A 68‐year‐old man with a history of congestive heart failure and hypertension presented to the emergency department with fatigue and dyspnea of 3 weeks duration. Physical examination was consistent with heart failure. In addition, a right upper extremity resting tremor was noticed. An electrocardiogram (ECG) revealed an atrial flutter with a conduction ratio of 4:1 (Figure 1A). He denied palpitations or a previous history of atrial flutter/fibrillation. Unlike typical atrial flutter, these flutter like waves were distinctly absent in lead III, the only limb lead not connected to the right arm.

Figure 1

While holding the patient's right arm to control the tremor, a second ECG tracing was obtained. As expected the flutter like waves disappeared (Figure 1B). These ECG findings were attributed to the patient's tremor. A neurological consultation established a clinical diagnosis of Parkinson's disease. His congestive heart failure (CHF) was treated with increasing diuretics and appropriate treatment for Parkinson's disease was initiated.

A 68‐year‐old man with a history of congestive heart failure and hypertension presented to the emergency department with fatigue and dyspnea of 3 weeks duration. Physical examination was consistent with heart failure. In addition, a right upper extremity resting tremor was noticed. An electrocardiogram (ECG) revealed an atrial flutter with a conduction ratio of 4:1 (Figure 1A). He denied palpitations or a previous history of atrial flutter/fibrillation. Unlike typical atrial flutter, these flutter like waves were distinctly absent in lead III, the only limb lead not connected to the right arm.

Figure 1

While holding the patient's right arm to control the tremor, a second ECG tracing was obtained. As expected the flutter like waves disappeared (Figure 1B). These ECG findings were attributed to the patient's tremor. A neurological consultation established a clinical diagnosis of Parkinson's disease. His congestive heart failure (CHF) was treated with increasing diuretics and appropriate treatment for Parkinson's disease was initiated.

A 68‐year‐old man with a history of congestive heart failure and hypertension presented to the emergency department with fatigue and dyspnea of 3 weeks duration. Physical examination was consistent with heart failure. In addition, a right upper extremity resting tremor was noticed. An electrocardiogram (ECG) revealed an atrial flutter with a conduction ratio of 4:1 (Figure 1A). He denied palpitations or a previous history of atrial flutter/fibrillation. Unlike typical atrial flutter, these flutter like waves were distinctly absent in lead III, the only limb lead not connected to the right arm.

Figure 1

While holding the patient's right arm to control the tremor, a second ECG tracing was obtained. As expected the flutter like waves disappeared (Figure 1B). These ECG findings were attributed to the patient's tremor. A neurological consultation established a clinical diagnosis of Parkinson's disease. His congestive heart failure (CHF) was treated with increasing diuretics and appropriate treatment for Parkinson's disease was initiated.

Quality of care in US hospitals is inconsistent and often below accepted standards.1 This observation has catalyzed a number of performance measurement initiatives intended to publicize gaps and spur quality improvement.2 As the field has evolved, organizational factors such as teaching status, ownership model, nurse staffing levels, and hospital volume have been found to be associated with performance on quality measures.1, 3‐7 Hospitalists represent a more recent change in the organization of inpatient care8 that may impact hospital‐level performance. In fact, most hospitals provide financial support to hospitalists, not only for hopes of improving efficiency, but also for improving quality and safety.9

Only a few single‐site studies have examined the impact of hospitalists on quality of care for common medical conditions (ie, pneumonia, congestive heart failure, and acute myocardial infarction), and each has focused on patient‐level effects. Rifkin et al.10, 11 did not find differences between hospitalists' and nonhospitalists' patients in terms of pneumonia process measures. Roytman et al.12 found hospitalists more frequently prescribed afterload‐reducing agents for congestive heart failure (CHF), but other studies have shown no differences in care quality for heart failure.13, 14 Importantly, no studies have examined the role of hospitalists in the care of patients with acute myocardial infarction (AMI). In addition, studies have not addressed the effect of hospitalists at the hospital level to understand whether hospitalists have broader system‐level effects reflected by overall hospital performance.

We hypothesized that the presence of hospitalists within a hospital would be associated with improvements in hospital‐level adherence to publicly reported quality process measures, and having a greater percentage of patients admitted by hospitalists would be associated with improved performance. To test these hypotheses, we linked data from a statewide census of hospitalists with data collected as part of a hospital quality‐reporting initiative.

Materials and Methods

Results

Discussion

In this cross‐sectional analysis of hospitals participating in a voluntary quality reporting initiative, hospitals with at least 1 hospitalist group had fewer missed discharge care process measures for CHF, even after adjusting for hospital‐level characteristics. In addition, as the estimated percentage of patients admitted by hospitalists increased, the percentage of missed quality opportunities decreased across all measures. The observed relationships were most apparent for measures that could be completed at any time during the hospitalization and at discharge. While it is likely that hospitalists are a marker of a hospital's ability to invest in systems (and as a result, care improvement initiatives), the presence of a potential dose‐response relationship suggests that hospitalists themselves may have a role in improving processes of care.

Our study suggests a generally positive, but mixed, picture of hospitalists' effects on quality process measure performance. Lack of uniformity across measures may depend on the timing of the process measure (eg, whether or not the process is measured at admission or discharge). For example, in contrast to admission process measures, we more commonly observed a positive association between hospitalists and care quality on process measures targeting processes that generally took place later in hospitalization or at discharge. Many admission process measures (eg, door to antibiotic time, blood cultures, and appropriate initial antibiotics) likely occurred prior to hospitalist involvement in most cases and were instead under the direction of emergency medicine physicians. Performance on these measures would not be expected to relate to use of hospitalists, and that is what we observed.

In addition to the timing of when a process was measured or took place, associations between hospitalists and care quality vary by disease. The apparent variation in impact of hospitalists by disease (more impact for cardiac conditions, less for pneumonia) may relate primarily to the characteristics of the processes of care that were measured for each condition. For example, one‐half of the pneumonia process measures related to care occurring within a few hours of admission, while the other one‐half (smoking cessation advice and streptococcal and influenza vaccines) were often administered per protocol or by nonphysician providers.26‐29 However, more of the cardiac measures required physician action (eg, prescription of an ACE‐I at discharge). Alternatively, unmeasured confounders important in the delivery of cardiac care might play an important role in the relationship between hospitalists and cardiac process measure performance.

Our approach to defining hospitalists bears mention as well. While a dichotomous measure of having hospitalists available was only statistically significant for the single CHF discharge composite measure, our measure of hospitalist availabilitythe percentage of patients admitted by hospitalistswas more strongly associated with a larger number of quality measures. Contrast between the dichotomous and continuous measures may have statistical explanations (the power to see differences between 2 groups is more limited with use of a binary predictor, which itself can be subject to bias),30 but may also indicate a dose‐response relationship. A larger number of admissions to hospitalists may help standardize practices, as care is concentrated in a smaller number of physicians' hands. Moreover, larger hospitalist programs may be more likely to have implemented care standardization or quality improvement processes or to have been incorporated into (or lead) hospitals' quality infrastructures. Finally, presence of larger hospitalist groups may be a marker for a hospital's capacity to make hospital‐wide investments in improvement. However, the association between the percentage of patients admitted by hospitalists and care quality persisted even after adjustment for many measures plausibly associated with ability to invest in care quality.

Our study has several limitations. First, although we used a widely accepted definition of hospitalists endorsed by the Society of Hospital Medicine, there are no gold standard definitions for a hospitalist's job description or skill set. As a result, it is possible that a model utilizing rotating internists (from a multispecialty group) might have been misidentified as a hospitalist model. Second, our findings represent a convenience sample of hospitals in a voluntary reporting initiative (CHART) and may not be applicable to hospitals that are less able to participate in such an endeavor. CHART hospitals are recognized to be better performers than the overall California population of hospitals, potentially decreasing variability in our quality of care measures.2 Third, there were significant differences between our comparison groups within the CHART hospitals, including sample size. Although we attempted to adjust our analyses for many important potential confounders and applied conservative measures to assess statistical significance, given the baseline differences, we cannot rule out the possibility of residual confounding by unmeasured factors. Fourth, as described above, this observational study cannot provide robust evidence to support conclusions regarding causality. Fifth, the estimation of the percent of patients admitted by hospitalists is unvalidated and based upon self‐reported and incomplete (41% of respondents) data. We are somewhat reassured by the fact that respondents and nonresponders were similar across all hospital characteristics, as well as outcomes. Sixth, misclassification of the estimated percentage of patients admitted by hospitalists may have influenced our results. Although possible, misclassification often biases results toward the null, potentially weakening any observed association. Given that our respondents were not aware of our hypotheses, there is no reason to expect recall issues to bias the results one way or the other. Finally, for many performance measures, overall performance was excellent among all hospitals (eg, aspirin at admission) with limited variability, thus limiting the ability to assess for differences.

In summary, in a large, cross‐sectional study of California hospitals participating in a voluntary quality reporting initiative, the presence of hospitalists was associated with modest improvements in hospital‐level performance of quality process measures. In addition, we found a relationship between the percentage of patients admitted by hospitalists and improved process measure adherence. Although we cannot determine causality, our data support the hypothesis that dedicated hospital physicians can positively affect the quality of care. Future research should examine this relationship in other settings and should address causality using broader measures of quality including both processes and outcomes.

Quality of care in US hospitals is inconsistent and often below accepted standards.1 This observation has catalyzed a number of performance measurement initiatives intended to publicize gaps and spur quality improvement.2 As the field has evolved, organizational factors such as teaching status, ownership model, nurse staffing levels, and hospital volume have been found to be associated with performance on quality measures.1, 3‐7 Hospitalists represent a more recent change in the organization of inpatient care8 that may impact hospital‐level performance. In fact, most hospitals provide financial support to hospitalists, not only for hopes of improving efficiency, but also for improving quality and safety.9

Only a few single‐site studies have examined the impact of hospitalists on quality of care for common medical conditions (ie, pneumonia, congestive heart failure, and acute myocardial infarction), and each has focused on patient‐level effects. Rifkin et al.10, 11 did not find differences between hospitalists' and nonhospitalists' patients in terms of pneumonia process measures. Roytman et al.12 found hospitalists more frequently prescribed afterload‐reducing agents for congestive heart failure (CHF), but other studies have shown no differences in care quality for heart failure.13, 14 Importantly, no studies have examined the role of hospitalists in the care of patients with acute myocardial infarction (AMI). In addition, studies have not addressed the effect of hospitalists at the hospital level to understand whether hospitalists have broader system‐level effects reflected by overall hospital performance.

We hypothesized that the presence of hospitalists within a hospital would be associated with improvements in hospital‐level adherence to publicly reported quality process measures, and having a greater percentage of patients admitted by hospitalists would be associated with improved performance. To test these hypotheses, we linked data from a statewide census of hospitalists with data collected as part of a hospital quality‐reporting initiative.

Materials and Methods

Results

Discussion

In this cross‐sectional analysis of hospitals participating in a voluntary quality reporting initiative, hospitals with at least 1 hospitalist group had fewer missed discharge care process measures for CHF, even after adjusting for hospital‐level characteristics. In addition, as the estimated percentage of patients admitted by hospitalists increased, the percentage of missed quality opportunities decreased across all measures. The observed relationships were most apparent for measures that could be completed at any time during the hospitalization and at discharge. While it is likely that hospitalists are a marker of a hospital's ability to invest in systems (and as a result, care improvement initiatives), the presence of a potential dose‐response relationship suggests that hospitalists themselves may have a role in improving processes of care.

Our study suggests a generally positive, but mixed, picture of hospitalists' effects on quality process measure performance. Lack of uniformity across measures may depend on the timing of the process measure (eg, whether or not the process is measured at admission or discharge). For example, in contrast to admission process measures, we more commonly observed a positive association between hospitalists and care quality on process measures targeting processes that generally took place later in hospitalization or at discharge. Many admission process measures (eg, door to antibiotic time, blood cultures, and appropriate initial antibiotics) likely occurred prior to hospitalist involvement in most cases and were instead under the direction of emergency medicine physicians. Performance on these measures would not be expected to relate to use of hospitalists, and that is what we observed.

In addition to the timing of when a process was measured or took place, associations between hospitalists and care quality vary by disease. The apparent variation in impact of hospitalists by disease (more impact for cardiac conditions, less for pneumonia) may relate primarily to the characteristics of the processes of care that were measured for each condition. For example, one‐half of the pneumonia process measures related to care occurring within a few hours of admission, while the other one‐half (smoking cessation advice and streptococcal and influenza vaccines) were often administered per protocol or by nonphysician providers.26‐29 However, more of the cardiac measures required physician action (eg, prescription of an ACE‐I at discharge). Alternatively, unmeasured confounders important in the delivery of cardiac care might play an important role in the relationship between hospitalists and cardiac process measure performance.

Our approach to defining hospitalists bears mention as well. While a dichotomous measure of having hospitalists available was only statistically significant for the single CHF discharge composite measure, our measure of hospitalist availabilitythe percentage of patients admitted by hospitalistswas more strongly associated with a larger number of quality measures. Contrast between the dichotomous and continuous measures may have statistical explanations (the power to see differences between 2 groups is more limited with use of a binary predictor, which itself can be subject to bias),30 but may also indicate a dose‐response relationship. A larger number of admissions to hospitalists may help standardize practices, as care is concentrated in a smaller number of physicians' hands. Moreover, larger hospitalist programs may be more likely to have implemented care standardization or quality improvement processes or to have been incorporated into (or lead) hospitals' quality infrastructures. Finally, presence of larger hospitalist groups may be a marker for a hospital's capacity to make hospital‐wide investments in improvement. However, the association between the percentage of patients admitted by hospitalists and care quality persisted even after adjustment for many measures plausibly associated with ability to invest in care quality.

Our study has several limitations. First, although we used a widely accepted definition of hospitalists endorsed by the Society of Hospital Medicine, there are no gold standard definitions for a hospitalist's job description or skill set. As a result, it is possible that a model utilizing rotating internists (from a multispecialty group) might have been misidentified as a hospitalist model. Second, our findings represent a convenience sample of hospitals in a voluntary reporting initiative (CHART) and may not be applicable to hospitals that are less able to participate in such an endeavor. CHART hospitals are recognized to be better performers than the overall California population of hospitals, potentially decreasing variability in our quality of care measures.2 Third, there were significant differences between our comparison groups within the CHART hospitals, including sample size. Although we attempted to adjust our analyses for many important potential confounders and applied conservative measures to assess statistical significance, given the baseline differences, we cannot rule out the possibility of residual confounding by unmeasured factors. Fourth, as described above, this observational study cannot provide robust evidence to support conclusions regarding causality. Fifth, the estimation of the percent of patients admitted by hospitalists is unvalidated and based upon self‐reported and incomplete (41% of respondents) data. We are somewhat reassured by the fact that respondents and nonresponders were similar across all hospital characteristics, as well as outcomes. Sixth, misclassification of the estimated percentage of patients admitted by hospitalists may have influenced our results. Although possible, misclassification often biases results toward the null, potentially weakening any observed association. Given that our respondents were not aware of our hypotheses, there is no reason to expect recall issues to bias the results one way or the other. Finally, for many performance measures, overall performance was excellent among all hospitals (eg, aspirin at admission) with limited variability, thus limiting the ability to assess for differences.

In summary, in a large, cross‐sectional study of California hospitals participating in a voluntary quality reporting initiative, the presence of hospitalists was associated with modest improvements in hospital‐level performance of quality process measures. In addition, we found a relationship between the percentage of patients admitted by hospitalists and improved process measure adherence. Although we cannot determine causality, our data support the hypothesis that dedicated hospital physicians can positively affect the quality of care. Future research should examine this relationship in other settings and should address causality using broader measures of quality including both processes and outcomes.

Quality of care in US hospitals is inconsistent and often below accepted standards.1 This observation has catalyzed a number of performance measurement initiatives intended to publicize gaps and spur quality improvement.2 As the field has evolved, organizational factors such as teaching status, ownership model, nurse staffing levels, and hospital volume have been found to be associated with performance on quality measures.1, 3‐7 Hospitalists represent a more recent change in the organization of inpatient care8 that may impact hospital‐level performance. In fact, most hospitals provide financial support to hospitalists, not only for hopes of improving efficiency, but also for improving quality and safety.9

Only a few single‐site studies have examined the impact of hospitalists on quality of care for common medical conditions (ie, pneumonia, congestive heart failure, and acute myocardial infarction), and each has focused on patient‐level effects. Rifkin et al.10, 11 did not find differences between hospitalists' and nonhospitalists' patients in terms of pneumonia process measures. Roytman et al.12 found hospitalists more frequently prescribed afterload‐reducing agents for congestive heart failure (CHF), but other studies have shown no differences in care quality for heart failure.13, 14 Importantly, no studies have examined the role of hospitalists in the care of patients with acute myocardial infarction (AMI). In addition, studies have not addressed the effect of hospitalists at the hospital level to understand whether hospitalists have broader system‐level effects reflected by overall hospital performance.

We hypothesized that the presence of hospitalists within a hospital would be associated with improvements in hospital‐level adherence to publicly reported quality process measures, and having a greater percentage of patients admitted by hospitalists would be associated with improved performance. To test these hypotheses, we linked data from a statewide census of hospitalists with data collected as part of a hospital quality‐reporting initiative.

Materials and Methods

Results

Discussion

In this cross‐sectional analysis of hospitals participating in a voluntary quality reporting initiative, hospitals with at least 1 hospitalist group had fewer missed discharge care process measures for CHF, even after adjusting for hospital‐level characteristics. In addition, as the estimated percentage of patients admitted by hospitalists increased, the percentage of missed quality opportunities decreased across all measures. The observed relationships were most apparent for measures that could be completed at any time during the hospitalization and at discharge. While it is likely that hospitalists are a marker of a hospital's ability to invest in systems (and as a result, care improvement initiatives), the presence of a potential dose‐response relationship suggests that hospitalists themselves may have a role in improving processes of care.

Our study suggests a generally positive, but mixed, picture of hospitalists' effects on quality process measure performance. Lack of uniformity across measures may depend on the timing of the process measure (eg, whether or not the process is measured at admission or discharge). For example, in contrast to admission process measures, we more commonly observed a positive association between hospitalists and care quality on process measures targeting processes that generally took place later in hospitalization or at discharge. Many admission process measures (eg, door to antibiotic time, blood cultures, and appropriate initial antibiotics) likely occurred prior to hospitalist involvement in most cases and were instead under the direction of emergency medicine physicians. Performance on these measures would not be expected to relate to use of hospitalists, and that is what we observed.

In addition to the timing of when a process was measured or took place, associations between hospitalists and care quality vary by disease. The apparent variation in impact of hospitalists by disease (more impact for cardiac conditions, less for pneumonia) may relate primarily to the characteristics of the processes of care that were measured for each condition. For example, one‐half of the pneumonia process measures related to care occurring within a few hours of admission, while the other one‐half (smoking cessation advice and streptococcal and influenza vaccines) were often administered per protocol or by nonphysician providers.26‐29 However, more of the cardiac measures required physician action (eg, prescription of an ACE‐I at discharge). Alternatively, unmeasured confounders important in the delivery of cardiac care might play an important role in the relationship between hospitalists and cardiac process measure performance.

Our approach to defining hospitalists bears mention as well. While a dichotomous measure of having hospitalists available was only statistically significant for the single CHF discharge composite measure, our measure of hospitalist availabilitythe percentage of patients admitted by hospitalistswas more strongly associated with a larger number of quality measures. Contrast between the dichotomous and continuous measures may have statistical explanations (the power to see differences between 2 groups is more limited with use of a binary predictor, which itself can be subject to bias),30 but may also indicate a dose‐response relationship. A larger number of admissions to hospitalists may help standardize practices, as care is concentrated in a smaller number of physicians' hands. Moreover, larger hospitalist programs may be more likely to have implemented care standardization or quality improvement processes or to have been incorporated into (or lead) hospitals' quality infrastructures. Finally, presence of larger hospitalist groups may be a marker for a hospital's capacity to make hospital‐wide investments in improvement. However, the association between the percentage of patients admitted by hospitalists and care quality persisted even after adjustment for many measures plausibly associated with ability to invest in care quality.

Our study has several limitations. First, although we used a widely accepted definition of hospitalists endorsed by the Society of Hospital Medicine, there are no gold standard definitions for a hospitalist's job description or skill set. As a result, it is possible that a model utilizing rotating internists (from a multispecialty group) might have been misidentified as a hospitalist model. Second, our findings represent a convenience sample of hospitals in a voluntary reporting initiative (CHART) and may not be applicable to hospitals that are less able to participate in such an endeavor. CHART hospitals are recognized to be better performers than the overall California population of hospitals, potentially decreasing variability in our quality of care measures.2 Third, there were significant differences between our comparison groups within the CHART hospitals, including sample size. Although we attempted to adjust our analyses for many important potential confounders and applied conservative measures to assess statistical significance, given the baseline differences, we cannot rule out the possibility of residual confounding by unmeasured factors. Fourth, as described above, this observational study cannot provide robust evidence to support conclusions regarding causality. Fifth, the estimation of the percent of patients admitted by hospitalists is unvalidated and based upon self‐reported and incomplete (41% of respondents) data. We are somewhat reassured by the fact that respondents and nonresponders were similar across all hospital characteristics, as well as outcomes. Sixth, misclassification of the estimated percentage of patients admitted by hospitalists may have influenced our results. Although possible, misclassification often biases results toward the null, potentially weakening any observed association. Given that our respondents were not aware of our hypotheses, there is no reason to expect recall issues to bias the results one way or the other. Finally, for many performance measures, overall performance was excellent among all hospitals (eg, aspirin at admission) with limited variability, thus limiting the ability to assess for differences.

In summary, in a large, cross‐sectional study of California hospitals participating in a voluntary quality reporting initiative, the presence of hospitalists was associated with modest improvements in hospital‐level performance of quality process measures. In addition, we found a relationship between the percentage of patients admitted by hospitalists and improved process measure adherence. Although we cannot determine causality, our data support the hypothesis that dedicated hospital physicians can positively affect the quality of care. Future research should examine this relationship in other settings and should address causality using broader measures of quality including both processes and outcomes.

Hello Mrs. K, I'm Dr. Baru, I said. I squeezed the hand limply resting on her cover with my own gloved one.

The room was quiet except for the sighing of her ventilator in the background. She had a broad round face with high cheek bones. Her skin was wrinkle‐free except around the eyes and the corners of her mouth. She breathed peacefully through her tracheostomy. She slowly nodded her head when I squeezed her hand and her blue eyes shone as she smiled broadly. Her son stood at my side, his mouth set in a straight line his eyes gazing intently at his mom. His handshake was firm and brisk, a single downstroke.

I was called in as the Palliative Care consultant. She's in denial, I was told. There's nothing more that we can do for her here. She had already been in the hospital for 2 months but this was the first time I was meeting her and her son. With the help of the Polish interpreter and her son, who had become adept at reading her lips and translating her breathy rasps, I began to sift through all of the information that they had been told, all of the information they had gathered on their own, and what they understood.

Mrs. K was 59 years old. She'd given up her job as a kindergarten teacher and come from Poland 3 years ago to help care for her first grandchild. Two years later, her daughter‐in‐law delivered 2 more grandchildren: twins. Just weeks after the delivery, Mrs. K was diagnosed with multiple myeloma. She was told that her prognosis was good and that, with chemotherapy, she had years to live. She thought about returning home but she felt fine. Though she didn't yet qualify for any kind of insurance, she was getting good care in our County health system. Besides, her grandchildren meant everything to her and her son and daughter‐in‐law needed her now more than ever.

Five months into her treatment she developed pain in her neck and started noticing numbness in her hands. She immediately went to the hospital where she was found to have a tumor in her cervical spine. Despite early radiation and surgery, she was completely paralyzed and dependent on mechanical ventilation within a week. In a matter of days she had been torn from life as she knew it.

It became clear during our conversation that, though Mrs. K was not physically uncomfortable, being confined to the hospital was difficult for her. Her grandchildren couldn't visit because she was on contact precautions. She missed them deeply. She missed sitting on her porch drinking her morning coffee. Unable to move her head, she spent most of her day staring at the ceiling or at the TV watching shows in a language she couldn't understand. She had met countless doctors, nurses, and medical personnel, endured multiple complications, including a pulseless arrest, and had been placed in 3 different ICUs. Yet she was unwavering in her desire to remain on the ventilator and continue doing everything.

Both she and her son expected that the treatments she had been getting would help get her off the ventilator so that she could go home. I struggled to balance their hopes with the information at hand, exploring realistic goals.

Could she go to a nursing home and wait and see if she will recover? I know they have these for people that are on ventilators, her son asked.

This was not going to be possible given her disease. Her paralysis was complete and permanent. She would not recover the ability to breathe on her own. Besides, without insurance they would have to pay for these services. It was not realistic to hope for this.

Could we fly her back to Poland so that she can die and be buried there?

She was not stable enough to fly without medical assistance. Furthermore, she would need to be in contact isolation given the virulence of her uniquely resistant bacteria. The cost to arrange an air ambulance was exorbitant and unaffordable on her son's electrician salary. It was not realistic to hope for this.

Well, if she can't go back to Poland and can't get off of the ventilator, could we set up a ventilator at home so that we can care for her there until she dies?

I explained that this would require help from medical personnel trained in ventilator management. They would also need assistance from individuals trained in palliative care to insure that Mrs. K had adequate nursing and symptom relief, particularly in the event of a medical emergency or a complication with the ventilator. The family would be required to pay for these services out of pocket. After spending hours contacting hospice agencies, medical suppliers, nursing agencies, friends of the family, and community organizations it became clear that even though the family was willing to bankrupt themselves for their mother's care there was not a safe, affordable solution. It was not realistic to hope for this.

What can you do for me? Mrs. K asked.

My tongue sat numbly in my mouth. I felt a tide of shame and sorrow rising. Nothing! I thought, knowing not to say it. At times like this, we often fall back on training; I tried to force her into the round hole.

We can care for you here. We can insure that you are as comfortable as possible for whatever time you have left. We can shift our focus from life‐prolonging treatments to those that are purely focused on your comfort. We can stop those treatments that will interrupt the time that you spend with your family, and try to give you and your family as much space as possible to be together here in the ICU. As your death approaches we would work to keep you as comfortable and peaceful as possible and to allow your family to be with you here at your bedside, but we wouldn't try to prolong the dying process.

I don't WANT to die. I WANT to go home, I WANT to smell fresh air. I'm willing to take risks with my life for that, but not otherwise.

My impotence, my inability to give this woman any of the things that she was hoping for was overwhelming. Her suffering was overwhelming. I was unable to find a thread of hope in her world. I was failing my patient.

Feeling a sense of despair, I told her, I cannot help you breathe on your own or smell the fresh air or be with your grandchildren. None of those goals are realistic. I don't know what I can do for you, but I would like to continue to see you. I want to come back tomorrow.

I hope you do, she said, her blue eyes shining as she smiled softly.

I was struck by the words. Though my loftier goals had been frustrated, I realized that my efforts and my presence at her bedside alone were easing her sufferingthis is something that we all hope for.

Hello Mrs. K, I'm Dr. Baru, I said. I squeezed the hand limply resting on her cover with my own gloved one.

The room was quiet except for the sighing of her ventilator in the background. She had a broad round face with high cheek bones. Her skin was wrinkle‐free except around the eyes and the corners of her mouth. She breathed peacefully through her tracheostomy. She slowly nodded her head when I squeezed her hand and her blue eyes shone as she smiled broadly. Her son stood at my side, his mouth set in a straight line his eyes gazing intently at his mom. His handshake was firm and brisk, a single downstroke.

I was called in as the Palliative Care consultant. She's in denial, I was told. There's nothing more that we can do for her here. She had already been in the hospital for 2 months but this was the first time I was meeting her and her son. With the help of the Polish interpreter and her son, who had become adept at reading her lips and translating her breathy rasps, I began to sift through all of the information that they had been told, all of the information they had gathered on their own, and what they understood.

Mrs. K was 59 years old. She'd given up her job as a kindergarten teacher and come from Poland 3 years ago to help care for her first grandchild. Two years later, her daughter‐in‐law delivered 2 more grandchildren: twins. Just weeks after the delivery, Mrs. K was diagnosed with multiple myeloma. She was told that her prognosis was good and that, with chemotherapy, she had years to live. She thought about returning home but she felt fine. Though she didn't yet qualify for any kind of insurance, she was getting good care in our County health system. Besides, her grandchildren meant everything to her and her son and daughter‐in‐law needed her now more than ever.

Five months into her treatment she developed pain in her neck and started noticing numbness in her hands. She immediately went to the hospital where she was found to have a tumor in her cervical spine. Despite early radiation and surgery, she was completely paralyzed and dependent on mechanical ventilation within a week. In a matter of days she had been torn from life as she knew it.

It became clear during our conversation that, though Mrs. K was not physically uncomfortable, being confined to the hospital was difficult for her. Her grandchildren couldn't visit because she was on contact precautions. She missed them deeply. She missed sitting on her porch drinking her morning coffee. Unable to move her head, she spent most of her day staring at the ceiling or at the TV watching shows in a language she couldn't understand. She had met countless doctors, nurses, and medical personnel, endured multiple complications, including a pulseless arrest, and had been placed in 3 different ICUs. Yet she was unwavering in her desire to remain on the ventilator and continue doing everything.

Both she and her son expected that the treatments she had been getting would help get her off the ventilator so that she could go home. I struggled to balance their hopes with the information at hand, exploring realistic goals.

Could she go to a nursing home and wait and see if she will recover? I know they have these for people that are on ventilators, her son asked.

This was not going to be possible given her disease. Her paralysis was complete and permanent. She would not recover the ability to breathe on her own. Besides, without insurance they would have to pay for these services. It was not realistic to hope for this.

Could we fly her back to Poland so that she can die and be buried there?

She was not stable enough to fly without medical assistance. Furthermore, she would need to be in contact isolation given the virulence of her uniquely resistant bacteria. The cost to arrange an air ambulance was exorbitant and unaffordable on her son's electrician salary. It was not realistic to hope for this.

Well, if she can't go back to Poland and can't get off of the ventilator, could we set up a ventilator at home so that we can care for her there until she dies?

I explained that this would require help from medical personnel trained in ventilator management. They would also need assistance from individuals trained in palliative care to insure that Mrs. K had adequate nursing and symptom relief, particularly in the event of a medical emergency or a complication with the ventilator. The family would be required to pay for these services out of pocket. After spending hours contacting hospice agencies, medical suppliers, nursing agencies, friends of the family, and community organizations it became clear that even though the family was willing to bankrupt themselves for their mother's care there was not a safe, affordable solution. It was not realistic to hope for this.

What can you do for me? Mrs. K asked.

My tongue sat numbly in my mouth. I felt a tide of shame and sorrow rising. Nothing! I thought, knowing not to say it. At times like this, we often fall back on training; I tried to force her into the round hole.

We can care for you here. We can insure that you are as comfortable as possible for whatever time you have left. We can shift our focus from life‐prolonging treatments to those that are purely focused on your comfort. We can stop those treatments that will interrupt the time that you spend with your family, and try to give you and your family as much space as possible to be together here in the ICU. As your death approaches we would work to keep you as comfortable and peaceful as possible and to allow your family to be with you here at your bedside, but we wouldn't try to prolong the dying process.

I don't WANT to die. I WANT to go home, I WANT to smell fresh air. I'm willing to take risks with my life for that, but not otherwise.

My impotence, my inability to give this woman any of the things that she was hoping for was overwhelming. Her suffering was overwhelming. I was unable to find a thread of hope in her world. I was failing my patient.

Feeling a sense of despair, I told her, I cannot help you breathe on your own or smell the fresh air or be with your grandchildren. None of those goals are realistic. I don't know what I can do for you, but I would like to continue to see you. I want to come back tomorrow.

I hope you do, she said, her blue eyes shining as she smiled softly.

I was struck by the words. Though my loftier goals had been frustrated, I realized that my efforts and my presence at her bedside alone were easing her sufferingthis is something that we all hope for.

Hello Mrs. K, I'm Dr. Baru, I said. I squeezed the hand limply resting on her cover with my own gloved one.

The room was quiet except for the sighing of her ventilator in the background. She had a broad round face with high cheek bones. Her skin was wrinkle‐free except around the eyes and the corners of her mouth. She breathed peacefully through her tracheostomy. She slowly nodded her head when I squeezed her hand and her blue eyes shone as she smiled broadly. Her son stood at my side, his mouth set in a straight line his eyes gazing intently at his mom. His handshake was firm and brisk, a single downstroke.

I was called in as the Palliative Care consultant. She's in denial, I was told. There's nothing more that we can do for her here. She had already been in the hospital for 2 months but this was the first time I was meeting her and her son. With the help of the Polish interpreter and her son, who had become adept at reading her lips and translating her breathy rasps, I began to sift through all of the information that they had been told, all of the information they had gathered on their own, and what they understood.

Mrs. K was 59 years old. She'd given up her job as a kindergarten teacher and come from Poland 3 years ago to help care for her first grandchild. Two years later, her daughter‐in‐law delivered 2 more grandchildren: twins. Just weeks after the delivery, Mrs. K was diagnosed with multiple myeloma. She was told that her prognosis was good and that, with chemotherapy, she had years to live. She thought about returning home but she felt fine. Though she didn't yet qualify for any kind of insurance, she was getting good care in our County health system. Besides, her grandchildren meant everything to her and her son and daughter‐in‐law needed her now more than ever.

Five months into her treatment she developed pain in her neck and started noticing numbness in her hands. She immediately went to the hospital where she was found to have a tumor in her cervical spine. Despite early radiation and surgery, she was completely paralyzed and dependent on mechanical ventilation within a week. In a matter of days she had been torn from life as she knew it.

It became clear during our conversation that, though Mrs. K was not physically uncomfortable, being confined to the hospital was difficult for her. Her grandchildren couldn't visit because she was on contact precautions. She missed them deeply. She missed sitting on her porch drinking her morning coffee. Unable to move her head, she spent most of her day staring at the ceiling or at the TV watching shows in a language she couldn't understand. She had met countless doctors, nurses, and medical personnel, endured multiple complications, including a pulseless arrest, and had been placed in 3 different ICUs. Yet she was unwavering in her desire to remain on the ventilator and continue doing everything.

Both she and her son expected that the treatments she had been getting would help get her off the ventilator so that she could go home. I struggled to balance their hopes with the information at hand, exploring realistic goals.

Could she go to a nursing home and wait and see if she will recover? I know they have these for people that are on ventilators, her son asked.

This was not going to be possible given her disease. Her paralysis was complete and permanent. She would not recover the ability to breathe on her own. Besides, without insurance they would have to pay for these services. It was not realistic to hope for this.

Could we fly her back to Poland so that she can die and be buried there?

She was not stable enough to fly without medical assistance. Furthermore, she would need to be in contact isolation given the virulence of her uniquely resistant bacteria. The cost to arrange an air ambulance was exorbitant and unaffordable on her son's electrician salary. It was not realistic to hope for this.

Well, if she can't go back to Poland and can't get off of the ventilator, could we set up a ventilator at home so that we can care for her there until she dies?

I explained that this would require help from medical personnel trained in ventilator management. They would also need assistance from individuals trained in palliative care to insure that Mrs. K had adequate nursing and symptom relief, particularly in the event of a medical emergency or a complication with the ventilator. The family would be required to pay for these services out of pocket. After spending hours contacting hospice agencies, medical suppliers, nursing agencies, friends of the family, and community organizations it became clear that even though the family was willing to bankrupt themselves for their mother's care there was not a safe, affordable solution. It was not realistic to hope for this.

What can you do for me? Mrs. K asked.

My tongue sat numbly in my mouth. I felt a tide of shame and sorrow rising. Nothing! I thought, knowing not to say it. At times like this, we often fall back on training; I tried to force her into the round hole.

We can care for you here. We can insure that you are as comfortable as possible for whatever time you have left. We can shift our focus from life‐prolonging treatments to those that are purely focused on your comfort. We can stop those treatments that will interrupt the time that you spend with your family, and try to give you and your family as much space as possible to be together here in the ICU. As your death approaches we would work to keep you as comfortable and peaceful as possible and to allow your family to be with you here at your bedside, but we wouldn't try to prolong the dying process.

I don't WANT to die. I WANT to go home, I WANT to smell fresh air. I'm willing to take risks with my life for that, but not otherwise.

My impotence, my inability to give this woman any of the things that she was hoping for was overwhelming. Her suffering was overwhelming. I was unable to find a thread of hope in her world. I was failing my patient.

Feeling a sense of despair, I told her, I cannot help you breathe on your own or smell the fresh air or be with your grandchildren. None of those goals are realistic. I don't know what I can do for you, but I would like to continue to see you. I want to come back tomorrow.

I hope you do, she said, her blue eyes shining as she smiled softly.

I was struck by the words. Though my loftier goals had been frustrated, I realized that my efforts and my presence at her bedside alone were easing her sufferingthis is something that we all hope for.

A 60 year old man presented with a 1‐month history of severe, dry cough. His medical history is significant for lymphoma (4 years), bone marrow transplant (8 months), stable graft‐vs.‐host disease, with a negative bleeding history. He self‐medicated his cough with 6 to 8 tablets daily of an analgesic containing 250 mg acetaminophen, 65 mg caffeine, and 250 mg aspirin. He developed severe stabbing pain in his right lower quadrant exacerbated by coughing. Several days later he developed severe pain of his left upper abdomen. The abdominal wall was tender. Laboratory tests revealed a hematocrit of 31%, platelet count of 246,000, international normalized prothrombin ratio of 1.0, and activated partial thromboplastin ratio of 0.8. Abdominal computed tomography (CT) without contrast revealed 2 rectus sheath hematomas. A 4‐cm 2‐cm 2‐cm collection consistent with hematoma was contained within the left rectus abdominis muscle sheath above the level of the umbilicus (white arrow, Figure 1A, top). An additional collection measuring 6 cm 6 cm 4 cm was present below the umbilicus and between the posterior aspect of the right rectus abdominus muscle (black arrows, Figure 1A bottom; B and C) and the anterior surface of the bladder (arrowhead, Figure 1A and C). Aspirin was stopped and he was started on moxifloxacin for pneumonia. He was discharged 3 days later with improved cough resulting in improved pain control.

Figure 1

Rectus sheath hematoma may be contained with the fascia when above the arcuate line, located about 5 cm below the umbilicus. Below the arcuate line a rectus sheath hematoma may dissect posteriorly into the prevesicular space of Retzius. Both are seen in this case. Rectus sheath hematoma are usually associated with cough in the setting of coagulopathy, and may require transfusion or surgical evacuation. High dose aspirin, in this case up to 2000 mg per day, may lead to hemorrhage associated with intractable coughing.

A 60 year old man presented with a 1‐month history of severe, dry cough. His medical history is significant for lymphoma (4 years), bone marrow transplant (8 months), stable graft‐vs.‐host disease, with a negative bleeding history. He self‐medicated his cough with 6 to 8 tablets daily of an analgesic containing 250 mg acetaminophen, 65 mg caffeine, and 250 mg aspirin. He developed severe stabbing pain in his right lower quadrant exacerbated by coughing. Several days later he developed severe pain of his left upper abdomen. The abdominal wall was tender. Laboratory tests revealed a hematocrit of 31%, platelet count of 246,000, international normalized prothrombin ratio of 1.0, and activated partial thromboplastin ratio of 0.8. Abdominal computed tomography (CT) without contrast revealed 2 rectus sheath hematomas. A 4‐cm 2‐cm 2‐cm collection consistent with hematoma was contained within the left rectus abdominis muscle sheath above the level of the umbilicus (white arrow, Figure 1A, top). An additional collection measuring 6 cm 6 cm 4 cm was present below the umbilicus and between the posterior aspect of the right rectus abdominus muscle (black arrows, Figure 1A bottom; B and C) and the anterior surface of the bladder (arrowhead, Figure 1A and C). Aspirin was stopped and he was started on moxifloxacin for pneumonia. He was discharged 3 days later with improved cough resulting in improved pain control.

Figure 1

Rectus sheath hematoma may be contained with the fascia when above the arcuate line, located about 5 cm below the umbilicus. Below the arcuate line a rectus sheath hematoma may dissect posteriorly into the prevesicular space of Retzius. Both are seen in this case. Rectus sheath hematoma are usually associated with cough in the setting of coagulopathy, and may require transfusion or surgical evacuation. High dose aspirin, in this case up to 2000 mg per day, may lead to hemorrhage associated with intractable coughing.

A 60 year old man presented with a 1‐month history of severe, dry cough. His medical history is significant for lymphoma (4 years), bone marrow transplant (8 months), stable graft‐vs.‐host disease, with a negative bleeding history. He self‐medicated his cough with 6 to 8 tablets daily of an analgesic containing 250 mg acetaminophen, 65 mg caffeine, and 250 mg aspirin. He developed severe stabbing pain in his right lower quadrant exacerbated by coughing. Several days later he developed severe pain of his left upper abdomen. The abdominal wall was tender. Laboratory tests revealed a hematocrit of 31%, platelet count of 246,000, international normalized prothrombin ratio of 1.0, and activated partial thromboplastin ratio of 0.8. Abdominal computed tomography (CT) without contrast revealed 2 rectus sheath hematomas. A 4‐cm 2‐cm 2‐cm collection consistent with hematoma was contained within the left rectus abdominis muscle sheath above the level of the umbilicus (white arrow, Figure 1A, top). An additional collection measuring 6 cm 6 cm 4 cm was present below the umbilicus and between the posterior aspect of the right rectus abdominus muscle (black arrows, Figure 1A bottom; B and C) and the anterior surface of the bladder (arrowhead, Figure 1A and C). Aspirin was stopped and he was started on moxifloxacin for pneumonia. He was discharged 3 days later with improved cough resulting in improved pain control.

Figure 1

Rectus sheath hematoma may be contained with the fascia when above the arcuate line, located about 5 cm below the umbilicus. Below the arcuate line a rectus sheath hematoma may dissect posteriorly into the prevesicular space of Retzius. Both are seen in this case. Rectus sheath hematoma are usually associated with cough in the setting of coagulopathy, and may require transfusion or surgical evacuation. High dose aspirin, in this case up to 2000 mg per day, may lead to hemorrhage associated with intractable coughing.

A 24‐year‐old man presented to the emergency room with a 3‐month history of bright red blood per rectum and increasing fatigue. Review of systems was significant for intermittent hematuria, swelling, and pain in his lower extremities. He denied abdominal pain, nausea, or vomiting, and was otherwise asymptomatic. He was not taking any medicines. He said that he has had this bleeding problem on and off since he was a child. The chronic intermittent rectal bleeding usually resolved spontaneously. Previous treatments have consisted of blood transfusions, small bowel resections, and a partial colectomy.

Physical exam demonstrates a thin and well‐nourished African American male in no distress. Temperature 36.7C, blood pressure while sitting was 111/65 mmHg, with a pulse of 117 beats per minute; on standing his blood pressure was 103/54 mmHg, with a pulse of 137 beats per minute, and respirations were 18 breaths per minute. Abdominal examination revealed splenomegaly. Rectal exam revealed the presence of bright red blood. Other significant findings include unilateral limb skeletal asymmetry with the right upper and lower extremity being longer than the left side. There was significant hypertrophy of several digits of the hands and feet bilaterally (Figure 1). Notable was the presence of raised, hyperpigmented irregular linear plaques, extending from his right medial forearm to his chest and also from his abdomen to right medial thigh. Additional skin examination was remarkable for well‐demarcated, raised vascular areas on the lateral thighs and knees bilaterally (Figure 2), as well as the dorsum of both the feet. Laboratory workup was notable for hemoglobin of 2.7gm/dl, a hematocrit of 9%, and mean corpuscular volume (MCV) of 58 fl. Normal coagulation parameters, and profound iron deficiency (iron level 16 mcg/dl and ferritin 20 ng/ml).

Figure 1

Figure 2

Other routine laboratory results including coagulation parameters were unremarkable.

Discussion

Based on the classic examination findings and history of gastrointestinal bleeding, this patient has Klippel‐Trenaunay‐Weber syndrome (KTWS), which is characterized by cutaneous malformations of the capillary and venous systems, bony and soft tissue hypertrophy, and arteriovenous malformations (AVMs).1 Many patients with KTWS suffer recurrent bleeding from gastrointestinal AVMs.

Although involvement is usually unilateral, this patient had bilateral limb hypertrophy and hemangiomas. His nevus flammeus was unilateral and incidentally was present over the lower abdomen and posterior thigh and buttock, with significant underlying varices in the pelvis and rectum. His hematuria was secondary to AVMs in the bladder and resolved by itself. The size and extent of his pelvic and rectal varices presented a therapeutic challenge. With blood transfusions and a conservative approach, his bleeding diminished spontaneously. A rectal artery was thought to be contributing to the problem, so a prophylactic embolization was performed by interventional radiology. Follow‐up at 2 months revealed no further bleeding.

Hospitalists treat common causes of gastrointestinal (GI) bleeding such as ulcers, polyps, malignancies, varices, inflammatory bowel disease, AVMs, and, rarely, mucosal Kaposi sarcoma. However, they may occasionally encounter an adult with skin manifestations of a congenital cause of GI bleeding. The 4 most common congenital disorders with primary cutaneous manifestations that also involve the GI tract are reviewed below (also see Table 1).

Blue rubber bleb nevus syndrome, also known as Bean syndrome, is the rarest of these disorders, characterized by cutaneous and intestinal cavernous hemangiomas that may occasionally be painful and tender.2 Hemangiomas may measure from a few millimeters to approximately 5 cm and are raised, blue‐purple, and rubbery in consistency, with a wrinkled surface. They are usually located on the trunk, extremities, face, and any part of the GI tract, with the small intestine and distal colon being the most common sites involved. Given that the lesions may involve the full thickness of the bowel wall, surgery is often required, as less invasive measures such as endoscopic laser coagulation may be inadequate. Orthopedic problems such as scoliosis arise from pressure exerted by large vascular malformations.

Maffucci syndrome is characterized by skeletal and vascular malformations manifested as enchondromas in the metaphyseal and diaphyseal portion of long bones. The vascular lesions, which may involve mucous membranes or viscera, are compressible blue‐purple hemangiomas that follow the rate of the growth of the child. Limb deformities, pathological fractures, and malignant transformation into chondrosarcomas are common complications.3

Osler‐Weber‐Rendu syndrome is also known as hereditary hemorrhagic telangiectasia. In this disorder, mucocutaneous telangiectatic lesions usually develop by puberty and may involve the conjunctiva, respiratory tract, brain, liver, GI tract, and genitourinary (GU) tract. Most patients exhibit only epistaxis, yet massive hemorrhage may occur in the lung, GI tract, and GU tract. These hemorrhages can usually be managed by cautery or electrocoagulation but pulmonary and GI lesions may need excision.4

KTWS consists of the triad of cutaneous vascular malformations of the capillary, venous and lymphatic systems, bony and soft tissue hypertrophy, and venous varicosities in association with AVMs. The name Weber is added when patients have AVMs that are clinically significant; otherwise, it is simply known as Klippel‐Trenaunay syndrome. The most common cutaneous vascular lesion is a capillary hemangioma known as a nevus flammeus. The distribution of the nevus flammeus usually indicates underlying vascular malformations that may extend as deep as the bone, causing limb or digit hypertrophy, as seen in this patient.5 Delineation of the extent of vascular abnormalities is accomplished by noninvasive methods such as color ultrasonography, magnetic resonance imaging, and computer‐aided angiography. Symptomatic GI or GU involvement is rare (1%), but can cause significant hemorrhage.6 Surgical correction is often difficult and the lesions tend to recur.

In the largest published series of Klippel‐Trenaunay patients, followed over 30 years at the Mayo Clinic, most patients were treated conservatively, with surgery limited to epiphysiodesis to prevent excessive leg length in the affected limbs and selected superficial vein stripping in patients with large venous varicosities with preserved deep venous systems.7, 8 For the treatment of AVMs, nonsurgical measures such as foam embolization and radiotherapy are increasingly being used due to their safety and precise application.9, 10

A 24‐year‐old man presented to the emergency room with a 3‐month history of bright red blood per rectum and increasing fatigue. Review of systems was significant for intermittent hematuria, swelling, and pain in his lower extremities. He denied abdominal pain, nausea, or vomiting, and was otherwise asymptomatic. He was not taking any medicines. He said that he has had this bleeding problem on and off since he was a child. The chronic intermittent rectal bleeding usually resolved spontaneously. Previous treatments have consisted of blood transfusions, small bowel resections, and a partial colectomy.

Physical exam demonstrates a thin and well‐nourished African American male in no distress. Temperature 36.7C, blood pressure while sitting was 111/65 mmHg, with a pulse of 117 beats per minute; on standing his blood pressure was 103/54 mmHg, with a pulse of 137 beats per minute, and respirations were 18 breaths per minute. Abdominal examination revealed splenomegaly. Rectal exam revealed the presence of bright red blood. Other significant findings include unilateral limb skeletal asymmetry with the right upper and lower extremity being longer than the left side. There was significant hypertrophy of several digits of the hands and feet bilaterally (Figure 1). Notable was the presence of raised, hyperpigmented irregular linear plaques, extending from his right medial forearm to his chest and also from his abdomen to right medial thigh. Additional skin examination was remarkable for well‐demarcated, raised vascular areas on the lateral thighs and knees bilaterally (Figure 2), as well as the dorsum of both the feet. Laboratory workup was notable for hemoglobin of 2.7gm/dl, a hematocrit of 9%, and mean corpuscular volume (MCV) of 58 fl. Normal coagulation parameters, and profound iron deficiency (iron level 16 mcg/dl and ferritin 20 ng/ml).

Figure 1

Figure 2

Other routine laboratory results including coagulation parameters were unremarkable.

Discussion

Based on the classic examination findings and history of gastrointestinal bleeding, this patient has Klippel‐Trenaunay‐Weber syndrome (KTWS), which is characterized by cutaneous malformations of the capillary and venous systems, bony and soft tissue hypertrophy, and arteriovenous malformations (AVMs).1 Many patients with KTWS suffer recurrent bleeding from gastrointestinal AVMs.

Although involvement is usually unilateral, this patient had bilateral limb hypertrophy and hemangiomas. His nevus flammeus was unilateral and incidentally was present over the lower abdomen and posterior thigh and buttock, with significant underlying varices in the pelvis and rectum. His hematuria was secondary to AVMs in the bladder and resolved by itself. The size and extent of his pelvic and rectal varices presented a therapeutic challenge. With blood transfusions and a conservative approach, his bleeding diminished spontaneously. A rectal artery was thought to be contributing to the problem, so a prophylactic embolization was performed by interventional radiology. Follow‐up at 2 months revealed no further bleeding.

Hospitalists treat common causes of gastrointestinal (GI) bleeding such as ulcers, polyps, malignancies, varices, inflammatory bowel disease, AVMs, and, rarely, mucosal Kaposi sarcoma. However, they may occasionally encounter an adult with skin manifestations of a congenital cause of GI bleeding. The 4 most common congenital disorders with primary cutaneous manifestations that also involve the GI tract are reviewed below (also see Table 1).

Blue rubber bleb nevus syndrome, also known as Bean syndrome, is the rarest of these disorders, characterized by cutaneous and intestinal cavernous hemangiomas that may occasionally be painful and tender.2 Hemangiomas may measure from a few millimeters to approximately 5 cm and are raised, blue‐purple, and rubbery in consistency, with a wrinkled surface. They are usually located on the trunk, extremities, face, and any part of the GI tract, with the small intestine and distal colon being the most common sites involved. Given that the lesions may involve the full thickness of the bowel wall, surgery is often required, as less invasive measures such as endoscopic laser coagulation may be inadequate. Orthopedic problems such as scoliosis arise from pressure exerted by large vascular malformations.

Maffucci syndrome is characterized by skeletal and vascular malformations manifested as enchondromas in the metaphyseal and diaphyseal portion of long bones. The vascular lesions, which may involve mucous membranes or viscera, are compressible blue‐purple hemangiomas that follow the rate of the growth of the child. Limb deformities, pathological fractures, and malignant transformation into chondrosarcomas are common complications.3

Osler‐Weber‐Rendu syndrome is also known as hereditary hemorrhagic telangiectasia. In this disorder, mucocutaneous telangiectatic lesions usually develop by puberty and may involve the conjunctiva, respiratory tract, brain, liver, GI tract, and genitourinary (GU) tract. Most patients exhibit only epistaxis, yet massive hemorrhage may occur in the lung, GI tract, and GU tract. These hemorrhages can usually be managed by cautery or electrocoagulation but pulmonary and GI lesions may need excision.4

KTWS consists of the triad of cutaneous vascular malformations of the capillary, venous and lymphatic systems, bony and soft tissue hypertrophy, and venous varicosities in association with AVMs. The name Weber is added when patients have AVMs that are clinically significant; otherwise, it is simply known as Klippel‐Trenaunay syndrome. The most common cutaneous vascular lesion is a capillary hemangioma known as a nevus flammeus. The distribution of the nevus flammeus usually indicates underlying vascular malformations that may extend as deep as the bone, causing limb or digit hypertrophy, as seen in this patient.5 Delineation of the extent of vascular abnormalities is accomplished by noninvasive methods such as color ultrasonography, magnetic resonance imaging, and computer‐aided angiography. Symptomatic GI or GU involvement is rare (1%), but can cause significant hemorrhage.6 Surgical correction is often difficult and the lesions tend to recur.

In the largest published series of Klippel‐Trenaunay patients, followed over 30 years at the Mayo Clinic, most patients were treated conservatively, with surgery limited to epiphysiodesis to prevent excessive leg length in the affected limbs and selected superficial vein stripping in patients with large venous varicosities with preserved deep venous systems.7, 8 For the treatment of AVMs, nonsurgical measures such as foam embolization and radiotherapy are increasingly being used due to their safety and precise application.9, 10

A 24‐year‐old man presented to the emergency room with a 3‐month history of bright red blood per rectum and increasing fatigue. Review of systems was significant for intermittent hematuria, swelling, and pain in his lower extremities. He denied abdominal pain, nausea, or vomiting, and was otherwise asymptomatic. He was not taking any medicines. He said that he has had this bleeding problem on and off since he was a child. The chronic intermittent rectal bleeding usually resolved spontaneously. Previous treatments have consisted of blood transfusions, small bowel resections, and a partial colectomy.

Physical exam demonstrates a thin and well‐nourished African American male in no distress. Temperature 36.7C, blood pressure while sitting was 111/65 mmHg, with a pulse of 117 beats per minute; on standing his blood pressure was 103/54 mmHg, with a pulse of 137 beats per minute, and respirations were 18 breaths per minute. Abdominal examination revealed splenomegaly. Rectal exam revealed the presence of bright red blood. Other significant findings include unilateral limb skeletal asymmetry with the right upper and lower extremity being longer than the left side. There was significant hypertrophy of several digits of the hands and feet bilaterally (Figure 1). Notable was the presence of raised, hyperpigmented irregular linear plaques, extending from his right medial forearm to his chest and also from his abdomen to right medial thigh. Additional skin examination was remarkable for well‐demarcated, raised vascular areas on the lateral thighs and knees bilaterally (Figure 2), as well as the dorsum of both the feet. Laboratory workup was notable for hemoglobin of 2.7gm/dl, a hematocrit of 9%, and mean corpuscular volume (MCV) of 58 fl. Normal coagulation parameters, and profound iron deficiency (iron level 16 mcg/dl and ferritin 20 ng/ml).

Figure 1

Figure 2

Other routine laboratory results including coagulation parameters were unremarkable.

Discussion

Based on the classic examination findings and history of gastrointestinal bleeding, this patient has Klippel‐Trenaunay‐Weber syndrome (KTWS), which is characterized by cutaneous malformations of the capillary and venous systems, bony and soft tissue hypertrophy, and arteriovenous malformations (AVMs).1 Many patients with KTWS suffer recurrent bleeding from gastrointestinal AVMs.

Although involvement is usually unilateral, this patient had bilateral limb hypertrophy and hemangiomas. His nevus flammeus was unilateral and incidentally was present over the lower abdomen and posterior thigh and buttock, with significant underlying varices in the pelvis and rectum. His hematuria was secondary to AVMs in the bladder and resolved by itself. The size and extent of his pelvic and rectal varices presented a therapeutic challenge. With blood transfusions and a conservative approach, his bleeding diminished spontaneously. A rectal artery was thought to be contributing to the problem, so a prophylactic embolization was performed by interventional radiology. Follow‐up at 2 months revealed no further bleeding.

Hospitalists treat common causes of gastrointestinal (GI) bleeding such as ulcers, polyps, malignancies, varices, inflammatory bowel disease, AVMs, and, rarely, mucosal Kaposi sarcoma. However, they may occasionally encounter an adult with skin manifestations of a congenital cause of GI bleeding. The 4 most common congenital disorders with primary cutaneous manifestations that also involve the GI tract are reviewed below (also see Table 1).

Blue rubber bleb nevus syndrome, also known as Bean syndrome, is the rarest of these disorders, characterized by cutaneous and intestinal cavernous hemangiomas that may occasionally be painful and tender.2 Hemangiomas may measure from a few millimeters to approximately 5 cm and are raised, blue‐purple, and rubbery in consistency, with a wrinkled surface. They are usually located on the trunk, extremities, face, and any part of the GI tract, with the small intestine and distal colon being the most common sites involved. Given that the lesions may involve the full thickness of the bowel wall, surgery is often required, as less invasive measures such as endoscopic laser coagulation may be inadequate. Orthopedic problems such as scoliosis arise from pressure exerted by large vascular malformations.

Maffucci syndrome is characterized by skeletal and vascular malformations manifested as enchondromas in the metaphyseal and diaphyseal portion of long bones. The vascular lesions, which may involve mucous membranes or viscera, are compressible blue‐purple hemangiomas that follow the rate of the growth of the child. Limb deformities, pathological fractures, and malignant transformation into chondrosarcomas are common complications.3

Osler‐Weber‐Rendu syndrome is also known as hereditary hemorrhagic telangiectasia. In this disorder, mucocutaneous telangiectatic lesions usually develop by puberty and may involve the conjunctiva, respiratory tract, brain, liver, GI tract, and genitourinary (GU) tract. Most patients exhibit only epistaxis, yet massive hemorrhage may occur in the lung, GI tract, and GU tract. These hemorrhages can usually be managed by cautery or electrocoagulation but pulmonary and GI lesions may need excision.4

KTWS consists of the triad of cutaneous vascular malformations of the capillary, venous and lymphatic systems, bony and soft tissue hypertrophy, and venous varicosities in association with AVMs. The name Weber is added when patients have AVMs that are clinically significant; otherwise, it is simply known as Klippel‐Trenaunay syndrome. The most common cutaneous vascular lesion is a capillary hemangioma known as a nevus flammeus. The distribution of the nevus flammeus usually indicates underlying vascular malformations that may extend as deep as the bone, causing limb or digit hypertrophy, as seen in this patient.5 Delineation of the extent of vascular abnormalities is accomplished by noninvasive methods such as color ultrasonography, magnetic resonance imaging, and computer‐aided angiography. Symptomatic GI or GU involvement is rare (1%), but can cause significant hemorrhage.6 Surgical correction is often difficult and the lesions tend to recur.

In the largest published series of Klippel‐Trenaunay patients, followed over 30 years at the Mayo Clinic, most patients were treated conservatively, with surgery limited to epiphysiodesis to prevent excessive leg length in the affected limbs and selected superficial vein stripping in patients with large venous varicosities with preserved deep venous systems.7, 8 For the treatment of AVMs, nonsurgical measures such as foam embolization and radiotherapy are increasingly being used due to their safety and precise application.9, 10

A 61‐year‐old woman presented with 1 month of abdominal pain and bowel irregularity. Physical examination was normal. Computed tomography of the abdomen demonstrated a nonincarcerated hernia of the ascending colon through a 25‐mm bony defect in the superior aspect of the right iliac crest (Figure 1)a defect created 16 years prior when iliac crest bone graft harvest was performed during subtalar fusion. An open surgical approach was employed to reduce the hernia and close the defect with mesh. Her postoperative course was unremarkable. Her symptoms resolved entirely.

Figure 1

The iliac crest is the site utilized most frequently in orthopedic surgery for bone graft harvest. The literature suggests that up to 5% of these procedures may be complicated by symptomatic herniation of abdominal contents.1 Other procedural complications include: donor site pain, nerve injury (commonly the lateral femoral cutaneous nerve, manifesting as meralgia paresthetica), vascular disruption (including superior gluteal and lumbar arteries), hematoma, visible deformity, abnormal gait, and stress fracture.2, 3 Advanced age, obesity, muscular weakness, larger defect size, and full‐thickness harvest have been identified as risk factors.4 The hernia often presents as a soft‐tissue mass originating at the defect site, which may become more pronounced with cough. Auscultation over the area may reveal bowel sounds. The spectrum of associated abdominal symptoms ranges from mild discomfort to colicky pain and distention;3 up to 16% of patients present with acute signs of intestinal obstruction.1 Iliac herniation may be prevented by harvest of a partial thickness graft, whenever possible. Selective removal of bone from the anterior or posterior crest, rather than the middle, may also decrease the risk.4 Clinicians caring for patients with a history of iliac crest bone graft harvest should be familiar with this complication and consider prompt radiologic imaging when investigating new or unexplained abdominal symptoms.

A 61‐year‐old woman presented with 1 month of abdominal pain and bowel irregularity. Physical examination was normal. Computed tomography of the abdomen demonstrated a nonincarcerated hernia of the ascending colon through a 25‐mm bony defect in the superior aspect of the right iliac crest (Figure 1)a defect created 16 years prior when iliac crest bone graft harvest was performed during subtalar fusion. An open surgical approach was employed to reduce the hernia and close the defect with mesh. Her postoperative course was unremarkable. Her symptoms resolved entirely.

Figure 1

The iliac crest is the site utilized most frequently in orthopedic surgery for bone graft harvest. The literature suggests that up to 5% of these procedures may be complicated by symptomatic herniation of abdominal contents.1 Other procedural complications include: donor site pain, nerve injury (commonly the lateral femoral cutaneous nerve, manifesting as meralgia paresthetica), vascular disruption (including superior gluteal and lumbar arteries), hematoma, visible deformity, abnormal gait, and stress fracture.2, 3 Advanced age, obesity, muscular weakness, larger defect size, and full‐thickness harvest have been identified as risk factors.4 The hernia often presents as a soft‐tissue mass originating at the defect site, which may become more pronounced with cough. Auscultation over the area may reveal bowel sounds. The spectrum of associated abdominal symptoms ranges from mild discomfort to colicky pain and distention;3 up to 16% of patients present with acute signs of intestinal obstruction.1 Iliac herniation may be prevented by harvest of a partial thickness graft, whenever possible. Selective removal of bone from the anterior or posterior crest, rather than the middle, may also decrease the risk.4 Clinicians caring for patients with a history of iliac crest bone graft harvest should be familiar with this complication and consider prompt radiologic imaging when investigating new or unexplained abdominal symptoms.

A 61‐year‐old woman presented with 1 month of abdominal pain and bowel irregularity. Physical examination was normal. Computed tomography of the abdomen demonstrated a nonincarcerated hernia of the ascending colon through a 25‐mm bony defect in the superior aspect of the right iliac crest (Figure 1)a defect created 16 years prior when iliac crest bone graft harvest was performed during subtalar fusion. An open surgical approach was employed to reduce the hernia and close the defect with mesh. Her postoperative course was unremarkable. Her symptoms resolved entirely.

Figure 1

The iliac crest is the site utilized most frequently in orthopedic surgery for bone graft harvest. The literature suggests that up to 5% of these procedures may be complicated by symptomatic herniation of abdominal contents.1 Other procedural complications include: donor site pain, nerve injury (commonly the lateral femoral cutaneous nerve, manifesting as meralgia paresthetica), vascular disruption (including superior gluteal and lumbar arteries), hematoma, visible deformity, abnormal gait, and stress fracture.2, 3 Advanced age, obesity, muscular weakness, larger defect size, and full‐thickness harvest have been identified as risk factors.4 The hernia often presents as a soft‐tissue mass originating at the defect site, which may become more pronounced with cough. Auscultation over the area may reveal bowel sounds. The spectrum of associated abdominal symptoms ranges from mild discomfort to colicky pain and distention;3 up to 16% of patients present with acute signs of intestinal obstruction.1 Iliac herniation may be prevented by harvest of a partial thickness graft, whenever possible. Selective removal of bone from the anterior or posterior crest, rather than the middle, may also decrease the risk.4 Clinicians caring for patients with a history of iliac crest bone graft harvest should be familiar with this complication and consider prompt radiologic imaging when investigating new or unexplained abdominal symptoms.

Communication failures are a frequent cause of adverse events14; the Joint Commission (TJC) reports that such failures contributed to 65% of reported sentinel events.5 Strategies to improve communication have focused on implementing formal teamwork training programs and/or teaching specific communication skills.613 While these strategies largely address communication between healthcare providers, there is a growing emphasis on developing strategies to engage patients in their care, and improving communication with them and their families.

In 2007, TJC announced a new National Patient Safety Goal (NPSG) that encourage(s) patients' active involvement in their own care as a patient safety strategy.14 This builds upon a landmark Institute of Medicine report that highlighted patient‐centeredness as 1 of the 6 domains for delivering high‐quality care.15 Current literature on developing such patient‐centered strategies enumerates several approaches, including better access to health information, use of innovative technology solutions, and focused efforts at improving communication.1618

The placement of whiteboards in patient rooms is an increasingly common strategy to improve communication. These boards, typically placed on a wall near a patient's hospital bed, allow any number of providers to communicate a wide range of information. Both Kaiser Permanente's Nurse Knowledge Exchange program and the Institute for Healthcare Improvement's Transforming Care at the Bedside promote whiteboard use, though with little specific guidance about practical implementation.19^,20 Despite their growing prevalence, there is no published literature guiding the most effective uses of whiteboards, or describing their impact on communication, teamwork, or patient satisfaction and care. We present findings from a survey of patient whiteboard use on an academic medical service, and offer a series of recommendations based on our findings and experiences.

Methods

We anonymously surveyed bedside nurses from 3 inpatient medical units, internal medicine housestaff, and faculty from the Division of Hospital Medicine at the University of California, San Francisco (UCSF). We solicited experiences of physician and nursing leaders who were engaged in whiteboard interventions over the past 2 years to identify relevant topics for study. Their experiences were based on isolated unit‐based efforts to implement whiteboards through a variety of strategies (eg, whiteboard templates, simple identification of provider teams, goals for the day). Their input guided the survey development and the suggested recommendations. The topics identified were then translated into multiple‐choice questions, and further edited for clarity by the authors. A Likert scale was used that measured frequency of use, usefulness, and attitudes toward patient whiteboards. An open‐ended question seeking additional comments about patient whiteboards was also asked. The survey was administered to nurses at staff meetings and through physical mailboxes on their respective patient care units with a 1‐month collection period. The survey was administered to housestaff and attendings via e‐mail listserves using an online commercial survey administration tool.21 The nursing surveys were later entered into the same online survey administration tool, which ultimately provided summary reports and descriptive findings to meet the study objectives. Our project was reviewed and approved by the UCSF Committee on Human Research.

Results

Survey responses were collected from 104 nurse respondents (81% response rate), 118 internal medicine housestaff (74% response rate), and 31 hospitalists (86% response rate). Nurses were far more likely to write on whiteboards, read what was written on them, and find the related information useful (Figure 1A‐C). Nurses, housestaff, and attendings all believed the bedside nurse was the single most important provider name listed on a whiteboard. However, the respondents differed in their rated value of other providers listed on the whiteboard (Figure 2). Nurses gave higher ratings to the utility of having patient care assistants (PCAs) listed as compared to housestaff and attendings. Overall, respondents felt it would be less useful to list consultants and pharmacists than the nurse, attending, and housestaff. All of the respondents believed family contact information was the most useful information on a whiteboard, whereas more nurses rated a goal for the day and anticipated discharge date as more useful than housestaff and attendings (Figure 3).

Figure 1

Figure 2

Figure 3

From an operational standpoint, the majority of respondents felt that nurses should be responsible for the information on a whiteboard, nurses and physicians together should create goals for the day, and the greatest barrier to using whiteboards was not having pens easily available (Figure 4A‐C). Most respondents also agreed that using templated whiteboards (with predefined fields) to guide content would increase their use (Figure 4D). All respondents believed that whiteboard use could improve teamwork and communication as well as patient care (Figure 5). Respondents also offered a variety of specific comments in response to an open‐ended question about whiteboard use (Table 1).

Figure 4

Figure 5

Discussion

Our findings demonstrate the potential value of patient whiteboards, which is supported by the vast majority of respondents, who agreed their use may improve patient care and teamwork. It is also clear that whiteboard use is not achieving this potential or being used as a patient‐centered tool. This is best illustrated by findings of their low rate of use and completion among attendings and housestaff (Figure 1A, B) and the lack of consensus as to what information on the whiteboards is useful. Patient whiteboards require defined goals, thoughtful planning, regular monitoring, and ongoing evaluation. The challenges around effective adoption and implementation is perhaps more about ensuring compliance and completion rather than simply gaining buy‐in and engagement for their value.

While the differential use of whiteboards between nurses and physicians was not surprising, a few specific findings warrant further discussion. First, it is interesting that nurses rated their own names and that of PCAs as the most useful, while physicians rated the nurse's name as being of equal value to their own. This may speak to the role PCAs play for nurses in helping the latter provide bedside care, rather than a reflection of the nurses' perception of the value of PCAs for patients. Second, while all respondents rated highly the value of family contact information on the whiteboard, nurses valued a goal for the day and anticipated discharge date more highly than did physicians. These findings likely reflect that nurses desire an understanding about plans of care and if they are not communicated face‐to‐face as the most effective strategy,22 they should at least be spelled out clearly on a whiteboard. This is supported by evidence that better collaboration between nurses and physicians improves patient outcomes.23 It may also be that physicians place more value on their own progress notes (rather than whiteboards) as a vehicle for communicating daily goals and discharge planning.

Other practical considerations involve who owns it and, if we do create goals for the day, whose goals should they represent? The majority of nurse and physician responses advocated for nurses to be responsible for accurate and complete information being updated on whiteboards. A larger percentage of attendings favored shared responsibility of the whiteboard, which was reinforced by their support of having goals for the day created jointly by nurses and physicians. Interestingly, a much smaller percentage of respondents felt goals for the day should be driven by patients (or family members). These data may point to the different perspectives that each individual provider bringsphysician, nurse, pharmacist, discharge plannerwith their respective goals differing in nature. Finally, it is also interesting that while attendings and housestaff believed that whiteboards can improve patient care teamwork/communication (Figure 5), a much smaller percentage actually read what is on them (Figure 1B). This may reflect the unclear goals of whiteboards, its absence as part of daily workflow, the infrequency of updated information on them, or perhaps an institution‐specific phenomenon that we will use to drive further improvement strategies.

Selected respondent comments (Table 1) highlight important messages about whiteboard use and provide helpful context to the survey responses. We found that the goal of whiteboard use is not always clear; is it to improve communication among providers, to improve communication with patients, a tool to engage patients in their care, or some combination of the above? Without a clear goal, providers are left to wonder whether whiteboard use is simply another task or really an intervention to improve care. This may in part, or perhaps fully, explain the differences discovered in whiteboard use and practices among our surveyed providers.

If, however, one were to make clear that the goal of patient whiteboards is to engage patients in their care and help achieve an important NPSG, methods to implement their use become better guided. A limitation of our study is that we did not survey patients about their perceptions of whiteboards use, an important needs assessment that would further drive this patient‐centered intervention. Regardless, we can draw a number of lessons from our findings and devise a set of reasonable recommendations.

Recommendations

We provide the following set of recommendations for hospitals adopting patient whiteboards, drawing on our survey findings and experiences with implementation at our own institution. We also acknowledge the role that local hospital cultures may play in adopting whiteboard use, and our recommendations are simply guidelines that can be applied or used in planning efforts. We believe effective use of a patient whiteboard requires a patient‐centered approach and the following:

• 1

  Whiteboards should be placed in clear view of patients from their hospital bed

  A simple yet critical issue as placing a whiteboard behind a patient's bed or off to the side fails to provide them with a constant visual cue to engage in the information.

• 2

  Buy and fasten erasable pens to the whiteboards themselves

  In our institution, purchasing pens for each provider was a less effective strategy than simply affixing the pen to the whiteboard itself. A supply of erasable pens must be available at the nursing station to quickly replace those with fading ink.

• 3

  Create whiteboard templates

  Our findings and experience suggest that structured formats for whiteboards may be more effective in ensuring both important and accurate information gets included. Blank whiteboards lead to less standardization in practice and fail to create prompts for providers to both write and review the content available. Anecdotally, we created a number of whiteboards with templated information, and this did seem to increase the consistency, standardization, and ease of use.

• 4

  Whiteboard templates should include the following items:
• a

  Day and Date

  This serves to orient patients (and their families) as well as providers with the date of information written on the whiteboard. It is also an important mechanism to ensure information is updated daily.

• b

  Patient's name (or initials)

  With bed turnover (or patient transfers to different beds and units) commonplace in hospital care, we believe that listing the patient's name on the board prevents the potential for patients (and their families) or providers to mistakenly take information from a previous patient's care on the whiteboard for their own.

• c

  Bedside nurse

  This was noted as the most useful provider listed on a patient whiteboard, which is quite logical given the role bedside nurses play for hospitalized patients.

• d

  Primary physician(s) (attending, resident, and intern, if applicable)

  This was noted as the next most important provider(s) and perhaps increasingly important both in teaching and nonteaching settings where shift‐work and signouts are growing in frequency among physicians.

• e

  Goal for the day

  While this was not a consensus from our survey respondents, we believe patients (rather than providers) should ultimately guide determination of their goal for the day as this engages them directly with the planachieving a patient‐centered initiative. In our experience, an effective strategy was having the bedside nurse directly engage patients each morning to help place a goal for the day on their whiteboard.

• f

  Anticipated discharge date

  While understanding the potential for this date to change, we believe the benefits of having patients (and their families) thinking about discharge, rather than feeling surprised by it on the morning of discharge, serves as an important mechanism to bridge communication about the discharge process.24

• g

  Family member's contact information (phone number)

• h

  Questions for providers

  This last entry allows a space for families to engage the healthcare team and, once again, create an opportunity for clarification of treatment and discharge plans.

• 5

  Bedside nurses should facilitate writing and updating information on the whiteboard

  Without our survey findings, this might have generated debate or controversy over whether nurses should be burdened with one more task to their responsibilities. However, our nurse respondents embraced this responsibility with spontaneous comments about their patient advocate role, and stated that whiteboards can serve as a tool to assist in that responsibility. Furthermore, not a single nurse respondent stated as barrier to use that I didn't think it was my responsibility. Nonetheless, whiteboard use must be a shared communication tool and not simply a tool between nurse and patient. Practically, we would recommend that bedside nurses facilitate updating whiteboards each morning, at a time when they are already helping patients create a goal for the day. Other providers must be trained to review information on the whiteboard, engage patients about their specific goal, and share the responsibility of keeping the information on the whiteboard updated.

• 6

  Create a system for auditing utilization and providing feedback early during rollout

  We found that adoption was very slow at the outset. One strategy to consider is having designated auditors check whiteboards in each room, measuring weekly compliance and providing this feedback to nurse managers. This auditing process may help identify barriers that can be addressed quickly (eg, unavailability of pens).

Finally, it is important to comment on the confidentiality concerns often raised in the context of whiteboard use. Confidentiality concerns largely arise from personal health information being used without a patient's explicit consent. If our recommendations are adopted, they require whiteboard use to be a patient‐centered and patient‐driven initiative. The type of information on the whiteboard should be determined with sensitivity but also with consent of the patient. We have not experienced any concerns by patients or providers in this regard because patients are told about the goals of the whiteboard initiative with our above principles in mind.

Conclusions

Patient whiteboards may improve communication among members of the healthcare team (eg, nurses, physicians, and others) and between providers and their patients (and family members). Further investigation is warranted to determine if adopting our recommendations leads to improved communication, teamwork, or patient satisfaction and care. In the meantime, as many hospitals continue to install and implement whiteboards, we hope our recommendations, accompanied by an emphasis on creating a patient‐centered communication tool, offer a roadmap for considering best practices in their use.

Communication failures are a frequent cause of adverse events14; the Joint Commission (TJC) reports that such failures contributed to 65% of reported sentinel events.5 Strategies to improve communication have focused on implementing formal teamwork training programs and/or teaching specific communication skills.613 While these strategies largely address communication between healthcare providers, there is a growing emphasis on developing strategies to engage patients in their care, and improving communication with them and their families.

In 2007, TJC announced a new National Patient Safety Goal (NPSG) that encourage(s) patients' active involvement in their own care as a patient safety strategy.14 This builds upon a landmark Institute of Medicine report that highlighted patient‐centeredness as 1 of the 6 domains for delivering high‐quality care.15 Current literature on developing such patient‐centered strategies enumerates several approaches, including better access to health information, use of innovative technology solutions, and focused efforts at improving communication.1618

The placement of whiteboards in patient rooms is an increasingly common strategy to improve communication. These boards, typically placed on a wall near a patient's hospital bed, allow any number of providers to communicate a wide range of information. Both Kaiser Permanente's Nurse Knowledge Exchange program and the Institute for Healthcare Improvement's Transforming Care at the Bedside promote whiteboard use, though with little specific guidance about practical implementation.19^,20 Despite their growing prevalence, there is no published literature guiding the most effective uses of whiteboards, or describing their impact on communication, teamwork, or patient satisfaction and care. We present findings from a survey of patient whiteboard use on an academic medical service, and offer a series of recommendations based on our findings and experiences.

Methods

We anonymously surveyed bedside nurses from 3 inpatient medical units, internal medicine housestaff, and faculty from the Division of Hospital Medicine at the University of California, San Francisco (UCSF). We solicited experiences of physician and nursing leaders who were engaged in whiteboard interventions over the past 2 years to identify relevant topics for study. Their experiences were based on isolated unit‐based efforts to implement whiteboards through a variety of strategies (eg, whiteboard templates, simple identification of provider teams, goals for the day). Their input guided the survey development and the suggested recommendations. The topics identified were then translated into multiple‐choice questions, and further edited for clarity by the authors. A Likert scale was used that measured frequency of use, usefulness, and attitudes toward patient whiteboards. An open‐ended question seeking additional comments about patient whiteboards was also asked. The survey was administered to nurses at staff meetings and through physical mailboxes on their respective patient care units with a 1‐month collection period. The survey was administered to housestaff and attendings via e‐mail listserves using an online commercial survey administration tool.21 The nursing surveys were later entered into the same online survey administration tool, which ultimately provided summary reports and descriptive findings to meet the study objectives. Our project was reviewed and approved by the UCSF Committee on Human Research.

Results

Survey responses were collected from 104 nurse respondents (81% response rate), 118 internal medicine housestaff (74% response rate), and 31 hospitalists (86% response rate). Nurses were far more likely to write on whiteboards, read what was written on them, and find the related information useful (Figure 1A‐C). Nurses, housestaff, and attendings all believed the bedside nurse was the single most important provider name listed on a whiteboard. However, the respondents differed in their rated value of other providers listed on the whiteboard (Figure 2). Nurses gave higher ratings to the utility of having patient care assistants (PCAs) listed as compared to housestaff and attendings. Overall, respondents felt it would be less useful to list consultants and pharmacists than the nurse, attending, and housestaff. All of the respondents believed family contact information was the most useful information on a whiteboard, whereas more nurses rated a goal for the day and anticipated discharge date as more useful than housestaff and attendings (Figure 3).

Figure 1

Figure 2

Figure 3

From an operational standpoint, the majority of respondents felt that nurses should be responsible for the information on a whiteboard, nurses and physicians together should create goals for the day, and the greatest barrier to using whiteboards was not having pens easily available (Figure 4A‐C). Most respondents also agreed that using templated whiteboards (with predefined fields) to guide content would increase their use (Figure 4D). All respondents believed that whiteboard use could improve teamwork and communication as well as patient care (Figure 5). Respondents also offered a variety of specific comments in response to an open‐ended question about whiteboard use (Table 1).

Figure 4

Figure 5

Discussion

Our findings demonstrate the potential value of patient whiteboards, which is supported by the vast majority of respondents, who agreed their use may improve patient care and teamwork. It is also clear that whiteboard use is not achieving this potential or being used as a patient‐centered tool. This is best illustrated by findings of their low rate of use and completion among attendings and housestaff (Figure 1A, B) and the lack of consensus as to what information on the whiteboards is useful. Patient whiteboards require defined goals, thoughtful planning, regular monitoring, and ongoing evaluation. The challenges around effective adoption and implementation is perhaps more about ensuring compliance and completion rather than simply gaining buy‐in and engagement for their value.

While the differential use of whiteboards between nurses and physicians was not surprising, a few specific findings warrant further discussion. First, it is interesting that nurses rated their own names and that of PCAs as the most useful, while physicians rated the nurse's name as being of equal value to their own. This may speak to the role PCAs play for nurses in helping the latter provide bedside care, rather than a reflection of the nurses' perception of the value of PCAs for patients. Second, while all respondents rated highly the value of family contact information on the whiteboard, nurses valued a goal for the day and anticipated discharge date more highly than did physicians. These findings likely reflect that nurses desire an understanding about plans of care and if they are not communicated face‐to‐face as the most effective strategy,22 they should at least be spelled out clearly on a whiteboard. This is supported by evidence that better collaboration between nurses and physicians improves patient outcomes.23 It may also be that physicians place more value on their own progress notes (rather than whiteboards) as a vehicle for communicating daily goals and discharge planning.

Other practical considerations involve who owns it and, if we do create goals for the day, whose goals should they represent? The majority of nurse and physician responses advocated for nurses to be responsible for accurate and complete information being updated on whiteboards. A larger percentage of attendings favored shared responsibility of the whiteboard, which was reinforced by their support of having goals for the day created jointly by nurses and physicians. Interestingly, a much smaller percentage of respondents felt goals for the day should be driven by patients (or family members). These data may point to the different perspectives that each individual provider bringsphysician, nurse, pharmacist, discharge plannerwith their respective goals differing in nature. Finally, it is also interesting that while attendings and housestaff believed that whiteboards can improve patient care teamwork/communication (Figure 5), a much smaller percentage actually read what is on them (Figure 1B). This may reflect the unclear goals of whiteboards, its absence as part of daily workflow, the infrequency of updated information on them, or perhaps an institution‐specific phenomenon that we will use to drive further improvement strategies.

Selected respondent comments (Table 1) highlight important messages about whiteboard use and provide helpful context to the survey responses. We found that the goal of whiteboard use is not always clear; is it to improve communication among providers, to improve communication with patients, a tool to engage patients in their care, or some combination of the above? Without a clear goal, providers are left to wonder whether whiteboard use is simply another task or really an intervention to improve care. This may in part, or perhaps fully, explain the differences discovered in whiteboard use and practices among our surveyed providers.

If, however, one were to make clear that the goal of patient whiteboards is to engage patients in their care and help achieve an important NPSG, methods to implement their use become better guided. A limitation of our study is that we did not survey patients about their perceptions of whiteboards use, an important needs assessment that would further drive this patient‐centered intervention. Regardless, we can draw a number of lessons from our findings and devise a set of reasonable recommendations.

Recommendations

We provide the following set of recommendations for hospitals adopting patient whiteboards, drawing on our survey findings and experiences with implementation at our own institution. We also acknowledge the role that local hospital cultures may play in adopting whiteboard use, and our recommendations are simply guidelines that can be applied or used in planning efforts. We believe effective use of a patient whiteboard requires a patient‐centered approach and the following:

• 1

  Whiteboards should be placed in clear view of patients from their hospital bed

  A simple yet critical issue as placing a whiteboard behind a patient's bed or off to the side fails to provide them with a constant visual cue to engage in the information.

• 2

  Buy and fasten erasable pens to the whiteboards themselves

  In our institution, purchasing pens for each provider was a less effective strategy than simply affixing the pen to the whiteboard itself. A supply of erasable pens must be available at the nursing station to quickly replace those with fading ink.

• 3

  Create whiteboard templates

  Our findings and experience suggest that structured formats for whiteboards may be more effective in ensuring both important and accurate information gets included. Blank whiteboards lead to less standardization in practice and fail to create prompts for providers to both write and review the content available. Anecdotally, we created a number of whiteboards with templated information, and this did seem to increase the consistency, standardization, and ease of use.

• 4

  Whiteboard templates should include the following items:
• a

  Day and Date

  This serves to orient patients (and their families) as well as providers with the date of information written on the whiteboard. It is also an important mechanism to ensure information is updated daily.

• b

  Patient's name (or initials)

  With bed turnover (or patient transfers to different beds and units) commonplace in hospital care, we believe that listing the patient's name on the board prevents the potential for patients (and their families) or providers to mistakenly take information from a previous patient's care on the whiteboard for their own.

• c

  Bedside nurse

  This was noted as the most useful provider listed on a patient whiteboard, which is quite logical given the role bedside nurses play for hospitalized patients.

• d

  Primary physician(s) (attending, resident, and intern, if applicable)

  This was noted as the next most important provider(s) and perhaps increasingly important both in teaching and nonteaching settings where shift‐work and signouts are growing in frequency among physicians.

• e

  Goal for the day

  While this was not a consensus from our survey respondents, we believe patients (rather than providers) should ultimately guide determination of their goal for the day as this engages them directly with the planachieving a patient‐centered initiative. In our experience, an effective strategy was having the bedside nurse directly engage patients each morning to help place a goal for the day on their whiteboard.

• f

  Anticipated discharge date

  While understanding the potential for this date to change, we believe the benefits of having patients (and their families) thinking about discharge, rather than feeling surprised by it on the morning of discharge, serves as an important mechanism to bridge communication about the discharge process.24

• g

  Family member's contact information (phone number)

• h

  Questions for providers

  This last entry allows a space for families to engage the healthcare team and, once again, create an opportunity for clarification of treatment and discharge plans.

• 5

  Bedside nurses should facilitate writing and updating information on the whiteboard

  Without our survey findings, this might have generated debate or controversy over whether nurses should be burdened with one more task to their responsibilities. However, our nurse respondents embraced this responsibility with spontaneous comments about their patient advocate role, and stated that whiteboards can serve as a tool to assist in that responsibility. Furthermore, not a single nurse respondent stated as barrier to use that I didn't think it was my responsibility. Nonetheless, whiteboard use must be a shared communication tool and not simply a tool between nurse and patient. Practically, we would recommend that bedside nurses facilitate updating whiteboards each morning, at a time when they are already helping patients create a goal for the day. Other providers must be trained to review information on the whiteboard, engage patients about their specific goal, and share the responsibility of keeping the information on the whiteboard updated.

• 6

  Create a system for auditing utilization and providing feedback early during rollout

  We found that adoption was very slow at the outset. One strategy to consider is having designated auditors check whiteboards in each room, measuring weekly compliance and providing this feedback to nurse managers. This auditing process may help identify barriers that can be addressed quickly (eg, unavailability of pens).

Finally, it is important to comment on the confidentiality concerns often raised in the context of whiteboard use. Confidentiality concerns largely arise from personal health information being used without a patient's explicit consent. If our recommendations are adopted, they require whiteboard use to be a patient‐centered and patient‐driven initiative. The type of information on the whiteboard should be determined with sensitivity but also with consent of the patient. We have not experienced any concerns by patients or providers in this regard because patients are told about the goals of the whiteboard initiative with our above principles in mind.

Conclusions

Patient whiteboards may improve communication among members of the healthcare team (eg, nurses, physicians, and others) and between providers and their patients (and family members). Further investigation is warranted to determine if adopting our recommendations leads to improved communication, teamwork, or patient satisfaction and care. In the meantime, as many hospitals continue to install and implement whiteboards, we hope our recommendations, accompanied by an emphasis on creating a patient‐centered communication tool, offer a roadmap for considering best practices in their use.

Communication failures are a frequent cause of adverse events14; the Joint Commission (TJC) reports that such failures contributed to 65% of reported sentinel events.5 Strategies to improve communication have focused on implementing formal teamwork training programs and/or teaching specific communication skills.613 While these strategies largely address communication between healthcare providers, there is a growing emphasis on developing strategies to engage patients in their care, and improving communication with them and their families.

In 2007, TJC announced a new National Patient Safety Goal (NPSG) that encourage(s) patients' active involvement in their own care as a patient safety strategy.14 This builds upon a landmark Institute of Medicine report that highlighted patient‐centeredness as 1 of the 6 domains for delivering high‐quality care.15 Current literature on developing such patient‐centered strategies enumerates several approaches, including better access to health information, use of innovative technology solutions, and focused efforts at improving communication.1618

The placement of whiteboards in patient rooms is an increasingly common strategy to improve communication. These boards, typically placed on a wall near a patient's hospital bed, allow any number of providers to communicate a wide range of information. Both Kaiser Permanente's Nurse Knowledge Exchange program and the Institute for Healthcare Improvement's Transforming Care at the Bedside promote whiteboard use, though with little specific guidance about practical implementation.19^,20 Despite their growing prevalence, there is no published literature guiding the most effective uses of whiteboards, or describing their impact on communication, teamwork, or patient satisfaction and care. We present findings from a survey of patient whiteboard use on an academic medical service, and offer a series of recommendations based on our findings and experiences.

Methods

We anonymously surveyed bedside nurses from 3 inpatient medical units, internal medicine housestaff, and faculty from the Division of Hospital Medicine at the University of California, San Francisco (UCSF). We solicited experiences of physician and nursing leaders who were engaged in whiteboard interventions over the past 2 years to identify relevant topics for study. Their experiences were based on isolated unit‐based efforts to implement whiteboards through a variety of strategies (eg, whiteboard templates, simple identification of provider teams, goals for the day). Their input guided the survey development and the suggested recommendations. The topics identified were then translated into multiple‐choice questions, and further edited for clarity by the authors. A Likert scale was used that measured frequency of use, usefulness, and attitudes toward patient whiteboards. An open‐ended question seeking additional comments about patient whiteboards was also asked. The survey was administered to nurses at staff meetings and through physical mailboxes on their respective patient care units with a 1‐month collection period. The survey was administered to housestaff and attendings via e‐mail listserves using an online commercial survey administration tool.21 The nursing surveys were later entered into the same online survey administration tool, which ultimately provided summary reports and descriptive findings to meet the study objectives. Our project was reviewed and approved by the UCSF Committee on Human Research.

Results

Survey responses were collected from 104 nurse respondents (81% response rate), 118 internal medicine housestaff (74% response rate), and 31 hospitalists (86% response rate). Nurses were far more likely to write on whiteboards, read what was written on them, and find the related information useful (Figure 1A‐C). Nurses, housestaff, and attendings all believed the bedside nurse was the single most important provider name listed on a whiteboard. However, the respondents differed in their rated value of other providers listed on the whiteboard (Figure 2). Nurses gave higher ratings to the utility of having patient care assistants (PCAs) listed as compared to housestaff and attendings. Overall, respondents felt it would be less useful to list consultants and pharmacists than the nurse, attending, and housestaff. All of the respondents believed family contact information was the most useful information on a whiteboard, whereas more nurses rated a goal for the day and anticipated discharge date as more useful than housestaff and attendings (Figure 3).

Figure 1

Figure 2

Figure 3

From an operational standpoint, the majority of respondents felt that nurses should be responsible for the information on a whiteboard, nurses and physicians together should create goals for the day, and the greatest barrier to using whiteboards was not having pens easily available (Figure 4A‐C). Most respondents also agreed that using templated whiteboards (with predefined fields) to guide content would increase their use (Figure 4D). All respondents believed that whiteboard use could improve teamwork and communication as well as patient care (Figure 5). Respondents also offered a variety of specific comments in response to an open‐ended question about whiteboard use (Table 1).

Figure 4

Figure 5

Discussion

Our findings demonstrate the potential value of patient whiteboards, which is supported by the vast majority of respondents, who agreed their use may improve patient care and teamwork. It is also clear that whiteboard use is not achieving this potential or being used as a patient‐centered tool. This is best illustrated by findings of their low rate of use and completion among attendings and housestaff (Figure 1A, B) and the lack of consensus as to what information on the whiteboards is useful. Patient whiteboards require defined goals, thoughtful planning, regular monitoring, and ongoing evaluation. The challenges around effective adoption and implementation is perhaps more about ensuring compliance and completion rather than simply gaining buy‐in and engagement for their value.

While the differential use of whiteboards between nurses and physicians was not surprising, a few specific findings warrant further discussion. First, it is interesting that nurses rated their own names and that of PCAs as the most useful, while physicians rated the nurse's name as being of equal value to their own. This may speak to the role PCAs play for nurses in helping the latter provide bedside care, rather than a reflection of the nurses' perception of the value of PCAs for patients. Second, while all respondents rated highly the value of family contact information on the whiteboard, nurses valued a goal for the day and anticipated discharge date more highly than did physicians. These findings likely reflect that nurses desire an understanding about plans of care and if they are not communicated face‐to‐face as the most effective strategy,22 they should at least be spelled out clearly on a whiteboard. This is supported by evidence that better collaboration between nurses and physicians improves patient outcomes.23 It may also be that physicians place more value on their own progress notes (rather than whiteboards) as a vehicle for communicating daily goals and discharge planning.

Other practical considerations involve who owns it and, if we do create goals for the day, whose goals should they represent? The majority of nurse and physician responses advocated for nurses to be responsible for accurate and complete information being updated on whiteboards. A larger percentage of attendings favored shared responsibility of the whiteboard, which was reinforced by their support of having goals for the day created jointly by nurses and physicians. Interestingly, a much smaller percentage of respondents felt goals for the day should be driven by patients (or family members). These data may point to the different perspectives that each individual provider bringsphysician, nurse, pharmacist, discharge plannerwith their respective goals differing in nature. Finally, it is also interesting that while attendings and housestaff believed that whiteboards can improve patient care teamwork/communication (Figure 5), a much smaller percentage actually read what is on them (Figure 1B). This may reflect the unclear goals of whiteboards, its absence as part of daily workflow, the infrequency of updated information on them, or perhaps an institution‐specific phenomenon that we will use to drive further improvement strategies.

Selected respondent comments (Table 1) highlight important messages about whiteboard use and provide helpful context to the survey responses. We found that the goal of whiteboard use is not always clear; is it to improve communication among providers, to improve communication with patients, a tool to engage patients in their care, or some combination of the above? Without a clear goal, providers are left to wonder whether whiteboard use is simply another task or really an intervention to improve care. This may in part, or perhaps fully, explain the differences discovered in whiteboard use and practices among our surveyed providers.

If, however, one were to make clear that the goal of patient whiteboards is to engage patients in their care and help achieve an important NPSG, methods to implement their use become better guided. A limitation of our study is that we did not survey patients about their perceptions of whiteboards use, an important needs assessment that would further drive this patient‐centered intervention. Regardless, we can draw a number of lessons from our findings and devise a set of reasonable recommendations.

Recommendations

We provide the following set of recommendations for hospitals adopting patient whiteboards, drawing on our survey findings and experiences with implementation at our own institution. We also acknowledge the role that local hospital cultures may play in adopting whiteboard use, and our recommendations are simply guidelines that can be applied or used in planning efforts. We believe effective use of a patient whiteboard requires a patient‐centered approach and the following:

• 1

  Whiteboards should be placed in clear view of patients from their hospital bed

  A simple yet critical issue as placing a whiteboard behind a patient's bed or off to the side fails to provide them with a constant visual cue to engage in the information.

• 2

  Buy and fasten erasable pens to the whiteboards themselves

  In our institution, purchasing pens for each provider was a less effective strategy than simply affixing the pen to the whiteboard itself. A supply of erasable pens must be available at the nursing station to quickly replace those with fading ink.

• 3

  Create whiteboard templates

  Our findings and experience suggest that structured formats for whiteboards may be more effective in ensuring both important and accurate information gets included. Blank whiteboards lead to less standardization in practice and fail to create prompts for providers to both write and review the content available. Anecdotally, we created a number of whiteboards with templated information, and this did seem to increase the consistency, standardization, and ease of use.

• 4

  Whiteboard templates should include the following items:
• a

  Day and Date

  This serves to orient patients (and their families) as well as providers with the date of information written on the whiteboard. It is also an important mechanism to ensure information is updated daily.

• b

  Patient's name (or initials)

  With bed turnover (or patient transfers to different beds and units) commonplace in hospital care, we believe that listing the patient's name on the board prevents the potential for patients (and their families) or providers to mistakenly take information from a previous patient's care on the whiteboard for their own.

• c

  Bedside nurse

  This was noted as the most useful provider listed on a patient whiteboard, which is quite logical given the role bedside nurses play for hospitalized patients.

• d

  Primary physician(s) (attending, resident, and intern, if applicable)

  This was noted as the next most important provider(s) and perhaps increasingly important both in teaching and nonteaching settings where shift‐work and signouts are growing in frequency among physicians.

• e

  Goal for the day

  While this was not a consensus from our survey respondents, we believe patients (rather than providers) should ultimately guide determination of their goal for the day as this engages them directly with the planachieving a patient‐centered initiative. In our experience, an effective strategy was having the bedside nurse directly engage patients each morning to help place a goal for the day on their whiteboard.

• f

  Anticipated discharge date

  While understanding the potential for this date to change, we believe the benefits of having patients (and their families) thinking about discharge, rather than feeling surprised by it on the morning of discharge, serves as an important mechanism to bridge communication about the discharge process.24

• g

  Family member's contact information (phone number)

• h

  Questions for providers

  This last entry allows a space for families to engage the healthcare team and, once again, create an opportunity for clarification of treatment and discharge plans.

• 5

  Bedside nurses should facilitate writing and updating information on the whiteboard

  Without our survey findings, this might have generated debate or controversy over whether nurses should be burdened with one more task to their responsibilities. However, our nurse respondents embraced this responsibility with spontaneous comments about their patient advocate role, and stated that whiteboards can serve as a tool to assist in that responsibility. Furthermore, not a single nurse respondent stated as barrier to use that I didn't think it was my responsibility. Nonetheless, whiteboard use must be a shared communication tool and not simply a tool between nurse and patient. Practically, we would recommend that bedside nurses facilitate updating whiteboards each morning, at a time when they are already helping patients create a goal for the day. Other providers must be trained to review information on the whiteboard, engage patients about their specific goal, and share the responsibility of keeping the information on the whiteboard updated.

• 6

  Create a system for auditing utilization and providing feedback early during rollout

  We found that adoption was very slow at the outset. One strategy to consider is having designated auditors check whiteboards in each room, measuring weekly compliance and providing this feedback to nurse managers. This auditing process may help identify barriers that can be addressed quickly (eg, unavailability of pens).

Finally, it is important to comment on the confidentiality concerns often raised in the context of whiteboard use. Confidentiality concerns largely arise from personal health information being used without a patient's explicit consent. If our recommendations are adopted, they require whiteboard use to be a patient‐centered and patient‐driven initiative. The type of information on the whiteboard should be determined with sensitivity but also with consent of the patient. We have not experienced any concerns by patients or providers in this regard because patients are told about the goals of the whiteboard initiative with our above principles in mind.

Conclusions

Patient whiteboards may improve communication among members of the healthcare team (eg, nurses, physicians, and others) and between providers and their patients (and family members). Further investigation is warranted to determine if adopting our recommendations leads to improved communication, teamwork, or patient satisfaction and care. In the meantime, as many hospitals continue to install and implement whiteboards, we hope our recommendations, accompanied by an emphasis on creating a patient‐centered communication tool, offer a roadmap for considering best practices in their use.

Acute pancreatitis accounts for more than 220,000 hospital admissions in the United States annually.1 In the following review, we outline the etiology of acute pancreatitis, discuss its complications, and provide an updated review on its management for the hospitalized patient.

Etiology

Gallstone disease and excess alcohol ingestion are the most common causes of acute pancreatitis in the United States. Gallstones account for roughly 45% of all cases, and the pathogenesis is due to transient obstruction of the pancreatic duct orifice to the flow of pancreatic exocrine secretions.2 Excess alcohol ingestion accounts for approximately 35% of all cases, yet the pathogenesis here is less understood.3 Most theories suggest a direct toxic effect of the ethanol upon the pancreatic parenchyma or its neurovascular supply.4

There are many other less common causes of acute pancreatitis including toxins, drugs, infections, trauma, vascular insults, anatomic abnormalities, and metabolic derangements. Hypertriglyceridemia and hypercalcemia are both implicated in acute pancreatitis. Serum triglyceride levels >1000 mg/dL can precipitate an attack of acute pancreatitis though the pathogenesis is not clearly understood.5 Hypercalcemia is also an uncommon cause of acute pancreatitis, and is thought to result from deposition of calcium in the pancreatic duct and calcium activation of trypsinogen.6

Idiopathic pancreatitis occurs in up to 20% of patients with acute pancreatitis, and by definition, the cause is not established by history, physical examination, routine laboratory tests, or imaging. The majority of idiopathic cases of pancreatitis are thought to have a biliary source. In patients with gallbladder in situ, it is estimated that up to 75% acquire pancreatitis from microlithiasis, or biliary sludge and stone debris, that causes obstruction of the distal common bile and main pancreatic ducts. Conversely, sphincter of Oddi dysfunction (SOD) resulting in transient pancreatic ductal obstruction is felt to be the most common cause in those patients who have undergone a previous cholecystectomy.7

An emerging entity, autoimmune pancreatitis (AIP), is more commonly associated with chronic pancreatitis but may cause episodes of acute pancreatitis or mimic pancreatic carcinoma. Typically, the diagnosis is based on elevated levels of serum gammaglobulin subgroup 4 (IgG4) populations, along with characteristic findings on computed tomography (CT) scan (eg, narrowed or wispy main pancreatic duct and an enlarged pancreatic parenchyma). Core‐needle biopsy may confirm the diagnosis of AIP with lymphoplasmacytic infiltration and dense fibrosis.8 Since AIP can mimic pancreatic cancer, the diagnosis may not be made until the time of surgical resection.

Diagnosis

Along with characteristic symptoms, the diagnosis of acute pancreatitis is often based on elevated serum levels of pancreatic enzymes that are at least twice the normal level. Amylase and lipase are the most frequently used serum markers for acute pancreatitis, though their elevation is not pathognomonic for the presence of disease. These enzymes may not always be significantly elevated during times of acute inflammation, and elevation of the enzymes can come from nonpancreatic origins as well (Table 1). Although there is no gold standard for the diagnosis of acute pancreatitis, using serum lipase (>250 IU/L) in conjunction with amylase (>160 IU/L) improves the overall diagnostic sensitivity from 81% to 94%.9 Isoamylase levels can be used to distinguish among pancreatic, salivary, and macroamylasemia though this is not often used if pancreatitis is suspected clinically. Similarly, serum isolipase can be measured, though this is not readily available.

In order to improve the sensitivity and specificity of diagnosis, other tests have been studied to help predict disease presence and severity. Previously, serum tests for trypsin, elastase, phospholipase A2, and carboxylester lipase have all been evaluated but shown to have no significant improvement in diagnostic capability.1014 More recently, trypsinogen (a pancreatic proteinase) has proven to be a useful aid in the accurate diagnosis of acute disease. Trypsinogen undergoes activation into trypsin during acute pancreatic inflammation.3 It is comprised of 2 main isoenzymes (trypsinogen‐1 and trypsinogen‐2) that are secreted into the pancreatic fluid with a small proportion escaping into the circulation.15 Higher concentrations of trypsinogen‐1 are seen in healthy people, while higher concentrations of trypsinogen‐2 are seen in those with acute pancreatitis.16 Urinary trypsinogen‐2 dipstick tests detect acute pancreatitis more accurately than quantitative serum or urinary amylase, with a sensitivity as high as 94%, and a specificity of 95%.17 Studies have shown that in post‐endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis, serum trypsinogen‐2 levels begin to rise as early as 1 hour and peak at 6 hours.17 The Actim Pancreatitis (Medix Biomedica, Kauniainen, Finland) urine test strips measure concentrations of trypsinogen‐2 as low as 50 g/L, but is not a quantitative test and, thus, it does not predict severity. Some studies have advocated the use of urinary trypsinogen‐2 as a screening tool, with a positive result indicating a need for further evaluation of acute pancreatitis.1820 Urinary trypsinogen‐2 is less costly than serum tests, plus may result in additional cost savings with earlier patient discharge. Unfortunately, this test is not widely available for clinical use. Urinary trypsinogen activation peptide (TAP) is another test that has been studied in the diagnosis of acute pancreatitis, but may signify disease severity rather than the presence or absence of disease.21 Currently urinary assays for TAP are not widely available in the United States.

Choosing the Appropriate Imaging Modality

Along with the measurement of pancreatic release enzymes, abdominal imaging is often used, though not always necessary to confirm the diagnosis of acute pancreatitis. Imaging techniques such as CT, magnetic resonance imaging (MRI), and transabdominal ultrasonography may be used to rule out other causes of abdominal pain or elucidate the cause of the pancreatitis itself. Ultrasound may show pancreatic enlargement, diminished echogenicity, and possible adjacent fluid collections.22 In searching for evidence of gallstone pancreatitis, transabdominal ultrasound has a sensitivity of 67% and a specificity of 100%.23 However, it may be insensitive for detecting stones in the distal common bile duct near the ampulla due to acoustic interference from gas within the small bowel.24 Furthermore, ultrasound itself is operator‐dependent.

Contrast‐enhanced CT is the standard mode of imaging for diagnosing acute pancreatitis and provides superior imaging of the pancreas. Unfortunately it is more costly than ultrasound, involves radiation exposure, and requires intravenous contrast medium.25 Findings of acute pancreatitis frequently seen on CT include diffuse or segmental enlargement of the gland, irregular pancreatic contour, obliteration of peripancreatic fat planes, parenchymal heterogeneity, and ill‐defined fluid collections within the pancreas or in the lesser sac and pararenal spaces.26 CT scan may also be used to detect pancreatic necrosis, an important finding for the management and prognosis of this disease.27 Despite this, normal CT findings have been reported in patients with acute pancreatitis, and certain CT findings may be related to disease severity.25

Although MRI is less commonly used in the diagnosis of acute pancreatitis, it may provide a useful alternative to CT, especially in cases of renal failure or intravenous contrast hypersensitivity. When combined with magnetic resonance cholangiopancreatography (MRCP) imaging, MRI may even be able to detect a local area of pancreatic duct disruption.27 MRCP allows for a noninvasive cholangiogram and is frequently used to stratify patients who may benefit from ERCP. It can accurately identify common bile duct stones, with a higher sensitivity for choledocholithiasis than ultrasound or CT.2830 MRCP can also assist in the diagnosis of other disorders of the intrahepatic and extrahepatic biliary tree that may be related to the cause of pancreatitis. Overall, unless a patient has a contraindication, or the goal of the study is to diagnose choledocholithiasis, a contrast‐enhanced CT scan remains the imaging procedure of choice due to improved accessibility, lower cost, ease of performance, and increased sensitivity in the detection of gas bubbles (potentially indicating pancreatic infection).3133 Ordering a CT scan or other imaging at admission is not necessary in the diagnosis of acute pancreatitis if the patient's presentation is classic. At admission, however, a CT scan may be reasonable to exclude other serious causes of abdominal pain, such as a perforated ulcer. Imaging may also be ordered to define the cause of the episode of pancreatitis and to exclude occult malignancy. In addition, CT scan should be strongly considered in patients who do not improve within 2 to 3 days to assess for complications such as pancreatic necrosis, pseudocysts, or other complications.34

Most recently, endoscopic ultrasound (EUS) has risen to the forefront as a leader in accurate imaging of the pancreas and biliary tree. EUS is more sensitive than transabdominal ultrasound in detecting biliary stones,35 and it has been shown to have equivalent, and in some cases superior, sensitivity to ERCP and MRCP. Because EUS is able to detect smaller stones or sludge, it may have a role in those patients diagnosed with idiopathic pancreatitis.36 Like MRCP, EUS can also help stratify patients into those that are likely to benefit most from ERCP.37 Figure 1 reviews the evaluation of acute pancreatitis.

Figure 1

Prognosis

For the majority of patients with acute pancreatitis, the clinical course is mild and self‐limiting. In approximately 20% to 25% of patients, however, it is severe and associated with organ failure and significant morbidity and mortality.38, 39 Determining the severity of acute pancreatitis is critical, as patients at high‐risk for severe disease require closer monitoring and possible intervention. Several validated scoring systems are available that aim to predict the severity of acute pancreatitis including Ranson's criteria, the Imrie scoring system, the Acute Physiology and Chronic Health Evaluation (APACHE II) scale, and the CT Severity Index (CTSI) (Table 2).4043

In 1992, the Atlanta Classification of acute pancreatitis was developed to provide a rational approach in predicting disease severity, thus allowing for comparison between clinical trials. It defines severe acute pancreatitis (SAP) on the basis of standard clinical manifestations, a Ranson's score 3, an APACHE II score 8, and evidence of organ failure and intrapancreatic pathological findings.44 Serum markers such as C‐reactive protein (CRP), interleukin‐6, and phospholipase A2 have all been studied to predict severity; however, only CRP is widely available. A cutoff level of 150 mg/L at 48 hours distinguishes mild disease from SAP.45 Clinical findings such as thirst, poor urine output, progressive tachycardia, tachypnea, hypoxemia, confusion, and a lack of improvement in symptoms within the first 48 hours are warning signs of impending severe disease, and thus warrant consideration of admission to an intensive care unit (ICU).34

Natural History and Complications

Despite initial aggressive intensive care treatment, 30% to 50% of patients with SAP do not respond promptly to ICU treatment and develop persistent multisystem organ failure.39 Severe organ failure in the first week of onset of acute pancreatitis is closely linked to the development of pancreatic infection occurring within 2 weeks of the initiation of symptoms.46 Early multiorgan dysfunction triggers additional mechanisms that render bacterial translocation into clinically manifested sepsis and septic shock.39 In most studied series, infection (including bacteremia, fungemia, and pancreatic abscess) remains the leading cause of death in patients with acute pancreatitis, accounting for up to 80% of fatal cases.4749 While sepsis is the more frequent cause of death in patients surviving beyond 7 days, death occurring early in the course of disease is more likely to be from respiratory complications such as pulmonary edema.50

In the spectrum of acute pancreatitis, ongoing pancreatic injury can lead to pancreatic necrosis, fluid collections, pseudocyst formation, and pancreatic duct disruption (Figures 24).51 In patients hospitalized with acute pancreatitis, up to 57% will have peripancreatic fluid collections that are initially ill‐defined.44, 52 Typically, these fluid collections may be managed conservatively; however, if they continue to enlarge, cause persistent abdominal pain, become infected, or compress adjacent organs, they may require further intervention.53 Ductal disruption may be diagnosed when fluid collections have high levels of pancreatic amylase, and their presence may lead to the formation of pseudocysts, persistent ascites, or pleural effusions.54 Pancreatic pseudocysts usually require 4 weeks for complete formation, and they classically contain fluid only without significant solid debris.55 Formation typically occurs as a result of limited pancreatic necrosis causing a pancreatic duct leak with subsequent organization, or from areas of necrosis that liquefy over time.56 Both pancreatic pseudocysts and necrotic pancreatic tissue may become infected leading to abscess formation.51

Figure 2

Figure 3

Figure 4

Pancreatic necrosis is defined as diffuse or focal areas of nonviable pancreatic parenchyma, and it is seen in approximately 20% of patients with acute pancreatitis.44, 57 While pseudocyst formation takes approximately 1 month to occur, pancreatic necrosis can occur within the first few days of initial symptoms and is associated with an increase in complications leading to an increased risk of morbidity and mortality.58 More than 80% of deaths in acute pancreatitis are associated with the presence of pancreatic necrosis.39 Patients at highest risk for complications are those with necrosis involving more than 50% of the gland based on MRI or contrast‐enhanced CT scan.59, 60

Patients with pancreatic infection may have infected necrosis, pancreatic abscess, and/or infected pseudocysts.39 The microbes most frequently involved are gram‐negative organisms including Escherichia coli, Enterococcus, and Klebsiella.61 Recently, gram‐positive bacteria have been implicated in pancreatic infection.62 Fungal infection with Candida species is seen in up to 15% of patients with infected necrosis and is associated with more serious systemic complications.63 The use of prophylactic antibiotics may increase the risk of fungal infection. It may be challenging to distinguish between infected and sterile pancreatic necrosis; hence, needle aspiration under EUS or radiologic guidance may be required.61, 64

Management

Management of Pancreatic Necrosis

Sterile pancreatic necrosis is typically managed conservatively without drainage. Generally, CT scans are repeated every 7 to 10 days to assess the necrosis and to evaluate for further complications.32 Patients who are clinically unstable with fever, tachycardia, leukocytosis, or organ failure may require percutaneous sampling to evaluate for infected necrosis.33 If the pancreatic tissue is sterile, the patient is determined to have sterile necrosis. If the patient with sterile necrosis is clinically unstable then prophylactic antibiotics may be indicated. If the pancreatic tissue is infected, the patient is deemed to have infected necrosis and treatment with antibiotics and necrosectomy is often indicated, especially in those with a poor clinical state. The antibiotic chosen should have adequate penetration into the necrotic material, such as imipenem, meropenem, or a combination of quinolone and metronidazole.99

It may be challenging to distinguish between sterile and infected pancreatic necrosis. A CT scan is unable to differentiate them with certainty; though, intrapancreatic, retroperitoneal, or lesser sac gas may indicate infection.31 In addition, inducing infection within a previously sterile collection is a potential risk of percutaneous sampling. As a result, sampling should not be performed unless completely indicated.31

In patients with sterile pancreatic necrosis who are symptomatic with refractory abdominal pain, gastric outlet obstruction, or failure to thrive at 4 or more weeks following the onset of acute pancreatitis, drainage and/or debridement is usually indicated. Pancreatic necrosectomy for sterile pancreatic necrosis may be accomplished endoscopically, or more traditionally by a surgical approach.55 Although endoscopic drainage is less invasive, it is technically difficult and has a higher rate of complication in the hands of inexperienced operators.100 Careful selection and evaluation of patients undergoing endoscopic drainage procedures is necessary. Bleeding, perforation, infection, pancreatitis, aspiration, stent migration, and pancreatic ductal damage are all possible complications during the drainage of necrotic pancreatic fluid collections.55 If pancreatic necrosis is infected, surgical necrosectomy should be performed as this is the gold standard for infected necrosis when debridement is necessary.55 Figure 5 reviews the management of acute pancreatitis.

Figure 5

Conclusion

Acute pancreatitis is a common disease frequently caused by choledocholithiasis or excess alcohol ingestion. In idiopathic acute pancreatitis, microlithiasis and SOD should be considered. Though CT scan remains the imaging modality of choice, newer methods such as MRCP and EUS may help to provide additional and improved diagnostic information.

The management of acute pancreatitis is frequently challenging, and severity scales help to predict the likelihood of complications, determine necessary interventions, and guide the appropriate level of care. Nutrition is critical in patients with SAP, and enteral feeding is clearly preferred over TPN. Currently, prophylactic antibiotics do not appear to have a role in SAP. Finally, though not always straightforward, recommendations do exist to guide the management of many of the complications of acute pancreatitis, such as pseudocyst formation and necrotizing disease. A multidisciplinary approach should be used in managing patients with severe disease, and the primary inpatient physician should not hesitate to involve specialists, including gastroenterologists, radiologists, and surgeons.