Hospital Medicine

Ascites is the most common complication of cirrhosis and often leads to hospitalization. ¹ Paracentesis is recommended for all patients admitted with ascites and cirrhosis. ¹ Additionally, the Society of Hospital Medicine considers the ability to perform paracenteses a core competency for hospitalists. ² Although considered a safe procedure, major bleeding complications occur in 0.2% to 1.7% of paracenteses. ^3-7 Patients with cirrhosis form new abdominal wall vessels because of portal hypertension, and hemoperitoneum from the laceration of these vessels during paracentesis carries a high morbidity and mortality. ^6,8 Ultrasound guidance using a low-frequency ultrasound probe is currently standard practice for paracentesis and has been shown to reduce bleeding complications. ^9-11 However, the use of vascular ultrasound (high-frequency probe) is also recommended to identify blood vessels within the intended needle pathway to reduce bleeding, but no studies have been performed to demonstrate a benefit. ^3,11 This study aimed to evaluate whether this “2-probe technique” reduces paracentesis-related bleeding complications.

The procedure service at Cedars Sinai Medical Center (CSMC) in Los Angeles performs paracentesis regularly with ultrasound guidance. CSMC is a tertiary care, academic medical center with 861 licensed beds. We performed a pre- to postintervention study of consecutive patients (admitted and ambulatory) who underwent paracentesis done by 1 proceduralist (MJA) from the procedure service at CSMC from February 2010 through February 2016. From February 1, 2010, through August 2011, paracenteses were performed using only low-frequency, phased array ultrasound probes (preintervention group). From September 1, 2011, through February 2016, a 2-probe technique was used, whereby ultrasound interrogation of the abdomen using a low-frequency, phased array probe (to identify ascites) was supplemented with a second scan using a high-frequency, linear probe to identify vasculature within the planned needle path (postintervention group). As a standard part of quality assurance, CSMC documented all paracentesis-related complications from procedures performed by their center. Northwestern University investigators (JHB, EC, JF) independently evaluated these data to look at bleeding complications before and after the implementation of the 2-probe technique. The CSMC and Northwestern University institutional review boards approved this study.

Procedure Protocol

Each patient’s primary team or outpatient physician requested a consultation for paracentesis from the CSMC procedure service. All patient evaluations began with an abdominal ultrasound using the low-frequency probe to determine the presence of ascites and a potential window of access to the fluid. After September 1, 2011, the CSMC procedure service implemented the 2-probe technique to also evaluate the abdominal wall for the presence of vessels. Color flow Doppler ultrasound further helped to differentiate blood vessels as necessary. The optimal window was then marked on the abdominal wall, and the paracentesis was performed. Per the routine of the CSMC procedure service, antiplatelet or anticoagulant medications were not held for paracenteses.

Measurement

All data were collected prospectively at the time of the procedure, including the volume of fluid removed, the number of needle passes required, and whether the patient was on antiplatelet or anticoagulant medications (including warfarin, direct oral anticoagulants, thrombin inhibitors, heparin, or low molecular weight heparins). Patients were followed for complications for up to 24 hours after the procedure or until a clinical question of a complication was reconciled. Minor bleeding was defined as new serosanguinous fluid on repeat paracentesis not associated with hemodynamic changes, local bruising or bleeding at the site, or abdominal wall hematoma. Major bleeding was defined by the development of hemodynamic instability or by reaccumulation of fluid on ultrasound within 24 hours postparacentesis and one of the following: an associated hemoglobin drop of greater than 2 g/dl, blood seen on repeat paracentesis, blood density fluid on a computed tomography scan, or the lack of an alternative explanation. All data were recorded in a handheld database (HanDbase; DDH Software, Wellington, FL).

A query of the electronic medical record was performed to obtain patient demographics and relevant clinical information, including age, sex, body mass index, International Normalized Ratio (INR), partial thromboplastin time (PTT), platelet counts (10³/uL, hematocrit (%) and creatinine (mg/dl). Our query for laboratory data retrieved the closest laboratory entry up to 48 hours before the procedure.

Statistical Analysis

We used a χ² test, Student t test, or Kruskal-Wallis test to compare demographic and clinical characteristics of procedure patients between the 2 study groups (pre- and postintervention). Major and minor bleeding were compared between the 2 groups using the χ² test.¹² We used the χ² test instead of the Fisher’s exact test for several reasons. The usual rule is that the Fisher’s exact test is necessary when 1 or more expected outcome values are less than 5. However, McDonald argues that the χ² test should be used with large sample sizes (more than 1000) in lieu of the outcome-value-of-5 rule.¹² The Fisher’s exact test also assumes that the row and column totals are fixed. However, the outcomes in our study were not fixed because any patient could have a bleeding complication during each procedure. When row and column totals are not fixed, only 5% of the time will a P value be less than 0.05, and the Fisher’s exact test is too conservative.¹² We performed all statistical analyses using IBM SPSS Statistics Version 22 (IBM Corp, Armonk, NY).

Patient demographic and clinical information can be found in the Table. The proceduralist (MJA) performed a total of 5777 paracenteses (1000 preintervention, 4777 postintervention) on 1639 patients. Four hundred eighty-nine (10.2%) vascular anomalies were identified within the intended needle path in the postintervention group (Figure). More patients in the preintervention group were on aspirin (93 [9.3%] vs 230 [4.8%]; P < 0.001) and therapeutic intravenous anticoagulants (33 [3.3%] vs 89 [1.9%]; P = 0.004), while more patients in the postintervention group were on both an antiplatelet and oral anticoagulant (1 [0.1%] vs 38 [0.8%]; P = 0.015) and subcutaneous prophylactic anticoagulants (184 [18.4%] vs 1120 [23.4%]; P = 0.001) at the time of the procedure. There were no other differences between groups with antiplatelet or anticoagulant drugs. We found no difference in minor bleeding between pre- and postintervention groups. Major bleeding was lower after the 2-probe technique was implemented (3 [0.3%] vs 4 [0.08%]; P = 0.07). There were no between-group differences in INR, PTT, or platelet counts among major bleeders. One patient in the postintervention group had hemodynamic instability and dropped his hemoglobin by 3.8 g/dl at 7 hours after the procedure. This was unexplained, as the patient had no abdominal symptoms or findings on examination. The patient received several liters of fluid before ultimately dying, and the primary team considered sepsis as a possible cause, but no postmortem examination was performed. This was the only death attributed to a major bleeding complication. We included this patient in our analysis because the cause of his demise was not completely clear. However, excluding this patient would change the results from a trend to a statistically significant difference between groups (3 [0.3%] vs 3 [0.06%]; P = 0.03).

To our knowledge, we report the largest series of paracentesis prospectively evaluated for bleeding complications, and this is the first study to evaluate whether adding a vascular ultrasound (high-frequency probe) avoids major bleeding. In our series, up to 10% of patients had abnormal vessels seen with a vascular ultrasound that were within the original intended trajectory path of the needle. These vessels were also likely present yet invisible when ultrasound-guided paracentesis using only the standard, low-frequency probe was being performed. It is unknown whether these vessels are routinely traversed with the needle, nicked, or narrowly avoided during paracenteses performed using only a low-frequency probe.

Procedure-related bleeding may not be completely avoidable, despite using the vascular probe. Some authors have suggested that the mechanism of bleeding is more related to the rapid reduction in intraperitoneal pressure, which increases the gradient across vessel walls, resulting in rupture and bleeding.⁶ However, in our series, using vascular ultrasound also reduced major bleeding to numbers lower than those historically reported in the literature (0.2%).^3-4 Our preintervention number needed to harm was 333 procedures to cause 1 major bleed, compared to 1250 (or 1666 using the 3-patient bleeding analysis) in the postintervention group. In 2008, 150,000 Medicare beneficiaries underwent paracentesis.¹³ Using our study analysis, if vascular ultrasound was used on these patients, up to 360 major bleeds may have been prevented, along with a corresponding reduction in unnecessary morbidity and mortality.

Our study has several limitations. First, it was limited to 1 center with 1 very experienced proceduralist. Although it is possible that the reduction in major bleeding may have been due to the increasing experience of the proceduralist over time, we do not think that this is likely because he had already performed thousands of paracenteses over 9 years before the start of our study. Second, major bleeding was rare and therefore precluded a multivariate analysis to control for temporal trends that might have occurred in our pre- to poststudy design. Statistically significant demographic and clinical variable differences between groups were likely not clinically meaningful. Although more patients were on intravenous anticoagulants in the preintervention group, coagulopathy or low platelets do not increase the bleeding risk during paracenteses,^1,8 and there was no clinical difference in INR, PTT, or platelets between groups (Table). Third, it is possible that unmeasured characteristics contributed to more patient complications in the preintervention group. Finally, we were unable to evaluate length of stay and mortality differences between groups that might have been attributable to the procedure because of the low number of major bleeding complications and the inability to perform a multivariate analysis.

Our results suggest that using the 2-probe technique to predetermine the needle path before performing paracentesis might prevent major bleeding. Based on our findings, we believe that the addition of a vascular ultrasound during paracentesis should be considered by all hospitalists.

Acknowledgments

The authors acknowledge Drs. Douglas Vaughan and Kevin O’Leary for their support and encouragement of this work. They would also like to thank the Cedars-Sinai Enterprise Information Systems Department for assistance with their data query.

Disclosure

The authors have no relevant financial disclosures or conflicts of interest to report.

Ascites is the most common complication of cirrhosis and often leads to hospitalization. ¹ Paracentesis is recommended for all patients admitted with ascites and cirrhosis. ¹ Additionally, the Society of Hospital Medicine considers the ability to perform paracenteses a core competency for hospitalists. ² Although considered a safe procedure, major bleeding complications occur in 0.2% to 1.7% of paracenteses. ^3-7 Patients with cirrhosis form new abdominal wall vessels because of portal hypertension, and hemoperitoneum from the laceration of these vessels during paracentesis carries a high morbidity and mortality. ^6,8 Ultrasound guidance using a low-frequency ultrasound probe is currently standard practice for paracentesis and has been shown to reduce bleeding complications. ^9-11 However, the use of vascular ultrasound (high-frequency probe) is also recommended to identify blood vessels within the intended needle pathway to reduce bleeding, but no studies have been performed to demonstrate a benefit. ^3,11 This study aimed to evaluate whether this “2-probe technique” reduces paracentesis-related bleeding complications.

The procedure service at Cedars Sinai Medical Center (CSMC) in Los Angeles performs paracentesis regularly with ultrasound guidance. CSMC is a tertiary care, academic medical center with 861 licensed beds. We performed a pre- to postintervention study of consecutive patients (admitted and ambulatory) who underwent paracentesis done by 1 proceduralist (MJA) from the procedure service at CSMC from February 2010 through February 2016. From February 1, 2010, through August 2011, paracenteses were performed using only low-frequency, phased array ultrasound probes (preintervention group). From September 1, 2011, through February 2016, a 2-probe technique was used, whereby ultrasound interrogation of the abdomen using a low-frequency, phased array probe (to identify ascites) was supplemented with a second scan using a high-frequency, linear probe to identify vasculature within the planned needle path (postintervention group). As a standard part of quality assurance, CSMC documented all paracentesis-related complications from procedures performed by their center. Northwestern University investigators (JHB, EC, JF) independently evaluated these data to look at bleeding complications before and after the implementation of the 2-probe technique. The CSMC and Northwestern University institutional review boards approved this study.

Procedure Protocol

Each patient’s primary team or outpatient physician requested a consultation for paracentesis from the CSMC procedure service. All patient evaluations began with an abdominal ultrasound using the low-frequency probe to determine the presence of ascites and a potential window of access to the fluid. After September 1, 2011, the CSMC procedure service implemented the 2-probe technique to also evaluate the abdominal wall for the presence of vessels. Color flow Doppler ultrasound further helped to differentiate blood vessels as necessary. The optimal window was then marked on the abdominal wall, and the paracentesis was performed. Per the routine of the CSMC procedure service, antiplatelet or anticoagulant medications were not held for paracenteses.

Measurement

All data were collected prospectively at the time of the procedure, including the volume of fluid removed, the number of needle passes required, and whether the patient was on antiplatelet or anticoagulant medications (including warfarin, direct oral anticoagulants, thrombin inhibitors, heparin, or low molecular weight heparins). Patients were followed for complications for up to 24 hours after the procedure or until a clinical question of a complication was reconciled. Minor bleeding was defined as new serosanguinous fluid on repeat paracentesis not associated with hemodynamic changes, local bruising or bleeding at the site, or abdominal wall hematoma. Major bleeding was defined by the development of hemodynamic instability or by reaccumulation of fluid on ultrasound within 24 hours postparacentesis and one of the following: an associated hemoglobin drop of greater than 2 g/dl, blood seen on repeat paracentesis, blood density fluid on a computed tomography scan, or the lack of an alternative explanation. All data were recorded in a handheld database (HanDbase; DDH Software, Wellington, FL).

A query of the electronic medical record was performed to obtain patient demographics and relevant clinical information, including age, sex, body mass index, International Normalized Ratio (INR), partial thromboplastin time (PTT), platelet counts (10³/uL, hematocrit (%) and creatinine (mg/dl). Our query for laboratory data retrieved the closest laboratory entry up to 48 hours before the procedure.

Statistical Analysis

We used a χ² test, Student t test, or Kruskal-Wallis test to compare demographic and clinical characteristics of procedure patients between the 2 study groups (pre- and postintervention). Major and minor bleeding were compared between the 2 groups using the χ² test.¹² We used the χ² test instead of the Fisher’s exact test for several reasons. The usual rule is that the Fisher’s exact test is necessary when 1 or more expected outcome values are less than 5. However, McDonald argues that the χ² test should be used with large sample sizes (more than 1000) in lieu of the outcome-value-of-5 rule.¹² The Fisher’s exact test also assumes that the row and column totals are fixed. However, the outcomes in our study were not fixed because any patient could have a bleeding complication during each procedure. When row and column totals are not fixed, only 5% of the time will a P value be less than 0.05, and the Fisher’s exact test is too conservative.¹² We performed all statistical analyses using IBM SPSS Statistics Version 22 (IBM Corp, Armonk, NY).

Patient demographic and clinical information can be found in the Table. The proceduralist (MJA) performed a total of 5777 paracenteses (1000 preintervention, 4777 postintervention) on 1639 patients. Four hundred eighty-nine (10.2%) vascular anomalies were identified within the intended needle path in the postintervention group (Figure). More patients in the preintervention group were on aspirin (93 [9.3%] vs 230 [4.8%]; P < 0.001) and therapeutic intravenous anticoagulants (33 [3.3%] vs 89 [1.9%]; P = 0.004), while more patients in the postintervention group were on both an antiplatelet and oral anticoagulant (1 [0.1%] vs 38 [0.8%]; P = 0.015) and subcutaneous prophylactic anticoagulants (184 [18.4%] vs 1120 [23.4%]; P = 0.001) at the time of the procedure. There were no other differences between groups with antiplatelet or anticoagulant drugs. We found no difference in minor bleeding between pre- and postintervention groups. Major bleeding was lower after the 2-probe technique was implemented (3 [0.3%] vs 4 [0.08%]; P = 0.07). There were no between-group differences in INR, PTT, or platelet counts among major bleeders. One patient in the postintervention group had hemodynamic instability and dropped his hemoglobin by 3.8 g/dl at 7 hours after the procedure. This was unexplained, as the patient had no abdominal symptoms or findings on examination. The patient received several liters of fluid before ultimately dying, and the primary team considered sepsis as a possible cause, but no postmortem examination was performed. This was the only death attributed to a major bleeding complication. We included this patient in our analysis because the cause of his demise was not completely clear. However, excluding this patient would change the results from a trend to a statistically significant difference between groups (3 [0.3%] vs 3 [0.06%]; P = 0.03).

To our knowledge, we report the largest series of paracentesis prospectively evaluated for bleeding complications, and this is the first study to evaluate whether adding a vascular ultrasound (high-frequency probe) avoids major bleeding. In our series, up to 10% of patients had abnormal vessels seen with a vascular ultrasound that were within the original intended trajectory path of the needle. These vessels were also likely present yet invisible when ultrasound-guided paracentesis using only the standard, low-frequency probe was being performed. It is unknown whether these vessels are routinely traversed with the needle, nicked, or narrowly avoided during paracenteses performed using only a low-frequency probe.

Procedure-related bleeding may not be completely avoidable, despite using the vascular probe. Some authors have suggested that the mechanism of bleeding is more related to the rapid reduction in intraperitoneal pressure, which increases the gradient across vessel walls, resulting in rupture and bleeding.⁶ However, in our series, using vascular ultrasound also reduced major bleeding to numbers lower than those historically reported in the literature (0.2%).^3-4 Our preintervention number needed to harm was 333 procedures to cause 1 major bleed, compared to 1250 (or 1666 using the 3-patient bleeding analysis) in the postintervention group. In 2008, 150,000 Medicare beneficiaries underwent paracentesis.¹³ Using our study analysis, if vascular ultrasound was used on these patients, up to 360 major bleeds may have been prevented, along with a corresponding reduction in unnecessary morbidity and mortality.

Our study has several limitations. First, it was limited to 1 center with 1 very experienced proceduralist. Although it is possible that the reduction in major bleeding may have been due to the increasing experience of the proceduralist over time, we do not think that this is likely because he had already performed thousands of paracenteses over 9 years before the start of our study. Second, major bleeding was rare and therefore precluded a multivariate analysis to control for temporal trends that might have occurred in our pre- to poststudy design. Statistically significant demographic and clinical variable differences between groups were likely not clinically meaningful. Although more patients were on intravenous anticoagulants in the preintervention group, coagulopathy or low platelets do not increase the bleeding risk during paracenteses,^1,8 and there was no clinical difference in INR, PTT, or platelets between groups (Table). Third, it is possible that unmeasured characteristics contributed to more patient complications in the preintervention group. Finally, we were unable to evaluate length of stay and mortality differences between groups that might have been attributable to the procedure because of the low number of major bleeding complications and the inability to perform a multivariate analysis.

Our results suggest that using the 2-probe technique to predetermine the needle path before performing paracentesis might prevent major bleeding. Based on our findings, we believe that the addition of a vascular ultrasound during paracentesis should be considered by all hospitalists.

Acknowledgments

The authors acknowledge Drs. Douglas Vaughan and Kevin O’Leary for their support and encouragement of this work. They would also like to thank the Cedars-Sinai Enterprise Information Systems Department for assistance with their data query.

Disclosure

The authors have no relevant financial disclosures or conflicts of interest to report.

Ascites is the most common complication of cirrhosis and often leads to hospitalization. ¹ Paracentesis is recommended for all patients admitted with ascites and cirrhosis. ¹ Additionally, the Society of Hospital Medicine considers the ability to perform paracenteses a core competency for hospitalists. ² Although considered a safe procedure, major bleeding complications occur in 0.2% to 1.7% of paracenteses. ^3-7 Patients with cirrhosis form new abdominal wall vessels because of portal hypertension, and hemoperitoneum from the laceration of these vessels during paracentesis carries a high morbidity and mortality. ^6,8 Ultrasound guidance using a low-frequency ultrasound probe is currently standard practice for paracentesis and has been shown to reduce bleeding complications. ^9-11 However, the use of vascular ultrasound (high-frequency probe) is also recommended to identify blood vessels within the intended needle pathway to reduce bleeding, but no studies have been performed to demonstrate a benefit. ^3,11 This study aimed to evaluate whether this “2-probe technique” reduces paracentesis-related bleeding complications.

The procedure service at Cedars Sinai Medical Center (CSMC) in Los Angeles performs paracentesis regularly with ultrasound guidance. CSMC is a tertiary care, academic medical center with 861 licensed beds. We performed a pre- to postintervention study of consecutive patients (admitted and ambulatory) who underwent paracentesis done by 1 proceduralist (MJA) from the procedure service at CSMC from February 2010 through February 2016. From February 1, 2010, through August 2011, paracenteses were performed using only low-frequency, phased array ultrasound probes (preintervention group). From September 1, 2011, through February 2016, a 2-probe technique was used, whereby ultrasound interrogation of the abdomen using a low-frequency, phased array probe (to identify ascites) was supplemented with a second scan using a high-frequency, linear probe to identify vasculature within the planned needle path (postintervention group). As a standard part of quality assurance, CSMC documented all paracentesis-related complications from procedures performed by their center. Northwestern University investigators (JHB, EC, JF) independently evaluated these data to look at bleeding complications before and after the implementation of the 2-probe technique. The CSMC and Northwestern University institutional review boards approved this study.

Procedure Protocol

Each patient’s primary team or outpatient physician requested a consultation for paracentesis from the CSMC procedure service. All patient evaluations began with an abdominal ultrasound using the low-frequency probe to determine the presence of ascites and a potential window of access to the fluid. After September 1, 2011, the CSMC procedure service implemented the 2-probe technique to also evaluate the abdominal wall for the presence of vessels. Color flow Doppler ultrasound further helped to differentiate blood vessels as necessary. The optimal window was then marked on the abdominal wall, and the paracentesis was performed. Per the routine of the CSMC procedure service, antiplatelet or anticoagulant medications were not held for paracenteses.

Measurement

All data were collected prospectively at the time of the procedure, including the volume of fluid removed, the number of needle passes required, and whether the patient was on antiplatelet or anticoagulant medications (including warfarin, direct oral anticoagulants, thrombin inhibitors, heparin, or low molecular weight heparins). Patients were followed for complications for up to 24 hours after the procedure or until a clinical question of a complication was reconciled. Minor bleeding was defined as new serosanguinous fluid on repeat paracentesis not associated with hemodynamic changes, local bruising or bleeding at the site, or abdominal wall hematoma. Major bleeding was defined by the development of hemodynamic instability or by reaccumulation of fluid on ultrasound within 24 hours postparacentesis and one of the following: an associated hemoglobin drop of greater than 2 g/dl, blood seen on repeat paracentesis, blood density fluid on a computed tomography scan, or the lack of an alternative explanation. All data were recorded in a handheld database (HanDbase; DDH Software, Wellington, FL).

A query of the electronic medical record was performed to obtain patient demographics and relevant clinical information, including age, sex, body mass index, International Normalized Ratio (INR), partial thromboplastin time (PTT), platelet counts (10³/uL, hematocrit (%) and creatinine (mg/dl). Our query for laboratory data retrieved the closest laboratory entry up to 48 hours before the procedure.

Statistical Analysis

We used a χ² test, Student t test, or Kruskal-Wallis test to compare demographic and clinical characteristics of procedure patients between the 2 study groups (pre- and postintervention). Major and minor bleeding were compared between the 2 groups using the χ² test.¹² We used the χ² test instead of the Fisher’s exact test for several reasons. The usual rule is that the Fisher’s exact test is necessary when 1 or more expected outcome values are less than 5. However, McDonald argues that the χ² test should be used with large sample sizes (more than 1000) in lieu of the outcome-value-of-5 rule.¹² The Fisher’s exact test also assumes that the row and column totals are fixed. However, the outcomes in our study were not fixed because any patient could have a bleeding complication during each procedure. When row and column totals are not fixed, only 5% of the time will a P value be less than 0.05, and the Fisher’s exact test is too conservative.¹² We performed all statistical analyses using IBM SPSS Statistics Version 22 (IBM Corp, Armonk, NY).

Patient demographic and clinical information can be found in the Table. The proceduralist (MJA) performed a total of 5777 paracenteses (1000 preintervention, 4777 postintervention) on 1639 patients. Four hundred eighty-nine (10.2%) vascular anomalies were identified within the intended needle path in the postintervention group (Figure). More patients in the preintervention group were on aspirin (93 [9.3%] vs 230 [4.8%]; P < 0.001) and therapeutic intravenous anticoagulants (33 [3.3%] vs 89 [1.9%]; P = 0.004), while more patients in the postintervention group were on both an antiplatelet and oral anticoagulant (1 [0.1%] vs 38 [0.8%]; P = 0.015) and subcutaneous prophylactic anticoagulants (184 [18.4%] vs 1120 [23.4%]; P = 0.001) at the time of the procedure. There were no other differences between groups with antiplatelet or anticoagulant drugs. We found no difference in minor bleeding between pre- and postintervention groups. Major bleeding was lower after the 2-probe technique was implemented (3 [0.3%] vs 4 [0.08%]; P = 0.07). There were no between-group differences in INR, PTT, or platelet counts among major bleeders. One patient in the postintervention group had hemodynamic instability and dropped his hemoglobin by 3.8 g/dl at 7 hours after the procedure. This was unexplained, as the patient had no abdominal symptoms or findings on examination. The patient received several liters of fluid before ultimately dying, and the primary team considered sepsis as a possible cause, but no postmortem examination was performed. This was the only death attributed to a major bleeding complication. We included this patient in our analysis because the cause of his demise was not completely clear. However, excluding this patient would change the results from a trend to a statistically significant difference between groups (3 [0.3%] vs 3 [0.06%]; P = 0.03).

To our knowledge, we report the largest series of paracentesis prospectively evaluated for bleeding complications, and this is the first study to evaluate whether adding a vascular ultrasound (high-frequency probe) avoids major bleeding. In our series, up to 10% of patients had abnormal vessels seen with a vascular ultrasound that were within the original intended trajectory path of the needle. These vessels were also likely present yet invisible when ultrasound-guided paracentesis using only the standard, low-frequency probe was being performed. It is unknown whether these vessels are routinely traversed with the needle, nicked, or narrowly avoided during paracenteses performed using only a low-frequency probe.

Procedure-related bleeding may not be completely avoidable, despite using the vascular probe. Some authors have suggested that the mechanism of bleeding is more related to the rapid reduction in intraperitoneal pressure, which increases the gradient across vessel walls, resulting in rupture and bleeding.⁶ However, in our series, using vascular ultrasound also reduced major bleeding to numbers lower than those historically reported in the literature (0.2%).^3-4 Our preintervention number needed to harm was 333 procedures to cause 1 major bleed, compared to 1250 (or 1666 using the 3-patient bleeding analysis) in the postintervention group. In 2008, 150,000 Medicare beneficiaries underwent paracentesis.¹³ Using our study analysis, if vascular ultrasound was used on these patients, up to 360 major bleeds may have been prevented, along with a corresponding reduction in unnecessary morbidity and mortality.

Our study has several limitations. First, it was limited to 1 center with 1 very experienced proceduralist. Although it is possible that the reduction in major bleeding may have been due to the increasing experience of the proceduralist over time, we do not think that this is likely because he had already performed thousands of paracenteses over 9 years before the start of our study. Second, major bleeding was rare and therefore precluded a multivariate analysis to control for temporal trends that might have occurred in our pre- to poststudy design. Statistically significant demographic and clinical variable differences between groups were likely not clinically meaningful. Although more patients were on intravenous anticoagulants in the preintervention group, coagulopathy or low platelets do not increase the bleeding risk during paracenteses,^1,8 and there was no clinical difference in INR, PTT, or platelets between groups (Table). Third, it is possible that unmeasured characteristics contributed to more patient complications in the preintervention group. Finally, we were unable to evaluate length of stay and mortality differences between groups that might have been attributable to the procedure because of the low number of major bleeding complications and the inability to perform a multivariate analysis.

Our results suggest that using the 2-probe technique to predetermine the needle path before performing paracentesis might prevent major bleeding. Based on our findings, we believe that the addition of a vascular ultrasound during paracentesis should be considered by all hospitalists.

Acknowledgments

The authors acknowledge Drs. Douglas Vaughan and Kevin O’Leary for their support and encouragement of this work. They would also like to thank the Cedars-Sinai Enterprise Information Systems Department for assistance with their data query.

Disclosure

The authors have no relevant financial disclosures or conflicts of interest to report.

The value placed on bedside clinical observation in the decision-making process of a patient’s illness has been diminished by today’s armamentarium of sophisticated technology. Increasing reliance is now placed on the result of nonspecific tests in preference to bedside clinical judgement in the diagnostic and management process. While diagnostic investigations have undoubtedly provided great advancements in medical care, they come at time and financial costs. Physicians should therefore continue to be encouraged to make clinical decisions based on their bedside assessment.

With hospital overcrowding a significant problem within the healthcare system and the expectation that it will worsen with an ageing population, identifying factors that predict patient suitability for discharge has become an important focus for clinicians.^1,2 There exists a paucity of literature predicting discharge suitability of general surgical patients admitted through the emergency department (ED). Furthermore, despite the extensive research into the effectiveness of discharge planning,³ little research has been conducted to describe positive predictive indicators for discharge. Observations made during surgical rounds have led the authors to consider that individuals who are using a smartphone during their bedside assessment may be clinically well enough for discharge.

The aim of this study was to assess whether the clinical assessment of an acute surgical patient could be usefully augmented by the observation of the active use of smartphones (the smartphone sign) and whether this could be used as a surrogate marker to indicate a patient’s well-being and suitability for same-day discharge from the hospital in acute surgical patients.

Design and Setting

This was a prospective observational study performed over 2 periods at a tertiary hospital in South Australia, Australia. At our institution, acute surgical patients are admitted to the acute surgical unit (ASU) from the ED by junior surgical doctors. Patients are then reviewed by the on-call surgical consultant, who implements management plans or advises discharge on 2 occasions per day.

Participants

All patients admitted under the ASU were considered eligible for the study. Exclusion criteria included patients that (i) required immediate surgical intervention (defined as time of review to theatre of less than 4 hours) and (ii) had immediate admission to the intensive care unit.

Consultant surgeons are employed within a general surgical subspecialty, including upper gastrointestinal, hepatobiliary, breast and endocrine, and colorectal. All surgeons from each team partake in the general surgery on-call roster. Each surgeon was included at least once within the observation periods. Experience of consultant surgeons ranged from 5 years of postfellowship experience to surgeons with more than 30 years of experience, with the majority having more than 10 years of postfellowship experience.

Patients were stratified into 2 distinct cohorts upon consultant review: smartphone positive (spP) was defined as a patient who was using a smartphone or who had their phone on their bed; a patient was classified as smartphone negative (spN) if they did not fulfil these criteria. The presence or absence of a smartphone was recorded by the authors, who were present on consultant ward rounds but not involved in the decision-making process of patient care. In order to minimize bias, only 1 surgeon (PGD) was aware that the study was being conducted and all patients were blinded to the study. Additional information that was collected included patient demographics, requirement for surgery, and length of stay (LOS). A patient who was discharged on the same day as the consultant review was considered to be discharged on day 1, all other patients were considered to have LOS greater than 1 day. Requirement for surgery was defined as a patient who underwent a surgical procedure in an operating suite. Thirty-day unplanned readmission rates for all patients were examined. Readmission to another public hospital within the state was also included within the readmission data.

Observation Periods

An initial 4-week pilot study was conducted to assess for a possible association between spP and same-day discharge. A second 8-week study period was undertaken 1 year later accounting for the employment of the authors at the study’s institution. Unless stated, the results described are the accumulation of both study periods.

Statistical Analysis

As this is the first study of its kind, no prior estimates of numbers were known. After 2 weeks of data collection, data were analyzed in order to provide an estimate of the total number of patients required to provide a statistically valid result (α = 0.05; power = 0.80). Sample size was calculated to be 40 subjects. It was agreed that in order to make the study as robust as possible, data should be collected for the 2 observation periods.

Demographic data are presented as means with standard deviations (SDs) or frequencies with percentages. A 2-sample Student t test was used to compare the age of spP and spN patients. A χ² test and logistic regressions were used to assess the association between smartphone status and patient demographics, LOS, and requirement for surgery. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). A P value of <0.05 was considered significant. All data were analyzed by using R 3.2.3 (R Foundation for Statistical Computing, Vienna, Austria).

During the 2 observation periods, a total of 227 eligible surgical admissions were observed with complete data for 221 patients. Six patients were excluded as their smartphone status was not recorded. The study sample represents our population of interest within an ASU, and we had complete data for 97.4% of participants with a 100% follow-up. There was no significant effect of study between the 2 observation periods (χ² = 140.19; P = 0.10). The mean age of patients was 50.24 years. Further demographic data are presented in Table 1. Twenty-five (11.3%) patients were spP and 196 (88.7%) were spN. Fifty-two (23.5%) patients were discharged home on day 1, and 169 (76.5%) had admissions longer than 1 day (see Figure). Sixty (27%) patients underwent surgery during their admission. Twenty-two patients had unplanned readmissions; only 1 of these patients had been observed to be spP.

There was a statistically significant difference in ages between the spP and spN groups (t = 8.40; P < 0.0005), with the average age of spP patients being 31.84 years compared with 52.58 years for spN patients. There was no statistical difference between gender and smartphone status (χ² = 1.78; P = 0.18; Table 2).

For those patients discharged home on day 1, there was a statistically significant association with being spP (χ² = 14.55, P = 0.0001). Patients who were spP were 5.29 times more likely to be discharged on day 1 (95% CI, 2.24-12.84). Of the variables analyzed, only gender failed to demonstrate an effect on discharge home on day 1 (Table 3). Overall, the presence of a smartphone was found to have a sensitivity of 56.0% (95% CI, 34.93-75.60) and a specificity of 80.6% (95% CI, 74.37-85.90) in regard to same-day discharge. However, it was found to have a negative predictive value of 93.49% (95% CI, 88.65-96.71).

When examining readmission rates, only 4% of spP patients were readmitted versus 10.7% of spN patients. Accounting for variables, spP patients were 4 times less likely to be readmitted, though this was not statistically significant (OR 4.02; 95% CI, 0.43-37.2; P = 0.22). Furthermore, when examining only those patients discharged on day 1, smartphone status was not a predictor of readmission (OR 0.94; 95% CI, 0.06-15.2; P = 0 .97).

To mitigate the effect of age, analysis was conducted excluding those aged over 55 years (the previous retirement age in Australia), leaving 131 patients for analysis. The average age of spP patients was 31.8 years (SD 10.0) compared with 36.7 years (SD 10.9) for spN patients, representing a significant difference (t = 2.14; P = 0.04); 51.1% of patients were male, 19.1% of patients were spP, 26.0% of patients proceeded to an operation, the oldest spP was 51 years, and 29.0% of patients were discharged home on day 1. There was no difference in gender and smartphone status (χ² = 0.33; P = 0.6). When analyzing those discharged on day 1, again spP patients were more likely to be discharged home (χ² = 9.4; P = 0.002), and spP patients were 3.6 times more likely to be discharged home on day 1.

There were 4 spP patients who underwent an operation. Two patients had an incision and drainage of a perianal abscess, 1 patient underwent a laparotomy for an internal hernia after recently undergoing a Roux-en-Y gastric bypass at another hospital, and the final patient underwent a laparoscopic appendicectomy. One of these patients was still discharged home on day 1.

As J. A. Lindsay⁴ said, “For one mistake made for not knowing, ten mistakes are made for not looking.” At medical school, we are taught the finer techniques of the physical examination in order to support our diagnosis made from the history. It is not until we are experienced clinicians do we develop the clinical acumen and ability to tell an unwell patient from a well patient at a glance—colloquially known as the “end of the bed” assessment. In the pretechnology era, a well patient could frequently be seen reading their book, eg, the “novel-sign.” With the advent of the smartphone and electronic devices upon which novels can be read, statuses updated, and locations “checked into” (ie, the modern “vital signs”), the book sign may be a thing of the past. However, the ability for the clinician to assess a patient’s wellness is still crucial, and the value of any additional “physical signs” need to be estimated.

We observed a cohort of patients through a busy ASU in a tertiary hospital in South Australia, Australia. Acute surgical patients admitted to the hospital who were observed to be on their phones upon consultant review were more than 5 times likely to be discharged that same day. To the best of our knowledge, this is the first study to prospectively collect data to assess a frequently used but unevaluated clinical observation.

The use of a smartphone can tell us a lot about an individual’s physiology. We can assume the individual’s airway and breathing are adequate, allowing enough oxygen to reach the lungs and subsequently circulate. The individual is usually sitting up in bed and thus has an adequate blood pressure and blood oxygenation that can maintain cerebral perfusion. They have the cognitive and cerebral processing in place to function the device, and we can examine their cerebellar function by looking for fine-motor movements.

Mobile phone ownership is pervasive within Australia,⁵ with a conservative estimated 85.7% of the population (20.57 million people of a total population of approximately 24 million) owning a mobile phone and an estimated 50% to 79% of mobile phone ownership being of a smartphone.^6,7 This ownership is not just limited to the young, with 74% of Australians over 65 owning or using a mobile phone.⁸ Despite this high phone ownership among those over 65, it is still significantly less than their younger counterparts and may be one reason for the absence of spP in those older than 51 years. A key point in the study is that overall phone ownership was not known, and, thus, it is not possible to determine the proportion of spN patients who were negative because they did not own a phone. However, based on general population data, the incidence of spP patients was well below that seen in the community (11.3%)⁵ and even when excluding those over 55, the percentage of spP patients only rose to 19.1%. Unsurprisingly, increasing age was associated with a decreased likelihood of being spP (P < 0.0005), as younger people are more likely to own a phone.⁸ There was no association with gender (P = 0.18). There are a number of explanations that may explain the lower than expected percentage of spP patients, including the inability for the patient to gather their possessions during a medical emergency, patients storing their phones prior to doctor review (72%-85% of Australians report talking on phones in public places to be rude or intrusive⁵), but more importantly, that our hypothesis that patients were too unwell to use their device appears to hold true.

There are potential alternate reasons other than smartphone status that may account for patients being discharged home on day 1. While there was no association seen with gender, the need for an operation prolonged a patient’s stay (OR 1.64; 95% CI, 0.046-0.46), and there was a trend seen with increasing age (OR 0.98; 95% CI, 0.96-1.00). Neither of these 2 demographics are unsurprising: increasing age is associated with increasing medical comorbidities and thus complexity; even the simplest of operations require a postprocedure observation period, automatically increasing their LOS. Additionally, measured demographics are limited and there may be further unmeasured reasons that account for earlier discharge.

The other key component to this study is the value of the physical examination, albeit only assessing 1 component: the general inspection. In their review of the value of the physical examination of the cardiovascular system, Elder et al. highlight an important point: in traditional teaching, the value of a physical sign is compared with a diagnostic reference, typically imaging or an invasive test.⁹ They argue that this definition undervalues the physical examination and list other values aside from accuracy including accessibility, contribution to clinical care beyond diagnoses, cost effectiveness, patients’ safety, patients’ perceptions, and pedagogic value; and they argue that the physical examination should always be considered in regard to the clinical context—in this case, the newly admitted general surgical patient.

The assessment of the presence or absence of a smartphone is readily performed upon general inspection and is easily visible; general inspection of the patient and failure to observe the clinical sign when present are 2 of the greatest errors associated with physical examination.¹⁰ Furthermore, given its unique status as a physical sign, the authors’ opinion and experience is that it is readily teachable. McGee states, “…a fundamental lesson [in regards to teaching] is that the diagnosis of many clinical problems, despite modern testing, still depends primarily on what the clinician sees, hears, and feels.”¹¹ In their article, Paley et al. found that more than 80% of patients admitted from the ED under internal medicine could be accurately diagnosed based largely on history and examination alone and concluded that basic clinical skills are sufficient for achieving an accurate diagnosis in most cases.¹² Although Paley et al. were assisted with basic tests (such as electrocardiogram and basic haematological and biochemistry results), the point of clinical skills is not lost. Furthermore, this assessment was made in a group of patients generally considered to be complex in contrast to the “standard” appendicitis or cholecystitis patient that makes up a significant proportion of general surgical patients.

There are a number of limitations to this study, however, including smartphones that may have been missed during the observational period. Potential confounding variables such as socioeconomic status and the overall smartphone ownership of our subjects were not known. We did not ask all admitted patients whether they owned a phone or whether they had a phone in their possession. Knowledge of those who owned phones but were not in possession of them could strengthen our argument that spN patients were not using their phone because they were unwell, rather than just not having access to it.

However, this study has a number of strengths, including a large sample size and data that were prospectively collected by a method and in a setting that was the same for all participants. Clear and appropriate definitions were used, which minimizes misclassification bias. Participants and decision makers were blinded to the study, and potentially confounding variables such as age and sex were accounted for.

Assessing the suitability for discharge from the hospital is a decision encountered by hospital-based clinicians every day. These skills are not taught, but are rather learned as a junior doctor acquires experience. It is unlikely that protocols will be developed to aid identification of potential discharges from an acute surgical ward; acute surgical conditions are too varied and dynamic to be able to pool all data. We continue to rely on our own and fellow colleagues’ (doctors, nurses, and other staff) input and assessment. However, our study has shown that it is possible to identify and quantify clinical findings that are already regularly used, albeit potentially subconsciously, to assess suitability for discharge. We have shown in this large, prospectively collected observational study that if a surgical patient is seen using their electronic device, they are more likely to be safe to go home. Thus, surgeons can reliably use this observation as a trigger to consider discharging the patient following a more thorough assessment.

While these observations might appear to be rather a simplistic way of trying to quantify whether or not a patient is fit for discharge, any clues that hint towards a patient’s well-being should be taken into account when making an overall assessment. The active use of a smartphone is one such measure.

Acknowledgments

The authors thank Emma Knight and Nancy Briggs from the Data Management & Analysis Centre, Discipline of Public Health, University of Adelaide.

Disclosure

No author nor the institution received any payment or services from a third party for any aspect of the submitted work and report no conflict of interest. There are no reported financial relationships with any entities by any of the authors. There are no patents pending based upon this publication. There are no relationships or activities that readers could perceive to have influenced, or give the appearance of influencing, the submitted work. The corresponding author is not in receipt of a research scholarship. The paper is not based on a previous communication.

The value placed on bedside clinical observation in the decision-making process of a patient’s illness has been diminished by today’s armamentarium of sophisticated technology. Increasing reliance is now placed on the result of nonspecific tests in preference to bedside clinical judgement in the diagnostic and management process. While diagnostic investigations have undoubtedly provided great advancements in medical care, they come at time and financial costs. Physicians should therefore continue to be encouraged to make clinical decisions based on their bedside assessment.

With hospital overcrowding a significant problem within the healthcare system and the expectation that it will worsen with an ageing population, identifying factors that predict patient suitability for discharge has become an important focus for clinicians.^1,2 There exists a paucity of literature predicting discharge suitability of general surgical patients admitted through the emergency department (ED). Furthermore, despite the extensive research into the effectiveness of discharge planning,³ little research has been conducted to describe positive predictive indicators for discharge. Observations made during surgical rounds have led the authors to consider that individuals who are using a smartphone during their bedside assessment may be clinically well enough for discharge.

The aim of this study was to assess whether the clinical assessment of an acute surgical patient could be usefully augmented by the observation of the active use of smartphones (the smartphone sign) and whether this could be used as a surrogate marker to indicate a patient’s well-being and suitability for same-day discharge from the hospital in acute surgical patients.

Design and Setting

This was a prospective observational study performed over 2 periods at a tertiary hospital in South Australia, Australia. At our institution, acute surgical patients are admitted to the acute surgical unit (ASU) from the ED by junior surgical doctors. Patients are then reviewed by the on-call surgical consultant, who implements management plans or advises discharge on 2 occasions per day.

Participants

All patients admitted under the ASU were considered eligible for the study. Exclusion criteria included patients that (i) required immediate surgical intervention (defined as time of review to theatre of less than 4 hours) and (ii) had immediate admission to the intensive care unit.

Consultant surgeons are employed within a general surgical subspecialty, including upper gastrointestinal, hepatobiliary, breast and endocrine, and colorectal. All surgeons from each team partake in the general surgery on-call roster. Each surgeon was included at least once within the observation periods. Experience of consultant surgeons ranged from 5 years of postfellowship experience to surgeons with more than 30 years of experience, with the majority having more than 10 years of postfellowship experience.

Patients were stratified into 2 distinct cohorts upon consultant review: smartphone positive (spP) was defined as a patient who was using a smartphone or who had their phone on their bed; a patient was classified as smartphone negative (spN) if they did not fulfil these criteria. The presence or absence of a smartphone was recorded by the authors, who were present on consultant ward rounds but not involved in the decision-making process of patient care. In order to minimize bias, only 1 surgeon (PGD) was aware that the study was being conducted and all patients were blinded to the study. Additional information that was collected included patient demographics, requirement for surgery, and length of stay (LOS). A patient who was discharged on the same day as the consultant review was considered to be discharged on day 1, all other patients were considered to have LOS greater than 1 day. Requirement for surgery was defined as a patient who underwent a surgical procedure in an operating suite. Thirty-day unplanned readmission rates for all patients were examined. Readmission to another public hospital within the state was also included within the readmission data.

Observation Periods

An initial 4-week pilot study was conducted to assess for a possible association between spP and same-day discharge. A second 8-week study period was undertaken 1 year later accounting for the employment of the authors at the study’s institution. Unless stated, the results described are the accumulation of both study periods.

Statistical Analysis

As this is the first study of its kind, no prior estimates of numbers were known. After 2 weeks of data collection, data were analyzed in order to provide an estimate of the total number of patients required to provide a statistically valid result (α = 0.05; power = 0.80). Sample size was calculated to be 40 subjects. It was agreed that in order to make the study as robust as possible, data should be collected for the 2 observation periods.

Demographic data are presented as means with standard deviations (SDs) or frequencies with percentages. A 2-sample Student t test was used to compare the age of spP and spN patients. A χ² test and logistic regressions were used to assess the association between smartphone status and patient demographics, LOS, and requirement for surgery. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). A P value of <0.05 was considered significant. All data were analyzed by using R 3.2.3 (R Foundation for Statistical Computing, Vienna, Austria).

During the 2 observation periods, a total of 227 eligible surgical admissions were observed with complete data for 221 patients. Six patients were excluded as their smartphone status was not recorded. The study sample represents our population of interest within an ASU, and we had complete data for 97.4% of participants with a 100% follow-up. There was no significant effect of study between the 2 observation periods (χ² = 140.19; P = 0.10). The mean age of patients was 50.24 years. Further demographic data are presented in Table 1. Twenty-five (11.3%) patients were spP and 196 (88.7%) were spN. Fifty-two (23.5%) patients were discharged home on day 1, and 169 (76.5%) had admissions longer than 1 day (see Figure). Sixty (27%) patients underwent surgery during their admission. Twenty-two patients had unplanned readmissions; only 1 of these patients had been observed to be spP.

There was a statistically significant difference in ages between the spP and spN groups (t = 8.40; P < 0.0005), with the average age of spP patients being 31.84 years compared with 52.58 years for spN patients. There was no statistical difference between gender and smartphone status (χ² = 1.78; P = 0.18; Table 2).

For those patients discharged home on day 1, there was a statistically significant association with being spP (χ² = 14.55, P = 0.0001). Patients who were spP were 5.29 times more likely to be discharged on day 1 (95% CI, 2.24-12.84). Of the variables analyzed, only gender failed to demonstrate an effect on discharge home on day 1 (Table 3). Overall, the presence of a smartphone was found to have a sensitivity of 56.0% (95% CI, 34.93-75.60) and a specificity of 80.6% (95% CI, 74.37-85.90) in regard to same-day discharge. However, it was found to have a negative predictive value of 93.49% (95% CI, 88.65-96.71).

When examining readmission rates, only 4% of spP patients were readmitted versus 10.7% of spN patients. Accounting for variables, spP patients were 4 times less likely to be readmitted, though this was not statistically significant (OR 4.02; 95% CI, 0.43-37.2; P = 0.22). Furthermore, when examining only those patients discharged on day 1, smartphone status was not a predictor of readmission (OR 0.94; 95% CI, 0.06-15.2; P = 0 .97).

To mitigate the effect of age, analysis was conducted excluding those aged over 55 years (the previous retirement age in Australia), leaving 131 patients for analysis. The average age of spP patients was 31.8 years (SD 10.0) compared with 36.7 years (SD 10.9) for spN patients, representing a significant difference (t = 2.14; P = 0.04); 51.1% of patients were male, 19.1% of patients were spP, 26.0% of patients proceeded to an operation, the oldest spP was 51 years, and 29.0% of patients were discharged home on day 1. There was no difference in gender and smartphone status (χ² = 0.33; P = 0.6). When analyzing those discharged on day 1, again spP patients were more likely to be discharged home (χ² = 9.4; P = 0.002), and spP patients were 3.6 times more likely to be discharged home on day 1.

There were 4 spP patients who underwent an operation. Two patients had an incision and drainage of a perianal abscess, 1 patient underwent a laparotomy for an internal hernia after recently undergoing a Roux-en-Y gastric bypass at another hospital, and the final patient underwent a laparoscopic appendicectomy. One of these patients was still discharged home on day 1.

As J. A. Lindsay⁴ said, “For one mistake made for not knowing, ten mistakes are made for not looking.” At medical school, we are taught the finer techniques of the physical examination in order to support our diagnosis made from the history. It is not until we are experienced clinicians do we develop the clinical acumen and ability to tell an unwell patient from a well patient at a glance—colloquially known as the “end of the bed” assessment. In the pretechnology era, a well patient could frequently be seen reading their book, eg, the “novel-sign.” With the advent of the smartphone and electronic devices upon which novels can be read, statuses updated, and locations “checked into” (ie, the modern “vital signs”), the book sign may be a thing of the past. However, the ability for the clinician to assess a patient’s wellness is still crucial, and the value of any additional “physical signs” need to be estimated.

We observed a cohort of patients through a busy ASU in a tertiary hospital in South Australia, Australia. Acute surgical patients admitted to the hospital who were observed to be on their phones upon consultant review were more than 5 times likely to be discharged that same day. To the best of our knowledge, this is the first study to prospectively collect data to assess a frequently used but unevaluated clinical observation.

The use of a smartphone can tell us a lot about an individual’s physiology. We can assume the individual’s airway and breathing are adequate, allowing enough oxygen to reach the lungs and subsequently circulate. The individual is usually sitting up in bed and thus has an adequate blood pressure and blood oxygenation that can maintain cerebral perfusion. They have the cognitive and cerebral processing in place to function the device, and we can examine their cerebellar function by looking for fine-motor movements.

Mobile phone ownership is pervasive within Australia,⁵ with a conservative estimated 85.7% of the population (20.57 million people of a total population of approximately 24 million) owning a mobile phone and an estimated 50% to 79% of mobile phone ownership being of a smartphone.^6,7 This ownership is not just limited to the young, with 74% of Australians over 65 owning or using a mobile phone.⁸ Despite this high phone ownership among those over 65, it is still significantly less than their younger counterparts and may be one reason for the absence of spP in those older than 51 years. A key point in the study is that overall phone ownership was not known, and, thus, it is not possible to determine the proportion of spN patients who were negative because they did not own a phone. However, based on general population data, the incidence of spP patients was well below that seen in the community (11.3%)⁵ and even when excluding those over 55, the percentage of spP patients only rose to 19.1%. Unsurprisingly, increasing age was associated with a decreased likelihood of being spP (P < 0.0005), as younger people are more likely to own a phone.⁸ There was no association with gender (P = 0.18). There are a number of explanations that may explain the lower than expected percentage of spP patients, including the inability for the patient to gather their possessions during a medical emergency, patients storing their phones prior to doctor review (72%-85% of Australians report talking on phones in public places to be rude or intrusive⁵), but more importantly, that our hypothesis that patients were too unwell to use their device appears to hold true.

There are potential alternate reasons other than smartphone status that may account for patients being discharged home on day 1. While there was no association seen with gender, the need for an operation prolonged a patient’s stay (OR 1.64; 95% CI, 0.046-0.46), and there was a trend seen with increasing age (OR 0.98; 95% CI, 0.96-1.00). Neither of these 2 demographics are unsurprising: increasing age is associated with increasing medical comorbidities and thus complexity; even the simplest of operations require a postprocedure observation period, automatically increasing their LOS. Additionally, measured demographics are limited and there may be further unmeasured reasons that account for earlier discharge.

The other key component to this study is the value of the physical examination, albeit only assessing 1 component: the general inspection. In their review of the value of the physical examination of the cardiovascular system, Elder et al. highlight an important point: in traditional teaching, the value of a physical sign is compared with a diagnostic reference, typically imaging or an invasive test.⁹ They argue that this definition undervalues the physical examination and list other values aside from accuracy including accessibility, contribution to clinical care beyond diagnoses, cost effectiveness, patients’ safety, patients’ perceptions, and pedagogic value; and they argue that the physical examination should always be considered in regard to the clinical context—in this case, the newly admitted general surgical patient.

The assessment of the presence or absence of a smartphone is readily performed upon general inspection and is easily visible; general inspection of the patient and failure to observe the clinical sign when present are 2 of the greatest errors associated with physical examination.¹⁰ Furthermore, given its unique status as a physical sign, the authors’ opinion and experience is that it is readily teachable. McGee states, “…a fundamental lesson [in regards to teaching] is that the diagnosis of many clinical problems, despite modern testing, still depends primarily on what the clinician sees, hears, and feels.”¹¹ In their article, Paley et al. found that more than 80% of patients admitted from the ED under internal medicine could be accurately diagnosed based largely on history and examination alone and concluded that basic clinical skills are sufficient for achieving an accurate diagnosis in most cases.¹² Although Paley et al. were assisted with basic tests (such as electrocardiogram and basic haematological and biochemistry results), the point of clinical skills is not lost. Furthermore, this assessment was made in a group of patients generally considered to be complex in contrast to the “standard” appendicitis or cholecystitis patient that makes up a significant proportion of general surgical patients.

There are a number of limitations to this study, however, including smartphones that may have been missed during the observational period. Potential confounding variables such as socioeconomic status and the overall smartphone ownership of our subjects were not known. We did not ask all admitted patients whether they owned a phone or whether they had a phone in their possession. Knowledge of those who owned phones but were not in possession of them could strengthen our argument that spN patients were not using their phone because they were unwell, rather than just not having access to it.

However, this study has a number of strengths, including a large sample size and data that were prospectively collected by a method and in a setting that was the same for all participants. Clear and appropriate definitions were used, which minimizes misclassification bias. Participants and decision makers were blinded to the study, and potentially confounding variables such as age and sex were accounted for.

Assessing the suitability for discharge from the hospital is a decision encountered by hospital-based clinicians every day. These skills are not taught, but are rather learned as a junior doctor acquires experience. It is unlikely that protocols will be developed to aid identification of potential discharges from an acute surgical ward; acute surgical conditions are too varied and dynamic to be able to pool all data. We continue to rely on our own and fellow colleagues’ (doctors, nurses, and other staff) input and assessment. However, our study has shown that it is possible to identify and quantify clinical findings that are already regularly used, albeit potentially subconsciously, to assess suitability for discharge. We have shown in this large, prospectively collected observational study that if a surgical patient is seen using their electronic device, they are more likely to be safe to go home. Thus, surgeons can reliably use this observation as a trigger to consider discharging the patient following a more thorough assessment.

While these observations might appear to be rather a simplistic way of trying to quantify whether or not a patient is fit for discharge, any clues that hint towards a patient’s well-being should be taken into account when making an overall assessment. The active use of a smartphone is one such measure.

Acknowledgments

The authors thank Emma Knight and Nancy Briggs from the Data Management & Analysis Centre, Discipline of Public Health, University of Adelaide.

Disclosure

No author nor the institution received any payment or services from a third party for any aspect of the submitted work and report no conflict of interest. There are no reported financial relationships with any entities by any of the authors. There are no patents pending based upon this publication. There are no relationships or activities that readers could perceive to have influenced, or give the appearance of influencing, the submitted work. The corresponding author is not in receipt of a research scholarship. The paper is not based on a previous communication.

The value placed on bedside clinical observation in the decision-making process of a patient’s illness has been diminished by today’s armamentarium of sophisticated technology. Increasing reliance is now placed on the result of nonspecific tests in preference to bedside clinical judgement in the diagnostic and management process. While diagnostic investigations have undoubtedly provided great advancements in medical care, they come at time and financial costs. Physicians should therefore continue to be encouraged to make clinical decisions based on their bedside assessment.

With hospital overcrowding a significant problem within the healthcare system and the expectation that it will worsen with an ageing population, identifying factors that predict patient suitability for discharge has become an important focus for clinicians.^1,2 There exists a paucity of literature predicting discharge suitability of general surgical patients admitted through the emergency department (ED). Furthermore, despite the extensive research into the effectiveness of discharge planning,³ little research has been conducted to describe positive predictive indicators for discharge. Observations made during surgical rounds have led the authors to consider that individuals who are using a smartphone during their bedside assessment may be clinically well enough for discharge.

The aim of this study was to assess whether the clinical assessment of an acute surgical patient could be usefully augmented by the observation of the active use of smartphones (the smartphone sign) and whether this could be used as a surrogate marker to indicate a patient’s well-being and suitability for same-day discharge from the hospital in acute surgical patients.

Design and Setting

This was a prospective observational study performed over 2 periods at a tertiary hospital in South Australia, Australia. At our institution, acute surgical patients are admitted to the acute surgical unit (ASU) from the ED by junior surgical doctors. Patients are then reviewed by the on-call surgical consultant, who implements management plans or advises discharge on 2 occasions per day.

Participants

All patients admitted under the ASU were considered eligible for the study. Exclusion criteria included patients that (i) required immediate surgical intervention (defined as time of review to theatre of less than 4 hours) and (ii) had immediate admission to the intensive care unit.

Consultant surgeons are employed within a general surgical subspecialty, including upper gastrointestinal, hepatobiliary, breast and endocrine, and colorectal. All surgeons from each team partake in the general surgery on-call roster. Each surgeon was included at least once within the observation periods. Experience of consultant surgeons ranged from 5 years of postfellowship experience to surgeons with more than 30 years of experience, with the majority having more than 10 years of postfellowship experience.

Patients were stratified into 2 distinct cohorts upon consultant review: smartphone positive (spP) was defined as a patient who was using a smartphone or who had their phone on their bed; a patient was classified as smartphone negative (spN) if they did not fulfil these criteria. The presence or absence of a smartphone was recorded by the authors, who were present on consultant ward rounds but not involved in the decision-making process of patient care. In order to minimize bias, only 1 surgeon (PGD) was aware that the study was being conducted and all patients were blinded to the study. Additional information that was collected included patient demographics, requirement for surgery, and length of stay (LOS). A patient who was discharged on the same day as the consultant review was considered to be discharged on day 1, all other patients were considered to have LOS greater than 1 day. Requirement for surgery was defined as a patient who underwent a surgical procedure in an operating suite. Thirty-day unplanned readmission rates for all patients were examined. Readmission to another public hospital within the state was also included within the readmission data.

Observation Periods

An initial 4-week pilot study was conducted to assess for a possible association between spP and same-day discharge. A second 8-week study period was undertaken 1 year later accounting for the employment of the authors at the study’s institution. Unless stated, the results described are the accumulation of both study periods.

Statistical Analysis

As this is the first study of its kind, no prior estimates of numbers were known. After 2 weeks of data collection, data were analyzed in order to provide an estimate of the total number of patients required to provide a statistically valid result (α = 0.05; power = 0.80). Sample size was calculated to be 40 subjects. It was agreed that in order to make the study as robust as possible, data should be collected for the 2 observation periods.

Demographic data are presented as means with standard deviations (SDs) or frequencies with percentages. A 2-sample Student t test was used to compare the age of spP and spN patients. A χ² test and logistic regressions were used to assess the association between smartphone status and patient demographics, LOS, and requirement for surgery. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). A P value of <0.05 was considered significant. All data were analyzed by using R 3.2.3 (R Foundation for Statistical Computing, Vienna, Austria).

During the 2 observation periods, a total of 227 eligible surgical admissions were observed with complete data for 221 patients. Six patients were excluded as their smartphone status was not recorded. The study sample represents our population of interest within an ASU, and we had complete data for 97.4% of participants with a 100% follow-up. There was no significant effect of study between the 2 observation periods (χ² = 140.19; P = 0.10). The mean age of patients was 50.24 years. Further demographic data are presented in Table 1. Twenty-five (11.3%) patients were spP and 196 (88.7%) were spN. Fifty-two (23.5%) patients were discharged home on day 1, and 169 (76.5%) had admissions longer than 1 day (see Figure). Sixty (27%) patients underwent surgery during their admission. Twenty-two patients had unplanned readmissions; only 1 of these patients had been observed to be spP.

There was a statistically significant difference in ages between the spP and spN groups (t = 8.40; P < 0.0005), with the average age of spP patients being 31.84 years compared with 52.58 years for spN patients. There was no statistical difference between gender and smartphone status (χ² = 1.78; P = 0.18; Table 2).

For those patients discharged home on day 1, there was a statistically significant association with being spP (χ² = 14.55, P = 0.0001). Patients who were spP were 5.29 times more likely to be discharged on day 1 (95% CI, 2.24-12.84). Of the variables analyzed, only gender failed to demonstrate an effect on discharge home on day 1 (Table 3). Overall, the presence of a smartphone was found to have a sensitivity of 56.0% (95% CI, 34.93-75.60) and a specificity of 80.6% (95% CI, 74.37-85.90) in regard to same-day discharge. However, it was found to have a negative predictive value of 93.49% (95% CI, 88.65-96.71).

When examining readmission rates, only 4% of spP patients were readmitted versus 10.7% of spN patients. Accounting for variables, spP patients were 4 times less likely to be readmitted, though this was not statistically significant (OR 4.02; 95% CI, 0.43-37.2; P = 0.22). Furthermore, when examining only those patients discharged on day 1, smartphone status was not a predictor of readmission (OR 0.94; 95% CI, 0.06-15.2; P = 0 .97).

To mitigate the effect of age, analysis was conducted excluding those aged over 55 years (the previous retirement age in Australia), leaving 131 patients for analysis. The average age of spP patients was 31.8 years (SD 10.0) compared with 36.7 years (SD 10.9) for spN patients, representing a significant difference (t = 2.14; P = 0.04); 51.1% of patients were male, 19.1% of patients were spP, 26.0% of patients proceeded to an operation, the oldest spP was 51 years, and 29.0% of patients were discharged home on day 1. There was no difference in gender and smartphone status (χ² = 0.33; P = 0.6). When analyzing those discharged on day 1, again spP patients were more likely to be discharged home (χ² = 9.4; P = 0.002), and spP patients were 3.6 times more likely to be discharged home on day 1.

There were 4 spP patients who underwent an operation. Two patients had an incision and drainage of a perianal abscess, 1 patient underwent a laparotomy for an internal hernia after recently undergoing a Roux-en-Y gastric bypass at another hospital, and the final patient underwent a laparoscopic appendicectomy. One of these patients was still discharged home on day 1.

As J. A. Lindsay⁴ said, “For one mistake made for not knowing, ten mistakes are made for not looking.” At medical school, we are taught the finer techniques of the physical examination in order to support our diagnosis made from the history. It is not until we are experienced clinicians do we develop the clinical acumen and ability to tell an unwell patient from a well patient at a glance—colloquially known as the “end of the bed” assessment. In the pretechnology era, a well patient could frequently be seen reading their book, eg, the “novel-sign.” With the advent of the smartphone and electronic devices upon which novels can be read, statuses updated, and locations “checked into” (ie, the modern “vital signs”), the book sign may be a thing of the past. However, the ability for the clinician to assess a patient’s wellness is still crucial, and the value of any additional “physical signs” need to be estimated.

We observed a cohort of patients through a busy ASU in a tertiary hospital in South Australia, Australia. Acute surgical patients admitted to the hospital who were observed to be on their phones upon consultant review were more than 5 times likely to be discharged that same day. To the best of our knowledge, this is the first study to prospectively collect data to assess a frequently used but unevaluated clinical observation.

The use of a smartphone can tell us a lot about an individual’s physiology. We can assume the individual’s airway and breathing are adequate, allowing enough oxygen to reach the lungs and subsequently circulate. The individual is usually sitting up in bed and thus has an adequate blood pressure and blood oxygenation that can maintain cerebral perfusion. They have the cognitive and cerebral processing in place to function the device, and we can examine their cerebellar function by looking for fine-motor movements.

Mobile phone ownership is pervasive within Australia,⁵ with a conservative estimated 85.7% of the population (20.57 million people of a total population of approximately 24 million) owning a mobile phone and an estimated 50% to 79% of mobile phone ownership being of a smartphone.^6,7 This ownership is not just limited to the young, with 74% of Australians over 65 owning or using a mobile phone.⁸ Despite this high phone ownership among those over 65, it is still significantly less than their younger counterparts and may be one reason for the absence of spP in those older than 51 years. A key point in the study is that overall phone ownership was not known, and, thus, it is not possible to determine the proportion of spN patients who were negative because they did not own a phone. However, based on general population data, the incidence of spP patients was well below that seen in the community (11.3%)⁵ and even when excluding those over 55, the percentage of spP patients only rose to 19.1%. Unsurprisingly, increasing age was associated with a decreased likelihood of being spP (P < 0.0005), as younger people are more likely to own a phone.⁸ There was no association with gender (P = 0.18). There are a number of explanations that may explain the lower than expected percentage of spP patients, including the inability for the patient to gather their possessions during a medical emergency, patients storing their phones prior to doctor review (72%-85% of Australians report talking on phones in public places to be rude or intrusive⁵), but more importantly, that our hypothesis that patients were too unwell to use their device appears to hold true.

There are potential alternate reasons other than smartphone status that may account for patients being discharged home on day 1. While there was no association seen with gender, the need for an operation prolonged a patient’s stay (OR 1.64; 95% CI, 0.046-0.46), and there was a trend seen with increasing age (OR 0.98; 95% CI, 0.96-1.00). Neither of these 2 demographics are unsurprising: increasing age is associated with increasing medical comorbidities and thus complexity; even the simplest of operations require a postprocedure observation period, automatically increasing their LOS. Additionally, measured demographics are limited and there may be further unmeasured reasons that account for earlier discharge.

The other key component to this study is the value of the physical examination, albeit only assessing 1 component: the general inspection. In their review of the value of the physical examination of the cardiovascular system, Elder et al. highlight an important point: in traditional teaching, the value of a physical sign is compared with a diagnostic reference, typically imaging or an invasive test.⁹ They argue that this definition undervalues the physical examination and list other values aside from accuracy including accessibility, contribution to clinical care beyond diagnoses, cost effectiveness, patients’ safety, patients’ perceptions, and pedagogic value; and they argue that the physical examination should always be considered in regard to the clinical context—in this case, the newly admitted general surgical patient.

The assessment of the presence or absence of a smartphone is readily performed upon general inspection and is easily visible; general inspection of the patient and failure to observe the clinical sign when present are 2 of the greatest errors associated with physical examination.¹⁰ Furthermore, given its unique status as a physical sign, the authors’ opinion and experience is that it is readily teachable. McGee states, “…a fundamental lesson [in regards to teaching] is that the diagnosis of many clinical problems, despite modern testing, still depends primarily on what the clinician sees, hears, and feels.”¹¹ In their article, Paley et al. found that more than 80% of patients admitted from the ED under internal medicine could be accurately diagnosed based largely on history and examination alone and concluded that basic clinical skills are sufficient for achieving an accurate diagnosis in most cases.¹² Although Paley et al. were assisted with basic tests (such as electrocardiogram and basic haematological and biochemistry results), the point of clinical skills is not lost. Furthermore, this assessment was made in a group of patients generally considered to be complex in contrast to the “standard” appendicitis or cholecystitis patient that makes up a significant proportion of general surgical patients.

There are a number of limitations to this study, however, including smartphones that may have been missed during the observational period. Potential confounding variables such as socioeconomic status and the overall smartphone ownership of our subjects were not known. We did not ask all admitted patients whether they owned a phone or whether they had a phone in their possession. Knowledge of those who owned phones but were not in possession of them could strengthen our argument that spN patients were not using their phone because they were unwell, rather than just not having access to it.

However, this study has a number of strengths, including a large sample size and data that were prospectively collected by a method and in a setting that was the same for all participants. Clear and appropriate definitions were used, which minimizes misclassification bias. Participants and decision makers were blinded to the study, and potentially confounding variables such as age and sex were accounted for.

Assessing the suitability for discharge from the hospital is a decision encountered by hospital-based clinicians every day. These skills are not taught, but are rather learned as a junior doctor acquires experience. It is unlikely that protocols will be developed to aid identification of potential discharges from an acute surgical ward; acute surgical conditions are too varied and dynamic to be able to pool all data. We continue to rely on our own and fellow colleagues’ (doctors, nurses, and other staff) input and assessment. However, our study has shown that it is possible to identify and quantify clinical findings that are already regularly used, albeit potentially subconsciously, to assess suitability for discharge. We have shown in this large, prospectively collected observational study that if a surgical patient is seen using their electronic device, they are more likely to be safe to go home. Thus, surgeons can reliably use this observation as a trigger to consider discharging the patient following a more thorough assessment.

While these observations might appear to be rather a simplistic way of trying to quantify whether or not a patient is fit for discharge, any clues that hint towards a patient’s well-being should be taken into account when making an overall assessment. The active use of a smartphone is one such measure.

Acknowledgments

The authors thank Emma Knight and Nancy Briggs from the Data Management & Analysis Centre, Discipline of Public Health, University of Adelaide.

Disclosure

No author nor the institution received any payment or services from a third party for any aspect of the submitted work and report no conflict of interest. There are no reported financial relationships with any entities by any of the authors. There are no patents pending based upon this publication. There are no relationships or activities that readers could perceive to have influenced, or give the appearance of influencing, the submitted work. The corresponding author is not in receipt of a research scholarship. The paper is not based on a previous communication.

Annual healthcare costs in the United States are over $3 trillion and are garnering significant national attention.¹ The United States spends approximately 2.5 times more per capita on healthcare when compared to other developed nations.² One source of unnecessary cost in healthcare is defensive medicine. Defensive medicine has been defined by Congress as occurring “when doctors order tests, procedures, or visits, or avoid certain high-risk patients or procedures, primarily (but not necessarily) because of concern about malpractice liability.”³

Though difficult to assess, in 1 study, defensive medicine was estimated to cost $45 billion annually.⁴ While general agreement exists that physicians practice defensive medicine, the extent of defensive practices and the subsequent impact on healthcare costs remain unclear. This is especially true for a group of clinicians that is rapidly increasing in number: hospitalists. Currently, there are more than 50,000 hospitalists in the United States,⁵ yet the prevalence of defensive medicine in this relatively new specialty is unknown. Inpatient care is complex and time constraints can impede establishing an optimal therapeutic relationship with the patient, potentially raising liability fears. We therefore sought to quantify hospitalist physician estimates of the cost of defensive medicine and assess correlates of their estimates. As being sued might spur defensive behaviors, we also assessed how many hospitalists reported being sued and whether this was associated with their estimates of defensive medicine.

Survey Questionnaire

In a previously published survey-based analysis, we reported on physician practice and overuse for 2 common scenarios in hospital medicine: preoperative evaluation and management of uncomplicated syncope.⁶ After responding to the vignettes, each physician was asked to provide demographic and employment information and malpractice history. In addition, they were asked the following: In your best estimation, what percentage of healthcare-related resources (eg, hospital admissions, diagnostic testing, treatment) are spent purely because of defensive medicine concerns? __________% resources

Survey Sample & Administration

The survey was sent to a sample of 1753 hospitalists, randomly identified through the Society of Hospital Medicine’s (SHM) database of members and annual meeting attendees. It is estimated that almost 30% of practicing hospitalists in the United States are members of the SHM.⁵ A full description of the sampling methodology was previously published.⁶ Selected hospitalists were mailed surveys, a $20 financial incentive, and subsequent reminders between June and October 2011.

The study was exempted from institutional review board review by the University of Michigan and the VA Ann Arbor Healthcare System.

Variables

The primary outcome of interest was the response to the “% resources” estimated to be spent on defensive medicine. This was analyzed as a continuous variable. Independent variables included the following: VA employment, malpractice insurance payer, employer, history of malpractice lawsuit, sex, race, and years practicing as a physician.

Statistical Analysis

Analyses were conducted using SAS, version 9.4 (SAS Institute). Descriptive statistics were first calculated for all variables. Next, bivariable comparisons between the outcome variables and other variables of interest were performed. Multivariable comparisons were made using linear regression for the outcome of estimated resources spent on defensive medicine. A P value of < 0.05 was considered statistically significant.

Of the 1753 surveys mailed, 253 were excluded due to incorrect addresses or because the recipients were not practicing hospitalists. A total of 1020 were completed and returned, yielding a 68% response rate (1020 out of 1500 eligible). The hospitalist respondents were in practice for an average of 11 years (range 1-40 years). Respondents represented all 50 states and had a diverse background of experience and demographic characteristics, which has been previously described.⁶

Resources Estimated Spent on Defensive Medicine

Hospitalists reported, on average, that they believed defensive medicine accounted for 37.5% (standard deviation, 20.2%) of all healthcare spending. Results from the multivariable regression are presented in the Table. Hospitalists affiliated with a VA hospital reported 5.5% less in resources spent on defensive medicine than those not affiliated with a VA hospital (32.2% VA vs 37.7% non-VA, P = 0.025). For every 10 years in practice, the estimate of resources spent on defensive medicine decreased by 3% (P = 0.003). Those who were male (36.4% male vs 39.4% female, P = 0.023) and non-Hispanic white (32.5% non-Hispanic white vs 44.7% other, P ≤ 0.001) also estimated less resources spent on defensive medicine. We did not find an association between a hospitalist reporting being sued and their perception of resources spent on defensive medicine.  

Risk of Being Sued

Over a quarter of our sample (25.6%) reported having been sued at least once for medical malpractice. The proportion of hospitalists that reported a history of being sued generally increased with more years of practice (Figure). For those who had been in practice for at least 20 years, more than half (55%) had been sued at least once during their career.

In a national survey, hospitalists estimated that almost 40% of all healthcare-related resources are spent purely because of defensive medicine concerns. This estimate was affected by personal demographic and employment factors. Our second major finding is that over one-quarter of a large random sample of hospitalist physicians reported being sued for malpractice.

Hospitalist perceptions of defensive medicine varied significantly based on employment at a VA hospital, with VA-affiliated hospitalists reporting less estimated spending on defensive medicine. This effect may reflect a less litigious environment within the VA, even though physicians practicing within the VA can be reported to the National Practitioner Data Bank.⁷ The different environment may be due to the VA’s patient mix (VA patients tend to be poorer, older, sicker, and have more mental illness)⁸; however, it could also be due to its de facto practice of a form of enterprise liability, in which, by law, the VA assumes responsibility for negligence, sheltering its physicians from direct liability.

We also found that the higher the number of years a hospitalist reported practicing, the lower the perception of resources being spent on defensive medicine. The reason for this finding is unclear. There has been a recent focus on high-value care and overspending, and perhaps younger hospitalists are more aware of these initiatives and thus have higher estimates. Additionally, non-Hispanic white male respondents estimated a lower amount spent on defensive medicine compared with other respondents. This is consistent with previous studies of risk perception which have noted a “white male effect” in which white males generally perceive a wide range of risks to be lower than female and non-white individuals, likely due to sociopolitical factors.⁹ Here, the white male effect is particularly interesting, considering that male physicians are almost 2.5 times as likely as female physicians to report being sued.¹⁰

Similar to prior studies,¹¹ there was no association with personal liability claim experience and perceived resources spent on defensive medicine. It is unclear why personal experience of being sued does not appear to be associated with perceptions of defensive medicine practice. It is possible that the fear of being sued is worse than the actual experience or that physicians believe that lawsuits are either random events or inevitable and, as a result, do not change their practice patterns.

The lifetime risk of being named in a malpractice suit is substantial for hospitalists: in our study, over half of hospitalists in practice for 20 years or more reported they had been sued. This corresponds with the projection made by Jena and colleagues,¹² which estimated that 55% of internal medicine physicians will be sued by the age of 45, a number just slightly higher than the average for all physicians.

Our study has important limitations. Our sample was of hospitalists and therefore may not be reflective of other medical specialties. Second, due to the nature of the study design, the responses to spending on defensive medicine may not represent actual practice. Third, we did not confirm details such as place of employment or history of lawsuit, and this may be subject to recall bias. However, physicians are unlikely to forget having been sued. Finally, this survey is observational and cross-sectional. Our data imply association rather than causation. Without longitudinal data, it is impossible to know if years of practice correlate with perceived defensive medicine spending due to a generational effect or a longitudinal effect (such as more confidence in diagnostic skills with more years of practice).

Despite these limitations, our survey has important policy implications. First, we found that defensive medicine is perceived by hospitalists to be costly. Although physicians likely overestimated the cost (37.5%, or an estimated $1 trillion is far higher than previous estimates of approximately 2% of all healthcare spending),⁴ it also demonstrates the extent to which physicians feel as though the medical care that is provided may be unnecessary. Second, at least a quarter of hospitalist physicians have been sued, and the risk of being named as a defendant in a lawsuit increases the longer they have been in clinical practice.

Given these findings, policies aimed to reduce the practice of defensive medicine may help the rising costs of healthcare. Reducing defensive medicine requires decreasing physician fears of liability and related reporting. Traditional tort reforms (with the exception of damage caps) have not been proven to do this. And damage caps can be inequitable, hard to pass, and even found to be unconstitutional in some states.¹³ However, other reform options hold promise in reducing liability fears, including enterprise liability, safe harbor legislation, and health courts.¹³ Finally, shared decision-making models may also provide a method to reduce defensive fears as well.⁶

Acknowledgments

The authors thank the Society of Hospital Medicine, Dr. Scott Flanders, Andrew Hickner, and David Ratz for their assistance with this project.

Disclosure

The authors received financial support from the Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs Health Services Research and Development Center for Clinical Management Research, the University of Michigan Specialist-Hospitalist Allied Research Program, and the Ann Arbor University of Michigan VA Patient Safety Enhancement Program.

Disclaimer

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs, or the Society of Hospital Medicine.

Annual healthcare costs in the United States are over $3 trillion and are garnering significant national attention.¹ The United States spends approximately 2.5 times more per capita on healthcare when compared to other developed nations.² One source of unnecessary cost in healthcare is defensive medicine. Defensive medicine has been defined by Congress as occurring “when doctors order tests, procedures, or visits, or avoid certain high-risk patients or procedures, primarily (but not necessarily) because of concern about malpractice liability.”³

Though difficult to assess, in 1 study, defensive medicine was estimated to cost $45 billion annually.⁴ While general agreement exists that physicians practice defensive medicine, the extent of defensive practices and the subsequent impact on healthcare costs remain unclear. This is especially true for a group of clinicians that is rapidly increasing in number: hospitalists. Currently, there are more than 50,000 hospitalists in the United States,⁵ yet the prevalence of defensive medicine in this relatively new specialty is unknown. Inpatient care is complex and time constraints can impede establishing an optimal therapeutic relationship with the patient, potentially raising liability fears. We therefore sought to quantify hospitalist physician estimates of the cost of defensive medicine and assess correlates of their estimates. As being sued might spur defensive behaviors, we also assessed how many hospitalists reported being sued and whether this was associated with their estimates of defensive medicine.

Survey Questionnaire

In a previously published survey-based analysis, we reported on physician practice and overuse for 2 common scenarios in hospital medicine: preoperative evaluation and management of uncomplicated syncope.⁶ After responding to the vignettes, each physician was asked to provide demographic and employment information and malpractice history. In addition, they were asked the following: In your best estimation, what percentage of healthcare-related resources (eg, hospital admissions, diagnostic testing, treatment) are spent purely because of defensive medicine concerns? __________% resources

Survey Sample & Administration

The survey was sent to a sample of 1753 hospitalists, randomly identified through the Society of Hospital Medicine’s (SHM) database of members and annual meeting attendees. It is estimated that almost 30% of practicing hospitalists in the United States are members of the SHM.⁵ A full description of the sampling methodology was previously published.⁶ Selected hospitalists were mailed surveys, a $20 financial incentive, and subsequent reminders between June and October 2011.

The study was exempted from institutional review board review by the University of Michigan and the VA Ann Arbor Healthcare System.

Variables

The primary outcome of interest was the response to the “% resources” estimated to be spent on defensive medicine. This was analyzed as a continuous variable. Independent variables included the following: VA employment, malpractice insurance payer, employer, history of malpractice lawsuit, sex, race, and years practicing as a physician.

Statistical Analysis

Analyses were conducted using SAS, version 9.4 (SAS Institute). Descriptive statistics were first calculated for all variables. Next, bivariable comparisons between the outcome variables and other variables of interest were performed. Multivariable comparisons were made using linear regression for the outcome of estimated resources spent on defensive medicine. A P value of < 0.05 was considered statistically significant.

Of the 1753 surveys mailed, 253 were excluded due to incorrect addresses or because the recipients were not practicing hospitalists. A total of 1020 were completed and returned, yielding a 68% response rate (1020 out of 1500 eligible). The hospitalist respondents were in practice for an average of 11 years (range 1-40 years). Respondents represented all 50 states and had a diverse background of experience and demographic characteristics, which has been previously described.⁶

Resources Estimated Spent on Defensive Medicine

Hospitalists reported, on average, that they believed defensive medicine accounted for 37.5% (standard deviation, 20.2%) of all healthcare spending. Results from the multivariable regression are presented in the Table. Hospitalists affiliated with a VA hospital reported 5.5% less in resources spent on defensive medicine than those not affiliated with a VA hospital (32.2% VA vs 37.7% non-VA, P = 0.025). For every 10 years in practice, the estimate of resources spent on defensive medicine decreased by 3% (P = 0.003). Those who were male (36.4% male vs 39.4% female, P = 0.023) and non-Hispanic white (32.5% non-Hispanic white vs 44.7% other, P ≤ 0.001) also estimated less resources spent on defensive medicine. We did not find an association between a hospitalist reporting being sued and their perception of resources spent on defensive medicine.  

Risk of Being Sued

Over a quarter of our sample (25.6%) reported having been sued at least once for medical malpractice. The proportion of hospitalists that reported a history of being sued generally increased with more years of practice (Figure). For those who had been in practice for at least 20 years, more than half (55%) had been sued at least once during their career.

In a national survey, hospitalists estimated that almost 40% of all healthcare-related resources are spent purely because of defensive medicine concerns. This estimate was affected by personal demographic and employment factors. Our second major finding is that over one-quarter of a large random sample of hospitalist physicians reported being sued for malpractice.

Hospitalist perceptions of defensive medicine varied significantly based on employment at a VA hospital, with VA-affiliated hospitalists reporting less estimated spending on defensive medicine. This effect may reflect a less litigious environment within the VA, even though physicians practicing within the VA can be reported to the National Practitioner Data Bank.⁷ The different environment may be due to the VA’s patient mix (VA patients tend to be poorer, older, sicker, and have more mental illness)⁸; however, it could also be due to its de facto practice of a form of enterprise liability, in which, by law, the VA assumes responsibility for negligence, sheltering its physicians from direct liability.

We also found that the higher the number of years a hospitalist reported practicing, the lower the perception of resources being spent on defensive medicine. The reason for this finding is unclear. There has been a recent focus on high-value care and overspending, and perhaps younger hospitalists are more aware of these initiatives and thus have higher estimates. Additionally, non-Hispanic white male respondents estimated a lower amount spent on defensive medicine compared with other respondents. This is consistent with previous studies of risk perception which have noted a “white male effect” in which white males generally perceive a wide range of risks to be lower than female and non-white individuals, likely due to sociopolitical factors.⁹ Here, the white male effect is particularly interesting, considering that male physicians are almost 2.5 times as likely as female physicians to report being sued.¹⁰

Similar to prior studies,¹¹ there was no association with personal liability claim experience and perceived resources spent on defensive medicine. It is unclear why personal experience of being sued does not appear to be associated with perceptions of defensive medicine practice. It is possible that the fear of being sued is worse than the actual experience or that physicians believe that lawsuits are either random events or inevitable and, as a result, do not change their practice patterns.

The lifetime risk of being named in a malpractice suit is substantial for hospitalists: in our study, over half of hospitalists in practice for 20 years or more reported they had been sued. This corresponds with the projection made by Jena and colleagues,¹² which estimated that 55% of internal medicine physicians will be sued by the age of 45, a number just slightly higher than the average for all physicians.

Our study has important limitations. Our sample was of hospitalists and therefore may not be reflective of other medical specialties. Second, due to the nature of the study design, the responses to spending on defensive medicine may not represent actual practice. Third, we did not confirm details such as place of employment or history of lawsuit, and this may be subject to recall bias. However, physicians are unlikely to forget having been sued. Finally, this survey is observational and cross-sectional. Our data imply association rather than causation. Without longitudinal data, it is impossible to know if years of practice correlate with perceived defensive medicine spending due to a generational effect or a longitudinal effect (such as more confidence in diagnostic skills with more years of practice).

Despite these limitations, our survey has important policy implications. First, we found that defensive medicine is perceived by hospitalists to be costly. Although physicians likely overestimated the cost (37.5%, or an estimated $1 trillion is far higher than previous estimates of approximately 2% of all healthcare spending),⁴ it also demonstrates the extent to which physicians feel as though the medical care that is provided may be unnecessary. Second, at least a quarter of hospitalist physicians have been sued, and the risk of being named as a defendant in a lawsuit increases the longer they have been in clinical practice.

Given these findings, policies aimed to reduce the practice of defensive medicine may help the rising costs of healthcare. Reducing defensive medicine requires decreasing physician fears of liability and related reporting. Traditional tort reforms (with the exception of damage caps) have not been proven to do this. And damage caps can be inequitable, hard to pass, and even found to be unconstitutional in some states.¹³ However, other reform options hold promise in reducing liability fears, including enterprise liability, safe harbor legislation, and health courts.¹³ Finally, shared decision-making models may also provide a method to reduce defensive fears as well.⁶

Acknowledgments

The authors thank the Society of Hospital Medicine, Dr. Scott Flanders, Andrew Hickner, and David Ratz for their assistance with this project.

Disclosure

The authors received financial support from the Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs Health Services Research and Development Center for Clinical Management Research, the University of Michigan Specialist-Hospitalist Allied Research Program, and the Ann Arbor University of Michigan VA Patient Safety Enhancement Program.

Disclaimer

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs, or the Society of Hospital Medicine.

Annual healthcare costs in the United States are over $3 trillion and are garnering significant national attention.¹ The United States spends approximately 2.5 times more per capita on healthcare when compared to other developed nations.² One source of unnecessary cost in healthcare is defensive medicine. Defensive medicine has been defined by Congress as occurring “when doctors order tests, procedures, or visits, or avoid certain high-risk patients or procedures, primarily (but not necessarily) because of concern about malpractice liability.”³

Though difficult to assess, in 1 study, defensive medicine was estimated to cost $45 billion annually.⁴ While general agreement exists that physicians practice defensive medicine, the extent of defensive practices and the subsequent impact on healthcare costs remain unclear. This is especially true for a group of clinicians that is rapidly increasing in number: hospitalists. Currently, there are more than 50,000 hospitalists in the United States,⁵ yet the prevalence of defensive medicine in this relatively new specialty is unknown. Inpatient care is complex and time constraints can impede establishing an optimal therapeutic relationship with the patient, potentially raising liability fears. We therefore sought to quantify hospitalist physician estimates of the cost of defensive medicine and assess correlates of their estimates. As being sued might spur defensive behaviors, we also assessed how many hospitalists reported being sued and whether this was associated with their estimates of defensive medicine.

Survey Questionnaire

In a previously published survey-based analysis, we reported on physician practice and overuse for 2 common scenarios in hospital medicine: preoperative evaluation and management of uncomplicated syncope.⁶ After responding to the vignettes, each physician was asked to provide demographic and employment information and malpractice history. In addition, they were asked the following: In your best estimation, what percentage of healthcare-related resources (eg, hospital admissions, diagnostic testing, treatment) are spent purely because of defensive medicine concerns? __________% resources

Survey Sample & Administration

The survey was sent to a sample of 1753 hospitalists, randomly identified through the Society of Hospital Medicine’s (SHM) database of members and annual meeting attendees. It is estimated that almost 30% of practicing hospitalists in the United States are members of the SHM.⁵ A full description of the sampling methodology was previously published.⁶ Selected hospitalists were mailed surveys, a $20 financial incentive, and subsequent reminders between June and October 2011.

The study was exempted from institutional review board review by the University of Michigan and the VA Ann Arbor Healthcare System.

Variables

The primary outcome of interest was the response to the “% resources” estimated to be spent on defensive medicine. This was analyzed as a continuous variable. Independent variables included the following: VA employment, malpractice insurance payer, employer, history of malpractice lawsuit, sex, race, and years practicing as a physician.

Statistical Analysis

Analyses were conducted using SAS, version 9.4 (SAS Institute). Descriptive statistics were first calculated for all variables. Next, bivariable comparisons between the outcome variables and other variables of interest were performed. Multivariable comparisons were made using linear regression for the outcome of estimated resources spent on defensive medicine. A P value of < 0.05 was considered statistically significant.

Of the 1753 surveys mailed, 253 were excluded due to incorrect addresses or because the recipients were not practicing hospitalists. A total of 1020 were completed and returned, yielding a 68% response rate (1020 out of 1500 eligible). The hospitalist respondents were in practice for an average of 11 years (range 1-40 years). Respondents represented all 50 states and had a diverse background of experience and demographic characteristics, which has been previously described.⁶

Resources Estimated Spent on Defensive Medicine

Hospitalists reported, on average, that they believed defensive medicine accounted for 37.5% (standard deviation, 20.2%) of all healthcare spending. Results from the multivariable regression are presented in the Table. Hospitalists affiliated with a VA hospital reported 5.5% less in resources spent on defensive medicine than those not affiliated with a VA hospital (32.2% VA vs 37.7% non-VA, P = 0.025). For every 10 years in practice, the estimate of resources spent on defensive medicine decreased by 3% (P = 0.003). Those who were male (36.4% male vs 39.4% female, P = 0.023) and non-Hispanic white (32.5% non-Hispanic white vs 44.7% other, P ≤ 0.001) also estimated less resources spent on defensive medicine. We did not find an association between a hospitalist reporting being sued and their perception of resources spent on defensive medicine.  

Risk of Being Sued

Over a quarter of our sample (25.6%) reported having been sued at least once for medical malpractice. The proportion of hospitalists that reported a history of being sued generally increased with more years of practice (Figure). For those who had been in practice for at least 20 years, more than half (55%) had been sued at least once during their career.

In a national survey, hospitalists estimated that almost 40% of all healthcare-related resources are spent purely because of defensive medicine concerns. This estimate was affected by personal demographic and employment factors. Our second major finding is that over one-quarter of a large random sample of hospitalist physicians reported being sued for malpractice.

Hospitalist perceptions of defensive medicine varied significantly based on employment at a VA hospital, with VA-affiliated hospitalists reporting less estimated spending on defensive medicine. This effect may reflect a less litigious environment within the VA, even though physicians practicing within the VA can be reported to the National Practitioner Data Bank.⁷ The different environment may be due to the VA’s patient mix (VA patients tend to be poorer, older, sicker, and have more mental illness)⁸; however, it could also be due to its de facto practice of a form of enterprise liability, in which, by law, the VA assumes responsibility for negligence, sheltering its physicians from direct liability.

We also found that the higher the number of years a hospitalist reported practicing, the lower the perception of resources being spent on defensive medicine. The reason for this finding is unclear. There has been a recent focus on high-value care and overspending, and perhaps younger hospitalists are more aware of these initiatives and thus have higher estimates. Additionally, non-Hispanic white male respondents estimated a lower amount spent on defensive medicine compared with other respondents. This is consistent with previous studies of risk perception which have noted a “white male effect” in which white males generally perceive a wide range of risks to be lower than female and non-white individuals, likely due to sociopolitical factors.⁹ Here, the white male effect is particularly interesting, considering that male physicians are almost 2.5 times as likely as female physicians to report being sued.¹⁰

Similar to prior studies,¹¹ there was no association with personal liability claim experience and perceived resources spent on defensive medicine. It is unclear why personal experience of being sued does not appear to be associated with perceptions of defensive medicine practice. It is possible that the fear of being sued is worse than the actual experience or that physicians believe that lawsuits are either random events or inevitable and, as a result, do not change their practice patterns.

The lifetime risk of being named in a malpractice suit is substantial for hospitalists: in our study, over half of hospitalists in practice for 20 years or more reported they had been sued. This corresponds with the projection made by Jena and colleagues,¹² which estimated that 55% of internal medicine physicians will be sued by the age of 45, a number just slightly higher than the average for all physicians.

Our study has important limitations. Our sample was of hospitalists and therefore may not be reflective of other medical specialties. Second, due to the nature of the study design, the responses to spending on defensive medicine may not represent actual practice. Third, we did not confirm details such as place of employment or history of lawsuit, and this may be subject to recall bias. However, physicians are unlikely to forget having been sued. Finally, this survey is observational and cross-sectional. Our data imply association rather than causation. Without longitudinal data, it is impossible to know if years of practice correlate with perceived defensive medicine spending due to a generational effect or a longitudinal effect (such as more confidence in diagnostic skills with more years of practice).

Despite these limitations, our survey has important policy implications. First, we found that defensive medicine is perceived by hospitalists to be costly. Although physicians likely overestimated the cost (37.5%, or an estimated $1 trillion is far higher than previous estimates of approximately 2% of all healthcare spending),⁴ it also demonstrates the extent to which physicians feel as though the medical care that is provided may be unnecessary. Second, at least a quarter of hospitalist physicians have been sued, and the risk of being named as a defendant in a lawsuit increases the longer they have been in clinical practice.

Given these findings, policies aimed to reduce the practice of defensive medicine may help the rising costs of healthcare. Reducing defensive medicine requires decreasing physician fears of liability and related reporting. Traditional tort reforms (with the exception of damage caps) have not been proven to do this. And damage caps can be inequitable, hard to pass, and even found to be unconstitutional in some states.¹³ However, other reform options hold promise in reducing liability fears, including enterprise liability, safe harbor legislation, and health courts.¹³ Finally, shared decision-making models may also provide a method to reduce defensive fears as well.⁶

Acknowledgments

The authors thank the Society of Hospital Medicine, Dr. Scott Flanders, Andrew Hickner, and David Ratz for their assistance with this project.

Disclosure

The authors received financial support from the Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs Health Services Research and Development Center for Clinical Management Research, the University of Michigan Specialist-Hospitalist Allied Research Program, and the Ann Arbor University of Michigan VA Patient Safety Enhancement Program.

Disclaimer

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of Blue Cross Blue Shield of Michigan Foundation, the Department of Veterans Affairs, or the Society of Hospital Medicine.

The belief that corneas that have been preserved for > 7 days are not viable for transplantation is not based on evidence, says Jonathan Lass, MD. In fact, he led a study that found corneas can be preserved safely for 11 days without negative impact on the success of transplantation. In the Cornea Preservation Time Study, funded by the National Eye Institute, Lass and other researchers looked at 3-year graft success rates among 1,090 participants (1,330 eyes) who underwent transplantation via Descemet’s stripping automated endothelial keratoplasty by 70 surgeons at 40 surgical sites. Most of the patients had Fuchs’ endothelial corneal dystrophy, a progressive disease.

The researchers were “unable to conclude” that the success rates were the same for corneas preserved for 8 to 14 days, versus up to 7 days (92% vs 95%). However, they found that much of the difference between the groups was accounted for by patients receiving corneas preserved for 12 to 14 days.

In a separate analysis, the researchers looked to see if differences in corneal preservation time affected endothelial cell loss after 3 years. They found that corneas preserved for up to 7 days had a 37% loss of cells versus 40% in those preserved for 8 to 14 days. A closer look at the data showed that the effect of corneal preservation time on the loss of endothelial cells was comparable from 4 to 13 days.

Dr. Lass emphasizes that while patients who received the older corneas had lower success rates, even those success rates were “impressively high” at 89%.

Donor corneas are not in short supply in the U.S. Outside the U.S., however, corneal disease is the third leading cause of blindness and corneal donor tissue is scarce.

The belief that corneas that have been preserved for > 7 days are not viable for transplantation is not based on evidence, says Jonathan Lass, MD. In fact, he led a study that found corneas can be preserved safely for 11 days without negative impact on the success of transplantation. In the Cornea Preservation Time Study, funded by the National Eye Institute, Lass and other researchers looked at 3-year graft success rates among 1,090 participants (1,330 eyes) who underwent transplantation via Descemet’s stripping automated endothelial keratoplasty by 70 surgeons at 40 surgical sites. Most of the patients had Fuchs’ endothelial corneal dystrophy, a progressive disease.

The researchers were “unable to conclude” that the success rates were the same for corneas preserved for 8 to 14 days, versus up to 7 days (92% vs 95%). However, they found that much of the difference between the groups was accounted for by patients receiving corneas preserved for 12 to 14 days.

In a separate analysis, the researchers looked to see if differences in corneal preservation time affected endothelial cell loss after 3 years. They found that corneas preserved for up to 7 days had a 37% loss of cells versus 40% in those preserved for 8 to 14 days. A closer look at the data showed that the effect of corneal preservation time on the loss of endothelial cells was comparable from 4 to 13 days.

Dr. Lass emphasizes that while patients who received the older corneas had lower success rates, even those success rates were “impressively high” at 89%.

Donor corneas are not in short supply in the U.S. Outside the U.S., however, corneal disease is the third leading cause of blindness and corneal donor tissue is scarce.

The belief that corneas that have been preserved for > 7 days are not viable for transplantation is not based on evidence, says Jonathan Lass, MD. In fact, he led a study that found corneas can be preserved safely for 11 days without negative impact on the success of transplantation. In the Cornea Preservation Time Study, funded by the National Eye Institute, Lass and other researchers looked at 3-year graft success rates among 1,090 participants (1,330 eyes) who underwent transplantation via Descemet’s stripping automated endothelial keratoplasty by 70 surgeons at 40 surgical sites. Most of the patients had Fuchs’ endothelial corneal dystrophy, a progressive disease.

The researchers were “unable to conclude” that the success rates were the same for corneas preserved for 8 to 14 days, versus up to 7 days (92% vs 95%). However, they found that much of the difference between the groups was accounted for by patients receiving corneas preserved for 12 to 14 days.

In a separate analysis, the researchers looked to see if differences in corneal preservation time affected endothelial cell loss after 3 years. They found that corneas preserved for up to 7 days had a 37% loss of cells versus 40% in those preserved for 8 to 14 days. A closer look at the data showed that the effect of corneal preservation time on the loss of endothelial cells was comparable from 4 to 13 days.

Dr. Lass emphasizes that while patients who received the older corneas had lower success rates, even those success rates were “impressively high” at 89%.

Donor corneas are not in short supply in the U.S. Outside the U.S., however, corneal disease is the third leading cause of blindness and corneal donor tissue is scarce.

The aim of palliative care (PC) is to improve quality of life for patients facing serious, life-threatening illness and their families.¹ Due to insufficient numbers of PC specialists to meet the PC needs for every hospitalized patient,² all hospitalists should maintain basic PC skills as recognized by PC being a core competency for hospitalists.^3,4

We summarize and critique PC research articles published between January 1, 2016, and December 31, 2016, that have a high likelihood of impacting the practice of hospital medicine. We hand searched 15 journals and conducted a MEDLINE keyword search of PC terms (see Table). All titles and/or abstracts were reviewed and selected for full review based on the following factors: palliative medicine content, scientific rigor, impact on practice, and relevance to hospital medicine. Fifty-five articles were individually reviewed and scored by all authors according to rigor, impact, and relevance. Articles were ranked according to their mean scores, and 9 articles were chosen for inclusion through consensus discussion.

Antipsychotics Were Inferior to a Placebo in Treating Nonterminal Delirium

Agar MR, Lawlor PG, Quinn S, et al. Efficacy of oral risperidone, haloperidol, or placebo for symptoms of delirium among patients in palliative care: a randomized clinical trial. JAMA Intern Med. 2017;177(1):34-42.

Background

Delirium is highly prevalent in PC and is associated with significant distress.⁵ Antipsychotics are widely used for symptoms of delirium, although current evidence does not support this practice in hospitalized adults.^6,7

Findings

This was a double-blind, parallel-arm, placebo randomized controlled trial (RCT) of 247 patients with delirium with an estimated life expectancy of ≥7 days in 11 PC or hospice units across Australia. Patients were randomized to receive risperidone, haloperidol, or a placebo in addition to nonpharmacological management of delirium. Delirium symptom scores after 3 days of treatment, the use of midazolam as a rescue medication, and the presence of extrapyramidal symptoms (EPS) were measured. The risperidone and haloperidol arms had significantly higher delirium symptom scores (P = .02 and P = .009, respectively), mean EPS symptoms (P < .001), and more use of rescue midazolam than the placebo arm. Mortality was higher for antipsychotics, with a hazard ratio of 1.73 for haloperidol (P = .003), 1.29 for risperidone (P = .14), and 1.47 for any antipsychotic (P = .01).

Cautions

The study population was elderly (mean age >70 years) with mild delirium scores. The use of antipsychotics was associated with more benzodiazepine use, which could itself worsen delirium. As patients with clinician-predicted life expectancy of <7 days were excluded, findings cannot be extrapolated to the treatment of terminal delirium, which can often be more symptomatic and difficult to treat.

Implications

Avoid scheduled antipsychotics in patients with nonterminal delirium, as they can increase risk of harm without advantages, over nonpharmacologic interventions.

Low-Dose Morphine Was Superior to Weak Opioids in the Treatment of Moderate Cancer Pain

Bandieri E, Romero M, Ripamonti CI, et al. Randomized trial of low-dose morphine versus weak opioids in moderate cancer pain. J Clin Oncol. 2016;34(5):436-442.

Background

The World Health Organization guidelines recommend the use of weak opioids (WOs), such as codeine or tramadol, as a sequential step in the management of cancer pain.⁸ This strategy has not been tested against low doses of stronger opioids.

Findings

In this multicenter, open-label RCT, 240 patients in Italy were randomized and stratified by age (<75 years or ≥75 years) to either the WO group or low-dose morphine (M) group. The primary outcome measure was a reduction in pain intensity by 20% or more. Secondary outcomes included an improvement in symptom scores, a ≥30% and ≥50% reduction in pain, increased opioid dosage, and adverse side effects. Compared with the WO group, the M group had more patients with a 20% reduction in pain (88.2% vs 54.7%; P < .001), more evidence of pain control in the first week (80.9% vs 43.6%; P < .001), more patients with a ≥30% and ≥50% reduction in pain, and less need to switch to a stronger opioid (15.5% vs 35.0%; P = .001) or require dose increases. Adverse effects were similar in both groups.

Cautions

Patients with chronic kidney disease (CKD) were excluded due to concerns about the accumulation of morphine metabolites. Additionally, this study was open label, increasing the risk of bias.

Implications

Low-dose morphine should be considered over the use of WOs to achieve better and more rapid pain control in patients without CKD.

The Use of Methadone as a Coanalgesic May Improve Moderate Cancer Pain

Courtemanche F, Dao D, Gagné F, et al. Methadone as a coanalgesic for palliative care cancer patients. J Palliat Med. 2016;19(9):972-978.

Background

Methadone is effective at treating cancer pain and is often utilized when patients have neuropathic pain, fail to respond to traditional opioids, or have renal failure.^9,10 However, its long half-life and many drug interactions make methadone challenging to use.

Findings

This cohort study looked at 153 inpatient or outpatient PC patients in Montreal who received methadone as a coanalgesic for cancer pain. The patients’ median morphine equivalent dose was 120 mg when initiating methadone. The median starting dose of methadone was 3 mg per day. Of patients, 49.3% had a significant response (≥30% pain reduction), with a median response time of 7 days, and 30.1% achieved a substantial response (≥50% pain reduction), with a median response time of 3 days. Patients with higher initial pain scores were more likely to respond to adjuvant methadone. Those who had not responded after a week of methadone were unlikely to respond despite dose escalations. Adverse effects included drowsiness (51.4%), confusion (27.4%), constipation (24.7%), nausea (19.9%), and myoclonia (16.4%).

Cautions

This was an observational study with retrospective data, leading to higher levels of missing data. A high rate of adverse side effects was reported (90.4%). Further study is needed to validate and reproduce the findings.

Implications

The use of adjuvant low-dose methadone may be considered in patients with moderate pain despite high-dose opioids. If a response is not seen within 7 days, then methadone use should be reconsidered.

Many Hospitalized Patients on Comfort Care Still Receive Antimicrobials

Merel SE, Meier CA, McKinney CM, Pottinger PS. Antimicrobial use in patients on a comfort care protocol: a retrospective cohort study. J Palliat Med. 2016;19(11):1210-1214.

Background

It is unknown how often patients who are hospitalized at the end of life continue to receive antimicrobials and what factors are associated with antimicrobial use.

Findings

This retrospective cohort study of 1881 hospitalized adults transitioned to a comfort care order (CCO) set at 2 academic medical centers found that 77% of these patients received antimicrobials during their hospital stay (62.4% at 24 hours prior to CCO). Of the 711 still alive at ≥24 hours after CCO, 111 (15.6%) were still on antimicrobials, with that proportion remaining stable for the remainder of hospitalization. In comparing those who did and did not receive antimicrobials after 24 hours of CCO, the presence of a documented infection was not significantly different after adjusting for age. Those with a cancer diagnosis (adjusted risk ratio [ARR] = 1.44: P = .04), a longer length of stay (≥7 days vs <7 days; ARR = 1.49; P = .05), and those discharged home (ARR 2.93; P < .001) or to a facility (ARR 3.63; P < .001) versus dying in the hospital were more likely to be on antimicrobials 24 hours after CCO. Compared with those on a medicine service, patients in the medical and surgical intensive care units (ICUs) were less likely to receive antimicrobials (medical ICU ARR = 0.32; P = .01; surgical ICU and/or neuro-ICU ARR = 0.32; P = .02). The most commonly administered antimicrobials were fluoroquinolones and vancomycin.

Cautions

Only 111 patients were still on antimicrobials at 24 hours, which limited analysis. Investigators relied on retrospective data for medication administration and diagnoses.

Implications

Further work is needed to understand and address the expectations of clinicians, patients, and families regarding the role of antimicrobials at the end of life.

Video Decision Aids Improved Rates of Advance Care Planning and Hospice Use and Decreased Costs

Volandes, AE, Paasche-Orlow MK, Davis AD et al. Use of video decision aids to promote advance care planning in Hilo, Hawai‘i. J Gen Intern Med. 2016;31(9):1035-1040.

Background

Advance care planning (ACP) can be enhanced with the use of video decision aids, which may help address scalability and cost.¹¹ The Hawaii Medical Service Association began an initiative to improve ACP rates, which included a financial incentive. Clinician training and patient access to ACP videos were implemented 1 year into this campaign, which was intended for patients with late-stage disease.

Findings

This study tested the impact of the video intervention on the rates of ACP documentation in Hilo, Hawaii, along with secondary outcomes of hospice use, hospital deaths, and costs. The intervention was sequentially rolled out to Hilo Medical Center (HMC), followed by hospice and primary care practices. Following the video introduction, the proportion of patients discharged from HMC with ACP documentation markedly increased (3.2% to 39.9%; P < .001). The percentage of hospital patients discharged to hospice increased from 5.7% to 13.8% (P < .001). Overall admissions to the Hospice of Hilo increased at a greater rate than in other parts of Hawaii. After the intervention in Hilo, the in-hospital death rate among patients >65 years old declined slightly (P = .14), while in the rest of the state, the rate remained essentially unchanged. ACP planning did not reduce healthcare costs at the end of life, but costs seemed to increase more slowly in Hilo after the intervention than they did in the rest of Hawaii (P < .05).

Cautions

This report relies on before-and-after comparisons, with potential confounding by a background pay-for-quality initiative; however, the timing of the changes in outcomes correlates well with the introduction of the videos. ACP videos have been studied in other settings, so the intervention is likely generalizable to other states.

Implications

A widespread distribution of ACP videos and training for physicians in their use may lead to significant increases in ACP documentation and other beneficial clinical outcomes for patients and health systems.

A Standardized Palliative Care-Led Intervention Did Not Improve Psychological Outcomes in Families of Patients with Chronic Critical Illness

Carson SS, Cox CE, Wallenstein S, et al. Effect of palliative care-led meetings for families of patients with chronic critical illness: a randomized clinical trial. JAMA. 2016;316(1):51-62.

Background

Chronic critical illness (CCI) occurs when a patient neither recovers nor dies for days to weeks after an acute illness requiring aggressive intensive care. CCI is associated with poor patient and family outcomes.¹² Does a protocol-driven support and information meeting led by PC providers improve these outcomes?

Findings

This multicenter RCT compared 130 CCI patients (184 surrogates) who received a structured intervention to 126 patients (181 surrogates) with usual care. The structured intervention was led by PC clinicians in order to provide supportive conversations and information about CCI and prognosis compared with the usual intensivist communication. The support and information team met with the families of patients in the intervention group after day 7 of mechanical ventilation (MV) and again 10 days later. Both the intervention and control groups received validated information about CCI, and all were eligible for specialty PC consultation, as indicated. The primary outcome of the study was the Hospital Anxiety and Depression Scale (HADS) at 90-day follow-up with the surrogates. Secondary endpoints included posttraumatic stress disorder (PTSD) assessment and other communication measures as well as patient outcomes (hospital mortality, 90-day survival, length of stay, and days of MV). At least 1 meeting took place for 89% of patients (82% of surrogates) in the intervention arm. Fewer patients in the intervention arm had nonstudy PC consultations (13% vs 22%). Ninety-day HADS results were similar in the 2 groups. PTSD symptoms, however, were higher in the intervention group (Impact of Event Scale-Revised score: 25.9 for intervention and 21.3 for control; intergroup difference 4.6 [95% confidence interval, 0.01-9.10]). There were no statistically significant differences among the patient-focused measures, including survival.

Cautions

Although the teams contained skilled clinicians led by PC practitioners, this was not an ordinary PC intervention. The intervention included information and emotional support meetings alone rather than support from a PC team driven by clinical considerations. This study included surrogates of patients with CCI but not other conditions.

Implications

Protocol-driven support and information meetings may not improve, and may slightly worsen, outcomes in families of patients with CCI. This study did not evaluate and should not be applied to clinically indicated, specialty PC consultation in the ICU.

Caregivers of Patients Surviving Prolonged Critical Illness Experience High and Persistent Rates of Depression

Cameron JI, Chu LM, Matte A, et al. One-year outcomes in caregivers of critically ill patients. N Engl J Med. 2016;374(19):1831-1841.

Background

More than half of patients with a CCI require caregiver support 1 year after hospitalization.¹³ Caregivers provide tremendous physical and psychosocial support to their loved ones, but that care is often associated with significant burden.¹⁴

Findings

This prospective parallel cohort study followed caregivers of surviving patients ventilated for at least 7 days from 10 academic hospitals in Canada. The prevalence of depression (Center for Epidemiologic Studies–Depression scale ≥16) in this cohort of 280 caregivers (70% were women) was 67%, 49%, 43%, and 43% at the survey intervals of 7 days, 3 months, 6 months, and 12 months after ICU discharge, respectively. Using latent-class linear mixed models, the investigators identified 2 groups of caregivers: those whose depressive symptoms decreased over time (84%) and those whose depressive symptoms persisted at a high level for the year (16%). Patient characteristics (such as age, comorbidity, sex, and functional status) were not associated with caregiver outcomes. Younger caregiver age, greater effect of patient care on other activities, less social support, less mastery (sense of control), and less personal growth were associated with worse caregiver mental health outcomes.

Cautions

Although this is a high-quality prospective study, causality of caregiving on the high rates of depressive symptoms cannot be confirmed without a control group or knowledge of the caregivers’ mental health status prior to the episode of prolonged critical illness.

Implications

Patient critical illness may have serious impacts on caregiver health and well-being. Hospitalists should be attentive to factors associated with caregiver vulnerability and offer support. Improving caregivers’ sense of control and social support may be targets for interventions.

People with Non-normative Sexuality or Gender Face Additional Barriers and Stressors with Partner Loss

Bristowe K, Marshall S, Harding R. The bereavement experiences of lesbian, gay, bisexual and/or trans* people who have lost a partner: A systematic review, thematic synthesis and modelling of the literature. Palliat Med. 2016;30(8):730-744.

Background

Grief and bereavement impact individuals differently as they adjust to a death. Increasingly, it is recognized that lesbian, gay, bisexual, and/or transgender (LGBT) communities may face additional barriers when interacting with the healthcare system. This review sought to identify and appraise the evidence of the bereavement experiences among LGBT communities.

Findings

This systematic review summarized quantitative and qualitative data from 23 articles (13 studies). The synthesis noted that the pain associated with the loss of a partner was a universal experience regardless of sexual identity or gender history. Additional barriers and stressors of bereavement were reported for LGBT people, including homophobia, failure to acknowledge the relationship, additional legal and financial issues, and the shadow of human immunodeficiency virus (HIV) or acquired immunodeficiency syndrome (AIDS). LGBT people turned to additional resources for bereavement help: professional support, social and familial support, and societal and community support. Caregiver bereavement support experiences were shaped by whether the relationships were disclosed and accepted (acceptance-disclosure model).

Cautions

The quantitative data was mostly from the 1990s and described the context of HIV/AIDS. The qualitative studies, however, were done in the last decade. Very little research was available for transgender or bisexual caregivers.

Implications

People who identify as LGBT face additional barriers and stressors with the loss of a partner. The described acceptance-disclosure model may help providers be mindful of the additional barriers to LGBT bereavement support.

Physician Trainees Experience Significant Moral Distress with Futile Treatments

Dzeng E, Colaianni A, Roland M, et al. Moral distress amongst American physician trainees regarding futile treatments at the end of life: a qualitative study. J Gen Intern Med. 2016;31(1):93-99.

Background

Physician trainees are often faced with ethical challenges in providing end-of-life care. These ethical challenges can create confusion and conflict about the balance between the benefits and burdens experienced by patients.

Findings

The authors used semistructured, in-depth, qualitative interviews of 22 internal medicine trainees from 3 academic medical centers. An analysis of these interviews revealed several themes. Trainees reported moral distress when (1) many of the treatments provided in end-of-life care (ie, feeding tubes in advanced dementia) were perceived to be futile; (2) they felt obligated to provide end-of-life care that was not in the patient’s best interest, leading to “torture” or “suffering” for the patient; (3) they provided care they felt not to be in the patient’s best interest; (4) they perceived themselves to be powerless to affect change in these dilemmas; (5) they attributed some of their powerlessness to the hierarchy of their academic institutions; and (6) they feared that dehumanization and cynicism would be required to endure this distress.

Cautions

Resident recruitment occurred by solicitation, which may invite bias. Generalizability of qualitative studies to other settings can be limited.

Implications

Trainees may experience several dimensions of moral distress in end-of-life care. These findings challenge training programs to find ways to reduce the dehumanization, sense of powerlessness, and cynicism that this distress may cause.

Disclosure

The authors declare that they have no relevant financial conflicts of interest.

The aim of palliative care (PC) is to improve quality of life for patients facing serious, life-threatening illness and their families.¹ Due to insufficient numbers of PC specialists to meet the PC needs for every hospitalized patient,² all hospitalists should maintain basic PC skills as recognized by PC being a core competency for hospitalists.^3,4

We summarize and critique PC research articles published between January 1, 2016, and December 31, 2016, that have a high likelihood of impacting the practice of hospital medicine. We hand searched 15 journals and conducted a MEDLINE keyword search of PC terms (see Table). All titles and/or abstracts were reviewed and selected for full review based on the following factors: palliative medicine content, scientific rigor, impact on practice, and relevance to hospital medicine. Fifty-five articles were individually reviewed and scored by all authors according to rigor, impact, and relevance. Articles were ranked according to their mean scores, and 9 articles were chosen for inclusion through consensus discussion.

Antipsychotics Were Inferior to a Placebo in Treating Nonterminal Delirium

Agar MR, Lawlor PG, Quinn S, et al. Efficacy of oral risperidone, haloperidol, or placebo for symptoms of delirium among patients in palliative care: a randomized clinical trial. JAMA Intern Med. 2017;177(1):34-42.

Background

Delirium is highly prevalent in PC and is associated with significant distress.⁵ Antipsychotics are widely used for symptoms of delirium, although current evidence does not support this practice in hospitalized adults.^6,7

Findings

This was a double-blind, parallel-arm, placebo randomized controlled trial (RCT) of 247 patients with delirium with an estimated life expectancy of ≥7 days in 11 PC or hospice units across Australia. Patients were randomized to receive risperidone, haloperidol, or a placebo in addition to nonpharmacological management of delirium. Delirium symptom scores after 3 days of treatment, the use of midazolam as a rescue medication, and the presence of extrapyramidal symptoms (EPS) were measured. The risperidone and haloperidol arms had significantly higher delirium symptom scores (P = .02 and P = .009, respectively), mean EPS symptoms (P < .001), and more use of rescue midazolam than the placebo arm. Mortality was higher for antipsychotics, with a hazard ratio of 1.73 for haloperidol (P = .003), 1.29 for risperidone (P = .14), and 1.47 for any antipsychotic (P = .01).

Cautions

The study population was elderly (mean age >70 years) with mild delirium scores. The use of antipsychotics was associated with more benzodiazepine use, which could itself worsen delirium. As patients with clinician-predicted life expectancy of <7 days were excluded, findings cannot be extrapolated to the treatment of terminal delirium, which can often be more symptomatic and difficult to treat.

Implications

Avoid scheduled antipsychotics in patients with nonterminal delirium, as they can increase risk of harm without advantages, over nonpharmacologic interventions.

Low-Dose Morphine Was Superior to Weak Opioids in the Treatment of Moderate Cancer Pain

Bandieri E, Romero M, Ripamonti CI, et al. Randomized trial of low-dose morphine versus weak opioids in moderate cancer pain. J Clin Oncol. 2016;34(5):436-442.

Background

The World Health Organization guidelines recommend the use of weak opioids (WOs), such as codeine or tramadol, as a sequential step in the management of cancer pain.⁸ This strategy has not been tested against low doses of stronger opioids.

Findings

In this multicenter, open-label RCT, 240 patients in Italy were randomized and stratified by age (<75 years or ≥75 years) to either the WO group or low-dose morphine (M) group. The primary outcome measure was a reduction in pain intensity by 20% or more. Secondary outcomes included an improvement in symptom scores, a ≥30% and ≥50% reduction in pain, increased opioid dosage, and adverse side effects. Compared with the WO group, the M group had more patients with a 20% reduction in pain (88.2% vs 54.7%; P < .001), more evidence of pain control in the first week (80.9% vs 43.6%; P < .001), more patients with a ≥30% and ≥50% reduction in pain, and less need to switch to a stronger opioid (15.5% vs 35.0%; P = .001) or require dose increases. Adverse effects were similar in both groups.

Cautions

Patients with chronic kidney disease (CKD) were excluded due to concerns about the accumulation of morphine metabolites. Additionally, this study was open label, increasing the risk of bias.

Implications

Low-dose morphine should be considered over the use of WOs to achieve better and more rapid pain control in patients without CKD.

The Use of Methadone as a Coanalgesic May Improve Moderate Cancer Pain

Courtemanche F, Dao D, Gagné F, et al. Methadone as a coanalgesic for palliative care cancer patients. J Palliat Med. 2016;19(9):972-978.

Background

Methadone is effective at treating cancer pain and is often utilized when patients have neuropathic pain, fail to respond to traditional opioids, or have renal failure.^9,10 However, its long half-life and many drug interactions make methadone challenging to use.

Findings

This cohort study looked at 153 inpatient or outpatient PC patients in Montreal who received methadone as a coanalgesic for cancer pain. The patients’ median morphine equivalent dose was 120 mg when initiating methadone. The median starting dose of methadone was 3 mg per day. Of patients, 49.3% had a significant response (≥30% pain reduction), with a median response time of 7 days, and 30.1% achieved a substantial response (≥50% pain reduction), with a median response time of 3 days. Patients with higher initial pain scores were more likely to respond to adjuvant methadone. Those who had not responded after a week of methadone were unlikely to respond despite dose escalations. Adverse effects included drowsiness (51.4%), confusion (27.4%), constipation (24.7%), nausea (19.9%), and myoclonia (16.4%).

Cautions

This was an observational study with retrospective data, leading to higher levels of missing data. A high rate of adverse side effects was reported (90.4%). Further study is needed to validate and reproduce the findings.

Implications

The use of adjuvant low-dose methadone may be considered in patients with moderate pain despite high-dose opioids. If a response is not seen within 7 days, then methadone use should be reconsidered.

Many Hospitalized Patients on Comfort Care Still Receive Antimicrobials

Merel SE, Meier CA, McKinney CM, Pottinger PS. Antimicrobial use in patients on a comfort care protocol: a retrospective cohort study. J Palliat Med. 2016;19(11):1210-1214.

Background

It is unknown how often patients who are hospitalized at the end of life continue to receive antimicrobials and what factors are associated with antimicrobial use.

Findings

This retrospective cohort study of 1881 hospitalized adults transitioned to a comfort care order (CCO) set at 2 academic medical centers found that 77% of these patients received antimicrobials during their hospital stay (62.4% at 24 hours prior to CCO). Of the 711 still alive at ≥24 hours after CCO, 111 (15.6%) were still on antimicrobials, with that proportion remaining stable for the remainder of hospitalization. In comparing those who did and did not receive antimicrobials after 24 hours of CCO, the presence of a documented infection was not significantly different after adjusting for age. Those with a cancer diagnosis (adjusted risk ratio [ARR] = 1.44: P = .04), a longer length of stay (≥7 days vs <7 days; ARR = 1.49; P = .05), and those discharged home (ARR 2.93; P < .001) or to a facility (ARR 3.63; P < .001) versus dying in the hospital were more likely to be on antimicrobials 24 hours after CCO. Compared with those on a medicine service, patients in the medical and surgical intensive care units (ICUs) were less likely to receive antimicrobials (medical ICU ARR = 0.32; P = .01; surgical ICU and/or neuro-ICU ARR = 0.32; P = .02). The most commonly administered antimicrobials were fluoroquinolones and vancomycin.

Cautions

Only 111 patients were still on antimicrobials at 24 hours, which limited analysis. Investigators relied on retrospective data for medication administration and diagnoses.

Implications

Further work is needed to understand and address the expectations of clinicians, patients, and families regarding the role of antimicrobials at the end of life.

Video Decision Aids Improved Rates of Advance Care Planning and Hospice Use and Decreased Costs

Volandes, AE, Paasche-Orlow MK, Davis AD et al. Use of video decision aids to promote advance care planning in Hilo, Hawai‘i. J Gen Intern Med. 2016;31(9):1035-1040.

Background

Advance care planning (ACP) can be enhanced with the use of video decision aids, which may help address scalability and cost.¹¹ The Hawaii Medical Service Association began an initiative to improve ACP rates, which included a financial incentive. Clinician training and patient access to ACP videos were implemented 1 year into this campaign, which was intended for patients with late-stage disease.

Findings

This study tested the impact of the video intervention on the rates of ACP documentation in Hilo, Hawaii, along with secondary outcomes of hospice use, hospital deaths, and costs. The intervention was sequentially rolled out to Hilo Medical Center (HMC), followed by hospice and primary care practices. Following the video introduction, the proportion of patients discharged from HMC with ACP documentation markedly increased (3.2% to 39.9%; P < .001). The percentage of hospital patients discharged to hospice increased from 5.7% to 13.8% (P < .001). Overall admissions to the Hospice of Hilo increased at a greater rate than in other parts of Hawaii. After the intervention in Hilo, the in-hospital death rate among patients >65 years old declined slightly (P = .14), while in the rest of the state, the rate remained essentially unchanged. ACP planning did not reduce healthcare costs at the end of life, but costs seemed to increase more slowly in Hilo after the intervention than they did in the rest of Hawaii (P < .05).

Cautions

This report relies on before-and-after comparisons, with potential confounding by a background pay-for-quality initiative; however, the timing of the changes in outcomes correlates well with the introduction of the videos. ACP videos have been studied in other settings, so the intervention is likely generalizable to other states.

Implications

A widespread distribution of ACP videos and training for physicians in their use may lead to significant increases in ACP documentation and other beneficial clinical outcomes for patients and health systems.

A Standardized Palliative Care-Led Intervention Did Not Improve Psychological Outcomes in Families of Patients with Chronic Critical Illness

Carson SS, Cox CE, Wallenstein S, et al. Effect of palliative care-led meetings for families of patients with chronic critical illness: a randomized clinical trial. JAMA. 2016;316(1):51-62.

Background

Chronic critical illness (CCI) occurs when a patient neither recovers nor dies for days to weeks after an acute illness requiring aggressive intensive care. CCI is associated with poor patient and family outcomes.¹² Does a protocol-driven support and information meeting led by PC providers improve these outcomes?

Findings

This multicenter RCT compared 130 CCI patients (184 surrogates) who received a structured intervention to 126 patients (181 surrogates) with usual care. The structured intervention was led by PC clinicians in order to provide supportive conversations and information about CCI and prognosis compared with the usual intensivist communication. The support and information team met with the families of patients in the intervention group after day 7 of mechanical ventilation (MV) and again 10 days later. Both the intervention and control groups received validated information about CCI, and all were eligible for specialty PC consultation, as indicated. The primary outcome of the study was the Hospital Anxiety and Depression Scale (HADS) at 90-day follow-up with the surrogates. Secondary endpoints included posttraumatic stress disorder (PTSD) assessment and other communication measures as well as patient outcomes (hospital mortality, 90-day survival, length of stay, and days of MV). At least 1 meeting took place for 89% of patients (82% of surrogates) in the intervention arm. Fewer patients in the intervention arm had nonstudy PC consultations (13% vs 22%). Ninety-day HADS results were similar in the 2 groups. PTSD symptoms, however, were higher in the intervention group (Impact of Event Scale-Revised score: 25.9 for intervention and 21.3 for control; intergroup difference 4.6 [95% confidence interval, 0.01-9.10]). There were no statistically significant differences among the patient-focused measures, including survival.

Cautions

Although the teams contained skilled clinicians led by PC practitioners, this was not an ordinary PC intervention. The intervention included information and emotional support meetings alone rather than support from a PC team driven by clinical considerations. This study included surrogates of patients with CCI but not other conditions.

Implications

Protocol-driven support and information meetings may not improve, and may slightly worsen, outcomes in families of patients with CCI. This study did not evaluate and should not be applied to clinically indicated, specialty PC consultation in the ICU.

Caregivers of Patients Surviving Prolonged Critical Illness Experience High and Persistent Rates of Depression

Cameron JI, Chu LM, Matte A, et al. One-year outcomes in caregivers of critically ill patients. N Engl J Med. 2016;374(19):1831-1841.

Background

More than half of patients with a CCI require caregiver support 1 year after hospitalization.¹³ Caregivers provide tremendous physical and psychosocial support to their loved ones, but that care is often associated with significant burden.¹⁴

Findings

This prospective parallel cohort study followed caregivers of surviving patients ventilated for at least 7 days from 10 academic hospitals in Canada. The prevalence of depression (Center for Epidemiologic Studies–Depression scale ≥16) in this cohort of 280 caregivers (70% were women) was 67%, 49%, 43%, and 43% at the survey intervals of 7 days, 3 months, 6 months, and 12 months after ICU discharge, respectively. Using latent-class linear mixed models, the investigators identified 2 groups of caregivers: those whose depressive symptoms decreased over time (84%) and those whose depressive symptoms persisted at a high level for the year (16%). Patient characteristics (such as age, comorbidity, sex, and functional status) were not associated with caregiver outcomes. Younger caregiver age, greater effect of patient care on other activities, less social support, less mastery (sense of control), and less personal growth were associated with worse caregiver mental health outcomes.

Cautions

Although this is a high-quality prospective study, causality of caregiving on the high rates of depressive symptoms cannot be confirmed without a control group or knowledge of the caregivers’ mental health status prior to the episode of prolonged critical illness.

Implications

Patient critical illness may have serious impacts on caregiver health and well-being. Hospitalists should be attentive to factors associated with caregiver vulnerability and offer support. Improving caregivers’ sense of control and social support may be targets for interventions.

People with Non-normative Sexuality or Gender Face Additional Barriers and Stressors with Partner Loss

Bristowe K, Marshall S, Harding R. The bereavement experiences of lesbian, gay, bisexual and/or trans* people who have lost a partner: A systematic review, thematic synthesis and modelling of the literature. Palliat Med. 2016;30(8):730-744.

Background

Grief and bereavement impact individuals differently as they adjust to a death. Increasingly, it is recognized that lesbian, gay, bisexual, and/or transgender (LGBT) communities may face additional barriers when interacting with the healthcare system. This review sought to identify and appraise the evidence of the bereavement experiences among LGBT communities.

Findings

This systematic review summarized quantitative and qualitative data from 23 articles (13 studies). The synthesis noted that the pain associated with the loss of a partner was a universal experience regardless of sexual identity or gender history. Additional barriers and stressors of bereavement were reported for LGBT people, including homophobia, failure to acknowledge the relationship, additional legal and financial issues, and the shadow of human immunodeficiency virus (HIV) or acquired immunodeficiency syndrome (AIDS). LGBT people turned to additional resources for bereavement help: professional support, social and familial support, and societal and community support. Caregiver bereavement support experiences were shaped by whether the relationships were disclosed and accepted (acceptance-disclosure model).

Cautions

The quantitative data was mostly from the 1990s and described the context of HIV/AIDS. The qualitative studies, however, were done in the last decade. Very little research was available for transgender or bisexual caregivers.

Implications

People who identify as LGBT face additional barriers and stressors with the loss of a partner. The described acceptance-disclosure model may help providers be mindful of the additional barriers to LGBT bereavement support.

Physician Trainees Experience Significant Moral Distress with Futile Treatments

Dzeng E, Colaianni A, Roland M, et al. Moral distress amongst American physician trainees regarding futile treatments at the end of life: a qualitative study. J Gen Intern Med. 2016;31(1):93-99.

Background

Physician trainees are often faced with ethical challenges in providing end-of-life care. These ethical challenges can create confusion and conflict about the balance between the benefits and burdens experienced by patients.

Findings

The authors used semistructured, in-depth, qualitative interviews of 22 internal medicine trainees from 3 academic medical centers. An analysis of these interviews revealed several themes. Trainees reported moral distress when (1) many of the treatments provided in end-of-life care (ie, feeding tubes in advanced dementia) were perceived to be futile; (2) they felt obligated to provide end-of-life care that was not in the patient’s best interest, leading to “torture” or “suffering” for the patient; (3) they provided care they felt not to be in the patient’s best interest; (4) they perceived themselves to be powerless to affect change in these dilemmas; (5) they attributed some of their powerlessness to the hierarchy of their academic institutions; and (6) they feared that dehumanization and cynicism would be required to endure this distress.

Cautions

Resident recruitment occurred by solicitation, which may invite bias. Generalizability of qualitative studies to other settings can be limited.

Implications

Trainees may experience several dimensions of moral distress in end-of-life care. These findings challenge training programs to find ways to reduce the dehumanization, sense of powerlessness, and cynicism that this distress may cause.

Disclosure

The authors declare that they have no relevant financial conflicts of interest.

The aim of palliative care (PC) is to improve quality of life for patients facing serious, life-threatening illness and their families.¹ Due to insufficient numbers of PC specialists to meet the PC needs for every hospitalized patient,² all hospitalists should maintain basic PC skills as recognized by PC being a core competency for hospitalists.^3,4

We summarize and critique PC research articles published between January 1, 2016, and December 31, 2016, that have a high likelihood of impacting the practice of hospital medicine. We hand searched 15 journals and conducted a MEDLINE keyword search of PC terms (see Table). All titles and/or abstracts were reviewed and selected for full review based on the following factors: palliative medicine content, scientific rigor, impact on practice, and relevance to hospital medicine. Fifty-five articles were individually reviewed and scored by all authors according to rigor, impact, and relevance. Articles were ranked according to their mean scores, and 9 articles were chosen for inclusion through consensus discussion.

Antipsychotics Were Inferior to a Placebo in Treating Nonterminal Delirium

Agar MR, Lawlor PG, Quinn S, et al. Efficacy of oral risperidone, haloperidol, or placebo for symptoms of delirium among patients in palliative care: a randomized clinical trial. JAMA Intern Med. 2017;177(1):34-42.

Background

Delirium is highly prevalent in PC and is associated with significant distress.⁵ Antipsychotics are widely used for symptoms of delirium, although current evidence does not support this practice in hospitalized adults.^6,7

Findings

This was a double-blind, parallel-arm, placebo randomized controlled trial (RCT) of 247 patients with delirium with an estimated life expectancy of ≥7 days in 11 PC or hospice units across Australia. Patients were randomized to receive risperidone, haloperidol, or a placebo in addition to nonpharmacological management of delirium. Delirium symptom scores after 3 days of treatment, the use of midazolam as a rescue medication, and the presence of extrapyramidal symptoms (EPS) were measured. The risperidone and haloperidol arms had significantly higher delirium symptom scores (P = .02 and P = .009, respectively), mean EPS symptoms (P < .001), and more use of rescue midazolam than the placebo arm. Mortality was higher for antipsychotics, with a hazard ratio of 1.73 for haloperidol (P = .003), 1.29 for risperidone (P = .14), and 1.47 for any antipsychotic (P = .01).

Cautions

The study population was elderly (mean age >70 years) with mild delirium scores. The use of antipsychotics was associated with more benzodiazepine use, which could itself worsen delirium. As patients with clinician-predicted life expectancy of <7 days were excluded, findings cannot be extrapolated to the treatment of terminal delirium, which can often be more symptomatic and difficult to treat.

Implications

Avoid scheduled antipsychotics in patients with nonterminal delirium, as they can increase risk of harm without advantages, over nonpharmacologic interventions.

Low-Dose Morphine Was Superior to Weak Opioids in the Treatment of Moderate Cancer Pain

Bandieri E, Romero M, Ripamonti CI, et al. Randomized trial of low-dose morphine versus weak opioids in moderate cancer pain. J Clin Oncol. 2016;34(5):436-442.

Background

The World Health Organization guidelines recommend the use of weak opioids (WOs), such as codeine or tramadol, as a sequential step in the management of cancer pain.⁸ This strategy has not been tested against low doses of stronger opioids.

Findings

In this multicenter, open-label RCT, 240 patients in Italy were randomized and stratified by age (<75 years or ≥75 years) to either the WO group or low-dose morphine (M) group. The primary outcome measure was a reduction in pain intensity by 20% or more. Secondary outcomes included an improvement in symptom scores, a ≥30% and ≥50% reduction in pain, increased opioid dosage, and adverse side effects. Compared with the WO group, the M group had more patients with a 20% reduction in pain (88.2% vs 54.7%; P < .001), more evidence of pain control in the first week (80.9% vs 43.6%; P < .001), more patients with a ≥30% and ≥50% reduction in pain, and less need to switch to a stronger opioid (15.5% vs 35.0%; P = .001) or require dose increases. Adverse effects were similar in both groups.

Cautions

Patients with chronic kidney disease (CKD) were excluded due to concerns about the accumulation of morphine metabolites. Additionally, this study was open label, increasing the risk of bias.

Implications

Low-dose morphine should be considered over the use of WOs to achieve better and more rapid pain control in patients without CKD.

The Use of Methadone as a Coanalgesic May Improve Moderate Cancer Pain

Courtemanche F, Dao D, Gagné F, et al. Methadone as a coanalgesic for palliative care cancer patients. J Palliat Med. 2016;19(9):972-978.

Background

Methadone is effective at treating cancer pain and is often utilized when patients have neuropathic pain, fail to respond to traditional opioids, or have renal failure.^9,10 However, its long half-life and many drug interactions make methadone challenging to use.

Findings

This cohort study looked at 153 inpatient or outpatient PC patients in Montreal who received methadone as a coanalgesic for cancer pain. The patients’ median morphine equivalent dose was 120 mg when initiating methadone. The median starting dose of methadone was 3 mg per day. Of patients, 49.3% had a significant response (≥30% pain reduction), with a median response time of 7 days, and 30.1% achieved a substantial response (≥50% pain reduction), with a median response time of 3 days. Patients with higher initial pain scores were more likely to respond to adjuvant methadone. Those who had not responded after a week of methadone were unlikely to respond despite dose escalations. Adverse effects included drowsiness (51.4%), confusion (27.4%), constipation (24.7%), nausea (19.9%), and myoclonia (16.4%).

Cautions

This was an observational study with retrospective data, leading to higher levels of missing data. A high rate of adverse side effects was reported (90.4%). Further study is needed to validate and reproduce the findings.

Implications

The use of adjuvant low-dose methadone may be considered in patients with moderate pain despite high-dose opioids. If a response is not seen within 7 days, then methadone use should be reconsidered.

Many Hospitalized Patients on Comfort Care Still Receive Antimicrobials

Merel SE, Meier CA, McKinney CM, Pottinger PS. Antimicrobial use in patients on a comfort care protocol: a retrospective cohort study. J Palliat Med. 2016;19(11):1210-1214.

Background

It is unknown how often patients who are hospitalized at the end of life continue to receive antimicrobials and what factors are associated with antimicrobial use.

Findings

This retrospective cohort study of 1881 hospitalized adults transitioned to a comfort care order (CCO) set at 2 academic medical centers found that 77% of these patients received antimicrobials during their hospital stay (62.4% at 24 hours prior to CCO). Of the 711 still alive at ≥24 hours after CCO, 111 (15.6%) were still on antimicrobials, with that proportion remaining stable for the remainder of hospitalization. In comparing those who did and did not receive antimicrobials after 24 hours of CCO, the presence of a documented infection was not significantly different after adjusting for age. Those with a cancer diagnosis (adjusted risk ratio [ARR] = 1.44: P = .04), a longer length of stay (≥7 days vs <7 days; ARR = 1.49; P = .05), and those discharged home (ARR 2.93; P < .001) or to a facility (ARR 3.63; P < .001) versus dying in the hospital were more likely to be on antimicrobials 24 hours after CCO. Compared with those on a medicine service, patients in the medical and surgical intensive care units (ICUs) were less likely to receive antimicrobials (medical ICU ARR = 0.32; P = .01; surgical ICU and/or neuro-ICU ARR = 0.32; P = .02). The most commonly administered antimicrobials were fluoroquinolones and vancomycin.

Cautions

Only 111 patients were still on antimicrobials at 24 hours, which limited analysis. Investigators relied on retrospective data for medication administration and diagnoses.

Implications

Further work is needed to understand and address the expectations of clinicians, patients, and families regarding the role of antimicrobials at the end of life.

Video Decision Aids Improved Rates of Advance Care Planning and Hospice Use and Decreased Costs

Volandes, AE, Paasche-Orlow MK, Davis AD et al. Use of video decision aids to promote advance care planning in Hilo, Hawai‘i. J Gen Intern Med. 2016;31(9):1035-1040.

Background

Advance care planning (ACP) can be enhanced with the use of video decision aids, which may help address scalability and cost.¹¹ The Hawaii Medical Service Association began an initiative to improve ACP rates, which included a financial incentive. Clinician training and patient access to ACP videos were implemented 1 year into this campaign, which was intended for patients with late-stage disease.

Findings

This study tested the impact of the video intervention on the rates of ACP documentation in Hilo, Hawaii, along with secondary outcomes of hospice use, hospital deaths, and costs. The intervention was sequentially rolled out to Hilo Medical Center (HMC), followed by hospice and primary care practices. Following the video introduction, the proportion of patients discharged from HMC with ACP documentation markedly increased (3.2% to 39.9%; P < .001). The percentage of hospital patients discharged to hospice increased from 5.7% to 13.8% (P < .001). Overall admissions to the Hospice of Hilo increased at a greater rate than in other parts of Hawaii. After the intervention in Hilo, the in-hospital death rate among patients >65 years old declined slightly (P = .14), while in the rest of the state, the rate remained essentially unchanged. ACP planning did not reduce healthcare costs at the end of life, but costs seemed to increase more slowly in Hilo after the intervention than they did in the rest of Hawaii (P < .05).

Cautions

This report relies on before-and-after comparisons, with potential confounding by a background pay-for-quality initiative; however, the timing of the changes in outcomes correlates well with the introduction of the videos. ACP videos have been studied in other settings, so the intervention is likely generalizable to other states.

Implications

A widespread distribution of ACP videos and training for physicians in their use may lead to significant increases in ACP documentation and other beneficial clinical outcomes for patients and health systems.

A Standardized Palliative Care-Led Intervention Did Not Improve Psychological Outcomes in Families of Patients with Chronic Critical Illness

Carson SS, Cox CE, Wallenstein S, et al. Effect of palliative care-led meetings for families of patients with chronic critical illness: a randomized clinical trial. JAMA. 2016;316(1):51-62.

Background

Chronic critical illness (CCI) occurs when a patient neither recovers nor dies for days to weeks after an acute illness requiring aggressive intensive care. CCI is associated with poor patient and family outcomes.¹² Does a protocol-driven support and information meeting led by PC providers improve these outcomes?

Findings

This multicenter RCT compared 130 CCI patients (184 surrogates) who received a structured intervention to 126 patients (181 surrogates) with usual care. The structured intervention was led by PC clinicians in order to provide supportive conversations and information about CCI and prognosis compared with the usual intensivist communication. The support and information team met with the families of patients in the intervention group after day 7 of mechanical ventilation (MV) and again 10 days later. Both the intervention and control groups received validated information about CCI, and all were eligible for specialty PC consultation, as indicated. The primary outcome of the study was the Hospital Anxiety and Depression Scale (HADS) at 90-day follow-up with the surrogates. Secondary endpoints included posttraumatic stress disorder (PTSD) assessment and other communication measures as well as patient outcomes (hospital mortality, 90-day survival, length of stay, and days of MV). At least 1 meeting took place for 89% of patients (82% of surrogates) in the intervention arm. Fewer patients in the intervention arm had nonstudy PC consultations (13% vs 22%). Ninety-day HADS results were similar in the 2 groups. PTSD symptoms, however, were higher in the intervention group (Impact of Event Scale-Revised score: 25.9 for intervention and 21.3 for control; intergroup difference 4.6 [95% confidence interval, 0.01-9.10]). There were no statistically significant differences among the patient-focused measures, including survival.

Cautions

Although the teams contained skilled clinicians led by PC practitioners, this was not an ordinary PC intervention. The intervention included information and emotional support meetings alone rather than support from a PC team driven by clinical considerations. This study included surrogates of patients with CCI but not other conditions.

Implications

Protocol-driven support and information meetings may not improve, and may slightly worsen, outcomes in families of patients with CCI. This study did not evaluate and should not be applied to clinically indicated, specialty PC consultation in the ICU.

Caregivers of Patients Surviving Prolonged Critical Illness Experience High and Persistent Rates of Depression

Cameron JI, Chu LM, Matte A, et al. One-year outcomes in caregivers of critically ill patients. N Engl J Med. 2016;374(19):1831-1841.

Background

More than half of patients with a CCI require caregiver support 1 year after hospitalization.¹³ Caregivers provide tremendous physical and psychosocial support to their loved ones, but that care is often associated with significant burden.¹⁴

Findings

This prospective parallel cohort study followed caregivers of surviving patients ventilated for at least 7 days from 10 academic hospitals in Canada. The prevalence of depression (Center for Epidemiologic Studies–Depression scale ≥16) in this cohort of 280 caregivers (70% were women) was 67%, 49%, 43%, and 43% at the survey intervals of 7 days, 3 months, 6 months, and 12 months after ICU discharge, respectively. Using latent-class linear mixed models, the investigators identified 2 groups of caregivers: those whose depressive symptoms decreased over time (84%) and those whose depressive symptoms persisted at a high level for the year (16%). Patient characteristics (such as age, comorbidity, sex, and functional status) were not associated with caregiver outcomes. Younger caregiver age, greater effect of patient care on other activities, less social support, less mastery (sense of control), and less personal growth were associated with worse caregiver mental health outcomes.

Cautions

Although this is a high-quality prospective study, causality of caregiving on the high rates of depressive symptoms cannot be confirmed without a control group or knowledge of the caregivers’ mental health status prior to the episode of prolonged critical illness.

Implications

Patient critical illness may have serious impacts on caregiver health and well-being. Hospitalists should be attentive to factors associated with caregiver vulnerability and offer support. Improving caregivers’ sense of control and social support may be targets for interventions.

People with Non-normative Sexuality or Gender Face Additional Barriers and Stressors with Partner Loss

Bristowe K, Marshall S, Harding R. The bereavement experiences of lesbian, gay, bisexual and/or trans* people who have lost a partner: A systematic review, thematic synthesis and modelling of the literature. Palliat Med. 2016;30(8):730-744.

Background

Grief and bereavement impact individuals differently as they adjust to a death. Increasingly, it is recognized that lesbian, gay, bisexual, and/or transgender (LGBT) communities may face additional barriers when interacting with the healthcare system. This review sought to identify and appraise the evidence of the bereavement experiences among LGBT communities.

Findings

This systematic review summarized quantitative and qualitative data from 23 articles (13 studies). The synthesis noted that the pain associated with the loss of a partner was a universal experience regardless of sexual identity or gender history. Additional barriers and stressors of bereavement were reported for LGBT people, including homophobia, failure to acknowledge the relationship, additional legal and financial issues, and the shadow of human immunodeficiency virus (HIV) or acquired immunodeficiency syndrome (AIDS). LGBT people turned to additional resources for bereavement help: professional support, social and familial support, and societal and community support. Caregiver bereavement support experiences were shaped by whether the relationships were disclosed and accepted (acceptance-disclosure model).

Cautions

The quantitative data was mostly from the 1990s and described the context of HIV/AIDS. The qualitative studies, however, were done in the last decade. Very little research was available for transgender or bisexual caregivers.

Implications

People who identify as LGBT face additional barriers and stressors with the loss of a partner. The described acceptance-disclosure model may help providers be mindful of the additional barriers to LGBT bereavement support.

Physician Trainees Experience Significant Moral Distress with Futile Treatments

Dzeng E, Colaianni A, Roland M, et al. Moral distress amongst American physician trainees regarding futile treatments at the end of life: a qualitative study. J Gen Intern Med. 2016;31(1):93-99.

Background

Physician trainees are often faced with ethical challenges in providing end-of-life care. These ethical challenges can create confusion and conflict about the balance between the benefits and burdens experienced by patients.

Findings

The authors used semistructured, in-depth, qualitative interviews of 22 internal medicine trainees from 3 academic medical centers. An analysis of these interviews revealed several themes. Trainees reported moral distress when (1) many of the treatments provided in end-of-life care (ie, feeding tubes in advanced dementia) were perceived to be futile; (2) they felt obligated to provide end-of-life care that was not in the patient’s best interest, leading to “torture” or “suffering” for the patient; (3) they provided care they felt not to be in the patient’s best interest; (4) they perceived themselves to be powerless to affect change in these dilemmas; (5) they attributed some of their powerlessness to the hierarchy of their academic institutions; and (6) they feared that dehumanization and cynicism would be required to endure this distress.

Cautions

Resident recruitment occurred by solicitation, which may invite bias. Generalizability of qualitative studies to other settings can be limited.

Implications

Trainees may experience several dimensions of moral distress in end-of-life care. These findings challenge training programs to find ways to reduce the dehumanization, sense of powerlessness, and cynicism that this distress may cause.

Disclosure

The authors declare that they have no relevant financial conflicts of interest.

Engaging patients and their caregivers in care improves health outcomes^1-3 and is endorsed by leading healthcare organizations as essential to improving care quality and safety.^4-6 Patient engagement emphasizes that patients, caregivers, and healthcare providers work together to “promote and support active patient and public involvement in health and healthcare and to strengthen their influence on healthcare decisions.”⁷ Patient portals, web-based personal health records linked to electronic health record (EHR) data, are intended to promote engagement by providing patients and their caregivers with timely electronic access to their healthcare information and supporting communication through secure messaging with their healthcare team.⁸ The use of patient portals has also been suggested as a way for patients and/or caregivers to identify and intercept medical errors, thus having the potential to also improve patient safety.^8,9

As a requirement for meaningful use, access to health information through patient portals in the ambulatory setting has increased dramatically.¹⁰ Studies evaluating the use of these patient portals to promote patient-centered care are growing, but evidence supporting their impact on improved health outcomes is currently insufficient.^11-15 Although research and policy focus on the use of patient portals in the ambulatory setting, recent literature suggests that patient portals may be used to share inpatient clinical information to engage patients and their caregivers during their hospitalization.^16-18 Before the widespread use of patient portals in the inpatient setting is endorsed, systematic research is needed to understand optimal portal design requirements, if and how these portals are used, and whether their use provides value to the hospitalized patient and/or caregiver.⁸

Prior literature summarized early findings regarding the use of various technologies designed to engage hospitalized patients.^17,19,20 In this systematic review, we describe the emerging literature examining the design, use, and impact of inpatient portals for hospitalized patients and/or caregivers over the last 10 years. Inpatient portals are defined here as electronic patient portals tethered to EHRs that are designed to provide hospitalized patients and/or caregivers secure access to personalized, inpatient clinical information with the intent of engaging them in their hospital care. After analyzing and summarizing these data, we then identify knowledge gaps and potential future research directions.

Search Strategy, Study Selection, and Analysis

This systematic review included available, peer-reviewed, and grey literature published from January 1, 2006, to August 8, 2017, in PubMed, Web of Science (including the Institute of Electrical and Electronics Engineers Xplore), Cochrane, CINAHLPlus, and Scopus databases. Terms and phrases, including those found in the Medical Subject Heading (MeSH) index, were used to identify studies evaluating (1) patient portals (“health record, personal [MeSH],” “personal health record,” “patient portal,” “inpatient portal,” “ipad,” “tablet,” or “bedside information technology”), (2) engagement (“engagement,” “empowerment,” “participation,” “activation,” or “self-efficacy”), and (3) in the hospital (“inpatient [MeSH],” “hospital [MeSH],” “hospitalized patient [MeSH],” or “unit”). MeSH terms were used when applicable. Based on previous literature, free-text terms were also used when subject headings were not applied consistently, such as with terms related to engagement.^17,21 Studies were excluded if they were not written in English, if they evaluated portals exclusively in the emergency department or ambulatory setting, and/or if they described future study protocols. Studies describing general inpatient technology or evaluating portals used in the hospital but not tethered to inpatient EHR clinical data were also excluded.

By using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines,²² 2 researchers (M.K. and P.H.) completed the literature search and potential article screening. Results were aggregated and studies were screened and excluded from full review based on title and abstract information. Additional studies were included after reference list review. During a full review of included studies, 2 researchers independently extracted data, including the study objective, design, setting, sample, data collection instruments, outcomes, and a description of results. Guided by our study objective, findings were reconciled by consensus and analyzed and described according to the following 3 themes: (1) inpatient portal design, (2) inpatient portal use and usability, and (3) the impact of inpatient portal use on patient or caregiver and healthcare team outcomes as defined by retrieved studies.

The quality of studies was evaluated by the same 2 researchers independently by using the Downs and Black checklist for assessing the methodological quality of randomized and nonrandomized healthcare interventions.²³ Qualitative studies describing the development of portal prototypes and/or portal redesign efforts were excluded from these analyses. Discrepancies were resolved by consensus.Because of the wide variability in study designs, populations, and outcomes, a meta-analysis of pooled data was not performed.

Of the 731 studies identified through database searching and reference review, 36 were included for full-text review and 17 met inclusion criteria (Figure; Table 1). Studies excluded after full-text review described portal use outside of the inpatient setting, portals not linked to hospital EHR clinical data, portals not designed for inpatients, and/or inpatient technology in general. The inpatient portal platforms, hardware used, and functionalities varied within included studies (Table 2). The majority of studies used custom, web-based inpatient portal applications on tablet computers. Most provided information about the patients’ hospital medications, healthcare team, and education about their condition and/or a medical glossary. Many included the patient’s schedule, hospital problem list, discharge information, and a way to keep notes.

There has been a recent increase in inpatient portal study publication, with 9 studies published during or after 2016. Five were conducted in the pediatric setting and all but 1³⁰ with English-speaking participants. Twelve studies were qualitative, many of which were conducted in multiple phases by using semi-structured interviews and/or focus groups to develop or redesign inpatient portals. Of the remaining studies, 3 used a cross-sectional design, 1 used a before and after design without a control group, and 1 was a nonrandomized trial. Studies were rated as having medium-to-high risk of bias because of design flaws (Table 1 in supplementary Appendix). Because many studies were small pilot studies and all were single-centered studies, the generalizability of findings to different healthcare settings or patient populations is limited.

Inpatient Portal Design

Most included studies evaluated patient and/or caregiver information needs to design and/or enhance inpatient portals.^16,24-37 In 1 study, patients described an overall lack of information provided in the hospital and insufficient time to understand and remember information, which, when shared, was often presented by using medical terminology.³⁰ They wanted information to help them understand their daily hospital routine, confirm and compare medications and test results, learn about care, and prepare for discharge. Participants in multiple studies echoed these results, indicating the need for a schedule of upcoming clinical events (eg, medication administration, procedures, imaging), secure and timely clinical information (eg, list of diagnoses and medications, test results), personalized education, a medical glossary, discharge information, and a way to take notes and recognize and communicate with providers.

Patients also requested further information transparency,^34,37 including physicians’ notes, radiology results, operative reports, and billing information, along with general hospital information,¹⁶ meal ordering,³³ and video conferencing.²⁷ ln designing and refining an inpatient medication-tracking tool, participants identified the need for information about medication dosage, frequency, timing, administration method, criticality, alternative medications or forms, and education.^26,36 Patients and/or caregivers also indicated interest in communicating with inpatient providers by using the portal.^{16,27,28,30-37} In 1 study, patients highlighted the need to be involved in care plan development,²⁷ which led to portal refinement to allow for patient-generated data entry, including care goals and a way to communicate real-time concerns and feedback.²⁸

Studies also considered healthcare team perspectives to inform portal design.^{25,26,28,30,35,37} Although information needs usually overlapped, patient and healthcare team priorities differed in some areas. Although patients wanted to “know what was going to happen to them,” nurses in 1 study were more concerned about providing information to protect patients, such as safety and precaution materials.²⁵ Similarly, when designing a medication-tracking tool, patients sought information that helped them understand what to expect, while pharmacists focused on medication safety and providing information that fit their workflow (eg, abstract medication schedules).³⁶

Identified study data raised important portal interface design considerations. Results suggested clinical data should be presented by using simple displays,²⁸ accommodating real-time information. Participants recommended links^16,29 to personalized patient-friendly³⁷ education accessed with minimal steps.²⁶ Interfaces may be personalized for target users, such as patient or proxy and younger or older individuals. For example, older patients reported less familiarity with touch screens, internal keyboards, and handwriting recognition, favoring voice recognition for recording notes.²⁷ This raised questions about how portals can be designed to best maintain patient privacy.²⁵ Interface design, such as navigation, also relied heavily on hardware choice, such as tablet versus mobile phone.²⁸

Inpatient Portal Use and Usability

Most patient and/or caregiver participants in included studies were interested in using an inpatient portal, used it when offered, found it easy to use, useful, and/or were satisfied with it.^16,18,24-37 Most used and liked functionalities that provided healthcare team, test result, and medication information.^22,33,37 In the 1 identified controlled trial,¹⁸ researchers evaluated an inpatient portal given to adult inpatients that included a problem list, schedule, medication list, and healthcare team information. Of the intervention unit patients, 80% used the portal, 76% indicated it was easy to use, and 71% thought it provided useful information. When a portal was given to 239 adult patients and caregivers in another study, 66% sent a total of 291 messages to the healthcare team.³¹ Of these, 153 provided feedback, 76 expressed preferences, and 16 communicated concerns. In a pediatric study, an inpatient portal was given to 296 parents who sent a total of 36 messages and 176 requests.³³ Messages sent included information regarding caregiver needs, questions, updates, and/or positive endorsements of the healthcare team and/or care.

Impact of Inpatient Portal Use

Multiple studies evaluated the impact of inpatient portal use on patient and/or caregiver engagement, empowerment, activation, and/or knowledge, which had mixed results. Most adult patients interviewed in one study had positive experiences using a portal to answer their questions between physician visits and learn about, remember, and engage in care.³⁷ A majority of adult inpatient portal users in another study agreed that portal use helped them feel in control and understand their condition; however, they did not report having improved discharge timing knowledge.²⁹ In a pediatric study, most parent inpatient portal users agreed use improved their ability to monitor, understand, and make decisions about their child’s care.³³ In the controlled trial,¹⁸ a higher percentage of portal intervention patients could identify their physician or role; however, patient activation was not statistically different between intervention and control patients.

Results from included studies also evaluated the impact of portal use on communication. Some suggest inpatient portal use may replace and/or facilitate verbal communication between patients, caregivers, and providers.³⁵ In a pediatric study, 51% of parent portal users reported it gave them the information they needed, reducing the amount of questions they had for their healthcare team.³³ Similarly 43% of 14 adult inpatient portal users in another study thought the portal could replace at least some face-to-face communication.³⁷ Some providers indicated portal use enhanced rounding discussion quality.³⁵ Another study suggested that patient-provider communication via electronic messaging may provide benefits for some patients and not others.³⁷

Multiple studies evaluated patient, caregiver, and/or healthcare team perceptions of the impact of inpatient portal use on detection of errors and patient safety.^29,31,33,35 In adult inpatients, 6% agreed portal use could help them find errors.²⁹ In a pediatric study, 8% reported finding at least 1 medication error by using the portal, and 89% thought use reduced errors in their child’s care.³³ One patient in a qualitative study of adult inpatients cited an example of a dosing error discovered by using the portal.³⁷ Healthcare providers in another study also reported that use facilitated patient error identification.³⁵

Included studies evaluated the potential impact of portal use on patient anxiety, confusion, and/or worry, and the work of healthcare teams. In 1 study, nurses voiced concerns about giving information subject to change or that couldn’t always be achieved because of competing hospital priorities, such as discharge timing.²⁵ They also worried about giving medical information that would create cognitive overload for patients and/or require professional interpretation. Although providers in another study perceived little negative impact on their workflow after portal implementation, they worried about the potential of adding other information to the portal.³⁵ For example, they were concerned that the future release of abnormal test results or sensitive data would lead to confusion and more time spent answering patient questions. Physicians also worried that secure messaging could be overused by patients, would be used to inappropriately express acute concerns, or might adversely affect verbal communication. Providers in 2 studies expressed concerns about potential negative implications of portal use on their work before implementation, which were subsequently reduced after portal implementation.^29,38 Conversely, no parent portal users in another study thought portal information was confusing.³³ One parent participant noted portal use may actually decrease anxiety: “Access to their medical information gives patients and their caregivers perspective and insight into their hospital care and empowers them with knowledge about [what is going on], which reduces anxiety.”³⁷

We identified multiple studies evaluating the design, use, and impact of inpatient patient portals for hospitalized patients and caregivers. Based on the information needs identified by patients and healthcare team participants, multiple key content and design recommendations are suggested, including presenting (1) timely, personalized clinical and educational information in lay terms, (2) the care trajectory, including care plan and patient schedule, and (3) a way to recognize and communicate with the inpatient healthcare team. Design challenges still exist, such as translating medical terminology from EHRs into patient-friendly language, proxy access, and portal integration across transitions. Data from identified studies suggest hospitalized patients and caregivers are interested in and willing to use inpatient portals, but there is less information about the use of each functionality. Evidence supporting the role of inpatient portal use in improving patient and/or caregiver engagement, knowledge, communication, and the quality and safety of care is currently limited. Included studies indicate that healthcare team members had concerns about using portals to share clinical information and communicate electronically in the hospital. The extent to which these concerns translate to demonstrable problems remains to be seen.

Early studies focus on patient and caregiver information needs and portal interface design. Although the necessity for certain core functionalities and design requirements are becoming clear,²⁰ best practices regarding the amount and timing of information released (eg, physician notes, lab results), optimal hardware decisions (eg, large-screen displays, hospital-owned tablets, bring-your-own-device model), and details around secure-messaging implementation in the acute hospital setting are still lacking. Future work is needed to understand optimal patient-provider communication architectures that support improved synchronous and asynchronous messaging and privacy-preserving approaches to the design of these systems to handle patient-generated data as it becomes more commonplace. Although patient participants in these studies were generally satisfied using inpatient portals, many indicated the need for even more transparency, such as the release of results in real time and inclusion of physician notes (even if they could not be fully comprehended).³⁷ As the movement of sharing notes with patients in the ambulatory setting grows,³⁹ it will inevitably extend to the inpatient setting.⁴⁰ Further research is needed to understand the impact of increased transparency on health outcomes, patient anxiety, and inpatient healthcare team workload. Although the majority of studies described the design and/or use of custom portal platforms, EHR vendors are now developing inpatient portals that integrate into preexisting systems (eg, MyChart Bedside, Epic Systems). This will increase the likelihood of broad inpatient portal adoption and may facilitate multicenter trials evaluating the impact of their use.

The next steps will need to focus on the evaluation of specific inpatient portal functionalities and the impact of their use on objective process and outcome measures by using rigorous, experimental study designs. Akin to ambulatory portal research, measures of interest will include patient activation,^41,42 patient and/or caregiver satisfaction,⁴³ care processes (eg, length of stay, readmissions), and patient safety (eg, safety perceptions, adverse drug events, hospital-acquired conditions, and diagnostic errors). More than a mechanism for unidirectional sharing information from providers to the patient, inpatient portals will also provide a platform for the reciprocal exchange of information from the patient to the provider through patient-generated data, such as goal setting and feedback. Patients may play a larger role in reporting hospital satisfaction in real time, reconciling medications, contributing to the treatment plan, and identifying medical errors. As portals are integrated across the care continuum,²⁰ our understanding of their impact may become more clear.

In this review, only 5 studies were conducted in the pediatric hospital setting.^24,32-34,38 With hospitalized children experiencing 3 times more harm from medical errors than adults,⁴⁴ engaging parents in inpatient care to improve safety has become a national priority.⁴⁵ Giving patient portals, or “parent portals,” to parents of hospitalized children may provide a unique opportunity to share healthcare information and promote engagement, a direction for future study. There is also a research gap in evaluating adolescent inpatient portal use. Future portals may be designed to incentivize young children to learn about their hospitalization through games linked to health-related education.

Finally, as patients and caregivers begin using inpatient portals, there will almost certainly be consequences for healthcare teams. Understanding and anticipating human and work system factors influencing inpatient portal adoption and use from the perspectives of both patients and healthcare teams are needed.^46,47 Engaging healthcare team members as valuable stakeholders during implementation and measuring the impact of portal use on their workload is necessary, especially as portal use spreads beyond pilot units. The success of inpatient portals is dependent upon both the positive benefits for patients and their acceptance by healthcare teams.⁴⁸

Limitations exist in conducting a systematic literature review.⁴⁹ The conceptual definition of a portal for hospitalized patients and patient/caregiver engagement is evolving; therefore, our definition may not have captured all relevant studies. We intentionally did not include all inpatient technology, as we were interested in a narrow definition of portals designed for inpatients that provided clinical information from the inpatient EHR. Because of rapid technology changes, we also limited our search to studies published within the last 10 years; prior literature has been described elsewhere.¹⁷ We excluded non-English language studies, limiting our ability to capture the full scope of inpatient portal research. These patients already experience healthcare delivery disparities, widened by the inaccessibility of innovative health information technologies.⁵⁰ Future studies would be enhanced with the inclusion of these participants.

Inpatient portal research is in its infancy but growing rapidly. Studies to date are primarily focused on portal design and have small sample sizes. Early findings suggest that patients and caregivers are, in general, enthusiastic about using inpatient portals. Further research is needed, however, to determine the impact of inpatient portal use on patient engagement and hospital-care quality, safety, and cost.

Disclosure

This work was supported by a Department of Pediatrics Research and Development Grant at the University of Wisconsin School of Medicine and Public Health. This publication was also supported by the Clinical and Translational Science Award program through the National Center for Advancing Translational Sciences, grant UL1TR000427. Dr. Hoonakker’s involvement was also partially supported by the National Science Foundation, grant CMMI 1536987. Funding sources had no involvement in study design, analysis, or interpretation of data. The authors have no conflicts of interest to declare.

Engaging patients and their caregivers in care improves health outcomes^1-3 and is endorsed by leading healthcare organizations as essential to improving care quality and safety.^4-6 Patient engagement emphasizes that patients, caregivers, and healthcare providers work together to “promote and support active patient and public involvement in health and healthcare and to strengthen their influence on healthcare decisions.”⁷ Patient portals, web-based personal health records linked to electronic health record (EHR) data, are intended to promote engagement by providing patients and their caregivers with timely electronic access to their healthcare information and supporting communication through secure messaging with their healthcare team.⁸ The use of patient portals has also been suggested as a way for patients and/or caregivers to identify and intercept medical errors, thus having the potential to also improve patient safety.^8,9

As a requirement for meaningful use, access to health information through patient portals in the ambulatory setting has increased dramatically.¹⁰ Studies evaluating the use of these patient portals to promote patient-centered care are growing, but evidence supporting their impact on improved health outcomes is currently insufficient.^11-15 Although research and policy focus on the use of patient portals in the ambulatory setting, recent literature suggests that patient portals may be used to share inpatient clinical information to engage patients and their caregivers during their hospitalization.^16-18 Before the widespread use of patient portals in the inpatient setting is endorsed, systematic research is needed to understand optimal portal design requirements, if and how these portals are used, and whether their use provides value to the hospitalized patient and/or caregiver.⁸

Prior literature summarized early findings regarding the use of various technologies designed to engage hospitalized patients.^17,19,20 In this systematic review, we describe the emerging literature examining the design, use, and impact of inpatient portals for hospitalized patients and/or caregivers over the last 10 years. Inpatient portals are defined here as electronic patient portals tethered to EHRs that are designed to provide hospitalized patients and/or caregivers secure access to personalized, inpatient clinical information with the intent of engaging them in their hospital care. After analyzing and summarizing these data, we then identify knowledge gaps and potential future research directions.

Search Strategy, Study Selection, and Analysis

This systematic review included available, peer-reviewed, and grey literature published from January 1, 2006, to August 8, 2017, in PubMed, Web of Science (including the Institute of Electrical and Electronics Engineers Xplore), Cochrane, CINAHLPlus, and Scopus databases. Terms and phrases, including those found in the Medical Subject Heading (MeSH) index, were used to identify studies evaluating (1) patient portals (“health record, personal [MeSH],” “personal health record,” “patient portal,” “inpatient portal,” “ipad,” “tablet,” or “bedside information technology”), (2) engagement (“engagement,” “empowerment,” “participation,” “activation,” or “self-efficacy”), and (3) in the hospital (“inpatient [MeSH],” “hospital [MeSH],” “hospitalized patient [MeSH],” or “unit”). MeSH terms were used when applicable. Based on previous literature, free-text terms were also used when subject headings were not applied consistently, such as with terms related to engagement.^17,21 Studies were excluded if they were not written in English, if they evaluated portals exclusively in the emergency department or ambulatory setting, and/or if they described future study protocols. Studies describing general inpatient technology or evaluating portals used in the hospital but not tethered to inpatient EHR clinical data were also excluded.

By using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines,²² 2 researchers (M.K. and P.H.) completed the literature search and potential article screening. Results were aggregated and studies were screened and excluded from full review based on title and abstract information. Additional studies were included after reference list review. During a full review of included studies, 2 researchers independently extracted data, including the study objective, design, setting, sample, data collection instruments, outcomes, and a description of results. Guided by our study objective, findings were reconciled by consensus and analyzed and described according to the following 3 themes: (1) inpatient portal design, (2) inpatient portal use and usability, and (3) the impact of inpatient portal use on patient or caregiver and healthcare team outcomes as defined by retrieved studies.

The quality of studies was evaluated by the same 2 researchers independently by using the Downs and Black checklist for assessing the methodological quality of randomized and nonrandomized healthcare interventions.²³ Qualitative studies describing the development of portal prototypes and/or portal redesign efforts were excluded from these analyses. Discrepancies were resolved by consensus.Because of the wide variability in study designs, populations, and outcomes, a meta-analysis of pooled data was not performed.

Of the 731 studies identified through database searching and reference review, 36 were included for full-text review and 17 met inclusion criteria (Figure; Table 1). Studies excluded after full-text review described portal use outside of the inpatient setting, portals not linked to hospital EHR clinical data, portals not designed for inpatients, and/or inpatient technology in general. The inpatient portal platforms, hardware used, and functionalities varied within included studies (Table 2). The majority of studies used custom, web-based inpatient portal applications on tablet computers. Most provided information about the patients’ hospital medications, healthcare team, and education about their condition and/or a medical glossary. Many included the patient’s schedule, hospital problem list, discharge information, and a way to keep notes.

There has been a recent increase in inpatient portal study publication, with 9 studies published during or after 2016. Five were conducted in the pediatric setting and all but 1³⁰ with English-speaking participants. Twelve studies were qualitative, many of which were conducted in multiple phases by using semi-structured interviews and/or focus groups to develop or redesign inpatient portals. Of the remaining studies, 3 used a cross-sectional design, 1 used a before and after design without a control group, and 1 was a nonrandomized trial. Studies were rated as having medium-to-high risk of bias because of design flaws (Table 1 in supplementary Appendix). Because many studies were small pilot studies and all were single-centered studies, the generalizability of findings to different healthcare settings or patient populations is limited.

Inpatient Portal Design

Most included studies evaluated patient and/or caregiver information needs to design and/or enhance inpatient portals.^16,24-37 In 1 study, patients described an overall lack of information provided in the hospital and insufficient time to understand and remember information, which, when shared, was often presented by using medical terminology.³⁰ They wanted information to help them understand their daily hospital routine, confirm and compare medications and test results, learn about care, and prepare for discharge. Participants in multiple studies echoed these results, indicating the need for a schedule of upcoming clinical events (eg, medication administration, procedures, imaging), secure and timely clinical information (eg, list of diagnoses and medications, test results), personalized education, a medical glossary, discharge information, and a way to take notes and recognize and communicate with providers.

Patients also requested further information transparency,^34,37 including physicians’ notes, radiology results, operative reports, and billing information, along with general hospital information,¹⁶ meal ordering,³³ and video conferencing.²⁷ ln designing and refining an inpatient medication-tracking tool, participants identified the need for information about medication dosage, frequency, timing, administration method, criticality, alternative medications or forms, and education.^26,36 Patients and/or caregivers also indicated interest in communicating with inpatient providers by using the portal.^{16,27,28,30-37} In 1 study, patients highlighted the need to be involved in care plan development,²⁷ which led to portal refinement to allow for patient-generated data entry, including care goals and a way to communicate real-time concerns and feedback.²⁸

Studies also considered healthcare team perspectives to inform portal design.^{25,26,28,30,35,37} Although information needs usually overlapped, patient and healthcare team priorities differed in some areas. Although patients wanted to “know what was going to happen to them,” nurses in 1 study were more concerned about providing information to protect patients, such as safety and precaution materials.²⁵ Similarly, when designing a medication-tracking tool, patients sought information that helped them understand what to expect, while pharmacists focused on medication safety and providing information that fit their workflow (eg, abstract medication schedules).³⁶

Identified study data raised important portal interface design considerations. Results suggested clinical data should be presented by using simple displays,²⁸ accommodating real-time information. Participants recommended links^16,29 to personalized patient-friendly³⁷ education accessed with minimal steps.²⁶ Interfaces may be personalized for target users, such as patient or proxy and younger or older individuals. For example, older patients reported less familiarity with touch screens, internal keyboards, and handwriting recognition, favoring voice recognition for recording notes.²⁷ This raised questions about how portals can be designed to best maintain patient privacy.²⁵ Interface design, such as navigation, also relied heavily on hardware choice, such as tablet versus mobile phone.²⁸

Inpatient Portal Use and Usability

Most patient and/or caregiver participants in included studies were interested in using an inpatient portal, used it when offered, found it easy to use, useful, and/or were satisfied with it.^16,18,24-37 Most used and liked functionalities that provided healthcare team, test result, and medication information.^22,33,37 In the 1 identified controlled trial,¹⁸ researchers evaluated an inpatient portal given to adult inpatients that included a problem list, schedule, medication list, and healthcare team information. Of the intervention unit patients, 80% used the portal, 76% indicated it was easy to use, and 71% thought it provided useful information. When a portal was given to 239 adult patients and caregivers in another study, 66% sent a total of 291 messages to the healthcare team.³¹ Of these, 153 provided feedback, 76 expressed preferences, and 16 communicated concerns. In a pediatric study, an inpatient portal was given to 296 parents who sent a total of 36 messages and 176 requests.³³ Messages sent included information regarding caregiver needs, questions, updates, and/or positive endorsements of the healthcare team and/or care.

Impact of Inpatient Portal Use

Multiple studies evaluated the impact of inpatient portal use on patient and/or caregiver engagement, empowerment, activation, and/or knowledge, which had mixed results. Most adult patients interviewed in one study had positive experiences using a portal to answer their questions between physician visits and learn about, remember, and engage in care.³⁷ A majority of adult inpatient portal users in another study agreed that portal use helped them feel in control and understand their condition; however, they did not report having improved discharge timing knowledge.²⁹ In a pediatric study, most parent inpatient portal users agreed use improved their ability to monitor, understand, and make decisions about their child’s care.³³ In the controlled trial,¹⁸ a higher percentage of portal intervention patients could identify their physician or role; however, patient activation was not statistically different between intervention and control patients.

Results from included studies also evaluated the impact of portal use on communication. Some suggest inpatient portal use may replace and/or facilitate verbal communication between patients, caregivers, and providers.³⁵ In a pediatric study, 51% of parent portal users reported it gave them the information they needed, reducing the amount of questions they had for their healthcare team.³³ Similarly 43% of 14 adult inpatient portal users in another study thought the portal could replace at least some face-to-face communication.³⁷ Some providers indicated portal use enhanced rounding discussion quality.³⁵ Another study suggested that patient-provider communication via electronic messaging may provide benefits for some patients and not others.³⁷

Multiple studies evaluated patient, caregiver, and/or healthcare team perceptions of the impact of inpatient portal use on detection of errors and patient safety.^29,31,33,35 In adult inpatients, 6% agreed portal use could help them find errors.²⁹ In a pediatric study, 8% reported finding at least 1 medication error by using the portal, and 89% thought use reduced errors in their child’s care.³³ One patient in a qualitative study of adult inpatients cited an example of a dosing error discovered by using the portal.³⁷ Healthcare providers in another study also reported that use facilitated patient error identification.³⁵

Included studies evaluated the potential impact of portal use on patient anxiety, confusion, and/or worry, and the work of healthcare teams. In 1 study, nurses voiced concerns about giving information subject to change or that couldn’t always be achieved because of competing hospital priorities, such as discharge timing.²⁵ They also worried about giving medical information that would create cognitive overload for patients and/or require professional interpretation. Although providers in another study perceived little negative impact on their workflow after portal implementation, they worried about the potential of adding other information to the portal.³⁵ For example, they were concerned that the future release of abnormal test results or sensitive data would lead to confusion and more time spent answering patient questions. Physicians also worried that secure messaging could be overused by patients, would be used to inappropriately express acute concerns, or might adversely affect verbal communication. Providers in 2 studies expressed concerns about potential negative implications of portal use on their work before implementation, which were subsequently reduced after portal implementation.^29,38 Conversely, no parent portal users in another study thought portal information was confusing.³³ One parent participant noted portal use may actually decrease anxiety: “Access to their medical information gives patients and their caregivers perspective and insight into their hospital care and empowers them with knowledge about [what is going on], which reduces anxiety.”³⁷

We identified multiple studies evaluating the design, use, and impact of inpatient patient portals for hospitalized patients and caregivers. Based on the information needs identified by patients and healthcare team participants, multiple key content and design recommendations are suggested, including presenting (1) timely, personalized clinical and educational information in lay terms, (2) the care trajectory, including care plan and patient schedule, and (3) a way to recognize and communicate with the inpatient healthcare team. Design challenges still exist, such as translating medical terminology from EHRs into patient-friendly language, proxy access, and portal integration across transitions. Data from identified studies suggest hospitalized patients and caregivers are interested in and willing to use inpatient portals, but there is less information about the use of each functionality. Evidence supporting the role of inpatient portal use in improving patient and/or caregiver engagement, knowledge, communication, and the quality and safety of care is currently limited. Included studies indicate that healthcare team members had concerns about using portals to share clinical information and communicate electronically in the hospital. The extent to which these concerns translate to demonstrable problems remains to be seen.

Early studies focus on patient and caregiver information needs and portal interface design. Although the necessity for certain core functionalities and design requirements are becoming clear,²⁰ best practices regarding the amount and timing of information released (eg, physician notes, lab results), optimal hardware decisions (eg, large-screen displays, hospital-owned tablets, bring-your-own-device model), and details around secure-messaging implementation in the acute hospital setting are still lacking. Future work is needed to understand optimal patient-provider communication architectures that support improved synchronous and asynchronous messaging and privacy-preserving approaches to the design of these systems to handle patient-generated data as it becomes more commonplace. Although patient participants in these studies were generally satisfied using inpatient portals, many indicated the need for even more transparency, such as the release of results in real time and inclusion of physician notes (even if they could not be fully comprehended).³⁷ As the movement of sharing notes with patients in the ambulatory setting grows,³⁹ it will inevitably extend to the inpatient setting.⁴⁰ Further research is needed to understand the impact of increased transparency on health outcomes, patient anxiety, and inpatient healthcare team workload. Although the majority of studies described the design and/or use of custom portal platforms, EHR vendors are now developing inpatient portals that integrate into preexisting systems (eg, MyChart Bedside, Epic Systems). This will increase the likelihood of broad inpatient portal adoption and may facilitate multicenter trials evaluating the impact of their use.

The next steps will need to focus on the evaluation of specific inpatient portal functionalities and the impact of their use on objective process and outcome measures by using rigorous, experimental study designs. Akin to ambulatory portal research, measures of interest will include patient activation,^41,42 patient and/or caregiver satisfaction,⁴³ care processes (eg, length of stay, readmissions), and patient safety (eg, safety perceptions, adverse drug events, hospital-acquired conditions, and diagnostic errors). More than a mechanism for unidirectional sharing information from providers to the patient, inpatient portals will also provide a platform for the reciprocal exchange of information from the patient to the provider through patient-generated data, such as goal setting and feedback. Patients may play a larger role in reporting hospital satisfaction in real time, reconciling medications, contributing to the treatment plan, and identifying medical errors. As portals are integrated across the care continuum,²⁰ our understanding of their impact may become more clear.

In this review, only 5 studies were conducted in the pediatric hospital setting.^24,32-34,38 With hospitalized children experiencing 3 times more harm from medical errors than adults,⁴⁴ engaging parents in inpatient care to improve safety has become a national priority.⁴⁵ Giving patient portals, or “parent portals,” to parents of hospitalized children may provide a unique opportunity to share healthcare information and promote engagement, a direction for future study. There is also a research gap in evaluating adolescent inpatient portal use. Future portals may be designed to incentivize young children to learn about their hospitalization through games linked to health-related education.

Finally, as patients and caregivers begin using inpatient portals, there will almost certainly be consequences for healthcare teams. Understanding and anticipating human and work system factors influencing inpatient portal adoption and use from the perspectives of both patients and healthcare teams are needed.^46,47 Engaging healthcare team members as valuable stakeholders during implementation and measuring the impact of portal use on their workload is necessary, especially as portal use spreads beyond pilot units. The success of inpatient portals is dependent upon both the positive benefits for patients and their acceptance by healthcare teams.⁴⁸

Limitations exist in conducting a systematic literature review.⁴⁹ The conceptual definition of a portal for hospitalized patients and patient/caregiver engagement is evolving; therefore, our definition may not have captured all relevant studies. We intentionally did not include all inpatient technology, as we were interested in a narrow definition of portals designed for inpatients that provided clinical information from the inpatient EHR. Because of rapid technology changes, we also limited our search to studies published within the last 10 years; prior literature has been described elsewhere.¹⁷ We excluded non-English language studies, limiting our ability to capture the full scope of inpatient portal research. These patients already experience healthcare delivery disparities, widened by the inaccessibility of innovative health information technologies.⁵⁰ Future studies would be enhanced with the inclusion of these participants.

Inpatient portal research is in its infancy but growing rapidly. Studies to date are primarily focused on portal design and have small sample sizes. Early findings suggest that patients and caregivers are, in general, enthusiastic about using inpatient portals. Further research is needed, however, to determine the impact of inpatient portal use on patient engagement and hospital-care quality, safety, and cost.

Disclosure

This work was supported by a Department of Pediatrics Research and Development Grant at the University of Wisconsin School of Medicine and Public Health. This publication was also supported by the Clinical and Translational Science Award program through the National Center for Advancing Translational Sciences, grant UL1TR000427. Dr. Hoonakker’s involvement was also partially supported by the National Science Foundation, grant CMMI 1536987. Funding sources had no involvement in study design, analysis, or interpretation of data. The authors have no conflicts of interest to declare.

Engaging patients and their caregivers in care improves health outcomes^1-3 and is endorsed by leading healthcare organizations as essential to improving care quality and safety.^4-6 Patient engagement emphasizes that patients, caregivers, and healthcare providers work together to “promote and support active patient and public involvement in health and healthcare and to strengthen their influence on healthcare decisions.”⁷ Patient portals, web-based personal health records linked to electronic health record (EHR) data, are intended to promote engagement by providing patients and their caregivers with timely electronic access to their healthcare information and supporting communication through secure messaging with their healthcare team.⁸ The use of patient portals has also been suggested as a way for patients and/or caregivers to identify and intercept medical errors, thus having the potential to also improve patient safety.^8,9

As a requirement for meaningful use, access to health information through patient portals in the ambulatory setting has increased dramatically.¹⁰ Studies evaluating the use of these patient portals to promote patient-centered care are growing, but evidence supporting their impact on improved health outcomes is currently insufficient.^11-15 Although research and policy focus on the use of patient portals in the ambulatory setting, recent literature suggests that patient portals may be used to share inpatient clinical information to engage patients and their caregivers during their hospitalization.^16-18 Before the widespread use of patient portals in the inpatient setting is endorsed, systematic research is needed to understand optimal portal design requirements, if and how these portals are used, and whether their use provides value to the hospitalized patient and/or caregiver.⁸

Prior literature summarized early findings regarding the use of various technologies designed to engage hospitalized patients.^17,19,20 In this systematic review, we describe the emerging literature examining the design, use, and impact of inpatient portals for hospitalized patients and/or caregivers over the last 10 years. Inpatient portals are defined here as electronic patient portals tethered to EHRs that are designed to provide hospitalized patients and/or caregivers secure access to personalized, inpatient clinical information with the intent of engaging them in their hospital care. After analyzing and summarizing these data, we then identify knowledge gaps and potential future research directions.

Search Strategy, Study Selection, and Analysis

This systematic review included available, peer-reviewed, and grey literature published from January 1, 2006, to August 8, 2017, in PubMed, Web of Science (including the Institute of Electrical and Electronics Engineers Xplore), Cochrane, CINAHLPlus, and Scopus databases. Terms and phrases, including those found in the Medical Subject Heading (MeSH) index, were used to identify studies evaluating (1) patient portals (“health record, personal [MeSH],” “personal health record,” “patient portal,” “inpatient portal,” “ipad,” “tablet,” or “bedside information technology”), (2) engagement (“engagement,” “empowerment,” “participation,” “activation,” or “self-efficacy”), and (3) in the hospital (“inpatient [MeSH],” “hospital [MeSH],” “hospitalized patient [MeSH],” or “unit”). MeSH terms were used when applicable. Based on previous literature, free-text terms were also used when subject headings were not applied consistently, such as with terms related to engagement.^17,21 Studies were excluded if they were not written in English, if they evaluated portals exclusively in the emergency department or ambulatory setting, and/or if they described future study protocols. Studies describing general inpatient technology or evaluating portals used in the hospital but not tethered to inpatient EHR clinical data were also excluded.

By using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines,²² 2 researchers (M.K. and P.H.) completed the literature search and potential article screening. Results were aggregated and studies were screened and excluded from full review based on title and abstract information. Additional studies were included after reference list review. During a full review of included studies, 2 researchers independently extracted data, including the study objective, design, setting, sample, data collection instruments, outcomes, and a description of results. Guided by our study objective, findings were reconciled by consensus and analyzed and described according to the following 3 themes: (1) inpatient portal design, (2) inpatient portal use and usability, and (3) the impact of inpatient portal use on patient or caregiver and healthcare team outcomes as defined by retrieved studies.

The quality of studies was evaluated by the same 2 researchers independently by using the Downs and Black checklist for assessing the methodological quality of randomized and nonrandomized healthcare interventions.²³ Qualitative studies describing the development of portal prototypes and/or portal redesign efforts were excluded from these analyses. Discrepancies were resolved by consensus.Because of the wide variability in study designs, populations, and outcomes, a meta-analysis of pooled data was not performed.

Of the 731 studies identified through database searching and reference review, 36 were included for full-text review and 17 met inclusion criteria (Figure; Table 1). Studies excluded after full-text review described portal use outside of the inpatient setting, portals not linked to hospital EHR clinical data, portals not designed for inpatients, and/or inpatient technology in general. The inpatient portal platforms, hardware used, and functionalities varied within included studies (Table 2). The majority of studies used custom, web-based inpatient portal applications on tablet computers. Most provided information about the patients’ hospital medications, healthcare team, and education about their condition and/or a medical glossary. Many included the patient’s schedule, hospital problem list, discharge information, and a way to keep notes.

There has been a recent increase in inpatient portal study publication, with 9 studies published during or after 2016. Five were conducted in the pediatric setting and all but 1³⁰ with English-speaking participants. Twelve studies were qualitative, many of which were conducted in multiple phases by using semi-structured interviews and/or focus groups to develop or redesign inpatient portals. Of the remaining studies, 3 used a cross-sectional design, 1 used a before and after design without a control group, and 1 was a nonrandomized trial. Studies were rated as having medium-to-high risk of bias because of design flaws (Table 1 in supplementary Appendix). Because many studies were small pilot studies and all were single-centered studies, the generalizability of findings to different healthcare settings or patient populations is limited.

Inpatient Portal Design

Most included studies evaluated patient and/or caregiver information needs to design and/or enhance inpatient portals.^16,24-37 In 1 study, patients described an overall lack of information provided in the hospital and insufficient time to understand and remember information, which, when shared, was often presented by using medical terminology.³⁰ They wanted information to help them understand their daily hospital routine, confirm and compare medications and test results, learn about care, and prepare for discharge. Participants in multiple studies echoed these results, indicating the need for a schedule of upcoming clinical events (eg, medication administration, procedures, imaging), secure and timely clinical information (eg, list of diagnoses and medications, test results), personalized education, a medical glossary, discharge information, and a way to take notes and recognize and communicate with providers.

Patients also requested further information transparency,^34,37 including physicians’ notes, radiology results, operative reports, and billing information, along with general hospital information,¹⁶ meal ordering,³³ and video conferencing.²⁷ ln designing and refining an inpatient medication-tracking tool, participants identified the need for information about medication dosage, frequency, timing, administration method, criticality, alternative medications or forms, and education.^26,36 Patients and/or caregivers also indicated interest in communicating with inpatient providers by using the portal.^{16,27,28,30-37} In 1 study, patients highlighted the need to be involved in care plan development,²⁷ which led to portal refinement to allow for patient-generated data entry, including care goals and a way to communicate real-time concerns and feedback.²⁸

Studies also considered healthcare team perspectives to inform portal design.^{25,26,28,30,35,37} Although information needs usually overlapped, patient and healthcare team priorities differed in some areas. Although patients wanted to “know what was going to happen to them,” nurses in 1 study were more concerned about providing information to protect patients, such as safety and precaution materials.²⁵ Similarly, when designing a medication-tracking tool, patients sought information that helped them understand what to expect, while pharmacists focused on medication safety and providing information that fit their workflow (eg, abstract medication schedules).³⁶

Identified study data raised important portal interface design considerations. Results suggested clinical data should be presented by using simple displays,²⁸ accommodating real-time information. Participants recommended links^16,29 to personalized patient-friendly³⁷ education accessed with minimal steps.²⁶ Interfaces may be personalized for target users, such as patient or proxy and younger or older individuals. For example, older patients reported less familiarity with touch screens, internal keyboards, and handwriting recognition, favoring voice recognition for recording notes.²⁷ This raised questions about how portals can be designed to best maintain patient privacy.²⁵ Interface design, such as navigation, also relied heavily on hardware choice, such as tablet versus mobile phone.²⁸

Inpatient Portal Use and Usability

Most patient and/or caregiver participants in included studies were interested in using an inpatient portal, used it when offered, found it easy to use, useful, and/or were satisfied with it.^16,18,24-37 Most used and liked functionalities that provided healthcare team, test result, and medication information.^22,33,37 In the 1 identified controlled trial,¹⁸ researchers evaluated an inpatient portal given to adult inpatients that included a problem list, schedule, medication list, and healthcare team information. Of the intervention unit patients, 80% used the portal, 76% indicated it was easy to use, and 71% thought it provided useful information. When a portal was given to 239 adult patients and caregivers in another study, 66% sent a total of 291 messages to the healthcare team.³¹ Of these, 153 provided feedback, 76 expressed preferences, and 16 communicated concerns. In a pediatric study, an inpatient portal was given to 296 parents who sent a total of 36 messages and 176 requests.³³ Messages sent included information regarding caregiver needs, questions, updates, and/or positive endorsements of the healthcare team and/or care.

Impact of Inpatient Portal Use

Multiple studies evaluated the impact of inpatient portal use on patient and/or caregiver engagement, empowerment, activation, and/or knowledge, which had mixed results. Most adult patients interviewed in one study had positive experiences using a portal to answer their questions between physician visits and learn about, remember, and engage in care.³⁷ A majority of adult inpatient portal users in another study agreed that portal use helped them feel in control and understand their condition; however, they did not report having improved discharge timing knowledge.²⁹ In a pediatric study, most parent inpatient portal users agreed use improved their ability to monitor, understand, and make decisions about their child’s care.³³ In the controlled trial,¹⁸ a higher percentage of portal intervention patients could identify their physician or role; however, patient activation was not statistically different between intervention and control patients.

Results from included studies also evaluated the impact of portal use on communication. Some suggest inpatient portal use may replace and/or facilitate verbal communication between patients, caregivers, and providers.³⁵ In a pediatric study, 51% of parent portal users reported it gave them the information they needed, reducing the amount of questions they had for their healthcare team.³³ Similarly 43% of 14 adult inpatient portal users in another study thought the portal could replace at least some face-to-face communication.³⁷ Some providers indicated portal use enhanced rounding discussion quality.³⁵ Another study suggested that patient-provider communication via electronic messaging may provide benefits for some patients and not others.³⁷

Multiple studies evaluated patient, caregiver, and/or healthcare team perceptions of the impact of inpatient portal use on detection of errors and patient safety.^29,31,33,35 In adult inpatients, 6% agreed portal use could help them find errors.²⁹ In a pediatric study, 8% reported finding at least 1 medication error by using the portal, and 89% thought use reduced errors in their child’s care.³³ One patient in a qualitative study of adult inpatients cited an example of a dosing error discovered by using the portal.³⁷ Healthcare providers in another study also reported that use facilitated patient error identification.³⁵

Included studies evaluated the potential impact of portal use on patient anxiety, confusion, and/or worry, and the work of healthcare teams. In 1 study, nurses voiced concerns about giving information subject to change or that couldn’t always be achieved because of competing hospital priorities, such as discharge timing.²⁵ They also worried about giving medical information that would create cognitive overload for patients and/or require professional interpretation. Although providers in another study perceived little negative impact on their workflow after portal implementation, they worried about the potential of adding other information to the portal.³⁵ For example, they were concerned that the future release of abnormal test results or sensitive data would lead to confusion and more time spent answering patient questions. Physicians also worried that secure messaging could be overused by patients, would be used to inappropriately express acute concerns, or might adversely affect verbal communication. Providers in 2 studies expressed concerns about potential negative implications of portal use on their work before implementation, which were subsequently reduced after portal implementation.^29,38 Conversely, no parent portal users in another study thought portal information was confusing.³³ One parent participant noted portal use may actually decrease anxiety: “Access to their medical information gives patients and their caregivers perspective and insight into their hospital care and empowers them with knowledge about [what is going on], which reduces anxiety.”³⁷

We identified multiple studies evaluating the design, use, and impact of inpatient patient portals for hospitalized patients and caregivers. Based on the information needs identified by patients and healthcare team participants, multiple key content and design recommendations are suggested, including presenting (1) timely, personalized clinical and educational information in lay terms, (2) the care trajectory, including care plan and patient schedule, and (3) a way to recognize and communicate with the inpatient healthcare team. Design challenges still exist, such as translating medical terminology from EHRs into patient-friendly language, proxy access, and portal integration across transitions. Data from identified studies suggest hospitalized patients and caregivers are interested in and willing to use inpatient portals, but there is less information about the use of each functionality. Evidence supporting the role of inpatient portal use in improving patient and/or caregiver engagement, knowledge, communication, and the quality and safety of care is currently limited. Included studies indicate that healthcare team members had concerns about using portals to share clinical information and communicate electronically in the hospital. The extent to which these concerns translate to demonstrable problems remains to be seen.

Early studies focus on patient and caregiver information needs and portal interface design. Although the necessity for certain core functionalities and design requirements are becoming clear,²⁰ best practices regarding the amount and timing of information released (eg, physician notes, lab results), optimal hardware decisions (eg, large-screen displays, hospital-owned tablets, bring-your-own-device model), and details around secure-messaging implementation in the acute hospital setting are still lacking. Future work is needed to understand optimal patient-provider communication architectures that support improved synchronous and asynchronous messaging and privacy-preserving approaches to the design of these systems to handle patient-generated data as it becomes more commonplace. Although patient participants in these studies were generally satisfied using inpatient portals, many indicated the need for even more transparency, such as the release of results in real time and inclusion of physician notes (even if they could not be fully comprehended).³⁷ As the movement of sharing notes with patients in the ambulatory setting grows,³⁹ it will inevitably extend to the inpatient setting.⁴⁰ Further research is needed to understand the impact of increased transparency on health outcomes, patient anxiety, and inpatient healthcare team workload. Although the majority of studies described the design and/or use of custom portal platforms, EHR vendors are now developing inpatient portals that integrate into preexisting systems (eg, MyChart Bedside, Epic Systems). This will increase the likelihood of broad inpatient portal adoption and may facilitate multicenter trials evaluating the impact of their use.

The next steps will need to focus on the evaluation of specific inpatient portal functionalities and the impact of their use on objective process and outcome measures by using rigorous, experimental study designs. Akin to ambulatory portal research, measures of interest will include patient activation,^41,42 patient and/or caregiver satisfaction,⁴³ care processes (eg, length of stay, readmissions), and patient safety (eg, safety perceptions, adverse drug events, hospital-acquired conditions, and diagnostic errors). More than a mechanism for unidirectional sharing information from providers to the patient, inpatient portals will also provide a platform for the reciprocal exchange of information from the patient to the provider through patient-generated data, such as goal setting and feedback. Patients may play a larger role in reporting hospital satisfaction in real time, reconciling medications, contributing to the treatment plan, and identifying medical errors. As portals are integrated across the care continuum,²⁰ our understanding of their impact may become more clear.

In this review, only 5 studies were conducted in the pediatric hospital setting.^24,32-34,38 With hospitalized children experiencing 3 times more harm from medical errors than adults,⁴⁴ engaging parents in inpatient care to improve safety has become a national priority.⁴⁵ Giving patient portals, or “parent portals,” to parents of hospitalized children may provide a unique opportunity to share healthcare information and promote engagement, a direction for future study. There is also a research gap in evaluating adolescent inpatient portal use. Future portals may be designed to incentivize young children to learn about their hospitalization through games linked to health-related education.

Finally, as patients and caregivers begin using inpatient portals, there will almost certainly be consequences for healthcare teams. Understanding and anticipating human and work system factors influencing inpatient portal adoption and use from the perspectives of both patients and healthcare teams are needed.^46,47 Engaging healthcare team members as valuable stakeholders during implementation and measuring the impact of portal use on their workload is necessary, especially as portal use spreads beyond pilot units. The success of inpatient portals is dependent upon both the positive benefits for patients and their acceptance by healthcare teams.⁴⁸

Limitations exist in conducting a systematic literature review.⁴⁹ The conceptual definition of a portal for hospitalized patients and patient/caregiver engagement is evolving; therefore, our definition may not have captured all relevant studies. We intentionally did not include all inpatient technology, as we were interested in a narrow definition of portals designed for inpatients that provided clinical information from the inpatient EHR. Because of rapid technology changes, we also limited our search to studies published within the last 10 years; prior literature has been described elsewhere.¹⁷ We excluded non-English language studies, limiting our ability to capture the full scope of inpatient portal research. These patients already experience healthcare delivery disparities, widened by the inaccessibility of innovative health information technologies.⁵⁰ Future studies would be enhanced with the inclusion of these participants.

Inpatient portal research is in its infancy but growing rapidly. Studies to date are primarily focused on portal design and have small sample sizes. Early findings suggest that patients and caregivers are, in general, enthusiastic about using inpatient portals. Further research is needed, however, to determine the impact of inpatient portal use on patient engagement and hospital-care quality, safety, and cost.

Disclosure

This work was supported by a Department of Pediatrics Research and Development Grant at the University of Wisconsin School of Medicine and Public Health. This publication was also supported by the Clinical and Translational Science Award program through the National Center for Advancing Translational Sciences, grant UL1TR000427. Dr. Hoonakker’s involvement was also partially supported by the National Science Foundation, grant CMMI 1536987. Funding sources had no involvement in study design, analysis, or interpretation of data. The authors have no conflicts of interest to declare.

Hospitalists’ performance is routinely evaluated by third-party payers, employers, and patients. As hospitalist programs mature, there is a need to develop processes to identify, internally measure, and report on individual and group performance. We know from Society of Hospital Medicine (SHM) data that a significant amount of hospitalists’ total compensation is at least partially based on performance. Often this is based at least in part on quality data. In 2006, SHM issued a white paper detailing the key elements of a successful performance monitoring and reporting process.^1,2 Recommendations included the identification of meaningful operational and clinical performance metrics, and the ability to monitor and report both group and individual metrics was highlighted as an essential component. There is evidence that comparison of individual provider performance with that of their peers is a necessary element of successful provider dashboards.³ Additionally, regular feedback and a clear, visual presentation of the data are important components of successful provider feedback dashboards.^3-6

Much of the literature regarding provider feedback dashboards has been based in the outpatient setting. The majority of these dashboards focus on the management of chronic illnesses (eg, diabetes and hypertension), rates of preventative care services (eg, colonoscopy or mammogram), or avoidance of unnecessary care (eg, antibiotics for sinusitis).^4,5 Unlike in the outpatient setting, in which 1 provider often provides a majority of the care for a given episode of care, hospitalized patients are often cared for by multiple providers, challenging the appropriate attribution of patient-level metrics to specific providers. Under the standard approach, an entire hospitalization is attributed to 1 physician, generally the attending of record for the hospitalization, which may be the admitting provider or the discharging provider, depending on the approach used by the hospital. However, assigning responsibility for an entire hospitalization to a provider who may have only seen the patient for a small percentage of a hospitalization may jeopardize the validity of metrics. As provider metrics are increasingly being used for compensation, it is important to ensure that the method for attribution correctly identifies the providers caring for patients. To our knowledge there is no gold standard approach for attributing metrics to providers when patients are cared for by multiple providers, and the standard attending of record–based approach may lack face validity in many cases.

We aimed to develop and operationalize a system to more fairly attribute patient-level data to individual providers across a single hospitalization even when multiple providers cared for the patient. We then compared our methodology to the standard approach, in which the attending of record receives full attribution for each metric, to determine the difference on a provider level between the 2 models.

Clinical Setting

The Johns Hopkins Hospital is a 1145-bed, tertiary-care hospital. Over the years of this project, the Johns Hopkins Hospitalist Program was an approximately 20-physician group providing care in a variety of settings, including a dedicated hospitalist floor, where this metrics program was initiated. Hospitalists in this setting work Monday through Friday, with 1 hospitalist and a moonlighter covering on the weekends. Admissions are performed by an admitter, and overnight care is provided by a nocturnist. Initially 17 beds, this unit expanded to 24 beds in June 2012. For the purposes of this article, we included all general medicine patients admitted to this floor between July 1, 2010, and June 30, 2014, who were cared for by hospitalists. During this period, all patients were inpatients; no patients were admitted under observation status. All of these patients were cared for by hospitalists without housestaff or advanced practitioners. Since 2014, the metrics program has been expanded to other hospitalist-run services in the hospital, but for simplicity, we have not presented these more recent data.

Individual Provider Metrics

Metrics were chosen to reflect institutional quality and efficiency priorities. Our choice of metrics was restricted to those that (1) plausibly reflect provider performance, at least in part, and (2) could be accessed in electronic form (without any manual chart review). Whenever possible, we chose metrics with objective data. Additionally, because funding for this effort was provided by the hospital, we sought to ensure that enough of the metrics were related to cost to justify ongoing hospital support of the project. SAS 9.2 (SAS Institute Inc, Cary, NC) was used to calculate metric weights. Specific metrics included American College of Chest Physicians (ACCP)–compliant venous thromboembolism (VTE) prophylaxis,⁷ observed-to-expected length of stay (LOS) ratio, percentage of discharges per day, discharges before 3 pm, depth of coding, patient satisfaction, readmissions, communication with the primary care provider, and time to signature for discharge summaries (Table 1).

Appropriate prophylaxis for VTE was calculated by using an algorithm embedded within the computerized provider order entry system, which assessed the prescription of ACCP-compliant VTE prophylaxis within 24 hours following admission. This included a risk assessment, and credit was given for no prophylaxis and/or mechanical and/or pharmacologic prophylaxis per the ACCP guidelines.⁷

Observed-to-expected LOS was defined by using the University HealthSystem Consortium (UHC; now Vizient Inc) expected LOS for the given calendar year. This approach incorporates patient diagnoses, demographics, and other administrative variables to define an expected LOS for each patient.

The percent of patients discharged per day was defined from billing data as the percentage of a provider’s evaluation and management charges that were the final charge of a patient’s stay (regardless of whether a discharge day service was coded).

Discharge prior to 3 pm was defined from administrative data as the time a patient was discharged from the electronic medical system.

Depth of coding was defined as the number of coded diagnoses submitted to the Maryland Health Services Cost Review Commission for determining payment and was viewed as an indicator of the thoroughness of provider documentation.

Patient satisfaction was defined at the patient level (for those patients who turned in patient satisfaction surveys) as the pooled value of the 5 provider questions on the hospital’s patient satisfaction survey administered by Press Ganey: “time the physician spent with you,” “did the physician show concern for your questions/worries,” “did the physician keep you informed,” “friendliness/courtesy of the physician,” and “skill of the physician.”⁸

Readmission rates were defined as same-hospital readmissions divided by the total number of patients discharged by a given provider, with exclusions based on the Centers for Medicare and Medicaid Services hospital-wide, all-cause readmission measure.¹ The expected same-hospital readmission rate was defined for each patient as the observed readmission rate in the entire UHC (Vizient) data set for all patients with the same All Patient Refined Diagnosis Related Group and severity of illness, as we have described previously.⁹

Communication with the primary care provider was the only self-reported metric used. It was based on a mandatory prompt on the discharge worksheet in the electronic medical record (EMR). Successful communication with the outpatient provider was defined as verbal or electronic communication by the hospitalist with the outpatient provider. Partial (50%) credit was given for providers who attempted but were unsuccessful in communicating with the outpatient provider, for patients for whom the provider had access to the Johns Hopkins EMR system, and for planned admissions without new or important information to convey. No credit was given for providers who indicated that communication was not indicated, who indicated that a patient and/or family would update the provider, or who indicated that the discharge summary would be sufficient.⁹ Because the discharge worksheet could be initiated at any time during the hospitalization, providers could document communication with the outpatient provider at any point during hospitalization.

Discharge summary turnaround was defined as the average number of days elapsed between the day of discharge and the signing of the discharge summary in the EMR.

Assigning Ownership of Patients to Individual Providers

Using billing data, we assigned ownership of patient care based on the type, timing, and number of charges that occurred during each hospitalization (Figure 1). Eligible charges included all history and physical (codes 99221, 99222, and 99223), subsequent care (codes 99231, 99232, and 99233), and discharge charges (codes 99238 and 99239).

By using a unique identifier assigned for each hospitalization, professional fees submitted by providers were used to identify which provider saw the patient on the admission day, discharge day, as well as subsequent care days. Providers’ productivity, bonus supplements, and policy compliance were determined by using billing data, which encouraged the prompt submittal of charges.

The provider who billed the admission history and physical (codes 99221, 99222, and 99223) within 1 calendar date of the patient’s initial admission was defined as the admitting provider. Patients transferred to the hospitalist service from other services were not assigned an admitting hospitalist. The sole metric assigned to the admitting hospitalist was ACCP-compliant VTE prophylaxis.

The provider who billed the final subsequent care or discharge code (codes 99231, 99232, 99233, 99238, and 99239) within 1 calendar date of discharge was defined as the discharging provider. For hospitalizations characterized by a single provider charge (eg, for patients admitted and discharged on the same day), the provider billing this charge was assigned as both the admitting and discharging physician. Patients upgraded to the intensive care unit (ICU) were not counted as a discharge unless the patient was downgraded and discharged from the hospitalist service. The discharging provider was assigned responsibility for the time of discharge, the percent of patients discharged per day, the discharge summary turnaround time, and hospital readmissions.

Metrics that were assigned to multiple providers for a single hospitalization were termed “provider day–weighted” metrics. The formula for calculating the weight for each provider day–weighted metric was as follows: weight for provider A = [number of daily charges billed by provider A] divided by [LOS +1]. The initial hospital day was counted as day 0. LOS plus 1 was used to recognize that a typical hospitalization will have a charge on the day of admission (day 0) and a charge on the day of discharge such that an LOS of 2 days (eg, a patient admitted on Monday and discharged on Wednesday) will have 3 daily charges. Provider day–weighted metrics included patient satisfaction, communication with the outpatient provider, depth of coding, and observed-to-expected LOS.

Our billing software prevented providers from the same group from billing multiple daily charges, thus ensuring that there were no duplicated charges submitted for a given day.

Presenting Results

Providers were only shown data from the day-weighted approach. For ease of visual interpretation, scores for each metric were scaled ordinally from 1 (worst performance) to 9 (best performance; Table 1). Data were displayed in a dashboard format on a password-protected website for each provider to view his or her own data relative to that of the hospitalist peer group. The dashboard was implemented in this format on July 1, 2011. Data were updated quarterly (Figure 2).

Results were displayed in a polyhedral or spider-web graph (Figure 2). Provider and group metrics were scaled according to predefined benchmarks established for each metric and standardized to a scale ranging from 1 to 9. The scale for each metric was set based on examining historical data and group median performance on the metrics to ensure that there was a range of performance (ie, to avoid having most hospitalists scoring a 1 or 9). Scaling thresholds were periodically adjusted as appropriate to maintain good visual discrimination. Higher scores (creating a larger-volume polygon) are desirable even for metrics such as LOS, for which a low value is desirable. Both a spider-web graph and trends over time were available to the provider (Figure 2). These graphs display a comparison of the individual provider scores for each metric to the hospitalist group average for that metric.

Comparison with the Standard (Attending of Record) Method of Attribution

For the purposes of this report, we sought to determine whether there were meaningful differences between our day-weighted approach versus the standard method of attribution, in which the attending of record is assigned responsibility for each metric that would not have been attributed to the discharging attending under both methods. Our goal was to determine where and whether there was a meaningful difference between the 2 methodologies, recognizing that the degree of difference between these 2 methodologies might vary in other institutions and settings. In our hospital, the attending of record is generally the discharging attending. In order to compare the 2 methodologies, we arbitrarily picked 2015 to retrospectively evaluate the differences between these 2 methods of attribution. We did not display or provide data using the standard methodology to providers at any point; this approach was used only for the purposes of this report. Because these metrics are intended to evaluate relative provider performance, we assigned a percentile to each provider for his or her performance on the given metric using our attribution methodology and then, similarly, assigned a percentile to each provider using the standard methodology. This yielded 2 percentile scores for each provider and each metric. We then compared these percentile ranks for providers in 2 ways: (1) we determined how often providers who scored in the top half of the group for a given metric (above the 50th percentile) also scored in the top half of the group for that metric by using the other calculation method, and (2) we calculated the absolute value of the difference in percentiles between the 2 methods to characterize the impact on a provider’s ranking for that metric that might result from switching to the other method. For instance, if a provider scored at the 20th percentile for the group in patient satisfaction with 1 attribution method and scored at the 40th percentile for the group in patient satisfaction using the other method, the absolute change in percentile would be 20 percentile points. But, this provider would still be below the 50th percentile by both methods (concordant bottom half performance). We did not perform this comparison for metrics assigned to the discharging provider (such as discharge summary turnaround time or readmissions) because the attending of record designation is assigned to the discharging provider at our hospital.

The dashboard was successfully operationalized on July 1, 2011, with displays visible to providers as shown in Figure 2. Consistent with the principles of providing effective performance feedback to providers, the display simultaneously showed providers their individual performance as well as the performance of their peers. Providers were able to view their spider-web plot for prior quarters. Not shown are additional views that allowed providers to see quarterly trends in their data versus their peers across several fiscal years. Also available to providers was their ranking relative to their peers for each metric; specific peers were deidentified in the display.

There was notable discordance between provider rankings between the 2 methodologies, as shown in Table 2. Provider performance above or below the median was concordant 56% to 75% of the time (depending on the particular metric), indicating substantial discordance because top-half or bottom-half concordance would be expected to occur by chance 50% of the time. Although the provider percentile differences between the 2 methods tended to be modest for most providers (the median difference between the methods was 13 to 22 percentile points for the various metrics), there were some providers for whom the method of calculation dramatically impacted their rankings. For 5 of the 6 metrics we examined, at least 1 provider had a 50-percentile or greater change in his or her ranking based on the method used. This indicates that at least some providers would have had markedly different scores relative to their peers had we used the alternative methodology (Table 2). In VTE prophylaxis, for example, at least 1 provider had a 94-percentile change in his or her ranking; similarly, a provider had an 88-perentile change in his or her LOS ranking between the 2 methodologies.

We found that it is possible to assign metrics across 1 hospital stay to multiple providers by using billing data. We also found a meaningful discrepancy in how well providers scored (relative to their peers) based on the method used for attribution. These results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

As hospitalist programs and providers in general are increasingly being asked to develop dashboards to monitor individual and group performance, correctly attributing care to providers is likely to become increasingly important. Experts agree that principles of effective provider performance dashboards include ranking individual provider performance relative to peers, clearly displaying data in an easily accessible format, and ensuring that data can be credibly attributed to the individual provider.^3,4,6 However, there appears to be no gold standard method for attribution, especially in the inpatient setting. Our results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

Several limitations of our findings are important to consider. First, our program is a relatively small, academic group with handoffs that typically occur every 1 to 2 weeks and sometimes with additional handoffs on weekends. Different care patterns and settings might impact the utility of our attribution methodology relative to the standard methodology. Additionally, it is important to note that the relative merits of the different methodologies cannot be ascertained from our comparison. We can demonstrate discordance between the attribution methodologies, but we cannot say that 1 method is correct and the other is flawed. Although we believe that our day-weighted approach feels fairer to providers based on group input and feedback, we did not conduct a formal survey to examine providers’ preferences for the standard versus day-weighted approaches. The appropriateness of a particular attribution method needs to be assessed locally and may vary based on the clinical setting. For instance, on a service in which patients are admitted for procedures, it may make more sense to attribute the outcome of the case to the proceduralist even if that provider did not bill for the patient’s care on a daily basis. Finally, the computational requirements of our methodology are not trivial and require linking billing data with administrative patient-level data, which may be challenging to operationalize in some institutions.

These limitations aside, we believe that our attribution methodology has face validity. For example, a provider might be justifiably frustrated if, using the standard methodology, he or she is charged with the LOS of a patient who had been hospitalized for months, particularly if that patient is discharged shortly after the provider assumes care. Our method addresses this type of misattribution. Particularly when individual provider compensation is based on performance on metrics (as is the case at our institution), optimizing provider attribution to particular patients may be important, and face validity may be required for group buy-in.

In summary, we have demonstrated that it is possible to use billing data to assign ownership of patients to multiple providers over 1 hospital stay. This could be applied to other hospitalist programs as well as other healthcare settings in which multiple providers care for patients during 1 healthcare encounter (eg, ICUs).

Disclosure

The authors declare they have no relevant conflicts of interest.

Hospitalists’ performance is routinely evaluated by third-party payers, employers, and patients. As hospitalist programs mature, there is a need to develop processes to identify, internally measure, and report on individual and group performance. We know from Society of Hospital Medicine (SHM) data that a significant amount of hospitalists’ total compensation is at least partially based on performance. Often this is based at least in part on quality data. In 2006, SHM issued a white paper detailing the key elements of a successful performance monitoring and reporting process.^1,2 Recommendations included the identification of meaningful operational and clinical performance metrics, and the ability to monitor and report both group and individual metrics was highlighted as an essential component. There is evidence that comparison of individual provider performance with that of their peers is a necessary element of successful provider dashboards.³ Additionally, regular feedback and a clear, visual presentation of the data are important components of successful provider feedback dashboards.^3-6

Much of the literature regarding provider feedback dashboards has been based in the outpatient setting. The majority of these dashboards focus on the management of chronic illnesses (eg, diabetes and hypertension), rates of preventative care services (eg, colonoscopy or mammogram), or avoidance of unnecessary care (eg, antibiotics for sinusitis).^4,5 Unlike in the outpatient setting, in which 1 provider often provides a majority of the care for a given episode of care, hospitalized patients are often cared for by multiple providers, challenging the appropriate attribution of patient-level metrics to specific providers. Under the standard approach, an entire hospitalization is attributed to 1 physician, generally the attending of record for the hospitalization, which may be the admitting provider or the discharging provider, depending on the approach used by the hospital. However, assigning responsibility for an entire hospitalization to a provider who may have only seen the patient for a small percentage of a hospitalization may jeopardize the validity of metrics. As provider metrics are increasingly being used for compensation, it is important to ensure that the method for attribution correctly identifies the providers caring for patients. To our knowledge there is no gold standard approach for attributing metrics to providers when patients are cared for by multiple providers, and the standard attending of record–based approach may lack face validity in many cases.

We aimed to develop and operationalize a system to more fairly attribute patient-level data to individual providers across a single hospitalization even when multiple providers cared for the patient. We then compared our methodology to the standard approach, in which the attending of record receives full attribution for each metric, to determine the difference on a provider level between the 2 models.

Clinical Setting

The Johns Hopkins Hospital is a 1145-bed, tertiary-care hospital. Over the years of this project, the Johns Hopkins Hospitalist Program was an approximately 20-physician group providing care in a variety of settings, including a dedicated hospitalist floor, where this metrics program was initiated. Hospitalists in this setting work Monday through Friday, with 1 hospitalist and a moonlighter covering on the weekends. Admissions are performed by an admitter, and overnight care is provided by a nocturnist. Initially 17 beds, this unit expanded to 24 beds in June 2012. For the purposes of this article, we included all general medicine patients admitted to this floor between July 1, 2010, and June 30, 2014, who were cared for by hospitalists. During this period, all patients were inpatients; no patients were admitted under observation status. All of these patients were cared for by hospitalists without housestaff or advanced practitioners. Since 2014, the metrics program has been expanded to other hospitalist-run services in the hospital, but for simplicity, we have not presented these more recent data.

Individual Provider Metrics

Metrics were chosen to reflect institutional quality and efficiency priorities. Our choice of metrics was restricted to those that (1) plausibly reflect provider performance, at least in part, and (2) could be accessed in electronic form (without any manual chart review). Whenever possible, we chose metrics with objective data. Additionally, because funding for this effort was provided by the hospital, we sought to ensure that enough of the metrics were related to cost to justify ongoing hospital support of the project. SAS 9.2 (SAS Institute Inc, Cary, NC) was used to calculate metric weights. Specific metrics included American College of Chest Physicians (ACCP)–compliant venous thromboembolism (VTE) prophylaxis,⁷ observed-to-expected length of stay (LOS) ratio, percentage of discharges per day, discharges before 3 pm, depth of coding, patient satisfaction, readmissions, communication with the primary care provider, and time to signature for discharge summaries (Table 1).

Appropriate prophylaxis for VTE was calculated by using an algorithm embedded within the computerized provider order entry system, which assessed the prescription of ACCP-compliant VTE prophylaxis within 24 hours following admission. This included a risk assessment, and credit was given for no prophylaxis and/or mechanical and/or pharmacologic prophylaxis per the ACCP guidelines.⁷

Observed-to-expected LOS was defined by using the University HealthSystem Consortium (UHC; now Vizient Inc) expected LOS for the given calendar year. This approach incorporates patient diagnoses, demographics, and other administrative variables to define an expected LOS for each patient.

The percent of patients discharged per day was defined from billing data as the percentage of a provider’s evaluation and management charges that were the final charge of a patient’s stay (regardless of whether a discharge day service was coded).

Discharge prior to 3 pm was defined from administrative data as the time a patient was discharged from the electronic medical system.

Depth of coding was defined as the number of coded diagnoses submitted to the Maryland Health Services Cost Review Commission for determining payment and was viewed as an indicator of the thoroughness of provider documentation.

Patient satisfaction was defined at the patient level (for those patients who turned in patient satisfaction surveys) as the pooled value of the 5 provider questions on the hospital’s patient satisfaction survey administered by Press Ganey: “time the physician spent with you,” “did the physician show concern for your questions/worries,” “did the physician keep you informed,” “friendliness/courtesy of the physician,” and “skill of the physician.”⁸

Readmission rates were defined as same-hospital readmissions divided by the total number of patients discharged by a given provider, with exclusions based on the Centers for Medicare and Medicaid Services hospital-wide, all-cause readmission measure.¹ The expected same-hospital readmission rate was defined for each patient as the observed readmission rate in the entire UHC (Vizient) data set for all patients with the same All Patient Refined Diagnosis Related Group and severity of illness, as we have described previously.⁹

Communication with the primary care provider was the only self-reported metric used. It was based on a mandatory prompt on the discharge worksheet in the electronic medical record (EMR). Successful communication with the outpatient provider was defined as verbal or electronic communication by the hospitalist with the outpatient provider. Partial (50%) credit was given for providers who attempted but were unsuccessful in communicating with the outpatient provider, for patients for whom the provider had access to the Johns Hopkins EMR system, and for planned admissions without new or important information to convey. No credit was given for providers who indicated that communication was not indicated, who indicated that a patient and/or family would update the provider, or who indicated that the discharge summary would be sufficient.⁹ Because the discharge worksheet could be initiated at any time during the hospitalization, providers could document communication with the outpatient provider at any point during hospitalization.

Discharge summary turnaround was defined as the average number of days elapsed between the day of discharge and the signing of the discharge summary in the EMR.

Assigning Ownership of Patients to Individual Providers

Using billing data, we assigned ownership of patient care based on the type, timing, and number of charges that occurred during each hospitalization (Figure 1). Eligible charges included all history and physical (codes 99221, 99222, and 99223), subsequent care (codes 99231, 99232, and 99233), and discharge charges (codes 99238 and 99239).

By using a unique identifier assigned for each hospitalization, professional fees submitted by providers were used to identify which provider saw the patient on the admission day, discharge day, as well as subsequent care days. Providers’ productivity, bonus supplements, and policy compliance were determined by using billing data, which encouraged the prompt submittal of charges.

The provider who billed the admission history and physical (codes 99221, 99222, and 99223) within 1 calendar date of the patient’s initial admission was defined as the admitting provider. Patients transferred to the hospitalist service from other services were not assigned an admitting hospitalist. The sole metric assigned to the admitting hospitalist was ACCP-compliant VTE prophylaxis.

The provider who billed the final subsequent care or discharge code (codes 99231, 99232, 99233, 99238, and 99239) within 1 calendar date of discharge was defined as the discharging provider. For hospitalizations characterized by a single provider charge (eg, for patients admitted and discharged on the same day), the provider billing this charge was assigned as both the admitting and discharging physician. Patients upgraded to the intensive care unit (ICU) were not counted as a discharge unless the patient was downgraded and discharged from the hospitalist service. The discharging provider was assigned responsibility for the time of discharge, the percent of patients discharged per day, the discharge summary turnaround time, and hospital readmissions.

Metrics that were assigned to multiple providers for a single hospitalization were termed “provider day–weighted” metrics. The formula for calculating the weight for each provider day–weighted metric was as follows: weight for provider A = [number of daily charges billed by provider A] divided by [LOS +1]. The initial hospital day was counted as day 0. LOS plus 1 was used to recognize that a typical hospitalization will have a charge on the day of admission (day 0) and a charge on the day of discharge such that an LOS of 2 days (eg, a patient admitted on Monday and discharged on Wednesday) will have 3 daily charges. Provider day–weighted metrics included patient satisfaction, communication with the outpatient provider, depth of coding, and observed-to-expected LOS.

Our billing software prevented providers from the same group from billing multiple daily charges, thus ensuring that there were no duplicated charges submitted for a given day.

Presenting Results

Providers were only shown data from the day-weighted approach. For ease of visual interpretation, scores for each metric were scaled ordinally from 1 (worst performance) to 9 (best performance; Table 1). Data were displayed in a dashboard format on a password-protected website for each provider to view his or her own data relative to that of the hospitalist peer group. The dashboard was implemented in this format on July 1, 2011. Data were updated quarterly (Figure 2).

Results were displayed in a polyhedral or spider-web graph (Figure 2). Provider and group metrics were scaled according to predefined benchmarks established for each metric and standardized to a scale ranging from 1 to 9. The scale for each metric was set based on examining historical data and group median performance on the metrics to ensure that there was a range of performance (ie, to avoid having most hospitalists scoring a 1 or 9). Scaling thresholds were periodically adjusted as appropriate to maintain good visual discrimination. Higher scores (creating a larger-volume polygon) are desirable even for metrics such as LOS, for which a low value is desirable. Both a spider-web graph and trends over time were available to the provider (Figure 2). These graphs display a comparison of the individual provider scores for each metric to the hospitalist group average for that metric.

Comparison with the Standard (Attending of Record) Method of Attribution

For the purposes of this report, we sought to determine whether there were meaningful differences between our day-weighted approach versus the standard method of attribution, in which the attending of record is assigned responsibility for each metric that would not have been attributed to the discharging attending under both methods. Our goal was to determine where and whether there was a meaningful difference between the 2 methodologies, recognizing that the degree of difference between these 2 methodologies might vary in other institutions and settings. In our hospital, the attending of record is generally the discharging attending. In order to compare the 2 methodologies, we arbitrarily picked 2015 to retrospectively evaluate the differences between these 2 methods of attribution. We did not display or provide data using the standard methodology to providers at any point; this approach was used only for the purposes of this report. Because these metrics are intended to evaluate relative provider performance, we assigned a percentile to each provider for his or her performance on the given metric using our attribution methodology and then, similarly, assigned a percentile to each provider using the standard methodology. This yielded 2 percentile scores for each provider and each metric. We then compared these percentile ranks for providers in 2 ways: (1) we determined how often providers who scored in the top half of the group for a given metric (above the 50th percentile) also scored in the top half of the group for that metric by using the other calculation method, and (2) we calculated the absolute value of the difference in percentiles between the 2 methods to characterize the impact on a provider’s ranking for that metric that might result from switching to the other method. For instance, if a provider scored at the 20th percentile for the group in patient satisfaction with 1 attribution method and scored at the 40th percentile for the group in patient satisfaction using the other method, the absolute change in percentile would be 20 percentile points. But, this provider would still be below the 50th percentile by both methods (concordant bottom half performance). We did not perform this comparison for metrics assigned to the discharging provider (such as discharge summary turnaround time or readmissions) because the attending of record designation is assigned to the discharging provider at our hospital.

The dashboard was successfully operationalized on July 1, 2011, with displays visible to providers as shown in Figure 2. Consistent with the principles of providing effective performance feedback to providers, the display simultaneously showed providers their individual performance as well as the performance of their peers. Providers were able to view their spider-web plot for prior quarters. Not shown are additional views that allowed providers to see quarterly trends in their data versus their peers across several fiscal years. Also available to providers was their ranking relative to their peers for each metric; specific peers were deidentified in the display.

There was notable discordance between provider rankings between the 2 methodologies, as shown in Table 2. Provider performance above or below the median was concordant 56% to 75% of the time (depending on the particular metric), indicating substantial discordance because top-half or bottom-half concordance would be expected to occur by chance 50% of the time. Although the provider percentile differences between the 2 methods tended to be modest for most providers (the median difference between the methods was 13 to 22 percentile points for the various metrics), there were some providers for whom the method of calculation dramatically impacted their rankings. For 5 of the 6 metrics we examined, at least 1 provider had a 50-percentile or greater change in his or her ranking based on the method used. This indicates that at least some providers would have had markedly different scores relative to their peers had we used the alternative methodology (Table 2). In VTE prophylaxis, for example, at least 1 provider had a 94-percentile change in his or her ranking; similarly, a provider had an 88-perentile change in his or her LOS ranking between the 2 methodologies.

We found that it is possible to assign metrics across 1 hospital stay to multiple providers by using billing data. We also found a meaningful discrepancy in how well providers scored (relative to their peers) based on the method used for attribution. These results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

As hospitalist programs and providers in general are increasingly being asked to develop dashboards to monitor individual and group performance, correctly attributing care to providers is likely to become increasingly important. Experts agree that principles of effective provider performance dashboards include ranking individual provider performance relative to peers, clearly displaying data in an easily accessible format, and ensuring that data can be credibly attributed to the individual provider.^3,4,6 However, there appears to be no gold standard method for attribution, especially in the inpatient setting. Our results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

Several limitations of our findings are important to consider. First, our program is a relatively small, academic group with handoffs that typically occur every 1 to 2 weeks and sometimes with additional handoffs on weekends. Different care patterns and settings might impact the utility of our attribution methodology relative to the standard methodology. Additionally, it is important to note that the relative merits of the different methodologies cannot be ascertained from our comparison. We can demonstrate discordance between the attribution methodologies, but we cannot say that 1 method is correct and the other is flawed. Although we believe that our day-weighted approach feels fairer to providers based on group input and feedback, we did not conduct a formal survey to examine providers’ preferences for the standard versus day-weighted approaches. The appropriateness of a particular attribution method needs to be assessed locally and may vary based on the clinical setting. For instance, on a service in which patients are admitted for procedures, it may make more sense to attribute the outcome of the case to the proceduralist even if that provider did not bill for the patient’s care on a daily basis. Finally, the computational requirements of our methodology are not trivial and require linking billing data with administrative patient-level data, which may be challenging to operationalize in some institutions.

These limitations aside, we believe that our attribution methodology has face validity. For example, a provider might be justifiably frustrated if, using the standard methodology, he or she is charged with the LOS of a patient who had been hospitalized for months, particularly if that patient is discharged shortly after the provider assumes care. Our method addresses this type of misattribution. Particularly when individual provider compensation is based on performance on metrics (as is the case at our institution), optimizing provider attribution to particular patients may be important, and face validity may be required for group buy-in.

In summary, we have demonstrated that it is possible to use billing data to assign ownership of patients to multiple providers over 1 hospital stay. This could be applied to other hospitalist programs as well as other healthcare settings in which multiple providers care for patients during 1 healthcare encounter (eg, ICUs).

Disclosure

The authors declare they have no relevant conflicts of interest.

Hospitalists’ performance is routinely evaluated by third-party payers, employers, and patients. As hospitalist programs mature, there is a need to develop processes to identify, internally measure, and report on individual and group performance. We know from Society of Hospital Medicine (SHM) data that a significant amount of hospitalists’ total compensation is at least partially based on performance. Often this is based at least in part on quality data. In 2006, SHM issued a white paper detailing the key elements of a successful performance monitoring and reporting process.^1,2 Recommendations included the identification of meaningful operational and clinical performance metrics, and the ability to monitor and report both group and individual metrics was highlighted as an essential component. There is evidence that comparison of individual provider performance with that of their peers is a necessary element of successful provider dashboards.³ Additionally, regular feedback and a clear, visual presentation of the data are important components of successful provider feedback dashboards.^3-6

Much of the literature regarding provider feedback dashboards has been based in the outpatient setting. The majority of these dashboards focus on the management of chronic illnesses (eg, diabetes and hypertension), rates of preventative care services (eg, colonoscopy or mammogram), or avoidance of unnecessary care (eg, antibiotics for sinusitis).^4,5 Unlike in the outpatient setting, in which 1 provider often provides a majority of the care for a given episode of care, hospitalized patients are often cared for by multiple providers, challenging the appropriate attribution of patient-level metrics to specific providers. Under the standard approach, an entire hospitalization is attributed to 1 physician, generally the attending of record for the hospitalization, which may be the admitting provider or the discharging provider, depending on the approach used by the hospital. However, assigning responsibility for an entire hospitalization to a provider who may have only seen the patient for a small percentage of a hospitalization may jeopardize the validity of metrics. As provider metrics are increasingly being used for compensation, it is important to ensure that the method for attribution correctly identifies the providers caring for patients. To our knowledge there is no gold standard approach for attributing metrics to providers when patients are cared for by multiple providers, and the standard attending of record–based approach may lack face validity in many cases.

We aimed to develop and operationalize a system to more fairly attribute patient-level data to individual providers across a single hospitalization even when multiple providers cared for the patient. We then compared our methodology to the standard approach, in which the attending of record receives full attribution for each metric, to determine the difference on a provider level between the 2 models.

Clinical Setting

The Johns Hopkins Hospital is a 1145-bed, tertiary-care hospital. Over the years of this project, the Johns Hopkins Hospitalist Program was an approximately 20-physician group providing care in a variety of settings, including a dedicated hospitalist floor, where this metrics program was initiated. Hospitalists in this setting work Monday through Friday, with 1 hospitalist and a moonlighter covering on the weekends. Admissions are performed by an admitter, and overnight care is provided by a nocturnist. Initially 17 beds, this unit expanded to 24 beds in June 2012. For the purposes of this article, we included all general medicine patients admitted to this floor between July 1, 2010, and June 30, 2014, who were cared for by hospitalists. During this period, all patients were inpatients; no patients were admitted under observation status. All of these patients were cared for by hospitalists without housestaff or advanced practitioners. Since 2014, the metrics program has been expanded to other hospitalist-run services in the hospital, but for simplicity, we have not presented these more recent data.

Individual Provider Metrics

Metrics were chosen to reflect institutional quality and efficiency priorities. Our choice of metrics was restricted to those that (1) plausibly reflect provider performance, at least in part, and (2) could be accessed in electronic form (without any manual chart review). Whenever possible, we chose metrics with objective data. Additionally, because funding for this effort was provided by the hospital, we sought to ensure that enough of the metrics were related to cost to justify ongoing hospital support of the project. SAS 9.2 (SAS Institute Inc, Cary, NC) was used to calculate metric weights. Specific metrics included American College of Chest Physicians (ACCP)–compliant venous thromboembolism (VTE) prophylaxis,⁷ observed-to-expected length of stay (LOS) ratio, percentage of discharges per day, discharges before 3 pm, depth of coding, patient satisfaction, readmissions, communication with the primary care provider, and time to signature for discharge summaries (Table 1).

Appropriate prophylaxis for VTE was calculated by using an algorithm embedded within the computerized provider order entry system, which assessed the prescription of ACCP-compliant VTE prophylaxis within 24 hours following admission. This included a risk assessment, and credit was given for no prophylaxis and/or mechanical and/or pharmacologic prophylaxis per the ACCP guidelines.⁷

Observed-to-expected LOS was defined by using the University HealthSystem Consortium (UHC; now Vizient Inc) expected LOS for the given calendar year. This approach incorporates patient diagnoses, demographics, and other administrative variables to define an expected LOS for each patient.

The percent of patients discharged per day was defined from billing data as the percentage of a provider’s evaluation and management charges that were the final charge of a patient’s stay (regardless of whether a discharge day service was coded).

Discharge prior to 3 pm was defined from administrative data as the time a patient was discharged from the electronic medical system.

Depth of coding was defined as the number of coded diagnoses submitted to the Maryland Health Services Cost Review Commission for determining payment and was viewed as an indicator of the thoroughness of provider documentation.

Patient satisfaction was defined at the patient level (for those patients who turned in patient satisfaction surveys) as the pooled value of the 5 provider questions on the hospital’s patient satisfaction survey administered by Press Ganey: “time the physician spent with you,” “did the physician show concern for your questions/worries,” “did the physician keep you informed,” “friendliness/courtesy of the physician,” and “skill of the physician.”⁸

Readmission rates were defined as same-hospital readmissions divided by the total number of patients discharged by a given provider, with exclusions based on the Centers for Medicare and Medicaid Services hospital-wide, all-cause readmission measure.¹ The expected same-hospital readmission rate was defined for each patient as the observed readmission rate in the entire UHC (Vizient) data set for all patients with the same All Patient Refined Diagnosis Related Group and severity of illness, as we have described previously.⁹

Communication with the primary care provider was the only self-reported metric used. It was based on a mandatory prompt on the discharge worksheet in the electronic medical record (EMR). Successful communication with the outpatient provider was defined as verbal or electronic communication by the hospitalist with the outpatient provider. Partial (50%) credit was given for providers who attempted but were unsuccessful in communicating with the outpatient provider, for patients for whom the provider had access to the Johns Hopkins EMR system, and for planned admissions without new or important information to convey. No credit was given for providers who indicated that communication was not indicated, who indicated that a patient and/or family would update the provider, or who indicated that the discharge summary would be sufficient.⁹ Because the discharge worksheet could be initiated at any time during the hospitalization, providers could document communication with the outpatient provider at any point during hospitalization.

Discharge summary turnaround was defined as the average number of days elapsed between the day of discharge and the signing of the discharge summary in the EMR.

Assigning Ownership of Patients to Individual Providers

Using billing data, we assigned ownership of patient care based on the type, timing, and number of charges that occurred during each hospitalization (Figure 1). Eligible charges included all history and physical (codes 99221, 99222, and 99223), subsequent care (codes 99231, 99232, and 99233), and discharge charges (codes 99238 and 99239).

By using a unique identifier assigned for each hospitalization, professional fees submitted by providers were used to identify which provider saw the patient on the admission day, discharge day, as well as subsequent care days. Providers’ productivity, bonus supplements, and policy compliance were determined by using billing data, which encouraged the prompt submittal of charges.

The provider who billed the admission history and physical (codes 99221, 99222, and 99223) within 1 calendar date of the patient’s initial admission was defined as the admitting provider. Patients transferred to the hospitalist service from other services were not assigned an admitting hospitalist. The sole metric assigned to the admitting hospitalist was ACCP-compliant VTE prophylaxis.

The provider who billed the final subsequent care or discharge code (codes 99231, 99232, 99233, 99238, and 99239) within 1 calendar date of discharge was defined as the discharging provider. For hospitalizations characterized by a single provider charge (eg, for patients admitted and discharged on the same day), the provider billing this charge was assigned as both the admitting and discharging physician. Patients upgraded to the intensive care unit (ICU) were not counted as a discharge unless the patient was downgraded and discharged from the hospitalist service. The discharging provider was assigned responsibility for the time of discharge, the percent of patients discharged per day, the discharge summary turnaround time, and hospital readmissions.

Metrics that were assigned to multiple providers for a single hospitalization were termed “provider day–weighted” metrics. The formula for calculating the weight for each provider day–weighted metric was as follows: weight for provider A = [number of daily charges billed by provider A] divided by [LOS +1]. The initial hospital day was counted as day 0. LOS plus 1 was used to recognize that a typical hospitalization will have a charge on the day of admission (day 0) and a charge on the day of discharge such that an LOS of 2 days (eg, a patient admitted on Monday and discharged on Wednesday) will have 3 daily charges. Provider day–weighted metrics included patient satisfaction, communication with the outpatient provider, depth of coding, and observed-to-expected LOS.

Our billing software prevented providers from the same group from billing multiple daily charges, thus ensuring that there were no duplicated charges submitted for a given day.

Presenting Results

Providers were only shown data from the day-weighted approach. For ease of visual interpretation, scores for each metric were scaled ordinally from 1 (worst performance) to 9 (best performance; Table 1). Data were displayed in a dashboard format on a password-protected website for each provider to view his or her own data relative to that of the hospitalist peer group. The dashboard was implemented in this format on July 1, 2011. Data were updated quarterly (Figure 2).

Results were displayed in a polyhedral or spider-web graph (Figure 2). Provider and group metrics were scaled according to predefined benchmarks established for each metric and standardized to a scale ranging from 1 to 9. The scale for each metric was set based on examining historical data and group median performance on the metrics to ensure that there was a range of performance (ie, to avoid having most hospitalists scoring a 1 or 9). Scaling thresholds were periodically adjusted as appropriate to maintain good visual discrimination. Higher scores (creating a larger-volume polygon) are desirable even for metrics such as LOS, for which a low value is desirable. Both a spider-web graph and trends over time were available to the provider (Figure 2). These graphs display a comparison of the individual provider scores for each metric to the hospitalist group average for that metric.

Comparison with the Standard (Attending of Record) Method of Attribution

For the purposes of this report, we sought to determine whether there were meaningful differences between our day-weighted approach versus the standard method of attribution, in which the attending of record is assigned responsibility for each metric that would not have been attributed to the discharging attending under both methods. Our goal was to determine where and whether there was a meaningful difference between the 2 methodologies, recognizing that the degree of difference between these 2 methodologies might vary in other institutions and settings. In our hospital, the attending of record is generally the discharging attending. In order to compare the 2 methodologies, we arbitrarily picked 2015 to retrospectively evaluate the differences between these 2 methods of attribution. We did not display or provide data using the standard methodology to providers at any point; this approach was used only for the purposes of this report. Because these metrics are intended to evaluate relative provider performance, we assigned a percentile to each provider for his or her performance on the given metric using our attribution methodology and then, similarly, assigned a percentile to each provider using the standard methodology. This yielded 2 percentile scores for each provider and each metric. We then compared these percentile ranks for providers in 2 ways: (1) we determined how often providers who scored in the top half of the group for a given metric (above the 50th percentile) also scored in the top half of the group for that metric by using the other calculation method, and (2) we calculated the absolute value of the difference in percentiles between the 2 methods to characterize the impact on a provider’s ranking for that metric that might result from switching to the other method. For instance, if a provider scored at the 20th percentile for the group in patient satisfaction with 1 attribution method and scored at the 40th percentile for the group in patient satisfaction using the other method, the absolute change in percentile would be 20 percentile points. But, this provider would still be below the 50th percentile by both methods (concordant bottom half performance). We did not perform this comparison for metrics assigned to the discharging provider (such as discharge summary turnaround time or readmissions) because the attending of record designation is assigned to the discharging provider at our hospital.

The dashboard was successfully operationalized on July 1, 2011, with displays visible to providers as shown in Figure 2. Consistent with the principles of providing effective performance feedback to providers, the display simultaneously showed providers their individual performance as well as the performance of their peers. Providers were able to view their spider-web plot for prior quarters. Not shown are additional views that allowed providers to see quarterly trends in their data versus their peers across several fiscal years. Also available to providers was their ranking relative to their peers for each metric; specific peers were deidentified in the display.

There was notable discordance between provider rankings between the 2 methodologies, as shown in Table 2. Provider performance above or below the median was concordant 56% to 75% of the time (depending on the particular metric), indicating substantial discordance because top-half or bottom-half concordance would be expected to occur by chance 50% of the time. Although the provider percentile differences between the 2 methods tended to be modest for most providers (the median difference between the methods was 13 to 22 percentile points for the various metrics), there were some providers for whom the method of calculation dramatically impacted their rankings. For 5 of the 6 metrics we examined, at least 1 provider had a 50-percentile or greater change in his or her ranking based on the method used. This indicates that at least some providers would have had markedly different scores relative to their peers had we used the alternative methodology (Table 2). In VTE prophylaxis, for example, at least 1 provider had a 94-percentile change in his or her ranking; similarly, a provider had an 88-perentile change in his or her LOS ranking between the 2 methodologies.

We found that it is possible to assign metrics across 1 hospital stay to multiple providers by using billing data. We also found a meaningful discrepancy in how well providers scored (relative to their peers) based on the method used for attribution. These results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

As hospitalist programs and providers in general are increasingly being asked to develop dashboards to monitor individual and group performance, correctly attributing care to providers is likely to become increasingly important. Experts agree that principles of effective provider performance dashboards include ranking individual provider performance relative to peers, clearly displaying data in an easily accessible format, and ensuring that data can be credibly attributed to the individual provider.^3,4,6 However, there appears to be no gold standard method for attribution, especially in the inpatient setting. Our results imply that hospitals should consider attributing performance metrics based on ascribed ownership from billing data and not just from attending of record status.

Several limitations of our findings are important to consider. First, our program is a relatively small, academic group with handoffs that typically occur every 1 to 2 weeks and sometimes with additional handoffs on weekends. Different care patterns and settings might impact the utility of our attribution methodology relative to the standard methodology. Additionally, it is important to note that the relative merits of the different methodologies cannot be ascertained from our comparison. We can demonstrate discordance between the attribution methodologies, but we cannot say that 1 method is correct and the other is flawed. Although we believe that our day-weighted approach feels fairer to providers based on group input and feedback, we did not conduct a formal survey to examine providers’ preferences for the standard versus day-weighted approaches. The appropriateness of a particular attribution method needs to be assessed locally and may vary based on the clinical setting. For instance, on a service in which patients are admitted for procedures, it may make more sense to attribute the outcome of the case to the proceduralist even if that provider did not bill for the patient’s care on a daily basis. Finally, the computational requirements of our methodology are not trivial and require linking billing data with administrative patient-level data, which may be challenging to operationalize in some institutions.

These limitations aside, we believe that our attribution methodology has face validity. For example, a provider might be justifiably frustrated if, using the standard methodology, he or she is charged with the LOS of a patient who had been hospitalized for months, particularly if that patient is discharged shortly after the provider assumes care. Our method addresses this type of misattribution. Particularly when individual provider compensation is based on performance on metrics (as is the case at our institution), optimizing provider attribution to particular patients may be important, and face validity may be required for group buy-in.

In summary, we have demonstrated that it is possible to use billing data to assign ownership of patients to multiple providers over 1 hospital stay. This could be applied to other hospitalist programs as well as other healthcare settings in which multiple providers care for patients during 1 healthcare encounter (eg, ICUs).

Disclosure

The authors declare they have no relevant conflicts of interest.

A previously healthy 30-year-old woman presented to the emergency department with 2 weeks of weakness.

True muscle weakness must be distinguished from the more common causes of asthenia. Many systemic disorders produce fatigue, with resulting functional limitation that is often interpreted by patients as weakness. Initial history should focus on conditions producing fatigue, such as cardiopulmonary disease, anemia, connective tissue disease, depression or cachexia related to malignancy, infection, or other inflammatory states. Careful questioning may reveal evidence of dyspnea, poor exercise tolerance, or joint pain as an alternative to actual loss of muscle power. If true weakness is still suspected, attention should be focused on the pattern, onset, anatomic site, and progression of weakness. Muscle weakness is often characterized by difficulty with specific tasks, such as climbing stairs, rising from a chair, raising a hand, or using cutlery. The physical examination is critical in determining whether weakness is due to true loss of motor power. The differential diagnosis of weakness is broad and includes neurologic, infectious, endocrine, inflammatory, genetic, metabolic, and drug-induced etiologies.

She initially experienced 3 days of mild cramps and soreness in her thighs. She then developed weakness that began in her thighs and progressed to involve her lower legs and upper and lower arms. She had difficulty combing her hair. She required the use of her arms to get up from a chair. She grasped onto objects to aid in ambulation around the house. In addition, she described 1 year of moderate fatigue but no fever, weight loss, dyspnea, dysphagia, visual changes, paresthesias, bowel or bladder incontinence, back pain, or preceding gastrointestinal or respiratory illness. She had experienced diffuse intermittent hives, most prominent in her chest and upper arms, for the past several weeks.

History certainly supports true weakness but will need to be confirmed on examination. The distribution began as proximal but now appears diffuse. The presence of myalgia and cramping raises the possibility of noninflammatory myopathies, which are usually more insidious in onset. A severe electrolyte disturbance would be possible, based on the diffuse nature of weakness that was preceded by cramping. The distribution of weakness and lack of bowel or bladder incontinence is reassuring and suggests against a spinal cord disorder; however, a high index of suspicion must be maintained for myelopathy because delayed treatment might result in irreversible paralysis.

The patient’s course also includes hives. Common causes of hives include infections and allergic reactions to medications, foods, and insect stings. Urticaria may also result from systemic disorders, such as vasculitis, lupus, lymphoma, mastocytosis, and paraproteinemias, which can be associated with weakness and fatigue. Although severe weakness in combination with hives makes an infectious and allergic reaction less likely, we still seek to ascertain if the evolving chief complaints of weakness and hives are the result of a single unifying and evolving multisystem disorder or are distinct and unrelated processes.

Her past medical history included fibromyalgia, kidney stones, and gastroesophageal reflux disease. One week prior to presentation, she was prescribed prednisone 60 mg daily for the treatment of hives; the dose had been tapered to 40 mg at presentation, with mild improvement of hives. She recently started doxepin for fibromyalgia and insomnia. She lived at home with her husband and 8-year-old child. She worked as a clerk in a pest control office and denied any pesticide exposure. She denied tobacco, alcohol, or illicit drug use. Her family history included systemic lupus erythematosus (SLE) in her mother and maternal aunt.

Glucocorticoids are associated with myopathy; however, the weakness preceded steroid therapy. Thus, unless there was unknown exposure to high-dose steroid medication to treat recurrent episodes of urticaria earlier in her course, glucocorticoid-related myopathy is unlikely. Fibromyalgia might cause the perception of weakness from pain. However, the history of difficulty combing her hair and rising from a chair suggests actual loss of motor power. The side effects of her medications, such as newly started doxepin, must be reviewed. A family history of SLE raises concern for rheumatologic conditions; however, one might expect improvement with steroid therapy.

On physical examination, her temperature was 36.9 °C, blood pressure 126/93 mmHg, pulse 81 beats per minute, respiratory rate 16 breaths per minute, and oxygen saturation 100% on ambient air. Her cardiopulmonary examination was normal. Her abdomen was nontender and without hepatosplenomegaly. Her strength was 2 out of 5 in proximal and distal legs, bilaterally, and 4 out of 5 in proximal and distal upper extremities. She had normal muscle tone without fasciculations or atrophy. Her joints were without edema, erythema, or impaired range of motion. She had normal sensation to light touch in arms and legs. Her reflexes were 2+ in the patellar, Achilles, and brachioradialis tendons. She had no lymphadenopathy, mucosal ulcerations, or alopecia. A skin examination revealed smooth, slightly elevated, and faded pink wheals that were diffuse but most prominent in upper arms and chest.

Physical examination confirms the presence of true muscle weakness. The differential diagnosis is narrowed by several findings, both positive and negative, elicited in the examination. The diffuse nature of the weakness eliminates focal central nervous system lesions, such as stroke, intracranial mass lesions, or demyelinating white matter foci. Combining this finding with normal reflexes and history of preceding myalgias makes electrolyte-induced and inflammatory (eg, polymyositis) myopathies more likely. The normal deep tendon reflexes and the absence of a delayed relaxation phase lower the likelihood of hypothyroidism. 

Diseases originating from the neuromuscular junction, such as myasthenia gravis, may also present with weakness and normal reflexes, although this pattern of weakness would be atypical; myasthenia gravis classically presents with fatigable weakness and ocular findings of diplopia and/or ptosis. First-tier testing should include a complete blood count to evaluate for eosinophilia, comprehensive metabolic panel, and urinalysis for myoglobinuria, thyroid stimulating hormone, and muscle enzymes. 


Results of a complete blood count demonstrated a leukocyte count of 16.1 k/uL with 82% neutrophils, 13% lymphocytes, 5% monocytes, and 0% eosinophils. Hemoglobin was 13.2 g/dL, and platelet count 226 k/uL. Sodium was 136 mmol/L, potassium 1.5 mmol/L, chloride 115 mmol/L, bicarbonate 12 mmol/L, blood urea nitrogen 26 mg/dL, creatinine 1.0 mg/dL (baseline creatinine: 0.6), and glucose 102 mg/dL. Calcium was 9.4 mg/dL, magnesium 2.6 mg/dL, phosphorus 1.8 mg/dL, CK 501 U/L (normal: 40-230), and TSH 5.48 uIU/mL (normal: 0.5-4). Aspartate aminotransferase was 64 U/L, alanine aminotransferase 23 U/L, alkaline phosphatase 66 U/L, bilirubin 0.9 mg/dL, albumin 3.8 g/dL, and total protein 8.7 g/dL (normal: 6.2-7.8). Human immunodeficiency virus antibody screen was negative. An electrocardiogram revealed normal sinus rhythm, flattened T waves, and prominent U waves.

Potassium losses are classically categorized into 1 of 3 groups: renal losses, gastrointestinal losses, or transcellular shifts. Without a clear history of diuretic use, renal losses may not be apparent on history and examination. In contrast, gastrointestinal losses are almost always evidenced by a history of vomiting and/or diarrhea, with rare exceptions, including unreported laxative abuse or surreptitious vomiting. Transcellular potassium shifts can be seen in states of increased insulin or beta-adrenergic activity and alkalosis and result from both primary and secondary causes of hypokalemic periodic paralysis.

The presence of a reduced serum bicarbonate and elevated chloride concentration suggests a normal anion gap metabolic acidosis. Many conditions associated with normal anion gap metabolic acidosis are evident by history, such as diarrhea. In enigmatic cases such as this, it will be important to take a stepwise approach that includes an evaluation for urinary potassium losses and assessment of acid-base status. An unexplained normal anion gap metabolic acidosis combined with hypokalemia raises suspicion for a distal renal tubular acidosis (RTA). Additional testing to evaluate for a possible RTA should include the assessment of urinary electrolytes and urinary pH. The hypokalemia explains her weakness, but the etiology of such profound hypokalemia is not evident, nor is it clear how it relates to her hives.

The severity of the hypokalemia, combined with electrocardiogram changes, necessitates rapid intravenous potassium repletion, telemetry monitoring, and frequent serum potassium measurement. Treatment of her metabolic acidosis is more nuanced and depends upon both the severity of disturbance and the suspicion of whether the etiology is transcellular shift, potassium depletion, or both.

Urine studies demonstrated a urine specific gravity of 1.006 (normal: 1.001-1.030), urine pH was 6.5 (normal: 5-6.5), trace leukocyte esterase, negative nitrite, 30 mg/dL of protein (normal: <15), sodium 64 mmol/L (normal: 40-220), potassium 17 mmol/L (normal: 25-125), and chloride 71 mmol/L (normal: 110-250). Urine microscopy demonstrated 3 red blood cells per high power field (normal: 0-1), 4 white blood cells per high power field (normal: 0-4), 4+ bacteria per high power field, and no red blood cell casts. Urine protein-to-creatinine ratio was 1.6. C3 and C4 complement levels were 53 mg/dL (normal: 80-165) and 12 mg/dL (normal: 15-49), respectively. C-reactive protein was <0.5 (normal: 0-0.9), and erythrocyte sedimentation rate was 16 mm/hour (normal: 0-20).

A calculation of the urine anion gap (UAG; [urine sodium + urine potassium] – urine chloride) yields a UAG of 10 mq/L. A positive UAG, together with a nongap metabolic acidosis, should prompt the consideration of RTA. The normal renal response to acidosis is to reduce the urine pH to less than 5.3 through an increase in hydrogen ion excretion in the form of ammonium. A urine pH of 6.5 is highly suggestive of type 1 (distal) RTA and its associated impairment of distal acidification. Treatment with sodium bicarbonate to correct the acidosis and associated complications is warranted.

A distal RTA would account for her past medical history of renal stones. Acidemia promotes both increased calcium phosphate release from bone (with subsequent hypercalciuria) and enhanced citrate reabsorption in the proximal renal tubules, leading to decreased urinary citrate. Citrate inhibits calcium stone formation. The increased calcium load to renal tubules in addition to decreased urinary citrate both lead to increased precipitation of calcium stones in the genitourinary tract.

A diagnosis of distal RTA should prompt evaluation for specific etiologies, such as Sjögren’s syndrome or SLE. While not diagnostic of any specific condition, low C3 and C4 levels suggest immune complex formation with related complement consumption, contributing to hypocomplementemia. The diagnosis of RTA may occur among patients with Sjögren’s syndrome in the absence of overt evidence of sicca syndrome (xerostomia and keratoconjunctivitis sicca). Other etiologies of distal RTA include conditions leading to hypercalciuria, such as hyperparathyroidism and idiopathic hypercalciuria, hereditary causes, toxins such as toluene, and drugs such as amphotericin B, lithium carbonate, and ibuprofen.

Her antinuclear antibody titer was >1:1280 (normal: <80). Anti-SSA and -SSB antibodies were both positive, with a titer >100 (normal: <20). Rheumatoid factor was positive at 22 IU/mL (normal: 0-14). Anti-smith, anti-double stranded DNA, and anti-ribonucleoprotein antibodies were negative.

Sjögren’s syndrome appears to be the ultimate etiology of this patient’s distal RTA. The diagnosis of Sjögren’s is more classically made in the presence of lacrimal and/or salivary dysfunction and confirmed with compatible autoantibodies. In the absence of dry eyes or dry mouth, attention should be focused on her skin findings. Cutaneous vasculitis does occur in a small percentage of Sjögren’s syndrome cases. Urticarial lesions have been reported in this subset, and skin biopsy would further support the diagnosis.

Treatment of Sjögren’s syndrome with immunosuppressive therapy may ameliorate renal parenchymal pathology and improve her profound metabolic disturbances.

On further questioning, she described several months of mild xerostomia, which resulted in increased consumption of fluids. She did not have keratoconjunctivitis sicca. Biopsy of her urticarial rash demonstrated a leukocytoclastic vasculitis with eosinophilic infiltration (Figure 1). Renal biopsy with hematoxylin and eosin staining, immunofluorescence, and electron microscopy demonstrated an immune complex-mediated glomerulonephritis and moderate tubulointerstitial nephritis (Figure 2). A diagnosis of Sjögren’s syndrome was made based on the patient’s xerostomia, high titers of antinuclear antibodies, SSA and SSB antibodies, positive rheumatoid factor, hypocomplementemia, and systemic manifestations associated with Sjögren’s syndrome, including distal RTA, nephrolithiasis, and hives, with histologic evidence of leukocytoclastic vasculitis.

She received aggressive potassium and bicarbonate repletion and, several days later, had normalization of both. Her weakness and myalgia rapidly improved concomitantly with the correction of her hypokalemia. Five days later she was ambulating independently and discharged with potassium citrate and prednisone therapy. She had improved fatigue and rash at a 1-month follow-up with rheumatology. As an outpatient, she was started on azathioprine and slowly tapered off her steroids. Over the next several months, she had normal potassium, bicarbonate, and renal function, although she did require lithotripsy for an obstructive renal stone.

RTA should be considered in the differential diagnosis of an unexplained normal anion gap metabolic acidosis. There are 3 major types of RTAs, with different characteristics. In type 1 (distal) RTA, the primary defect is impaired distal acidification of the urine. Distal RTA commonly presents with hypokalemia, calciuria (often presenting as renal stones), and a positive UAG.¹ In type 2 (proximal) RTA, the primary defect is impaired bicarbonate reabsorption, leading to bicarbonate wasting in the urine. Proximal RTAs can be secondary to an isolated defect in bicarbonate reabsorption or generalized proximal renal tubule dysfunction (Fanconi syndrome).¹ A type 4 RTA is characterized by hypoaldosteronism, presenting usually with a mild nonanion gap metabolic acidosis and hyperkalemia. This patient’s history of renal stones, hypokalemia, and positive UAG supported a type 1 (distal) RTA. Distal RTA is often idiopathic, but initial evaluation should include a review of medications and investigation into an underlying systemic disorder (eg, plasma cell dyscrasia or autoimmune disease). This would include eliciting a possible history of xerostomia and xerophthalmia, together with testing of SSA (Ro) and SSB (La) antibodies, to assess for Sjögren’s syndrome. In addition, checking serum calcium to assess for hyperparathyroidism or familial idiopathic hypercalciuria and a review of medications, such as lithium and amphotericin,¹ may uncover other secondary causes of distal RTA.

While Sjögren’s syndrome primarily affects salivary and lacrimal glands, leading to dry mouth and dry eyes, respectively, extraglandular manifestations are common, with fatigue and arthralgia occurring in half of patients. Extra-glandular involvement also often includes the skin and kidneys but can affect several other organ systems, including the central nervous system, heart, lungs, bone marrow, and lymph nodes.²

There are many cutaneous manifestations of Sjögren’s syndrome.³ Xerosis, or xeroderma, is the most common and is characterized by dry, scaly skin. Cutaneous vasculitis can occur in 10% of patients with Sjögren’s syndrome and often presents with palpable purpura or diffuse urticarial lesions, as in our patient.⁴ Erythematous maculopapules and cryoglobulinemic vasculitis may also occur.⁴ A less common skin manifestation is annular erythema, presenting as an indurated, ring-like lesion.⁵

Chronic tubulointerstitial nephritis is the most common renal manifestation of Sjögren’s syndrome.⁶ This often pre-sents with a mild elevated serum creatinine and a distal RTA, leading to hypokalemia, as in the case discussed. Distal RTA is well described, occurring in one-quarter of patients with Sjögren’s syndrome.⁷ The pathophysiology leading to distal RTA in Sjögren’s syndrome is thought to arise from autoimmune injury to the H(+)-ATPase pump in the renal collecting tubules, leading to decreased distal proton secretion.^8,9 Younger adults with Sjögren’s syndrome, in the third and fourth decades of life, have a predilection to develop tubulointerstitial inflammation, distal RTA, and nephrolithiasis, as in the present case.⁶ Sjögren’s syndrome less commonly presents with membranoproliferative glomerulonephritis or membranous nephropathy.^10,11 Cryoglobulinemia-associated hypocomplementemia and glomerulonephritis may also occur with Sjögren’s syndrome, yet glomerular lesions are less common than is tubulointerstitial inflammation. The patient discussed had proteinuria and evidence of immune complex-mediated glomerulonephritis.

Treatment of sicca symptoms is generally supportive. It includes artificial tears, encouragement of good hydration, salivary stimulants, and maintaining good oral hygiene. Pilocarpine, a cholinergic parasympathomimetic agent, is approved by the Food and Drug Administration to treat dry mouth associated with Sjögren’s syndrome. The treatment of extraglandular manifestations depends on the organ(s) involved. More severe presentations, such as vasculitis and glomerulonephritis, often require immunosuppressive therapy with systemic glucocorticoids, cyclophosphamide, azathioprine, or other immunosuppressive agents,¹² including rituximab. RTA often necessitates treatment with oral bicarbonate and supplemental potassium repletion.

The base rate of disease (ie, prevalence of disease) influences a diagnostician’s pretest probability of a given diagnosis. The discussant briefly considered rare causes of hives (eg, vasculitis) but appropriately fine-tuned their differential for the patient’s hypokalemia and RTA. Once the diagnosis of Sjögren’s syndrome was made with certainty, the clinician was able to revisit the patient’s rash with a new lens. Urticarial vasculitis suddenly became a plausible consideration, despite its rarity (compared to allergic causes of hives) because of the direct link to the underlying autoimmune condition, which affected both the proximal muscles and distal nephrons.

• Evaluation of patients with weakness starts with determining true muscle weakness (ie, pathology involving the brain, spinal cord, peripheral nerve, neuromuscular junction, and/or muscle) from asthenia.
• Distal RTA should be considered in patients with a nonanion gap metabolic acidosis and hypokalemia.
• Sjögren’s syndrome has many extraglandular clinical manifestations, including vasculitis, urticaria, tubulointerstitial renal inflammation, glomerulonephritis, and lymphoma.

Acknowledgment

The authors thank Virgilius Cornea, MD, for his interpretation of the pathologic images.

Disclosure

Dr. Manesh is supported by the Jeremiah A. Barondess Fellowship in the Clinical Transaction of the New York Academy of Medicine, in collaboration with the Accreditation Council for Graduate Medical Education (ACGME). The authors declare no conflicts of interests.

A previously healthy 30-year-old woman presented to the emergency department with 2 weeks of weakness.

True muscle weakness must be distinguished from the more common causes of asthenia. Many systemic disorders produce fatigue, with resulting functional limitation that is often interpreted by patients as weakness. Initial history should focus on conditions producing fatigue, such as cardiopulmonary disease, anemia, connective tissue disease, depression or cachexia related to malignancy, infection, or other inflammatory states. Careful questioning may reveal evidence of dyspnea, poor exercise tolerance, or joint pain as an alternative to actual loss of muscle power. If true weakness is still suspected, attention should be focused on the pattern, onset, anatomic site, and progression of weakness. Muscle weakness is often characterized by difficulty with specific tasks, such as climbing stairs, rising from a chair, raising a hand, or using cutlery. The physical examination is critical in determining whether weakness is due to true loss of motor power. The differential diagnosis of weakness is broad and includes neurologic, infectious, endocrine, inflammatory, genetic, metabolic, and drug-induced etiologies.

She initially experienced 3 days of mild cramps and soreness in her thighs. She then developed weakness that began in her thighs and progressed to involve her lower legs and upper and lower arms. She had difficulty combing her hair. She required the use of her arms to get up from a chair. She grasped onto objects to aid in ambulation around the house. In addition, she described 1 year of moderate fatigue but no fever, weight loss, dyspnea, dysphagia, visual changes, paresthesias, bowel or bladder incontinence, back pain, or preceding gastrointestinal or respiratory illness. She had experienced diffuse intermittent hives, most prominent in her chest and upper arms, for the past several weeks.

History certainly supports true weakness but will need to be confirmed on examination. The distribution began as proximal but now appears diffuse. The presence of myalgia and cramping raises the possibility of noninflammatory myopathies, which are usually more insidious in onset. A severe electrolyte disturbance would be possible, based on the diffuse nature of weakness that was preceded by cramping. The distribution of weakness and lack of bowel or bladder incontinence is reassuring and suggests against a spinal cord disorder; however, a high index of suspicion must be maintained for myelopathy because delayed treatment might result in irreversible paralysis.

The patient’s course also includes hives. Common causes of hives include infections and allergic reactions to medications, foods, and insect stings. Urticaria may also result from systemic disorders, such as vasculitis, lupus, lymphoma, mastocytosis, and paraproteinemias, which can be associated with weakness and fatigue. Although severe weakness in combination with hives makes an infectious and allergic reaction less likely, we still seek to ascertain if the evolving chief complaints of weakness and hives are the result of a single unifying and evolving multisystem disorder or are distinct and unrelated processes.

Her past medical history included fibromyalgia, kidney stones, and gastroesophageal reflux disease. One week prior to presentation, she was prescribed prednisone 60 mg daily for the treatment of hives; the dose had been tapered to 40 mg at presentation, with mild improvement of hives. She recently started doxepin for fibromyalgia and insomnia. She lived at home with her husband and 8-year-old child. She worked as a clerk in a pest control office and denied any pesticide exposure. She denied tobacco, alcohol, or illicit drug use. Her family history included systemic lupus erythematosus (SLE) in her mother and maternal aunt.

Glucocorticoids are associated with myopathy; however, the weakness preceded steroid therapy. Thus, unless there was unknown exposure to high-dose steroid medication to treat recurrent episodes of urticaria earlier in her course, glucocorticoid-related myopathy is unlikely. Fibromyalgia might cause the perception of weakness from pain. However, the history of difficulty combing her hair and rising from a chair suggests actual loss of motor power. The side effects of her medications, such as newly started doxepin, must be reviewed. A family history of SLE raises concern for rheumatologic conditions; however, one might expect improvement with steroid therapy.

On physical examination, her temperature was 36.9 °C, blood pressure 126/93 mmHg, pulse 81 beats per minute, respiratory rate 16 breaths per minute, and oxygen saturation 100% on ambient air. Her cardiopulmonary examination was normal. Her abdomen was nontender and without hepatosplenomegaly. Her strength was 2 out of 5 in proximal and distal legs, bilaterally, and 4 out of 5 in proximal and distal upper extremities. She had normal muscle tone without fasciculations or atrophy. Her joints were without edema, erythema, or impaired range of motion. She had normal sensation to light touch in arms and legs. Her reflexes were 2+ in the patellar, Achilles, and brachioradialis tendons. She had no lymphadenopathy, mucosal ulcerations, or alopecia. A skin examination revealed smooth, slightly elevated, and faded pink wheals that were diffuse but most prominent in upper arms and chest.

Physical examination confirms the presence of true muscle weakness. The differential diagnosis is narrowed by several findings, both positive and negative, elicited in the examination. The diffuse nature of the weakness eliminates focal central nervous system lesions, such as stroke, intracranial mass lesions, or demyelinating white matter foci. Combining this finding with normal reflexes and history of preceding myalgias makes electrolyte-induced and inflammatory (eg, polymyositis) myopathies more likely. The normal deep tendon reflexes and the absence of a delayed relaxation phase lower the likelihood of hypothyroidism. 

Diseases originating from the neuromuscular junction, such as myasthenia gravis, may also present with weakness and normal reflexes, although this pattern of weakness would be atypical; myasthenia gravis classically presents with fatigable weakness and ocular findings of diplopia and/or ptosis. First-tier testing should include a complete blood count to evaluate for eosinophilia, comprehensive metabolic panel, and urinalysis for myoglobinuria, thyroid stimulating hormone, and muscle enzymes. 


Results of a complete blood count demonstrated a leukocyte count of 16.1 k/uL with 82% neutrophils, 13% lymphocytes, 5% monocytes, and 0% eosinophils. Hemoglobin was 13.2 g/dL, and platelet count 226 k/uL. Sodium was 136 mmol/L, potassium 1.5 mmol/L, chloride 115 mmol/L, bicarbonate 12 mmol/L, blood urea nitrogen 26 mg/dL, creatinine 1.0 mg/dL (baseline creatinine: 0.6), and glucose 102 mg/dL. Calcium was 9.4 mg/dL, magnesium 2.6 mg/dL, phosphorus 1.8 mg/dL, CK 501 U/L (normal: 40-230), and TSH 5.48 uIU/mL (normal: 0.5-4). Aspartate aminotransferase was 64 U/L, alanine aminotransferase 23 U/L, alkaline phosphatase 66 U/L, bilirubin 0.9 mg/dL, albumin 3.8 g/dL, and total protein 8.7 g/dL (normal: 6.2-7.8). Human immunodeficiency virus antibody screen was negative. An electrocardiogram revealed normal sinus rhythm, flattened T waves, and prominent U waves.

Potassium losses are classically categorized into 1 of 3 groups: renal losses, gastrointestinal losses, or transcellular shifts. Without a clear history of diuretic use, renal losses may not be apparent on history and examination. In contrast, gastrointestinal losses are almost always evidenced by a history of vomiting and/or diarrhea, with rare exceptions, including unreported laxative abuse or surreptitious vomiting. Transcellular potassium shifts can be seen in states of increased insulin or beta-adrenergic activity and alkalosis and result from both primary and secondary causes of hypokalemic periodic paralysis.

The presence of a reduced serum bicarbonate and elevated chloride concentration suggests a normal anion gap metabolic acidosis. Many conditions associated with normal anion gap metabolic acidosis are evident by history, such as diarrhea. In enigmatic cases such as this, it will be important to take a stepwise approach that includes an evaluation for urinary potassium losses and assessment of acid-base status. An unexplained normal anion gap metabolic acidosis combined with hypokalemia raises suspicion for a distal renal tubular acidosis (RTA). Additional testing to evaluate for a possible RTA should include the assessment of urinary electrolytes and urinary pH. The hypokalemia explains her weakness, but the etiology of such profound hypokalemia is not evident, nor is it clear how it relates to her hives.

The severity of the hypokalemia, combined with electrocardiogram changes, necessitates rapid intravenous potassium repletion, telemetry monitoring, and frequent serum potassium measurement. Treatment of her metabolic acidosis is more nuanced and depends upon both the severity of disturbance and the suspicion of whether the etiology is transcellular shift, potassium depletion, or both.

Urine studies demonstrated a urine specific gravity of 1.006 (normal: 1.001-1.030), urine pH was 6.5 (normal: 5-6.5), trace leukocyte esterase, negative nitrite, 30 mg/dL of protein (normal: <15), sodium 64 mmol/L (normal: 40-220), potassium 17 mmol/L (normal: 25-125), and chloride 71 mmol/L (normal: 110-250). Urine microscopy demonstrated 3 red blood cells per high power field (normal: 0-1), 4 white blood cells per high power field (normal: 0-4), 4+ bacteria per high power field, and no red blood cell casts. Urine protein-to-creatinine ratio was 1.6. C3 and C4 complement levels were 53 mg/dL (normal: 80-165) and 12 mg/dL (normal: 15-49), respectively. C-reactive protein was <0.5 (normal: 0-0.9), and erythrocyte sedimentation rate was 16 mm/hour (normal: 0-20).

A calculation of the urine anion gap (UAG; [urine sodium + urine potassium] – urine chloride) yields a UAG of 10 mq/L. A positive UAG, together with a nongap metabolic acidosis, should prompt the consideration of RTA. The normal renal response to acidosis is to reduce the urine pH to less than 5.3 through an increase in hydrogen ion excretion in the form of ammonium. A urine pH of 6.5 is highly suggestive of type 1 (distal) RTA and its associated impairment of distal acidification. Treatment with sodium bicarbonate to correct the acidosis and associated complications is warranted.

A distal RTA would account for her past medical history of renal stones. Acidemia promotes both increased calcium phosphate release from bone (with subsequent hypercalciuria) and enhanced citrate reabsorption in the proximal renal tubules, leading to decreased urinary citrate. Citrate inhibits calcium stone formation. The increased calcium load to renal tubules in addition to decreased urinary citrate both lead to increased precipitation of calcium stones in the genitourinary tract.

A diagnosis of distal RTA should prompt evaluation for specific etiologies, such as Sjögren’s syndrome or SLE. While not diagnostic of any specific condition, low C3 and C4 levels suggest immune complex formation with related complement consumption, contributing to hypocomplementemia. The diagnosis of RTA may occur among patients with Sjögren’s syndrome in the absence of overt evidence of sicca syndrome (xerostomia and keratoconjunctivitis sicca). Other etiologies of distal RTA include conditions leading to hypercalciuria, such as hyperparathyroidism and idiopathic hypercalciuria, hereditary causes, toxins such as toluene, and drugs such as amphotericin B, lithium carbonate, and ibuprofen.

Her antinuclear antibody titer was >1:1280 (normal: <80). Anti-SSA and -SSB antibodies were both positive, with a titer >100 (normal: <20). Rheumatoid factor was positive at 22 IU/mL (normal: 0-14). Anti-smith, anti-double stranded DNA, and anti-ribonucleoprotein antibodies were negative.

Sjögren’s syndrome appears to be the ultimate etiology of this patient’s distal RTA. The diagnosis of Sjögren’s is more classically made in the presence of lacrimal and/or salivary dysfunction and confirmed with compatible autoantibodies. In the absence of dry eyes or dry mouth, attention should be focused on her skin findings. Cutaneous vasculitis does occur in a small percentage of Sjögren’s syndrome cases. Urticarial lesions have been reported in this subset, and skin biopsy would further support the diagnosis.

Treatment of Sjögren’s syndrome with immunosuppressive therapy may ameliorate renal parenchymal pathology and improve her profound metabolic disturbances.

On further questioning, she described several months of mild xerostomia, which resulted in increased consumption of fluids. She did not have keratoconjunctivitis sicca. Biopsy of her urticarial rash demonstrated a leukocytoclastic vasculitis with eosinophilic infiltration (Figure 1). Renal biopsy with hematoxylin and eosin staining, immunofluorescence, and electron microscopy demonstrated an immune complex-mediated glomerulonephritis and moderate tubulointerstitial nephritis (Figure 2). A diagnosis of Sjögren’s syndrome was made based on the patient’s xerostomia, high titers of antinuclear antibodies, SSA and SSB antibodies, positive rheumatoid factor, hypocomplementemia, and systemic manifestations associated with Sjögren’s syndrome, including distal RTA, nephrolithiasis, and hives, with histologic evidence of leukocytoclastic vasculitis.

She received aggressive potassium and bicarbonate repletion and, several days later, had normalization of both. Her weakness and myalgia rapidly improved concomitantly with the correction of her hypokalemia. Five days later she was ambulating independently and discharged with potassium citrate and prednisone therapy. She had improved fatigue and rash at a 1-month follow-up with rheumatology. As an outpatient, she was started on azathioprine and slowly tapered off her steroids. Over the next several months, she had normal potassium, bicarbonate, and renal function, although she did require lithotripsy for an obstructive renal stone.

RTA should be considered in the differential diagnosis of an unexplained normal anion gap metabolic acidosis. There are 3 major types of RTAs, with different characteristics. In type 1 (distal) RTA, the primary defect is impaired distal acidification of the urine. Distal RTA commonly presents with hypokalemia, calciuria (often presenting as renal stones), and a positive UAG.¹ In type 2 (proximal) RTA, the primary defect is impaired bicarbonate reabsorption, leading to bicarbonate wasting in the urine. Proximal RTAs can be secondary to an isolated defect in bicarbonate reabsorption or generalized proximal renal tubule dysfunction (Fanconi syndrome).¹ A type 4 RTA is characterized by hypoaldosteronism, presenting usually with a mild nonanion gap metabolic acidosis and hyperkalemia. This patient’s history of renal stones, hypokalemia, and positive UAG supported a type 1 (distal) RTA. Distal RTA is often idiopathic, but initial evaluation should include a review of medications and investigation into an underlying systemic disorder (eg, plasma cell dyscrasia or autoimmune disease). This would include eliciting a possible history of xerostomia and xerophthalmia, together with testing of SSA (Ro) and SSB (La) antibodies, to assess for Sjögren’s syndrome. In addition, checking serum calcium to assess for hyperparathyroidism or familial idiopathic hypercalciuria and a review of medications, such as lithium and amphotericin,¹ may uncover other secondary causes of distal RTA.

While Sjögren’s syndrome primarily affects salivary and lacrimal glands, leading to dry mouth and dry eyes, respectively, extraglandular manifestations are common, with fatigue and arthralgia occurring in half of patients. Extra-glandular involvement also often includes the skin and kidneys but can affect several other organ systems, including the central nervous system, heart, lungs, bone marrow, and lymph nodes.²

There are many cutaneous manifestations of Sjögren’s syndrome.³ Xerosis, or xeroderma, is the most common and is characterized by dry, scaly skin. Cutaneous vasculitis can occur in 10% of patients with Sjögren’s syndrome and often presents with palpable purpura or diffuse urticarial lesions, as in our patient.⁴ Erythematous maculopapules and cryoglobulinemic vasculitis may also occur.⁴ A less common skin manifestation is annular erythema, presenting as an indurated, ring-like lesion.⁵

Chronic tubulointerstitial nephritis is the most common renal manifestation of Sjögren’s syndrome.⁶ This often pre-sents with a mild elevated serum creatinine and a distal RTA, leading to hypokalemia, as in the case discussed. Distal RTA is well described, occurring in one-quarter of patients with Sjögren’s syndrome.⁷ The pathophysiology leading to distal RTA in Sjögren’s syndrome is thought to arise from autoimmune injury to the H(+)-ATPase pump in the renal collecting tubules, leading to decreased distal proton secretion.^8,9 Younger adults with Sjögren’s syndrome, in the third and fourth decades of life, have a predilection to develop tubulointerstitial inflammation, distal RTA, and nephrolithiasis, as in the present case.⁶ Sjögren’s syndrome less commonly presents with membranoproliferative glomerulonephritis or membranous nephropathy.^10,11 Cryoglobulinemia-associated hypocomplementemia and glomerulonephritis may also occur with Sjögren’s syndrome, yet glomerular lesions are less common than is tubulointerstitial inflammation. The patient discussed had proteinuria and evidence of immune complex-mediated glomerulonephritis.

Treatment of sicca symptoms is generally supportive. It includes artificial tears, encouragement of good hydration, salivary stimulants, and maintaining good oral hygiene. Pilocarpine, a cholinergic parasympathomimetic agent, is approved by the Food and Drug Administration to treat dry mouth associated with Sjögren’s syndrome. The treatment of extraglandular manifestations depends on the organ(s) involved. More severe presentations, such as vasculitis and glomerulonephritis, often require immunosuppressive therapy with systemic glucocorticoids, cyclophosphamide, azathioprine, or other immunosuppressive agents,¹² including rituximab. RTA often necessitates treatment with oral bicarbonate and supplemental potassium repletion.

The base rate of disease (ie, prevalence of disease) influences a diagnostician’s pretest probability of a given diagnosis. The discussant briefly considered rare causes of hives (eg, vasculitis) but appropriately fine-tuned their differential for the patient’s hypokalemia and RTA. Once the diagnosis of Sjögren’s syndrome was made with certainty, the clinician was able to revisit the patient’s rash with a new lens. Urticarial vasculitis suddenly became a plausible consideration, despite its rarity (compared to allergic causes of hives) because of the direct link to the underlying autoimmune condition, which affected both the proximal muscles and distal nephrons.

• Evaluation of patients with weakness starts with determining true muscle weakness (ie, pathology involving the brain, spinal cord, peripheral nerve, neuromuscular junction, and/or muscle) from asthenia.
• Distal RTA should be considered in patients with a nonanion gap metabolic acidosis and hypokalemia.
• Sjögren’s syndrome has many extraglandular clinical manifestations, including vasculitis, urticaria, tubulointerstitial renal inflammation, glomerulonephritis, and lymphoma.

Acknowledgment

The authors thank Virgilius Cornea, MD, for his interpretation of the pathologic images.

Disclosure

Dr. Manesh is supported by the Jeremiah A. Barondess Fellowship in the Clinical Transaction of the New York Academy of Medicine, in collaboration with the Accreditation Council for Graduate Medical Education (ACGME). The authors declare no conflicts of interests.

A previously healthy 30-year-old woman presented to the emergency department with 2 weeks of weakness.

True muscle weakness must be distinguished from the more common causes of asthenia. Many systemic disorders produce fatigue, with resulting functional limitation that is often interpreted by patients as weakness. Initial history should focus on conditions producing fatigue, such as cardiopulmonary disease, anemia, connective tissue disease, depression or cachexia related to malignancy, infection, or other inflammatory states. Careful questioning may reveal evidence of dyspnea, poor exercise tolerance, or joint pain as an alternative to actual loss of muscle power. If true weakness is still suspected, attention should be focused on the pattern, onset, anatomic site, and progression of weakness. Muscle weakness is often characterized by difficulty with specific tasks, such as climbing stairs, rising from a chair, raising a hand, or using cutlery. The physical examination is critical in determining whether weakness is due to true loss of motor power. The differential diagnosis of weakness is broad and includes neurologic, infectious, endocrine, inflammatory, genetic, metabolic, and drug-induced etiologies.

She initially experienced 3 days of mild cramps and soreness in her thighs. She then developed weakness that began in her thighs and progressed to involve her lower legs and upper and lower arms. She had difficulty combing her hair. She required the use of her arms to get up from a chair. She grasped onto objects to aid in ambulation around the house. In addition, she described 1 year of moderate fatigue but no fever, weight loss, dyspnea, dysphagia, visual changes, paresthesias, bowel or bladder incontinence, back pain, or preceding gastrointestinal or respiratory illness. She had experienced diffuse intermittent hives, most prominent in her chest and upper arms, for the past several weeks.

History certainly supports true weakness but will need to be confirmed on examination. The distribution began as proximal but now appears diffuse. The presence of myalgia and cramping raises the possibility of noninflammatory myopathies, which are usually more insidious in onset. A severe electrolyte disturbance would be possible, based on the diffuse nature of weakness that was preceded by cramping. The distribution of weakness and lack of bowel or bladder incontinence is reassuring and suggests against a spinal cord disorder; however, a high index of suspicion must be maintained for myelopathy because delayed treatment might result in irreversible paralysis.

The patient’s course also includes hives. Common causes of hives include infections and allergic reactions to medications, foods, and insect stings. Urticaria may also result from systemic disorders, such as vasculitis, lupus, lymphoma, mastocytosis, and paraproteinemias, which can be associated with weakness and fatigue. Although severe weakness in combination with hives makes an infectious and allergic reaction less likely, we still seek to ascertain if the evolving chief complaints of weakness and hives are the result of a single unifying and evolving multisystem disorder or are distinct and unrelated processes.

Her past medical history included fibromyalgia, kidney stones, and gastroesophageal reflux disease. One week prior to presentation, she was prescribed prednisone 60 mg daily for the treatment of hives; the dose had been tapered to 40 mg at presentation, with mild improvement of hives. She recently started doxepin for fibromyalgia and insomnia. She lived at home with her husband and 8-year-old child. She worked as a clerk in a pest control office and denied any pesticide exposure. She denied tobacco, alcohol, or illicit drug use. Her family history included systemic lupus erythematosus (SLE) in her mother and maternal aunt.

Glucocorticoids are associated with myopathy; however, the weakness preceded steroid therapy. Thus, unless there was unknown exposure to high-dose steroid medication to treat recurrent episodes of urticaria earlier in her course, glucocorticoid-related myopathy is unlikely. Fibromyalgia might cause the perception of weakness from pain. However, the history of difficulty combing her hair and rising from a chair suggests actual loss of motor power. The side effects of her medications, such as newly started doxepin, must be reviewed. A family history of SLE raises concern for rheumatologic conditions; however, one might expect improvement with steroid therapy.

On physical examination, her temperature was 36.9 °C, blood pressure 126/93 mmHg, pulse 81 beats per minute, respiratory rate 16 breaths per minute, and oxygen saturation 100% on ambient air. Her cardiopulmonary examination was normal. Her abdomen was nontender and without hepatosplenomegaly. Her strength was 2 out of 5 in proximal and distal legs, bilaterally, and 4 out of 5 in proximal and distal upper extremities. She had normal muscle tone without fasciculations or atrophy. Her joints were without edema, erythema, or impaired range of motion. She had normal sensation to light touch in arms and legs. Her reflexes were 2+ in the patellar, Achilles, and brachioradialis tendons. She had no lymphadenopathy, mucosal ulcerations, or alopecia. A skin examination revealed smooth, slightly elevated, and faded pink wheals that were diffuse but most prominent in upper arms and chest.

Physical examination confirms the presence of true muscle weakness. The differential diagnosis is narrowed by several findings, both positive and negative, elicited in the examination. The diffuse nature of the weakness eliminates focal central nervous system lesions, such as stroke, intracranial mass lesions, or demyelinating white matter foci. Combining this finding with normal reflexes and history of preceding myalgias makes electrolyte-induced and inflammatory (eg, polymyositis) myopathies more likely. The normal deep tendon reflexes and the absence of a delayed relaxation phase lower the likelihood of hypothyroidism. 

Diseases originating from the neuromuscular junction, such as myasthenia gravis, may also present with weakness and normal reflexes, although this pattern of weakness would be atypical; myasthenia gravis classically presents with fatigable weakness and ocular findings of diplopia and/or ptosis. First-tier testing should include a complete blood count to evaluate for eosinophilia, comprehensive metabolic panel, and urinalysis for myoglobinuria, thyroid stimulating hormone, and muscle enzymes. 


Results of a complete blood count demonstrated a leukocyte count of 16.1 k/uL with 82% neutrophils, 13% lymphocytes, 5% monocytes, and 0% eosinophils. Hemoglobin was 13.2 g/dL, and platelet count 226 k/uL. Sodium was 136 mmol/L, potassium 1.5 mmol/L, chloride 115 mmol/L, bicarbonate 12 mmol/L, blood urea nitrogen 26 mg/dL, creatinine 1.0 mg/dL (baseline creatinine: 0.6), and glucose 102 mg/dL. Calcium was 9.4 mg/dL, magnesium 2.6 mg/dL, phosphorus 1.8 mg/dL, CK 501 U/L (normal: 40-230), and TSH 5.48 uIU/mL (normal: 0.5-4). Aspartate aminotransferase was 64 U/L, alanine aminotransferase 23 U/L, alkaline phosphatase 66 U/L, bilirubin 0.9 mg/dL, albumin 3.8 g/dL, and total protein 8.7 g/dL (normal: 6.2-7.8). Human immunodeficiency virus antibody screen was negative. An electrocardiogram revealed normal sinus rhythm, flattened T waves, and prominent U waves.

Potassium losses are classically categorized into 1 of 3 groups: renal losses, gastrointestinal losses, or transcellular shifts. Without a clear history of diuretic use, renal losses may not be apparent on history and examination. In contrast, gastrointestinal losses are almost always evidenced by a history of vomiting and/or diarrhea, with rare exceptions, including unreported laxative abuse or surreptitious vomiting. Transcellular potassium shifts can be seen in states of increased insulin or beta-adrenergic activity and alkalosis and result from both primary and secondary causes of hypokalemic periodic paralysis.

The presence of a reduced serum bicarbonate and elevated chloride concentration suggests a normal anion gap metabolic acidosis. Many conditions associated with normal anion gap metabolic acidosis are evident by history, such as diarrhea. In enigmatic cases such as this, it will be important to take a stepwise approach that includes an evaluation for urinary potassium losses and assessment of acid-base status. An unexplained normal anion gap metabolic acidosis combined with hypokalemia raises suspicion for a distal renal tubular acidosis (RTA). Additional testing to evaluate for a possible RTA should include the assessment of urinary electrolytes and urinary pH. The hypokalemia explains her weakness, but the etiology of such profound hypokalemia is not evident, nor is it clear how it relates to her hives.

The severity of the hypokalemia, combined with electrocardiogram changes, necessitates rapid intravenous potassium repletion, telemetry monitoring, and frequent serum potassium measurement. Treatment of her metabolic acidosis is more nuanced and depends upon both the severity of disturbance and the suspicion of whether the etiology is transcellular shift, potassium depletion, or both.

Urine studies demonstrated a urine specific gravity of 1.006 (normal: 1.001-1.030), urine pH was 6.5 (normal: 5-6.5), trace leukocyte esterase, negative nitrite, 30 mg/dL of protein (normal: <15), sodium 64 mmol/L (normal: 40-220), potassium 17 mmol/L (normal: 25-125), and chloride 71 mmol/L (normal: 110-250). Urine microscopy demonstrated 3 red blood cells per high power field (normal: 0-1), 4 white blood cells per high power field (normal: 0-4), 4+ bacteria per high power field, and no red blood cell casts. Urine protein-to-creatinine ratio was 1.6. C3 and C4 complement levels were 53 mg/dL (normal: 80-165) and 12 mg/dL (normal: 15-49), respectively. C-reactive protein was <0.5 (normal: 0-0.9), and erythrocyte sedimentation rate was 16 mm/hour (normal: 0-20).

A calculation of the urine anion gap (UAG; [urine sodium + urine potassium] – urine chloride) yields a UAG of 10 mq/L. A positive UAG, together with a nongap metabolic acidosis, should prompt the consideration of RTA. The normal renal response to acidosis is to reduce the urine pH to less than 5.3 through an increase in hydrogen ion excretion in the form of ammonium. A urine pH of 6.5 is highly suggestive of type 1 (distal) RTA and its associated impairment of distal acidification. Treatment with sodium bicarbonate to correct the acidosis and associated complications is warranted.

A distal RTA would account for her past medical history of renal stones. Acidemia promotes both increased calcium phosphate release from bone (with subsequent hypercalciuria) and enhanced citrate reabsorption in the proximal renal tubules, leading to decreased urinary citrate. Citrate inhibits calcium stone formation. The increased calcium load to renal tubules in addition to decreased urinary citrate both lead to increased precipitation of calcium stones in the genitourinary tract.

A diagnosis of distal RTA should prompt evaluation for specific etiologies, such as Sjögren’s syndrome or SLE. While not diagnostic of any specific condition, low C3 and C4 levels suggest immune complex formation with related complement consumption, contributing to hypocomplementemia. The diagnosis of RTA may occur among patients with Sjögren’s syndrome in the absence of overt evidence of sicca syndrome (xerostomia and keratoconjunctivitis sicca). Other etiologies of distal RTA include conditions leading to hypercalciuria, such as hyperparathyroidism and idiopathic hypercalciuria, hereditary causes, toxins such as toluene, and drugs such as amphotericin B, lithium carbonate, and ibuprofen.

Her antinuclear antibody titer was >1:1280 (normal: <80). Anti-SSA and -SSB antibodies were both positive, with a titer >100 (normal: <20). Rheumatoid factor was positive at 22 IU/mL (normal: 0-14). Anti-smith, anti-double stranded DNA, and anti-ribonucleoprotein antibodies were negative.

Sjögren’s syndrome appears to be the ultimate etiology of this patient’s distal RTA. The diagnosis of Sjögren’s is more classically made in the presence of lacrimal and/or salivary dysfunction and confirmed with compatible autoantibodies. In the absence of dry eyes or dry mouth, attention should be focused on her skin findings. Cutaneous vasculitis does occur in a small percentage of Sjögren’s syndrome cases. Urticarial lesions have been reported in this subset, and skin biopsy would further support the diagnosis.

Treatment of Sjögren’s syndrome with immunosuppressive therapy may ameliorate renal parenchymal pathology and improve her profound metabolic disturbances.

On further questioning, she described several months of mild xerostomia, which resulted in increased consumption of fluids. She did not have keratoconjunctivitis sicca. Biopsy of her urticarial rash demonstrated a leukocytoclastic vasculitis with eosinophilic infiltration (Figure 1). Renal biopsy with hematoxylin and eosin staining, immunofluorescence, and electron microscopy demonstrated an immune complex-mediated glomerulonephritis and moderate tubulointerstitial nephritis (Figure 2). A diagnosis of Sjögren’s syndrome was made based on the patient’s xerostomia, high titers of antinuclear antibodies, SSA and SSB antibodies, positive rheumatoid factor, hypocomplementemia, and systemic manifestations associated with Sjögren’s syndrome, including distal RTA, nephrolithiasis, and hives, with histologic evidence of leukocytoclastic vasculitis.

She received aggressive potassium and bicarbonate repletion and, several days later, had normalization of both. Her weakness and myalgia rapidly improved concomitantly with the correction of her hypokalemia. Five days later she was ambulating independently and discharged with potassium citrate and prednisone therapy. She had improved fatigue and rash at a 1-month follow-up with rheumatology. As an outpatient, she was started on azathioprine and slowly tapered off her steroids. Over the next several months, she had normal potassium, bicarbonate, and renal function, although she did require lithotripsy for an obstructive renal stone.

RTA should be considered in the differential diagnosis of an unexplained normal anion gap metabolic acidosis. There are 3 major types of RTAs, with different characteristics. In type 1 (distal) RTA, the primary defect is impaired distal acidification of the urine. Distal RTA commonly presents with hypokalemia, calciuria (often presenting as renal stones), and a positive UAG.¹ In type 2 (proximal) RTA, the primary defect is impaired bicarbonate reabsorption, leading to bicarbonate wasting in the urine. Proximal RTAs can be secondary to an isolated defect in bicarbonate reabsorption or generalized proximal renal tubule dysfunction (Fanconi syndrome).¹ A type 4 RTA is characterized by hypoaldosteronism, presenting usually with a mild nonanion gap metabolic acidosis and hyperkalemia. This patient’s history of renal stones, hypokalemia, and positive UAG supported a type 1 (distal) RTA. Distal RTA is often idiopathic, but initial evaluation should include a review of medications and investigation into an underlying systemic disorder (eg, plasma cell dyscrasia or autoimmune disease). This would include eliciting a possible history of xerostomia and xerophthalmia, together with testing of SSA (Ro) and SSB (La) antibodies, to assess for Sjögren’s syndrome. In addition, checking serum calcium to assess for hyperparathyroidism or familial idiopathic hypercalciuria and a review of medications, such as lithium and amphotericin,¹ may uncover other secondary causes of distal RTA.

While Sjögren’s syndrome primarily affects salivary and lacrimal glands, leading to dry mouth and dry eyes, respectively, extraglandular manifestations are common, with fatigue and arthralgia occurring in half of patients. Extra-glandular involvement also often includes the skin and kidneys but can affect several other organ systems, including the central nervous system, heart, lungs, bone marrow, and lymph nodes.²

There are many cutaneous manifestations of Sjögren’s syndrome.³ Xerosis, or xeroderma, is the most common and is characterized by dry, scaly skin. Cutaneous vasculitis can occur in 10% of patients with Sjögren’s syndrome and often presents with palpable purpura or diffuse urticarial lesions, as in our patient.⁴ Erythematous maculopapules and cryoglobulinemic vasculitis may also occur.⁴ A less common skin manifestation is annular erythema, presenting as an indurated, ring-like lesion.⁵

Chronic tubulointerstitial nephritis is the most common renal manifestation of Sjögren’s syndrome.⁶ This often pre-sents with a mild elevated serum creatinine and a distal RTA, leading to hypokalemia, as in the case discussed. Distal RTA is well described, occurring in one-quarter of patients with Sjögren’s syndrome.⁷ The pathophysiology leading to distal RTA in Sjögren’s syndrome is thought to arise from autoimmune injury to the H(+)-ATPase pump in the renal collecting tubules, leading to decreased distal proton secretion.^8,9 Younger adults with Sjögren’s syndrome, in the third and fourth decades of life, have a predilection to develop tubulointerstitial inflammation, distal RTA, and nephrolithiasis, as in the present case.⁶ Sjögren’s syndrome less commonly presents with membranoproliferative glomerulonephritis or membranous nephropathy.^10,11 Cryoglobulinemia-associated hypocomplementemia and glomerulonephritis may also occur with Sjögren’s syndrome, yet glomerular lesions are less common than is tubulointerstitial inflammation. The patient discussed had proteinuria and evidence of immune complex-mediated glomerulonephritis.

Treatment of sicca symptoms is generally supportive. It includes artificial tears, encouragement of good hydration, salivary stimulants, and maintaining good oral hygiene. Pilocarpine, a cholinergic parasympathomimetic agent, is approved by the Food and Drug Administration to treat dry mouth associated with Sjögren’s syndrome. The treatment of extraglandular manifestations depends on the organ(s) involved. More severe presentations, such as vasculitis and glomerulonephritis, often require immunosuppressive therapy with systemic glucocorticoids, cyclophosphamide, azathioprine, or other immunosuppressive agents,¹² including rituximab. RTA often necessitates treatment with oral bicarbonate and supplemental potassium repletion.

The base rate of disease (ie, prevalence of disease) influences a diagnostician’s pretest probability of a given diagnosis. The discussant briefly considered rare causes of hives (eg, vasculitis) but appropriately fine-tuned their differential for the patient’s hypokalemia and RTA. Once the diagnosis of Sjögren’s syndrome was made with certainty, the clinician was able to revisit the patient’s rash with a new lens. Urticarial vasculitis suddenly became a plausible consideration, despite its rarity (compared to allergic causes of hives) because of the direct link to the underlying autoimmune condition, which affected both the proximal muscles and distal nephrons.

• Evaluation of patients with weakness starts with determining true muscle weakness (ie, pathology involving the brain, spinal cord, peripheral nerve, neuromuscular junction, and/or muscle) from asthenia.
• Distal RTA should be considered in patients with a nonanion gap metabolic acidosis and hypokalemia.
• Sjögren’s syndrome has many extraglandular clinical manifestations, including vasculitis, urticaria, tubulointerstitial renal inflammation, glomerulonephritis, and lymphoma.

Acknowledgment

The authors thank Virgilius Cornea, MD, for his interpretation of the pathologic images.

Disclosure

Dr. Manesh is supported by the Jeremiah A. Barondess Fellowship in the Clinical Transaction of the New York Academy of Medicine, in collaboration with the Accreditation Council for Graduate Medical Education (ACGME). The authors declare no conflicts of interests.

Palliative care is specialized medical care focused on providing relief from the symptoms, pain, and stress of a serious illness. The goal is to improve the quality of life for both the patient and the family. In all settings, palliative care has been found to improve patients’ quality of life,^1,2 improve family satisfaction and well-being,³ reduce resource utilization and costs,⁴ and, in some studies, increase the length of life for seriously ill patients.⁵

Given the frequency with which seriously ill patients are hospitalized, hospitalists are well positioned to identify those who could benefit from palliative care interventions.⁶ Hospitalists routinely use primary palliative care skills, including pain and symptom management and skilled care planning conversations. For complex cases, such as patients with intractable symptoms or major family conflict, hospitalists may refer to specialist palliative care teams for consultation.

The Society of Hospital Medicine (SHM) defines the key primary palliative care responsibilities for hospitalists as (1) leading discussions on the goals of care and advance care planning with patients and families, (2) screening and treating common physical symptoms, and (3) referring patients to community services to provide support postdischarge.⁷ According to data in the National Palliative Care Registry,⁸ 48% of all palliative care referrals in 2015 came from hospitalists, which is more than double the percentage of referrals from any other specialty.⁹

In a recent survey conducted by SHM about serious illness communication, 53% of hospitalists reported concerns about a patient or family’s understanding of their prognosis, and 50% indicated that they do not feel confident managing family conflict.¹⁰

Context

Patients with multiple serious chronic conditions are often forced to rely on emergency services when crises, such as uncontrolled pain or dyspnea exacerbation, occur after hours, resulting in the revolving-door hospitalizations that typically characterize their care.¹¹ As the prevalence of serious illness rises and the shift to value-based payment accelerates, hospitals are under increasing pressure to deliver efficient and high-quality services that meet the needs of seriously ill patients. The integration of standardized palliative care screening and assessment enables hospitalists and other providers to identify high-need individuals and match services and delivery models to needs, whether it be respite care for an exhausted and overwhelmed family caregiver or a home protocol for managing recurrent dyspnea crises for a patient with chronic obstructive pulmonary disease (COPD). This process improves the quality of care and quality of life, and in doing so, prevents the need for costly crisis care.

Reducing Readmissions

By identifying patients in need of extra symptom management support, or those at a turning point requiring discussion about achievable priorities for care, hospitalists can avert crises for patients earlier in the disease trajectory either by managing the patient’s palliative needs themselves or by connecting patients with specialty palliative care services as needed. This leads to a better quality of life (and survival in some studies) for both patients and their families^1,3,5 and reduces unnecessary emergency department (ED) and hospital use.¹² Hospitalists providing palliative care can also reduce readmissions by improving care coordination, including clinical communication and medication reconciliation after discharge.¹³

A 2015 Harvard Business Review study found that the quality of communication in the hospital is the strongest independent predictor of readmissions when combined with process-of-care improvements, such as standardized patient screening and assessment of family caregiver capacity.¹⁴ While medical education prepares physicians to deliver evidence-based medical care, it currently offers little to no training in communication skills, despite mounting evidence that this is a critical component of quality healthcare.

Cost Savings

Hospital palliative care teams are associated with significant hospital cost savings that result from aligning care with patient priorities, leading, in turn, to reduced nonbeneficial hospital imaging, medications, procedures, and length of stay.¹⁵ See the table^16,17 for examples of cost and quality outcomes of specialist palliative care provision and evidence supporting each outcome.^18-25

Multiple studies consistently demonstrate that inpatient palliative care teams reduce hospital costs.²⁶ One randomized controlled trial investigating the impact of an inpatient palliative care service found that patients who received care from the palliative care team reported greater satisfaction with their care, had fewer intensive care unit admissions, had more advanced directives at hospital discharge, longer hospice length of stay, and lower total healthcare costs (a net difference of $6766 per patient).²³

Research shows that the earlier palliative care is provided, the greater the impact on the subsequent course of care,²⁷ suggesting that hospitalists who provide frontline palliative care interventions as early as possible in a seriously ill patient’s stay will be able to provide higher quality care with lower overall costs. Notably, the majority of research on cost savings associated with palliative care has focused on the impact of specialist palliative care teams, and further research is needed to understand the economic impact of primary palliative care provision.

Improving Satisfaction

Shifting to value-based payment means that the patient and family experience determine an increasingly large percentage of hospital and provider reimbursement. Palliative care approaches, such as family caregiver assessment and support, access to 24/7 assistance after discharge, and person-centered care by an interdisciplinary team, improve performance in all of these measures. Communication skills training improves patient satisfaction scores, and skilled discussions about achievable priorities for care are associated with better quality of life, reduced nonbeneficial and burdensome treatments, and an increase in goal-concordant care.¹⁹ Communication skills training has also been shown to reduce burnout and improve empathy among physicians.^28,29

Though more evidence is needed to understand the impact of primary palliative care provision by hospitalists, the strong evidence on the benefits of specialty palliative care suggests that the skilled provision of primary palliative care by hospitalists will result in higher quality, higher value care. A number of training options exist for midcareer hospital medicine clinicians, including both in-person and online training in communication and other palliative care skills.

• The Center to Advance Palliative Care (CAPC) is a membership organization that offers online continuing education unit and continuing medical education courses on communication skills, pain and symptom management, caregiver support, and care coordination. CAPC also offers courses on palliative interventions for patients with dementia, COPD, and heart failure.
• SHM is actively invested in engaging hospitalists in palliative care skills training. SHM provides free toolkits on a variety of topics within the palliative care domain, including pain management, postacute care transitions, and opioid safety. The recently released Serious Illness Communication toolkit offers background on the role of hospitalists in palliative care provision, a pathway for fitting goals-of-care conversations into hospitalist workflow and recommended metrics and training resources. SHM also uses a mentored implementation model in which expert physicians mentor hospital team members on best practices in palliative care. SHM’s Palliative Care Task Force seeks to identify educational activities for hospitalists and create opportunities to integrate palliative care in hospital medicine.³⁰
• The Serious Illness Care Program at Ariadne Labs in Boston aims to facilitate conversations between clinicians and seriously ill patients through its Serious Illness Conversation Guide, combined with technical assistance on workflow redesign to help clinicians conduct and document serious illness conversations.
• VitalTalk specializes in clinical communication education. Through online and in-person train-the-trainer programs, VitalTalk equips clinicians to lead communication training programs at their home institutions.
• The Education in Palliative and End-of-Life Care Program and End-of-Life Nursing Education Consortium (ELNEC) uses a train-the-trainer approach to educate providers in palliative care clinical competencies and increase the reach of primary palliative care provision. ELNEC workshops are complemented by a curriculum of online clinical training modules.

Though palliative care skills training is a necessary first step, hospitalists also cite lack of time, difficulty finding records of previous patient discussions, and frequent handoffs as among the barriers to integrating palliative care into their practice.¹⁰ Studies examining the process of palliative care and hospital culture change have found that barriers to palliative care integration include a culture of aggressive care in EDs, lack of standardized patient identification criteria, and limited knowledge about and staffing for palliative care.³¹ These data indicate the need for system changes that enable hospitalists to operationalize palliative care principles.

Health systems must implement systems and processes that routinize palliative care, making it part of the mainstream course of care for seriously ill patients and their caregivers. This includes developing systems for the identification of patients with palliative care needs, embedding palliative care assessment and referral into clinical workflows, and enabling standardized palliative care documentation in electronic medical records. While palliative care skills training is essential, investment in systems change is no less critical to embedding palliative care practices in clinical norms across specialties.

Hospitalists can use a palliative approach to improve care quality and quality of life for seriously ill patients while helping to avoid preventable and unnecessary 911 calls, ED visits, and hospitalizations. The shift towards value-based payment is a strong incentive for hospitals and hospitalists to direct resources toward practices that improve the quality of life and care for the highest-need patients and their families. When equipped with the tools they need to provide palliative care, either themselves or in collaboration with palliative care teams, hospitalists have the opportunity to profoundly redirect the experience of care for seriously ill patients and their families.

Disclosure

The authors declared no conflicts of interest.

Palliative care is specialized medical care focused on providing relief from the symptoms, pain, and stress of a serious illness. The goal is to improve the quality of life for both the patient and the family. In all settings, palliative care has been found to improve patients’ quality of life,^1,2 improve family satisfaction and well-being,³ reduce resource utilization and costs,⁴ and, in some studies, increase the length of life for seriously ill patients.⁵

Given the frequency with which seriously ill patients are hospitalized, hospitalists are well positioned to identify those who could benefit from palliative care interventions.⁶ Hospitalists routinely use primary palliative care skills, including pain and symptom management and skilled care planning conversations. For complex cases, such as patients with intractable symptoms or major family conflict, hospitalists may refer to specialist palliative care teams for consultation.

The Society of Hospital Medicine (SHM) defines the key primary palliative care responsibilities for hospitalists as (1) leading discussions on the goals of care and advance care planning with patients and families, (2) screening and treating common physical symptoms, and (3) referring patients to community services to provide support postdischarge.⁷ According to data in the National Palliative Care Registry,⁸ 48% of all palliative care referrals in 2015 came from hospitalists, which is more than double the percentage of referrals from any other specialty.⁹

In a recent survey conducted by SHM about serious illness communication, 53% of hospitalists reported concerns about a patient or family’s understanding of their prognosis, and 50% indicated that they do not feel confident managing family conflict.¹⁰

Context

Patients with multiple serious chronic conditions are often forced to rely on emergency services when crises, such as uncontrolled pain or dyspnea exacerbation, occur after hours, resulting in the revolving-door hospitalizations that typically characterize their care.¹¹ As the prevalence of serious illness rises and the shift to value-based payment accelerates, hospitals are under increasing pressure to deliver efficient and high-quality services that meet the needs of seriously ill patients. The integration of standardized palliative care screening and assessment enables hospitalists and other providers to identify high-need individuals and match services and delivery models to needs, whether it be respite care for an exhausted and overwhelmed family caregiver or a home protocol for managing recurrent dyspnea crises for a patient with chronic obstructive pulmonary disease (COPD). This process improves the quality of care and quality of life, and in doing so, prevents the need for costly crisis care.

Reducing Readmissions

By identifying patients in need of extra symptom management support, or those at a turning point requiring discussion about achievable priorities for care, hospitalists can avert crises for patients earlier in the disease trajectory either by managing the patient’s palliative needs themselves or by connecting patients with specialty palliative care services as needed. This leads to a better quality of life (and survival in some studies) for both patients and their families^1,3,5 and reduces unnecessary emergency department (ED) and hospital use.¹² Hospitalists providing palliative care can also reduce readmissions by improving care coordination, including clinical communication and medication reconciliation after discharge.¹³

A 2015 Harvard Business Review study found that the quality of communication in the hospital is the strongest independent predictor of readmissions when combined with process-of-care improvements, such as standardized patient screening and assessment of family caregiver capacity.¹⁴ While medical education prepares physicians to deliver evidence-based medical care, it currently offers little to no training in communication skills, despite mounting evidence that this is a critical component of quality healthcare.

Cost Savings

Hospital palliative care teams are associated with significant hospital cost savings that result from aligning care with patient priorities, leading, in turn, to reduced nonbeneficial hospital imaging, medications, procedures, and length of stay.¹⁵ See the table^16,17 for examples of cost and quality outcomes of specialist palliative care provision and evidence supporting each outcome.^18-25

Multiple studies consistently demonstrate that inpatient palliative care teams reduce hospital costs.²⁶ One randomized controlled trial investigating the impact of an inpatient palliative care service found that patients who received care from the palliative care team reported greater satisfaction with their care, had fewer intensive care unit admissions, had more advanced directives at hospital discharge, longer hospice length of stay, and lower total healthcare costs (a net difference of $6766 per patient).²³

Research shows that the earlier palliative care is provided, the greater the impact on the subsequent course of care,²⁷ suggesting that hospitalists who provide frontline palliative care interventions as early as possible in a seriously ill patient’s stay will be able to provide higher quality care with lower overall costs. Notably, the majority of research on cost savings associated with palliative care has focused on the impact of specialist palliative care teams, and further research is needed to understand the economic impact of primary palliative care provision.

Improving Satisfaction

Shifting to value-based payment means that the patient and family experience determine an increasingly large percentage of hospital and provider reimbursement. Palliative care approaches, such as family caregiver assessment and support, access to 24/7 assistance after discharge, and person-centered care by an interdisciplinary team, improve performance in all of these measures. Communication skills training improves patient satisfaction scores, and skilled discussions about achievable priorities for care are associated with better quality of life, reduced nonbeneficial and burdensome treatments, and an increase in goal-concordant care.¹⁹ Communication skills training has also been shown to reduce burnout and improve empathy among physicians.^28,29

Though more evidence is needed to understand the impact of primary palliative care provision by hospitalists, the strong evidence on the benefits of specialty palliative care suggests that the skilled provision of primary palliative care by hospitalists will result in higher quality, higher value care. A number of training options exist for midcareer hospital medicine clinicians, including both in-person and online training in communication and other palliative care skills.

• The Center to Advance Palliative Care (CAPC) is a membership organization that offers online continuing education unit and continuing medical education courses on communication skills, pain and symptom management, caregiver support, and care coordination. CAPC also offers courses on palliative interventions for patients with dementia, COPD, and heart failure.
• SHM is actively invested in engaging hospitalists in palliative care skills training. SHM provides free toolkits on a variety of topics within the palliative care domain, including pain management, postacute care transitions, and opioid safety. The recently released Serious Illness Communication toolkit offers background on the role of hospitalists in palliative care provision, a pathway for fitting goals-of-care conversations into hospitalist workflow and recommended metrics and training resources. SHM also uses a mentored implementation model in which expert physicians mentor hospital team members on best practices in palliative care. SHM’s Palliative Care Task Force seeks to identify educational activities for hospitalists and create opportunities to integrate palliative care in hospital medicine.³⁰
• The Serious Illness Care Program at Ariadne Labs in Boston aims to facilitate conversations between clinicians and seriously ill patients through its Serious Illness Conversation Guide, combined with technical assistance on workflow redesign to help clinicians conduct and document serious illness conversations.
• VitalTalk specializes in clinical communication education. Through online and in-person train-the-trainer programs, VitalTalk equips clinicians to lead communication training programs at their home institutions.
• The Education in Palliative and End-of-Life Care Program and End-of-Life Nursing Education Consortium (ELNEC) uses a train-the-trainer approach to educate providers in palliative care clinical competencies and increase the reach of primary palliative care provision. ELNEC workshops are complemented by a curriculum of online clinical training modules.

Though palliative care skills training is a necessary first step, hospitalists also cite lack of time, difficulty finding records of previous patient discussions, and frequent handoffs as among the barriers to integrating palliative care into their practice.¹⁰ Studies examining the process of palliative care and hospital culture change have found that barriers to palliative care integration include a culture of aggressive care in EDs, lack of standardized patient identification criteria, and limited knowledge about and staffing for palliative care.³¹ These data indicate the need for system changes that enable hospitalists to operationalize palliative care principles.

Health systems must implement systems and processes that routinize palliative care, making it part of the mainstream course of care for seriously ill patients and their caregivers. This includes developing systems for the identification of patients with palliative care needs, embedding palliative care assessment and referral into clinical workflows, and enabling standardized palliative care documentation in electronic medical records. While palliative care skills training is essential, investment in systems change is no less critical to embedding palliative care practices in clinical norms across specialties.

Hospitalists can use a palliative approach to improve care quality and quality of life for seriously ill patients while helping to avoid preventable and unnecessary 911 calls, ED visits, and hospitalizations. The shift towards value-based payment is a strong incentive for hospitals and hospitalists to direct resources toward practices that improve the quality of life and care for the highest-need patients and their families. When equipped with the tools they need to provide palliative care, either themselves or in collaboration with palliative care teams, hospitalists have the opportunity to profoundly redirect the experience of care for seriously ill patients and their families.

Disclosure

The authors declared no conflicts of interest.

Palliative care is specialized medical care focused on providing relief from the symptoms, pain, and stress of a serious illness. The goal is to improve the quality of life for both the patient and the family. In all settings, palliative care has been found to improve patients’ quality of life,^1,2 improve family satisfaction and well-being,³ reduce resource utilization and costs,⁴ and, in some studies, increase the length of life for seriously ill patients.⁵

Given the frequency with which seriously ill patients are hospitalized, hospitalists are well positioned to identify those who could benefit from palliative care interventions.⁶ Hospitalists routinely use primary palliative care skills, including pain and symptom management and skilled care planning conversations. For complex cases, such as patients with intractable symptoms or major family conflict, hospitalists may refer to specialist palliative care teams for consultation.

The Society of Hospital Medicine (SHM) defines the key primary palliative care responsibilities for hospitalists as (1) leading discussions on the goals of care and advance care planning with patients and families, (2) screening and treating common physical symptoms, and (3) referring patients to community services to provide support postdischarge.⁷ According to data in the National Palliative Care Registry,⁸ 48% of all palliative care referrals in 2015 came from hospitalists, which is more than double the percentage of referrals from any other specialty.⁹

In a recent survey conducted by SHM about serious illness communication, 53% of hospitalists reported concerns about a patient or family’s understanding of their prognosis, and 50% indicated that they do not feel confident managing family conflict.¹⁰

Context

Patients with multiple serious chronic conditions are often forced to rely on emergency services when crises, such as uncontrolled pain or dyspnea exacerbation, occur after hours, resulting in the revolving-door hospitalizations that typically characterize their care.¹¹ As the prevalence of serious illness rises and the shift to value-based payment accelerates, hospitals are under increasing pressure to deliver efficient and high-quality services that meet the needs of seriously ill patients. The integration of standardized palliative care screening and assessment enables hospitalists and other providers to identify high-need individuals and match services and delivery models to needs, whether it be respite care for an exhausted and overwhelmed family caregiver or a home protocol for managing recurrent dyspnea crises for a patient with chronic obstructive pulmonary disease (COPD). This process improves the quality of care and quality of life, and in doing so, prevents the need for costly crisis care.

Reducing Readmissions

By identifying patients in need of extra symptom management support, or those at a turning point requiring discussion about achievable priorities for care, hospitalists can avert crises for patients earlier in the disease trajectory either by managing the patient’s palliative needs themselves or by connecting patients with specialty palliative care services as needed. This leads to a better quality of life (and survival in some studies) for both patients and their families^1,3,5 and reduces unnecessary emergency department (ED) and hospital use.¹² Hospitalists providing palliative care can also reduce readmissions by improving care coordination, including clinical communication and medication reconciliation after discharge.¹³

A 2015 Harvard Business Review study found that the quality of communication in the hospital is the strongest independent predictor of readmissions when combined with process-of-care improvements, such as standardized patient screening and assessment of family caregiver capacity.¹⁴ While medical education prepares physicians to deliver evidence-based medical care, it currently offers little to no training in communication skills, despite mounting evidence that this is a critical component of quality healthcare.

Cost Savings

Hospital palliative care teams are associated with significant hospital cost savings that result from aligning care with patient priorities, leading, in turn, to reduced nonbeneficial hospital imaging, medications, procedures, and length of stay.¹⁵ See the table^16,17 for examples of cost and quality outcomes of specialist palliative care provision and evidence supporting each outcome.^18-25

Multiple studies consistently demonstrate that inpatient palliative care teams reduce hospital costs.²⁶ One randomized controlled trial investigating the impact of an inpatient palliative care service found that patients who received care from the palliative care team reported greater satisfaction with their care, had fewer intensive care unit admissions, had more advanced directives at hospital discharge, longer hospice length of stay, and lower total healthcare costs (a net difference of $6766 per patient).²³

Research shows that the earlier palliative care is provided, the greater the impact on the subsequent course of care,²⁷ suggesting that hospitalists who provide frontline palliative care interventions as early as possible in a seriously ill patient’s stay will be able to provide higher quality care with lower overall costs. Notably, the majority of research on cost savings associated with palliative care has focused on the impact of specialist palliative care teams, and further research is needed to understand the economic impact of primary palliative care provision.

Improving Satisfaction

Shifting to value-based payment means that the patient and family experience determine an increasingly large percentage of hospital and provider reimbursement. Palliative care approaches, such as family caregiver assessment and support, access to 24/7 assistance after discharge, and person-centered care by an interdisciplinary team, improve performance in all of these measures. Communication skills training improves patient satisfaction scores, and skilled discussions about achievable priorities for care are associated with better quality of life, reduced nonbeneficial and burdensome treatments, and an increase in goal-concordant care.¹⁹ Communication skills training has also been shown to reduce burnout and improve empathy among physicians.^28,29

Though more evidence is needed to understand the impact of primary palliative care provision by hospitalists, the strong evidence on the benefits of specialty palliative care suggests that the skilled provision of primary palliative care by hospitalists will result in higher quality, higher value care. A number of training options exist for midcareer hospital medicine clinicians, including both in-person and online training in communication and other palliative care skills.

• The Center to Advance Palliative Care (CAPC) is a membership organization that offers online continuing education unit and continuing medical education courses on communication skills, pain and symptom management, caregiver support, and care coordination. CAPC also offers courses on palliative interventions for patients with dementia, COPD, and heart failure.
• SHM is actively invested in engaging hospitalists in palliative care skills training. SHM provides free toolkits on a variety of topics within the palliative care domain, including pain management, postacute care transitions, and opioid safety. The recently released Serious Illness Communication toolkit offers background on the role of hospitalists in palliative care provision, a pathway for fitting goals-of-care conversations into hospitalist workflow and recommended metrics and training resources. SHM also uses a mentored implementation model in which expert physicians mentor hospital team members on best practices in palliative care. SHM’s Palliative Care Task Force seeks to identify educational activities for hospitalists and create opportunities to integrate palliative care in hospital medicine.³⁰
• The Serious Illness Care Program at Ariadne Labs in Boston aims to facilitate conversations between clinicians and seriously ill patients through its Serious Illness Conversation Guide, combined with technical assistance on workflow redesign to help clinicians conduct and document serious illness conversations.
• VitalTalk specializes in clinical communication education. Through online and in-person train-the-trainer programs, VitalTalk equips clinicians to lead communication training programs at their home institutions.
• The Education in Palliative and End-of-Life Care Program and End-of-Life Nursing Education Consortium (ELNEC) uses a train-the-trainer approach to educate providers in palliative care clinical competencies and increase the reach of primary palliative care provision. ELNEC workshops are complemented by a curriculum of online clinical training modules.

Though palliative care skills training is a necessary first step, hospitalists also cite lack of time, difficulty finding records of previous patient discussions, and frequent handoffs as among the barriers to integrating palliative care into their practice.¹⁰ Studies examining the process of palliative care and hospital culture change have found that barriers to palliative care integration include a culture of aggressive care in EDs, lack of standardized patient identification criteria, and limited knowledge about and staffing for palliative care.³¹ These data indicate the need for system changes that enable hospitalists to operationalize palliative care principles.

Health systems must implement systems and processes that routinize palliative care, making it part of the mainstream course of care for seriously ill patients and their caregivers. This includes developing systems for the identification of patients with palliative care needs, embedding palliative care assessment and referral into clinical workflows, and enabling standardized palliative care documentation in electronic medical records. While palliative care skills training is essential, investment in systems change is no less critical to embedding palliative care practices in clinical norms across specialties.

Hospitalists can use a palliative approach to improve care quality and quality of life for seriously ill patients while helping to avoid preventable and unnecessary 911 calls, ED visits, and hospitalizations. The shift towards value-based payment is a strong incentive for hospitals and hospitalists to direct resources toward practices that improve the quality of life and care for the highest-need patients and their families. When equipped with the tools they need to provide palliative care, either themselves or in collaboration with palliative care teams, hospitalists have the opportunity to profoundly redirect the experience of care for seriously ill patients and their families.

Disclosure

The authors declared no conflicts of interest.

Pruritus is a subjective report of itching, which can be caused by dermatologic or systemic conditions. Pruritus accounts for about 5% of all skin adverse drug reactions (ADRs) after administration.¹ Mechanisms by which medication-induced pruritus occurs are still unknown and have been understudied. Treatment modalities also have been understudied; however, the understood method for treatment is cessation of the causative agent.²

Anticoagulants commonly are used in several conditions, including prevention of ischemic cerebrovascular accident (CVA) in patients with atrial fibrillation (AF) as well as for the treatment and prevention of deep vein thrombosis (DVT) and pulmonary embolism (PE).³ Traditionally, warfarin was the gold standard anticoagulant. With the relatively recent approval of several direct oral anticoagulants (DOACs), such as rivaroxaban, apixaban, and dabigatran, the landscape of anticoagulation is changing. One benefit of using DOACs as opposed to warfarin is that they often require less frequent laboratory monitoring. However, rare ADRs not detected during clinical trials have appeared following drug approval.⁴

In a VA anticoagulation clinic that managed more than 60 patients taking DOACs, the authors identified 2 cases of pruritus, possibly associated with DOAC agents. These 2 cases point to a higher incidence rate than the rate reported in the rivaroxaban package insert (2%).⁵ Of note, pruritus is not mentioned in the apixaban package insert.⁶

Patient 1 Case Presentation

Patient 1 was a 69-year-old male with AF who was switched to rivaroxaban after 5 years of warfarin therapy. Past medical history included iron deficiency anemia, hypertension, type 2 diabetes mellitus, systolic heart failure, hyperlipidemia, hepatic steatosis, benign prostatic hyperplasia, and gastroesophageal reflux disease. The patient reported “itching all over” soon after initiation of rivaroxaban and that the itching was intolerable and always began 90 to 120 minutes after each dose of rivaroxaban with no associated rash.

After about 6 months of treatment with rivaroxaban, the patient was switched to apixaban; however, the pruritus persisted even after the switch. The onset of itching had similar timing with regard to the apixaban doses. When apixaban was initiated, the patient also was started on amiodarone and tamsulosin. A full pharmacotherapy review of the patient’s medication list for the incidence of pruritus was conducted. Regarding amiodarone and tamsulosin, incidence of pruritus was < 1%.^7,8 Neither agent had yet been started during the rivaroxaban therapy; therefore, it was unlikely that either of these 2 medications were the causative agent of the pruritic ADR.

In response to the itching, the patient was given diphenhydramine 25 mg twice daily, taken with each dose of apixaban. Shortly thereafter, the patient reported that diphenhydramine lessened the severity of the pruritus. He was switched to loratadine 10 mg twice daily with each dose of apixaban, to avoid drowsiness as well as the increased anticholinergic ADRs of first-generation antihistamines. The patient reported that the itching was tolerable.

Patient 2 Case Presentation

Patient 2 was a 63-year-old male with AF and hypertension who was initially started on rivaroxaban and reported pruritus after about 1 month. Despite the uncomfortable itch, the patient elected to continue therapy and began diphenhydramine 25 mg daily with each dose of rivaroxaban. Diphenhydramine seemed to improve the pruritus but did not completely alleviate it. While on rivaroxaban, the patient experienced an acute drop in hemoglobin; however, no source of bleeding was found. Although the pruritus was largely resolved, he was switched to apixaban due to its favorable bleeding profile.⁹ The patient continued to have pruritus on apixaban; however, he reported that pruritus was less severe than it had been while taking rivaroxaban. The patient restarted on diphenhydramine twice daily with each dose of apixaban and reported cessation of pruritus.

Discussion

After observing both cases in relation to the timing between the administration of a DOAC and onset of pruritus, it seemed likely that the causative agent could be the DOAC. A Naranjo Nomogram was used to determine the likeliness of each drug to be the causative agent of the ADR.¹⁰ This nomogram is scaled from a low score of -4 to a high score of 13. Patients 1 and 2 had a score of 4, which is reflective of a possible ADR (score 1-4). Using the nomogram as well as the subjective information provided by the patients, it is reasonable to conclude that the pruritus was possibly associated with the use of the DOACs. Nonadherence to anticoagulants may lead to potentially fatal adverse outcomes, such as stroke. Medication-associated pruritus could lead to medication nonadherence if left unaddressed. It is notable that prescribing an antihistamine that is taken at the time of the anticoagulant dose allowed these patients with possible DOAC-associated pruritus to continue therapy with the selected anticoagulant. Further research on this topic is needed.

Pruritus is a subjective report of itching, which can be caused by dermatologic or systemic conditions. Pruritus accounts for about 5% of all skin adverse drug reactions (ADRs) after administration.¹ Mechanisms by which medication-induced pruritus occurs are still unknown and have been understudied. Treatment modalities also have been understudied; however, the understood method for treatment is cessation of the causative agent.²

Anticoagulants commonly are used in several conditions, including prevention of ischemic cerebrovascular accident (CVA) in patients with atrial fibrillation (AF) as well as for the treatment and prevention of deep vein thrombosis (DVT) and pulmonary embolism (PE).³ Traditionally, warfarin was the gold standard anticoagulant. With the relatively recent approval of several direct oral anticoagulants (DOACs), such as rivaroxaban, apixaban, and dabigatran, the landscape of anticoagulation is changing. One benefit of using DOACs as opposed to warfarin is that they often require less frequent laboratory monitoring. However, rare ADRs not detected during clinical trials have appeared following drug approval.⁴

In a VA anticoagulation clinic that managed more than 60 patients taking DOACs, the authors identified 2 cases of pruritus, possibly associated with DOAC agents. These 2 cases point to a higher incidence rate than the rate reported in the rivaroxaban package insert (2%).⁵ Of note, pruritus is not mentioned in the apixaban package insert.⁶

Patient 1 Case Presentation

Patient 1 was a 69-year-old male with AF who was switched to rivaroxaban after 5 years of warfarin therapy. Past medical history included iron deficiency anemia, hypertension, type 2 diabetes mellitus, systolic heart failure, hyperlipidemia, hepatic steatosis, benign prostatic hyperplasia, and gastroesophageal reflux disease. The patient reported “itching all over” soon after initiation of rivaroxaban and that the itching was intolerable and always began 90 to 120 minutes after each dose of rivaroxaban with no associated rash.

After about 6 months of treatment with rivaroxaban, the patient was switched to apixaban; however, the pruritus persisted even after the switch. The onset of itching had similar timing with regard to the apixaban doses. When apixaban was initiated, the patient also was started on amiodarone and tamsulosin. A full pharmacotherapy review of the patient’s medication list for the incidence of pruritus was conducted. Regarding amiodarone and tamsulosin, incidence of pruritus was < 1%.^7,8 Neither agent had yet been started during the rivaroxaban therapy; therefore, it was unlikely that either of these 2 medications were the causative agent of the pruritic ADR.

In response to the itching, the patient was given diphenhydramine 25 mg twice daily, taken with each dose of apixaban. Shortly thereafter, the patient reported that diphenhydramine lessened the severity of the pruritus. He was switched to loratadine 10 mg twice daily with each dose of apixaban, to avoid drowsiness as well as the increased anticholinergic ADRs of first-generation antihistamines. The patient reported that the itching was tolerable.

Patient 2 Case Presentation

Patient 2 was a 63-year-old male with AF and hypertension who was initially started on rivaroxaban and reported pruritus after about 1 month. Despite the uncomfortable itch, the patient elected to continue therapy and began diphenhydramine 25 mg daily with each dose of rivaroxaban. Diphenhydramine seemed to improve the pruritus but did not completely alleviate it. While on rivaroxaban, the patient experienced an acute drop in hemoglobin; however, no source of bleeding was found. Although the pruritus was largely resolved, he was switched to apixaban due to its favorable bleeding profile.⁹ The patient continued to have pruritus on apixaban; however, he reported that pruritus was less severe than it had been while taking rivaroxaban. The patient restarted on diphenhydramine twice daily with each dose of apixaban and reported cessation of pruritus.

Discussion

After observing both cases in relation to the timing between the administration of a DOAC and onset of pruritus, it seemed likely that the causative agent could be the DOAC. A Naranjo Nomogram was used to determine the likeliness of each drug to be the causative agent of the ADR.¹⁰ This nomogram is scaled from a low score of -4 to a high score of 13. Patients 1 and 2 had a score of 4, which is reflective of a possible ADR (score 1-4). Using the nomogram as well as the subjective information provided by the patients, it is reasonable to conclude that the pruritus was possibly associated with the use of the DOACs. Nonadherence to anticoagulants may lead to potentially fatal adverse outcomes, such as stroke. Medication-associated pruritus could lead to medication nonadherence if left unaddressed. It is notable that prescribing an antihistamine that is taken at the time of the anticoagulant dose allowed these patients with possible DOAC-associated pruritus to continue therapy with the selected anticoagulant. Further research on this topic is needed.

Pruritus is a subjective report of itching, which can be caused by dermatologic or systemic conditions. Pruritus accounts for about 5% of all skin adverse drug reactions (ADRs) after administration.¹ Mechanisms by which medication-induced pruritus occurs are still unknown and have been understudied. Treatment modalities also have been understudied; however, the understood method for treatment is cessation of the causative agent.²

Anticoagulants commonly are used in several conditions, including prevention of ischemic cerebrovascular accident (CVA) in patients with atrial fibrillation (AF) as well as for the treatment and prevention of deep vein thrombosis (DVT) and pulmonary embolism (PE).³ Traditionally, warfarin was the gold standard anticoagulant. With the relatively recent approval of several direct oral anticoagulants (DOACs), such as rivaroxaban, apixaban, and dabigatran, the landscape of anticoagulation is changing. One benefit of using DOACs as opposed to warfarin is that they often require less frequent laboratory monitoring. However, rare ADRs not detected during clinical trials have appeared following drug approval.⁴

In a VA anticoagulation clinic that managed more than 60 patients taking DOACs, the authors identified 2 cases of pruritus, possibly associated with DOAC agents. These 2 cases point to a higher incidence rate than the rate reported in the rivaroxaban package insert (2%).⁵ Of note, pruritus is not mentioned in the apixaban package insert.⁶

Patient 1 Case Presentation

Patient 1 was a 69-year-old male with AF who was switched to rivaroxaban after 5 years of warfarin therapy. Past medical history included iron deficiency anemia, hypertension, type 2 diabetes mellitus, systolic heart failure, hyperlipidemia, hepatic steatosis, benign prostatic hyperplasia, and gastroesophageal reflux disease. The patient reported “itching all over” soon after initiation of rivaroxaban and that the itching was intolerable and always began 90 to 120 minutes after each dose of rivaroxaban with no associated rash.

After about 6 months of treatment with rivaroxaban, the patient was switched to apixaban; however, the pruritus persisted even after the switch. The onset of itching had similar timing with regard to the apixaban doses. When apixaban was initiated, the patient also was started on amiodarone and tamsulosin. A full pharmacotherapy review of the patient’s medication list for the incidence of pruritus was conducted. Regarding amiodarone and tamsulosin, incidence of pruritus was < 1%.^7,8 Neither agent had yet been started during the rivaroxaban therapy; therefore, it was unlikely that either of these 2 medications were the causative agent of the pruritic ADR.

In response to the itching, the patient was given diphenhydramine 25 mg twice daily, taken with each dose of apixaban. Shortly thereafter, the patient reported that diphenhydramine lessened the severity of the pruritus. He was switched to loratadine 10 mg twice daily with each dose of apixaban, to avoid drowsiness as well as the increased anticholinergic ADRs of first-generation antihistamines. The patient reported that the itching was tolerable.

Patient 2 Case Presentation

Patient 2 was a 63-year-old male with AF and hypertension who was initially started on rivaroxaban and reported pruritus after about 1 month. Despite the uncomfortable itch, the patient elected to continue therapy and began diphenhydramine 25 mg daily with each dose of rivaroxaban. Diphenhydramine seemed to improve the pruritus but did not completely alleviate it. While on rivaroxaban, the patient experienced an acute drop in hemoglobin; however, no source of bleeding was found. Although the pruritus was largely resolved, he was switched to apixaban due to its favorable bleeding profile.⁹ The patient continued to have pruritus on apixaban; however, he reported that pruritus was less severe than it had been while taking rivaroxaban. The patient restarted on diphenhydramine twice daily with each dose of apixaban and reported cessation of pruritus.

Discussion

After observing both cases in relation to the timing between the administration of a DOAC and onset of pruritus, it seemed likely that the causative agent could be the DOAC. A Naranjo Nomogram was used to determine the likeliness of each drug to be the causative agent of the ADR.¹⁰ This nomogram is scaled from a low score of -4 to a high score of 13. Patients 1 and 2 had a score of 4, which is reflective of a possible ADR (score 1-4). Using the nomogram as well as the subjective information provided by the patients, it is reasonable to conclude that the pruritus was possibly associated with the use of the DOACs. Nonadherence to anticoagulants may lead to potentially fatal adverse outcomes, such as stroke. Medication-associated pruritus could lead to medication nonadherence if left unaddressed. It is notable that prescribing an antihistamine that is taken at the time of the anticoagulant dose allowed these patients with possible DOAC-associated pruritus to continue therapy with the selected anticoagulant. Further research on this topic is needed.