis of cancer, ICU with MOF, 2 noncancer hospice guidelines, and age were all predictive of 1‐year mortality. Parameter estimates were similar in 3 of the criteria compared to the VA setting. Differences in patient characteristics may have caused the differences in the estimates. Gender was additionally tested but not significant in any model. One‐year survival was significantly lower for those who met 1 of the CARING criteria versus those who did not (Figure 1).
Based on the framework from the original CARING criteria analysis, a scoring rule was developed using the regression results of this validation cohort. To predict a high probability of 1‐year mortality, sensitivity was set to 58% and specificity was set at 86% (error rate=17%). Medium to high probability was set with a sensitivity of 73% and specificity of 72% (error rate=28%). The coefficients from the regression model of the entire study cohort were converted to scores for each of the CARING criteria. The scores are as follows: 0.5 points for admission from a nursing home, 1 point for 2 hospital admissions in the past year for a chronic illness, 10 points for primary diagnosis of cancer, 10 points for ICU admission with MOF, and 14 points for 2 noncancer hospice guidelines. For every age category increase, 2 points are assigned so that 0 points for age <55 years, 2 points for ages 56 to 65 years, 4 points for ages 66 to 75 years, and 6 points for >75 years. Points for individual risk factors were proportional to s (ie, log odds) in the logistic regression model for death at 1 year. Although no linear transformation exists between s and probabilities (of death at 1 year), the aggregated points for combinations of risk factors shown in Table 4 follow the probabilities in an approximately linear fashion, so that different degrees of risk of death can be represented contiguously (as highlighted by differently shaded regions in the scoring matrix) (Table 4). The scoring matrix allows for quick identification for patients at high risk for 1‐year mortality. In this non‐VA setting with healthier patients, low risk is defined at a lower probability threshold (0.1) compared to the VA setting (0.175).
| CARING Criteria Components | |||||||
|---|---|---|---|---|---|---|---|
| None | Resident in a Nursing Home | Admitted to the Hospital 2 Times in the Past Year | Resident in a Nursing Home Admitted to the Hospital 2 Times in the Past Year | Primary Diagnosis of Cancer | ICU Admission With MOF | Noncancer Hospice Guidelines | |
| |||||||
| Age | |||||||
| 55 years | 0 | 0.5 | 1 | 1.5 | 10 | ||
| 5565 years | 2 | 2.5 | 3 | 3.5 | 10 | ||
| 6675 years | 4 | 4.5 | 5 | 5.5 | 10 | ||
| >75 years | 6 | 6.5 | 7 | 7.5 | 10 | ||
| Risk | |||||||
| Low | 3.5 | Probability<0.1 | |||||
| Medium | 46.5 | 0.1probability <0.175 | |||||
| High | 7 | Probability0.175 | |||||
DISCUSSION
The CARING criteria are a practical prognostic tool that can be easily and rapidly applied to patients admitted to the hospital to estimate risk of death in 1 year, with the goal of identifying patients who may benefit most from incorporating palliative interventions into their plan of care. This study validated the CARING criteria in a tertiary referral university hospital and safety‐net hospital setting, demonstrating applicability in a much broader population than the VA hospital of the original CARING criteria study. The population studied represented a younger population by over 10 years, a more equitable proportion of males to females, a broader ethnic diversity, and lower 1‐year deaths rates than the original study. Despite the broader representation of the population, the significance of each of the individual CARING criterion was maintained except for 2 hospital admissions in the past year for a chronic illness (admission from a nursing home did not meet significance in either study as a sole criterion). As with the original study, meeting 2 of the NHPCO noncancer hospice guidelines demonstrated the highest risk of 1‐year mortality following index hospitalization, followed by primary diagnosis of cancer and ICU admission with MOF. Advancing age, also similar to the original study, conferred increased risk across the criterion.
Hospitalists could be an effective target for utilizing the CARING criteria because they are frequently the first‐line providers in the hospital setting. With the national shortage of palliative care specialists, hospitalists need to be able to identify when a patient has a limited life expectancy so they will be better equipped to make clinical decisions that are aligned with their patients' values, preferences, and goals of care. With the realization that not addressing advance care planning and patient goals of care may be considered medical errors, primary palliative care skills become alarmingly more important as priorities for hospitalists to obtain and feel comfortable using in daily practice.
The CARING criteria are directly applicable to patients who are seen by hospitalists. Other prognostic indices have focused on select patient populations, such as the elderly,[10, 11, 12] require collection of data that are not readily available on admission or would not otherwise be obtained,[10, 13] or apply to patients post‐hospital discharge, thereby missing the opportunity to make an impact earlier in the disease trajectory and incorporate palliative care into the hospital plan of care when key discussions about goals of care and preferences should be encouraged.
Additionally, the CARING criteria could easily be incorporated as a trigger for palliative care consults on hospital admission. Palliative care consults tend to happen late in a hospital stay, limiting the effectiveness of the palliative care team. A trigger system for hospitalists and other primary providers on hospital admission would lend to more effective timing of palliative measures being incorporated into the plan of care. Palliative care consults would not only be initiated earlier, but could be targeted for the more complex and sick patients with the highest risk of death in the next year.
In the time‐pressured environment, the presence of any 1 of the CARING criteria can act as a trigger to begin incorporating primary palliative care measures into the plan of care. The admitting hospitalist provider (ie, physician, nurse practitioner, physician assistant) could access the CARING criteria through an electronic health record prompt when admitting patients. When a more detailed assessment of mortality risk is helpful, the hospitalist can use the scoring matrix, which combines age with the individual criterion to calculate patients at medium or high risk of death within 1 year. Limited resources can then be directed to the patients with the greatest need. Patients with a focused care need, such as advance care planning or hospice referral, can be directed to the social worker or case manager. More complicated patients may be referred to a specialty palliative care team.
Several limitations to this study are recognized, including the small sample size of patients meeting criterion for ICU with MOF in the academic center study cohort. The patient data were collected during a transition time when the university hospital moved to a new campus, resulting in an ICU at each campus that housed patients with differing levels of illness severity, which may have contributed to the lower acuity ICU patient observed. Although we advocate the simplicity of the CARING criteria, the NHPCO noncancer hospice guidelines are more complicated, as they incorporates 8 broad categories of chronic illness. The hospice guidelines may not be general knowledge to the hospitalist or other primary providers. ePrognosis (
CONCLUSION
The CARING criteria are a simple, practical prognostic tool predictive of death within 1 year that has been validated in a broad population of hospitalized patients. The criteria hold up in a younger, healthier population that is more diverse by age, gender, and ethnicity than the VA population. With ready access to critical prognostic information on hospital admission, clinicians will be better informed to make decisions that are aligned with their patients' values, preferences, and goals of care.
Disclosure
Nothing to report.
- , , . Predicting death: an empirical evaluation of predictive tools for mortality. Arch Intern Med. 2011;171:1721–1726.
- , . Extent and determinants of error in physicians' prognoses in terminally ill patients: prospective cohort study. West J Med. 2000;172:310–313.
- , , , et al. A systematic review of physicians' survival predictions in terminally ill cancer patients. BMJ. 2003;327:195–198.
- , . Attitude and self‐reported practice regarding prognostication in a national sample of internists. Arch Intern Med. 1998;158:2389–2395.
- , , , et al. Discussing prognosis: balancing hope and realism. Cancer J. 2010;16:461–466.
- , , , . Acute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today's critically ill patients. Crit Care Med. 2006;34:1297–1310.
- , , , , . SAPS 3 admission score: an external validation in a general intensive care population. Intensive Care Med. 2008;34:1873–1877.
- , , , , , . Prospective validation of the intensive care unit admission Mortality Probability Model (MPM0‐III). Crit Care Med. 2009;37:1619–1623.
- , , , , , . A practical tool to identify patients who may benefit from a palliative approach: the CARING criteria. J Pain Symptom Manage. 2006;31:285–292.
- , , , et al. Prediction of survival for older hospitalized patients: the HELP survival model. J Am Geriatr Soc. 2000;48:S16–S24.
- , , , et al. Development and validation of a multidimensional prognostic index for one‐year mortality from comprehensive geriatric assessment in hospitalized older patients. Rejuvenation Res. 2008;11:151–161.
- , , , et al. Burden of illness score for elderly persons: risk adjustment incorporating the cumulative impact of diseases, physiologic abnormalities, and functional impairments. Med Care. 2003;41:70–83.
- , , , et al. The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to understand prognoses and preferences for outcomes and risks of treatments. Ann Intern Med. 1995;122:191–203.
- , , . Predicting death: an empirical evaluation of predictive tools for mortality. Arch Intern Med. 2011;171:1721–1726.
- , . Extent and determinants of error in physicians' prognoses in terminally ill patients: prospective cohort study. West J Med. 2000;172:310–313.
- , , , et al. A systematic review of physicians' survival predictions in terminally ill cancer patients. BMJ. 2003;327:195–198.
- , . Attitude and self‐reported practice regarding prognostication in a national sample of internists. Arch Intern Med. 1998;158:2389–2395.
- , , , et al. Discussing prognosis: balancing hope and realism. Cancer J. 2010;16:461–466.
- , , , . Acute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today's critically ill patients. Crit Care Med. 2006;34:1297–1310.
- , , , , . SAPS 3 admission score: an external validation in a general intensive care population. Intensive Care Med. 2008;34:1873–1877.
- , , , , , . Prospective validation of the intensive care unit admission Mortality Probability Model (MPM0‐III). Crit Care Med. 2009;37:1619–1623.
- , , , , , . A practical tool to identify patients who may benefit from a palliative approach: the CARING criteria. J Pain Symptom Manage. 2006;31:285–292.
- , , , et al. Prediction of survival for older hospitalized patients: the HELP survival model. J Am Geriatr Soc. 2000;48:S16–S24.
- , , , et al. Development and validation of a multidimensional prognostic index for one‐year mortality from comprehensive geriatric assessment in hospitalized older patients. Rejuvenation Res. 2008;11:151–161.
- , , , et al. Burden of illness score for elderly persons: risk adjustment incorporating the cumulative impact of diseases, physiologic abnormalities, and functional impairments. Med Care. 2003;41:70–83.
- , , , et al. The SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to understand prognoses and preferences for outcomes and risks of treatments. Ann Intern Med. 1995;122:191–203.
© 2013 Society of Hospital Medicine
Patients at Risk for Readmission
Unplanned hospital readmissions are common, costly, and potentially avoidable. Approximately 20% of Medicare patients are readmitted within 30 days of discharge.[1] Readmission rates are estimated to be similarly high in other population subgroups,[2, 3, 4] with approximately 80% of patients[1, 5, 6] readmitted to the original discharging hospital. A recent systematic review suggested that 27% of readmissions may be preventable.[7]
Hospital readmissions have increasingly been viewed as a correctable marker of poor quality care and have been adopted by a number of organizations as quality indicators.[8, 9, 10] As a result, hospitals have important internal and external motivations to address readmissions. Identification of patients at high risk for readmissions may be an important first step toward preventing them. In particular, readmission risk assessment could be used to help providers target the delivery of resource‐intensive transitional care interventions[11, 12, 13, 14] to patients with the greatest needs. Such an approach is appealing because it allows hospitals to focus scarce resources where the impact may be greatest and provides a starting point for organizations struggling to develop robust models of transitional care delivery.
Electronic health records (EHRs) may prove to be an important component of strategies designed to risk stratify patients at the point of care. Algorithms integrated into the EHR that automatically generate risk predictions have the potential to (1) improve provider time efficiency by automating the prediction process, (2) improve consistency of data collection and risk score calculation, (3) increase adoption through improved usability, and (4) provide clinically important information in real‐time to all healthcare team members caring for a hospitalized patient.
We thus sought to derive a predictive model for 30‐day readmissions using data reliably present in our EHR at the time of admission, and integrate this predictive model into our hospital's EHR to create an automated prediction tool that identifies on admission patients at high risk for readmission within 30 days of discharge. In addition, we prospectively validated this model using the 12‐month period after implementation and examined the impact on readmissions.
METHODS
Setting
The University of Pennsylvania Health System (UPHS) includes 3 hospitals, with a combined capacity of over 1500 beds and 70,000 annual admissions. All hospitals currently utilize Sunrise Clinical Manager version 5.5 (Allscripts, Chicago, IL) as their EHR. The study sample included all adult admissions to any of the 3 UPHS hospitals during the study period. Admissions to short procedure, rehabilitation, and hospice units were excluded. The study received expedited approval and a HIPAA waiver from the University of Pennsylvania institutional review board.
Development of Predictive Model
The UPHS Center for Evidence‐based Practice[15, 16] performed a systematic review to identify factors associated with hospital readmission within 30 days of discharge. We then examined the data available from our hospital EHR at the time of admission for those factors identified in the review. Using different threshold values and look‐back periods, we developed and tested 30 candidate prediction models using these variables alone and in combination (Table 1). Prediction models were evaluated using 24 months of historical data between August 1, 2009 and August 1, 2011.
Implementation
An automated readmission risk flag was then integrated into the EHR. Patients classified as being at high risk for readmission with the automated prediction model were flagged in the EHR on admission (Figure 1A). The flag can be double‐clicked to display a separate screen with information relevant to discharge planning including inpatient and emergency department (ED) visits in the prior 12 months, as well as information about the primary team, length of stay, and admitting problem associated with those admissions (Figure 1B). The prediction model was integrated into our EHR using Arden Syntax for Medical Logic Modules.[17] The readmission risk screen involved presenting the provider with a new screen and was thus developed in Microsoft .NET using C# and Windows Forms (Microsoft Corp., Redmond, WA).
The flag was visible on the patient lists of all providers who utilized the EHR. This included but was not limited to nurses, social workers, unit pharmacists, and physicians. At the time of implementation, educational events regarding the readmission risk flag were provided in forums targeting administrators, pharmacists, social workers, and housestaff. Information about the flag and recommendations for use were distributed through emails and broadcast screensaver messages disseminated throughout the inpatient units of the health system. Providers were asked to pay special attention to discharge planning for patients triggering the readmission risk flag, including medication reconciliation by pharmacists for these patients prior to discharge, and arrangement of available home services by social workers.
The risk flag was 1 of 4 classes of interventions developed and endorsed by the health system in its efforts to reduce readmissions. Besides risk stratification, the other classes were: interdisciplinary rounding, patient education, and discharge communication. None of the interventions alone were expected to decrease readmissions, but as all 4 classes of interventions were implemented and performed routinely, the expectation was that they would work in concert to reduce readmissions.
Analysis
The primary outcome was all‐cause hospital readmissions in the healthcare system within 30 days of discharge. Although this outcome is commonly used both in the literature and as a quality metric, significant debate persists as to the appropriateness of this metric.[18] Many of the factors driving 30‐day readmissions may be dependent on factors outside of the discharging hospital's control and it has been argued that nearer‐term, nonelective readmission rates may provide a more meaningful quality metric.[18] Seven‐day unplanned readmissions were thus used as a secondary outcome measure for this study.
Sensitivity, specificity, predictive value, C statistic, F score (the harmonic mean of positive predictive value and sensitivity),[19] and screen‐positive rate were calculated for each of the 30 prediction models evaluated using the historical data. The prediction model with the best balance of F score and screen‐positive rate was selected as the prediction model to be integrated into the EHR. Prospective validation of the selected prediction model was performed using the 12‐month period following implementation of the risk flag (September 2011September 2012).
To assess the impact of the automated prediction model on monthly readmission rate, we used the 24‐month period immediately before and the 12‐month period immediately after implementation of the readmission risk flag. Segmented regression analysis was performed testing for changes in level and slope of readmission rates between preimplementation and postimplementation time periods. This quasiexperimental interrupted time series methodology[20] allows us to control for secular trends in readmission rates and to assess the preimplementation trend (secular trend), the difference in rates immediately before and after the implementation (immediate effect), and the postimplementation change over time (sustained effect). We used Cochrane‐Orcutt estimation[21] to correct for serial autocorrelation.
All analyses were performed using Stata 12.1 software (Stata Corp, College Station, TX).
RESULTS
Predictors of Readmission
Our systematic review of the literature identified several patient and healthcare utilization patterns predictive of 30‐day readmission risk. Utilization factors included length of stay, number of prior admissions, previous 30‐day readmissions, and previous ED visits. Patient characteristics included number of comorbidities, living alone, and payor. Evidence was inconsistent regarding threshold values for these variables.
Many variables readily available in our EHR were either found by the systematic review not to be reliably predictive of 30‐day readmission (including age and gender) or were not readily or reliably available on admission (including length of stay and payor). At the time of implementation, our EHR did not include vital sign or nursing assessment variables, so these were not considered for inclusion in our model.
Of the available variables, 3 were consistently accurate and available in the EHR at the time of patient admission: prior hospital admission, emergency department visit, and 30‐day readmission within UPHS. We then developed 30 candidate prediction models using a combination of these variables, including 1 and 2 prior admissions, ED visits, and 30‐day readmissions in the 6 and 12 months preceding the index visit (Table 1).
Development and Validation
We used 24 months of retrospective data, which included 120,396 discharges with 17,337 thirty‐day readmissions (14.4% 30‐day all‐cause readmission rate) to test the candidate prediction models. A single risk factor, 2 inpatient admissions in the past 12 months, was found to have the best balance of sensitivity (40%), positive predictive value (31%), and proportion of patients flagged (18%) (Table 1).
| Sensitivity | Specificity | C Statistic | PPV | NPV | Screen Positive | F Score | |
|---|---|---|---|---|---|---|---|
| |||||||
| Retrospective Evaluation of Prediction Rules Lookback period: 6 months | |||||||
| Prior Admissions | |||||||
| 1 | 53% | 74% | 0.640 | 26% | 91% | 30% | 0.350 |
| 2 | 32% | 90% | 0.610 | 35% | 89% | 13% | 0.333 |
| 3 | 20% | 96% | 0.578 | 44% | 88% | 7% | 0.274 |
| Prior ED Visits | |||||||
| 1 | 31% | 81% | 0.558 | 21% | 87% | 21% | 0.252 |
| 2 | 13% | 93% | 0.532 | 25% | 87% | 8% | 0.172 |
| 3 | 7% | 97% | 0.519 | 27% | 86% | 4% | 0.111 |
| Prior 30‐day Readmissions | |||||||
| 1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.347 |
| 2 | 21% | 95% | 0.582 | 43% | 88% | 7% | 0.284 |
| 3 | 13% | 98% | 0.555 | 53% | 87% | 4% | 0.208 |
| Combined Rules | |||||||
| Admit1 & ED1 | 22% | 92% | 0.568 | 31% | 88% | 10% | 0.255 |
| Admit2 & ED1 | 15% | 96% | 0.556 | 40% | 87% | 5% | 0.217 |
| Admit1 & 30‐day1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.346 |
| Admit2 & 30‐day1 | 29% | 92% | 0.603 | 37% | 89% | 11% | 0.324 |
| 30‐day1 & ED1 | 17% | 95% | 0.559 | 37% | 87% | 6% | 0.229 |
| 30‐day1 & ED2 | 8% | 98% | 0.527 | 40% | 86% | 3% | 0.132 |
| Lookback period: 12 months | |||||||
| Prior Admission | |||||||
| 1 | 60% | 68% | 0.593 | 24% | 91% | 36% | 0.340 |
| 2a | 40% | 85% | 0.624 | 31% | 89% | 18% | 0.354 |
| 3 | 28% | 92% | 0.600 | 37% | 88% | 11% | 0.318 |
| Prior ED Visit | |||||||
| 1 | 38% | 74% | 0.560 | 20% | 88% | 28% | 0.260 |
| 2 | 20% | 88% | 0.544 | 23% | 87% | 13% | 0.215 |
| 3 | 8% | 96% | 0.523 | 27% | 86% | 4% | 0.126 |
| Prior 30‐day Readmission | |||||||
| 1 | 43% | 84% | 0.630 | 30% | 90% | 20% | 0.353 |
| 2 | 24% | 94% | 0.592 | 41% | 88% | 9% | 0.305 |
| 3 | 11% | 98% | 0.548 | 54% | 87% | 3% | 0.186 |
| Combined Rules | |||||||
| Admit1 & ED1 | 29% | 87% | 0.580 | 27% | 88% | 15% | 0.281 |
| Admit2 & ED1 | 22% | 93% | 0.574 | 34% | 88% | 9% | 0.266 |
| Admit1 & 30‐day1 | 42% | 84% | 0.630 | 30% | 90% | 14% | 0.353 |
| Admit2 & 30‐day1 | 34% | 89% | 0.615 | 34% | 89% | 14% | 0.341 |
| 30‐day1 & ED1 | 21% | 93% | 0.569 | 35% | 88% | 9% | 0.261 |
| 30‐day1 & ED2 | 13% | 96% | 0.545 | 37% | 87% | 5% | 0.187 |
| Prospective Evaluation of Prediction Rule | |||||||
| 30‐Day All‐Cause | 39% | 84% | 0.614 | 30% | 89% | 18% | 0.339 |
Prospective validation of the prediction model was performed using the 12‐month period directly following readmission risk flag implementation. During this period, the 30‐day all‐cause readmission rate was 15.1%. Sensitivity (39%), positive predictive value (30%), and proportion of patients flagged (18%) were consistent with the values derived from the retrospective data, supporting the reproducibility and predictive stability of the chosen risk prediction model (Table 1). The C statistic of the model was also consistent between the retrospective and prospective datasets (0.62 and 0.61, respectively).
Readmission Rates
The mean 30‐day all‐cause readmission rate for the 24‐month period prior to the intervention was 14.4%, whereas the mean for the 12‐month period after the implementation was 15.1%. Thirty‐day all‐cause and 7‐day unplanned monthly readmission rates do not appear to have been impacted by the intervention (Figure 2). There was no evidence for either an immediate or sustained effect (Table 2).
| Hospital | Preimplementation Period | Immediate Effect | Postimplementation Period | P Value Change in Trenda | |||||
|---|---|---|---|---|---|---|---|---|---|
| Monthly % Change in Readmission Rates | P Value | Immediate % Change | P Value | Monthly % Change in Readmission Rates | P Value | ||||
| |||||||||
| 30‐Day All‐Cause Readmission Rates | |||||||||
| Hosp A | 0.023 | Stable | 0.153 | 0.480 | 0.991 | 0.100 | Increasing | 0.044 | 0.134 |
| Hosp B | 0.061 | Increasing | 0.002 | 0.492 | 0.125 | 0.060 | Stable | 0.296 | 0.048 |
| Hosp C | 0.026 | Stable | 0.413 | 0.447 | 0.585 | 0.046 | Stable | 0.629 | 0.476 |
| Health System | 0.032 | Increasing | 0.014 | 0.344 | 0.302 | 0.026 | Stable | 0.499 | 0.881 |
| 7‐Day Unplanned Readmission Rates | |||||||||
| Hosp A | 0.004 | Stable | 0.642 | 0.271 | 0.417 | 0.005 | Stable | 0.891 | 0.967 |
| Hosp B | 0.012 | Stable | 0.201 | 0.298 | 0.489 | 0.038 | Stable | 0.429 | 0.602 |
| Hosp C | 0.008 | Stable | 0.213 | 0.353 | 0.204 | 0.004 | Stable | 0.895 | 0.899 |
| Health System | 0.005 | Stable | 0.358 | 0.003 | 0.990 | 0.010 | Stable | 0.712 | 0.583 |
DISCUSSION
In this proof‐of‐concept study, we demonstrated the feasibility of an automated readmission risk prediction model integrated into a health system's EHR for a mixed population of hospitalized medical and surgical patients. To our knowledge, this is the first study in a general population of hospitalized patients to examine the impact of providing readmission risk assessment on readmission rates. We used a simple prediction model potentially generalizable to EHRs and healthcare populations beyond our own.
Existing risk prediction models for hospital readmission have important limitations and are difficult to implement in clinical practice.[22] Prediction models for hospital readmission are often dependent on retrospective claims data, developed for specific patient populations, and not designed for use early in the course of hospitalization when transitional care interventions can be initiated.[22] In addition, the time required to gather the necessary data and calculate the risk score remains a barrier to the adoption of prediction models in practice. By automating the process of readmission risk prediction, we were able to help integrate risk assessment into the healthcare process across many providers in a large multihospital healthcare organization. This has allowed us to consistently share risk assessment in real time with all members of the inpatient team, facilitating a team‐based approach to discharge planning.[23]
Two prior studies have developed readmission risk prediction models designed to be implemented into the EHR. Amarasingham et al.[24] developed and implemented[25] a heart failure‐specific prediction model based on the 18‐item Tabak mortality score.[26] Bradley et al.[27] studied in a broader population of medicine and surgery patients the predictive ability of a 26‐item score that utilized vital sign, cardiac rhythm, and nursing assessment data. Although EHRs are developing rapidly, currently the majority of EHRs do not support the use of many of the variables used in these models. In addition, both models were complex, raising concerns about generalizability to other healthcare settings and populations.
A distinctive characteristic of our model is its simplicity. We were cognizant of the realities of running a prediction model in a high‐volume production environment and the diminishing returns of adding more variables. We thus favored simplicity at all stages of model development, with the associated belief that complexity could be added with future iterations once feasibility had been established. Finally, we were aware that we were constructing a medical decision support tool rather than a simple classifier.[26] As such, the optimal model was not purely driven by discriminative ability, but also by our subjective assessment of the optimal trade‐off between sensitivity and specificity (the test‐treatment threshold) for such a model.[26] To facilitate model assessment, we thus categorized the potential predictor variables and evaluated the test characteristics of each combination of categorized variables. Although the C statistic of a model using continuous variables will be higher than a model using categorical values, model performance at the chosen trade‐off point is unlikely to be different.
Although the overall predictive ability of our model was fair, we found that it was associated with clinically meaningful differences in readmission rates between those triggering and not triggering the flag. The 30‐day all‐cause readmission rate in the 12‐month prospective sample was 15.1%, yet among those flagged as being at high risk for readmission the readmission rate was 30.4%. Given resource constraints and the need to selectively apply potentially costly care transition interventions, this may in practice translate into a meaningful discriminative ability.
Readmission rates did not change significantly during the study period. A number of plausible reasons for this exist, including: (1) the current model may not exhibit sufficient predictive ability to classify those at high risk or impact the behavior of providers appropriately, (2) those patients classified as high risk of readmission may not be at high risk of readmissions that are preventable, (3) information provided by the model may not yet routinely be used such that it can affect care, or (4) providing readmission risk assessment alone is not sufficient to influence readmission rates, and the other interventions or organizational changes necessary to impact care of those defined as high risk have not yet been implemented or are not yet being performed routinely. If the primary reasons for our results are those outlined in numbers 3 or 4, then readmission rates should improve over time as the risk flag becomes more routinely used, and those interventions necessary to impact readmission rates of those defined as high risk are implemented and performed.
Limitations
There are several limitations of this intervention. First, the prediction model was developed using 30‐day all‐cause readmissions, rather than attempting to identify potentially preventable readmissions. Thirty‐day readmission rates may not be a good proxy for preventable readmissions,[18] and as a consequence, the ability to predict 30‐day readmissions may not ensure that a prediction model is able to predict preventable readmissions. Nonetheless, 30‐day readmission rates remain the most commonly used quality metric.
Second, the impact of the risk flag on provider behavior is uncertain. We did not formally assess how the readmission risk flag was used by healthcare team members. Informal assessment has, however, revealed that the readmission risk flag is gradually being adopted by different members of the care team including unit‐based pharmacists who are using the flag to prioritize the delivery of medication education, social workers who are using the flag to prompt providers to consider higher level services for patients at high risk of readmission, and patient navigators who are using the flag to prioritize follow‐up phone calls. As a result, we hope that the flag will ultimately improve the processes of care for high‐risk patients.
Third, we did not capture readmissions to hospitals outside of our healthcare system and have therefore underestimated the readmission rate in our population. However, our assessment of the effect of the risk flag on readmissions focused on relative readmission rates over time, and the use of the interrupted time series methodology should protect against secular changes in outside hospital readmission rates that were not associated with the intervention.
Fourth, it is possible that the prediction model implemented could be significantly improved by including additional variables or data available during the hospital stay. However, simple classification models using a single variable have repeatedly been shown to have the ability to compete favorably with state‐of‐the‐art multivariable classification models.[28]
Fifth, our study was limited to a single academic health system, and our experience may not be generalizable to smaller healthcare systems with limited EHR systems. However, the simplicity of our prediction model and the integration into a commercial EHR may improve the generalizability of our experience to other healthcare settings. Additionally, partly due to recent policy initiatives, the adoption of integrated EHR systems by hospitals is expected to continue at a rapid rate and become the standard of care within the near future.[29]
CONCLUSION
An automated prediction model was effectively integrated into an existing EHR and was able to identify patients on admission who are at risk for readmission within 30 days of discharge. Future work will aim to further examine the impact of the flag on readmission rates, further refine the prediction model, and gather data on how providers and care teams use the information provided by the flag.
Disclosure
Dr. Umscheid‐s contribution to this project was supported in part by the National Center for Research Resources, Grant UL1RR024134, which is now at the National Center for Advancing Translational Sciences, Grant UL1TR000003. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
- , , . Rehospitalizations among patients in the Medicare fee‐for‐service program. N Engl J Med. 2009;360(14):1418–1428.
- , . Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med. 2000;160(8):1074–1081.
- , , , , . Do older rural and urban veterans experience different rates of unplanned readmission to VA and non‐VA hospitals? J Rural Health. 2009;25(1):62–69.
- , , . Cost, causes and rates of rehospitalization of preterm infants. J Perinatol. 2007;27(10):614–619.
- , , , . Redefining readmission risk factors for general medicine patients. J Hosp Med. 2011;6(2):54–60.
- , , , et al. Incidence and main factors associated with early unplanned hospital readmission among French medical inpatients aged 75 and over admitted through emergency units. Age Ageing. 2008;37(4):416–422.
- , , , , . Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ. 2011;183(7):E391–E402.
- Hospital Quality Alliance. Available at: http://www.hospitalqualityalliance.org/hospitalqualityalliance/qualitymeasures/qualitymeasures.html. Accessed March 6, 2013.
- Institute for Healthcare Improvement. Available at: http://www.ihi.org/explore/Readmissions/Pages/default.aspx. Accessed March 6, 2013.
- Centers for Medicare and Medicaid Services. Available at: http://www.cms.gov/Medicare/Quality‐Initiatives‐Patient‐Assessment‐Instruments/HospitalQualityInits/OutcomeMeasures.html. Accessed March 6, 2013.
- , , , et al. Comprehensive discharge planning and home follow‐up of hospitalized elders: a randomized clinical trial. JAMA. 1999;281(7):613–620.
- , , , , , . Preparing patients and caregivers to participate in care delivered across settings: the Care Transitions Intervention. J Am Geriatr Soc. 2004;52(11):1817–1825.
- , , , , . The importance of transitional care in achieving health reform. Health Aff (Millwood). 2011;30(4):746–754.
- , , , , . Interventions to reduce 30‐day rehospitalization: a systematic review. Ann Intern Med. 2011;155(8):520–528.
- University of Pennsylvania Health System Center for Evidence‐based Practice. Available at: http://www.uphs.upenn.edu/cep/. Accessed March 6, 2013.
- , , . Hospital‐based comparative effectiveness centers: translating research into practice to improve the quality, safety and value of patient care. J Gen Intern Med. 2010;25(12):1352–1355.
- . Writing Arden Syntax Medical Logic Modules. Comput Biol Med. 1994;24(5):331–363.
- , . Thirty‐day readmissions—truth and consequences. N Engl J Med. 2012;366(15):1366–1369.
- . Information Retrieval. 2nd ed. Oxford, UK: Butterworth‐Heinemann; 1979.
- , , , . Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther. 2002;27(4):299–309.
- , . Application of least squares regression to relationships containing auto‐correlated error terms. J Am Stat Assoc. 1949; 44:32–61.
- , , , et al. Risk prediction models for hospital readmission: a systematic review. JAMA. 2011;306(15):1688–1698.
- , , , et al. Core Principles and values of effective team‐based health care. Available at: https://www.nationalahec.org/pdfs/VSRT‐Team‐Based‐Care‐Principles‐Values.pdf. Accessed March 19, 2013.
- , , , et al. An automated model to identify heart failure patients at risk for 30‐day readmission or death using electronic medical record data. Med Care. 2010;48(11):981–988.
- , , , et al. Allocating scarce resources in real‐time to reduce heart failure readmissions: a prospective, controlled study [published online ahead of print July 31, 2013]. BMJ Qual Saf. doi:10.1136/bmjqs‐2013‐001901.
- , . The threshold approach to clinical decision making. N Engl J Med. 1980;302(20):1109–1117.
- , , , , . Identifying patients at increased risk for unplanned readmission. Med Care. 2013;51(9):761–766.
- . Very simple classification rules perform well on most commonly used datasets. Mach Learn. 1993;11(1):63–91.
- . Stimulating the adoption of health information technology. N Engl J Med. 2009;360(15):1477–1479.
Unplanned hospital readmissions are common, costly, and potentially avoidable. Approximately 20% of Medicare patients are readmitted within 30 days of discharge.[1] Readmission rates are estimated to be similarly high in other population subgroups,[2, 3, 4] with approximately 80% of patients[1, 5, 6] readmitted to the original discharging hospital. A recent systematic review suggested that 27% of readmissions may be preventable.[7]
Hospital readmissions have increasingly been viewed as a correctable marker of poor quality care and have been adopted by a number of organizations as quality indicators.[8, 9, 10] As a result, hospitals have important internal and external motivations to address readmissions. Identification of patients at high risk for readmissions may be an important first step toward preventing them. In particular, readmission risk assessment could be used to help providers target the delivery of resource‐intensive transitional care interventions[11, 12, 13, 14] to patients with the greatest needs. Such an approach is appealing because it allows hospitals to focus scarce resources where the impact may be greatest and provides a starting point for organizations struggling to develop robust models of transitional care delivery.
Electronic health records (EHRs) may prove to be an important component of strategies designed to risk stratify patients at the point of care. Algorithms integrated into the EHR that automatically generate risk predictions have the potential to (1) improve provider time efficiency by automating the prediction process, (2) improve consistency of data collection and risk score calculation, (3) increase adoption through improved usability, and (4) provide clinically important information in real‐time to all healthcare team members caring for a hospitalized patient.
We thus sought to derive a predictive model for 30‐day readmissions using data reliably present in our EHR at the time of admission, and integrate this predictive model into our hospital's EHR to create an automated prediction tool that identifies on admission patients at high risk for readmission within 30 days of discharge. In addition, we prospectively validated this model using the 12‐month period after implementation and examined the impact on readmissions.
METHODS
Setting
The University of Pennsylvania Health System (UPHS) includes 3 hospitals, with a combined capacity of over 1500 beds and 70,000 annual admissions. All hospitals currently utilize Sunrise Clinical Manager version 5.5 (Allscripts, Chicago, IL) as their EHR. The study sample included all adult admissions to any of the 3 UPHS hospitals during the study period. Admissions to short procedure, rehabilitation, and hospice units were excluded. The study received expedited approval and a HIPAA waiver from the University of Pennsylvania institutional review board.
Development of Predictive Model
The UPHS Center for Evidence‐based Practice[15, 16] performed a systematic review to identify factors associated with hospital readmission within 30 days of discharge. We then examined the data available from our hospital EHR at the time of admission for those factors identified in the review. Using different threshold values and look‐back periods, we developed and tested 30 candidate prediction models using these variables alone and in combination (Table 1). Prediction models were evaluated using 24 months of historical data between August 1, 2009 and August 1, 2011.
Implementation
An automated readmission risk flag was then integrated into the EHR. Patients classified as being at high risk for readmission with the automated prediction model were flagged in the EHR on admission (Figure 1A). The flag can be double‐clicked to display a separate screen with information relevant to discharge planning including inpatient and emergency department (ED) visits in the prior 12 months, as well as information about the primary team, length of stay, and admitting problem associated with those admissions (Figure 1B). The prediction model was integrated into our EHR using Arden Syntax for Medical Logic Modules.[17] The readmission risk screen involved presenting the provider with a new screen and was thus developed in Microsoft .NET using C# and Windows Forms (Microsoft Corp., Redmond, WA).
The flag was visible on the patient lists of all providers who utilized the EHR. This included but was not limited to nurses, social workers, unit pharmacists, and physicians. At the time of implementation, educational events regarding the readmission risk flag were provided in forums targeting administrators, pharmacists, social workers, and housestaff. Information about the flag and recommendations for use were distributed through emails and broadcast screensaver messages disseminated throughout the inpatient units of the health system. Providers were asked to pay special attention to discharge planning for patients triggering the readmission risk flag, including medication reconciliation by pharmacists for these patients prior to discharge, and arrangement of available home services by social workers.
The risk flag was 1 of 4 classes of interventions developed and endorsed by the health system in its efforts to reduce readmissions. Besides risk stratification, the other classes were: interdisciplinary rounding, patient education, and discharge communication. None of the interventions alone were expected to decrease readmissions, but as all 4 classes of interventions were implemented and performed routinely, the expectation was that they would work in concert to reduce readmissions.
Analysis
The primary outcome was all‐cause hospital readmissions in the healthcare system within 30 days of discharge. Although this outcome is commonly used both in the literature and as a quality metric, significant debate persists as to the appropriateness of this metric.[18] Many of the factors driving 30‐day readmissions may be dependent on factors outside of the discharging hospital's control and it has been argued that nearer‐term, nonelective readmission rates may provide a more meaningful quality metric.[18] Seven‐day unplanned readmissions were thus used as a secondary outcome measure for this study.
Sensitivity, specificity, predictive value, C statistic, F score (the harmonic mean of positive predictive value and sensitivity),[19] and screen‐positive rate were calculated for each of the 30 prediction models evaluated using the historical data. The prediction model with the best balance of F score and screen‐positive rate was selected as the prediction model to be integrated into the EHR. Prospective validation of the selected prediction model was performed using the 12‐month period following implementation of the risk flag (September 2011September 2012).
To assess the impact of the automated prediction model on monthly readmission rate, we used the 24‐month period immediately before and the 12‐month period immediately after implementation of the readmission risk flag. Segmented regression analysis was performed testing for changes in level and slope of readmission rates between preimplementation and postimplementation time periods. This quasiexperimental interrupted time series methodology[20] allows us to control for secular trends in readmission rates and to assess the preimplementation trend (secular trend), the difference in rates immediately before and after the implementation (immediate effect), and the postimplementation change over time (sustained effect). We used Cochrane‐Orcutt estimation[21] to correct for serial autocorrelation.
All analyses were performed using Stata 12.1 software (Stata Corp, College Station, TX).
RESULTS
Predictors of Readmission
Our systematic review of the literature identified several patient and healthcare utilization patterns predictive of 30‐day readmission risk. Utilization factors included length of stay, number of prior admissions, previous 30‐day readmissions, and previous ED visits. Patient characteristics included number of comorbidities, living alone, and payor. Evidence was inconsistent regarding threshold values for these variables.
Many variables readily available in our EHR were either found by the systematic review not to be reliably predictive of 30‐day readmission (including age and gender) or were not readily or reliably available on admission (including length of stay and payor). At the time of implementation, our EHR did not include vital sign or nursing assessment variables, so these were not considered for inclusion in our model.
Of the available variables, 3 were consistently accurate and available in the EHR at the time of patient admission: prior hospital admission, emergency department visit, and 30‐day readmission within UPHS. We then developed 30 candidate prediction models using a combination of these variables, including 1 and 2 prior admissions, ED visits, and 30‐day readmissions in the 6 and 12 months preceding the index visit (Table 1).
Development and Validation
We used 24 months of retrospective data, which included 120,396 discharges with 17,337 thirty‐day readmissions (14.4% 30‐day all‐cause readmission rate) to test the candidate prediction models. A single risk factor, 2 inpatient admissions in the past 12 months, was found to have the best balance of sensitivity (40%), positive predictive value (31%), and proportion of patients flagged (18%) (Table 1).
| Sensitivity | Specificity | C Statistic | PPV | NPV | Screen Positive | F Score | |
|---|---|---|---|---|---|---|---|
| |||||||
| Retrospective Evaluation of Prediction Rules Lookback period: 6 months | |||||||
| Prior Admissions | |||||||
| 1 | 53% | 74% | 0.640 | 26% | 91% | 30% | 0.350 |
| 2 | 32% | 90% | 0.610 | 35% | 89% | 13% | 0.333 |
| 3 | 20% | 96% | 0.578 | 44% | 88% | 7% | 0.274 |
| Prior ED Visits | |||||||
| 1 | 31% | 81% | 0.558 | 21% | 87% | 21% | 0.252 |
| 2 | 13% | 93% | 0.532 | 25% | 87% | 8% | 0.172 |
| 3 | 7% | 97% | 0.519 | 27% | 86% | 4% | 0.111 |
| Prior 30‐day Readmissions | |||||||
| 1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.347 |
| 2 | 21% | 95% | 0.582 | 43% | 88% | 7% | 0.284 |
| 3 | 13% | 98% | 0.555 | 53% | 87% | 4% | 0.208 |
| Combined Rules | |||||||
| Admit1 & ED1 | 22% | 92% | 0.568 | 31% | 88% | 10% | 0.255 |
| Admit2 & ED1 | 15% | 96% | 0.556 | 40% | 87% | 5% | 0.217 |
| Admit1 & 30‐day1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.346 |
| Admit2 & 30‐day1 | 29% | 92% | 0.603 | 37% | 89% | 11% | 0.324 |
| 30‐day1 & ED1 | 17% | 95% | 0.559 | 37% | 87% | 6% | 0.229 |
| 30‐day1 & ED2 | 8% | 98% | 0.527 | 40% | 86% | 3% | 0.132 |
| Lookback period: 12 months | |||||||
| Prior Admission | |||||||
| 1 | 60% | 68% | 0.593 | 24% | 91% | 36% | 0.340 |
| 2a | 40% | 85% | 0.624 | 31% | 89% | 18% | 0.354 |
| 3 | 28% | 92% | 0.600 | 37% | 88% | 11% | 0.318 |
| Prior ED Visit | |||||||
| 1 | 38% | 74% | 0.560 | 20% | 88% | 28% | 0.260 |
| 2 | 20% | 88% | 0.544 | 23% | 87% | 13% | 0.215 |
| 3 | 8% | 96% | 0.523 | 27% | 86% | 4% | 0.126 |
| Prior 30‐day Readmission | |||||||
| 1 | 43% | 84% | 0.630 | 30% | 90% | 20% | 0.353 |
| 2 | 24% | 94% | 0.592 | 41% | 88% | 9% | 0.305 |
| 3 | 11% | 98% | 0.548 | 54% | 87% | 3% | 0.186 |
| Combined Rules | |||||||
| Admit1 & ED1 | 29% | 87% | 0.580 | 27% | 88% | 15% | 0.281 |
| Admit2 & ED1 | 22% | 93% | 0.574 | 34% | 88% | 9% | 0.266 |
| Admit1 & 30‐day1 | 42% | 84% | 0.630 | 30% | 90% | 14% | 0.353 |
| Admit2 & 30‐day1 | 34% | 89% | 0.615 | 34% | 89% | 14% | 0.341 |
| 30‐day1 & ED1 | 21% | 93% | 0.569 | 35% | 88% | 9% | 0.261 |
| 30‐day1 & ED2 | 13% | 96% | 0.545 | 37% | 87% | 5% | 0.187 |
| Prospective Evaluation of Prediction Rule | |||||||
| 30‐Day All‐Cause | 39% | 84% | 0.614 | 30% | 89% | 18% | 0.339 |
Prospective validation of the prediction model was performed using the 12‐month period directly following readmission risk flag implementation. During this period, the 30‐day all‐cause readmission rate was 15.1%. Sensitivity (39%), positive predictive value (30%), and proportion of patients flagged (18%) were consistent with the values derived from the retrospective data, supporting the reproducibility and predictive stability of the chosen risk prediction model (Table 1). The C statistic of the model was also consistent between the retrospective and prospective datasets (0.62 and 0.61, respectively).
Readmission Rates
The mean 30‐day all‐cause readmission rate for the 24‐month period prior to the intervention was 14.4%, whereas the mean for the 12‐month period after the implementation was 15.1%. Thirty‐day all‐cause and 7‐day unplanned monthly readmission rates do not appear to have been impacted by the intervention (Figure 2). There was no evidence for either an immediate or sustained effect (Table 2).
| Hospital | Preimplementation Period | Immediate Effect | Postimplementation Period | P Value Change in Trenda | |||||
|---|---|---|---|---|---|---|---|---|---|
| Monthly % Change in Readmission Rates | P Value | Immediate % Change | P Value | Monthly % Change in Readmission Rates | P Value | ||||
| |||||||||
| 30‐Day All‐Cause Readmission Rates | |||||||||
| Hosp A | 0.023 | Stable | 0.153 | 0.480 | 0.991 | 0.100 | Increasing | 0.044 | 0.134 |
| Hosp B | 0.061 | Increasing | 0.002 | 0.492 | 0.125 | 0.060 | Stable | 0.296 | 0.048 |
| Hosp C | 0.026 | Stable | 0.413 | 0.447 | 0.585 | 0.046 | Stable | 0.629 | 0.476 |
| Health System | 0.032 | Increasing | 0.014 | 0.344 | 0.302 | 0.026 | Stable | 0.499 | 0.881 |
| 7‐Day Unplanned Readmission Rates | |||||||||
| Hosp A | 0.004 | Stable | 0.642 | 0.271 | 0.417 | 0.005 | Stable | 0.891 | 0.967 |
| Hosp B | 0.012 | Stable | 0.201 | 0.298 | 0.489 | 0.038 | Stable | 0.429 | 0.602 |
| Hosp C | 0.008 | Stable | 0.213 | 0.353 | 0.204 | 0.004 | Stable | 0.895 | 0.899 |
| Health System | 0.005 | Stable | 0.358 | 0.003 | 0.990 | 0.010 | Stable | 0.712 | 0.583 |
DISCUSSION
In this proof‐of‐concept study, we demonstrated the feasibility of an automated readmission risk prediction model integrated into a health system's EHR for a mixed population of hospitalized medical and surgical patients. To our knowledge, this is the first study in a general population of hospitalized patients to examine the impact of providing readmission risk assessment on readmission rates. We used a simple prediction model potentially generalizable to EHRs and healthcare populations beyond our own.
Existing risk prediction models for hospital readmission have important limitations and are difficult to implement in clinical practice.[22] Prediction models for hospital readmission are often dependent on retrospective claims data, developed for specific patient populations, and not designed for use early in the course of hospitalization when transitional care interventions can be initiated.[22] In addition, the time required to gather the necessary data and calculate the risk score remains a barrier to the adoption of prediction models in practice. By automating the process of readmission risk prediction, we were able to help integrate risk assessment into the healthcare process across many providers in a large multihospital healthcare organization. This has allowed us to consistently share risk assessment in real time with all members of the inpatient team, facilitating a team‐based approach to discharge planning.[23]
Two prior studies have developed readmission risk prediction models designed to be implemented into the EHR. Amarasingham et al.[24] developed and implemented[25] a heart failure‐specific prediction model based on the 18‐item Tabak mortality score.[26] Bradley et al.[27] studied in a broader population of medicine and surgery patients the predictive ability of a 26‐item score that utilized vital sign, cardiac rhythm, and nursing assessment data. Although EHRs are developing rapidly, currently the majority of EHRs do not support the use of many of the variables used in these models. In addition, both models were complex, raising concerns about generalizability to other healthcare settings and populations.
A distinctive characteristic of our model is its simplicity. We were cognizant of the realities of running a prediction model in a high‐volume production environment and the diminishing returns of adding more variables. We thus favored simplicity at all stages of model development, with the associated belief that complexity could be added with future iterations once feasibility had been established. Finally, we were aware that we were constructing a medical decision support tool rather than a simple classifier.[26] As such, the optimal model was not purely driven by discriminative ability, but also by our subjective assessment of the optimal trade‐off between sensitivity and specificity (the test‐treatment threshold) for such a model.[26] To facilitate model assessment, we thus categorized the potential predictor variables and evaluated the test characteristics of each combination of categorized variables. Although the C statistic of a model using continuous variables will be higher than a model using categorical values, model performance at the chosen trade‐off point is unlikely to be different.
Although the overall predictive ability of our model was fair, we found that it was associated with clinically meaningful differences in readmission rates between those triggering and not triggering the flag. The 30‐day all‐cause readmission rate in the 12‐month prospective sample was 15.1%, yet among those flagged as being at high risk for readmission the readmission rate was 30.4%. Given resource constraints and the need to selectively apply potentially costly care transition interventions, this may in practice translate into a meaningful discriminative ability.
Readmission rates did not change significantly during the study period. A number of plausible reasons for this exist, including: (1) the current model may not exhibit sufficient predictive ability to classify those at high risk or impact the behavior of providers appropriately, (2) those patients classified as high risk of readmission may not be at high risk of readmissions that are preventable, (3) information provided by the model may not yet routinely be used such that it can affect care, or (4) providing readmission risk assessment alone is not sufficient to influence readmission rates, and the other interventions or organizational changes necessary to impact care of those defined as high risk have not yet been implemented or are not yet being performed routinely. If the primary reasons for our results are those outlined in numbers 3 or 4, then readmission rates should improve over time as the risk flag becomes more routinely used, and those interventions necessary to impact readmission rates of those defined as high risk are implemented and performed.
Limitations
There are several limitations of this intervention. First, the prediction model was developed using 30‐day all‐cause readmissions, rather than attempting to identify potentially preventable readmissions. Thirty‐day readmission rates may not be a good proxy for preventable readmissions,[18] and as a consequence, the ability to predict 30‐day readmissions may not ensure that a prediction model is able to predict preventable readmissions. Nonetheless, 30‐day readmission rates remain the most commonly used quality metric.
Second, the impact of the risk flag on provider behavior is uncertain. We did not formally assess how the readmission risk flag was used by healthcare team members. Informal assessment has, however, revealed that the readmission risk flag is gradually being adopted by different members of the care team including unit‐based pharmacists who are using the flag to prioritize the delivery of medication education, social workers who are using the flag to prompt providers to consider higher level services for patients at high risk of readmission, and patient navigators who are using the flag to prioritize follow‐up phone calls. As a result, we hope that the flag will ultimately improve the processes of care for high‐risk patients.
Third, we did not capture readmissions to hospitals outside of our healthcare system and have therefore underestimated the readmission rate in our population. However, our assessment of the effect of the risk flag on readmissions focused on relative readmission rates over time, and the use of the interrupted time series methodology should protect against secular changes in outside hospital readmission rates that were not associated with the intervention.
Fourth, it is possible that the prediction model implemented could be significantly improved by including additional variables or data available during the hospital stay. However, simple classification models using a single variable have repeatedly been shown to have the ability to compete favorably with state‐of‐the‐art multivariable classification models.[28]
Fifth, our study was limited to a single academic health system, and our experience may not be generalizable to smaller healthcare systems with limited EHR systems. However, the simplicity of our prediction model and the integration into a commercial EHR may improve the generalizability of our experience to other healthcare settings. Additionally, partly due to recent policy initiatives, the adoption of integrated EHR systems by hospitals is expected to continue at a rapid rate and become the standard of care within the near future.[29]
CONCLUSION
An automated prediction model was effectively integrated into an existing EHR and was able to identify patients on admission who are at risk for readmission within 30 days of discharge. Future work will aim to further examine the impact of the flag on readmission rates, further refine the prediction model, and gather data on how providers and care teams use the information provided by the flag.
Disclosure
Dr. Umscheid‐s contribution to this project was supported in part by the National Center for Research Resources, Grant UL1RR024134, which is now at the National Center for Advancing Translational Sciences, Grant UL1TR000003. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Unplanned hospital readmissions are common, costly, and potentially avoidable. Approximately 20% of Medicare patients are readmitted within 30 days of discharge.[1] Readmission rates are estimated to be similarly high in other population subgroups,[2, 3, 4] with approximately 80% of patients[1, 5, 6] readmitted to the original discharging hospital. A recent systematic review suggested that 27% of readmissions may be preventable.[7]
Hospital readmissions have increasingly been viewed as a correctable marker of poor quality care and have been adopted by a number of organizations as quality indicators.[8, 9, 10] As a result, hospitals have important internal and external motivations to address readmissions. Identification of patients at high risk for readmissions may be an important first step toward preventing them. In particular, readmission risk assessment could be used to help providers target the delivery of resource‐intensive transitional care interventions[11, 12, 13, 14] to patients with the greatest needs. Such an approach is appealing because it allows hospitals to focus scarce resources where the impact may be greatest and provides a starting point for organizations struggling to develop robust models of transitional care delivery.
Electronic health records (EHRs) may prove to be an important component of strategies designed to risk stratify patients at the point of care. Algorithms integrated into the EHR that automatically generate risk predictions have the potential to (1) improve provider time efficiency by automating the prediction process, (2) improve consistency of data collection and risk score calculation, (3) increase adoption through improved usability, and (4) provide clinically important information in real‐time to all healthcare team members caring for a hospitalized patient.
We thus sought to derive a predictive model for 30‐day readmissions using data reliably present in our EHR at the time of admission, and integrate this predictive model into our hospital's EHR to create an automated prediction tool that identifies on admission patients at high risk for readmission within 30 days of discharge. In addition, we prospectively validated this model using the 12‐month period after implementation and examined the impact on readmissions.
METHODS
Setting
The University of Pennsylvania Health System (UPHS) includes 3 hospitals, with a combined capacity of over 1500 beds and 70,000 annual admissions. All hospitals currently utilize Sunrise Clinical Manager version 5.5 (Allscripts, Chicago, IL) as their EHR. The study sample included all adult admissions to any of the 3 UPHS hospitals during the study period. Admissions to short procedure, rehabilitation, and hospice units were excluded. The study received expedited approval and a HIPAA waiver from the University of Pennsylvania institutional review board.
Development of Predictive Model
The UPHS Center for Evidence‐based Practice[15, 16] performed a systematic review to identify factors associated with hospital readmission within 30 days of discharge. We then examined the data available from our hospital EHR at the time of admission for those factors identified in the review. Using different threshold values and look‐back periods, we developed and tested 30 candidate prediction models using these variables alone and in combination (Table 1). Prediction models were evaluated using 24 months of historical data between August 1, 2009 and August 1, 2011.
Implementation
An automated readmission risk flag was then integrated into the EHR. Patients classified as being at high risk for readmission with the automated prediction model were flagged in the EHR on admission (Figure 1A). The flag can be double‐clicked to display a separate screen with information relevant to discharge planning including inpatient and emergency department (ED) visits in the prior 12 months, as well as information about the primary team, length of stay, and admitting problem associated with those admissions (Figure 1B). The prediction model was integrated into our EHR using Arden Syntax for Medical Logic Modules.[17] The readmission risk screen involved presenting the provider with a new screen and was thus developed in Microsoft .NET using C# and Windows Forms (Microsoft Corp., Redmond, WA).
The flag was visible on the patient lists of all providers who utilized the EHR. This included but was not limited to nurses, social workers, unit pharmacists, and physicians. At the time of implementation, educational events regarding the readmission risk flag were provided in forums targeting administrators, pharmacists, social workers, and housestaff. Information about the flag and recommendations for use were distributed through emails and broadcast screensaver messages disseminated throughout the inpatient units of the health system. Providers were asked to pay special attention to discharge planning for patients triggering the readmission risk flag, including medication reconciliation by pharmacists for these patients prior to discharge, and arrangement of available home services by social workers.
The risk flag was 1 of 4 classes of interventions developed and endorsed by the health system in its efforts to reduce readmissions. Besides risk stratification, the other classes were: interdisciplinary rounding, patient education, and discharge communication. None of the interventions alone were expected to decrease readmissions, but as all 4 classes of interventions were implemented and performed routinely, the expectation was that they would work in concert to reduce readmissions.
Analysis
The primary outcome was all‐cause hospital readmissions in the healthcare system within 30 days of discharge. Although this outcome is commonly used both in the literature and as a quality metric, significant debate persists as to the appropriateness of this metric.[18] Many of the factors driving 30‐day readmissions may be dependent on factors outside of the discharging hospital's control and it has been argued that nearer‐term, nonelective readmission rates may provide a more meaningful quality metric.[18] Seven‐day unplanned readmissions were thus used as a secondary outcome measure for this study.
Sensitivity, specificity, predictive value, C statistic, F score (the harmonic mean of positive predictive value and sensitivity),[19] and screen‐positive rate were calculated for each of the 30 prediction models evaluated using the historical data. The prediction model with the best balance of F score and screen‐positive rate was selected as the prediction model to be integrated into the EHR. Prospective validation of the selected prediction model was performed using the 12‐month period following implementation of the risk flag (September 2011September 2012).
To assess the impact of the automated prediction model on monthly readmission rate, we used the 24‐month period immediately before and the 12‐month period immediately after implementation of the readmission risk flag. Segmented regression analysis was performed testing for changes in level and slope of readmission rates between preimplementation and postimplementation time periods. This quasiexperimental interrupted time series methodology[20] allows us to control for secular trends in readmission rates and to assess the preimplementation trend (secular trend), the difference in rates immediately before and after the implementation (immediate effect), and the postimplementation change over time (sustained effect). We used Cochrane‐Orcutt estimation[21] to correct for serial autocorrelation.
All analyses were performed using Stata 12.1 software (Stata Corp, College Station, TX).
RESULTS
Predictors of Readmission
Our systematic review of the literature identified several patient and healthcare utilization patterns predictive of 30‐day readmission risk. Utilization factors included length of stay, number of prior admissions, previous 30‐day readmissions, and previous ED visits. Patient characteristics included number of comorbidities, living alone, and payor. Evidence was inconsistent regarding threshold values for these variables.
Many variables readily available in our EHR were either found by the systematic review not to be reliably predictive of 30‐day readmission (including age and gender) or were not readily or reliably available on admission (including length of stay and payor). At the time of implementation, our EHR did not include vital sign or nursing assessment variables, so these were not considered for inclusion in our model.
Of the available variables, 3 were consistently accurate and available in the EHR at the time of patient admission: prior hospital admission, emergency department visit, and 30‐day readmission within UPHS. We then developed 30 candidate prediction models using a combination of these variables, including 1 and 2 prior admissions, ED visits, and 30‐day readmissions in the 6 and 12 months preceding the index visit (Table 1).
Development and Validation
We used 24 months of retrospective data, which included 120,396 discharges with 17,337 thirty‐day readmissions (14.4% 30‐day all‐cause readmission rate) to test the candidate prediction models. A single risk factor, 2 inpatient admissions in the past 12 months, was found to have the best balance of sensitivity (40%), positive predictive value (31%), and proportion of patients flagged (18%) (Table 1).
| Sensitivity | Specificity | C Statistic | PPV | NPV | Screen Positive | F Score | |
|---|---|---|---|---|---|---|---|
| |||||||
| Retrospective Evaluation of Prediction Rules Lookback period: 6 months | |||||||
| Prior Admissions | |||||||
| 1 | 53% | 74% | 0.640 | 26% | 91% | 30% | 0.350 |
| 2 | 32% | 90% | 0.610 | 35% | 89% | 13% | 0.333 |
| 3 | 20% | 96% | 0.578 | 44% | 88% | 7% | 0.274 |
| Prior ED Visits | |||||||
| 1 | 31% | 81% | 0.558 | 21% | 87% | 21% | 0.252 |
| 2 | 13% | 93% | 0.532 | 25% | 87% | 8% | 0.172 |
| 3 | 7% | 97% | 0.519 | 27% | 86% | 4% | 0.111 |
| Prior 30‐day Readmissions | |||||||
| 1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.347 |
| 2 | 21% | 95% | 0.582 | 43% | 88% | 7% | 0.284 |
| 3 | 13% | 98% | 0.555 | 53% | 87% | 4% | 0.208 |
| Combined Rules | |||||||
| Admit1 & ED1 | 22% | 92% | 0.568 | 31% | 88% | 10% | 0.255 |
| Admit2 & ED1 | 15% | 96% | 0.556 | 40% | 87% | 5% | 0.217 |
| Admit1 & 30‐day1 | 39% | 85% | 0.623 | 31% | 89% | 18% | 0.346 |
| Admit2 & 30‐day1 | 29% | 92% | 0.603 | 37% | 89% | 11% | 0.324 |
| 30‐day1 & ED1 | 17% | 95% | 0.559 | 37% | 87% | 6% | 0.229 |
| 30‐day1 & ED2 | 8% | 98% | 0.527 | 40% | 86% | 3% | 0.132 |
| Lookback period: 12 months | |||||||
| Prior Admission | |||||||
| 1 | 60% | 68% | 0.593 | 24% | 91% | 36% | 0.340 |
| 2a | 40% | 85% | 0.624 | 31% | 89% | 18% | 0.354 |
| 3 | 28% | 92% | 0.600 | 37% | 88% | 11% | 0.318 |
| Prior ED Visit | |||||||
| 1 | 38% | 74% | 0.560 | 20% | 88% | 28% | 0.260 |
| 2 | 20% | 88% | 0.544 | 23% | 87% | 13% | 0.215 |
| 3 | 8% | 96% | 0.523 | 27% | 86% | 4% | 0.126 |
| Prior 30‐day Readmission | |||||||
| 1 | 43% | 84% | 0.630 | 30% | 90% | 20% | 0.353 |
| 2 | 24% | 94% | 0.592 | 41% | 88% | 9% | 0.305 |
| 3 | 11% | 98% | 0.548 | 54% | 87% | 3% | 0.186 |
| Combined Rules | |||||||
| Admit1 & ED1 | 29% | 87% | 0.580 | 27% | 88% | 15% | 0.281 |
| Admit2 & ED1 | 22% | 93% | 0.574 | 34% | 88% | 9% | 0.266 |
| Admit1 & 30‐day1 | 42% | 84% | 0.630 | 30% | 90% | 14% | 0.353 |
| Admit2 & 30‐day1 | 34% | 89% | 0.615 | 34% | 89% | 14% | 0.341 |
| 30‐day1 & ED1 | 21% | 93% | 0.569 | 35% | 88% | 9% | 0.261 |
| 30‐day1 & ED2 | 13% | 96% | 0.545 | 37% | 87% | 5% | 0.187 |
| Prospective Evaluation of Prediction Rule | |||||||
| 30‐Day All‐Cause | 39% | 84% | 0.614 | 30% | 89% | 18% | 0.339 |
Prospective validation of the prediction model was performed using the 12‐month period directly following readmission risk flag implementation. During this period, the 30‐day all‐cause readmission rate was 15.1%. Sensitivity (39%), positive predictive value (30%), and proportion of patients flagged (18%) were consistent with the values derived from the retrospective data, supporting the reproducibility and predictive stability of the chosen risk prediction model (Table 1). The C statistic of the model was also consistent between the retrospective and prospective datasets (0.62 and 0.61, respectively).
Readmission Rates
The mean 30‐day all‐cause readmission rate for the 24‐month period prior to the intervention was 14.4%, whereas the mean for the 12‐month period after the implementation was 15.1%. Thirty‐day all‐cause and 7‐day unplanned monthly readmission rates do not appear to have been impacted by the intervention (Figure 2). There was no evidence for either an immediate or sustained effect (Table 2).
| Hospital | Preimplementation Period | Immediate Effect | Postimplementation Period | P Value Change in Trenda | |||||
|---|---|---|---|---|---|---|---|---|---|
| Monthly % Change in Readmission Rates | P Value | Immediate % Change | P Value | Monthly % Change in Readmission Rates | P Value | ||||
| |||||||||
| 30‐Day All‐Cause Readmission Rates | |||||||||
| Hosp A | 0.023 | Stable | 0.153 | 0.480 | 0.991 | 0.100 | Increasing | 0.044 | 0.134 |
| Hosp B | 0.061 | Increasing | 0.002 | 0.492 | 0.125 | 0.060 | Stable | 0.296 | 0.048 |
| Hosp C | 0.026 | Stable | 0.413 | 0.447 | 0.585 | 0.046 | Stable | 0.629 | 0.476 |
| Health System | 0.032 | Increasing | 0.014 | 0.344 | 0.302 | 0.026 | Stable | 0.499 | 0.881 |
| 7‐Day Unplanned Readmission Rates | |||||||||
| Hosp A | 0.004 | Stable | 0.642 | 0.271 | 0.417 | 0.005 | Stable | 0.891 | 0.967 |
| Hosp B | 0.012 | Stable | 0.201 | 0.298 | 0.489 | 0.038 | Stable | 0.429 | 0.602 |
| Hosp C | 0.008 | Stable | 0.213 | 0.353 | 0.204 | 0.004 | Stable | 0.895 | 0.899 |
| Health System | 0.005 | Stable | 0.358 | 0.003 | 0.990 | 0.010 | Stable | 0.712 | 0.583 |
DISCUSSION
In this proof‐of‐concept study, we demonstrated the feasibility of an automated readmission risk prediction model integrated into a health system's EHR for a mixed population of hospitalized medical and surgical patients. To our knowledge, this is the first study in a general population of hospitalized patients to examine the impact of providing readmission risk assessment on readmission rates. We used a simple prediction model potentially generalizable to EHRs and healthcare populations beyond our own.
Existing risk prediction models for hospital readmission have important limitations and are difficult to implement in clinical practice.[22] Prediction models for hospital readmission are often dependent on retrospective claims data, developed for specific patient populations, and not designed for use early in the course of hospitalization when transitional care interventions can be initiated.[22] In addition, the time required to gather the necessary data and calculate the risk score remains a barrier to the adoption of prediction models in practice. By automating the process of readmission risk prediction, we were able to help integrate risk assessment into the healthcare process across many providers in a large multihospital healthcare organization. This has allowed us to consistently share risk assessment in real time with all members of the inpatient team, facilitating a team‐based approach to discharge planning.[23]
Two prior studies have developed readmission risk prediction models designed to be implemented into the EHR. Amarasingham et al.[24] developed and implemented[25] a heart failure‐specific prediction model based on the 18‐item Tabak mortality score.[26] Bradley et al.[27] studied in a broader population of medicine and surgery patients the predictive ability of a 26‐item score that utilized vital sign, cardiac rhythm, and nursing assessment data. Although EHRs are developing rapidly, currently the majority of EHRs do not support the use of many of the variables used in these models. In addition, both models were complex, raising concerns about generalizability to other healthcare settings and populations.
A distinctive characteristic of our model is its simplicity. We were cognizant of the realities of running a prediction model in a high‐volume production environment and the diminishing returns of adding more variables. We thus favored simplicity at all stages of model development, with the associated belief that complexity could be added with future iterations once feasibility had been established. Finally, we were aware that we were constructing a medical decision support tool rather than a simple classifier.[26] As such, the optimal model was not purely driven by discriminative ability, but also by our subjective assessment of the optimal trade‐off between sensitivity and specificity (the test‐treatment threshold) for such a model.[26] To facilitate model assessment, we thus categorized the potential predictor variables and evaluated the test characteristics of each combination of categorized variables. Although the C statistic of a model using continuous variables will be higher than a model using categorical values, model performance at the chosen trade‐off point is unlikely to be different.
Although the overall predictive ability of our model was fair, we found that it was associated with clinically meaningful differences in readmission rates between those triggering and not triggering the flag. The 30‐day all‐cause readmission rate in the 12‐month prospective sample was 15.1%, yet among those flagged as being at high risk for readmission the readmission rate was 30.4%. Given resource constraints and the need to selectively apply potentially costly care transition interventions, this may in practice translate into a meaningful discriminative ability.
Readmission rates did not change significantly during the study period. A number of plausible reasons for this exist, including: (1) the current model may not exhibit sufficient predictive ability to classify those at high risk or impact the behavior of providers appropriately, (2) those patients classified as high risk of readmission may not be at high risk of readmissions that are preventable, (3) information provided by the model may not yet routinely be used such that it can affect care, or (4) providing readmission risk assessment alone is not sufficient to influence readmission rates, and the other interventions or organizational changes necessary to impact care of those defined as high risk have not yet been implemented or are not yet being performed routinely. If the primary reasons for our results are those outlined in numbers 3 or 4, then readmission rates should improve over time as the risk flag becomes more routinely used, and those interventions necessary to impact readmission rates of those defined as high risk are implemented and performed.
Limitations
There are several limitations of this intervention. First, the prediction model was developed using 30‐day all‐cause readmissions, rather than attempting to identify potentially preventable readmissions. Thirty‐day readmission rates may not be a good proxy for preventable readmissions,[18] and as a consequence, the ability to predict 30‐day readmissions may not ensure that a prediction model is able to predict preventable readmissions. Nonetheless, 30‐day readmission rates remain the most commonly used quality metric.
Second, the impact of the risk flag on provider behavior is uncertain. We did not formally assess how the readmission risk flag was used by healthcare team members. Informal assessment has, however, revealed that the readmission risk flag is gradually being adopted by different members of the care team including unit‐based pharmacists who are using the flag to prioritize the delivery of medication education, social workers who are using the flag to prompt providers to consider higher level services for patients at high risk of readmission, and patient navigators who are using the flag to prioritize follow‐up phone calls. As a result, we hope that the flag will ultimately improve the processes of care for high‐risk patients.
Third, we did not capture readmissions to hospitals outside of our healthcare system and have therefore underestimated the readmission rate in our population. However, our assessment of the effect of the risk flag on readmissions focused on relative readmission rates over time, and the use of the interrupted time series methodology should protect against secular changes in outside hospital readmission rates that were not associated with the intervention.
Fourth, it is possible that the prediction model implemented could be significantly improved by including additional variables or data available during the hospital stay. However, simple classification models using a single variable have repeatedly been shown to have the ability to compete favorably with state‐of‐the‐art multivariable classification models.[28]
Fifth, our study was limited to a single academic health system, and our experience may not be generalizable to smaller healthcare systems with limited EHR systems. However, the simplicity of our prediction model and the integration into a commercial EHR may improve the generalizability of our experience to other healthcare settings. Additionally, partly due to recent policy initiatives, the adoption of integrated EHR systems by hospitals is expected to continue at a rapid rate and become the standard of care within the near future.[29]
CONCLUSION
An automated prediction model was effectively integrated into an existing EHR and was able to identify patients on admission who are at risk for readmission within 30 days of discharge. Future work will aim to further examine the impact of the flag on readmission rates, further refine the prediction model, and gather data on how providers and care teams use the information provided by the flag.
Disclosure
Dr. Umscheid‐s contribution to this project was supported in part by the National Center for Research Resources, Grant UL1RR024134, which is now at the National Center for Advancing Translational Sciences, Grant UL1TR000003. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
- , , . Rehospitalizations among patients in the Medicare fee‐for‐service program. N Engl J Med. 2009;360(14):1418–1428.
- , . Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med. 2000;160(8):1074–1081.
- , , , , . Do older rural and urban veterans experience different rates of unplanned readmission to VA and non‐VA hospitals? J Rural Health. 2009;25(1):62–69.
- , , . Cost, causes and rates of rehospitalization of preterm infants. J Perinatol. 2007;27(10):614–619.
- , , , . Redefining readmission risk factors for general medicine patients. J Hosp Med. 2011;6(2):54–60.
- , , , et al. Incidence and main factors associated with early unplanned hospital readmission among French medical inpatients aged 75 and over admitted through emergency units. Age Ageing. 2008;37(4):416–422.
- , , , , . Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ. 2011;183(7):E391–E402.
- Hospital Quality Alliance. Available at: http://www.hospitalqualityalliance.org/hospitalqualityalliance/qualitymeasures/qualitymeasures.html. Accessed March 6, 2013.
- Institute for Healthcare Improvement. Available at: http://www.ihi.org/explore/Readmissions/Pages/default.aspx. Accessed March 6, 2013.
- Centers for Medicare and Medicaid Services. Available at: http://www.cms.gov/Medicare/Quality‐Initiatives‐Patient‐Assessment‐Instruments/HospitalQualityInits/OutcomeMeasures.html. Accessed March 6, 2013.
- , , , et al. Comprehensive discharge planning and home follow‐up of hospitalized elders: a randomized clinical trial. JAMA. 1999;281(7):613–620.
- , , , , , . Preparing patients and caregivers to participate in care delivered across settings: the Care Transitions Intervention. J Am Geriatr Soc. 2004;52(11):1817–1825.
- , , , , . The importance of transitional care in achieving health reform. Health Aff (Millwood). 2011;30(4):746–754.
- , , , , . Interventions to reduce 30‐day rehospitalization: a systematic review. Ann Intern Med. 2011;155(8):520–528.
- University of Pennsylvania Health System Center for Evidence‐based Practice. Available at: http://www.uphs.upenn.edu/cep/. Accessed March 6, 2013.
- , , . Hospital‐based comparative effectiveness centers: translating research into practice to improve the quality, safety and value of patient care. J Gen Intern Med. 2010;25(12):1352–1355.
- . Writing Arden Syntax Medical Logic Modules. Comput Biol Med. 1994;24(5):331–363.
- , . Thirty‐day readmissions—truth and consequences. N Engl J Med. 2012;366(15):1366–1369.
- . Information Retrieval. 2nd ed. Oxford, UK: Butterworth‐Heinemann; 1979.
- , , , . Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther. 2002;27(4):299–309.
- , . Application of least squares regression to relationships containing auto‐correlated error terms. J Am Stat Assoc. 1949; 44:32–61.
- , , , et al. Risk prediction models for hospital readmission: a systematic review. JAMA. 2011;306(15):1688–1698.
- , , , et al. Core Principles and values of effective team‐based health care. Available at: https://www.nationalahec.org/pdfs/VSRT‐Team‐Based‐Care‐Principles‐Values.pdf. Accessed March 19, 2013.
- , , , et al. An automated model to identify heart failure patients at risk for 30‐day readmission or death using electronic medical record data. Med Care. 2010;48(11):981–988.
- , , , et al. Allocating scarce resources in real‐time to reduce heart failure readmissions: a prospective, controlled study [published online ahead of print July 31, 2013]. BMJ Qual Saf. doi:10.1136/bmjqs‐2013‐001901.
- , . The threshold approach to clinical decision making. N Engl J Med. 1980;302(20):1109–1117.
- , , , , . Identifying patients at increased risk for unplanned readmission. Med Care. 2013;51(9):761–766.
- . Very simple classification rules perform well on most commonly used datasets. Mach Learn. 1993;11(1):63–91.
- . Stimulating the adoption of health information technology. N Engl J Med. 2009;360(15):1477–1479.
- , , . Rehospitalizations among patients in the Medicare fee‐for‐service program. N Engl J Med. 2009;360(14):1418–1428.
- , . Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med. 2000;160(8):1074–1081.
- , , , , . Do older rural and urban veterans experience different rates of unplanned readmission to VA and non‐VA hospitals? J Rural Health. 2009;25(1):62–69.
- , , . Cost, causes and rates of rehospitalization of preterm infants. J Perinatol. 2007;27(10):614–619.
- , , , . Redefining readmission risk factors for general medicine patients. J Hosp Med. 2011;6(2):54–60.
- , , , et al. Incidence and main factors associated with early unplanned hospital readmission among French medical inpatients aged 75 and over admitted through emergency units. Age Ageing. 2008;37(4):416–422.
- , , , , . Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ. 2011;183(7):E391–E402.
- Hospital Quality Alliance. Available at: http://www.hospitalqualityalliance.org/hospitalqualityalliance/qualitymeasures/qualitymeasures.html. Accessed March 6, 2013.
- Institute for Healthcare Improvement. Available at: http://www.ihi.org/explore/Readmissions/Pages/default.aspx. Accessed March 6, 2013.
- Centers for Medicare and Medicaid Services. Available at: http://www.cms.gov/Medicare/Quality‐Initiatives‐Patient‐Assessment‐Instruments/HospitalQualityInits/OutcomeMeasures.html. Accessed March 6, 2013.
- , , , et al. Comprehensive discharge planning and home follow‐up of hospitalized elders: a randomized clinical trial. JAMA. 1999;281(7):613–620.
- , , , , , . Preparing patients and caregivers to participate in care delivered across settings: the Care Transitions Intervention. J Am Geriatr Soc. 2004;52(11):1817–1825.
- , , , , . The importance of transitional care in achieving health reform. Health Aff (Millwood). 2011;30(4):746–754.
- , , , , . Interventions to reduce 30‐day rehospitalization: a systematic review. Ann Intern Med. 2011;155(8):520–528.
- University of Pennsylvania Health System Center for Evidence‐based Practice. Available at: http://www.uphs.upenn.edu/cep/. Accessed March 6, 2013.
- , , . Hospital‐based comparative effectiveness centers: translating research into practice to improve the quality, safety and value of patient care. J Gen Intern Med. 2010;25(12):1352–1355.
- . Writing Arden Syntax Medical Logic Modules. Comput Biol Med. 1994;24(5):331–363.
- , . Thirty‐day readmissions—truth and consequences. N Engl J Med. 2012;366(15):1366–1369.
- . Information Retrieval. 2nd ed. Oxford, UK: Butterworth‐Heinemann; 1979.
- , , , . Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther. 2002;27(4):299–309.
- , . Application of least squares regression to relationships containing auto‐correlated error terms. J Am Stat Assoc. 1949; 44:32–61.
- , , , et al. Risk prediction models for hospital readmission: a systematic review. JAMA. 2011;306(15):1688–1698.
- , , , et al. Core Principles and values of effective team‐based health care. Available at: https://www.nationalahec.org/pdfs/VSRT‐Team‐Based‐Care‐Principles‐Values.pdf. Accessed March 19, 2013.
- , , , et al. An automated model to identify heart failure patients at risk for 30‐day readmission or death using electronic medical record data. Med Care. 2010;48(11):981–988.
- , , , et al. Allocating scarce resources in real‐time to reduce heart failure readmissions: a prospective, controlled study [published online ahead of print July 31, 2013]. BMJ Qual Saf. doi:10.1136/bmjqs‐2013‐001901.
- , . The threshold approach to clinical decision making. N Engl J Med. 1980;302(20):1109–1117.
- , , , , . Identifying patients at increased risk for unplanned readmission. Med Care. 2013;51(9):761–766.
- . Very simple classification rules perform well on most commonly used datasets. Mach Learn. 1993;11(1):63–91.
- . Stimulating the adoption of health information technology. N Engl J Med. 2009;360(15):1477–1479.
© 2013 Society of Hospital Medicine
Caring for Patients on Insulin Pumps
Delivery of insulin via continuous subcutaneous insulin infusion (CSII), or insulin pump, has gained wide acceptance. It is estimated that 400,000 patients with type 1 diabetes mellitus (T1DM) are using insulin pumps.[1] A registry for T1DM in the United States indicated that 50% of the 25,833 participants were using an insulin pump.[2] Use of CSII in type 2 DM (T2DM) is also increasing.[3]
Patients with DM are 3 as likely to be hospitalized than patients without DM.[4] Twenty percent to 30% of adult hospitalized patients have a known diagnosis of DM.[5] It is therefore to be expected that patients on CSII will be seen with increased frequency in the hospital setting. This leads to potential difficultiesinpatient healthcare providers may not be familiar with insulin pumps, and patients may not be aware of complexities associated with pump usage in the hospital.
This article will review CSII usage in the hospital, offering strategies for management in partnership with the patient based on our experiences and processes developed in our institution.
SHOULD CONTINUOUS SUBCUTANEOUS INSULIN INFUSION BE CONTINUED IN THE HOSPITAL?
The American Diabetes Association advocates (1) allowing patients who are physically and mentally able to continue CSII when hospitalized, (2) having a hospital policy for CSII use, and (3) having hospital personnel with expertise on pump management.[6] The American Association of Clinical Endocrinologists echoes much of the same and suggests contacting the specialist responsible for the pump in the ambulatory setting for decisions on adjustments in the hospitalized patient,[7] which at times may not be feasible.
The logic and benefits of basal‐bolus insulin dosing (ie, giving basal insulin to account for fasting requirements, plus bolus insulin to cover nutritional and correctional needs) have been well‐described.[8, 9, 10] In randomized clinical trials on patients admitted to general medical and surgical floors, basal‐bolus insulin (long‐acting basal insulin plus mealtime fast‐acting insulin injections) resulted in better glycemic control and reduced infection rates compared with sliding‐scale therapy (waiting for high blood glucoses before giving insulin, instead of giving it proactively to prevent hyperglycemia).[9, 10] At present, insulin delivery via the insulin pump is the best commercially available method to deliver insulin in a basal‐bolus manner in ambulatory patients. It thus makes sense to continue CSII in the hospital if patients are able to manage their pumps, though there are no randomized trials answering this question as of yet.
Studies on insulin pumps in the hospital are sparse. In one group's latest retrospective study of 136 patients over a 6‐year period, CSII was continued during the entire hospital stay in 65% of the hospitalizations, was used intermittently in 20%, and was discontinued with alternative insulin regimens given in 15%.[11] Mean glucose was 178 47 mg/dL (mean standard deviation), with no significant difference between groups. There were fewer episodes of severe hyperglycemia among those who continued on the pump compared with the other groups, and fewer episodes of hypoglycemia in those who continued on vs those taken off the pump. There was 1 episode of an infusion catheter kinking, resulting in nonfatal hyperglycemia, but no reported pump‐site infections, mechanical pump failure, or diabetic ketoacidosis (DKA) among patients remaining on CSII.
CLINICAL VIGNETTES
The following cases illustrate potential challenges with CSII use that we have encountered in the hospital.
The Patient Needing Transition to Multiple Daily Insulin Injections
A 29‐year‐old male with T1DM, on CSII, was admitted for fever and chills. His latest glycated hemoglobin (HbA1c) level was 6.8%. His glucose levels started rising, and he wished to be taken off the insulin pump. He was started by the primary team on multiple daily insulin injections (MDII) with insulin glargine and insulin lispro. His glucose levels continued to rise, so an intravenous (IV) insulin infusion was started. Endocrinology was then consulted. The patient's condition was concerning for the potential development of DKA, so he was kept on IV insulin. When he was ready for transitioning to subcutaneous insulin, the pump had been taken home by a family member, and the patient could not recall his CSII basal rates but knew his total basal insulin dose, carbohydrate ratio, and sensitivity factor. Endocrinology assisted in transitioning him from the insulin infusion to MDII based on these recalled doses. When the insulin pump was available, the pump settings were interrogated, and he was transitioned back to it.
Key points:
- Having key hospital personnel trained on CSII, including interrogating the pump's settings, facilitates the transitioning of these patients from one hospital unit, or level of care, to another.
- Accessing the pump's settings involves pushing several buttons on the insulin pump. Because key hospital personnel will encounter patients on different insulin pumps, it may be helpful to keep menu maps handy as a quick reference. A menu map will show at a glance where certain information can be found, such as the basal insulin rate or the sensitivity factor (see examples in Figures 1 and 2).[12, 13]
- Knowing the HbA1c will help determine if pump use has been effective.
The Patient With Technical Problems
An 84‐year‐old gentleman with T2DM was admitted for heart failure and aortic valve replacement. His HbA1c was 6.2%, and he had had several outpatient hypoglycemic events. While on CSII in the hospital, his point of care testing (POCT) glucose readings ranged from 105 to 260 mg/dL. On the afternoon of the third hospital day, POCT readings stayed above 220 mg/dL and rose to 348 mg/dL on the fourth hospital morning, despite multiple blousing and changing the insulin, insertion site, reservoir cartridge, and pump tubing. There was no evidence of infection and no medication change that would have impacted glucose levels. Review of his procedures revealed that he had undergone computed tomography (CT) on the morning of hospital day 3 and wore his pump while being scanned. The pump company was notified.
Key points:
- Patients, and medical and nursing staff, should be reminded to remove insulin pumps for CT scans, magnetic resonance imaging, x‐rays, or other tests with high electromagnetic fields.
- If there is a suspicion of pump malfunction from such a procedure, notify the pump company.
The Patient Who Can Benefit From Inpatient Education
A 70‐year‐old female with T1DM was admitted for heart failure. The patient had been using CSII for 20 years. Her latest HbA1c was 6.9%. She had 1 hypoglycemic event every 1 to 2 weeks. In the hospital, she experienced 2 hypoglycemic events within 3 days, both around bedtime. It was discovered that the patient was giving a bolus of insulin for elevated glucose levels based on the hospital POCT, and when the meal arrived (3060 minutes later), she again delivered a bolus based on her own glucometer reading plus insulin based on the carbohydrates in her meal. The patient was then instructed to request the POCT when her meal tray arrived, and she was taught how to use the pump's built‐in calculator. Glucose excursions improved.
Key points:
- Patients on CSII, though able to exercise autonomy in managing their insulin doses, may also need assistance in dosing insulin properly.
- Although pump education is ideally done on an outpatient basis, hospital‐based providers may encounter patients who need reinforcement of their training while hospitalized. Hospital personnel trained on insulin pumps (such as physicians, nurse practitioners, physician assistants, and certified diabetes educators) can help augment the patient's knowledge while in the hospital. In the absence of such key personnel, patient safety has to be addressed with re‐evaluation of the need to discontinue the pump and switch to multiple doses of subcutaneous insulin.
STEPS IN TAKING CARE OF PATIENTS ON CONTINUOUS SUBCUTANEOUS INSULIN INFUSION
Initial Patient Assessment
On admission, patients are asked whether they use an insulin pump. This is included in the nursing assessment form. If they do, the physician is notified.
The insulin pump might be missed unless specifically asked for because (1) the insulin pump may be thought of more as a device rather than a medication, and (2) the insulin pump may be worn in less obvious areas, not only on the abdomen where providers are more apt to detect it.
Hospital Policy and Insulin Orders
Written hospital policies on how to safely manage patients presenting with an insulin pump will delineate patients who can safely be allowed to continue on the pump, and the responsibilities that come along with this. Our institution has such a policy. Experts from both the legal and biomedical engineering departments were consulted when the policy was crafted. Patients must be fully alert, able, and willing to self‐manage the pump. General contraindications to pump use in the hospital, such as altered mental status or DKA, are listed in Table 1. In addition, patients in the intensive care units are best managed on an IV insulin infusion during their critical illness, in keeping with several society guidelines.[14] Controlling severe hyperglycemia and DKA with multiple boluses through the insulin pump can potentially lead to stacking of insulin with subsequent hypoglycemia.
| Altered state of consciousness |
| Suicidal ideation |
| Prolonged instability of blood glucose levels |
| Diabetic ketoacidosis |
| Patient/family inability or refusal to participate in own care |
| Insulin‐pump malfunction |
| Lack of appropriate supplies for the insulin pump |
| Other circumstances as identified by the physician, resident, or licensed independent practitioner |
In our institution, a computerized insulin pump order set has to be activated. Apart from insulin, POCT, and hypoglycemia‐management orders, this order set contains documents aimed at balancing patient autonomy with delivery of appropriate and safe medical care that the bedside nurse goes over with the patient (Table 2). By policy at our institution, insulin should be dispensed from the hospital's pharmacy (except for that already in the pump), so the order set is linked to the pharmacy and a 3‐mL insulin vial is delivered to the hospital floor and stored in the patient‐specific medication bin. The order set triggers an Endocrinology consult so that the patient can be assessed by key trained personnel.
|
| CSII pump therapy patient agreement |
| Delineates the conditions for continuing on CSII and those for whom it may be discontinued |
| Terms of use and release of liability for patient‐owned equipment |
| Delineates the patient's responsibility for the pump and supplies |
| Patient‐maintained flow sheet for inpatient CSII |
| Includes blood glucose levels (obtained by nurse or patient‐care assistant with the hospital glucose meter) |
| Includes insulin doses (basal, bolus) |
| Includes carbohydrate intake in grams |
Patient Diagnosis
It is important to try to distinguish T1DM vs T2DM, as patients with T1DM are prone to ketoacidosis when the pump is disconnected.
Patient Assessment by the Endocrinology Consult Service
Hospitalized patients on the pump have varying degrees of pump knowledge and skill sets. We have encountered highly trained patients who meticulously count their carbohydrates and double‐check the insulin doses calculated by the built‐in pump calculator, and those who have knowledge gaps because their physicians, and not they themselves, change pump settings at the clinic visits.
Therefore, the Endocrinology consult team members (comprising physicians, nurse practitioners, and certified diabetes educators) go through the following items to be able to order the insulin correctly, assess whether patients are still able and willing to continue on their pump despite their illness, formulate alternative insulin regimens as needed, or help empower patients who may have forgotten some aspects of pump management:
- Insulin pump manufacturer/model.
- Insulin used in the pump.
- Often fast‐acting insulin (lispro, aspart, or glulisine).
- Some patients use regular insulin.
- A few patients use U500 insulin (5 more potent than other insulins).
- Insulin doses/pump settings.
Patients are assessed for:
- Hypoglycemia awareness.
- Previous glucose control.
- Bolus calculation (either using the built‐in calculator, computing this mentally, or using a different calculator).
- Ability to deliver a bolus (including vision and dexterity challenges).
- Ability to change the basal rate, or set a temporary rate, and suspend insulin delivery.
Discussion on Options for Inpatient Management
After assessment, education is provided as needed. If there are concerns on the part of the patient, the primary team, or the Endocrinology team about safe continuation of CSII during the hospitalization, then alternative insulin regimens are discussed. Patients who cannot access their basal rates and cannot adjust the doses are not able to self‐manage; they should be taken off the pump and treated with multiple subcutaneous insulin doses. Conversion to MDII is discussed under Interruption of Continuous Subcutaneous Insulin Infusion for Short and Prolonged Periods.
Provision of Pump Information for Hospital Healthcare Providers
Users of CSII, even if perfectly competent using their pumps in the ambulatory setting, may need assistance in the hospital for various reasons. They may not know what to do for surgical or radiologic procedures (discussed below) and may not be familiar with hospital policies involving CSII. Hospital providers trained on insulin pumps may need a refresher on locating a particular pump setting.
The provider can call the toll‐free number on the back of the pump for assistance (Table 3). Insulin‐pump companies also have menu maps to aid in finding information on pump settings (samples shown in Figures 1 and 2).[12, 13] Documentation of the patient's pump settings will assist in CSII dose changes during the acute illness or assist in switching to MDII if needed. The following information need to be collected:
| Animas Corporation | 877‐937‐7867 |
| Insulet Corporation | 800‐591‐3455 |
| Medtronic | 800‐826‐2099 |
| Roche Diagnostics | 800‐688‐4578 |
Basal Rate
This is the hourly insulin rate delivered for the patient's insulin needs even when not eating. The patient might have one or multiple basal rates in a day, or a different pattern on some days. Because the patient's activity in the hospital will be different from the usual ambulatory activity, we recommend that patients choose only 1 pattern.
Bolus
This is the insulin to cover meals or to correct for hyperglycemia, or both. The patient has to activate buttons for delivery. The patient may or may not be using the built‐in pump calculator.
Carbohydrate Ratio
This is the amount of insulin per quantity of carbohydrate consumed. When patients are initially placed on the insulin pump, they are given a carbohydrate ratio that is derived from a calculation called the rule of 500. In the rule of 500, the number 500 is divided by the patient's total daily insulin dose while on multiple subcutaneous insulin shots. For example, if the patient was on insulin glargine 13 units daily and insulin lispro 4 units 3 daily with meals, 500 divided by 25 gives us a carbohydrate ratio of 20 grams of carbohydrate for 1 unit of insulin (or conversely called insulin‐to‐carbohydrate ratio of 1 unit of insulin for every 20 grams of carbohydrate).
This often comes out to 1 unit for every 1530 grams of carbohydrates in patients with T1DM, and 1 unit for every 515 grams of carbohydrate in patients with T2DM, reflecting the need for a higher insulin dose in the latter.
It is best to ask the patient how many units he or she usually takes for a meal, or to present the patient with an example of a meal and ask how much he or she would take. We have encountered a patient whose carbohydrate ratio was 1, but upon further inquiry, the patient demonstrated that he actually bolused 1 unit for every 1 serving (or 1 unit for every 15 grams) of carbohydrate.
Sensitivity Factor
This is the amount of insulin that would bring the blood glucose to goal. For example, if the patient requires 1 unit of insulin to bring down the blood glucose from 170 mg/dL to 120 mg/dL, then the sensitivity factor of 50 would be seen on the pump screen. Similar to the carbohydrate ratio, a sensitivity factor is calculated when patients are initially placed on the insulin pump. This time, the rule of 1800 is used, where the number 1800 is divided by the patient's total daily insulin dose. In patients with T1DM, this often comes to 30100 mg/dL per 1 unit of insulin; or, conversely, 1 unit for every 30100 mg/dL glucose. For patients with T2DM, this is often 1 unit for every 1025 mg/dL glucose.
This insulin dose is given in addition to the dose resulting from the carbohydrate ratio, or alone if the patient is not eating.
Target
This is the blood glucose goal for the patient, which might be too tight in the presence of acute illness, and therefore would have to be modified. The American Diabetes Association, Endocrine Society, and American Association of Clinical Endocrinologists recommend premeal glucose targets of <140 mg/dL in hospitalized noncritically ill patients on insulin, with re‐evaluation of the insulin dose when premeal glucose levels fall below 100 mg/dL and dose adjustment if they are <70mg/dLunless there is an obvious explanation, such as a missed meal.[14, 15]
Point‐of‐Care Testing for Glucose Monitoring
Our policy specifies that the hospital glucose meter is the meter of record upon which dose adjustments are based. Point‐of‐care testing is performed by our patient care nursing assistants or bedside nurses. The timing is typically before meals, at bedtime, between 2 and 3 AM, and with allowance for other times that patients are used to checking when they were at home, such as after meals. Frequent POCT has to be especially borne in mind for patients with hypoglycemia unawareness. Some patients are used to checking with their own home glucose meters in between these times, and we do work with them with the understanding that dose‐change decisions are based on the hospital glucose meter readings.
Dose Adjustments
Continuous subcutaneous insulin infusion dose adjustments for hypoglycemia and hyperglycemia are usually done in 10% to 20% decrements/emncrements. Our Endocrinology team discusses these with the patients and ensures that the new settings are entered into the pump and into the order set.
INTERRUPTION OF CONTINUOUS SUBCUTANEOUS INSULIN INFUSION FOR SHORT AND PROLONGED PERIODS
Patients with T1DM should not be left without basal insulin. However, pump interruption for 30 minutes to an hour often does not lead to problems. Beyond an hour and certainly closer to 2 to 3 hours off the pump, the patient should be given a subcutaneous insulin injection if the patient is left without easy access to the insulin pump.
The subcutaneous insulin dose for temporary pump suspension can be roughly calculated as hourly basal rate multiplier, where the multiplier is the number of hours that the patient is expected to be disconnected from the pump (for example, hourly basal rate of 0.85 unit/hour 3 hours = 2.55 units, which can be rounded off to 2 or 3 units depending on the patient's general glucose control).
When it has been determined that the patient should come off the pump for substantial periods of time, then subcutaneous insulin injections should be given. This is imperative for the prevention of DKA in patients with T1DM, and highly recommended for maintenance of good glycemic control for patients with T2DM.
Basal Dose
In most cases, these situations stretch for greater than 24 hours, such as surgery and the anticipated recovery from anesthesia. We favor long‐acting insulin for basal needs, given 2 hours before discontinuing the pump. The total basal insulin dose per day can be given as the starting long‐acting insulin dose and then adjusted as needed. The total daily basal insulin dose can be retrieved from the insulin pump. In one study on T1DM patients using insulin lispro on the pump, total daily basal dose was given as insulin glargine without adverse effects.[16] If there is concern for hypoglycemia, then the dose can be reduced by 10% to 20%. Care should be made to ascertain that the basal insulin delivered via the pump is appropriate.
There are several ways to estimate this:
- If the daily total basal and the total prandial insulin requirements approximate a 50:50 ratio, then the basal rate via the pump is appropriate.
- If the daily total basal rate via the pump is similar to a weight‐based estimate of the basal dose (often 0.150.2 units/kg/day in patients with T1DM, 0.20.3 units/kg/day for T2DM, and higher in both cases with longer duration of DM or greater insulin resistance), then the basal rate via the pump is appropriate.
Bolus Dose
Patients can still continue to calculate their sensitivity factor and carbohydrate ratio and request for the equivalent dose of insulin. In the ideal situation, bedside nurses would be taught how to calculate this ratio and dose rapid‐acting insulin accordingly should the patient need to come off the insulin pump. Because of the logistic difficulties of making this uniform in our institution, we have worked instead on providing patients with information on the carbohydrate content of their meal tray. If the pump is discontinued, the patient would continue to calculate their prandial insulin based on their carbohydrates ratio and indicate to the nurse how much he/she would need. Our subcutaneous insulin orders for MDII allow for us to put a range of insulin doses based on the patient's typical insulin needs for mealtime.
Pump Removal for Certain Hospital Procedures
Patients may not remember that the pump has to be removed before entering high‐radiation areas. The pump owner manuals tell patients not to use the pump when going for magnetic resonance imaging, CT scans, or x‐rays, or near equipment with high electromagnetic fields.
Interrogation of the Pump
If there is concern about pump malfunction, patients should be switched to MDII. The pump company can be contacted for pump interrogation and provision of a temporary pump (Table 3).
Pump Disconnection
In our institution, the Radiology department has signage instructing patients to inform the technician if they are wearing an insulin pump. The pump is handed off to a family member or stored until the procedure is over. Another option is to leave it with the bedside nurse or the floor nurse manager for safekeeping. This is less ideal, because the wait for the radiologic procedure might take longer than expected and the patient is left without any insulin on board.
Interruption of Continuous Subcutaneous Insulin Infusion for Surgical Procedures
In our institution, the anesthesiology department has worked with the Endocrinology, Surgery and Medicine departments regarding patients with insulin pumps. Discontinuation of CSII is recommended for surgical procedures longer than 1 hour; patients are asked to continue on the insulin pump until they are taken to the preoperative suite, at which point they are placed on IV insulin infusion. Ideally, there should be an overlap of 1530 minutes. Providing an alternative continuous source of insulin during pump interruption is important, especially for patients with T1DM.[17]
Pump Resumption
Once the patient is ready to resume the pump, any subcutaneous insulin that was delivered and might still be active has to be accounted for and subtracted from the basal pump dose so that hypoglycemia is avoided. An alternative would be to wait until the last basal subcutaneous insulin dose is expected to be cleared before restarting CSII.
SUMMARY
As patients on an insulin pump are increasingly seen in the hospital, inpatient providers have to be able to adapt to these patients' needs. Inpatient providers need to have a working knowledge of the insulin pump. Alternative methods of insulin delivery will have to be discussed with the patient to assure continued safety in the hospital.
Disclosures
Dr. Lansang has served as a Sanofi Advisory Board member.
- http://www.rncos.com/Press_Releases/US‐to‐Dominate‐the‐Global‐Insulin‐Pump‐Market.htm. Accessed on October 25, 2013.
- , , , , , ;T1D Exchange Clinic Network. The T1D exchange clinic registry. J Clin Endocrinol Metab. 2012;97(12):4383–4389.
- , , , et al. Resource utilization with insulin pump therapy for type 2 diabetes mellitus. Am J Manag Care. 2010;16(12):892–896.
- , , , , . Diabetes‐related hospitalization and hospital‐related utilization. In: Diabetes in America. 2nd ed. Bethesda, MD: National Diabetes Data Group, National Institute of Diabetes and Digestive and Kidney Diseases; 1995:553–569. Available at: http://diabetes.niddk.nih.gov/dm/pubs/america/pdf/chapter27.pdf. Accessed February 21, 2013.
- , , , , , . Hyperglycemia: an independent marker of in‐hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978–982.
- American Diabetes Association. Standards of medical care in diabetes—2012. Diabetes Care. 2012;35(suppl 1):S11–S63.
- , , , et al;AACE Insulin Pump Management Task Force. Statement by the American Association of Clinical Endocrinologists Consensus Panel on insulin pump management. Endocr Pract. 2010;16(5):746–762.
- , , , et al;American Diabetes Association Diabetes in Hospitals Writing Committee. Management of diabetes and hyperglycemia in hospitals [published correction appears in Diabetes Care. 2004;27(5):1255]. Diabetes Care. 2004;27(2):553–591.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes (RABBIT 2 trial). Diabetes Care. 2007;30(9):2181–2186.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes undergoing general surgery (RABBIT 2 surgery). Diabetes Care. 2011;34(2):256–261.
- , , , , , . Transitioning insulin pump therapy from the outpatient to the inpatient setting: a review of 6 years' experience with 253 cases. J Diabetes Sci Technol. 2012;6(5):995–1002.
- Medtronic Paradigm Revel insulin pump [menu map]. Northridge, CA: Medtronic. Available at: http://www.medtronicdiabetes.com/sites/default/files/library/download‐library/workbooks/x23_menu_map.pdf. Updated January 22, 2010. Accessed February 2013.
- OneTouch Ping insulin pump [menu map]. West Chester, PA: Animas Corporation.
- , , , et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Endocr Pract. 2009;15(4):353–369.
- , , , et al. Management of hyperglycemia in hospitalized patients in non‐critical care setting: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38.
- , , , . Switch to multiple daily injections with insulin glargine and insulin lispro from continuous subcutaneous insulin infusion with insulin lispro: a randomized, open‐label study using a continuous glucose monitoring system. Endocr Pract. 2005;11(3):157–164.
- , , , , . Perioperative glycemic management in insulin pump patients undergoing noncardiac surgery. Curr Pharm Des. 2012;18(38):6204–6214.
Delivery of insulin via continuous subcutaneous insulin infusion (CSII), or insulin pump, has gained wide acceptance. It is estimated that 400,000 patients with type 1 diabetes mellitus (T1DM) are using insulin pumps.[1] A registry for T1DM in the United States indicated that 50% of the 25,833 participants were using an insulin pump.[2] Use of CSII in type 2 DM (T2DM) is also increasing.[3]
Patients with DM are 3 as likely to be hospitalized than patients without DM.[4] Twenty percent to 30% of adult hospitalized patients have a known diagnosis of DM.[5] It is therefore to be expected that patients on CSII will be seen with increased frequency in the hospital setting. This leads to potential difficultiesinpatient healthcare providers may not be familiar with insulin pumps, and patients may not be aware of complexities associated with pump usage in the hospital.
This article will review CSII usage in the hospital, offering strategies for management in partnership with the patient based on our experiences and processes developed in our institution.
SHOULD CONTINUOUS SUBCUTANEOUS INSULIN INFUSION BE CONTINUED IN THE HOSPITAL?
The American Diabetes Association advocates (1) allowing patients who are physically and mentally able to continue CSII when hospitalized, (2) having a hospital policy for CSII use, and (3) having hospital personnel with expertise on pump management.[6] The American Association of Clinical Endocrinologists echoes much of the same and suggests contacting the specialist responsible for the pump in the ambulatory setting for decisions on adjustments in the hospitalized patient,[7] which at times may not be feasible.
The logic and benefits of basal‐bolus insulin dosing (ie, giving basal insulin to account for fasting requirements, plus bolus insulin to cover nutritional and correctional needs) have been well‐described.[8, 9, 10] In randomized clinical trials on patients admitted to general medical and surgical floors, basal‐bolus insulin (long‐acting basal insulin plus mealtime fast‐acting insulin injections) resulted in better glycemic control and reduced infection rates compared with sliding‐scale therapy (waiting for high blood glucoses before giving insulin, instead of giving it proactively to prevent hyperglycemia).[9, 10] At present, insulin delivery via the insulin pump is the best commercially available method to deliver insulin in a basal‐bolus manner in ambulatory patients. It thus makes sense to continue CSII in the hospital if patients are able to manage their pumps, though there are no randomized trials answering this question as of yet.
Studies on insulin pumps in the hospital are sparse. In one group's latest retrospective study of 136 patients over a 6‐year period, CSII was continued during the entire hospital stay in 65% of the hospitalizations, was used intermittently in 20%, and was discontinued with alternative insulin regimens given in 15%.[11] Mean glucose was 178 47 mg/dL (mean standard deviation), with no significant difference between groups. There were fewer episodes of severe hyperglycemia among those who continued on the pump compared with the other groups, and fewer episodes of hypoglycemia in those who continued on vs those taken off the pump. There was 1 episode of an infusion catheter kinking, resulting in nonfatal hyperglycemia, but no reported pump‐site infections, mechanical pump failure, or diabetic ketoacidosis (DKA) among patients remaining on CSII.
CLINICAL VIGNETTES
The following cases illustrate potential challenges with CSII use that we have encountered in the hospital.
The Patient Needing Transition to Multiple Daily Insulin Injections
A 29‐year‐old male with T1DM, on CSII, was admitted for fever and chills. His latest glycated hemoglobin (HbA1c) level was 6.8%. His glucose levels started rising, and he wished to be taken off the insulin pump. He was started by the primary team on multiple daily insulin injections (MDII) with insulin glargine and insulin lispro. His glucose levels continued to rise, so an intravenous (IV) insulin infusion was started. Endocrinology was then consulted. The patient's condition was concerning for the potential development of DKA, so he was kept on IV insulin. When he was ready for transitioning to subcutaneous insulin, the pump had been taken home by a family member, and the patient could not recall his CSII basal rates but knew his total basal insulin dose, carbohydrate ratio, and sensitivity factor. Endocrinology assisted in transitioning him from the insulin infusion to MDII based on these recalled doses. When the insulin pump was available, the pump settings were interrogated, and he was transitioned back to it.
Key points:
- Having key hospital personnel trained on CSII, including interrogating the pump's settings, facilitates the transitioning of these patients from one hospital unit, or level of care, to another.
- Accessing the pump's settings involves pushing several buttons on the insulin pump. Because key hospital personnel will encounter patients on different insulin pumps, it may be helpful to keep menu maps handy as a quick reference. A menu map will show at a glance where certain information can be found, such as the basal insulin rate or the sensitivity factor (see examples in Figures 1 and 2).[12, 13]
- Knowing the HbA1c will help determine if pump use has been effective.
The Patient With Technical Problems
An 84‐year‐old gentleman with T2DM was admitted for heart failure and aortic valve replacement. His HbA1c was 6.2%, and he had had several outpatient hypoglycemic events. While on CSII in the hospital, his point of care testing (POCT) glucose readings ranged from 105 to 260 mg/dL. On the afternoon of the third hospital day, POCT readings stayed above 220 mg/dL and rose to 348 mg/dL on the fourth hospital morning, despite multiple blousing and changing the insulin, insertion site, reservoir cartridge, and pump tubing. There was no evidence of infection and no medication change that would have impacted glucose levels. Review of his procedures revealed that he had undergone computed tomography (CT) on the morning of hospital day 3 and wore his pump while being scanned. The pump company was notified.
Key points:
- Patients, and medical and nursing staff, should be reminded to remove insulin pumps for CT scans, magnetic resonance imaging, x‐rays, or other tests with high electromagnetic fields.
- If there is a suspicion of pump malfunction from such a procedure, notify the pump company.
The Patient Who Can Benefit From Inpatient Education
A 70‐year‐old female with T1DM was admitted for heart failure. The patient had been using CSII for 20 years. Her latest HbA1c was 6.9%. She had 1 hypoglycemic event every 1 to 2 weeks. In the hospital, she experienced 2 hypoglycemic events within 3 days, both around bedtime. It was discovered that the patient was giving a bolus of insulin for elevated glucose levels based on the hospital POCT, and when the meal arrived (3060 minutes later), she again delivered a bolus based on her own glucometer reading plus insulin based on the carbohydrates in her meal. The patient was then instructed to request the POCT when her meal tray arrived, and she was taught how to use the pump's built‐in calculator. Glucose excursions improved.
Key points:
- Patients on CSII, though able to exercise autonomy in managing their insulin doses, may also need assistance in dosing insulin properly.
- Although pump education is ideally done on an outpatient basis, hospital‐based providers may encounter patients who need reinforcement of their training while hospitalized. Hospital personnel trained on insulin pumps (such as physicians, nurse practitioners, physician assistants, and certified diabetes educators) can help augment the patient's knowledge while in the hospital. In the absence of such key personnel, patient safety has to be addressed with re‐evaluation of the need to discontinue the pump and switch to multiple doses of subcutaneous insulin.
STEPS IN TAKING CARE OF PATIENTS ON CONTINUOUS SUBCUTANEOUS INSULIN INFUSION
Initial Patient Assessment
On admission, patients are asked whether they use an insulin pump. This is included in the nursing assessment form. If they do, the physician is notified.
The insulin pump might be missed unless specifically asked for because (1) the insulin pump may be thought of more as a device rather than a medication, and (2) the insulin pump may be worn in less obvious areas, not only on the abdomen where providers are more apt to detect it.
Hospital Policy and Insulin Orders
Written hospital policies on how to safely manage patients presenting with an insulin pump will delineate patients who can safely be allowed to continue on the pump, and the responsibilities that come along with this. Our institution has such a policy. Experts from both the legal and biomedical engineering departments were consulted when the policy was crafted. Patients must be fully alert, able, and willing to self‐manage the pump. General contraindications to pump use in the hospital, such as altered mental status or DKA, are listed in Table 1. In addition, patients in the intensive care units are best managed on an IV insulin infusion during their critical illness, in keeping with several society guidelines.[14] Controlling severe hyperglycemia and DKA with multiple boluses through the insulin pump can potentially lead to stacking of insulin with subsequent hypoglycemia.
| Altered state of consciousness |
| Suicidal ideation |
| Prolonged instability of blood glucose levels |
| Diabetic ketoacidosis |
| Patient/family inability or refusal to participate in own care |
| Insulin‐pump malfunction |
| Lack of appropriate supplies for the insulin pump |
| Other circumstances as identified by the physician, resident, or licensed independent practitioner |
In our institution, a computerized insulin pump order set has to be activated. Apart from insulin, POCT, and hypoglycemia‐management orders, this order set contains documents aimed at balancing patient autonomy with delivery of appropriate and safe medical care that the bedside nurse goes over with the patient (Table 2). By policy at our institution, insulin should be dispensed from the hospital's pharmacy (except for that already in the pump), so the order set is linked to the pharmacy and a 3‐mL insulin vial is delivered to the hospital floor and stored in the patient‐specific medication bin. The order set triggers an Endocrinology consult so that the patient can be assessed by key trained personnel.
|
| CSII pump therapy patient agreement |
| Delineates the conditions for continuing on CSII and those for whom it may be discontinued |
| Terms of use and release of liability for patient‐owned equipment |
| Delineates the patient's responsibility for the pump and supplies |
| Patient‐maintained flow sheet for inpatient CSII |
| Includes blood glucose levels (obtained by nurse or patient‐care assistant with the hospital glucose meter) |
| Includes insulin doses (basal, bolus) |
| Includes carbohydrate intake in grams |
Patient Diagnosis
It is important to try to distinguish T1DM vs T2DM, as patients with T1DM are prone to ketoacidosis when the pump is disconnected.
Patient Assessment by the Endocrinology Consult Service
Hospitalized patients on the pump have varying degrees of pump knowledge and skill sets. We have encountered highly trained patients who meticulously count their carbohydrates and double‐check the insulin doses calculated by the built‐in pump calculator, and those who have knowledge gaps because their physicians, and not they themselves, change pump settings at the clinic visits.
Therefore, the Endocrinology consult team members (comprising physicians, nurse practitioners, and certified diabetes educators) go through the following items to be able to order the insulin correctly, assess whether patients are still able and willing to continue on their pump despite their illness, formulate alternative insulin regimens as needed, or help empower patients who may have forgotten some aspects of pump management:
- Insulin pump manufacturer/model.
- Insulin used in the pump.
- Often fast‐acting insulin (lispro, aspart, or glulisine).
- Some patients use regular insulin.
- A few patients use U500 insulin (5 more potent than other insulins).
- Insulin doses/pump settings.
Patients are assessed for:
- Hypoglycemia awareness.
- Previous glucose control.
- Bolus calculation (either using the built‐in calculator, computing this mentally, or using a different calculator).
- Ability to deliver a bolus (including vision and dexterity challenges).
- Ability to change the basal rate, or set a temporary rate, and suspend insulin delivery.
Discussion on Options for Inpatient Management
After assessment, education is provided as needed. If there are concerns on the part of the patient, the primary team, or the Endocrinology team about safe continuation of CSII during the hospitalization, then alternative insulin regimens are discussed. Patients who cannot access their basal rates and cannot adjust the doses are not able to self‐manage; they should be taken off the pump and treated with multiple subcutaneous insulin doses. Conversion to MDII is discussed under Interruption of Continuous Subcutaneous Insulin Infusion for Short and Prolonged Periods.
Provision of Pump Information for Hospital Healthcare Providers
Users of CSII, even if perfectly competent using their pumps in the ambulatory setting, may need assistance in the hospital for various reasons. They may not know what to do for surgical or radiologic procedures (discussed below) and may not be familiar with hospital policies involving CSII. Hospital providers trained on insulin pumps may need a refresher on locating a particular pump setting.
The provider can call the toll‐free number on the back of the pump for assistance (Table 3). Insulin‐pump companies also have menu maps to aid in finding information on pump settings (samples shown in Figures 1 and 2).[12, 13] Documentation of the patient's pump settings will assist in CSII dose changes during the acute illness or assist in switching to MDII if needed. The following information need to be collected:
| Animas Corporation | 877‐937‐7867 |
| Insulet Corporation | 800‐591‐3455 |
| Medtronic | 800‐826‐2099 |
| Roche Diagnostics | 800‐688‐4578 |
Basal Rate
This is the hourly insulin rate delivered for the patient's insulin needs even when not eating. The patient might have one or multiple basal rates in a day, or a different pattern on some days. Because the patient's activity in the hospital will be different from the usual ambulatory activity, we recommend that patients choose only 1 pattern.
Bolus
This is the insulin to cover meals or to correct for hyperglycemia, or both. The patient has to activate buttons for delivery. The patient may or may not be using the built‐in pump calculator.
Carbohydrate Ratio
This is the amount of insulin per quantity of carbohydrate consumed. When patients are initially placed on the insulin pump, they are given a carbohydrate ratio that is derived from a calculation called the rule of 500. In the rule of 500, the number 500 is divided by the patient's total daily insulin dose while on multiple subcutaneous insulin shots. For example, if the patient was on insulin glargine 13 units daily and insulin lispro 4 units 3 daily with meals, 500 divided by 25 gives us a carbohydrate ratio of 20 grams of carbohydrate for 1 unit of insulin (or conversely called insulin‐to‐carbohydrate ratio of 1 unit of insulin for every 20 grams of carbohydrate).
This often comes out to 1 unit for every 1530 grams of carbohydrates in patients with T1DM, and 1 unit for every 515 grams of carbohydrate in patients with T2DM, reflecting the need for a higher insulin dose in the latter.
It is best to ask the patient how many units he or she usually takes for a meal, or to present the patient with an example of a meal and ask how much he or she would take. We have encountered a patient whose carbohydrate ratio was 1, but upon further inquiry, the patient demonstrated that he actually bolused 1 unit for every 1 serving (or 1 unit for every 15 grams) of carbohydrate.
Sensitivity Factor
This is the amount of insulin that would bring the blood glucose to goal. For example, if the patient requires 1 unit of insulin to bring down the blood glucose from 170 mg/dL to 120 mg/dL, then the sensitivity factor of 50 would be seen on the pump screen. Similar to the carbohydrate ratio, a sensitivity factor is calculated when patients are initially placed on the insulin pump. This time, the rule of 1800 is used, where the number 1800 is divided by the patient's total daily insulin dose. In patients with T1DM, this often comes to 30100 mg/dL per 1 unit of insulin; or, conversely, 1 unit for every 30100 mg/dL glucose. For patients with T2DM, this is often 1 unit for every 1025 mg/dL glucose.
This insulin dose is given in addition to the dose resulting from the carbohydrate ratio, or alone if the patient is not eating.
Target
This is the blood glucose goal for the patient, which might be too tight in the presence of acute illness, and therefore would have to be modified. The American Diabetes Association, Endocrine Society, and American Association of Clinical Endocrinologists recommend premeal glucose targets of <140 mg/dL in hospitalized noncritically ill patients on insulin, with re‐evaluation of the insulin dose when premeal glucose levels fall below 100 mg/dL and dose adjustment if they are <70mg/dLunless there is an obvious explanation, such as a missed meal.[14, 15]
Point‐of‐Care Testing for Glucose Monitoring
Our policy specifies that the hospital glucose meter is the meter of record upon which dose adjustments are based. Point‐of‐care testing is performed by our patient care nursing assistants or bedside nurses. The timing is typically before meals, at bedtime, between 2 and 3 AM, and with allowance for other times that patients are used to checking when they were at home, such as after meals. Frequent POCT has to be especially borne in mind for patients with hypoglycemia unawareness. Some patients are used to checking with their own home glucose meters in between these times, and we do work with them with the understanding that dose‐change decisions are based on the hospital glucose meter readings.
Dose Adjustments
Continuous subcutaneous insulin infusion dose adjustments for hypoglycemia and hyperglycemia are usually done in 10% to 20% decrements/emncrements. Our Endocrinology team discusses these with the patients and ensures that the new settings are entered into the pump and into the order set.
INTERRUPTION OF CONTINUOUS SUBCUTANEOUS INSULIN INFUSION FOR SHORT AND PROLONGED PERIODS
Patients with T1DM should not be left without basal insulin. However, pump interruption for 30 minutes to an hour often does not lead to problems. Beyond an hour and certainly closer to 2 to 3 hours off the pump, the patient should be given a subcutaneous insulin injection if the patient is left without easy access to the insulin pump.
The subcutaneous insulin dose for temporary pump suspension can be roughly calculated as hourly basal rate multiplier, where the multiplier is the number of hours that the patient is expected to be disconnected from the pump (for example, hourly basal rate of 0.85 unit/hour 3 hours = 2.55 units, which can be rounded off to 2 or 3 units depending on the patient's general glucose control).
When it has been determined that the patient should come off the pump for substantial periods of time, then subcutaneous insulin injections should be given. This is imperative for the prevention of DKA in patients with T1DM, and highly recommended for maintenance of good glycemic control for patients with T2DM.
Basal Dose
In most cases, these situations stretch for greater than 24 hours, such as surgery and the anticipated recovery from anesthesia. We favor long‐acting insulin for basal needs, given 2 hours before discontinuing the pump. The total basal insulin dose per day can be given as the starting long‐acting insulin dose and then adjusted as needed. The total daily basal insulin dose can be retrieved from the insulin pump. In one study on T1DM patients using insulin lispro on the pump, total daily basal dose was given as insulin glargine without adverse effects.[16] If there is concern for hypoglycemia, then the dose can be reduced by 10% to 20%. Care should be made to ascertain that the basal insulin delivered via the pump is appropriate.
There are several ways to estimate this:
- If the daily total basal and the total prandial insulin requirements approximate a 50:50 ratio, then the basal rate via the pump is appropriate.
- If the daily total basal rate via the pump is similar to a weight‐based estimate of the basal dose (often 0.150.2 units/kg/day in patients with T1DM, 0.20.3 units/kg/day for T2DM, and higher in both cases with longer duration of DM or greater insulin resistance), then the basal rate via the pump is appropriate.
Bolus Dose
Patients can still continue to calculate their sensitivity factor and carbohydrate ratio and request for the equivalent dose of insulin. In the ideal situation, bedside nurses would be taught how to calculate this ratio and dose rapid‐acting insulin accordingly should the patient need to come off the insulin pump. Because of the logistic difficulties of making this uniform in our institution, we have worked instead on providing patients with information on the carbohydrate content of their meal tray. If the pump is discontinued, the patient would continue to calculate their prandial insulin based on their carbohydrates ratio and indicate to the nurse how much he/she would need. Our subcutaneous insulin orders for MDII allow for us to put a range of insulin doses based on the patient's typical insulin needs for mealtime.
Pump Removal for Certain Hospital Procedures
Patients may not remember that the pump has to be removed before entering high‐radiation areas. The pump owner manuals tell patients not to use the pump when going for magnetic resonance imaging, CT scans, or x‐rays, or near equipment with high electromagnetic fields.
Interrogation of the Pump
If there is concern about pump malfunction, patients should be switched to MDII. The pump company can be contacted for pump interrogation and provision of a temporary pump (Table 3).
Pump Disconnection
In our institution, the Radiology department has signage instructing patients to inform the technician if they are wearing an insulin pump. The pump is handed off to a family member or stored until the procedure is over. Another option is to leave it with the bedside nurse or the floor nurse manager for safekeeping. This is less ideal, because the wait for the radiologic procedure might take longer than expected and the patient is left without any insulin on board.
Interruption of Continuous Subcutaneous Insulin Infusion for Surgical Procedures
In our institution, the anesthesiology department has worked with the Endocrinology, Surgery and Medicine departments regarding patients with insulin pumps. Discontinuation of CSII is recommended for surgical procedures longer than 1 hour; patients are asked to continue on the insulin pump until they are taken to the preoperative suite, at which point they are placed on IV insulin infusion. Ideally, there should be an overlap of 1530 minutes. Providing an alternative continuous source of insulin during pump interruption is important, especially for patients with T1DM.[17]
Pump Resumption
Once the patient is ready to resume the pump, any subcutaneous insulin that was delivered and might still be active has to be accounted for and subtracted from the basal pump dose so that hypoglycemia is avoided. An alternative would be to wait until the last basal subcutaneous insulin dose is expected to be cleared before restarting CSII.
SUMMARY
As patients on an insulin pump are increasingly seen in the hospital, inpatient providers have to be able to adapt to these patients' needs. Inpatient providers need to have a working knowledge of the insulin pump. Alternative methods of insulin delivery will have to be discussed with the patient to assure continued safety in the hospital.
Disclosures
Dr. Lansang has served as a Sanofi Advisory Board member.
Delivery of insulin via continuous subcutaneous insulin infusion (CSII), or insulin pump, has gained wide acceptance. It is estimated that 400,000 patients with type 1 diabetes mellitus (T1DM) are using insulin pumps.[1] A registry for T1DM in the United States indicated that 50% of the 25,833 participants were using an insulin pump.[2] Use of CSII in type 2 DM (T2DM) is also increasing.[3]
Patients with DM are 3 as likely to be hospitalized than patients without DM.[4] Twenty percent to 30% of adult hospitalized patients have a known diagnosis of DM.[5] It is therefore to be expected that patients on CSII will be seen with increased frequency in the hospital setting. This leads to potential difficultiesinpatient healthcare providers may not be familiar with insulin pumps, and patients may not be aware of complexities associated with pump usage in the hospital.
This article will review CSII usage in the hospital, offering strategies for management in partnership with the patient based on our experiences and processes developed in our institution.
SHOULD CONTINUOUS SUBCUTANEOUS INSULIN INFUSION BE CONTINUED IN THE HOSPITAL?
The American Diabetes Association advocates (1) allowing patients who are physically and mentally able to continue CSII when hospitalized, (2) having a hospital policy for CSII use, and (3) having hospital personnel with expertise on pump management.[6] The American Association of Clinical Endocrinologists echoes much of the same and suggests contacting the specialist responsible for the pump in the ambulatory setting for decisions on adjustments in the hospitalized patient,[7] which at times may not be feasible.
The logic and benefits of basal‐bolus insulin dosing (ie, giving basal insulin to account for fasting requirements, plus bolus insulin to cover nutritional and correctional needs) have been well‐described.[8, 9, 10] In randomized clinical trials on patients admitted to general medical and surgical floors, basal‐bolus insulin (long‐acting basal insulin plus mealtime fast‐acting insulin injections) resulted in better glycemic control and reduced infection rates compared with sliding‐scale therapy (waiting for high blood glucoses before giving insulin, instead of giving it proactively to prevent hyperglycemia).[9, 10] At present, insulin delivery via the insulin pump is the best commercially available method to deliver insulin in a basal‐bolus manner in ambulatory patients. It thus makes sense to continue CSII in the hospital if patients are able to manage their pumps, though there are no randomized trials answering this question as of yet.
Studies on insulin pumps in the hospital are sparse. In one group's latest retrospective study of 136 patients over a 6‐year period, CSII was continued during the entire hospital stay in 65% of the hospitalizations, was used intermittently in 20%, and was discontinued with alternative insulin regimens given in 15%.[11] Mean glucose was 178 47 mg/dL (mean standard deviation), with no significant difference between groups. There were fewer episodes of severe hyperglycemia among those who continued on the pump compared with the other groups, and fewer episodes of hypoglycemia in those who continued on vs those taken off the pump. There was 1 episode of an infusion catheter kinking, resulting in nonfatal hyperglycemia, but no reported pump‐site infections, mechanical pump failure, or diabetic ketoacidosis (DKA) among patients remaining on CSII.
CLINICAL VIGNETTES
The following cases illustrate potential challenges with CSII use that we have encountered in the hospital.
The Patient Needing Transition to Multiple Daily Insulin Injections
A 29‐year‐old male with T1DM, on CSII, was admitted for fever and chills. His latest glycated hemoglobin (HbA1c) level was 6.8%. His glucose levels started rising, and he wished to be taken off the insulin pump. He was started by the primary team on multiple daily insulin injections (MDII) with insulin glargine and insulin lispro. His glucose levels continued to rise, so an intravenous (IV) insulin infusion was started. Endocrinology was then consulted. The patient's condition was concerning for the potential development of DKA, so he was kept on IV insulin. When he was ready for transitioning to subcutaneous insulin, the pump had been taken home by a family member, and the patient could not recall his CSII basal rates but knew his total basal insulin dose, carbohydrate ratio, and sensitivity factor. Endocrinology assisted in transitioning him from the insulin infusion to MDII based on these recalled doses. When the insulin pump was available, the pump settings were interrogated, and he was transitioned back to it.
Key points:
- Having key hospital personnel trained on CSII, including interrogating the pump's settings, facilitates the transitioning of these patients from one hospital unit, or level of care, to another.
- Accessing the pump's settings involves pushing several buttons on the insulin pump. Because key hospital personnel will encounter patients on different insulin pumps, it may be helpful to keep menu maps handy as a quick reference. A menu map will show at a glance where certain information can be found, such as the basal insulin rate or the sensitivity factor (see examples in Figures 1 and 2).[12, 13]
- Knowing the HbA1c will help determine if pump use has been effective.
The Patient With Technical Problems
An 84‐year‐old gentleman with T2DM was admitted for heart failure and aortic valve replacement. His HbA1c was 6.2%, and he had had several outpatient hypoglycemic events. While on CSII in the hospital, his point of care testing (POCT) glucose readings ranged from 105 to 260 mg/dL. On the afternoon of the third hospital day, POCT readings stayed above 220 mg/dL and rose to 348 mg/dL on the fourth hospital morning, despite multiple blousing and changing the insulin, insertion site, reservoir cartridge, and pump tubing. There was no evidence of infection and no medication change that would have impacted glucose levels. Review of his procedures revealed that he had undergone computed tomography (CT) on the morning of hospital day 3 and wore his pump while being scanned. The pump company was notified.
Key points:
- Patients, and medical and nursing staff, should be reminded to remove insulin pumps for CT scans, magnetic resonance imaging, x‐rays, or other tests with high electromagnetic fields.
- If there is a suspicion of pump malfunction from such a procedure, notify the pump company.
The Patient Who Can Benefit From Inpatient Education
A 70‐year‐old female with T1DM was admitted for heart failure. The patient had been using CSII for 20 years. Her latest HbA1c was 6.9%. She had 1 hypoglycemic event every 1 to 2 weeks. In the hospital, she experienced 2 hypoglycemic events within 3 days, both around bedtime. It was discovered that the patient was giving a bolus of insulin for elevated glucose levels based on the hospital POCT, and when the meal arrived (3060 minutes later), she again delivered a bolus based on her own glucometer reading plus insulin based on the carbohydrates in her meal. The patient was then instructed to request the POCT when her meal tray arrived, and she was taught how to use the pump's built‐in calculator. Glucose excursions improved.
Key points:
- Patients on CSII, though able to exercise autonomy in managing their insulin doses, may also need assistance in dosing insulin properly.
- Although pump education is ideally done on an outpatient basis, hospital‐based providers may encounter patients who need reinforcement of their training while hospitalized. Hospital personnel trained on insulin pumps (such as physicians, nurse practitioners, physician assistants, and certified diabetes educators) can help augment the patient's knowledge while in the hospital. In the absence of such key personnel, patient safety has to be addressed with re‐evaluation of the need to discontinue the pump and switch to multiple doses of subcutaneous insulin.
STEPS IN TAKING CARE OF PATIENTS ON CONTINUOUS SUBCUTANEOUS INSULIN INFUSION
Initial Patient Assessment
On admission, patients are asked whether they use an insulin pump. This is included in the nursing assessment form. If they do, the physician is notified.
The insulin pump might be missed unless specifically asked for because (1) the insulin pump may be thought of more as a device rather than a medication, and (2) the insulin pump may be worn in less obvious areas, not only on the abdomen where providers are more apt to detect it.
Hospital Policy and Insulin Orders
Written hospital policies on how to safely manage patients presenting with an insulin pump will delineate patients who can safely be allowed to continue on the pump, and the responsibilities that come along with this. Our institution has such a policy. Experts from both the legal and biomedical engineering departments were consulted when the policy was crafted. Patients must be fully alert, able, and willing to self‐manage the pump. General contraindications to pump use in the hospital, such as altered mental status or DKA, are listed in Table 1. In addition, patients in the intensive care units are best managed on an IV insulin infusion during their critical illness, in keeping with several society guidelines.[14] Controlling severe hyperglycemia and DKA with multiple boluses through the insulin pump can potentially lead to stacking of insulin with subsequent hypoglycemia.
| Altered state of consciousness |
| Suicidal ideation |
| Prolonged instability of blood glucose levels |
| Diabetic ketoacidosis |
| Patient/family inability or refusal to participate in own care |
| Insulin‐pump malfunction |
| Lack of appropriate supplies for the insulin pump |
| Other circumstances as identified by the physician, resident, or licensed independent practitioner |
In our institution, a computerized insulin pump order set has to be activated. Apart from insulin, POCT, and hypoglycemia‐management orders, this order set contains documents aimed at balancing patient autonomy with delivery of appropriate and safe medical care that the bedside nurse goes over with the patient (Table 2). By policy at our institution, insulin should be dispensed from the hospital's pharmacy (except for that already in the pump), so the order set is linked to the pharmacy and a 3‐mL insulin vial is delivered to the hospital floor and stored in the patient‐specific medication bin. The order set triggers an Endocrinology consult so that the patient can be assessed by key trained personnel.
|
| CSII pump therapy patient agreement |
| Delineates the conditions for continuing on CSII and those for whom it may be discontinued |
| Terms of use and release of liability for patient‐owned equipment |
| Delineates the patient's responsibility for the pump and supplies |
| Patient‐maintained flow sheet for inpatient CSII |
| Includes blood glucose levels (obtained by nurse or patient‐care assistant with the hospital glucose meter) |
| Includes insulin doses (basal, bolus) |
| Includes carbohydrate intake in grams |
Patient Diagnosis
It is important to try to distinguish T1DM vs T2DM, as patients with T1DM are prone to ketoacidosis when the pump is disconnected.
Patient Assessment by the Endocrinology Consult Service
Hospitalized patients on the pump have varying degrees of pump knowledge and skill sets. We have encountered highly trained patients who meticulously count their carbohydrates and double‐check the insulin doses calculated by the built‐in pump calculator, and those who have knowledge gaps because their physicians, and not they themselves, change pump settings at the clinic visits.
Therefore, the Endocrinology consult team members (comprising physicians, nurse practitioners, and certified diabetes educators) go through the following items to be able to order the insulin correctly, assess whether patients are still able and willing to continue on their pump despite their illness, formulate alternative insulin regimens as needed, or help empower patients who may have forgotten some aspects of pump management:
- Insulin pump manufacturer/model.
- Insulin used in the pump.
- Often fast‐acting insulin (lispro, aspart, or glulisine).
- Some patients use regular insulin.
- A few patients use U500 insulin (5 more potent than other insulins).
- Insulin doses/pump settings.
Patients are assessed for:
- Hypoglycemia awareness.
- Previous glucose control.
- Bolus calculation (either using the built‐in calculator, computing this mentally, or using a different calculator).
- Ability to deliver a bolus (including vision and dexterity challenges).
- Ability to change the basal rate, or set a temporary rate, and suspend insulin delivery.
Discussion on Options for Inpatient Management
After assessment, education is provided as needed. If there are concerns on the part of the patient, the primary team, or the Endocrinology team about safe continuation of CSII during the hospitalization, then alternative insulin regimens are discussed. Patients who cannot access their basal rates and cannot adjust the doses are not able to self‐manage; they should be taken off the pump and treated with multiple subcutaneous insulin doses. Conversion to MDII is discussed under Interruption of Continuous Subcutaneous Insulin Infusion for Short and Prolonged Periods.
Provision of Pump Information for Hospital Healthcare Providers
Users of CSII, even if perfectly competent using their pumps in the ambulatory setting, may need assistance in the hospital for various reasons. They may not know what to do for surgical or radiologic procedures (discussed below) and may not be familiar with hospital policies involving CSII. Hospital providers trained on insulin pumps may need a refresher on locating a particular pump setting.
The provider can call the toll‐free number on the back of the pump for assistance (Table 3). Insulin‐pump companies also have menu maps to aid in finding information on pump settings (samples shown in Figures 1 and 2).[12, 13] Documentation of the patient's pump settings will assist in CSII dose changes during the acute illness or assist in switching to MDII if needed. The following information need to be collected:
| Animas Corporation | 877‐937‐7867 |
| Insulet Corporation | 800‐591‐3455 |
| Medtronic | 800‐826‐2099 |
| Roche Diagnostics | 800‐688‐4578 |
Basal Rate
This is the hourly insulin rate delivered for the patient's insulin needs even when not eating. The patient might have one or multiple basal rates in a day, or a different pattern on some days. Because the patient's activity in the hospital will be different from the usual ambulatory activity, we recommend that patients choose only 1 pattern.
Bolus
This is the insulin to cover meals or to correct for hyperglycemia, or both. The patient has to activate buttons for delivery. The patient may or may not be using the built‐in pump calculator.
Carbohydrate Ratio
This is the amount of insulin per quantity of carbohydrate consumed. When patients are initially placed on the insulin pump, they are given a carbohydrate ratio that is derived from a calculation called the rule of 500. In the rule of 500, the number 500 is divided by the patient's total daily insulin dose while on multiple subcutaneous insulin shots. For example, if the patient was on insulin glargine 13 units daily and insulin lispro 4 units 3 daily with meals, 500 divided by 25 gives us a carbohydrate ratio of 20 grams of carbohydrate for 1 unit of insulin (or conversely called insulin‐to‐carbohydrate ratio of 1 unit of insulin for every 20 grams of carbohydrate).
This often comes out to 1 unit for every 1530 grams of carbohydrates in patients with T1DM, and 1 unit for every 515 grams of carbohydrate in patients with T2DM, reflecting the need for a higher insulin dose in the latter.
It is best to ask the patient how many units he or she usually takes for a meal, or to present the patient with an example of a meal and ask how much he or she would take. We have encountered a patient whose carbohydrate ratio was 1, but upon further inquiry, the patient demonstrated that he actually bolused 1 unit for every 1 serving (or 1 unit for every 15 grams) of carbohydrate.
Sensitivity Factor
This is the amount of insulin that would bring the blood glucose to goal. For example, if the patient requires 1 unit of insulin to bring down the blood glucose from 170 mg/dL to 120 mg/dL, then the sensitivity factor of 50 would be seen on the pump screen. Similar to the carbohydrate ratio, a sensitivity factor is calculated when patients are initially placed on the insulin pump. This time, the rule of 1800 is used, where the number 1800 is divided by the patient's total daily insulin dose. In patients with T1DM, this often comes to 30100 mg/dL per 1 unit of insulin; or, conversely, 1 unit for every 30100 mg/dL glucose. For patients with T2DM, this is often 1 unit for every 1025 mg/dL glucose.
This insulin dose is given in addition to the dose resulting from the carbohydrate ratio, or alone if the patient is not eating.
Target
This is the blood glucose goal for the patient, which might be too tight in the presence of acute illness, and therefore would have to be modified. The American Diabetes Association, Endocrine Society, and American Association of Clinical Endocrinologists recommend premeal glucose targets of <140 mg/dL in hospitalized noncritically ill patients on insulin, with re‐evaluation of the insulin dose when premeal glucose levels fall below 100 mg/dL and dose adjustment if they are <70mg/dLunless there is an obvious explanation, such as a missed meal.[14, 15]
Point‐of‐Care Testing for Glucose Monitoring
Our policy specifies that the hospital glucose meter is the meter of record upon which dose adjustments are based. Point‐of‐care testing is performed by our patient care nursing assistants or bedside nurses. The timing is typically before meals, at bedtime, between 2 and 3 AM, and with allowance for other times that patients are used to checking when they were at home, such as after meals. Frequent POCT has to be especially borne in mind for patients with hypoglycemia unawareness. Some patients are used to checking with their own home glucose meters in between these times, and we do work with them with the understanding that dose‐change decisions are based on the hospital glucose meter readings.
Dose Adjustments
Continuous subcutaneous insulin infusion dose adjustments for hypoglycemia and hyperglycemia are usually done in 10% to 20% decrements/emncrements. Our Endocrinology team discusses these with the patients and ensures that the new settings are entered into the pump and into the order set.
INTERRUPTION OF CONTINUOUS SUBCUTANEOUS INSULIN INFUSION FOR SHORT AND PROLONGED PERIODS
Patients with T1DM should not be left without basal insulin. However, pump interruption for 30 minutes to an hour often does not lead to problems. Beyond an hour and certainly closer to 2 to 3 hours off the pump, the patient should be given a subcutaneous insulin injection if the patient is left without easy access to the insulin pump.
The subcutaneous insulin dose for temporary pump suspension can be roughly calculated as hourly basal rate multiplier, where the multiplier is the number of hours that the patient is expected to be disconnected from the pump (for example, hourly basal rate of 0.85 unit/hour 3 hours = 2.55 units, which can be rounded off to 2 or 3 units depending on the patient's general glucose control).
When it has been determined that the patient should come off the pump for substantial periods of time, then subcutaneous insulin injections should be given. This is imperative for the prevention of DKA in patients with T1DM, and highly recommended for maintenance of good glycemic control for patients with T2DM.
Basal Dose
In most cases, these situations stretch for greater than 24 hours, such as surgery and the anticipated recovery from anesthesia. We favor long‐acting insulin for basal needs, given 2 hours before discontinuing the pump. The total basal insulin dose per day can be given as the starting long‐acting insulin dose and then adjusted as needed. The total daily basal insulin dose can be retrieved from the insulin pump. In one study on T1DM patients using insulin lispro on the pump, total daily basal dose was given as insulin glargine without adverse effects.[16] If there is concern for hypoglycemia, then the dose can be reduced by 10% to 20%. Care should be made to ascertain that the basal insulin delivered via the pump is appropriate.
There are several ways to estimate this:
- If the daily total basal and the total prandial insulin requirements approximate a 50:50 ratio, then the basal rate via the pump is appropriate.
- If the daily total basal rate via the pump is similar to a weight‐based estimate of the basal dose (often 0.150.2 units/kg/day in patients with T1DM, 0.20.3 units/kg/day for T2DM, and higher in both cases with longer duration of DM or greater insulin resistance), then the basal rate via the pump is appropriate.
Bolus Dose
Patients can still continue to calculate their sensitivity factor and carbohydrate ratio and request for the equivalent dose of insulin. In the ideal situation, bedside nurses would be taught how to calculate this ratio and dose rapid‐acting insulin accordingly should the patient need to come off the insulin pump. Because of the logistic difficulties of making this uniform in our institution, we have worked instead on providing patients with information on the carbohydrate content of their meal tray. If the pump is discontinued, the patient would continue to calculate their prandial insulin based on their carbohydrates ratio and indicate to the nurse how much he/she would need. Our subcutaneous insulin orders for MDII allow for us to put a range of insulin doses based on the patient's typical insulin needs for mealtime.
Pump Removal for Certain Hospital Procedures
Patients may not remember that the pump has to be removed before entering high‐radiation areas. The pump owner manuals tell patients not to use the pump when going for magnetic resonance imaging, CT scans, or x‐rays, or near equipment with high electromagnetic fields.
Interrogation of the Pump
If there is concern about pump malfunction, patients should be switched to MDII. The pump company can be contacted for pump interrogation and provision of a temporary pump (Table 3).
Pump Disconnection
In our institution, the Radiology department has signage instructing patients to inform the technician if they are wearing an insulin pump. The pump is handed off to a family member or stored until the procedure is over. Another option is to leave it with the bedside nurse or the floor nurse manager for safekeeping. This is less ideal, because the wait for the radiologic procedure might take longer than expected and the patient is left without any insulin on board.
Interruption of Continuous Subcutaneous Insulin Infusion for Surgical Procedures
In our institution, the anesthesiology department has worked with the Endocrinology, Surgery and Medicine departments regarding patients with insulin pumps. Discontinuation of CSII is recommended for surgical procedures longer than 1 hour; patients are asked to continue on the insulin pump until they are taken to the preoperative suite, at which point they are placed on IV insulin infusion. Ideally, there should be an overlap of 1530 minutes. Providing an alternative continuous source of insulin during pump interruption is important, especially for patients with T1DM.[17]
Pump Resumption
Once the patient is ready to resume the pump, any subcutaneous insulin that was delivered and might still be active has to be accounted for and subtracted from the basal pump dose so that hypoglycemia is avoided. An alternative would be to wait until the last basal subcutaneous insulin dose is expected to be cleared before restarting CSII.
SUMMARY
As patients on an insulin pump are increasingly seen in the hospital, inpatient providers have to be able to adapt to these patients' needs. Inpatient providers need to have a working knowledge of the insulin pump. Alternative methods of insulin delivery will have to be discussed with the patient to assure continued safety in the hospital.
Disclosures
Dr. Lansang has served as a Sanofi Advisory Board member.
- http://www.rncos.com/Press_Releases/US‐to‐Dominate‐the‐Global‐Insulin‐Pump‐Market.htm. Accessed on October 25, 2013.
- , , , , , ;T1D Exchange Clinic Network. The T1D exchange clinic registry. J Clin Endocrinol Metab. 2012;97(12):4383–4389.
- , , , et al. Resource utilization with insulin pump therapy for type 2 diabetes mellitus. Am J Manag Care. 2010;16(12):892–896.
- , , , , . Diabetes‐related hospitalization and hospital‐related utilization. In: Diabetes in America. 2nd ed. Bethesda, MD: National Diabetes Data Group, National Institute of Diabetes and Digestive and Kidney Diseases; 1995:553–569. Available at: http://diabetes.niddk.nih.gov/dm/pubs/america/pdf/chapter27.pdf. Accessed February 21, 2013.
- , , , , , . Hyperglycemia: an independent marker of in‐hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978–982.
- American Diabetes Association. Standards of medical care in diabetes—2012. Diabetes Care. 2012;35(suppl 1):S11–S63.
- , , , et al;AACE Insulin Pump Management Task Force. Statement by the American Association of Clinical Endocrinologists Consensus Panel on insulin pump management. Endocr Pract. 2010;16(5):746–762.
- , , , et al;American Diabetes Association Diabetes in Hospitals Writing Committee. Management of diabetes and hyperglycemia in hospitals [published correction appears in Diabetes Care. 2004;27(5):1255]. Diabetes Care. 2004;27(2):553–591.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes (RABBIT 2 trial). Diabetes Care. 2007;30(9):2181–2186.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes undergoing general surgery (RABBIT 2 surgery). Diabetes Care. 2011;34(2):256–261.
- , , , , , . Transitioning insulin pump therapy from the outpatient to the inpatient setting: a review of 6 years' experience with 253 cases. J Diabetes Sci Technol. 2012;6(5):995–1002.
- Medtronic Paradigm Revel insulin pump [menu map]. Northridge, CA: Medtronic. Available at: http://www.medtronicdiabetes.com/sites/default/files/library/download‐library/workbooks/x23_menu_map.pdf. Updated January 22, 2010. Accessed February 2013.
- OneTouch Ping insulin pump [menu map]. West Chester, PA: Animas Corporation.
- , , , et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Endocr Pract. 2009;15(4):353–369.
- , , , et al. Management of hyperglycemia in hospitalized patients in non‐critical care setting: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38.
- , , , . Switch to multiple daily injections with insulin glargine and insulin lispro from continuous subcutaneous insulin infusion with insulin lispro: a randomized, open‐label study using a continuous glucose monitoring system. Endocr Pract. 2005;11(3):157–164.
- , , , , . Perioperative glycemic management in insulin pump patients undergoing noncardiac surgery. Curr Pharm Des. 2012;18(38):6204–6214.
- http://www.rncos.com/Press_Releases/US‐to‐Dominate‐the‐Global‐Insulin‐Pump‐Market.htm. Accessed on October 25, 2013.
- , , , , , ;T1D Exchange Clinic Network. The T1D exchange clinic registry. J Clin Endocrinol Metab. 2012;97(12):4383–4389.
- , , , et al. Resource utilization with insulin pump therapy for type 2 diabetes mellitus. Am J Manag Care. 2010;16(12):892–896.
- , , , , . Diabetes‐related hospitalization and hospital‐related utilization. In: Diabetes in America. 2nd ed. Bethesda, MD: National Diabetes Data Group, National Institute of Diabetes and Digestive and Kidney Diseases; 1995:553–569. Available at: http://diabetes.niddk.nih.gov/dm/pubs/america/pdf/chapter27.pdf. Accessed February 21, 2013.
- , , , , , . Hyperglycemia: an independent marker of in‐hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978–982.
- American Diabetes Association. Standards of medical care in diabetes—2012. Diabetes Care. 2012;35(suppl 1):S11–S63.
- , , , et al;AACE Insulin Pump Management Task Force. Statement by the American Association of Clinical Endocrinologists Consensus Panel on insulin pump management. Endocr Pract. 2010;16(5):746–762.
- , , , et al;American Diabetes Association Diabetes in Hospitals Writing Committee. Management of diabetes and hyperglycemia in hospitals [published correction appears in Diabetes Care. 2004;27(5):1255]. Diabetes Care. 2004;27(2):553–591.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes (RABBIT 2 trial). Diabetes Care. 2007;30(9):2181–2186.
- , , , et al. Randomized study of basal‐bolus insulin therapy in the inpatient management of patients with type 2 diabetes undergoing general surgery (RABBIT 2 surgery). Diabetes Care. 2011;34(2):256–261.
- , , , , , . Transitioning insulin pump therapy from the outpatient to the inpatient setting: a review of 6 years' experience with 253 cases. J Diabetes Sci Technol. 2012;6(5):995–1002.
- Medtronic Paradigm Revel insulin pump [menu map]. Northridge, CA: Medtronic. Available at: http://www.medtronicdiabetes.com/sites/default/files/library/download‐library/workbooks/x23_menu_map.pdf. Updated January 22, 2010. Accessed February 2013.
- OneTouch Ping insulin pump [menu map]. West Chester, PA: Animas Corporation.
- , , , et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Endocr Pract. 2009;15(4):353–369.
- , , , et al. Management of hyperglycemia in hospitalized patients in non‐critical care setting: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38.
- , , , . Switch to multiple daily injections with insulin glargine and insulin lispro from continuous subcutaneous insulin infusion with insulin lispro: a randomized, open‐label study using a continuous glucose monitoring system. Endocr Pract. 2005;11(3):157–164.
- , , , , . Perioperative glycemic management in insulin pump patients undergoing noncardiac surgery. Curr Pharm Des. 2012;18(38):6204–6214.
Rounding Frequency and Length of Stay
Hospitals are facing growing pressures to operate more efficiently, spurring interest in improving patient flow from the emergency department (ED) to inpatient unit to home. Children's hospitals are often at high occupancy,[1] and EDs are increasingly an entry point for hospital admission.[2, 3]
Among children who require brief episodes of hospital‐based care, hospital processes, especially those associated with discharge, can greatly impact length of stay (LOS). Patients ready for discharge from inpatient units are typically identified through formal physician‐led rounds, in contrast to EDs where discharges occur on a more continual basis. Quantitative descriptions of rounding frequency and LOS are lacking.
The focus of this study was the population of children who had visits for select respiratory illnesses (e.g., asthma, bronchiolitis, pneumonia, and croup) for which there is general consensus regarding treatment, admission, and discharge criteria.[4, 5, 6, 7] The selected illnesses represent common reasons for ED visits[8] and hospitalizations.[9] Hospital stays for these conditions tend to be brief, often 1 to 2 days in duration,[10] and repeated assessments are necessary to determine suitability for discharge.
The primary objectives of this study were to compare discharge patterns in the ED and inpatient settings and to quantify the relationship between discharge timing and LOS in these different clinical settings. A simulation was then used to predict the effect of the timing and frequency of physician‐led rounds on hospital LOS. We hypothesized that increased frequency of simulated physician‐led rounds would lead to meaningful reductions in predicted hospital LOS for children admitted from the ED.
METHODS
Retrospective analyses were conducted using hospital administrative data from pediatric ED visits and resultant inpatient stays. The University of Michigan Institutional Review Board approved the study.
Setting
C. S. Mott Children's Hospital at the University of Michigan is a suburban academic, tertiary care hospital located in Ann Arbor, Michigan. The pediatric emergency department had approximately 20,000 visits per year during the study period, and children were treated in an 11‐bed area. There were 106 pediatric inpatient general care beds during the study years.
Visit Selection
ED visits made by children <18 years old during the 3‐year period between May 1, 2007 and April 30, 2010 were considered eligible for study. Visits were included in the study based on the presence of International Classification of Disease, 9th Revision, Clinical Modification (ICD‐9‐CM) codes for respiratory conditions (asthma [493.xx], pneumonia [480.x‐486.x], croup [464.4], bronchiolitis [466.x], acute bronchospasm [519.11], wheezing [786.07]) in the top 3 ED diagnoses. Visits were excluded if the hospital discharge diagnoses included ICD‐9‐CM codes determined a priori by the study team to represent severe illness (e.g., respiratory failure) or complicated medical conditions (e.g., ventilator dependence) unlikely to respond to short stay care (see Supporting Appendices 1 through 3 in the online version of this article). ED visits that resulted in an admission to the intensive care unit (ICU), a procedure, or telemetry were excluded because these children would be unlikely to respond to short stay care.
Variables
Subject demographic characteristics included age, gender, race, and payer. Visit‐related variables included date and time of ED arrival, date and time of ED disposition (to home or admission), and date and time of hospital discharge if admitted. Arrival date and time were used to determine season (winter: DecemberMarch, spring/summer: AprilAugust, fall: SeptemberNovember) and shift (day: 8 am4 pm, evening: 4 pmmidnight, night: midnight8 am).
Outcome
LOS was the primary outcome. ED LOS was calculated by subtracting the date and time of ED arrival from the date and time of ED disposition. Inpatient LOS was calculated by subtracting the date and time of arrival to the inpatient unit from the date and time of discharge from the hospital admission, discharge, transfer (ADT) system. Total LOS was calculated by subtracting the date and time of ED arrival from the date and time of discharge from either the ED or hospital as appropriate.
Analyses
Descriptive statistics were calculated for patient demographics and actual LOS. To describe discharge timing patterns in the ED and inpatient unit, 24 24, 576‐cell matrices were produced with arrival hour along a horizontal axis and discharge hour along a vertical axis. Visits were grouped by hour of ED or inpatient arrival and the number of these visits that were discharged in each hour was plotted in the appropriate cell in the matrix. Cells were then shaded according to the proportion of discharges that occurred at that hour, for each hour of arrival.
Finally, Monte Carlo simulations were designed to illustrate the impact of the timing and frequency of rounds on inpatient LOS. Historical arrival rates were utilized in the simulations. Because medication administration times were not available in the dataset, active treatment times were fixed at 12 hours and 20 hours. The fixed treatment time of 12 hours was based on research by McConnochie et al.,[11] which found that roughly 65% of children hospitalized for asthma received frequent nebulized treatments for 16 hours (two 8‐hour nursing shifts). Four hours were subtracted from this treatment time because our sample included ED treatment times, and local experience demonstrates that children receive at least 4 hours of ED treatment prior to transfer to an inpatient bed. The 20‐hour fixed treatment time was selected to simulate a sicker population of children.
We assumed that physician‐led rounds were the decision point leading to hospital discharge, and that, at minimum, rounds occurred once each morning. The simulation began with 1 physician‐led rounding session occurring at 9 am. Rounding sessions were then added to the model up to a maximum of 24 hourly rounds. Rounding every hour was considered to be analogous to the 24‐hour operation of an ED, in which patients are discharged at all hours. A 4‐hour lag time between the start of first morning rounds and patient discharge was assumed based on historical discharge times. The lag time for additional rounding times was varied, from a minimum of 2 hours for afternoon rounds to a maximum of 10 hours for evening rounds to allow the patient to sleep. Each simulation completed 10,000 iterations and 95% confidence intervals (CIs) were calculated. All analyses were conducted in Excel 2010 (Microsoft Corp., Redmond, WA).
RESULTS
Study Sample Characteristics
Of the 57,639 pediatric ED visits during the 3‐year study period, 5699 were diagnosed with respiratory conditions considered eligible for this study, and 5503 (96.6%) were included in analyses. Hospitalization was the outcome for 1285 (23.4%) ED visits. Demographic characteristics of the study sample are presented in Table 1. Almost half of the visits were by children <3 years old. Visits for respiratory conditions peaked in the fall and winter months. Daily peaks in ED arrivals occurred in the late afternoon.
| Characteristic | n | % |
|---|---|---|
| ||
| Male | 3291 | 59.8 |
| Age | ||
| 02 years | 2716 | 49.4 |
| 35 years | 1305 | 23.7 |
| 612 years | 1128 | 20.5 |
| >12 years | 354 | 6.4 |
| Season at presentation | ||
| Fall (SeptemberNovember) | 1779 | 32.3 |
| Winter (DecemberMarch) | 2429 | 44.1 |
| Spring/Summer (AprilAugust) | 1295 | 23.5 |
| Shift at time of ED arrival | ||
| Day (8 am4 pm) | 1939 | 35.2 |
| Evening (4 pm12 am) | 2296 | 41.7 |
| Overnight (12 am8 am) | 1268 | 23.0 |
| Payer type | ||
| Public | 1996 | 36.3 |
| Private | 3385 | 61.5 |
| Other | 122 | 2.2 |
| Admitted to general care unit | 1285 | 23.4 |
| ICD‐9 diagnoses | ||
| Asthma (493.xx) | 2229 | 40.5 |
| Pneumonia (480.x‐486.x) | 1006 | 18.3 |
| Croup (464.4) | 1048 | 19.0 |
| Bronchiolitis (466.x) | 605 | 11.0 |
| Other (519.11, 786.07, and multiple diagnoses) | 615 | 11.2 |
Length of Stay
For visits meeting study criteria, median ED LOS was 3.0 hours (interquartile range [IQR] 2.1‐4.1) among ED discharges and 5.1 hours (IQR 4.0‐6.6) among admissions. Of the inpatients, 8.4% were admitted and discharged on the same day, 37.3% were admitted for 1 night, 27.0% were admitted for 2 nights, and 27.3% were admitted for 3 or more nights. Median inpatient LOS was 41.2 hours (IQR 23.4‐66.4), with median total LOS of 46.7 hours (IQR 29.3‐71.5).
Arrival and Discharge Patterns
Figure 1 illustrates the relationship between the ED arrival and disposition times for all included visits. Figure 2 illustrates the relationship between the inpatient arrival and discharge times among the subset of visits that received inpatient care. Children with respiratory illness arrived to the ED at all hours of the day and night and generally were discharged 2 to 5 hours after arrival regardless of the time of day. In contrast, children admitted to the inpatient setting were most commonly discharged between 11 am and 6 pm, regardless of the time of admission.
Simulation
The influence of timing and frequency of rounds on inpatient LOS is shown in Table 2. Once‐daily rounds resulted in a predicted LOS of 27.24 hours (95% CI: 27.0727.41) for the 12‐hour active treatment time simulation and 36.93 hours (95% CI: 36.70‐37.16) for the 20‐hour active treatment time. There were incremental reductions in predicted LOS with each additional rounding session. When the number of rounding sessions reached 24, predicted LOS was 14.5 hours (95% CI: 14.49‐14.51) for the 12‐hour active treatment time simulation and 22.5 hours (95% CI: 22.49‐22.51) for the 20‐hour active treatment time.
| No. of Rounds | Time of Rounds | Rounds to Discharge (Hours) | 12‐Hour Active Treatment Time | 20‐Hour Active Treatment Time | ||
|---|---|---|---|---|---|---|
| Change in Length of Stay (Hours) | % of Reference Length of Stay | Change in Length of Stay (Hours) | % of Reference Length of Stay | |||
| 1 | 0900 | 4 | 27.24 (reference) | 36.93 (reference) | ||
| 2 | 0900 | 4 | 4.84 | 17.8% | 5.34 | 14.4% |
| 1500 | 4 | |||||
| 2 | 0900 | 4 | 5.31 | 19.5% | 5.90 | 16.0% |
| 1500 | 2 | |||||
| 3 | 0900 | 4 | 6.24 | 22.9% | 7.94 | 21.5% |
| 1500 | 2 | |||||
| 2100 | 10 | |||||
| 3 | 0900 | 4 | 7.32 | 26.9% | 10.49 | 28.4% |
| 1500 | 2 | |||||
| 2100 | 2 | |||||
| 24 | Hourly | 2 | 12.74 | 46.8% | 14.2 | 38.7% |
DISCUSSION
Our study illustrates dramatic differences in the discharge patterns from the ED and inpatient settings for children with respiratory illnesses. Although discharges from the ED occur at all hours, inpatient discharges were concentrated during midday. The time of discharge from the ED was highly related to the time of arrival, implying that any lag between discharge decision making and discharge was independent of time of arrival. The absence of a clear relationship between inpatient discharges and arrivals suggests that factors other than the clinical status of patients affect hospital LOS.
Because physician‐led rounds have traditionally been the point of decision‐making regarding inpatient discharge readiness, we hypothesized that increasing the frequency of rounds could reduce LOS. Within the parameters set in our simulation models, our results support this hypothesis. As the number of rounding sessions increased, hospital LOS decreased and the amount of time a child waited to be identified as ready for discharge approached zero. Simulating hourly rounding, analogous to the ED setting, resulted in the greatest reductions in inpatient LOS. Our findings have important implications for hospital operations and discharge policies given that children with respiratory conditions commonly experience short stay hospitalizations,[12, 13] but few children's hospitals have put in place models of care that differentiate the needs of short stay patients from those of inpatients requiring longer LOS.[14] To operationalize inpatient discharges as soon as a child is well enough to go home, parental expectations for discharge timing would need to be set, and discharge planning would need to begin at the start of each hospital stay.
Although there have been numerous studies on ED crowding, its causes, and potential solutions,[15, 16, 17, 18, 19, 20] this study is the first to our knowledge to demonstrate differences in discharge timing between the ED and inpatient general care unit. High levels of hospital occupancy decrease patient flow in both ED and inpatient units.[21, 22, 23, 24, 25] Shifting inpatient discharge for adult patients to earlier in the day can reduce or eliminate inpatient boarding in the ED.[26, 27] Evaluation of these relationships in pediatric populations are needed because of the unique care requirements of acutely ill children who often respond rapidly to hospital‐based treatment. In a recent opinion piece, Iantorno and Fieldston discourage hospitals from setting specified time targets for discharges and propose that high‐quality care includes afternoon and evening discharges.[28] The optimal timing for hospital discharges has not yet been defined, but our results indicate there is potential to reduce excess time spent in hospitals through the addition of rounds that would identify discharges throughout the day.
A simulation study based on hospital administrative data cannot determine a causal relationship between physician‐led rounds and hospital LOS. Still, our findings can generate discussions about hospital discharge policies and patient throughput initiatives. Hospitals can simulate other approaches that may apply to their institutional operations, such as changing rounding processes or adding an observation unit where discharges occur after hours. Hospitals could then pilot approaches locally. With the potential for unintended consequences of patient throughput initiatives that focus only on LOS as an outcome, pilot programs should be designed to track not only ED and inpatient LOS but also family satisfaction, access to follow‐up visits in primary care, need for reassessment in urgent care centers and EDs, and hospital readmissions.
The feasibility and acceptability of afternoon and evening rounds to improve patient throughput must be considered. Clinicians caring for hospitalized patients may have competing demands on their time, such as other clinical obligations, committee work, or academic pursuits that would make more frequent rounding unattractive. Physicians‐in‐training in teaching hospitals have educational requirements and duty‐hour restrictions that may limit their ability to round more frequently.[29, 30] There is also a need to define reasonable minimum standards for discharge processes (e.g., provision of patient education, discharge medications, and paperwork) for short‐stay patients and those requiring prolonged and/or complex hospitalizations. Streamlined discharge processes for short‐stay patients with simple illnesses may result in more efficient discharges. More efficient hospital discharges at flexible hours that are acceptable to families may require a culture shift among hospital staff. Although the palatability of off‐hours inpatient discharges has not been explored with patients and families, some may prefer this approach.
Limitations
There are several limitations to our study. First, the time of discharge is recorded differently for the ED and inpatient settings in our administrative dataset as a function of the electronic medical record systems in these 2 environments. Discharge from the ED reflects that time the patient left the ED but discharge from the inpatient setting reflects the time the patient exits the hospital ADT system. This likely biases our results toward longer LOS in the inpatient setting. We expect this would shift the band of inpatient discharges later in the day but do not expect this to alter the observed difference in discharge patterns between the ED and inpatient settings.
Second, administrative data do not provide information about the physicians making the discharge decisions or valid reasons a child would remain in the hospital after they improved clinically including time for teaching or care coordination. Our results therefore overestimate the amount of excess time associated with inpatient care. Future research is needed to determine the actual duration of time between when a child is clinically well enough for discharge home and when a child is actually discharged. There is also a need to understand discharge decision making and to identify the nonclinical factors that contribute to discharge delays.
Third, our data sample was taken from visits made by children to a pediatric ED nested within a general ED where children are admitted to an adjoining tertiary care academic children's hospital. Our results may not be generalizable to other settings in which children receive hospital‐based care, including freestanding children's hospitals and general hospitals that admit children.
Fourth, we focused our analyses on children with respiratory conditions, excluding visits with diagnoses suggestive of complex comorbid conditions and severe illnesses. Although we anticipate that children with these respiratory conditions treated in other hospitals will be similar to our population, our results may not hold for other conditions or across the full spectrum of severity for acute respiratory illness treated within EDs and inpatient units.
Finally, our simulation model makes assumptions, such as a fixed treatment times and discharge process length, which do not capture the clinical nuances of an individual child's response to hospital‐based treatment. Though much of the clinical complexity of hospital operations is not taken into account in the model, our intended purpose, to explore the influence of different rounding times on LOS, remains valid.
CONCLUSIONS
For children obtaining emergency care for respiratory illnesses, discharges from the ED occur around the clock, whereas discharges from the inpatient general care unit are concentrated during afternoon hours. Simulation models illustrate the potential to reduce hospital LOS by adding rounding sessions. Extending the hours of discharge for hospitalized children with respiratory illnesses may increase efficiency of care but could result in unintended consequences such as fewer opportunities for patient education.
Acknowledgments
The authors acknowledge the Center for Healthcare Engineering and Patient Safety, the Bonder Foundation, the Doctors Company Foundation, the Center for Research on Learning and Teaching, and the University of Michigan Summer Undergraduate Research in Engineering Program for their support of the students who contributed to this project.Disclosures: Funding for data analysis for this project was supported by a grant from the Center for Healthcare Research and Transformation (Dr. Macy). Support for the students working on this project (East, Burns, O'Gara, and Card) was provided through grants from the Center for Research and Teaching at the University of Michigan, the Center for Healthcare Engineering and Patient Safety at the University of Michigan, the Seth Bonder Foundation, and the Doctors Company Foundation. The authors have no financial or other conflicts of interest to disclose.
- , , , et al. Children's hospitals do not acutely respond to high occupancy. Pediatrics. 2010;125(5):974–981.
- , , , et al. Addressing inpatient crowding by smoothing occupancy at children's hospitals. J Hosp Med. 2011;6(8):462–468.
- , . The growing role of emergency departments in hospital admissions. N Engl J Med. 2012;367(5):391–393.
- , , . Managing asthma exacerbations in the emergency department: summary of the National Asthma Education and Prevention Program Expert Panel Report 3 guidelines for the management of asthma exacerbations. J Allergy Clin Immunol. 2009;124(2 suppl):S5–S14.
- American Academy of Pediatrics Subcommittee on Diagnosis and Management of Bronchiolitis. Diagnosis and management of bronchiolitis. Pediatrics. 2006;118(4):1774–1793.
- , , , et al. Executive summary: the management of community‐acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America. Clin Infect Dis. 2011;53(7):617–630.
- , , , , . Glucocorticoids for croup. Cochrane Database Syst Rev. 2011(1):CD001955.
- , , , et al. Epidemiology of a pediatric emergency medicine research network: the PECARN Core Data Project. Pediatr Emerg Care. 2006;22(10):689–699.
- , , , et al. Annual report on health care for children and youth in the United States: focus on trends in hospital use and quality. Acad Pediatr. 2011;11(4):263–279.
- , , , , , . Trends in high‐turnover stays among children hospitalized in the United States, 1993–2003. Pediatrics. 2009;123(3):996–1002.
- , , , , , . How commonly are children hospitalized for asthma eligible for care in alternative settings? Arch Pediatr Adolesc Med. 1999;153(1):49–55.
- , , , , . High turnover stays for pediatric asthma in the United States: analysis of the 2006 Kids' Inpatient Database. Med Care. 2010;48(9):827–833.
- , , , et al. Pediatric observation status: are we overlooking a growing population in children's hospitals? J Hosp Med. 2012;7(7):530–536.
- , , , et al. Differences in designations of observation care in US freestanding children's hospitals: are they virtual or real? J Hosp Med. 2012;7(4):287–293.
- , . Overcrowding in the nation's emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000;35(1):63–68.
- , , . Frequent overcrowding in U.S. emergency departments. Acad Emerg Med. 2001;8(2):151–155.
- , . Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emerg Med J. 2003;20(5):402–405.
- , , , et al. Solutions to emergency department 'boarding'[and crowding are underused and may need to be legislated. Health Aff (Millwood). 2012;31(8):1757–1766.
- . Managing emergency department overcrowding. Emerg Med Clin North Am. 2009;27(4):593–603, viii.
- Overcrowding crisis in our nation's emergency departments: is our safety net unraveling? Pediatrics. 2004;114(3):878–888.
- , , , et al. Crowding delays treatment and lengthens emergency department length of stay, even among high‐acuity patients. Ann Emerg Med. 2009;54(4):492–503.e494.
- . Do elective surgical and medical admissions impact emergency department length of stay measurements? Clin Invest Med. 2007;30(5):E177–E182.
- , , , . Unravelling relationships: hospital occupancy levels, discharge timing and emergency department access block. Emerg Med Australas. 2012;24(5):510–517.
- , , , . The effect of hospital bed occupancy on throughput in the pediatric emergency department. Ann Emerg Med. 2009;53(6):767–776.e763.
- , , , , . Impact of admission‐day crowding on the length of stay of pediatric hospitalizations. Pediatrics. 2008;121(4):e718–e730.
- , , , , , . The relationship between inpatient discharge timing and emergency department boarding. J Emerg Med. 2012;42(2):186–196.
- , , . Effect of coupling between emergency department and inpatient unit on the overcrowding in emergency department. Paper presented at: Winter Simulation Conference; 2007; Washington, DC.
- , . Hospitals are not hotels: high‐quality discharges occur around the clock. JAMA Pediatr. 2013;167(7):596–597.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , . Failure to thrive: pediatric residents weigh in on feasibility trial of the proposed 2008 institute of medicine work hour restrictions. J Grad Med Educ. 2009;1(2):181–184.
Hospitals are facing growing pressures to operate more efficiently, spurring interest in improving patient flow from the emergency department (ED) to inpatient unit to home. Children's hospitals are often at high occupancy,[1] and EDs are increasingly an entry point for hospital admission.[2, 3]
Among children who require brief episodes of hospital‐based care, hospital processes, especially those associated with discharge, can greatly impact length of stay (LOS). Patients ready for discharge from inpatient units are typically identified through formal physician‐led rounds, in contrast to EDs where discharges occur on a more continual basis. Quantitative descriptions of rounding frequency and LOS are lacking.
The focus of this study was the population of children who had visits for select respiratory illnesses (e.g., asthma, bronchiolitis, pneumonia, and croup) for which there is general consensus regarding treatment, admission, and discharge criteria.[4, 5, 6, 7] The selected illnesses represent common reasons for ED visits[8] and hospitalizations.[9] Hospital stays for these conditions tend to be brief, often 1 to 2 days in duration,[10] and repeated assessments are necessary to determine suitability for discharge.
The primary objectives of this study were to compare discharge patterns in the ED and inpatient settings and to quantify the relationship between discharge timing and LOS in these different clinical settings. A simulation was then used to predict the effect of the timing and frequency of physician‐led rounds on hospital LOS. We hypothesized that increased frequency of simulated physician‐led rounds would lead to meaningful reductions in predicted hospital LOS for children admitted from the ED.
METHODS
Retrospective analyses were conducted using hospital administrative data from pediatric ED visits and resultant inpatient stays. The University of Michigan Institutional Review Board approved the study.
Setting
C. S. Mott Children's Hospital at the University of Michigan is a suburban academic, tertiary care hospital located in Ann Arbor, Michigan. The pediatric emergency department had approximately 20,000 visits per year during the study period, and children were treated in an 11‐bed area. There were 106 pediatric inpatient general care beds during the study years.
Visit Selection
ED visits made by children <18 years old during the 3‐year period between May 1, 2007 and April 30, 2010 were considered eligible for study. Visits were included in the study based on the presence of International Classification of Disease, 9th Revision, Clinical Modification (ICD‐9‐CM) codes for respiratory conditions (asthma [493.xx], pneumonia [480.x‐486.x], croup [464.4], bronchiolitis [466.x], acute bronchospasm [519.11], wheezing [786.07]) in the top 3 ED diagnoses. Visits were excluded if the hospital discharge diagnoses included ICD‐9‐CM codes determined a priori by the study team to represent severe illness (e.g., respiratory failure) or complicated medical conditions (e.g., ventilator dependence) unlikely to respond to short stay care (see Supporting Appendices 1 through 3 in the online version of this article). ED visits that resulted in an admission to the intensive care unit (ICU), a procedure, or telemetry were excluded because these children would be unlikely to respond to short stay care.
Variables
Subject demographic characteristics included age, gender, race, and payer. Visit‐related variables included date and time of ED arrival, date and time of ED disposition (to home or admission), and date and time of hospital discharge if admitted. Arrival date and time were used to determine season (winter: DecemberMarch, spring/summer: AprilAugust, fall: SeptemberNovember) and shift (day: 8 am4 pm, evening: 4 pmmidnight, night: midnight8 am).
Outcome
LOS was the primary outcome. ED LOS was calculated by subtracting the date and time of ED arrival from the date and time of ED disposition. Inpatient LOS was calculated by subtracting the date and time of arrival to the inpatient unit from the date and time of discharge from the hospital admission, discharge, transfer (ADT) system. Total LOS was calculated by subtracting the date and time of ED arrival from the date and time of discharge from either the ED or hospital as appropriate.
Analyses
Descriptive statistics were calculated for patient demographics and actual LOS. To describe discharge timing patterns in the ED and inpatient unit, 24 24, 576‐cell matrices were produced with arrival hour along a horizontal axis and discharge hour along a vertical axis. Visits were grouped by hour of ED or inpatient arrival and the number of these visits that were discharged in each hour was plotted in the appropriate cell in the matrix. Cells were then shaded according to the proportion of discharges that occurred at that hour, for each hour of arrival.
Finally, Monte Carlo simulations were designed to illustrate the impact of the timing and frequency of rounds on inpatient LOS. Historical arrival rates were utilized in the simulations. Because medication administration times were not available in the dataset, active treatment times were fixed at 12 hours and 20 hours. The fixed treatment time of 12 hours was based on research by McConnochie et al.,[11] which found that roughly 65% of children hospitalized for asthma received frequent nebulized treatments for 16 hours (two 8‐hour nursing shifts). Four hours were subtracted from this treatment time because our sample included ED treatment times, and local experience demonstrates that children receive at least 4 hours of ED treatment prior to transfer to an inpatient bed. The 20‐hour fixed treatment time was selected to simulate a sicker population of children.
We assumed that physician‐led rounds were the decision point leading to hospital discharge, and that, at minimum, rounds occurred once each morning. The simulation began with 1 physician‐led rounding session occurring at 9 am. Rounding sessions were then added to the model up to a maximum of 24 hourly rounds. Rounding every hour was considered to be analogous to the 24‐hour operation of an ED, in which patients are discharged at all hours. A 4‐hour lag time between the start of first morning rounds and patient discharge was assumed based on historical discharge times. The lag time for additional rounding times was varied, from a minimum of 2 hours for afternoon rounds to a maximum of 10 hours for evening rounds to allow the patient to sleep. Each simulation completed 10,000 iterations and 95% confidence intervals (CIs) were calculated. All analyses were conducted in Excel 2010 (Microsoft Corp., Redmond, WA).
RESULTS
Study Sample Characteristics
Of the 57,639 pediatric ED visits during the 3‐year study period, 5699 were diagnosed with respiratory conditions considered eligible for this study, and 5503 (96.6%) were included in analyses. Hospitalization was the outcome for 1285 (23.4%) ED visits. Demographic characteristics of the study sample are presented in Table 1. Almost half of the visits were by children <3 years old. Visits for respiratory conditions peaked in the fall and winter months. Daily peaks in ED arrivals occurred in the late afternoon.
| Characteristic | n | % |
|---|---|---|
| ||
| Male | 3291 | 59.8 |
| Age | ||
| 02 years | 2716 | 49.4 |
| 35 years | 1305 | 23.7 |
| 612 years | 1128 | 20.5 |
| >12 years | 354 | 6.4 |
| Season at presentation | ||
| Fall (SeptemberNovember) | 1779 | 32.3 |
| Winter (DecemberMarch) | 2429 | 44.1 |
| Spring/Summer (AprilAugust) | 1295 | 23.5 |
| Shift at time of ED arrival | ||
| Day (8 am4 pm) | 1939 | 35.2 |
| Evening (4 pm12 am) | 2296 | 41.7 |
| Overnight (12 am8 am) | 1268 | 23.0 |
| Payer type | ||
| Public | 1996 | 36.3 |
| Private | 3385 | 61.5 |
| Other | 122 | 2.2 |
| Admitted to general care unit | 1285 | 23.4 |
| ICD‐9 diagnoses | ||
| Asthma (493.xx) | 2229 | 40.5 |
| Pneumonia (480.x‐486.x) | 1006 | 18.3 |
| Croup (464.4) | 1048 | 19.0 |
| Bronchiolitis (466.x) | 605 | 11.0 |
| Other (519.11, 786.07, and multiple diagnoses) | 615 | 11.2 |
Length of Stay
For visits meeting study criteria, median ED LOS was 3.0 hours (interquartile range [IQR] 2.1‐4.1) among ED discharges and 5.1 hours (IQR 4.0‐6.6) among admissions. Of the inpatients, 8.4% were admitted and discharged on the same day, 37.3% were admitted for 1 night, 27.0% were admitted for 2 nights, and 27.3% were admitted for 3 or more nights. Median inpatient LOS was 41.2 hours (IQR 23.4‐66.4), with median total LOS of 46.7 hours (IQR 29.3‐71.5).
Arrival and Discharge Patterns
Figure 1 illustrates the relationship between the ED arrival and disposition times for all included visits. Figure 2 illustrates the relationship between the inpatient arrival and discharge times among the subset of visits that received inpatient care. Children with respiratory illness arrived to the ED at all hours of the day and night and generally were discharged 2 to 5 hours after arrival regardless of the time of day. In contrast, children admitted to the inpatient setting were most commonly discharged between 11 am and 6 pm, regardless of the time of admission.
Simulation
The influence of timing and frequency of rounds on inpatient LOS is shown in Table 2. Once‐daily rounds resulted in a predicted LOS of 27.24 hours (95% CI: 27.0727.41) for the 12‐hour active treatment time simulation and 36.93 hours (95% CI: 36.70‐37.16) for the 20‐hour active treatment time. There were incremental reductions in predicted LOS with each additional rounding session. When the number of rounding sessions reached 24, predicted LOS was 14.5 hours (95% CI: 14.49‐14.51) for the 12‐hour active treatment time simulation and 22.5 hours (95% CI: 22.49‐22.51) for the 20‐hour active treatment time.
| No. of Rounds | Time of Rounds | Rounds to Discharge (Hours) | 12‐Hour Active Treatment Time | 20‐Hour Active Treatment Time | ||
|---|---|---|---|---|---|---|
| Change in Length of Stay (Hours) | % of Reference Length of Stay | Change in Length of Stay (Hours) | % of Reference Length of Stay | |||
| 1 | 0900 | 4 | 27.24 (reference) | 36.93 (reference) | ||
| 2 | 0900 | 4 | 4.84 | 17.8% | 5.34 | 14.4% |
| 1500 | 4 | |||||
| 2 | 0900 | 4 | 5.31 | 19.5% | 5.90 | 16.0% |
| 1500 | 2 | |||||
| 3 | 0900 | 4 | 6.24 | 22.9% | 7.94 | 21.5% |
| 1500 | 2 | |||||
| 2100 | 10 | |||||
| 3 | 0900 | 4 | 7.32 | 26.9% | 10.49 | 28.4% |
| 1500 | 2 | |||||
| 2100 | 2 | |||||
| 24 | Hourly | 2 | 12.74 | 46.8% | 14.2 | 38.7% |
DISCUSSION
Our study illustrates dramatic differences in the discharge patterns from the ED and inpatient settings for children with respiratory illnesses. Although discharges from the ED occur at all hours, inpatient discharges were concentrated during midday. The time of discharge from the ED was highly related to the time of arrival, implying that any lag between discharge decision making and discharge was independent of time of arrival. The absence of a clear relationship between inpatient discharges and arrivals suggests that factors other than the clinical status of patients affect hospital LOS.
Because physician‐led rounds have traditionally been the point of decision‐making regarding inpatient discharge readiness, we hypothesized that increasing the frequency of rounds could reduce LOS. Within the parameters set in our simulation models, our results support this hypothesis. As the number of rounding sessions increased, hospital LOS decreased and the amount of time a child waited to be identified as ready for discharge approached zero. Simulating hourly rounding, analogous to the ED setting, resulted in the greatest reductions in inpatient LOS. Our findings have important implications for hospital operations and discharge policies given that children with respiratory conditions commonly experience short stay hospitalizations,[12, 13] but few children's hospitals have put in place models of care that differentiate the needs of short stay patients from those of inpatients requiring longer LOS.[14] To operationalize inpatient discharges as soon as a child is well enough to go home, parental expectations for discharge timing would need to be set, and discharge planning would need to begin at the start of each hospital stay.
Although there have been numerous studies on ED crowding, its causes, and potential solutions,[15, 16, 17, 18, 19, 20] this study is the first to our knowledge to demonstrate differences in discharge timing between the ED and inpatient general care unit. High levels of hospital occupancy decrease patient flow in both ED and inpatient units.[21, 22, 23, 24, 25] Shifting inpatient discharge for adult patients to earlier in the day can reduce or eliminate inpatient boarding in the ED.[26, 27] Evaluation of these relationships in pediatric populations are needed because of the unique care requirements of acutely ill children who often respond rapidly to hospital‐based treatment. In a recent opinion piece, Iantorno and Fieldston discourage hospitals from setting specified time targets for discharges and propose that high‐quality care includes afternoon and evening discharges.[28] The optimal timing for hospital discharges has not yet been defined, but our results indicate there is potential to reduce excess time spent in hospitals through the addition of rounds that would identify discharges throughout the day.
A simulation study based on hospital administrative data cannot determine a causal relationship between physician‐led rounds and hospital LOS. Still, our findings can generate discussions about hospital discharge policies and patient throughput initiatives. Hospitals can simulate other approaches that may apply to their institutional operations, such as changing rounding processes or adding an observation unit where discharges occur after hours. Hospitals could then pilot approaches locally. With the potential for unintended consequences of patient throughput initiatives that focus only on LOS as an outcome, pilot programs should be designed to track not only ED and inpatient LOS but also family satisfaction, access to follow‐up visits in primary care, need for reassessment in urgent care centers and EDs, and hospital readmissions.
The feasibility and acceptability of afternoon and evening rounds to improve patient throughput must be considered. Clinicians caring for hospitalized patients may have competing demands on their time, such as other clinical obligations, committee work, or academic pursuits that would make more frequent rounding unattractive. Physicians‐in‐training in teaching hospitals have educational requirements and duty‐hour restrictions that may limit their ability to round more frequently.[29, 30] There is also a need to define reasonable minimum standards for discharge processes (e.g., provision of patient education, discharge medications, and paperwork) for short‐stay patients and those requiring prolonged and/or complex hospitalizations. Streamlined discharge processes for short‐stay patients with simple illnesses may result in more efficient discharges. More efficient hospital discharges at flexible hours that are acceptable to families may require a culture shift among hospital staff. Although the palatability of off‐hours inpatient discharges has not been explored with patients and families, some may prefer this approach.
Limitations
There are several limitations to our study. First, the time of discharge is recorded differently for the ED and inpatient settings in our administrative dataset as a function of the electronic medical record systems in these 2 environments. Discharge from the ED reflects that time the patient left the ED but discharge from the inpatient setting reflects the time the patient exits the hospital ADT system. This likely biases our results toward longer LOS in the inpatient setting. We expect this would shift the band of inpatient discharges later in the day but do not expect this to alter the observed difference in discharge patterns between the ED and inpatient settings.
Second, administrative data do not provide information about the physicians making the discharge decisions or valid reasons a child would remain in the hospital after they improved clinically including time for teaching or care coordination. Our results therefore overestimate the amount of excess time associated with inpatient care. Future research is needed to determine the actual duration of time between when a child is clinically well enough for discharge home and when a child is actually discharged. There is also a need to understand discharge decision making and to identify the nonclinical factors that contribute to discharge delays.
Third, our data sample was taken from visits made by children to a pediatric ED nested within a general ED where children are admitted to an adjoining tertiary care academic children's hospital. Our results may not be generalizable to other settings in which children receive hospital‐based care, including freestanding children's hospitals and general hospitals that admit children.
Fourth, we focused our analyses on children with respiratory conditions, excluding visits with diagnoses suggestive of complex comorbid conditions and severe illnesses. Although we anticipate that children with these respiratory conditions treated in other hospitals will be similar to our population, our results may not hold for other conditions or across the full spectrum of severity for acute respiratory illness treated within EDs and inpatient units.
Finally, our simulation model makes assumptions, such as a fixed treatment times and discharge process length, which do not capture the clinical nuances of an individual child's response to hospital‐based treatment. Though much of the clinical complexity of hospital operations is not taken into account in the model, our intended purpose, to explore the influence of different rounding times on LOS, remains valid.
CONCLUSIONS
For children obtaining emergency care for respiratory illnesses, discharges from the ED occur around the clock, whereas discharges from the inpatient general care unit are concentrated during afternoon hours. Simulation models illustrate the potential to reduce hospital LOS by adding rounding sessions. Extending the hours of discharge for hospitalized children with respiratory illnesses may increase efficiency of care but could result in unintended consequences such as fewer opportunities for patient education.
Acknowledgments
The authors acknowledge the Center for Healthcare Engineering and Patient Safety, the Bonder Foundation, the Doctors Company Foundation, the Center for Research on Learning and Teaching, and the University of Michigan Summer Undergraduate Research in Engineering Program for their support of the students who contributed to this project.Disclosures: Funding for data analysis for this project was supported by a grant from the Center for Healthcare Research and Transformation (Dr. Macy). Support for the students working on this project (East, Burns, O'Gara, and Card) was provided through grants from the Center for Research and Teaching at the University of Michigan, the Center for Healthcare Engineering and Patient Safety at the University of Michigan, the Seth Bonder Foundation, and the Doctors Company Foundation. The authors have no financial or other conflicts of interest to disclose.
Hospitals are facing growing pressures to operate more efficiently, spurring interest in improving patient flow from the emergency department (ED) to inpatient unit to home. Children's hospitals are often at high occupancy,[1] and EDs are increasingly an entry point for hospital admission.[2, 3]
Among children who require brief episodes of hospital‐based care, hospital processes, especially those associated with discharge, can greatly impact length of stay (LOS). Patients ready for discharge from inpatient units are typically identified through formal physician‐led rounds, in contrast to EDs where discharges occur on a more continual basis. Quantitative descriptions of rounding frequency and LOS are lacking.
The focus of this study was the population of children who had visits for select respiratory illnesses (e.g., asthma, bronchiolitis, pneumonia, and croup) for which there is general consensus regarding treatment, admission, and discharge criteria.[4, 5, 6, 7] The selected illnesses represent common reasons for ED visits[8] and hospitalizations.[9] Hospital stays for these conditions tend to be brief, often 1 to 2 days in duration,[10] and repeated assessments are necessary to determine suitability for discharge.
The primary objectives of this study were to compare discharge patterns in the ED and inpatient settings and to quantify the relationship between discharge timing and LOS in these different clinical settings. A simulation was then used to predict the effect of the timing and frequency of physician‐led rounds on hospital LOS. We hypothesized that increased frequency of simulated physician‐led rounds would lead to meaningful reductions in predicted hospital LOS for children admitted from the ED.
METHODS
Retrospective analyses were conducted using hospital administrative data from pediatric ED visits and resultant inpatient stays. The University of Michigan Institutional Review Board approved the study.
Setting
C. S. Mott Children's Hospital at the University of Michigan is a suburban academic, tertiary care hospital located in Ann Arbor, Michigan. The pediatric emergency department had approximately 20,000 visits per year during the study period, and children were treated in an 11‐bed area. There were 106 pediatric inpatient general care beds during the study years.
Visit Selection
ED visits made by children <18 years old during the 3‐year period between May 1, 2007 and April 30, 2010 were considered eligible for study. Visits were included in the study based on the presence of International Classification of Disease, 9th Revision, Clinical Modification (ICD‐9‐CM) codes for respiratory conditions (asthma [493.xx], pneumonia [480.x‐486.x], croup [464.4], bronchiolitis [466.x], acute bronchospasm [519.11], wheezing [786.07]) in the top 3 ED diagnoses. Visits were excluded if the hospital discharge diagnoses included ICD‐9‐CM codes determined a priori by the study team to represent severe illness (e.g., respiratory failure) or complicated medical conditions (e.g., ventilator dependence) unlikely to respond to short stay care (see Supporting Appendices 1 through 3 in the online version of this article). ED visits that resulted in an admission to the intensive care unit (ICU), a procedure, or telemetry were excluded because these children would be unlikely to respond to short stay care.
Variables
Subject demographic characteristics included age, gender, race, and payer. Visit‐related variables included date and time of ED arrival, date and time of ED disposition (to home or admission), and date and time of hospital discharge if admitted. Arrival date and time were used to determine season (winter: DecemberMarch, spring/summer: AprilAugust, fall: SeptemberNovember) and shift (day: 8 am4 pm, evening: 4 pmmidnight, night: midnight8 am).
Outcome
LOS was the primary outcome. ED LOS was calculated by subtracting the date and time of ED arrival from the date and time of ED disposition. Inpatient LOS was calculated by subtracting the date and time of arrival to the inpatient unit from the date and time of discharge from the hospital admission, discharge, transfer (ADT) system. Total LOS was calculated by subtracting the date and time of ED arrival from the date and time of discharge from either the ED or hospital as appropriate.
Analyses
Descriptive statistics were calculated for patient demographics and actual LOS. To describe discharge timing patterns in the ED and inpatient unit, 24 24, 576‐cell matrices were produced with arrival hour along a horizontal axis and discharge hour along a vertical axis. Visits were grouped by hour of ED or inpatient arrival and the number of these visits that were discharged in each hour was plotted in the appropriate cell in the matrix. Cells were then shaded according to the proportion of discharges that occurred at that hour, for each hour of arrival.
Finally, Monte Carlo simulations were designed to illustrate the impact of the timing and frequency of rounds on inpatient LOS. Historical arrival rates were utilized in the simulations. Because medication administration times were not available in the dataset, active treatment times were fixed at 12 hours and 20 hours. The fixed treatment time of 12 hours was based on research by McConnochie et al.,[11] which found that roughly 65% of children hospitalized for asthma received frequent nebulized treatments for 16 hours (two 8‐hour nursing shifts). Four hours were subtracted from this treatment time because our sample included ED treatment times, and local experience demonstrates that children receive at least 4 hours of ED treatment prior to transfer to an inpatient bed. The 20‐hour fixed treatment time was selected to simulate a sicker population of children.
We assumed that physician‐led rounds were the decision point leading to hospital discharge, and that, at minimum, rounds occurred once each morning. The simulation began with 1 physician‐led rounding session occurring at 9 am. Rounding sessions were then added to the model up to a maximum of 24 hourly rounds. Rounding every hour was considered to be analogous to the 24‐hour operation of an ED, in which patients are discharged at all hours. A 4‐hour lag time between the start of first morning rounds and patient discharge was assumed based on historical discharge times. The lag time for additional rounding times was varied, from a minimum of 2 hours for afternoon rounds to a maximum of 10 hours for evening rounds to allow the patient to sleep. Each simulation completed 10,000 iterations and 95% confidence intervals (CIs) were calculated. All analyses were conducted in Excel 2010 (Microsoft Corp., Redmond, WA).
RESULTS
Study Sample Characteristics
Of the 57,639 pediatric ED visits during the 3‐year study period, 5699 were diagnosed with respiratory conditions considered eligible for this study, and 5503 (96.6%) were included in analyses. Hospitalization was the outcome for 1285 (23.4%) ED visits. Demographic characteristics of the study sample are presented in Table 1. Almost half of the visits were by children <3 years old. Visits for respiratory conditions peaked in the fall and winter months. Daily peaks in ED arrivals occurred in the late afternoon.
| Characteristic | n | % |
|---|---|---|
| ||
| Male | 3291 | 59.8 |
| Age | ||
| 02 years | 2716 | 49.4 |
| 35 years | 1305 | 23.7 |
| 612 years | 1128 | 20.5 |
| >12 years | 354 | 6.4 |
| Season at presentation | ||
| Fall (SeptemberNovember) | 1779 | 32.3 |
| Winter (DecemberMarch) | 2429 | 44.1 |
| Spring/Summer (AprilAugust) | 1295 | 23.5 |
| Shift at time of ED arrival | ||
| Day (8 am4 pm) | 1939 | 35.2 |
| Evening (4 pm12 am) | 2296 | 41.7 |
| Overnight (12 am8 am) | 1268 | 23.0 |
| Payer type | ||
| Public | 1996 | 36.3 |
| Private | 3385 | 61.5 |
| Other | 122 | 2.2 |
| Admitted to general care unit | 1285 | 23.4 |
| ICD‐9 diagnoses | ||
| Asthma (493.xx) | 2229 | 40.5 |
| Pneumonia (480.x‐486.x) | 1006 | 18.3 |
| Croup (464.4) | 1048 | 19.0 |
| Bronchiolitis (466.x) | 605 | 11.0 |
| Other (519.11, 786.07, and multiple diagnoses) | 615 | 11.2 |
Length of Stay
For visits meeting study criteria, median ED LOS was 3.0 hours (interquartile range [IQR] 2.1‐4.1) among ED discharges and 5.1 hours (IQR 4.0‐6.6) among admissions. Of the inpatients, 8.4% were admitted and discharged on the same day, 37.3% were admitted for 1 night, 27.0% were admitted for 2 nights, and 27.3% were admitted for 3 or more nights. Median inpatient LOS was 41.2 hours (IQR 23.4‐66.4), with median total LOS of 46.7 hours (IQR 29.3‐71.5).
Arrival and Discharge Patterns
Figure 1 illustrates the relationship between the ED arrival and disposition times for all included visits. Figure 2 illustrates the relationship between the inpatient arrival and discharge times among the subset of visits that received inpatient care. Children with respiratory illness arrived to the ED at all hours of the day and night and generally were discharged 2 to 5 hours after arrival regardless of the time of day. In contrast, children admitted to the inpatient setting were most commonly discharged between 11 am and 6 pm, regardless of the time of admission.
Simulation
The influence of timing and frequency of rounds on inpatient LOS is shown in Table 2. Once‐daily rounds resulted in a predicted LOS of 27.24 hours (95% CI: 27.0727.41) for the 12‐hour active treatment time simulation and 36.93 hours (95% CI: 36.70‐37.16) for the 20‐hour active treatment time. There were incremental reductions in predicted LOS with each additional rounding session. When the number of rounding sessions reached 24, predicted LOS was 14.5 hours (95% CI: 14.49‐14.51) for the 12‐hour active treatment time simulation and 22.5 hours (95% CI: 22.49‐22.51) for the 20‐hour active treatment time.
| No. of Rounds | Time of Rounds | Rounds to Discharge (Hours) | 12‐Hour Active Treatment Time | 20‐Hour Active Treatment Time | ||
|---|---|---|---|---|---|---|
| Change in Length of Stay (Hours) | % of Reference Length of Stay | Change in Length of Stay (Hours) | % of Reference Length of Stay | |||
| 1 | 0900 | 4 | 27.24 (reference) | 36.93 (reference) | ||
| 2 | 0900 | 4 | 4.84 | 17.8% | 5.34 | 14.4% |
| 1500 | 4 | |||||
| 2 | 0900 | 4 | 5.31 | 19.5% | 5.90 | 16.0% |
| 1500 | 2 | |||||
| 3 | 0900 | 4 | 6.24 | 22.9% | 7.94 | 21.5% |
| 1500 | 2 | |||||
| 2100 | 10 | |||||
| 3 | 0900 | 4 | 7.32 | 26.9% | 10.49 | 28.4% |
| 1500 | 2 | |||||
| 2100 | 2 | |||||
| 24 | Hourly | 2 | 12.74 | 46.8% | 14.2 | 38.7% |
DISCUSSION
Our study illustrates dramatic differences in the discharge patterns from the ED and inpatient settings for children with respiratory illnesses. Although discharges from the ED occur at all hours, inpatient discharges were concentrated during midday. The time of discharge from the ED was highly related to the time of arrival, implying that any lag between discharge decision making and discharge was independent of time of arrival. The absence of a clear relationship between inpatient discharges and arrivals suggests that factors other than the clinical status of patients affect hospital LOS.
Because physician‐led rounds have traditionally been the point of decision‐making regarding inpatient discharge readiness, we hypothesized that increasing the frequency of rounds could reduce LOS. Within the parameters set in our simulation models, our results support this hypothesis. As the number of rounding sessions increased, hospital LOS decreased and the amount of time a child waited to be identified as ready for discharge approached zero. Simulating hourly rounding, analogous to the ED setting, resulted in the greatest reductions in inpatient LOS. Our findings have important implications for hospital operations and discharge policies given that children with respiratory conditions commonly experience short stay hospitalizations,[12, 13] but few children's hospitals have put in place models of care that differentiate the needs of short stay patients from those of inpatients requiring longer LOS.[14] To operationalize inpatient discharges as soon as a child is well enough to go home, parental expectations for discharge timing would need to be set, and discharge planning would need to begin at the start of each hospital stay.
Although there have been numerous studies on ED crowding, its causes, and potential solutions,[15, 16, 17, 18, 19, 20] this study is the first to our knowledge to demonstrate differences in discharge timing between the ED and inpatient general care unit. High levels of hospital occupancy decrease patient flow in both ED and inpatient units.[21, 22, 23, 24, 25] Shifting inpatient discharge for adult patients to earlier in the day can reduce or eliminate inpatient boarding in the ED.[26, 27] Evaluation of these relationships in pediatric populations are needed because of the unique care requirements of acutely ill children who often respond rapidly to hospital‐based treatment. In a recent opinion piece, Iantorno and Fieldston discourage hospitals from setting specified time targets for discharges and propose that high‐quality care includes afternoon and evening discharges.[28] The optimal timing for hospital discharges has not yet been defined, but our results indicate there is potential to reduce excess time spent in hospitals through the addition of rounds that would identify discharges throughout the day.
A simulation study based on hospital administrative data cannot determine a causal relationship between physician‐led rounds and hospital LOS. Still, our findings can generate discussions about hospital discharge policies and patient throughput initiatives. Hospitals can simulate other approaches that may apply to their institutional operations, such as changing rounding processes or adding an observation unit where discharges occur after hours. Hospitals could then pilot approaches locally. With the potential for unintended consequences of patient throughput initiatives that focus only on LOS as an outcome, pilot programs should be designed to track not only ED and inpatient LOS but also family satisfaction, access to follow‐up visits in primary care, need for reassessment in urgent care centers and EDs, and hospital readmissions.
The feasibility and acceptability of afternoon and evening rounds to improve patient throughput must be considered. Clinicians caring for hospitalized patients may have competing demands on their time, such as other clinical obligations, committee work, or academic pursuits that would make more frequent rounding unattractive. Physicians‐in‐training in teaching hospitals have educational requirements and duty‐hour restrictions that may limit their ability to round more frequently.[29, 30] There is also a need to define reasonable minimum standards for discharge processes (e.g., provision of patient education, discharge medications, and paperwork) for short‐stay patients and those requiring prolonged and/or complex hospitalizations. Streamlined discharge processes for short‐stay patients with simple illnesses may result in more efficient discharges. More efficient hospital discharges at flexible hours that are acceptable to families may require a culture shift among hospital staff. Although the palatability of off‐hours inpatient discharges has not been explored with patients and families, some may prefer this approach.
Limitations
There are several limitations to our study. First, the time of discharge is recorded differently for the ED and inpatient settings in our administrative dataset as a function of the electronic medical record systems in these 2 environments. Discharge from the ED reflects that time the patient left the ED but discharge from the inpatient setting reflects the time the patient exits the hospital ADT system. This likely biases our results toward longer LOS in the inpatient setting. We expect this would shift the band of inpatient discharges later in the day but do not expect this to alter the observed difference in discharge patterns between the ED and inpatient settings.
Second, administrative data do not provide information about the physicians making the discharge decisions or valid reasons a child would remain in the hospital after they improved clinically including time for teaching or care coordination. Our results therefore overestimate the amount of excess time associated with inpatient care. Future research is needed to determine the actual duration of time between when a child is clinically well enough for discharge home and when a child is actually discharged. There is also a need to understand discharge decision making and to identify the nonclinical factors that contribute to discharge delays.
Third, our data sample was taken from visits made by children to a pediatric ED nested within a general ED where children are admitted to an adjoining tertiary care academic children's hospital. Our results may not be generalizable to other settings in which children receive hospital‐based care, including freestanding children's hospitals and general hospitals that admit children.
Fourth, we focused our analyses on children with respiratory conditions, excluding visits with diagnoses suggestive of complex comorbid conditions and severe illnesses. Although we anticipate that children with these respiratory conditions treated in other hospitals will be similar to our population, our results may not hold for other conditions or across the full spectrum of severity for acute respiratory illness treated within EDs and inpatient units.
Finally, our simulation model makes assumptions, such as a fixed treatment times and discharge process length, which do not capture the clinical nuances of an individual child's response to hospital‐based treatment. Though much of the clinical complexity of hospital operations is not taken into account in the model, our intended purpose, to explore the influence of different rounding times on LOS, remains valid.
CONCLUSIONS
For children obtaining emergency care for respiratory illnesses, discharges from the ED occur around the clock, whereas discharges from the inpatient general care unit are concentrated during afternoon hours. Simulation models illustrate the potential to reduce hospital LOS by adding rounding sessions. Extending the hours of discharge for hospitalized children with respiratory illnesses may increase efficiency of care but could result in unintended consequences such as fewer opportunities for patient education.
Acknowledgments
The authors acknowledge the Center for Healthcare Engineering and Patient Safety, the Bonder Foundation, the Doctors Company Foundation, the Center for Research on Learning and Teaching, and the University of Michigan Summer Undergraduate Research in Engineering Program for their support of the students who contributed to this project.Disclosures: Funding for data analysis for this project was supported by a grant from the Center for Healthcare Research and Transformation (Dr. Macy). Support for the students working on this project (East, Burns, O'Gara, and Card) was provided through grants from the Center for Research and Teaching at the University of Michigan, the Center for Healthcare Engineering and Patient Safety at the University of Michigan, the Seth Bonder Foundation, and the Doctors Company Foundation. The authors have no financial or other conflicts of interest to disclose.
- , , , et al. Children's hospitals do not acutely respond to high occupancy. Pediatrics. 2010;125(5):974–981.
- , , , et al. Addressing inpatient crowding by smoothing occupancy at children's hospitals. J Hosp Med. 2011;6(8):462–468.
- , . The growing role of emergency departments in hospital admissions. N Engl J Med. 2012;367(5):391–393.
- , , . Managing asthma exacerbations in the emergency department: summary of the National Asthma Education and Prevention Program Expert Panel Report 3 guidelines for the management of asthma exacerbations. J Allergy Clin Immunol. 2009;124(2 suppl):S5–S14.
- American Academy of Pediatrics Subcommittee on Diagnosis and Management of Bronchiolitis. Diagnosis and management of bronchiolitis. Pediatrics. 2006;118(4):1774–1793.
- , , , et al. Executive summary: the management of community‐acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America. Clin Infect Dis. 2011;53(7):617–630.
- , , , , . Glucocorticoids for croup. Cochrane Database Syst Rev. 2011(1):CD001955.
- , , , et al. Epidemiology of a pediatric emergency medicine research network: the PECARN Core Data Project. Pediatr Emerg Care. 2006;22(10):689–699.
- , , , et al. Annual report on health care for children and youth in the United States: focus on trends in hospital use and quality. Acad Pediatr. 2011;11(4):263–279.
- , , , , , . Trends in high‐turnover stays among children hospitalized in the United States, 1993–2003. Pediatrics. 2009;123(3):996–1002.
- , , , , , . How commonly are children hospitalized for asthma eligible for care in alternative settings? Arch Pediatr Adolesc Med. 1999;153(1):49–55.
- , , , , . High turnover stays for pediatric asthma in the United States: analysis of the 2006 Kids' Inpatient Database. Med Care. 2010;48(9):827–833.
- , , , et al. Pediatric observation status: are we overlooking a growing population in children's hospitals? J Hosp Med. 2012;7(7):530–536.
- , , , et al. Differences in designations of observation care in US freestanding children's hospitals: are they virtual or real? J Hosp Med. 2012;7(4):287–293.
- , . Overcrowding in the nation's emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000;35(1):63–68.
- , , . Frequent overcrowding in U.S. emergency departments. Acad Emerg Med. 2001;8(2):151–155.
- , . Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emerg Med J. 2003;20(5):402–405.
- , , , et al. Solutions to emergency department 'boarding'[and crowding are underused and may need to be legislated. Health Aff (Millwood). 2012;31(8):1757–1766.
- . Managing emergency department overcrowding. Emerg Med Clin North Am. 2009;27(4):593–603, viii.
- Overcrowding crisis in our nation's emergency departments: is our safety net unraveling? Pediatrics. 2004;114(3):878–888.
- , , , et al. Crowding delays treatment and lengthens emergency department length of stay, even among high‐acuity patients. Ann Emerg Med. 2009;54(4):492–503.e494.
- . Do elective surgical and medical admissions impact emergency department length of stay measurements? Clin Invest Med. 2007;30(5):E177–E182.
- , , , . Unravelling relationships: hospital occupancy levels, discharge timing and emergency department access block. Emerg Med Australas. 2012;24(5):510–517.
- , , , . The effect of hospital bed occupancy on throughput in the pediatric emergency department. Ann Emerg Med. 2009;53(6):767–776.e763.
- , , , , . Impact of admission‐day crowding on the length of stay of pediatric hospitalizations. Pediatrics. 2008;121(4):e718–e730.
- , , , , , . The relationship between inpatient discharge timing and emergency department boarding. J Emerg Med. 2012;42(2):186–196.
- , , . Effect of coupling between emergency department and inpatient unit on the overcrowding in emergency department. Paper presented at: Winter Simulation Conference; 2007; Washington, DC.
- , . Hospitals are not hotels: high‐quality discharges occur around the clock. JAMA Pediatr. 2013;167(7):596–597.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , . Failure to thrive: pediatric residents weigh in on feasibility trial of the proposed 2008 institute of medicine work hour restrictions. J Grad Med Educ. 2009;1(2):181–184.
- , , , et al. Children's hospitals do not acutely respond to high occupancy. Pediatrics. 2010;125(5):974–981.
- , , , et al. Addressing inpatient crowding by smoothing occupancy at children's hospitals. J Hosp Med. 2011;6(8):462–468.
- , . The growing role of emergency departments in hospital admissions. N Engl J Med. 2012;367(5):391–393.
- , , . Managing asthma exacerbations in the emergency department: summary of the National Asthma Education and Prevention Program Expert Panel Report 3 guidelines for the management of asthma exacerbations. J Allergy Clin Immunol. 2009;124(2 suppl):S5–S14.
- American Academy of Pediatrics Subcommittee on Diagnosis and Management of Bronchiolitis. Diagnosis and management of bronchiolitis. Pediatrics. 2006;118(4):1774–1793.
- , , , et al. Executive summary: the management of community‐acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America. Clin Infect Dis. 2011;53(7):617–630.
- , , , , . Glucocorticoids for croup. Cochrane Database Syst Rev. 2011(1):CD001955.
- , , , et al. Epidemiology of a pediatric emergency medicine research network: the PECARN Core Data Project. Pediatr Emerg Care. 2006;22(10):689–699.
- , , , et al. Annual report on health care for children and youth in the United States: focus on trends in hospital use and quality. Acad Pediatr. 2011;11(4):263–279.
- , , , , , . Trends in high‐turnover stays among children hospitalized in the United States, 1993–2003. Pediatrics. 2009;123(3):996–1002.
- , , , , , . How commonly are children hospitalized for asthma eligible for care in alternative settings? Arch Pediatr Adolesc Med. 1999;153(1):49–55.
- , , , , . High turnover stays for pediatric asthma in the United States: analysis of the 2006 Kids' Inpatient Database. Med Care. 2010;48(9):827–833.
- , , , et al. Pediatric observation status: are we overlooking a growing population in children's hospitals? J Hosp Med. 2012;7(7):530–536.
- , , , et al. Differences in designations of observation care in US freestanding children's hospitals: are they virtual or real? J Hosp Med. 2012;7(4):287–293.
- , . Overcrowding in the nation's emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000;35(1):63–68.
- , , . Frequent overcrowding in U.S. emergency departments. Acad Emerg Med. 2001;8(2):151–155.
- , . Emergency department overcrowding in the United States: an emerging threat to patient safety and public health. Emerg Med J. 2003;20(5):402–405.
- , , , et al. Solutions to emergency department 'boarding'[and crowding are underused and may need to be legislated. Health Aff (Millwood). 2012;31(8):1757–1766.
- . Managing emergency department overcrowding. Emerg Med Clin North Am. 2009;27(4):593–603, viii.
- Overcrowding crisis in our nation's emergency departments: is our safety net unraveling? Pediatrics. 2004;114(3):878–888.
- , , , et al. Crowding delays treatment and lengthens emergency department length of stay, even among high‐acuity patients. Ann Emerg Med. 2009;54(4):492–503.e494.
- . Do elective surgical and medical admissions impact emergency department length of stay measurements? Clin Invest Med. 2007;30(5):E177–E182.
- , , , . Unravelling relationships: hospital occupancy levels, discharge timing and emergency department access block. Emerg Med Australas. 2012;24(5):510–517.
- , , , . The effect of hospital bed occupancy on throughput in the pediatric emergency department. Ann Emerg Med. 2009;53(6):767–776.e763.
- , , , , . Impact of admission‐day crowding on the length of stay of pediatric hospitalizations. Pediatrics. 2008;121(4):e718–e730.
- , , , , , . The relationship between inpatient discharge timing and emergency department boarding. J Emerg Med. 2012;42(2):186–196.
- , , . Effect of coupling between emergency department and inpatient unit on the overcrowding in emergency department. Paper presented at: Winter Simulation Conference; 2007; Washington, DC.
- , . Hospitals are not hotels: high‐quality discharges occur around the clock. JAMA Pediatr. 2013;167(7):596–597.
- , , , et al. Duty hour recommendations and implications for meeting the ACGME core competencies: views of residency directors. Mayo Clin Proc. 2011;86(3):185–191.
- , , , . Failure to thrive: pediatric residents weigh in on feasibility trial of the proposed 2008 institute of medicine work hour restrictions. J Grad Med Educ. 2009;1(2):181–184.
© 2013 Society of Hospital Medicine
