Editorial

After entertaining the possibility of acute intervention, the majority of hospitalists efforts in the management of patients with ischemic stroke involve identifying an etiology and initiating secondary prevention strategies. Other than evaluating stroke risk factors, workup has traditionally involved extracranial and intracranial vessel imaging, cardiac telemetry, and echocardiography. Even after exhaustive searches, no cause for stroke is found in nearly 25% of cases, leading to a recent focus on determining why these so‐called cryptogenic strokes happen and how to prevent their recurrence.[1, 2]

Echocardiography is commonly obtained in most patients with ischemic stroke, but its yield is probably modest at best. Although transesophageal echocardiography (TEE) may be superior to transthoracic echocardiography (TTE) for determining an etiology of stroke, whether these findings substantially change management remains debatable.[3, 4] In this issue of the Journal of Hospital Medicine, Marino and colleagues examined the yield of TEE in patients without a known cause of ischemic stroke following a normal TTE.[5] A possible cause of stroke was identified in 42%, including aortic plaques and patent foramen ovale (PFO), but in only 1 patient did this discovery change management.

Secondary prevention strategies in ischemic stroke outside of atrial fibrillation now almost exclusively involve antiplatelet medications.[6] Studies of secondary prevention in aortic arch atheromas, patients with depressed systolic function, and those with PFO have failed to demonstrate any strategy that is superior to antiplatelets, and therefore the bar is high to show that any TEE findings impact treatment other than obvious and rare smoking guns such as a rare valvular lesion, cardiac tumor, or atrial thrombus.[7, 8, 9]

What is more of a recent headline in stroke workup is the increasing emphasis on long‐term cardiac monitoring following discharge to detect those with atrial fibrillation, which likely comprise between 15% and 20% of cryptogenic stroke patients.[10] Finding atrial fibrillation clearly changes management and therefore has a higher yield than the vast majority of possible findings on echocardiography. Perhaps in patients in whom a TEE is being considered, extended monitoring should happen first as an outpatient, followed by TEE if the stroke etiology remains obscure. On the other hand, severe left atrial enlargement, thrombus in the atrium, or atrial spontaneous echo contrast (smoke) are features on echocardiography that might raise the suspicion of atrial fibrillation so high that anticoagulation could be considered while long‐term monitoring is being used to definitively prove an atrial arrhythmia.

The current study does have some limitations other than those inherent to its retrospective design. Patients were only included if they were older than 50 years. Some have advocated using TEE as the echocardiogram of choice in young stroke patients due to its perhaps higher yield in these individuals; this study does not address this strategy. At institutions such as ours, an abnormal TTE in a cryptogenic stroke patient is followed by a TEE, and this study again does not alter this approach, because only those with a normal TTE were included. The definition of a normal TTE used in the study was so narrow, including normal left ventricular systolic function, that a majority of stroke patients with vascular risk factors such as hypertension would have likely been excluded. Determining what features and quality of a TTE are so definitive that a TEE is not necessary will be an important thrust of additional research. However, because TEE shows a better view of the left atrial appendage, the aortic arch, and is probably a better shunt study compared with TTE, it is not clear if a normal TTE will ever be adequate to prevent this second more invasive study in selected patients.

At the heart of the matter for health systems is the cost‐effectiveness of any screening approach used to determine the etiology of acute ischemic stroke. Studies are underway that will likely demonstrate that long‐term monitoring for atrial fibrillation will be worth it. Although it is dubious that TEE would ever fall into the same category due to its low yield, one might imagine a scenario, as our workup for cryptogenic stroke becomes more and more complicated, where obtaining a TEE is cost‐effective simply because it gives an answer and therefore can halt further testing. Perhaps at the end of the day, a TEE will just allow us to say to our stroke patients that enough is enough.

Disclosures: Dr. Josephson receives personal compensation as Editor‐in‐Chief of the New England Journal of Medicine Journal Watch Neurology and in an editorial capacity for Continuum Audio.

After entertaining the possibility of acute intervention, the majority of hospitalists efforts in the management of patients with ischemic stroke involve identifying an etiology and initiating secondary prevention strategies. Other than evaluating stroke risk factors, workup has traditionally involved extracranial and intracranial vessel imaging, cardiac telemetry, and echocardiography. Even after exhaustive searches, no cause for stroke is found in nearly 25% of cases, leading to a recent focus on determining why these so‐called cryptogenic strokes happen and how to prevent their recurrence.[1, 2]

Echocardiography is commonly obtained in most patients with ischemic stroke, but its yield is probably modest at best. Although transesophageal echocardiography (TEE) may be superior to transthoracic echocardiography (TTE) for determining an etiology of stroke, whether these findings substantially change management remains debatable.[3, 4] In this issue of the Journal of Hospital Medicine, Marino and colleagues examined the yield of TEE in patients without a known cause of ischemic stroke following a normal TTE.[5] A possible cause of stroke was identified in 42%, including aortic plaques and patent foramen ovale (PFO), but in only 1 patient did this discovery change management.

Secondary prevention strategies in ischemic stroke outside of atrial fibrillation now almost exclusively involve antiplatelet medications.[6] Studies of secondary prevention in aortic arch atheromas, patients with depressed systolic function, and those with PFO have failed to demonstrate any strategy that is superior to antiplatelets, and therefore the bar is high to show that any TEE findings impact treatment other than obvious and rare smoking guns such as a rare valvular lesion, cardiac tumor, or atrial thrombus.[7, 8, 9]

What is more of a recent headline in stroke workup is the increasing emphasis on long‐term cardiac monitoring following discharge to detect those with atrial fibrillation, which likely comprise between 15% and 20% of cryptogenic stroke patients.[10] Finding atrial fibrillation clearly changes management and therefore has a higher yield than the vast majority of possible findings on echocardiography. Perhaps in patients in whom a TEE is being considered, extended monitoring should happen first as an outpatient, followed by TEE if the stroke etiology remains obscure. On the other hand, severe left atrial enlargement, thrombus in the atrium, or atrial spontaneous echo contrast (smoke) are features on echocardiography that might raise the suspicion of atrial fibrillation so high that anticoagulation could be considered while long‐term monitoring is being used to definitively prove an atrial arrhythmia.

The current study does have some limitations other than those inherent to its retrospective design. Patients were only included if they were older than 50 years. Some have advocated using TEE as the echocardiogram of choice in young stroke patients due to its perhaps higher yield in these individuals; this study does not address this strategy. At institutions such as ours, an abnormal TTE in a cryptogenic stroke patient is followed by a TEE, and this study again does not alter this approach, because only those with a normal TTE were included. The definition of a normal TTE used in the study was so narrow, including normal left ventricular systolic function, that a majority of stroke patients with vascular risk factors such as hypertension would have likely been excluded. Determining what features and quality of a TTE are so definitive that a TEE is not necessary will be an important thrust of additional research. However, because TEE shows a better view of the left atrial appendage, the aortic arch, and is probably a better shunt study compared with TTE, it is not clear if a normal TTE will ever be adequate to prevent this second more invasive study in selected patients.

At the heart of the matter for health systems is the cost‐effectiveness of any screening approach used to determine the etiology of acute ischemic stroke. Studies are underway that will likely demonstrate that long‐term monitoring for atrial fibrillation will be worth it. Although it is dubious that TEE would ever fall into the same category due to its low yield, one might imagine a scenario, as our workup for cryptogenic stroke becomes more and more complicated, where obtaining a TEE is cost‐effective simply because it gives an answer and therefore can halt further testing. Perhaps at the end of the day, a TEE will just allow us to say to our stroke patients that enough is enough.

Disclosures: Dr. Josephson receives personal compensation as Editor‐in‐Chief of the New England Journal of Medicine Journal Watch Neurology and in an editorial capacity for Continuum Audio.

After entertaining the possibility of acute intervention, the majority of hospitalists efforts in the management of patients with ischemic stroke involve identifying an etiology and initiating secondary prevention strategies. Other than evaluating stroke risk factors, workup has traditionally involved extracranial and intracranial vessel imaging, cardiac telemetry, and echocardiography. Even after exhaustive searches, no cause for stroke is found in nearly 25% of cases, leading to a recent focus on determining why these so‐called cryptogenic strokes happen and how to prevent their recurrence.[1, 2]

Echocardiography is commonly obtained in most patients with ischemic stroke, but its yield is probably modest at best. Although transesophageal echocardiography (TEE) may be superior to transthoracic echocardiography (TTE) for determining an etiology of stroke, whether these findings substantially change management remains debatable.[3, 4] In this issue of the Journal of Hospital Medicine, Marino and colleagues examined the yield of TEE in patients without a known cause of ischemic stroke following a normal TTE.[5] A possible cause of stroke was identified in 42%, including aortic plaques and patent foramen ovale (PFO), but in only 1 patient did this discovery change management.

Secondary prevention strategies in ischemic stroke outside of atrial fibrillation now almost exclusively involve antiplatelet medications.[6] Studies of secondary prevention in aortic arch atheromas, patients with depressed systolic function, and those with PFO have failed to demonstrate any strategy that is superior to antiplatelets, and therefore the bar is high to show that any TEE findings impact treatment other than obvious and rare smoking guns such as a rare valvular lesion, cardiac tumor, or atrial thrombus.[7, 8, 9]

What is more of a recent headline in stroke workup is the increasing emphasis on long‐term cardiac monitoring following discharge to detect those with atrial fibrillation, which likely comprise between 15% and 20% of cryptogenic stroke patients.[10] Finding atrial fibrillation clearly changes management and therefore has a higher yield than the vast majority of possible findings on echocardiography. Perhaps in patients in whom a TEE is being considered, extended monitoring should happen first as an outpatient, followed by TEE if the stroke etiology remains obscure. On the other hand, severe left atrial enlargement, thrombus in the atrium, or atrial spontaneous echo contrast (smoke) are features on echocardiography that might raise the suspicion of atrial fibrillation so high that anticoagulation could be considered while long‐term monitoring is being used to definitively prove an atrial arrhythmia.

The current study does have some limitations other than those inherent to its retrospective design. Patients were only included if they were older than 50 years. Some have advocated using TEE as the echocardiogram of choice in young stroke patients due to its perhaps higher yield in these individuals; this study does not address this strategy. At institutions such as ours, an abnormal TTE in a cryptogenic stroke patient is followed by a TEE, and this study again does not alter this approach, because only those with a normal TTE were included. The definition of a normal TTE used in the study was so narrow, including normal left ventricular systolic function, that a majority of stroke patients with vascular risk factors such as hypertension would have likely been excluded. Determining what features and quality of a TTE are so definitive that a TEE is not necessary will be an important thrust of additional research. However, because TEE shows a better view of the left atrial appendage, the aortic arch, and is probably a better shunt study compared with TTE, it is not clear if a normal TTE will ever be adequate to prevent this second more invasive study in selected patients.

At the heart of the matter for health systems is the cost‐effectiveness of any screening approach used to determine the etiology of acute ischemic stroke. Studies are underway that will likely demonstrate that long‐term monitoring for atrial fibrillation will be worth it. Although it is dubious that TEE would ever fall into the same category due to its low yield, one might imagine a scenario, as our workup for cryptogenic stroke becomes more and more complicated, where obtaining a TEE is cost‐effective simply because it gives an answer and therefore can halt further testing. Perhaps at the end of the day, a TEE will just allow us to say to our stroke patients that enough is enough.

Disclosures: Dr. Josephson receives personal compensation as Editor‐in‐Chief of the New England Journal of Medicine Journal Watch Neurology and in an editorial capacity for Continuum Audio.

Delirium is a form of acute brain failure that affects up to 64% of older hospitalized patients and is associated with a multitude of adverse outcomes.[1] Healthcare providers, regardless of clinical setting, do not identify delirium in approximately 75% of cases.[2, 3] The paucity of brief and simple delirium assessment tools has been a barrier to improving delirium recognition.

To address this unmet need, several ultrabrief (<30 seconds) delirium assessment tools have been recently studied. In this issue of the Journal of Hospital Medicine, Fick et al. evaluated 20 individual components of the 3‐minute diagnostic interview for delirium using the Confusion Assessment Method (3D‐CAM), which was recently validated in older hospitalized patients.[4, 5] They observed that the best‐performing single‐item delirium assessment was the months of the year backward (MOTYB) task from December to January. This task assesses for inattention, a cardinal feature of delirium. Using a cutoff of 1 or more errors, the MOTYB was 83% sensitive and 69% specific for delirium.[5] By adding name the day of the week, the sensitivity increased to 93% with similar specificity (64%). This supports research by O'Regan et al., who examined MOTYB, but defined a positive screen if they could not recite the months backward from December to July perfectly. They observed a sensitivity and specificity of 84% and 90%, respectively, in older hospitalized patients.[6]

The assessment of arousal, another feature of delirium, has also garnered significant interest as another ultrabrief delirium screening method. Arousal is the patient's responsiveness to the environment and can be assessed during routine clinical care. Fick et al. observed that impaired arousal using the 3D‐CAM was 19% sensitive for delirium. This is in contrast to others who have reported sensitivities of 64% to 84%.[7, 8, 9] The difference in sensitivity may in part be explained by the method used to detect arousal. The 3D‐CAM asks, Was the patient sleep/stuporous? or Was the patient hyperviglant? Previous studies used the Richmond Agitation Sedation Scale (RASS), an arousal scale based on eye contact and physical behaviors to assess patients from 5 (coma) to +4 (combative).[10] Therefore, it is important to consider the method of arousal assessment if using this feature for delirium screening.

These ultrabrief delirium assessments would be even more clinically useful if they identified patients at high risk for adverse outcomes. In this same journal issue, 2 studies evaluated the prognostic ability of several ultrabrief delirium assessments. Zadravecz et al. observed that an abnormal RASS was a moderately good predictor of 24‐hour mortality, with an area under the receiver operating characteristic curve of 0.82.[11] Yevchak et al. observed that an abnormal RASS or MOTYB was associated with longer hospital length of stays, increased in‐hospital mortality, and need for skilled nursing.[12]

Viewed as a whole, these studies represent a significant advancement in delirium measurement and have the potential to improve this quality‐of‐care issue. However, uncertainties still exist. (1) Can these ultrabrief delirium assessments be used as standalone assessments? Based upon current data, these assessments have a significant proportion of false negative and positive rates. The effect on such misclassification on patient outcomes and healthcare utilization needs to be clarified. Because of this concern, Fick et al. recommended performing a more specific delirium assessment in those who have a positive MOTYB screen.[5] (2) What is the optimal cutoff of the MOTYB task and does this cutoff vary in different patient populations? The optimal cutoff will depend on whether or not a more sensitive test (lower error threshold) or specific test (higher error threshold) is desired. The optimal cutoff may also depend on the patient population (eg, demented versus nondemented). (3) Most important to practicing hospitalist and patients, will introducing these ultrabrief delirium assessments improve delirium recognition and improve patient outcomes? The impetus for widespread implementation of these assessments would be strengthened if healthcare providers successfully applied these assessments in clinical practice and subsequently improved outcomes.

In conclusion, the MOTYB and the assessment of arousal may be reasonable alternatives to more conventional delirium screening, especially in clinical environments with significant time constraints. However, additional research is needed to better refine these instruments to the clinical environment they will be used and determine how they impact clinical care and patient outcomes.

Delirium is a form of acute brain failure that affects up to 64% of older hospitalized patients and is associated with a multitude of adverse outcomes.[1] Healthcare providers, regardless of clinical setting, do not identify delirium in approximately 75% of cases.[2, 3] The paucity of brief and simple delirium assessment tools has been a barrier to improving delirium recognition.

To address this unmet need, several ultrabrief (<30 seconds) delirium assessment tools have been recently studied. In this issue of the Journal of Hospital Medicine, Fick et al. evaluated 20 individual components of the 3‐minute diagnostic interview for delirium using the Confusion Assessment Method (3D‐CAM), which was recently validated in older hospitalized patients.[4, 5] They observed that the best‐performing single‐item delirium assessment was the months of the year backward (MOTYB) task from December to January. This task assesses for inattention, a cardinal feature of delirium. Using a cutoff of 1 or more errors, the MOTYB was 83% sensitive and 69% specific for delirium.[5] By adding name the day of the week, the sensitivity increased to 93% with similar specificity (64%). This supports research by O'Regan et al., who examined MOTYB, but defined a positive screen if they could not recite the months backward from December to July perfectly. They observed a sensitivity and specificity of 84% and 90%, respectively, in older hospitalized patients.[6]

The assessment of arousal, another feature of delirium, has also garnered significant interest as another ultrabrief delirium screening method. Arousal is the patient's responsiveness to the environment and can be assessed during routine clinical care. Fick et al. observed that impaired arousal using the 3D‐CAM was 19% sensitive for delirium. This is in contrast to others who have reported sensitivities of 64% to 84%.[7, 8, 9] The difference in sensitivity may in part be explained by the method used to detect arousal. The 3D‐CAM asks, Was the patient sleep/stuporous? or Was the patient hyperviglant? Previous studies used the Richmond Agitation Sedation Scale (RASS), an arousal scale based on eye contact and physical behaviors to assess patients from 5 (coma) to +4 (combative).[10] Therefore, it is important to consider the method of arousal assessment if using this feature for delirium screening.

These ultrabrief delirium assessments would be even more clinically useful if they identified patients at high risk for adverse outcomes. In this same journal issue, 2 studies evaluated the prognostic ability of several ultrabrief delirium assessments. Zadravecz et al. observed that an abnormal RASS was a moderately good predictor of 24‐hour mortality, with an area under the receiver operating characteristic curve of 0.82.[11] Yevchak et al. observed that an abnormal RASS or MOTYB was associated with longer hospital length of stays, increased in‐hospital mortality, and need for skilled nursing.[12]

Viewed as a whole, these studies represent a significant advancement in delirium measurement and have the potential to improve this quality‐of‐care issue. However, uncertainties still exist. (1) Can these ultrabrief delirium assessments be used as standalone assessments? Based upon current data, these assessments have a significant proportion of false negative and positive rates. The effect on such misclassification on patient outcomes and healthcare utilization needs to be clarified. Because of this concern, Fick et al. recommended performing a more specific delirium assessment in those who have a positive MOTYB screen.[5] (2) What is the optimal cutoff of the MOTYB task and does this cutoff vary in different patient populations? The optimal cutoff will depend on whether or not a more sensitive test (lower error threshold) or specific test (higher error threshold) is desired. The optimal cutoff may also depend on the patient population (eg, demented versus nondemented). (3) Most important to practicing hospitalist and patients, will introducing these ultrabrief delirium assessments improve delirium recognition and improve patient outcomes? The impetus for widespread implementation of these assessments would be strengthened if healthcare providers successfully applied these assessments in clinical practice and subsequently improved outcomes.

In conclusion, the MOTYB and the assessment of arousal may be reasonable alternatives to more conventional delirium screening, especially in clinical environments with significant time constraints. However, additional research is needed to better refine these instruments to the clinical environment they will be used and determine how they impact clinical care and patient outcomes.

Delirium is a form of acute brain failure that affects up to 64% of older hospitalized patients and is associated with a multitude of adverse outcomes.[1] Healthcare providers, regardless of clinical setting, do not identify delirium in approximately 75% of cases.[2, 3] The paucity of brief and simple delirium assessment tools has been a barrier to improving delirium recognition.

To address this unmet need, several ultrabrief (<30 seconds) delirium assessment tools have been recently studied. In this issue of the Journal of Hospital Medicine, Fick et al. evaluated 20 individual components of the 3‐minute diagnostic interview for delirium using the Confusion Assessment Method (3D‐CAM), which was recently validated in older hospitalized patients.[4, 5] They observed that the best‐performing single‐item delirium assessment was the months of the year backward (MOTYB) task from December to January. This task assesses for inattention, a cardinal feature of delirium. Using a cutoff of 1 or more errors, the MOTYB was 83% sensitive and 69% specific for delirium.[5] By adding name the day of the week, the sensitivity increased to 93% with similar specificity (64%). This supports research by O'Regan et al., who examined MOTYB, but defined a positive screen if they could not recite the months backward from December to July perfectly. They observed a sensitivity and specificity of 84% and 90%, respectively, in older hospitalized patients.[6]

The assessment of arousal, another feature of delirium, has also garnered significant interest as another ultrabrief delirium screening method. Arousal is the patient's responsiveness to the environment and can be assessed during routine clinical care. Fick et al. observed that impaired arousal using the 3D‐CAM was 19% sensitive for delirium. This is in contrast to others who have reported sensitivities of 64% to 84%.[7, 8, 9] The difference in sensitivity may in part be explained by the method used to detect arousal. The 3D‐CAM asks, Was the patient sleep/stuporous? or Was the patient hyperviglant? Previous studies used the Richmond Agitation Sedation Scale (RASS), an arousal scale based on eye contact and physical behaviors to assess patients from 5 (coma) to +4 (combative).[10] Therefore, it is important to consider the method of arousal assessment if using this feature for delirium screening.

These ultrabrief delirium assessments would be even more clinically useful if they identified patients at high risk for adverse outcomes. In this same journal issue, 2 studies evaluated the prognostic ability of several ultrabrief delirium assessments. Zadravecz et al. observed that an abnormal RASS was a moderately good predictor of 24‐hour mortality, with an area under the receiver operating characteristic curve of 0.82.[11] Yevchak et al. observed that an abnormal RASS or MOTYB was associated with longer hospital length of stays, increased in‐hospital mortality, and need for skilled nursing.[12]

Viewed as a whole, these studies represent a significant advancement in delirium measurement and have the potential to improve this quality‐of‐care issue. However, uncertainties still exist. (1) Can these ultrabrief delirium assessments be used as standalone assessments? Based upon current data, these assessments have a significant proportion of false negative and positive rates. The effect on such misclassification on patient outcomes and healthcare utilization needs to be clarified. Because of this concern, Fick et al. recommended performing a more specific delirium assessment in those who have a positive MOTYB screen.[5] (2) What is the optimal cutoff of the MOTYB task and does this cutoff vary in different patient populations? The optimal cutoff will depend on whether or not a more sensitive test (lower error threshold) or specific test (higher error threshold) is desired. The optimal cutoff may also depend on the patient population (eg, demented versus nondemented). (3) Most important to practicing hospitalist and patients, will introducing these ultrabrief delirium assessments improve delirium recognition and improve patient outcomes? The impetus for widespread implementation of these assessments would be strengthened if healthcare providers successfully applied these assessments in clinical practice and subsequently improved outcomes.

In conclusion, the MOTYB and the assessment of arousal may be reasonable alternatives to more conventional delirium screening, especially in clinical environments with significant time constraints. However, additional research is needed to better refine these instruments to the clinical environment they will be used and determine how they impact clinical care and patient outcomes.

In this issue of the Journal of Hospital Medicine, we introduce a new recurring feature, Choosing Wisely: Things We Do for No Reason. The series is based on a talk I have delivered for the past 4 years at the annual national meeting of the Society of Hospital Medicine, in which I highlight 4 diagnostic tests, therapies, or other clinical practices that are commonly performed even though they are of low value to our inpatients.

There are many reasons hospitalists order unnecessary tests or treatments, or employ unhelpful clinical practices. Unnecessary testing may occur when we are not familiar with the test itselfthe actual costs of the test, the operating characteristics of the test, or the evidence supporting its usefulness in specific situations. Some tests are ordered unnecessarily because we cannot retrieve usable results from a different hospital or even our own electronic medical records. We may order tests or treatments due to patient expectations, a perceived need to practice defensively, or economic incentives.

Finally, we may simply order tests because of our uncertainty in the absence of data or simply because they are traditional practices (the way we've always done it). Physicians often order tests and treatments and institute clinical practices learned in residency or fellowship training.[1, 2] Local norms and practices influence physician behavior.

We created Things We Do for No Reason (TWDFNR) as a platform for provocative discussions of practices that have become common parts of hospital care but have limited supporting evidence, or even have evidence refuting or justifiably challenging their value. Each article in TWDFNR will describe why the test, treatment, or other clinical practice is commonly employed, why it may not be of high value, in what circumstances it may actually be valuable, and what conclusions can be drawn from the evidence provided. TWDFNR pieces are not systematic reviews or meta‐analyses and do not represent black and white conclusions or clinical practice standards; they are meant as a starting place for research and active discussions among hospitalists and patients.

In many respects, the Choosing Wisely: Things We Do for No Reason series is an extension of the Choosing Wisely campaign created by the American Board of Internal Medicine Foundation. Like Choosing Wisely, we are focusing on individual tests, treatments, and other clinical practices that are not beneficial and are potentially harmful to patients. Practices discussed may not cause significant physical or financial harm at the time they are used, but they may have significant downstream effects.

The Choosing Wisely campaign has brilliantly identified 5 important hospital medicine low‐value practices, and we hope to identify many more. We hope this series will serve as a grassroots effort to uncover more Choosing Wisely‐type practices. As institutions create their own high‐value care committees, the Choosing Wisely: Things We Do for No Reason series can provide possible agenda items, or provide the opportunity for sites to carry out analyses of their own practices to see whether any of the TWDFNR topics provide local opportunities for implementing higher‐value practices.

Although we do not believe that reducing the low‐value practices that will appear in TWDFNR will, alone, solve our wasteful practices, we hope that highlighting them will remind individuals, institutions, and systems that targeting low‐value practices is a responsibility that we all must embrace. We accept that not everyone will agree that the practices we present are low value, but the conversation is important to have. We invite you to take part in the Choosing Wisely: Things We Do for No Reason conversation. Let us know whether you think the practices highlighted are low value or whether you disagree with the conclusions. We welcome unsolicited proposals for series topics submitted as a 500‐word prcis. Send us your prcis or ideas on low‐value adult or pediatric patient practices that we should highlight in this series by emailing us at [email protected].

Disclosure: Nothing to report.

In this issue of the Journal of Hospital Medicine, we introduce a new recurring feature, Choosing Wisely: Things We Do for No Reason. The series is based on a talk I have delivered for the past 4 years at the annual national meeting of the Society of Hospital Medicine, in which I highlight 4 diagnostic tests, therapies, or other clinical practices that are commonly performed even though they are of low value to our inpatients.

There are many reasons hospitalists order unnecessary tests or treatments, or employ unhelpful clinical practices. Unnecessary testing may occur when we are not familiar with the test itselfthe actual costs of the test, the operating characteristics of the test, or the evidence supporting its usefulness in specific situations. Some tests are ordered unnecessarily because we cannot retrieve usable results from a different hospital or even our own electronic medical records. We may order tests or treatments due to patient expectations, a perceived need to practice defensively, or economic incentives.

Finally, we may simply order tests because of our uncertainty in the absence of data or simply because they are traditional practices (the way we've always done it). Physicians often order tests and treatments and institute clinical practices learned in residency or fellowship training.[1, 2] Local norms and practices influence physician behavior.

We created Things We Do for No Reason (TWDFNR) as a platform for provocative discussions of practices that have become common parts of hospital care but have limited supporting evidence, or even have evidence refuting or justifiably challenging their value. Each article in TWDFNR will describe why the test, treatment, or other clinical practice is commonly employed, why it may not be of high value, in what circumstances it may actually be valuable, and what conclusions can be drawn from the evidence provided. TWDFNR pieces are not systematic reviews or meta‐analyses and do not represent black and white conclusions or clinical practice standards; they are meant as a starting place for research and active discussions among hospitalists and patients.

In many respects, the Choosing Wisely: Things We Do for No Reason series is an extension of the Choosing Wisely campaign created by the American Board of Internal Medicine Foundation. Like Choosing Wisely, we are focusing on individual tests, treatments, and other clinical practices that are not beneficial and are potentially harmful to patients. Practices discussed may not cause significant physical or financial harm at the time they are used, but they may have significant downstream effects.

The Choosing Wisely campaign has brilliantly identified 5 important hospital medicine low‐value practices, and we hope to identify many more. We hope this series will serve as a grassroots effort to uncover more Choosing Wisely‐type practices. As institutions create their own high‐value care committees, the Choosing Wisely: Things We Do for No Reason series can provide possible agenda items, or provide the opportunity for sites to carry out analyses of their own practices to see whether any of the TWDFNR topics provide local opportunities for implementing higher‐value practices.

Although we do not believe that reducing the low‐value practices that will appear in TWDFNR will, alone, solve our wasteful practices, we hope that highlighting them will remind individuals, institutions, and systems that targeting low‐value practices is a responsibility that we all must embrace. We accept that not everyone will agree that the practices we present are low value, but the conversation is important to have. We invite you to take part in the Choosing Wisely: Things We Do for No Reason conversation. Let us know whether you think the practices highlighted are low value or whether you disagree with the conclusions. We welcome unsolicited proposals for series topics submitted as a 500‐word prcis. Send us your prcis or ideas on low‐value adult or pediatric patient practices that we should highlight in this series by emailing us at [email protected].

Disclosure: Nothing to report.

In this issue of the Journal of Hospital Medicine, we introduce a new recurring feature, Choosing Wisely: Things We Do for No Reason. The series is based on a talk I have delivered for the past 4 years at the annual national meeting of the Society of Hospital Medicine, in which I highlight 4 diagnostic tests, therapies, or other clinical practices that are commonly performed even though they are of low value to our inpatients.

There are many reasons hospitalists order unnecessary tests or treatments, or employ unhelpful clinical practices. Unnecessary testing may occur when we are not familiar with the test itselfthe actual costs of the test, the operating characteristics of the test, or the evidence supporting its usefulness in specific situations. Some tests are ordered unnecessarily because we cannot retrieve usable results from a different hospital or even our own electronic medical records. We may order tests or treatments due to patient expectations, a perceived need to practice defensively, or economic incentives.

Finally, we may simply order tests because of our uncertainty in the absence of data or simply because they are traditional practices (the way we've always done it). Physicians often order tests and treatments and institute clinical practices learned in residency or fellowship training.[1, 2] Local norms and practices influence physician behavior.

We created Things We Do for No Reason (TWDFNR) as a platform for provocative discussions of practices that have become common parts of hospital care but have limited supporting evidence, or even have evidence refuting or justifiably challenging their value. Each article in TWDFNR will describe why the test, treatment, or other clinical practice is commonly employed, why it may not be of high value, in what circumstances it may actually be valuable, and what conclusions can be drawn from the evidence provided. TWDFNR pieces are not systematic reviews or meta‐analyses and do not represent black and white conclusions or clinical practice standards; they are meant as a starting place for research and active discussions among hospitalists and patients.

In many respects, the Choosing Wisely: Things We Do for No Reason series is an extension of the Choosing Wisely campaign created by the American Board of Internal Medicine Foundation. Like Choosing Wisely, we are focusing on individual tests, treatments, and other clinical practices that are not beneficial and are potentially harmful to patients. Practices discussed may not cause significant physical or financial harm at the time they are used, but they may have significant downstream effects.

The Choosing Wisely campaign has brilliantly identified 5 important hospital medicine low‐value practices, and we hope to identify many more. We hope this series will serve as a grassroots effort to uncover more Choosing Wisely‐type practices. As institutions create their own high‐value care committees, the Choosing Wisely: Things We Do for No Reason series can provide possible agenda items, or provide the opportunity for sites to carry out analyses of their own practices to see whether any of the TWDFNR topics provide local opportunities for implementing higher‐value practices.

Although we do not believe that reducing the low‐value practices that will appear in TWDFNR will, alone, solve our wasteful practices, we hope that highlighting them will remind individuals, institutions, and systems that targeting low‐value practices is a responsibility that we all must embrace. We accept that not everyone will agree that the practices we present are low value, but the conversation is important to have. We invite you to take part in the Choosing Wisely: Things We Do for No Reason conversation. Let us know whether you think the practices highlighted are low value or whether you disagree with the conclusions. We welcome unsolicited proposals for series topics submitted as a 500‐word prcis. Send us your prcis or ideas on low‐value adult or pediatric patient practices that we should highlight in this series by emailing us at [email protected].

Disclosure: Nothing to report.

The problems surrounding hospital observation care and associated audits by Recovery Audit Contractors are gaining increased attention from both Congress and the Centers for Medicare & Medicaid Services (CMS).[1, 2, 3, 4, 5] On August 6, 2015, President Obama signed the NOTICE (Notice of Observation Treatment and Implication for Care Eligibility) Act (P.L. 114‐42),[4] which will require all Medicare beneficiaries receiving observation services for over 24 hours to be informed of their outpatient status. However, providers and hospitals are currently unable to answer a question that patients will certainly ask: What will an observation stay cost me compared to the same stay billed as an inpatient?

In this issue of the Journal of Hospital Medicine, Kangovi et al.[6] get a step closer to answering this question. Using 2010 to 2012 Medicare data, Kangovi and colleagues studied patient out‐of‐pocket costs per Medicare benefit period and found that the mean financial self‐pay cost per beneficiary observation stay was less ($469.42) than the 2010 inpatient deductible ($1100),[7] although about 1 in 10 observation stays exceeded the inpatient deductible. For beneficiaries with multiple observation stays per benefit period, the mean cumulative self‐pay cost ($947.40) was also less than the inpatient deductible. However, for over a quarter of beneficiaries with multiple observation stays, the cumulative patient cost exceeded the inpatient deductible. The authors also found that black beneficiaries and those with more comorbidities were more likely to have multiple observation visits per benefit period, but higher out‐of‐pocket observation liability was associated with nonblack race, lower number of chronic conditions, and not being dual eligible.

Medicare beneficiaries hospitalized as inpatients are covered by Medicare Part A, with a single deductible per benefit period, and are eligible for skilled nursing facility (SNF) coverage after 3 consecutive inpatient midnights. Medicare patients hospitalized as outpatients, including those receiving observation services, are not eligible for SNF coverage, must pay the cost of many self‐administered pharmaceuticals, and are generally responsible for 20% of each service rendered, but with the per‐service out‐of‐pocket 20% deductible capped at the equivalent to the current Part A deductible. However, there is no cumulative limit on the total out‐of‐pocket cost for outpatient observation (Part B) hospitalizations.[8]

Put in a slightly different way, while [outpatient coverage] is designed to reflect the cost of caring for each individual beneficiary, [inpatient payment structure] is designed to reflect the cost of caring for an average beneficiary.[9] Because outpatient observation payments are made per service, Medicare and beneficiary payment amounts both increase as the number of services provided increases,[9] which creates a threshold where the number and complexity of outpatient services exceeds the average inpatient stay, resulting in out‐of‐pocket observation costs exceeding the inpatient deductible. It makes sense that this threshold is more likely to be reached when the costs of multiple observation stays are added. Therefore, we should not be surprised at the findings of Kangovi et al.,[6] nor at those of the Office of Inspector General (OIG)[9] using 2012 Medicare claims data, showing higher average out‐of‐pocket patient costs for short inpatient stays compared to observation stays, but with a significant minority of out‐of‐pocket observation patient stay costs exceeding the inpatient deductible.

Dr. Kangovi and colleagues should be applauded for their efforts to address this important Medicare beneficiary issue. Yet many questions remain. First, neither the OIG study nor Kangovi et al. fully included cost of self‐administered medications in calculating patients' out‐of‐pocket patient liability. Second, Kangovi and colleagues did not account for beneficiary posthospitalization SNF costs, which would be substantially higher for any patients who did not have a qualifying 3‐day inpatient stay, including all patients hospitalized under observation. Third, both reports used data predating the 2‐midnight rule, so it is unlikely that beneficiary costs are comparable under current policy. Fewer long (>48 hours) observation stays under the 2‐midnight rule should reduce beneficiary financial burden, though this is unconfirmed. However, certain shorter, high‐acuity, procedure‐based observation stays could be more costly for patients.[9] Fourth, Kangovi et al. also did not consider patients with both an inpatient stay and an observation stay in the same benefit period; these patients would be liable for both the inpatient deductible and the outpatient fees. Fifth, to be meaningful, comparison of beneficiary out‐of‐pocket liability for inpatient versus outpatient care must occur in the context of services rendered, similar to what was proposed by the House Ways and Means Subcommittee in their Hospital Improvements for Payment draft bill.[10] Absent this, we should not conclude from this study that observation care is delivered at a discounted rate for patients when it is possible that lower out‐of‐pocket payments simply reflect, on average, fewer services rendered per observation stay when compared to an inpatient stay. Finally, the association between race, socioeconomic status, chronic conditions, and inpatient and observation stays merits further investigation. How such hospitalizations may relate to larger costs associated with lack of appropriate follow‐up care, including costs for those who have adverse consequences when they curtail or forego SNF placement, must be considered.

Even if we accept these limitations and accede that out‐of‐pocket observation cost is, on average, less than inpatient, pitfalls of observation policy remain: a cap on out‐of‐pocket financial risk for hospital care and SNF coverage are protections only afforded to those Medicare beneficiaries hospitalized as inpatients. Although the aspect of CMS' 2‐midnight rule that presumes inpatient status if a there is a physician's expectation of a medically necessary hospitalization of 2 or more midnights mitigates, but does not eliminate, the observation policy problem of uncapped out‐of‐pocket financial liability, it does not address the lack of SNF coverage following outpatient hospitalization. Further action and answers need to come from both Congress and CMS. At a recent Senate Special Committee on Aging hearing, Elizabeth Warren emphasized that CMS must accurately determine Medicare beneficiary out‐of‐pocket cost for observation care so providers can answer this question that patients undoubtedly ask.[1] CMS should be called upon to make available estimates of beneficiary costs under the 2‐midnight rule that include pharmacy charges, copayments (in the context of services rendered), and SNF costs. In addition, data should extend past beneficiary liability to detail differences in outpatient versus inpatient hospital reimbursement, systematic recovery auditing costs, and the total financial impact of maintaining 2 distinct (inpatient and outpatient) hospital reimbursement systems.[11, 12]

Congress and CMS must ultimately go beyond cost estimates and actually reform the core problems in outpatient observation policy and the Recovery Audit program charged with enforcing status determinations. Congress should pass the Improving Access to Medicare Coverage Act of 2015 (H.R. 1571 and S. 843), which would guarantee SNF coverage for Medicare beneficiaries hospitalized for 3 consecutive midnights, regardless of whether those nights are inpatient or outpatient.[13] Recovery Audit reform bills in the House (H.R. 2156)[3] and under consideration in the Senate[2] should be strongly supported. In addition, Congress and CMS should consider legislation or regulation that would cap outpatient hospitalization out‐of‐pocket liability at the inpatient Medicare beneficiary deductible. Alternatively, policymakers could finally recognize the current observation versus inpatient system for what it is: a payment structure with little clinical relevance. When the same exact medical care has 2 different hospital reimbursement rates and 2 different patient out‐of‐pocket financial liabilities, it may be time for policymakers to eliminate the false distinction altogether.

Disclosure: Nothing to report.

The problems surrounding hospital observation care and associated audits by Recovery Audit Contractors are gaining increased attention from both Congress and the Centers for Medicare & Medicaid Services (CMS).[1, 2, 3, 4, 5] On August 6, 2015, President Obama signed the NOTICE (Notice of Observation Treatment and Implication for Care Eligibility) Act (P.L. 114‐42),[4] which will require all Medicare beneficiaries receiving observation services for over 24 hours to be informed of their outpatient status. However, providers and hospitals are currently unable to answer a question that patients will certainly ask: What will an observation stay cost me compared to the same stay billed as an inpatient?

In this issue of the Journal of Hospital Medicine, Kangovi et al.[6] get a step closer to answering this question. Using 2010 to 2012 Medicare data, Kangovi and colleagues studied patient out‐of‐pocket costs per Medicare benefit period and found that the mean financial self‐pay cost per beneficiary observation stay was less ($469.42) than the 2010 inpatient deductible ($1100),[7] although about 1 in 10 observation stays exceeded the inpatient deductible. For beneficiaries with multiple observation stays per benefit period, the mean cumulative self‐pay cost ($947.40) was also less than the inpatient deductible. However, for over a quarter of beneficiaries with multiple observation stays, the cumulative patient cost exceeded the inpatient deductible. The authors also found that black beneficiaries and those with more comorbidities were more likely to have multiple observation visits per benefit period, but higher out‐of‐pocket observation liability was associated with nonblack race, lower number of chronic conditions, and not being dual eligible.

Medicare beneficiaries hospitalized as inpatients are covered by Medicare Part A, with a single deductible per benefit period, and are eligible for skilled nursing facility (SNF) coverage after 3 consecutive inpatient midnights. Medicare patients hospitalized as outpatients, including those receiving observation services, are not eligible for SNF coverage, must pay the cost of many self‐administered pharmaceuticals, and are generally responsible for 20% of each service rendered, but with the per‐service out‐of‐pocket 20% deductible capped at the equivalent to the current Part A deductible. However, there is no cumulative limit on the total out‐of‐pocket cost for outpatient observation (Part B) hospitalizations.[8]

Put in a slightly different way, while [outpatient coverage] is designed to reflect the cost of caring for each individual beneficiary, [inpatient payment structure] is designed to reflect the cost of caring for an average beneficiary.[9] Because outpatient observation payments are made per service, Medicare and beneficiary payment amounts both increase as the number of services provided increases,[9] which creates a threshold where the number and complexity of outpatient services exceeds the average inpatient stay, resulting in out‐of‐pocket observation costs exceeding the inpatient deductible. It makes sense that this threshold is more likely to be reached when the costs of multiple observation stays are added. Therefore, we should not be surprised at the findings of Kangovi et al.,[6] nor at those of the Office of Inspector General (OIG)[9] using 2012 Medicare claims data, showing higher average out‐of‐pocket patient costs for short inpatient stays compared to observation stays, but with a significant minority of out‐of‐pocket observation patient stay costs exceeding the inpatient deductible.

Dr. Kangovi and colleagues should be applauded for their efforts to address this important Medicare beneficiary issue. Yet many questions remain. First, neither the OIG study nor Kangovi et al. fully included cost of self‐administered medications in calculating patients' out‐of‐pocket patient liability. Second, Kangovi and colleagues did not account for beneficiary posthospitalization SNF costs, which would be substantially higher for any patients who did not have a qualifying 3‐day inpatient stay, including all patients hospitalized under observation. Third, both reports used data predating the 2‐midnight rule, so it is unlikely that beneficiary costs are comparable under current policy. Fewer long (>48 hours) observation stays under the 2‐midnight rule should reduce beneficiary financial burden, though this is unconfirmed. However, certain shorter, high‐acuity, procedure‐based observation stays could be more costly for patients.[9] Fourth, Kangovi et al. also did not consider patients with both an inpatient stay and an observation stay in the same benefit period; these patients would be liable for both the inpatient deductible and the outpatient fees. Fifth, to be meaningful, comparison of beneficiary out‐of‐pocket liability for inpatient versus outpatient care must occur in the context of services rendered, similar to what was proposed by the House Ways and Means Subcommittee in their Hospital Improvements for Payment draft bill.[10] Absent this, we should not conclude from this study that observation care is delivered at a discounted rate for patients when it is possible that lower out‐of‐pocket payments simply reflect, on average, fewer services rendered per observation stay when compared to an inpatient stay. Finally, the association between race, socioeconomic status, chronic conditions, and inpatient and observation stays merits further investigation. How such hospitalizations may relate to larger costs associated with lack of appropriate follow‐up care, including costs for those who have adverse consequences when they curtail or forego SNF placement, must be considered.

Even if we accept these limitations and accede that out‐of‐pocket observation cost is, on average, less than inpatient, pitfalls of observation policy remain: a cap on out‐of‐pocket financial risk for hospital care and SNF coverage are protections only afforded to those Medicare beneficiaries hospitalized as inpatients. Although the aspect of CMS' 2‐midnight rule that presumes inpatient status if a there is a physician's expectation of a medically necessary hospitalization of 2 or more midnights mitigates, but does not eliminate, the observation policy problem of uncapped out‐of‐pocket financial liability, it does not address the lack of SNF coverage following outpatient hospitalization. Further action and answers need to come from both Congress and CMS. At a recent Senate Special Committee on Aging hearing, Elizabeth Warren emphasized that CMS must accurately determine Medicare beneficiary out‐of‐pocket cost for observation care so providers can answer this question that patients undoubtedly ask.[1] CMS should be called upon to make available estimates of beneficiary costs under the 2‐midnight rule that include pharmacy charges, copayments (in the context of services rendered), and SNF costs. In addition, data should extend past beneficiary liability to detail differences in outpatient versus inpatient hospital reimbursement, systematic recovery auditing costs, and the total financial impact of maintaining 2 distinct (inpatient and outpatient) hospital reimbursement systems.[11, 12]

Congress and CMS must ultimately go beyond cost estimates and actually reform the core problems in outpatient observation policy and the Recovery Audit program charged with enforcing status determinations. Congress should pass the Improving Access to Medicare Coverage Act of 2015 (H.R. 1571 and S. 843), which would guarantee SNF coverage for Medicare beneficiaries hospitalized for 3 consecutive midnights, regardless of whether those nights are inpatient or outpatient.[13] Recovery Audit reform bills in the House (H.R. 2156)[3] and under consideration in the Senate[2] should be strongly supported. In addition, Congress and CMS should consider legislation or regulation that would cap outpatient hospitalization out‐of‐pocket liability at the inpatient Medicare beneficiary deductible. Alternatively, policymakers could finally recognize the current observation versus inpatient system for what it is: a payment structure with little clinical relevance. When the same exact medical care has 2 different hospital reimbursement rates and 2 different patient out‐of‐pocket financial liabilities, it may be time for policymakers to eliminate the false distinction altogether.

Disclosure: Nothing to report.

The problems surrounding hospital observation care and associated audits by Recovery Audit Contractors are gaining increased attention from both Congress and the Centers for Medicare & Medicaid Services (CMS).[1, 2, 3, 4, 5] On August 6, 2015, President Obama signed the NOTICE (Notice of Observation Treatment and Implication for Care Eligibility) Act (P.L. 114‐42),[4] which will require all Medicare beneficiaries receiving observation services for over 24 hours to be informed of their outpatient status. However, providers and hospitals are currently unable to answer a question that patients will certainly ask: What will an observation stay cost me compared to the same stay billed as an inpatient?

In this issue of the Journal of Hospital Medicine, Kangovi et al.[6] get a step closer to answering this question. Using 2010 to 2012 Medicare data, Kangovi and colleagues studied patient out‐of‐pocket costs per Medicare benefit period and found that the mean financial self‐pay cost per beneficiary observation stay was less ($469.42) than the 2010 inpatient deductible ($1100),[7] although about 1 in 10 observation stays exceeded the inpatient deductible. For beneficiaries with multiple observation stays per benefit period, the mean cumulative self‐pay cost ($947.40) was also less than the inpatient deductible. However, for over a quarter of beneficiaries with multiple observation stays, the cumulative patient cost exceeded the inpatient deductible. The authors also found that black beneficiaries and those with more comorbidities were more likely to have multiple observation visits per benefit period, but higher out‐of‐pocket observation liability was associated with nonblack race, lower number of chronic conditions, and not being dual eligible.

Medicare beneficiaries hospitalized as inpatients are covered by Medicare Part A, with a single deductible per benefit period, and are eligible for skilled nursing facility (SNF) coverage after 3 consecutive inpatient midnights. Medicare patients hospitalized as outpatients, including those receiving observation services, are not eligible for SNF coverage, must pay the cost of many self‐administered pharmaceuticals, and are generally responsible for 20% of each service rendered, but with the per‐service out‐of‐pocket 20% deductible capped at the equivalent to the current Part A deductible. However, there is no cumulative limit on the total out‐of‐pocket cost for outpatient observation (Part B) hospitalizations.[8]

Put in a slightly different way, while [outpatient coverage] is designed to reflect the cost of caring for each individual beneficiary, [inpatient payment structure] is designed to reflect the cost of caring for an average beneficiary.[9] Because outpatient observation payments are made per service, Medicare and beneficiary payment amounts both increase as the number of services provided increases,[9] which creates a threshold where the number and complexity of outpatient services exceeds the average inpatient stay, resulting in out‐of‐pocket observation costs exceeding the inpatient deductible. It makes sense that this threshold is more likely to be reached when the costs of multiple observation stays are added. Therefore, we should not be surprised at the findings of Kangovi et al.,[6] nor at those of the Office of Inspector General (OIG)[9] using 2012 Medicare claims data, showing higher average out‐of‐pocket patient costs for short inpatient stays compared to observation stays, but with a significant minority of out‐of‐pocket observation patient stay costs exceeding the inpatient deductible.

Dr. Kangovi and colleagues should be applauded for their efforts to address this important Medicare beneficiary issue. Yet many questions remain. First, neither the OIG study nor Kangovi et al. fully included cost of self‐administered medications in calculating patients' out‐of‐pocket patient liability. Second, Kangovi and colleagues did not account for beneficiary posthospitalization SNF costs, which would be substantially higher for any patients who did not have a qualifying 3‐day inpatient stay, including all patients hospitalized under observation. Third, both reports used data predating the 2‐midnight rule, so it is unlikely that beneficiary costs are comparable under current policy. Fewer long (>48 hours) observation stays under the 2‐midnight rule should reduce beneficiary financial burden, though this is unconfirmed. However, certain shorter, high‐acuity, procedure‐based observation stays could be more costly for patients.[9] Fourth, Kangovi et al. also did not consider patients with both an inpatient stay and an observation stay in the same benefit period; these patients would be liable for both the inpatient deductible and the outpatient fees. Fifth, to be meaningful, comparison of beneficiary out‐of‐pocket liability for inpatient versus outpatient care must occur in the context of services rendered, similar to what was proposed by the House Ways and Means Subcommittee in their Hospital Improvements for Payment draft bill.[10] Absent this, we should not conclude from this study that observation care is delivered at a discounted rate for patients when it is possible that lower out‐of‐pocket payments simply reflect, on average, fewer services rendered per observation stay when compared to an inpatient stay. Finally, the association between race, socioeconomic status, chronic conditions, and inpatient and observation stays merits further investigation. How such hospitalizations may relate to larger costs associated with lack of appropriate follow‐up care, including costs for those who have adverse consequences when they curtail or forego SNF placement, must be considered.

Even if we accept these limitations and accede that out‐of‐pocket observation cost is, on average, less than inpatient, pitfalls of observation policy remain: a cap on out‐of‐pocket financial risk for hospital care and SNF coverage are protections only afforded to those Medicare beneficiaries hospitalized as inpatients. Although the aspect of CMS' 2‐midnight rule that presumes inpatient status if a there is a physician's expectation of a medically necessary hospitalization of 2 or more midnights mitigates, but does not eliminate, the observation policy problem of uncapped out‐of‐pocket financial liability, it does not address the lack of SNF coverage following outpatient hospitalization. Further action and answers need to come from both Congress and CMS. At a recent Senate Special Committee on Aging hearing, Elizabeth Warren emphasized that CMS must accurately determine Medicare beneficiary out‐of‐pocket cost for observation care so providers can answer this question that patients undoubtedly ask.[1] CMS should be called upon to make available estimates of beneficiary costs under the 2‐midnight rule that include pharmacy charges, copayments (in the context of services rendered), and SNF costs. In addition, data should extend past beneficiary liability to detail differences in outpatient versus inpatient hospital reimbursement, systematic recovery auditing costs, and the total financial impact of maintaining 2 distinct (inpatient and outpatient) hospital reimbursement systems.[11, 12]

Congress and CMS must ultimately go beyond cost estimates and actually reform the core problems in outpatient observation policy and the Recovery Audit program charged with enforcing status determinations. Congress should pass the Improving Access to Medicare Coverage Act of 2015 (H.R. 1571 and S. 843), which would guarantee SNF coverage for Medicare beneficiaries hospitalized for 3 consecutive midnights, regardless of whether those nights are inpatient or outpatient.[13] Recovery Audit reform bills in the House (H.R. 2156)[3] and under consideration in the Senate[2] should be strongly supported. In addition, Congress and CMS should consider legislation or regulation that would cap outpatient hospitalization out‐of‐pocket liability at the inpatient Medicare beneficiary deductible. Alternatively, policymakers could finally recognize the current observation versus inpatient system for what it is: a payment structure with little clinical relevance. When the same exact medical care has 2 different hospital reimbursement rates and 2 different patient out‐of‐pocket financial liabilities, it may be time for policymakers to eliminate the false distinction altogether.

Disclosure: Nothing to report.

Patient satisfaction has received increased attention in recent years, which we believe is well deserved and long overdue. Anyone who has been hospitalized, or has had a loved one hospitalized, can appreciate that there is room to improve the patient experience. Dedicating time and effort to improving the patient experience is consistent with our professional commitment to comfort, empathize, and partner with our patients. Though patient satisfaction itself is an outcome worthy of our attention, it is also positively associated with measures related to patient safety and clinical effectiveness.[1, 2] Moreover, patient satisfaction is the only publicly reported measure that represents the patient's voice,[3] and accounts for a substantial portion of the Centers for Medicare and Medicaid Services payment adjustments under the Hospital Value Based Purchasing Program.[4]

However, all healthcare professionals should understand some key fundamental issues related to the measurement of patient satisfaction. The survey from which data are publicly reported and used for hospital payment adjustment is the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, developed by the Agency for Healthcare Research and Quality.[5, 6] HCAHPS is sent to a random sample of 40% of hospitalized patients between 48 hours and 6 weeks after discharge. The HCAHPS survey uses ordinal response scales (eg, never, sometimes, usually, always) that generate highly skewed results toward favorable responses. Therefore, results are reported as the percent top box (ie, the percentage of responses in the most favorable category) rather than as a median score. The skewed distribution of results indicates that most patients are generally satisfied with care (ie, most respondents do not have an axe to grind), but also makes meaningful improvement difficult to achieve. Prior to public reporting and determination of effect on hospital payment, results are adjusted for mode of survey administration and patient mix. The same is not true when patient satisfaction data are used for internal purposes. Hospital leaders typically do not perform statistical adjustment and therefore need to be careful not to make apples‐to‐orangestype comparisons. For example, obstetric patient satisfaction scores should not be compared to general medical patient satisfaction scores, as these populations tend to rate satisfaction differently.

The HCAHPS survey questions are organized into domains of care, including satisfaction with nurses and satisfaction with doctors. Importantly, other healthcare team members may influence patients' perception in these domains. For example, a patient responding to nurse communication questions may also reflect on experiences with patient care technicians, social workers, and therapists. A patient responding to physician communication questions might also reflect on experiences with advanced practice providers. A common mistake is the practice of attributing satisfaction with doctors to the individual who served as the discharge physician. Many readers have likely seen patient satisfaction reports broken out by discharge physician with the expectation that giving this information to individual physicians will serve as useful formative feedback. The reality is that patients see many doctors during a hospitalization. To illustrate this point, we analyzed data from 420 patients admitted to our nonteaching hospitalist service who had completed an HCAHPS survey in 2014. We found that the discharge hospitalist accounted for only 34% of all physician encounters. Furthermore, research has shown that patients' experiences with specialist physicians also have a strong influence on their overall satisfaction with physicians.[7]

Having reliable patient satisfaction data on specific individuals would be a truly powerful formative assessment tool. In this issue of the Journal of Hospital Medicine, Banka and colleagues report on an impressive approach incorporating such a tool to give constructive feedback to physicians.[8] Since 2006, the study site had administered surveys to hospitalized patients that assess their satisfaction with specific resident physicians.[9] However, residency programs only reviewed the survey results with resident physicians about twice a year. The multifaceted intervention developed by Banka and colleagues included directly emailing the survey results to internal medicine resident physicians in real time while they were in service, a 1‐hour conference on best communication practices, and a reward program in which 3 residents were identified monthly to receive department‐wide recognition via email and a generous movie package. Using difference‐in‐differences regression analysis, the investigators compared changes in patient satisfaction results for internal medicine residents to results for residents from other specialties (who were not part of the intervention). The percentage of patients who gave top box responses to all 3 physician‐related questions and to the overall hospital rating was significantly higher for the internal medicine residents.

The findings from this study are important, because no prior study of an intervention, to our knowledge, has shown a significant improvement in patient satisfaction scores. In this study, feedback was believed to be the most powerful factor. The importance of meaningful, timely feedback in medical education is well recognized.[10] Without feedback there is poor insight into how intended results from specific actions compare with actual results. When feedback is lacking from external sources (in this case the voice of the patient), an uncontested sense of mastery develops, allowing mistakes to go uncorrected. This false sense of mastery contributes to an emotional and defensive response when performance is finally revealed to be less than optimal. The simple act of giving more timely feedback in this study encouraged self‐motivated reflection and practice change aimed at improving patient satisfaction, with remarkable results.

The study should inspire physician leaders from various hospital settings, and researchers, to develop and evaluate similar programs to improve patient satisfaction. We agree with the investigators that the approach should be multifaceted. Feedback to specific physicians is a powerful motivator, but needs to be combined with strategies to enhance communication skills. Brief conferences are less likely to have a lasting impact on behaviors than strategies like coaching and simulation based training.[11] Interventions should include recognition and reward to acknowledge exceptional performance and build friendly competition.

The biggest challenge to adopting an intervention such as the one used in the Banka study relates to the feasibility of implementing physician‐specific patient satisfaction reporting. Several survey instruments are available for use as tools to assess satisfaction with specific physicians.[9, 12, 13] However, who will administer these instruments? Most hospitals do not have undergraduate students available. Hospitals could use their volunteers, but this is not likely to be a sustainable solution. Hospitals could consider administering the survey via email, but many hospitals are just starting to collect patient email addresses and many patients do not use email. Once data are collected, who will conduct analyses and create comparative reports? Press Ganey recently developed a survey assessing satisfaction with specific hospitalists, using photographs, and offers the ability to create comparative reports.[14] Their service addresses the analytic challenge, but the quandary of survey administration remains.

In conclusion, we encourage hospital medicine leaders to develop and evaluate multifaceted interventions to improve patient satisfaction such as the one reported by Banka et al. Timely, specific feedback to physicians is an essential feature. The collection of physician‐specific data is a major challenge, but not an insurmountable one. Novel use of personnel and/or technology is likely to play a role in these efforts.

Disclosure: Nothing to report.

Patient satisfaction has received increased attention in recent years, which we believe is well deserved and long overdue. Anyone who has been hospitalized, or has had a loved one hospitalized, can appreciate that there is room to improve the patient experience. Dedicating time and effort to improving the patient experience is consistent with our professional commitment to comfort, empathize, and partner with our patients. Though patient satisfaction itself is an outcome worthy of our attention, it is also positively associated with measures related to patient safety and clinical effectiveness.[1, 2] Moreover, patient satisfaction is the only publicly reported measure that represents the patient's voice,[3] and accounts for a substantial portion of the Centers for Medicare and Medicaid Services payment adjustments under the Hospital Value Based Purchasing Program.[4]

However, all healthcare professionals should understand some key fundamental issues related to the measurement of patient satisfaction. The survey from which data are publicly reported and used for hospital payment adjustment is the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, developed by the Agency for Healthcare Research and Quality.[5, 6] HCAHPS is sent to a random sample of 40% of hospitalized patients between 48 hours and 6 weeks after discharge. The HCAHPS survey uses ordinal response scales (eg, never, sometimes, usually, always) that generate highly skewed results toward favorable responses. Therefore, results are reported as the percent top box (ie, the percentage of responses in the most favorable category) rather than as a median score. The skewed distribution of results indicates that most patients are generally satisfied with care (ie, most respondents do not have an axe to grind), but also makes meaningful improvement difficult to achieve. Prior to public reporting and determination of effect on hospital payment, results are adjusted for mode of survey administration and patient mix. The same is not true when patient satisfaction data are used for internal purposes. Hospital leaders typically do not perform statistical adjustment and therefore need to be careful not to make apples‐to‐orangestype comparisons. For example, obstetric patient satisfaction scores should not be compared to general medical patient satisfaction scores, as these populations tend to rate satisfaction differently.

The HCAHPS survey questions are organized into domains of care, including satisfaction with nurses and satisfaction with doctors. Importantly, other healthcare team members may influence patients' perception in these domains. For example, a patient responding to nurse communication questions may also reflect on experiences with patient care technicians, social workers, and therapists. A patient responding to physician communication questions might also reflect on experiences with advanced practice providers. A common mistake is the practice of attributing satisfaction with doctors to the individual who served as the discharge physician. Many readers have likely seen patient satisfaction reports broken out by discharge physician with the expectation that giving this information to individual physicians will serve as useful formative feedback. The reality is that patients see many doctors during a hospitalization. To illustrate this point, we analyzed data from 420 patients admitted to our nonteaching hospitalist service who had completed an HCAHPS survey in 2014. We found that the discharge hospitalist accounted for only 34% of all physician encounters. Furthermore, research has shown that patients' experiences with specialist physicians also have a strong influence on their overall satisfaction with physicians.[7]

Having reliable patient satisfaction data on specific individuals would be a truly powerful formative assessment tool. In this issue of the Journal of Hospital Medicine, Banka and colleagues report on an impressive approach incorporating such a tool to give constructive feedback to physicians.[8] Since 2006, the study site had administered surveys to hospitalized patients that assess their satisfaction with specific resident physicians.[9] However, residency programs only reviewed the survey results with resident physicians about twice a year. The multifaceted intervention developed by Banka and colleagues included directly emailing the survey results to internal medicine resident physicians in real time while they were in service, a 1‐hour conference on best communication practices, and a reward program in which 3 residents were identified monthly to receive department‐wide recognition via email and a generous movie package. Using difference‐in‐differences regression analysis, the investigators compared changes in patient satisfaction results for internal medicine residents to results for residents from other specialties (who were not part of the intervention). The percentage of patients who gave top box responses to all 3 physician‐related questions and to the overall hospital rating was significantly higher for the internal medicine residents.

The findings from this study are important, because no prior study of an intervention, to our knowledge, has shown a significant improvement in patient satisfaction scores. In this study, feedback was believed to be the most powerful factor. The importance of meaningful, timely feedback in medical education is well recognized.[10] Without feedback there is poor insight into how intended results from specific actions compare with actual results. When feedback is lacking from external sources (in this case the voice of the patient), an uncontested sense of mastery develops, allowing mistakes to go uncorrected. This false sense of mastery contributes to an emotional and defensive response when performance is finally revealed to be less than optimal. The simple act of giving more timely feedback in this study encouraged self‐motivated reflection and practice change aimed at improving patient satisfaction, with remarkable results.

The study should inspire physician leaders from various hospital settings, and researchers, to develop and evaluate similar programs to improve patient satisfaction. We agree with the investigators that the approach should be multifaceted. Feedback to specific physicians is a powerful motivator, but needs to be combined with strategies to enhance communication skills. Brief conferences are less likely to have a lasting impact on behaviors than strategies like coaching and simulation based training.[11] Interventions should include recognition and reward to acknowledge exceptional performance and build friendly competition.

The biggest challenge to adopting an intervention such as the one used in the Banka study relates to the feasibility of implementing physician‐specific patient satisfaction reporting. Several survey instruments are available for use as tools to assess satisfaction with specific physicians.[9, 12, 13] However, who will administer these instruments? Most hospitals do not have undergraduate students available. Hospitals could use their volunteers, but this is not likely to be a sustainable solution. Hospitals could consider administering the survey via email, but many hospitals are just starting to collect patient email addresses and many patients do not use email. Once data are collected, who will conduct analyses and create comparative reports? Press Ganey recently developed a survey assessing satisfaction with specific hospitalists, using photographs, and offers the ability to create comparative reports.[14] Their service addresses the analytic challenge, but the quandary of survey administration remains.

In conclusion, we encourage hospital medicine leaders to develop and evaluate multifaceted interventions to improve patient satisfaction such as the one reported by Banka et al. Timely, specific feedback to physicians is an essential feature. The collection of physician‐specific data is a major challenge, but not an insurmountable one. Novel use of personnel and/or technology is likely to play a role in these efforts.

Disclosure: Nothing to report.

Patient satisfaction has received increased attention in recent years, which we believe is well deserved and long overdue. Anyone who has been hospitalized, or has had a loved one hospitalized, can appreciate that there is room to improve the patient experience. Dedicating time and effort to improving the patient experience is consistent with our professional commitment to comfort, empathize, and partner with our patients. Though patient satisfaction itself is an outcome worthy of our attention, it is also positively associated with measures related to patient safety and clinical effectiveness.[1, 2] Moreover, patient satisfaction is the only publicly reported measure that represents the patient's voice,[3] and accounts for a substantial portion of the Centers for Medicare and Medicaid Services payment adjustments under the Hospital Value Based Purchasing Program.[4]

However, all healthcare professionals should understand some key fundamental issues related to the measurement of patient satisfaction. The survey from which data are publicly reported and used for hospital payment adjustment is the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, developed by the Agency for Healthcare Research and Quality.[5, 6] HCAHPS is sent to a random sample of 40% of hospitalized patients between 48 hours and 6 weeks after discharge. The HCAHPS survey uses ordinal response scales (eg, never, sometimes, usually, always) that generate highly skewed results toward favorable responses. Therefore, results are reported as the percent top box (ie, the percentage of responses in the most favorable category) rather than as a median score. The skewed distribution of results indicates that most patients are generally satisfied with care (ie, most respondents do not have an axe to grind), but also makes meaningful improvement difficult to achieve. Prior to public reporting and determination of effect on hospital payment, results are adjusted for mode of survey administration and patient mix. The same is not true when patient satisfaction data are used for internal purposes. Hospital leaders typically do not perform statistical adjustment and therefore need to be careful not to make apples‐to‐orangestype comparisons. For example, obstetric patient satisfaction scores should not be compared to general medical patient satisfaction scores, as these populations tend to rate satisfaction differently.

The HCAHPS survey questions are organized into domains of care, including satisfaction with nurses and satisfaction with doctors. Importantly, other healthcare team members may influence patients' perception in these domains. For example, a patient responding to nurse communication questions may also reflect on experiences with patient care technicians, social workers, and therapists. A patient responding to physician communication questions might also reflect on experiences with advanced practice providers. A common mistake is the practice of attributing satisfaction with doctors to the individual who served as the discharge physician. Many readers have likely seen patient satisfaction reports broken out by discharge physician with the expectation that giving this information to individual physicians will serve as useful formative feedback. The reality is that patients see many doctors during a hospitalization. To illustrate this point, we analyzed data from 420 patients admitted to our nonteaching hospitalist service who had completed an HCAHPS survey in 2014. We found that the discharge hospitalist accounted for only 34% of all physician encounters. Furthermore, research has shown that patients' experiences with specialist physicians also have a strong influence on their overall satisfaction with physicians.[7]

Having reliable patient satisfaction data on specific individuals would be a truly powerful formative assessment tool. In this issue of the Journal of Hospital Medicine, Banka and colleagues report on an impressive approach incorporating such a tool to give constructive feedback to physicians.[8] Since 2006, the study site had administered surveys to hospitalized patients that assess their satisfaction with specific resident physicians.[9] However, residency programs only reviewed the survey results with resident physicians about twice a year. The multifaceted intervention developed by Banka and colleagues included directly emailing the survey results to internal medicine resident physicians in real time while they were in service, a 1‐hour conference on best communication practices, and a reward program in which 3 residents were identified monthly to receive department‐wide recognition via email and a generous movie package. Using difference‐in‐differences regression analysis, the investigators compared changes in patient satisfaction results for internal medicine residents to results for residents from other specialties (who were not part of the intervention). The percentage of patients who gave top box responses to all 3 physician‐related questions and to the overall hospital rating was significantly higher for the internal medicine residents.

The findings from this study are important, because no prior study of an intervention, to our knowledge, has shown a significant improvement in patient satisfaction scores. In this study, feedback was believed to be the most powerful factor. The importance of meaningful, timely feedback in medical education is well recognized.[10] Without feedback there is poor insight into how intended results from specific actions compare with actual results. When feedback is lacking from external sources (in this case the voice of the patient), an uncontested sense of mastery develops, allowing mistakes to go uncorrected. This false sense of mastery contributes to an emotional and defensive response when performance is finally revealed to be less than optimal. The simple act of giving more timely feedback in this study encouraged self‐motivated reflection and practice change aimed at improving patient satisfaction, with remarkable results.

The study should inspire physician leaders from various hospital settings, and researchers, to develop and evaluate similar programs to improve patient satisfaction. We agree with the investigators that the approach should be multifaceted. Feedback to specific physicians is a powerful motivator, but needs to be combined with strategies to enhance communication skills. Brief conferences are less likely to have a lasting impact on behaviors than strategies like coaching and simulation based training.[11] Interventions should include recognition and reward to acknowledge exceptional performance and build friendly competition.

The biggest challenge to adopting an intervention such as the one used in the Banka study relates to the feasibility of implementing physician‐specific patient satisfaction reporting. Several survey instruments are available for use as tools to assess satisfaction with specific physicians.[9, 12, 13] However, who will administer these instruments? Most hospitals do not have undergraduate students available. Hospitals could use their volunteers, but this is not likely to be a sustainable solution. Hospitals could consider administering the survey via email, but many hospitals are just starting to collect patient email addresses and many patients do not use email. Once data are collected, who will conduct analyses and create comparative reports? Press Ganey recently developed a survey assessing satisfaction with specific hospitalists, using photographs, and offers the ability to create comparative reports.[14] Their service addresses the analytic challenge, but the quandary of survey administration remains.

In conclusion, we encourage hospital medicine leaders to develop and evaluate multifaceted interventions to improve patient satisfaction such as the one reported by Banka et al. Timely, specific feedback to physicians is an essential feature. The collection of physician‐specific data is a major challenge, but not an insurmountable one. Novel use of personnel and/or technology is likely to play a role in these efforts.

Disclosure: Nothing to report.

From a new concept to 44,000 practitioners in just 18 years,[1] there is no doubt that the word hospitalist is synonymous with innovation, leadership, growth, and change. Yet 2 articles in this month's Journal of Hospital Medicine prove that even our new field faces age‐old problems. Although women comprise half of all academic hospitalist and general internal medicine faculty, Burden et al.[2] showed that female hospitalists are less likely than male hospitalists to be division or section heads of hospital medicine, speakers at national meetings, and first or last authors on both research publications and editorials. This is made more concerning given that women are more likely to choose academic hospital medicine careers,[3] as they represent one‐third of all hospitalists but half of the academic hospitalist workforce.[2, 3] Findings in general internal medicine were similar, except that equal numbers of women and men were national meeting speakers and first authors on research publications (but not editorials). Weaver et al.[4] shed even more light on this disparity, and found that female hospitalists made $14,581 less per year than their male counterparts, even after adjusting for relevant differences. Weaver and colleagues also found other gender‐specific differences: women worked more nights and had fewer billable encounters per hospitalist shift than men.

Unfortunately, these trends are not new or limited to hospital medicine. For decades, almost equal numbers of women and men have entered medical school,[5] yet women are under‐represented in high status specialties,[6] less likely to be first or senior authors on original research studies compared to men,[7] less likely to be promoted,[8] and women physicians are consistently paid less than men across specialties.[9, 10] Simple analyses have not yet explained these disparities. Compared with men, women have similar leadership aspirations[11, 12] and are at least as effective as leaders.[13, 14, 15] Yet equity has not been attained.

Implicit bias research suggests that gender stereotypes influence women at all career stages.[16, 17, 18] For example, an elegant study conducted by Correll et al. identified a motherhood penalty, where indicating membership in the elementary school parent‐teacher organization on one's curriculum vitae hurt women's chances of employment and pay, but actually helped men.[19] Gender stereotypes exist, even among those who do not support their content. The universal reinforcement of such stereotypes over time leads to implicit but prescriptive rules about how women and men should act.[20] In particular, communal behaviors, including being cooperative, kind, and understanding, are typically associated with women, and agentic behaviors, including being ambitious and acting as a leader, are considered appropriate for men.[21] This leads to the think leader, think male phenomena, where we automatically associate men with leadership and higher status tasks (like first authorship or speaker invitations).[22, 23] Furthermore, acting against the stereotype (eg, a woman showing anger[24] or negotiating for more pay[25] or a man showing sadness[26]) can negatively impact wage and employment. Expecting social censure for violating gender norms, women develop a fear of the backlash that alone may shape behavior such that women may not express interest in having a high salary or negotiate for a raise.[27, 28, 29]

The specific system and institutional barriers that prevent female hospitalists from receiving equal pay and opportunities for leadership are not known, but one can surmise they are similar to those found in other specialties.[10, 30, 31] The findings of the studies of Burden et al.[2] and Weaver et al.[4] invite investigation of new questions specific to hospital medicine. Why are women in hospital medicine working more night shifts? Does this impact leadership or scholarship opportunities? Why are women documenting less productivity? Are they spending more time with patients, as they do in other settings?[32] What influences their practice choice? We would like to believe that there is something about hospital medicine that can explain the gender differences identified in these 2 studies. However, these data should prompt a serious and thorough examination of our specialty. We must accept that despite being a new specialty and a change leader, hospital medicine may not have escaped systematic gender bias that constrains the full participation and advancement of women.

But we believe that hospitalistsinnovators and change leaders in medicinewill be spurred to action to address the possibility of gender inequities. We do not need to know all of the causes to begin to address disparities, of every type, on an individual, institutional, and national level. As individuals, we can acknowledge that there are implicit assumptions that influence our decision making. No matter how unintentional, and even conflicting with evidence, these assumptions can lead us to judge women as less capable leaders than men or to automatically envision a high salary for a woman and man as different amounts. However, these automatic gender biases function as habits of mind, so they can be broken like any other unwanted habit.[33] Institutionally, we can also hold ourselves accountable for transparency in mentorship, leadership, scholarship, promotions, and wages to ensure diverse representation. We should routinely examine our practices to ensure the equitable hiring, pay, and promotion of our workforce.[18] National organizations and their respective journals should actively pursue diverse representation in leadership and membership on boards and committees, award nominees and recipients, and opportunities for invited editorials. Hospital medicinebeing young, innovative, and committed to changeis uniquely well suited to lead the charge for workforce equity. We can, and will, show the rest of medicine how it is done.

From a new concept to 44,000 practitioners in just 18 years,[1] there is no doubt that the word hospitalist is synonymous with innovation, leadership, growth, and change. Yet 2 articles in this month's Journal of Hospital Medicine prove that even our new field faces age‐old problems. Although women comprise half of all academic hospitalist and general internal medicine faculty, Burden et al.[2] showed that female hospitalists are less likely than male hospitalists to be division or section heads of hospital medicine, speakers at national meetings, and first or last authors on both research publications and editorials. This is made more concerning given that women are more likely to choose academic hospital medicine careers,[3] as they represent one‐third of all hospitalists but half of the academic hospitalist workforce.[2, 3] Findings in general internal medicine were similar, except that equal numbers of women and men were national meeting speakers and first authors on research publications (but not editorials). Weaver et al.[4] shed even more light on this disparity, and found that female hospitalists made $14,581 less per year than their male counterparts, even after adjusting for relevant differences. Weaver and colleagues also found other gender‐specific differences: women worked more nights and had fewer billable encounters per hospitalist shift than men.

Unfortunately, these trends are not new or limited to hospital medicine. For decades, almost equal numbers of women and men have entered medical school,[5] yet women are under‐represented in high status specialties,[6] less likely to be first or senior authors on original research studies compared to men,[7] less likely to be promoted,[8] and women physicians are consistently paid less than men across specialties.[9, 10] Simple analyses have not yet explained these disparities. Compared with men, women have similar leadership aspirations[11, 12] and are at least as effective as leaders.[13, 14, 15] Yet equity has not been attained.

Implicit bias research suggests that gender stereotypes influence women at all career stages.[16, 17, 18] For example, an elegant study conducted by Correll et al. identified a motherhood penalty, where indicating membership in the elementary school parent‐teacher organization on one's curriculum vitae hurt women's chances of employment and pay, but actually helped men.[19] Gender stereotypes exist, even among those who do not support their content. The universal reinforcement of such stereotypes over time leads to implicit but prescriptive rules about how women and men should act.[20] In particular, communal behaviors, including being cooperative, kind, and understanding, are typically associated with women, and agentic behaviors, including being ambitious and acting as a leader, are considered appropriate for men.[21] This leads to the think leader, think male phenomena, where we automatically associate men with leadership and higher status tasks (like first authorship or speaker invitations).[22, 23] Furthermore, acting against the stereotype (eg, a woman showing anger[24] or negotiating for more pay[25] or a man showing sadness[26]) can negatively impact wage and employment. Expecting social censure for violating gender norms, women develop a fear of the backlash that alone may shape behavior such that women may not express interest in having a high salary or negotiate for a raise.[27, 28, 29]

The specific system and institutional barriers that prevent female hospitalists from receiving equal pay and opportunities for leadership are not known, but one can surmise they are similar to those found in other specialties.[10, 30, 31] The findings of the studies of Burden et al.[2] and Weaver et al.[4] invite investigation of new questions specific to hospital medicine. Why are women in hospital medicine working more night shifts? Does this impact leadership or scholarship opportunities? Why are women documenting less productivity? Are they spending more time with patients, as they do in other settings?[32] What influences their practice choice? We would like to believe that there is something about hospital medicine that can explain the gender differences identified in these 2 studies. However, these data should prompt a serious and thorough examination of our specialty. We must accept that despite being a new specialty and a change leader, hospital medicine may not have escaped systematic gender bias that constrains the full participation and advancement of women.

But we believe that hospitalistsinnovators and change leaders in medicinewill be spurred to action to address the possibility of gender inequities. We do not need to know all of the causes to begin to address disparities, of every type, on an individual, institutional, and national level. As individuals, we can acknowledge that there are implicit assumptions that influence our decision making. No matter how unintentional, and even conflicting with evidence, these assumptions can lead us to judge women as less capable leaders than men or to automatically envision a high salary for a woman and man as different amounts. However, these automatic gender biases function as habits of mind, so they can be broken like any other unwanted habit.[33] Institutionally, we can also hold ourselves accountable for transparency in mentorship, leadership, scholarship, promotions, and wages to ensure diverse representation. We should routinely examine our practices to ensure the equitable hiring, pay, and promotion of our workforce.[18] National organizations and their respective journals should actively pursue diverse representation in leadership and membership on boards and committees, award nominees and recipients, and opportunities for invited editorials. Hospital medicinebeing young, innovative, and committed to changeis uniquely well suited to lead the charge for workforce equity. We can, and will, show the rest of medicine how it is done.

From a new concept to 44,000 practitioners in just 18 years,[1] there is no doubt that the word hospitalist is synonymous with innovation, leadership, growth, and change. Yet 2 articles in this month's Journal of Hospital Medicine prove that even our new field faces age‐old problems. Although women comprise half of all academic hospitalist and general internal medicine faculty, Burden et al.[2] showed that female hospitalists are less likely than male hospitalists to be division or section heads of hospital medicine, speakers at national meetings, and first or last authors on both research publications and editorials. This is made more concerning given that women are more likely to choose academic hospital medicine careers,[3] as they represent one‐third of all hospitalists but half of the academic hospitalist workforce.[2, 3] Findings in general internal medicine were similar, except that equal numbers of women and men were national meeting speakers and first authors on research publications (but not editorials). Weaver et al.[4] shed even more light on this disparity, and found that female hospitalists made $14,581 less per year than their male counterparts, even after adjusting for relevant differences. Weaver and colleagues also found other gender‐specific differences: women worked more nights and had fewer billable encounters per hospitalist shift than men.

Unfortunately, these trends are not new or limited to hospital medicine. For decades, almost equal numbers of women and men have entered medical school,[5] yet women are under‐represented in high status specialties,[6] less likely to be first or senior authors on original research studies compared to men,[7] less likely to be promoted,[8] and women physicians are consistently paid less than men across specialties.[9, 10] Simple analyses have not yet explained these disparities. Compared with men, women have similar leadership aspirations[11, 12] and are at least as effective as leaders.[13, 14, 15] Yet equity has not been attained.

Implicit bias research suggests that gender stereotypes influence women at all career stages.[16, 17, 18] For example, an elegant study conducted by Correll et al. identified a motherhood penalty, where indicating membership in the elementary school parent‐teacher organization on one's curriculum vitae hurt women's chances of employment and pay, but actually helped men.[19] Gender stereotypes exist, even among those who do not support their content. The universal reinforcement of such stereotypes over time leads to implicit but prescriptive rules about how women and men should act.[20] In particular, communal behaviors, including being cooperative, kind, and understanding, are typically associated with women, and agentic behaviors, including being ambitious and acting as a leader, are considered appropriate for men.[21] This leads to the think leader, think male phenomena, where we automatically associate men with leadership and higher status tasks (like first authorship or speaker invitations).[22, 23] Furthermore, acting against the stereotype (eg, a woman showing anger[24] or negotiating for more pay[25] or a man showing sadness[26]) can negatively impact wage and employment. Expecting social censure for violating gender norms, women develop a fear of the backlash that alone may shape behavior such that women may not express interest in having a high salary or negotiate for a raise.[27, 28, 29]

The specific system and institutional barriers that prevent female hospitalists from receiving equal pay and opportunities for leadership are not known, but one can surmise they are similar to those found in other specialties.[10, 30, 31] The findings of the studies of Burden et al.[2] and Weaver et al.[4] invite investigation of new questions specific to hospital medicine. Why are women in hospital medicine working more night shifts? Does this impact leadership or scholarship opportunities? Why are women documenting less productivity? Are they spending more time with patients, as they do in other settings?[32] What influences their practice choice? We would like to believe that there is something about hospital medicine that can explain the gender differences identified in these 2 studies. However, these data should prompt a serious and thorough examination of our specialty. We must accept that despite being a new specialty and a change leader, hospital medicine may not have escaped systematic gender bias that constrains the full participation and advancement of women.

But we believe that hospitalistsinnovators and change leaders in medicinewill be spurred to action to address the possibility of gender inequities. We do not need to know all of the causes to begin to address disparities, of every type, on an individual, institutional, and national level. As individuals, we can acknowledge that there are implicit assumptions that influence our decision making. No matter how unintentional, and even conflicting with evidence, these assumptions can lead us to judge women as less capable leaders than men or to automatically envision a high salary for a woman and man as different amounts. However, these automatic gender biases function as habits of mind, so they can be broken like any other unwanted habit.[33] Institutionally, we can also hold ourselves accountable for transparency in mentorship, leadership, scholarship, promotions, and wages to ensure diverse representation. We should routinely examine our practices to ensure the equitable hiring, pay, and promotion of our workforce.[18] National organizations and their respective journals should actively pursue diverse representation in leadership and membership on boards and committees, award nominees and recipients, and opportunities for invited editorials. Hospital medicinebeing young, innovative, and committed to changeis uniquely well suited to lead the charge for workforce equity. We can, and will, show the rest of medicine how it is done.

This year, we celebrate the 10th anniversary of this esteemed publication, and it is indeed an occasion for celebration. For those of us who were there at the creation of the hospitalist field, the establishment of a vibrant academic journal was a dream, one whose fulfillment was central to the legitimization of our field as a full‐fledged specialty. After a decade and 83 issues, the Journal of Hospital Medicine is a formidable source of information, cohesion, and pride.

The anniversary comes at a particularly interesting time for hospitals and hospitalists. Our field's lifeblood has been in trailblazing and continuous reinvention. We were the first physician specialty that embraced the mantra of systems thinking, as captured in our famous metaphor that we care for two sick patients: the person and the system. We were the first field that proudly, and without a hint of shame, allied ourselves with hospital leaders, believing that we were mutually dependent on one another, and that our ability to make change happen and stick was better if we were working with our institutions' leaders. In creating our professional society (and this journal), we took unusual pains to be inclusiveof academic and community‐based hospitalists, or hospitalists entering the field from a variety of backgrounds, of hospitalists caring for adults and kids, and of nonphysician providers.

Our efforts have paid off. Leaders as prominent as Don Berwick have observed that hospitalists have become the essential army of improvers in hospitals and healthcare systems. Hospitalists have made immense contributions at their own institutions, and are increasingly assuming leadership roles both locally and nationally. It is not a coincidence that Medicare's top physician (Patrick Conway) and the Surgeon General (Vivek Murthy) are both hospitalists. Although there have been a few bumps along the way, hospitalists are generally satisfied with their careers, respected by their colleagues, accepted by their patients, and pleased to be members of the fastest growing specialty in the history of modern medicine.

All of this should leave us all feeling warm, proud and more than a little nervous. We are now a mature medical specialty, no longer upstarts, and the natural inclination, in a changing world, will be to hunker down and protect what we have. Of course, some of that is reasonable and appropriate (for example, to fight for our fair share of a bundled payment pie),[1] but some of it will be wrong, even self‐defeating. The world of healthcare is changing fast, and our ability to stay relevant and indispensable will depend on our ability to evolve to meet new conditions and needs.

Let us consider some of the major trends playing out in healthcare. The biggest is the brisk and unmistakable shift from volume to value.[2] This is a trend we have been on top of, because this really has been our field's raison d'tre: improving value in the hospital by cutting costs and length of stay while improving (or at least keeping neutral) quality and safety.[3] However, a world under intense value pressure will work hard to move patients from hospital to less expensive postacute settings, and will insist on seamless handoffs between the hospital and such settings. Thoughtful hospital medicine groups are thinking hard about this trend, and many are placing colleagues in skilled nursing facilities, or at the very least tightening their connections to the postacute facilities in their healthcare ecosystem. We no longer have the luxury of confining our talents and energies to those things that take place within the 4 walls of the hospital.

Another trend is the digitization of healthcare, a trend turbocharged by $30 billion in federal incentive payments distributed between 2009 and 2014.[4] Here too, hospitalists have emerged as leaders in information technology (IT) implementations, and a disproportionate number of chief medical information officers and other IT leaders seem to be hospitalists. Splendid. But it is also up to us to help figure out how to use IT tools effectively. The notes have morphed into bloated, copy‐and‐pasteridden monstrosities: let us figure out what a good note should look like in the digital era, and then implement educational and system changes to create a new standard. We no longer go to radiology because we do not need to to see our films; let us think about what the loss of the collegial exchange with our radiology colleagues has cost, and then set out to develop new systems to reimagine it. Right now, big data are mostly hype and unrequited promise. Who better than hospitalists to dive in and start making sense of the data to predict risks or help point to better treatments?

Another trend is population health. Although I do not foresee a return to the Marcus Welby model of a kindly physician following the patient everywhere, I can imagine certain patients (mostly those with several social and clinical comorbidities and at least 3 admissions per year) who might be well served by a back‐to‐the‐future system in which a primary care provider follows them into the hospital, perhaps comanaging the patients with the on‐service hospitalist. David Meltzer, at the University of Chicago, is currently studying such a model, and I look forward to seeing his results.[5] Rather than rejecting such experiments as violating the usual hospitalist structure, we must embrace them, at least until the evidence is in.

In the end, the field of hospital medicine emerged and thrived because of the promise, and later the evidence, that our presence led to better quality, safety, patient experience, education, and efficiency. This mandate must remain our mantra, even if it means that we have to evolve our model in keeping with a changing healthcare landscape. The minute we stop evolving is the minute our field starts planting the seeds of its own destruction.

This year, we celebrate the 10th anniversary of this esteemed publication, and it is indeed an occasion for celebration. For those of us who were there at the creation of the hospitalist field, the establishment of a vibrant academic journal was a dream, one whose fulfillment was central to the legitimization of our field as a full‐fledged specialty. After a decade and 83 issues, the Journal of Hospital Medicine is a formidable source of information, cohesion, and pride.

The anniversary comes at a particularly interesting time for hospitals and hospitalists. Our field's lifeblood has been in trailblazing and continuous reinvention. We were the first physician specialty that embraced the mantra of systems thinking, as captured in our famous metaphor that we care for two sick patients: the person and the system. We were the first field that proudly, and without a hint of shame, allied ourselves with hospital leaders, believing that we were mutually dependent on one another, and that our ability to make change happen and stick was better if we were working with our institutions' leaders. In creating our professional society (and this journal), we took unusual pains to be inclusiveof academic and community‐based hospitalists, or hospitalists entering the field from a variety of backgrounds, of hospitalists caring for adults and kids, and of nonphysician providers.

Our efforts have paid off. Leaders as prominent as Don Berwick have observed that hospitalists have become the essential army of improvers in hospitals and healthcare systems. Hospitalists have made immense contributions at their own institutions, and are increasingly assuming leadership roles both locally and nationally. It is not a coincidence that Medicare's top physician (Patrick Conway) and the Surgeon General (Vivek Murthy) are both hospitalists. Although there have been a few bumps along the way, hospitalists are generally satisfied with their careers, respected by their colleagues, accepted by their patients, and pleased to be members of the fastest growing specialty in the history of modern medicine.

All of this should leave us all feeling warm, proud and more than a little nervous. We are now a mature medical specialty, no longer upstarts, and the natural inclination, in a changing world, will be to hunker down and protect what we have. Of course, some of that is reasonable and appropriate (for example, to fight for our fair share of a bundled payment pie),[1] but some of it will be wrong, even self‐defeating. The world of healthcare is changing fast, and our ability to stay relevant and indispensable will depend on our ability to evolve to meet new conditions and needs.

Let us consider some of the major trends playing out in healthcare. The biggest is the brisk and unmistakable shift from volume to value.[2] This is a trend we have been on top of, because this really has been our field's raison d'tre: improving value in the hospital by cutting costs and length of stay while improving (or at least keeping neutral) quality and safety.[3] However, a world under intense value pressure will work hard to move patients from hospital to less expensive postacute settings, and will insist on seamless handoffs between the hospital and such settings. Thoughtful hospital medicine groups are thinking hard about this trend, and many are placing colleagues in skilled nursing facilities, or at the very least tightening their connections to the postacute facilities in their healthcare ecosystem. We no longer have the luxury of confining our talents and energies to those things that take place within the 4 walls of the hospital.

Another trend is the digitization of healthcare, a trend turbocharged by $30 billion in federal incentive payments distributed between 2009 and 2014.[4] Here too, hospitalists have emerged as leaders in information technology (IT) implementations, and a disproportionate number of chief medical information officers and other IT leaders seem to be hospitalists. Splendid. But it is also up to us to help figure out how to use IT tools effectively. The notes have morphed into bloated, copy‐and‐pasteridden monstrosities: let us figure out what a good note should look like in the digital era, and then implement educational and system changes to create a new standard. We no longer go to radiology because we do not need to to see our films; let us think about what the loss of the collegial exchange with our radiology colleagues has cost, and then set out to develop new systems to reimagine it. Right now, big data are mostly hype and unrequited promise. Who better than hospitalists to dive in and start making sense of the data to predict risks or help point to better treatments?

Another trend is population health. Although I do not foresee a return to the Marcus Welby model of a kindly physician following the patient everywhere, I can imagine certain patients (mostly those with several social and clinical comorbidities and at least 3 admissions per year) who might be well served by a back‐to‐the‐future system in which a primary care provider follows them into the hospital, perhaps comanaging the patients with the on‐service hospitalist. David Meltzer, at the University of Chicago, is currently studying such a model, and I look forward to seeing his results.[5] Rather than rejecting such experiments as violating the usual hospitalist structure, we must embrace them, at least until the evidence is in.

In the end, the field of hospital medicine emerged and thrived because of the promise, and later the evidence, that our presence led to better quality, safety, patient experience, education, and efficiency. This mandate must remain our mantra, even if it means that we have to evolve our model in keeping with a changing healthcare landscape. The minute we stop evolving is the minute our field starts planting the seeds of its own destruction.

This year, we celebrate the 10th anniversary of this esteemed publication, and it is indeed an occasion for celebration. For those of us who were there at the creation of the hospitalist field, the establishment of a vibrant academic journal was a dream, one whose fulfillment was central to the legitimization of our field as a full‐fledged specialty. After a decade and 83 issues, the Journal of Hospital Medicine is a formidable source of information, cohesion, and pride.

The anniversary comes at a particularly interesting time for hospitals and hospitalists. Our field's lifeblood has been in trailblazing and continuous reinvention. We were the first physician specialty that embraced the mantra of systems thinking, as captured in our famous metaphor that we care for two sick patients: the person and the system. We were the first field that proudly, and without a hint of shame, allied ourselves with hospital leaders, believing that we were mutually dependent on one another, and that our ability to make change happen and stick was better if we were working with our institutions' leaders. In creating our professional society (and this journal), we took unusual pains to be inclusiveof academic and community‐based hospitalists, or hospitalists entering the field from a variety of backgrounds, of hospitalists caring for adults and kids, and of nonphysician providers.

Our efforts have paid off. Leaders as prominent as Don Berwick have observed that hospitalists have become the essential army of improvers in hospitals and healthcare systems. Hospitalists have made immense contributions at their own institutions, and are increasingly assuming leadership roles both locally and nationally. It is not a coincidence that Medicare's top physician (Patrick Conway) and the Surgeon General (Vivek Murthy) are both hospitalists. Although there have been a few bumps along the way, hospitalists are generally satisfied with their careers, respected by their colleagues, accepted by their patients, and pleased to be members of the fastest growing specialty in the history of modern medicine.

All of this should leave us all feeling warm, proud and more than a little nervous. We are now a mature medical specialty, no longer upstarts, and the natural inclination, in a changing world, will be to hunker down and protect what we have. Of course, some of that is reasonable and appropriate (for example, to fight for our fair share of a bundled payment pie),[1] but some of it will be wrong, even self‐defeating. The world of healthcare is changing fast, and our ability to stay relevant and indispensable will depend on our ability to evolve to meet new conditions and needs.

Let us consider some of the major trends playing out in healthcare. The biggest is the brisk and unmistakable shift from volume to value.[2] This is a trend we have been on top of, because this really has been our field's raison d'tre: improving value in the hospital by cutting costs and length of stay while improving (or at least keeping neutral) quality and safety.[3] However, a world under intense value pressure will work hard to move patients from hospital to less expensive postacute settings, and will insist on seamless handoffs between the hospital and such settings. Thoughtful hospital medicine groups are thinking hard about this trend, and many are placing colleagues in skilled nursing facilities, or at the very least tightening their connections to the postacute facilities in their healthcare ecosystem. We no longer have the luxury of confining our talents and energies to those things that take place within the 4 walls of the hospital.

Another trend is the digitization of healthcare, a trend turbocharged by $30 billion in federal incentive payments distributed between 2009 and 2014.[4] Here too, hospitalists have emerged as leaders in information technology (IT) implementations, and a disproportionate number of chief medical information officers and other IT leaders seem to be hospitalists. Splendid. But it is also up to us to help figure out how to use IT tools effectively. The notes have morphed into bloated, copy‐and‐pasteridden monstrosities: let us figure out what a good note should look like in the digital era, and then implement educational and system changes to create a new standard. We no longer go to radiology because we do not need to to see our films; let us think about what the loss of the collegial exchange with our radiology colleagues has cost, and then set out to develop new systems to reimagine it. Right now, big data are mostly hype and unrequited promise. Who better than hospitalists to dive in and start making sense of the data to predict risks or help point to better treatments?

Another trend is population health. Although I do not foresee a return to the Marcus Welby model of a kindly physician following the patient everywhere, I can imagine certain patients (mostly those with several social and clinical comorbidities and at least 3 admissions per year) who might be well served by a back‐to‐the‐future system in which a primary care provider follows them into the hospital, perhaps comanaging the patients with the on‐service hospitalist. David Meltzer, at the University of Chicago, is currently studying such a model, and I look forward to seeing his results.[5] Rather than rejecting such experiments as violating the usual hospitalist structure, we must embrace them, at least until the evidence is in.

In the end, the field of hospital medicine emerged and thrived because of the promise, and later the evidence, that our presence led to better quality, safety, patient experience, education, and efficiency. This mandate must remain our mantra, even if it means that we have to evolve our model in keeping with a changing healthcare landscape. The minute we stop evolving is the minute our field starts planting the seeds of its own destruction.