An 83‐year‐old man admitted for weakness, lethargy, and mental status changes was found to have human immunodeficiency virus (HIV) disease and cryptococcal meningitis. His hospital course was complicated by worsening hyponatremia (sodium < 136 mEq/L). By hospital day 6, the patient's serum sodium had declined to 127 mEq/L from his admission level of 133 mEq/L. The initial impression was that the patient had syndrome of inappropriate antidiuretic hormone (SIADH) and fluid restriction to less than 1500 mL per day was initiated. By hospital day 11, serum sodium continued to decline, to 123 mEq/L, despite fluid restriction.

The past medical history was remarkable for coronary artery disease, hypertension, hyperlipidemia, and anemia, but by self‐report he had not been taking any medications. His review of systems was positive for intermittent bouts of diarrhea.

Vital signs on day 11 included a temperature of 37.3C, blood pressure (BP) of 105/55 mm Hg, and pulse of 90 beats per minute. The BP on admission had been 145/86 mm Hg but had steadily declined with fluid restriction. On physical examination, he appeared thin and cachetic with no evidence of jugular venous distention, rales, or peripheral edema to suggest volume overload. He had been receiving 2 to 4 L of isotonic saline daily for 5 days before the fluid restriction was initiated. The urine output continuously exceeded his intake by at least 500 mL per day throughout his hospital course. His only inpatient medications were amphotericin B and flucytosine. For nutritional supplementation, he was receiving a high‐calorie supplement with free‐water flushes via a nasogastric tube.

Laboratory results revealed a serum sodium concentration of 123 mEq/L, serum potassium of 4.4 mEq/L, serum creatinine of 0.6 mg/dL, urine sodium of 139 mEq/L, serum osmolality of 272 mOsm/kg, and urine osmolality of 598 mOsm/kg (see Table 1). Urinalysis revealed a specific gravity of 1.030. A random serum cortisol level was 11.1 g/dL. A thyroid‐stimulating hormone (TSH) level was 1.32 IU/mL. Brain natriuretic peptide (BNP) was elevated, at 686 pg/mL. A fractional excretion of uric acid was also elevated, at 83.8%.

The clinical assessment was volume depletion given the high urine specific gravity, decreasing BP, and a negative fluid balance. The hyponatremia was determined to be due to sodium loss rather than dilution from inappropriate antidiuretic hormone secretion. Intravenous fluid (IVF) hydration with isotonic saline was initiated with a goal to keep the patient in positive fluid balance. The serum sodium level gradually improved to 140 mEq/L over the next 10 days. Attempts to decrease the rate of IVF resulted in a fall in serum sodium and improved when isotonic saline was increased. Eventually, the patient was placed on fludrocortisone, which normalized his urine output and serum sodium.

The response to the treatment regimen supported our diagnosis of cerebral salt wasting (CSW). The patient's serum sodium concentration upon discharge was 135 mEq/L.

Discussion

Our case illustrates the diagnostic challenge presented to physicians when they manage hyponatremia in the setting of a central nervous system (CNS) event. Hyponatremia (sodium < 136 mEq/L) has been associated with confusion, lethargy, seizures, coma, and even death.1 Hyponatremia has been reported to occur in up to 30% of the patients with subarachnoid hemorrhage.2, 3

SIADH is frequently the cause of hyponatremia in a patient with a concurrent intracranial process. However, CSW is an important diagnosis to consider and differentiate from SIADH. In a retrospective review of 316 patients with subarachnoid hemorrhage and hyponatremia, 69% were determined to be due to SIADH while 6.5% were from CSW.4 Both CSW and SIADH have been reported to occur in the setting of head trauma, intracranial or metastatic neoplasm, carcinomatous or infectious meningitis, subarachnoid hemorrhage, and CNS surgery. Cryptococcal meningitis as an etiology of CSW has not been previously reported.

The main differentiating feature between SIADH and CSW is that CSW is a dysfunction of renal sodium absorption whereas in SIADH renal sodium handling is intact. This also leads to a difference in the extracellular volume status. SIADH is associated with an increased to normal volume status whereas CSW is a volume‐depleted state. Our patient exhibited a low serum osmolality and a high urine osmolality in the context of hyponatremia, which is present in both CSW and SIADH. However, the clinical course and presentation suggested volume loss, specifically the diarrhea, high urine specific gravity, declining BP, and a negative fluid balance. Some other features that are helpful in determining the volume status may include orthostatic changes, tachycardia, and skin turgor.

Our patient had a low serum uric acid, which is also present in both SIADH and CSW. The key difference between the 2 is that while uric acid will improve with resolution of hyponatremia in SIADH, it will remain low in CSW, as in our patient's uric acid levels, which remained low after normalization of the serum sodium.

Finally, the hyponatremia improved with isotonic fluid repletion, which would not occur in SIADH. The majority of the CSW patients will respond to volume repletion alone, as CSW is a transient condition that will usually resolve in 3 to 4 weeks.3 However, a few patients may require fludrocortisone, as was needed in our patient.

The renal wasting of sodium in CSW is poorly understood. Some postulated mechanisms cite the disruption of sympathetic neural input to the kidney and natriuresis induced by natriuretic peptides. Natriuretic peptides, in particular BNP, have been reported to be elevated in patients with CSW.5, 6 Natriuretic peptides cause salt wasting by inhibition of sodium reabsorption in renal tubule and intramedullary collecting.5, 6 Renin and aldosterone release can also inhibited by the natriuretic peptides. BNP levels were elevated in our patient despite volume loss and no signs of congestive heart failure. Cardiac congestion is a possible etiology for the elevated BNP levels, which peaked to 900 pg/mL on hospital day 37. However, 3 days later the BNP levels declined to 222 pg/mL despite the fact that he was continually in positive fluid balance, suggesting that the BNP elevation was due to CSW and not heart failure.

Conclusions

Our case illustrates the diagnostic and management challenge of hyponatremia in the setting of a CNS event. Both SIADH and CSW are possible etiologies but it is important to make a differentiation. Levels of natriuretic peptides and changes in fractional excretion of uric acid may help differentiate between the 2 conditions.6 The key difference mechanistically is that CSW is due to sodium‐handling deficits, whereas in SIADH sodium‐handling is intact. It is essential to establish volume status since SIADH is a euvolemic to mildly hypervolemic state vs. CSW, which is a volume‐depleted state.7

CSW is well recognized in the neurosurgical arena. The hospitalist will encounter neurosurgical patients with increasing frequency, and thus having an understanding of this disorder, including its diagnosis and treatment, is key.

An 83‐year‐old man admitted for weakness, lethargy, and mental status changes was found to have human immunodeficiency virus (HIV) disease and cryptococcal meningitis. His hospital course was complicated by worsening hyponatremia (sodium < 136 mEq/L). By hospital day 6, the patient's serum sodium had declined to 127 mEq/L from his admission level of 133 mEq/L. The initial impression was that the patient had syndrome of inappropriate antidiuretic hormone (SIADH) and fluid restriction to less than 1500 mL per day was initiated. By hospital day 11, serum sodium continued to decline, to 123 mEq/L, despite fluid restriction.

The past medical history was remarkable for coronary artery disease, hypertension, hyperlipidemia, and anemia, but by self‐report he had not been taking any medications. His review of systems was positive for intermittent bouts of diarrhea.

Vital signs on day 11 included a temperature of 37.3C, blood pressure (BP) of 105/55 mm Hg, and pulse of 90 beats per minute. The BP on admission had been 145/86 mm Hg but had steadily declined with fluid restriction. On physical examination, he appeared thin and cachetic with no evidence of jugular venous distention, rales, or peripheral edema to suggest volume overload. He had been receiving 2 to 4 L of isotonic saline daily for 5 days before the fluid restriction was initiated. The urine output continuously exceeded his intake by at least 500 mL per day throughout his hospital course. His only inpatient medications were amphotericin B and flucytosine. For nutritional supplementation, he was receiving a high‐calorie supplement with free‐water flushes via a nasogastric tube.

Laboratory results revealed a serum sodium concentration of 123 mEq/L, serum potassium of 4.4 mEq/L, serum creatinine of 0.6 mg/dL, urine sodium of 139 mEq/L, serum osmolality of 272 mOsm/kg, and urine osmolality of 598 mOsm/kg (see Table 1). Urinalysis revealed a specific gravity of 1.030. A random serum cortisol level was 11.1 g/dL. A thyroid‐stimulating hormone (TSH) level was 1.32 IU/mL. Brain natriuretic peptide (BNP) was elevated, at 686 pg/mL. A fractional excretion of uric acid was also elevated, at 83.8%.

The clinical assessment was volume depletion given the high urine specific gravity, decreasing BP, and a negative fluid balance. The hyponatremia was determined to be due to sodium loss rather than dilution from inappropriate antidiuretic hormone secretion. Intravenous fluid (IVF) hydration with isotonic saline was initiated with a goal to keep the patient in positive fluid balance. The serum sodium level gradually improved to 140 mEq/L over the next 10 days. Attempts to decrease the rate of IVF resulted in a fall in serum sodium and improved when isotonic saline was increased. Eventually, the patient was placed on fludrocortisone, which normalized his urine output and serum sodium.

The response to the treatment regimen supported our diagnosis of cerebral salt wasting (CSW). The patient's serum sodium concentration upon discharge was 135 mEq/L.

Discussion

Our case illustrates the diagnostic challenge presented to physicians when they manage hyponatremia in the setting of a central nervous system (CNS) event. Hyponatremia (sodium < 136 mEq/L) has been associated with confusion, lethargy, seizures, coma, and even death.1 Hyponatremia has been reported to occur in up to 30% of the patients with subarachnoid hemorrhage.2, 3

SIADH is frequently the cause of hyponatremia in a patient with a concurrent intracranial process. However, CSW is an important diagnosis to consider and differentiate from SIADH. In a retrospective review of 316 patients with subarachnoid hemorrhage and hyponatremia, 69% were determined to be due to SIADH while 6.5% were from CSW.4 Both CSW and SIADH have been reported to occur in the setting of head trauma, intracranial or metastatic neoplasm, carcinomatous or infectious meningitis, subarachnoid hemorrhage, and CNS surgery. Cryptococcal meningitis as an etiology of CSW has not been previously reported.

The main differentiating feature between SIADH and CSW is that CSW is a dysfunction of renal sodium absorption whereas in SIADH renal sodium handling is intact. This also leads to a difference in the extracellular volume status. SIADH is associated with an increased to normal volume status whereas CSW is a volume‐depleted state. Our patient exhibited a low serum osmolality and a high urine osmolality in the context of hyponatremia, which is present in both CSW and SIADH. However, the clinical course and presentation suggested volume loss, specifically the diarrhea, high urine specific gravity, declining BP, and a negative fluid balance. Some other features that are helpful in determining the volume status may include orthostatic changes, tachycardia, and skin turgor.

Our patient had a low serum uric acid, which is also present in both SIADH and CSW. The key difference between the 2 is that while uric acid will improve with resolution of hyponatremia in SIADH, it will remain low in CSW, as in our patient's uric acid levels, which remained low after normalization of the serum sodium.

Finally, the hyponatremia improved with isotonic fluid repletion, which would not occur in SIADH. The majority of the CSW patients will respond to volume repletion alone, as CSW is a transient condition that will usually resolve in 3 to 4 weeks.3 However, a few patients may require fludrocortisone, as was needed in our patient.

The renal wasting of sodium in CSW is poorly understood. Some postulated mechanisms cite the disruption of sympathetic neural input to the kidney and natriuresis induced by natriuretic peptides. Natriuretic peptides, in particular BNP, have been reported to be elevated in patients with CSW.5, 6 Natriuretic peptides cause salt wasting by inhibition of sodium reabsorption in renal tubule and intramedullary collecting.5, 6 Renin and aldosterone release can also inhibited by the natriuretic peptides. BNP levels were elevated in our patient despite volume loss and no signs of congestive heart failure. Cardiac congestion is a possible etiology for the elevated BNP levels, which peaked to 900 pg/mL on hospital day 37. However, 3 days later the BNP levels declined to 222 pg/mL despite the fact that he was continually in positive fluid balance, suggesting that the BNP elevation was due to CSW and not heart failure.

Conclusions

Our case illustrates the diagnostic and management challenge of hyponatremia in the setting of a CNS event. Both SIADH and CSW are possible etiologies but it is important to make a differentiation. Levels of natriuretic peptides and changes in fractional excretion of uric acid may help differentiate between the 2 conditions.6 The key difference mechanistically is that CSW is due to sodium‐handling deficits, whereas in SIADH sodium‐handling is intact. It is essential to establish volume status since SIADH is a euvolemic to mildly hypervolemic state vs. CSW, which is a volume‐depleted state.7

CSW is well recognized in the neurosurgical arena. The hospitalist will encounter neurosurgical patients with increasing frequency, and thus having an understanding of this disorder, including its diagnosis and treatment, is key.

An 83‐year‐old man admitted for weakness, lethargy, and mental status changes was found to have human immunodeficiency virus (HIV) disease and cryptococcal meningitis. His hospital course was complicated by worsening hyponatremia (sodium < 136 mEq/L). By hospital day 6, the patient's serum sodium had declined to 127 mEq/L from his admission level of 133 mEq/L. The initial impression was that the patient had syndrome of inappropriate antidiuretic hormone (SIADH) and fluid restriction to less than 1500 mL per day was initiated. By hospital day 11, serum sodium continued to decline, to 123 mEq/L, despite fluid restriction.

The past medical history was remarkable for coronary artery disease, hypertension, hyperlipidemia, and anemia, but by self‐report he had not been taking any medications. His review of systems was positive for intermittent bouts of diarrhea.

Vital signs on day 11 included a temperature of 37.3C, blood pressure (BP) of 105/55 mm Hg, and pulse of 90 beats per minute. The BP on admission had been 145/86 mm Hg but had steadily declined with fluid restriction. On physical examination, he appeared thin and cachetic with no evidence of jugular venous distention, rales, or peripheral edema to suggest volume overload. He had been receiving 2 to 4 L of isotonic saline daily for 5 days before the fluid restriction was initiated. The urine output continuously exceeded his intake by at least 500 mL per day throughout his hospital course. His only inpatient medications were amphotericin B and flucytosine. For nutritional supplementation, he was receiving a high‐calorie supplement with free‐water flushes via a nasogastric tube.

Laboratory results revealed a serum sodium concentration of 123 mEq/L, serum potassium of 4.4 mEq/L, serum creatinine of 0.6 mg/dL, urine sodium of 139 mEq/L, serum osmolality of 272 mOsm/kg, and urine osmolality of 598 mOsm/kg (see Table 1). Urinalysis revealed a specific gravity of 1.030. A random serum cortisol level was 11.1 g/dL. A thyroid‐stimulating hormone (TSH) level was 1.32 IU/mL. Brain natriuretic peptide (BNP) was elevated, at 686 pg/mL. A fractional excretion of uric acid was also elevated, at 83.8%.

The clinical assessment was volume depletion given the high urine specific gravity, decreasing BP, and a negative fluid balance. The hyponatremia was determined to be due to sodium loss rather than dilution from inappropriate antidiuretic hormone secretion. Intravenous fluid (IVF) hydration with isotonic saline was initiated with a goal to keep the patient in positive fluid balance. The serum sodium level gradually improved to 140 mEq/L over the next 10 days. Attempts to decrease the rate of IVF resulted in a fall in serum sodium and improved when isotonic saline was increased. Eventually, the patient was placed on fludrocortisone, which normalized his urine output and serum sodium.

The response to the treatment regimen supported our diagnosis of cerebral salt wasting (CSW). The patient's serum sodium concentration upon discharge was 135 mEq/L.

Discussion

Our case illustrates the diagnostic challenge presented to physicians when they manage hyponatremia in the setting of a central nervous system (CNS) event. Hyponatremia (sodium < 136 mEq/L) has been associated with confusion, lethargy, seizures, coma, and even death.1 Hyponatremia has been reported to occur in up to 30% of the patients with subarachnoid hemorrhage.2, 3

SIADH is frequently the cause of hyponatremia in a patient with a concurrent intracranial process. However, CSW is an important diagnosis to consider and differentiate from SIADH. In a retrospective review of 316 patients with subarachnoid hemorrhage and hyponatremia, 69% were determined to be due to SIADH while 6.5% were from CSW.4 Both CSW and SIADH have been reported to occur in the setting of head trauma, intracranial or metastatic neoplasm, carcinomatous or infectious meningitis, subarachnoid hemorrhage, and CNS surgery. Cryptococcal meningitis as an etiology of CSW has not been previously reported.

The main differentiating feature between SIADH and CSW is that CSW is a dysfunction of renal sodium absorption whereas in SIADH renal sodium handling is intact. This also leads to a difference in the extracellular volume status. SIADH is associated with an increased to normal volume status whereas CSW is a volume‐depleted state. Our patient exhibited a low serum osmolality and a high urine osmolality in the context of hyponatremia, which is present in both CSW and SIADH. However, the clinical course and presentation suggested volume loss, specifically the diarrhea, high urine specific gravity, declining BP, and a negative fluid balance. Some other features that are helpful in determining the volume status may include orthostatic changes, tachycardia, and skin turgor.

Our patient had a low serum uric acid, which is also present in both SIADH and CSW. The key difference between the 2 is that while uric acid will improve with resolution of hyponatremia in SIADH, it will remain low in CSW, as in our patient's uric acid levels, which remained low after normalization of the serum sodium.

Finally, the hyponatremia improved with isotonic fluid repletion, which would not occur in SIADH. The majority of the CSW patients will respond to volume repletion alone, as CSW is a transient condition that will usually resolve in 3 to 4 weeks.3 However, a few patients may require fludrocortisone, as was needed in our patient.

The renal wasting of sodium in CSW is poorly understood. Some postulated mechanisms cite the disruption of sympathetic neural input to the kidney and natriuresis induced by natriuretic peptides. Natriuretic peptides, in particular BNP, have been reported to be elevated in patients with CSW.5, 6 Natriuretic peptides cause salt wasting by inhibition of sodium reabsorption in renal tubule and intramedullary collecting.5, 6 Renin and aldosterone release can also inhibited by the natriuretic peptides. BNP levels were elevated in our patient despite volume loss and no signs of congestive heart failure. Cardiac congestion is a possible etiology for the elevated BNP levels, which peaked to 900 pg/mL on hospital day 37. However, 3 days later the BNP levels declined to 222 pg/mL despite the fact that he was continually in positive fluid balance, suggesting that the BNP elevation was due to CSW and not heart failure.

Conclusions

Our case illustrates the diagnostic and management challenge of hyponatremia in the setting of a CNS event. Both SIADH and CSW are possible etiologies but it is important to make a differentiation. Levels of natriuretic peptides and changes in fractional excretion of uric acid may help differentiate between the 2 conditions.6 The key difference mechanistically is that CSW is due to sodium‐handling deficits, whereas in SIADH sodium‐handling is intact. It is essential to establish volume status since SIADH is a euvolemic to mildly hypervolemic state vs. CSW, which is a volume‐depleted state.7

CSW is well recognized in the neurosurgical arena. The hospitalist will encounter neurosurgical patients with increasing frequency, and thus having an understanding of this disorder, including its diagnosis and treatment, is key.

Wachter and Goldman1 first described hospitalists in 1996 as a new breed of physicians who devote blocks of time exclusively to the care of hospitalized patients. Since its definition, the hospitalist model has prompted 2 major debates. First, does the hospitalist system improve clinical efficiency, quality of care, cost effectiveness, and patient satisfaction? A series of large and small randomized trials have all but definitively proven the hospitalist model's advantage. Yet whether the hospitalist model is good for patient care has proven to remain contentious, as most recently demonstrated by the discussion between Williams2 and Centor3 and others like it.4, 5 What is clear in these exchanges is that the debate has shifted to the second great debate: does the hospitalist model pose inherent conflicts in clinical ethics? What are the implications of the purposeful discontinuity in care, the autonomy issues raised by mandatory hospitalist use, and the structural management issues that potentially pit hospitalists against patients in fiduciary and financial conflicts of interest? These important issues are certainly not new, and the hospitalist model has made much effort to address some of them.6, 7 This work aims to serve as a review of these important ethical concerns, demonstrating how some questions have been answered, while some remain unanswered.

The Hospitalist Model's Founding Premise

A growing threshold for hospital admission in the last 3 decades caused primary care physicians (PCPs) to see a diminishing number of inpatients. A survey in 1978 found that PCPs spent 40% of their time in the hospital, rounding on 10 patients per day.8 By 2001, PCPs spent 10% of their time in the hospital on average, and most PCPs rounded on fewer than 2 inpatients per day.9 The cost of inefficiencies associated with primary coordination of care in the hospital increasingly outweighed the tradeoff of preserving the patient‐PCP relationship in the hospital. Converging with increasing attention on cost controls through the restructuring of service provision, the hospitalist was born. Wachter10 argued that the hospitalist model could alleviate inpatient demands placed on PCPs while improving the outcomes and lowering the cost of care for hospitalized patients.

Early on there were setbacks to proving Wachter's10 case. Small studies found hospitalists to have higher hospital charges and longer length of stays.11 A survey of PCPs found only 56% were satisfied with communication with hospitalists and that most believed that patients generally preferred to be cared for in the hospital by their regular physician.12, 13 Meltzer and Herthko14 found 70% of people sampled said they would prefer care by their own physician to that of a hospitalist if they were hospitalized for a general medical condition. Yet this study found in a national random‐digit phone survey that only 10% of the respondents would pay $750 for their PCP to follow them to the hospital, the cost savings of the hospitalist system proven by the only 2 randomized trials performed at the time.15, 16 To 90% of respondents, the value of the PCP at the bedside was not worth the cost tradeoff to keep them there.

The meteoric rise in the number of hospitalists reflects the many studies and reviews that affirmed the premise that hospitalists improved inpatient efficiency without harmful effects on quality of care.17, 18 In a large retrospective cohort study of over 75,000 patients in 45 hospitals across the country, Lindenauer et al.19 found that hospitalists had a $268 lower cost when compared to internists, $125 lower cost when compared to family physicians, and a shorter hospital stay by about one‐half day when compared to both groups. The group found no significant difference in rates of death or readmission rates. While called modest in the text, these savings over time and volume add up for hospitals. Patients benefit from hospitalist care, researchers hypothesize, because of their familiarity with hospital systems, their increased availability to patients, and their experience with common hospital problems. Though the Lindenauer et al.19 study was criticized for design flaws, it prompted the editorialist McMahon20 to assert that the question was sufficiently answered, and it was time to move on away from the studies focusing on cost and comparing outcomes. As Wachter21 wrote, the demand for hospitalists is now relatively de‐linked from the field's original premiseefficiency advantagesand is now both more diversified and more robust. The model has become an accepted mode of care for hospitalized patients, with up to 20,000 hospitalists currently practicing in 29% of all hospitals and in over one‐half of hospitals with over 200 beds in the United States.22, 23

The Patient‐Physician Relationship

Purposeful discontinuity of care in the hospitalist system has the potential to diminish the doctor‐patient relationship.12 This relationship is built on a bond of loyalty, confidentiality, and trust. Handing off care to a hospitalist when the patient is most vulnerable can be viewed as a violation of this covenant. According to Meltzer,24 the hospitalist model pits Franicis Peabody's25 intimate personal relationship between patient and physician against Adam Smith et al.'s11 benefits of specialization. Peabody25 observed that physicians' lack of understanding of their patients as persons is especially acute in the hospital, where

This movement toward patient‐centered medicine fits into an ever‐growing sentiment to value the social as well as the physiological, a holistic approach to the patient as a person. This emphasis was the original justification for PCPs to coordinate increasingly specialized hospital care and translate recommendations suitable to patients. Can the long‐term relationship between patient and PCP be replaced by the hospital generalist, or would hospitalists be inherently deficient in their abilities to coordinate care appropriate for patients? Hospitalized patients are frequently in no position to make complex decisions regarding their care.26 Lo7 argues that PCPs who know patients over extended periods of time are in a better position to respect patient wishes by individualizing discussions with patients and checking that patients' decisions are consistent with their core values. The long‐term relationship is also critical for designing a complex discharge plan suitable to the patients' ability and resources. Information about long‐term patient compliance with medications is much more available to PCPs. Patients trust physicians to keep promises made concerning end‐of‐life issues, and these assurances are vulnerable during handoffs of care. Pantilat et al.6 provide a case study of an outpatient Do‐Not‐Resuscitate order ineffective in the hospital. These scenarios occur because most written advance directives are unavailable in acute situations, and when they are, hospitalists unfamiliar with the patient's wishes may hesitate to act on directives not specific enough to answer the acute clinical question.27

Hospitalists' broadened responsibility to systematically improve the care of patients may potentially improve end‐of‐life care. Patient values can be better communicated to hospitalists by encouraging inpatients to complete advance directive surveys and then asking hospitalists to discuss those directives with their patients.6 Significantly, Auerbach and Pantilat28 found that end‐of‐life care was improved with hospitalist care. This chart review found hospitalists more likely to have discussions with patients and their families regarding care and providing comfort care more frequently at the time of death than community‐based physicians. The authors hypothesize that hospitalists may have better communication with dying patients and their families because they spend more time in the hospital each day, using frequent meetings to better understand the preferences of patients. These preferences often require clarification and often change after admission, making previous discussions about end‐of‐life care with PCPs moot. Greater expertise in hospital care may also allow hospitalists to better recognize patients who are nearing death and may explain the fewer symptoms documented by Auerbach and Pantilat28 at the end of life among patients cared for by hospitalists compared to community‐based physicians.

Hospital medicine has taken continuity of care issues seriously, and responded by making pragmatic recommendations to preserve the patient‐PCP relationship in the hospital and assuage the perception that patients have been dropped. Harlan et al.29 identify important issues around good communication between pediatric hospitalists and PCPs including the content and timing of communication beneficial to the patient. Hospitalists can use a standard script for introducing themselves to patients, explaining their role, and their continued coordination with the PCP.30 PCPs can still be involved in the care of their patients in hospitals through continuity visits or phone calls with patients and through better communication with hospitalists.31 Generally, reimbursing PCPs for their increased role in the hospitalist system can encourage better communication with hospitalists.19 Potential disagreements between PCPs and hospitalist regarding the care of the patient can be resolved through explicit conflict resolution procedures within the hospitalist system.6

These procedural solutions are only as successful as they are used. A large review by Kripalani et al.32 found communication between hospitalists and PCPs occurred infrequently (3%‐20%), affecting the quality of care in approximately 25% of follow‐up visits and contributing to PCP dissatisfaction. Sharma et al.33 found that continuity visits decreased from 50.5% in 1996 to 39.8% in 2006. In a survey of patients cared for in a hospitalist system, Hruby et al.34 found that 33% of hospitalized patients had some contact with their PCP directly and 66% of patients were satisfied with the contact they or their relative had with their PCP. When probed, patient satisfaction is too vague a measurement to assess the complex value of the patient‐physician relationship. Studying these issues may require relying more on individualized narratives rather than generalized statistics, or may require years of follow‐up. As Centor3 argues, we need this broader perspective of the patient's experience in order to understand the effects of the hospitalist model on patient trust in their PCP and in their overall care. Studies by Davis et al.35 and Halpert et al.36 assert that rising quality of care and patient satisfaction with the hospitalist system rebuts coordination of care concerns. Yet we need more studies investigating the relationship between improved communication and patient outcomes, as evidence is currently conflicting on this subject.32, 37, 38

The Journal of Hospital Medicine has pursued this research agenda; the April 2009 issue presents several studies describing best practices in the discharging of hospitalized patients. Manning et al.39 describe a tool to assess patient mobility after discharge, and O'Leary et al.40 used electronic health records to create a better discharge summary. Project BOOST (Better Outcomes for Older Adults Through Safe Transitions) has shown improvements in discharge transition procedures41 and the use of transition coaches for vulnerable older patients has been proven cost‐effective and has been scaled up to more than 100 healthcare organizations.42, 43

Inpatient care handoff to PCPs is not entirely novel, as surgeons, oncologists, cardiologists, and other specialists have always grappled with continuity of care. It would be prudent to investigate what can be learned from these efforts, and which practices can be best applied to the hospitalist model. More longitudinal studies need to investigate the prevalence and success of the procedural recommendations to preserve the patient‐physician relationship. We need to know more about what works and what does not. How have hospitals found novel ways in implementing these approaches, and how can they be applied to a diversity of hospital settings? We need a better outcome measurement than patient or physician satisfaction for probing the subtleties of the patient‐physician relationship. There is a sizeable population that does not have a PCP to care for them before hospitalization or after discharge, and discussions about continuity of care must address these patients. Last, these best practices and patient centered values need to be incorporated into the core competencies of residencies and fellowships for a new generation of hospitalists.

Maintaining the continuity of the physician‐patient relationship is an integral part of the original premise of the hospitalist model. Importantly, Meltzer24 found that this discontinuity within the hospital has the potential to eliminate the savings of the hospitalist system. Yet concerns about continuity of care do not sufficiently encompass the complexand at times fragilerelationship between physician and patient. The survival of the physician‐patient relationship depends on the hospitalist model's affirmation of the values of coordination and Peabody's25 approach to patient‐centered care. If the hospitalist model is to thrive, it needs to emphasize its duty as steward of the PCP‐patient relationship as much as it focuses on efficiency and cost‐effectiveness.

Patient Autonomy

The mandatory transfer of patients into the hospitalist model raises serious ethical issues. A survey in 2000 of PCPs found that 23% were required to use hospitalists for all admissions.44 Other surveys found this prevalence to be as low as 2%.12 Nevertheless, several high profile cases of Health Maintenance Organizations (HMOs)Prudential HealthCareSouth Florida, Prudential, Humana, and Cigna Corporationall using mandatory hospitalists, prompted protests from professional organizations and there were even legislative efforts to ban the practice of the mandatory use of hospitalists in 2000 and 2001.45 Today, most insurance plans, as well as the Society of Hospital Medicine (SHM), support voluntary rather than mandatory hospitalist use.46 Yet while not mandatory, the hospitalist is the default provider in many settings, giving a de facto mandate for hospitalist care. As Royo et al.47 point out, the rise in physician employment by hospitals has facilitated a self‐selecting progression toward a structural network that closely resembles the mandatory model.

While PCPs and internists contested mandatory hospitalist plans as infringements on their autonomy, they overlooked the harm to the patient's autonomy. When healthy in the ambulatory setting, the patient has the opportunity to choose his or her doctor to provide longitudinal care. When the patient is admitted acutely to a hospital, the patient does not have the freedom to choose a physician; the patient is assigned to the hospitalist on duty that night. This call for patient autonomy is of utmost importance in the hospitalized patient, where patients are increasingly sicker, their diseases under a high‐powered lens, and their options diminished. This freedom of choice is integral to the patient‐physician partnership. Yet this freedom of choice is largely hindered by the employer's choice in the health plan for their employees or an individual's ability to pay for a health plan. These represent some of the many barriers to choice facing patients in the American model of health insurance.

As the hospitalist system grows to become the accepted mode of hospital care, more patients need to be informed about the transition of care to another physician and what steps are taken to ensure appropriate continuity of care. Transfers of patients from PCPs to hospitalists must be voluntary, with the decision left to patient care preferences.48 Educating patients in the outpatient setting about the hospitalist model, its benefits, risks, and alternatives, is necessary for them to make informed decisions about hospital care. This will require the collaboration of PCPs and hospitalists together. The continued success of the model depends on the nurturance of the partnership between the PCP, the hospitalist, and the patient.

Meltzer and Herthko14 have proposed that patients pay a premium for the option to choose a PCP that is not mandated to transfer their care to a hospitalist, in order to offset cost savings with the hospitalist system. Yet Meltzer and Herthko's14 study suggests that many patients could not afford to pay this premium and, in effect, patient autonomy would be preserved for the affluent. This raises the oft‐neglected professional ethic of justice for low‐income patients. Alexander and Lantos49 were resigned to see this infringement on patient autonomy as an inevitable consequence of balancing the desires of patients with the drive to lower cost and improve outcomes. If the hospitalist model grows to be the predominant mode of care, it is unclear if patient choice can survive. Investigators need to test whether the advantages of hospitalist care can coexist with voluntary programs. If it proves that they indeed cannot, then the hospitalist system will need to respond to concerned patients with honest answers and find pragmatic solutions to diminishing patient choice.

Conflict of Interest

The hospitalist system's main benefit of cost‐savings prompted Pantilat et al.6 to wonder whether hospitalists would face a conflict of interest between what is best for the patient and what financial incentives and utilization review encourage or require them to do. The financial support provided by many hospitals to meet the operating expenses of hospitalist programs is often associated with explicit or implicit incentives to reduce the length of hospital stay and costs.50 With hospitals employing hospitalists and increasingly pressuring them to decrease length of stay and discharge patients quickly, patients may have no advocate to protect them from discharge planners. Many hospitalists supplement their income by supervising discharge planners, and a dispute would put the hospitalist in the uncomfortable position of advocating for his patient against his employer and colleagues. While conflicts of interests occur in many managed care arrangements, they may be more acute in hospitalist systems. A weakened patient‐physician relationship may put the patients' best interest inferior to the employer's interests. Hospitalists do not immediately deal with adverse consequences of premature discharges in the outpatient setting and virtually no malpractice case law considers the obligations and practices of hospitalists in these settings.51

The SHM identified a core competency of hospitalists to

Ethical issues concerning conflict of interest remain unanswered, largely because no information about organizational features such as explicit incentives for reductions in length of stay is available to researchers or to patients. This is the wrong approach and only feeds the fear that hospitalists may weigh patients' best interest with financial incentives. Abbo and Volandes53 have argued that ambivalence to cost considerations is hazardous. If the hospitalist model cannot be forthright with the active considerations of costs in daily clinical practice, it is unlikely to truly make strides at cost savings, and may even raise the cost of care in the long run.

Jonsen et al.54 provide ethical standards for considering costs in clinical decisions. First, a physician's first priority should be to provide patient‐centered care that focuses on medical indications and patient preferences. Second, quality care does not mean all available care; quality care reflects what is not only diagnostically sound and technically correct, but also appropriate. Third, conflicts of interest are most vulnerable when there is a failing of the patient‐physician relationship. Health care organizations should expect physicians to argue for policies that provide all services that have a reasonable likelihood of benefiting the patient. Fourth, patient and physician autonomy and freedom of choice should be maximized within the limits of the system. Persons should be fully informed of the constraints of the system before choosing it. Plans need to disclose any financial incentive arrangements that exist between the plan and the physician. And incentive arrangements should be based on quality of care rather than on underutilization of care services. Fifth, the system should reflect principles of just distribution, ensuring that all who have a fair claim to service should receive it without discrimination. Last, capitation plans should share risks among physicians, not patients, while incentives are provided for improvements in access, prevention, and patient satisfaction.

Conflicts of interest have been a concern for as long as physicians have been paid for services. Fears about interference into the doctor‐patient relationship, whether they are from government or business, continue to stall real efforts to lower skyrocketing medical costs. The hospitalist model rebuts conflict of interest claims with improved outcomes, efficiency, and quality of care in the many reviews cited above. These arguments do prove that the hospitalist model's emphasis on medically indicated and appropriate care does address Jonsen et al.'s54 first and second standards. Yet, as Jonsen et al.54 point out, without strongly emphasizing the patient‐physician relationship and patient autonomy, it leaves itself vulnerable to creating conflicts of interest. Hospitalist systems need to be forthright about their explicit or implicit incentive structures and disclose this information to patients in a timely manner for them to make informed decisions. These incentives should be linked to quality of care and patient satisfaction, not cost savings. Last, hospitalist training programs should make ethical cost considerations a core competency of their curriculum.

Conclusions

Hospitalism was founded on the premise that it could improve the quality and reduce the cost of hospital care. Many randomized studies have all but definitively proven this original assertion. It is now time for the model to prove that these gains are not to the detriment of the patient‐physician relationship. Hospitalism must define itself as the steward of this relationship, valuing it as much as it values outcomes and costs. This is of particular concern in the United States as Medicare Part A (payment for inpatient care) is scheduled to go bankrupt in 2019, leading to potentially reasonable fears of hospital‐motivated cost containment.57

Investigators must find an outcome that encompasses the complexity of the patient‐physician relationship, and methods to improve it must be studied and improved upon. Preserving the patient‐physician relationship is a systemic issue, and full‐time hospitalists may be in the best position to implement systemic reforms to improve communication and continuity of care. Pham's56 case study of a hospitalist piecing together disparate parts of the patient's story illustrates this point. This should include more investigation into the prevalence of use and success of methods aimed at protecting the patient‐physician relationship at critical points in the handover of care. When proven successful, The SHM should propose new standards and safeguards to insure that these methods become standard practice in patient care. This effort, led by Snow et al.,57 is currently underway.

A hospitalist model that does not emphasize mitigating the effects of the diminishing patient‐physician relationship leaves itself exposed to further infringements on autonomy and choice. It is unclear whether patient autonomy and choice can coexist in a successful hospitalist system. The consequences of these unanswered ethical questions need to be explored. The professions of primary care need to be more proactive in educating patients about choice of care in hospitals, and hospitalists need to provide that choice, allowing voluntary programs in hospital care when feasible.

When combined, a wounded patient‐physician relationship and impaired patient autonomy leave the hospitalist model vulnerable to claims of financial and fiduciary conflict of interest. These concerns need not be inherent to the hospitalist systems, but hospitalists will need to be forthright and honest about incentives structures, and link them to quality of care and patient satisfaction, not to efficiency and cost savings.

It is indeed time for hospitalism to move onaway from proving its founding premise, and toward addressing these lingering ethical issues. Hospitalism's continued growth and success depends on it.

Wachter and Goldman1 first described hospitalists in 1996 as a new breed of physicians who devote blocks of time exclusively to the care of hospitalized patients. Since its definition, the hospitalist model has prompted 2 major debates. First, does the hospitalist system improve clinical efficiency, quality of care, cost effectiveness, and patient satisfaction? A series of large and small randomized trials have all but definitively proven the hospitalist model's advantage. Yet whether the hospitalist model is good for patient care has proven to remain contentious, as most recently demonstrated by the discussion between Williams2 and Centor3 and others like it.4, 5 What is clear in these exchanges is that the debate has shifted to the second great debate: does the hospitalist model pose inherent conflicts in clinical ethics? What are the implications of the purposeful discontinuity in care, the autonomy issues raised by mandatory hospitalist use, and the structural management issues that potentially pit hospitalists against patients in fiduciary and financial conflicts of interest? These important issues are certainly not new, and the hospitalist model has made much effort to address some of them.6, 7 This work aims to serve as a review of these important ethical concerns, demonstrating how some questions have been answered, while some remain unanswered.

The Hospitalist Model's Founding Premise

A growing threshold for hospital admission in the last 3 decades caused primary care physicians (PCPs) to see a diminishing number of inpatients. A survey in 1978 found that PCPs spent 40% of their time in the hospital, rounding on 10 patients per day.8 By 2001, PCPs spent 10% of their time in the hospital on average, and most PCPs rounded on fewer than 2 inpatients per day.9 The cost of inefficiencies associated with primary coordination of care in the hospital increasingly outweighed the tradeoff of preserving the patient‐PCP relationship in the hospital. Converging with increasing attention on cost controls through the restructuring of service provision, the hospitalist was born. Wachter10 argued that the hospitalist model could alleviate inpatient demands placed on PCPs while improving the outcomes and lowering the cost of care for hospitalized patients.

Early on there were setbacks to proving Wachter's10 case. Small studies found hospitalists to have higher hospital charges and longer length of stays.11 A survey of PCPs found only 56% were satisfied with communication with hospitalists and that most believed that patients generally preferred to be cared for in the hospital by their regular physician.12, 13 Meltzer and Herthko14 found 70% of people sampled said they would prefer care by their own physician to that of a hospitalist if they were hospitalized for a general medical condition. Yet this study found in a national random‐digit phone survey that only 10% of the respondents would pay $750 for their PCP to follow them to the hospital, the cost savings of the hospitalist system proven by the only 2 randomized trials performed at the time.15, 16 To 90% of respondents, the value of the PCP at the bedside was not worth the cost tradeoff to keep them there.

The meteoric rise in the number of hospitalists reflects the many studies and reviews that affirmed the premise that hospitalists improved inpatient efficiency without harmful effects on quality of care.17, 18 In a large retrospective cohort study of over 75,000 patients in 45 hospitals across the country, Lindenauer et al.19 found that hospitalists had a $268 lower cost when compared to internists, $125 lower cost when compared to family physicians, and a shorter hospital stay by about one‐half day when compared to both groups. The group found no significant difference in rates of death or readmission rates. While called modest in the text, these savings over time and volume add up for hospitals. Patients benefit from hospitalist care, researchers hypothesize, because of their familiarity with hospital systems, their increased availability to patients, and their experience with common hospital problems. Though the Lindenauer et al.19 study was criticized for design flaws, it prompted the editorialist McMahon20 to assert that the question was sufficiently answered, and it was time to move on away from the studies focusing on cost and comparing outcomes. As Wachter21 wrote, the demand for hospitalists is now relatively de‐linked from the field's original premiseefficiency advantagesand is now both more diversified and more robust. The model has become an accepted mode of care for hospitalized patients, with up to 20,000 hospitalists currently practicing in 29% of all hospitals and in over one‐half of hospitals with over 200 beds in the United States.22, 23

The Patient‐Physician Relationship

Purposeful discontinuity of care in the hospitalist system has the potential to diminish the doctor‐patient relationship.12 This relationship is built on a bond of loyalty, confidentiality, and trust. Handing off care to a hospitalist when the patient is most vulnerable can be viewed as a violation of this covenant. According to Meltzer,24 the hospitalist model pits Franicis Peabody's25 intimate personal relationship between patient and physician against Adam Smith et al.'s11 benefits of specialization. Peabody25 observed that physicians' lack of understanding of their patients as persons is especially acute in the hospital, where

This movement toward patient‐centered medicine fits into an ever‐growing sentiment to value the social as well as the physiological, a holistic approach to the patient as a person. This emphasis was the original justification for PCPs to coordinate increasingly specialized hospital care and translate recommendations suitable to patients. Can the long‐term relationship between patient and PCP be replaced by the hospital generalist, or would hospitalists be inherently deficient in their abilities to coordinate care appropriate for patients? Hospitalized patients are frequently in no position to make complex decisions regarding their care.26 Lo7 argues that PCPs who know patients over extended periods of time are in a better position to respect patient wishes by individualizing discussions with patients and checking that patients' decisions are consistent with their core values. The long‐term relationship is also critical for designing a complex discharge plan suitable to the patients' ability and resources. Information about long‐term patient compliance with medications is much more available to PCPs. Patients trust physicians to keep promises made concerning end‐of‐life issues, and these assurances are vulnerable during handoffs of care. Pantilat et al.6 provide a case study of an outpatient Do‐Not‐Resuscitate order ineffective in the hospital. These scenarios occur because most written advance directives are unavailable in acute situations, and when they are, hospitalists unfamiliar with the patient's wishes may hesitate to act on directives not specific enough to answer the acute clinical question.27

Hospitalists' broadened responsibility to systematically improve the care of patients may potentially improve end‐of‐life care. Patient values can be better communicated to hospitalists by encouraging inpatients to complete advance directive surveys and then asking hospitalists to discuss those directives with their patients.6 Significantly, Auerbach and Pantilat28 found that end‐of‐life care was improved with hospitalist care. This chart review found hospitalists more likely to have discussions with patients and their families regarding care and providing comfort care more frequently at the time of death than community‐based physicians. The authors hypothesize that hospitalists may have better communication with dying patients and their families because they spend more time in the hospital each day, using frequent meetings to better understand the preferences of patients. These preferences often require clarification and often change after admission, making previous discussions about end‐of‐life care with PCPs moot. Greater expertise in hospital care may also allow hospitalists to better recognize patients who are nearing death and may explain the fewer symptoms documented by Auerbach and Pantilat28 at the end of life among patients cared for by hospitalists compared to community‐based physicians.

Hospital medicine has taken continuity of care issues seriously, and responded by making pragmatic recommendations to preserve the patient‐PCP relationship in the hospital and assuage the perception that patients have been dropped. Harlan et al.29 identify important issues around good communication between pediatric hospitalists and PCPs including the content and timing of communication beneficial to the patient. Hospitalists can use a standard script for introducing themselves to patients, explaining their role, and their continued coordination with the PCP.30 PCPs can still be involved in the care of their patients in hospitals through continuity visits or phone calls with patients and through better communication with hospitalists.31 Generally, reimbursing PCPs for their increased role in the hospitalist system can encourage better communication with hospitalists.19 Potential disagreements between PCPs and hospitalist regarding the care of the patient can be resolved through explicit conflict resolution procedures within the hospitalist system.6

These procedural solutions are only as successful as they are used. A large review by Kripalani et al.32 found communication between hospitalists and PCPs occurred infrequently (3%‐20%), affecting the quality of care in approximately 25% of follow‐up visits and contributing to PCP dissatisfaction. Sharma et al.33 found that continuity visits decreased from 50.5% in 1996 to 39.8% in 2006. In a survey of patients cared for in a hospitalist system, Hruby et al.34 found that 33% of hospitalized patients had some contact with their PCP directly and 66% of patients were satisfied with the contact they or their relative had with their PCP. When probed, patient satisfaction is too vague a measurement to assess the complex value of the patient‐physician relationship. Studying these issues may require relying more on individualized narratives rather than generalized statistics, or may require years of follow‐up. As Centor3 argues, we need this broader perspective of the patient's experience in order to understand the effects of the hospitalist model on patient trust in their PCP and in their overall care. Studies by Davis et al.35 and Halpert et al.36 assert that rising quality of care and patient satisfaction with the hospitalist system rebuts coordination of care concerns. Yet we need more studies investigating the relationship between improved communication and patient outcomes, as evidence is currently conflicting on this subject.32, 37, 38

The Journal of Hospital Medicine has pursued this research agenda; the April 2009 issue presents several studies describing best practices in the discharging of hospitalized patients. Manning et al.39 describe a tool to assess patient mobility after discharge, and O'Leary et al.40 used electronic health records to create a better discharge summary. Project BOOST (Better Outcomes for Older Adults Through Safe Transitions) has shown improvements in discharge transition procedures41 and the use of transition coaches for vulnerable older patients has been proven cost‐effective and has been scaled up to more than 100 healthcare organizations.42, 43

Inpatient care handoff to PCPs is not entirely novel, as surgeons, oncologists, cardiologists, and other specialists have always grappled with continuity of care. It would be prudent to investigate what can be learned from these efforts, and which practices can be best applied to the hospitalist model. More longitudinal studies need to investigate the prevalence and success of the procedural recommendations to preserve the patient‐physician relationship. We need to know more about what works and what does not. How have hospitals found novel ways in implementing these approaches, and how can they be applied to a diversity of hospital settings? We need a better outcome measurement than patient or physician satisfaction for probing the subtleties of the patient‐physician relationship. There is a sizeable population that does not have a PCP to care for them before hospitalization or after discharge, and discussions about continuity of care must address these patients. Last, these best practices and patient centered values need to be incorporated into the core competencies of residencies and fellowships for a new generation of hospitalists.

Maintaining the continuity of the physician‐patient relationship is an integral part of the original premise of the hospitalist model. Importantly, Meltzer24 found that this discontinuity within the hospital has the potential to eliminate the savings of the hospitalist system. Yet concerns about continuity of care do not sufficiently encompass the complexand at times fragilerelationship between physician and patient. The survival of the physician‐patient relationship depends on the hospitalist model's affirmation of the values of coordination and Peabody's25 approach to patient‐centered care. If the hospitalist model is to thrive, it needs to emphasize its duty as steward of the PCP‐patient relationship as much as it focuses on efficiency and cost‐effectiveness.

Patient Autonomy

The mandatory transfer of patients into the hospitalist model raises serious ethical issues. A survey in 2000 of PCPs found that 23% were required to use hospitalists for all admissions.44 Other surveys found this prevalence to be as low as 2%.12 Nevertheless, several high profile cases of Health Maintenance Organizations (HMOs)Prudential HealthCareSouth Florida, Prudential, Humana, and Cigna Corporationall using mandatory hospitalists, prompted protests from professional organizations and there were even legislative efforts to ban the practice of the mandatory use of hospitalists in 2000 and 2001.45 Today, most insurance plans, as well as the Society of Hospital Medicine (SHM), support voluntary rather than mandatory hospitalist use.46 Yet while not mandatory, the hospitalist is the default provider in many settings, giving a de facto mandate for hospitalist care. As Royo et al.47 point out, the rise in physician employment by hospitals has facilitated a self‐selecting progression toward a structural network that closely resembles the mandatory model.

While PCPs and internists contested mandatory hospitalist plans as infringements on their autonomy, they overlooked the harm to the patient's autonomy. When healthy in the ambulatory setting, the patient has the opportunity to choose his or her doctor to provide longitudinal care. When the patient is admitted acutely to a hospital, the patient does not have the freedom to choose a physician; the patient is assigned to the hospitalist on duty that night. This call for patient autonomy is of utmost importance in the hospitalized patient, where patients are increasingly sicker, their diseases under a high‐powered lens, and their options diminished. This freedom of choice is integral to the patient‐physician partnership. Yet this freedom of choice is largely hindered by the employer's choice in the health plan for their employees or an individual's ability to pay for a health plan. These represent some of the many barriers to choice facing patients in the American model of health insurance.

As the hospitalist system grows to become the accepted mode of hospital care, more patients need to be informed about the transition of care to another physician and what steps are taken to ensure appropriate continuity of care. Transfers of patients from PCPs to hospitalists must be voluntary, with the decision left to patient care preferences.48 Educating patients in the outpatient setting about the hospitalist model, its benefits, risks, and alternatives, is necessary for them to make informed decisions about hospital care. This will require the collaboration of PCPs and hospitalists together. The continued success of the model depends on the nurturance of the partnership between the PCP, the hospitalist, and the patient.

Meltzer and Herthko14 have proposed that patients pay a premium for the option to choose a PCP that is not mandated to transfer their care to a hospitalist, in order to offset cost savings with the hospitalist system. Yet Meltzer and Herthko's14 study suggests that many patients could not afford to pay this premium and, in effect, patient autonomy would be preserved for the affluent. This raises the oft‐neglected professional ethic of justice for low‐income patients. Alexander and Lantos49 were resigned to see this infringement on patient autonomy as an inevitable consequence of balancing the desires of patients with the drive to lower cost and improve outcomes. If the hospitalist model grows to be the predominant mode of care, it is unclear if patient choice can survive. Investigators need to test whether the advantages of hospitalist care can coexist with voluntary programs. If it proves that they indeed cannot, then the hospitalist system will need to respond to concerned patients with honest answers and find pragmatic solutions to diminishing patient choice.

Conflict of Interest

The hospitalist system's main benefit of cost‐savings prompted Pantilat et al.6 to wonder whether hospitalists would face a conflict of interest between what is best for the patient and what financial incentives and utilization review encourage or require them to do. The financial support provided by many hospitals to meet the operating expenses of hospitalist programs is often associated with explicit or implicit incentives to reduce the length of hospital stay and costs.50 With hospitals employing hospitalists and increasingly pressuring them to decrease length of stay and discharge patients quickly, patients may have no advocate to protect them from discharge planners. Many hospitalists supplement their income by supervising discharge planners, and a dispute would put the hospitalist in the uncomfortable position of advocating for his patient against his employer and colleagues. While conflicts of interests occur in many managed care arrangements, they may be more acute in hospitalist systems. A weakened patient‐physician relationship may put the patients' best interest inferior to the employer's interests. Hospitalists do not immediately deal with adverse consequences of premature discharges in the outpatient setting and virtually no malpractice case law considers the obligations and practices of hospitalists in these settings.51

The SHM identified a core competency of hospitalists to

Ethical issues concerning conflict of interest remain unanswered, largely because no information about organizational features such as explicit incentives for reductions in length of stay is available to researchers or to patients. This is the wrong approach and only feeds the fear that hospitalists may weigh patients' best interest with financial incentives. Abbo and Volandes53 have argued that ambivalence to cost considerations is hazardous. If the hospitalist model cannot be forthright with the active considerations of costs in daily clinical practice, it is unlikely to truly make strides at cost savings, and may even raise the cost of care in the long run.

Jonsen et al.54 provide ethical standards for considering costs in clinical decisions. First, a physician's first priority should be to provide patient‐centered care that focuses on medical indications and patient preferences. Second, quality care does not mean all available care; quality care reflects what is not only diagnostically sound and technically correct, but also appropriate. Third, conflicts of interest are most vulnerable when there is a failing of the patient‐physician relationship. Health care organizations should expect physicians to argue for policies that provide all services that have a reasonable likelihood of benefiting the patient. Fourth, patient and physician autonomy and freedom of choice should be maximized within the limits of the system. Persons should be fully informed of the constraints of the system before choosing it. Plans need to disclose any financial incentive arrangements that exist between the plan and the physician. And incentive arrangements should be based on quality of care rather than on underutilization of care services. Fifth, the system should reflect principles of just distribution, ensuring that all who have a fair claim to service should receive it without discrimination. Last, capitation plans should share risks among physicians, not patients, while incentives are provided for improvements in access, prevention, and patient satisfaction.

Conflicts of interest have been a concern for as long as physicians have been paid for services. Fears about interference into the doctor‐patient relationship, whether they are from government or business, continue to stall real efforts to lower skyrocketing medical costs. The hospitalist model rebuts conflict of interest claims with improved outcomes, efficiency, and quality of care in the many reviews cited above. These arguments do prove that the hospitalist model's emphasis on medically indicated and appropriate care does address Jonsen et al.'s54 first and second standards. Yet, as Jonsen et al.54 point out, without strongly emphasizing the patient‐physician relationship and patient autonomy, it leaves itself vulnerable to creating conflicts of interest. Hospitalist systems need to be forthright about their explicit or implicit incentive structures and disclose this information to patients in a timely manner for them to make informed decisions. These incentives should be linked to quality of care and patient satisfaction, not cost savings. Last, hospitalist training programs should make ethical cost considerations a core competency of their curriculum.

Conclusions

Hospitalism was founded on the premise that it could improve the quality and reduce the cost of hospital care. Many randomized studies have all but definitively proven this original assertion. It is now time for the model to prove that these gains are not to the detriment of the patient‐physician relationship. Hospitalism must define itself as the steward of this relationship, valuing it as much as it values outcomes and costs. This is of particular concern in the United States as Medicare Part A (payment for inpatient care) is scheduled to go bankrupt in 2019, leading to potentially reasonable fears of hospital‐motivated cost containment.57

Investigators must find an outcome that encompasses the complexity of the patient‐physician relationship, and methods to improve it must be studied and improved upon. Preserving the patient‐physician relationship is a systemic issue, and full‐time hospitalists may be in the best position to implement systemic reforms to improve communication and continuity of care. Pham's56 case study of a hospitalist piecing together disparate parts of the patient's story illustrates this point. This should include more investigation into the prevalence of use and success of methods aimed at protecting the patient‐physician relationship at critical points in the handover of care. When proven successful, The SHM should propose new standards and safeguards to insure that these methods become standard practice in patient care. This effort, led by Snow et al.,57 is currently underway.

A hospitalist model that does not emphasize mitigating the effects of the diminishing patient‐physician relationship leaves itself exposed to further infringements on autonomy and choice. It is unclear whether patient autonomy and choice can coexist in a successful hospitalist system. The consequences of these unanswered ethical questions need to be explored. The professions of primary care need to be more proactive in educating patients about choice of care in hospitals, and hospitalists need to provide that choice, allowing voluntary programs in hospital care when feasible.

When combined, a wounded patient‐physician relationship and impaired patient autonomy leave the hospitalist model vulnerable to claims of financial and fiduciary conflict of interest. These concerns need not be inherent to the hospitalist systems, but hospitalists will need to be forthright and honest about incentives structures, and link them to quality of care and patient satisfaction, not to efficiency and cost savings.

It is indeed time for hospitalism to move onaway from proving its founding premise, and toward addressing these lingering ethical issues. Hospitalism's continued growth and success depends on it.

Wachter and Goldman1 first described hospitalists in 1996 as a new breed of physicians who devote blocks of time exclusively to the care of hospitalized patients. Since its definition, the hospitalist model has prompted 2 major debates. First, does the hospitalist system improve clinical efficiency, quality of care, cost effectiveness, and patient satisfaction? A series of large and small randomized trials have all but definitively proven the hospitalist model's advantage. Yet whether the hospitalist model is good for patient care has proven to remain contentious, as most recently demonstrated by the discussion between Williams2 and Centor3 and others like it.4, 5 What is clear in these exchanges is that the debate has shifted to the second great debate: does the hospitalist model pose inherent conflicts in clinical ethics? What are the implications of the purposeful discontinuity in care, the autonomy issues raised by mandatory hospitalist use, and the structural management issues that potentially pit hospitalists against patients in fiduciary and financial conflicts of interest? These important issues are certainly not new, and the hospitalist model has made much effort to address some of them.6, 7 This work aims to serve as a review of these important ethical concerns, demonstrating how some questions have been answered, while some remain unanswered.

The Hospitalist Model's Founding Premise

A growing threshold for hospital admission in the last 3 decades caused primary care physicians (PCPs) to see a diminishing number of inpatients. A survey in 1978 found that PCPs spent 40% of their time in the hospital, rounding on 10 patients per day.8 By 2001, PCPs spent 10% of their time in the hospital on average, and most PCPs rounded on fewer than 2 inpatients per day.9 The cost of inefficiencies associated with primary coordination of care in the hospital increasingly outweighed the tradeoff of preserving the patient‐PCP relationship in the hospital. Converging with increasing attention on cost controls through the restructuring of service provision, the hospitalist was born. Wachter10 argued that the hospitalist model could alleviate inpatient demands placed on PCPs while improving the outcomes and lowering the cost of care for hospitalized patients.

Early on there were setbacks to proving Wachter's10 case. Small studies found hospitalists to have higher hospital charges and longer length of stays.11 A survey of PCPs found only 56% were satisfied with communication with hospitalists and that most believed that patients generally preferred to be cared for in the hospital by their regular physician.12, 13 Meltzer and Herthko14 found 70% of people sampled said they would prefer care by their own physician to that of a hospitalist if they were hospitalized for a general medical condition. Yet this study found in a national random‐digit phone survey that only 10% of the respondents would pay $750 for their PCP to follow them to the hospital, the cost savings of the hospitalist system proven by the only 2 randomized trials performed at the time.15, 16 To 90% of respondents, the value of the PCP at the bedside was not worth the cost tradeoff to keep them there.

The meteoric rise in the number of hospitalists reflects the many studies and reviews that affirmed the premise that hospitalists improved inpatient efficiency without harmful effects on quality of care.17, 18 In a large retrospective cohort study of over 75,000 patients in 45 hospitals across the country, Lindenauer et al.19 found that hospitalists had a $268 lower cost when compared to internists, $125 lower cost when compared to family physicians, and a shorter hospital stay by about one‐half day when compared to both groups. The group found no significant difference in rates of death or readmission rates. While called modest in the text, these savings over time and volume add up for hospitals. Patients benefit from hospitalist care, researchers hypothesize, because of their familiarity with hospital systems, their increased availability to patients, and their experience with common hospital problems. Though the Lindenauer et al.19 study was criticized for design flaws, it prompted the editorialist McMahon20 to assert that the question was sufficiently answered, and it was time to move on away from the studies focusing on cost and comparing outcomes. As Wachter21 wrote, the demand for hospitalists is now relatively de‐linked from the field's original premiseefficiency advantagesand is now both more diversified and more robust. The model has become an accepted mode of care for hospitalized patients, with up to 20,000 hospitalists currently practicing in 29% of all hospitals and in over one‐half of hospitals with over 200 beds in the United States.22, 23

The Patient‐Physician Relationship

Purposeful discontinuity of care in the hospitalist system has the potential to diminish the doctor‐patient relationship.12 This relationship is built on a bond of loyalty, confidentiality, and trust. Handing off care to a hospitalist when the patient is most vulnerable can be viewed as a violation of this covenant. According to Meltzer,24 the hospitalist model pits Franicis Peabody's25 intimate personal relationship between patient and physician against Adam Smith et al.'s11 benefits of specialization. Peabody25 observed that physicians' lack of understanding of their patients as persons is especially acute in the hospital, where

This movement toward patient‐centered medicine fits into an ever‐growing sentiment to value the social as well as the physiological, a holistic approach to the patient as a person. This emphasis was the original justification for PCPs to coordinate increasingly specialized hospital care and translate recommendations suitable to patients. Can the long‐term relationship between patient and PCP be replaced by the hospital generalist, or would hospitalists be inherently deficient in their abilities to coordinate care appropriate for patients? Hospitalized patients are frequently in no position to make complex decisions regarding their care.26 Lo7 argues that PCPs who know patients over extended periods of time are in a better position to respect patient wishes by individualizing discussions with patients and checking that patients' decisions are consistent with their core values. The long‐term relationship is also critical for designing a complex discharge plan suitable to the patients' ability and resources. Information about long‐term patient compliance with medications is much more available to PCPs. Patients trust physicians to keep promises made concerning end‐of‐life issues, and these assurances are vulnerable during handoffs of care. Pantilat et al.6 provide a case study of an outpatient Do‐Not‐Resuscitate order ineffective in the hospital. These scenarios occur because most written advance directives are unavailable in acute situations, and when they are, hospitalists unfamiliar with the patient's wishes may hesitate to act on directives not specific enough to answer the acute clinical question.27

Hospitalists' broadened responsibility to systematically improve the care of patients may potentially improve end‐of‐life care. Patient values can be better communicated to hospitalists by encouraging inpatients to complete advance directive surveys and then asking hospitalists to discuss those directives with their patients.6 Significantly, Auerbach and Pantilat28 found that end‐of‐life care was improved with hospitalist care. This chart review found hospitalists more likely to have discussions with patients and their families regarding care and providing comfort care more frequently at the time of death than community‐based physicians. The authors hypothesize that hospitalists may have better communication with dying patients and their families because they spend more time in the hospital each day, using frequent meetings to better understand the preferences of patients. These preferences often require clarification and often change after admission, making previous discussions about end‐of‐life care with PCPs moot. Greater expertise in hospital care may also allow hospitalists to better recognize patients who are nearing death and may explain the fewer symptoms documented by Auerbach and Pantilat28 at the end of life among patients cared for by hospitalists compared to community‐based physicians.

Hospital medicine has taken continuity of care issues seriously, and responded by making pragmatic recommendations to preserve the patient‐PCP relationship in the hospital and assuage the perception that patients have been dropped. Harlan et al.29 identify important issues around good communication between pediatric hospitalists and PCPs including the content and timing of communication beneficial to the patient. Hospitalists can use a standard script for introducing themselves to patients, explaining their role, and their continued coordination with the PCP.30 PCPs can still be involved in the care of their patients in hospitals through continuity visits or phone calls with patients and through better communication with hospitalists.31 Generally, reimbursing PCPs for their increased role in the hospitalist system can encourage better communication with hospitalists.19 Potential disagreements between PCPs and hospitalist regarding the care of the patient can be resolved through explicit conflict resolution procedures within the hospitalist system.6

These procedural solutions are only as successful as they are used. A large review by Kripalani et al.32 found communication between hospitalists and PCPs occurred infrequently (3%‐20%), affecting the quality of care in approximately 25% of follow‐up visits and contributing to PCP dissatisfaction. Sharma et al.33 found that continuity visits decreased from 50.5% in 1996 to 39.8% in 2006. In a survey of patients cared for in a hospitalist system, Hruby et al.34 found that 33% of hospitalized patients had some contact with their PCP directly and 66% of patients were satisfied with the contact they or their relative had with their PCP. When probed, patient satisfaction is too vague a measurement to assess the complex value of the patient‐physician relationship. Studying these issues may require relying more on individualized narratives rather than generalized statistics, or may require years of follow‐up. As Centor3 argues, we need this broader perspective of the patient's experience in order to understand the effects of the hospitalist model on patient trust in their PCP and in their overall care. Studies by Davis et al.35 and Halpert et al.36 assert that rising quality of care and patient satisfaction with the hospitalist system rebuts coordination of care concerns. Yet we need more studies investigating the relationship between improved communication and patient outcomes, as evidence is currently conflicting on this subject.32, 37, 38

The Journal of Hospital Medicine has pursued this research agenda; the April 2009 issue presents several studies describing best practices in the discharging of hospitalized patients. Manning et al.39 describe a tool to assess patient mobility after discharge, and O'Leary et al.40 used electronic health records to create a better discharge summary. Project BOOST (Better Outcomes for Older Adults Through Safe Transitions) has shown improvements in discharge transition procedures41 and the use of transition coaches for vulnerable older patients has been proven cost‐effective and has been scaled up to more than 100 healthcare organizations.42, 43

Inpatient care handoff to PCPs is not entirely novel, as surgeons, oncologists, cardiologists, and other specialists have always grappled with continuity of care. It would be prudent to investigate what can be learned from these efforts, and which practices can be best applied to the hospitalist model. More longitudinal studies need to investigate the prevalence and success of the procedural recommendations to preserve the patient‐physician relationship. We need to know more about what works and what does not. How have hospitals found novel ways in implementing these approaches, and how can they be applied to a diversity of hospital settings? We need a better outcome measurement than patient or physician satisfaction for probing the subtleties of the patient‐physician relationship. There is a sizeable population that does not have a PCP to care for them before hospitalization or after discharge, and discussions about continuity of care must address these patients. Last, these best practices and patient centered values need to be incorporated into the core competencies of residencies and fellowships for a new generation of hospitalists.

Maintaining the continuity of the physician‐patient relationship is an integral part of the original premise of the hospitalist model. Importantly, Meltzer24 found that this discontinuity within the hospital has the potential to eliminate the savings of the hospitalist system. Yet concerns about continuity of care do not sufficiently encompass the complexand at times fragilerelationship between physician and patient. The survival of the physician‐patient relationship depends on the hospitalist model's affirmation of the values of coordination and Peabody's25 approach to patient‐centered care. If the hospitalist model is to thrive, it needs to emphasize its duty as steward of the PCP‐patient relationship as much as it focuses on efficiency and cost‐effectiveness.

Patient Autonomy

The mandatory transfer of patients into the hospitalist model raises serious ethical issues. A survey in 2000 of PCPs found that 23% were required to use hospitalists for all admissions.44 Other surveys found this prevalence to be as low as 2%.12 Nevertheless, several high profile cases of Health Maintenance Organizations (HMOs)Prudential HealthCareSouth Florida, Prudential, Humana, and Cigna Corporationall using mandatory hospitalists, prompted protests from professional organizations and there were even legislative efforts to ban the practice of the mandatory use of hospitalists in 2000 and 2001.45 Today, most insurance plans, as well as the Society of Hospital Medicine (SHM), support voluntary rather than mandatory hospitalist use.46 Yet while not mandatory, the hospitalist is the default provider in many settings, giving a de facto mandate for hospitalist care. As Royo et al.47 point out, the rise in physician employment by hospitals has facilitated a self‐selecting progression toward a structural network that closely resembles the mandatory model.

While PCPs and internists contested mandatory hospitalist plans as infringements on their autonomy, they overlooked the harm to the patient's autonomy. When healthy in the ambulatory setting, the patient has the opportunity to choose his or her doctor to provide longitudinal care. When the patient is admitted acutely to a hospital, the patient does not have the freedom to choose a physician; the patient is assigned to the hospitalist on duty that night. This call for patient autonomy is of utmost importance in the hospitalized patient, where patients are increasingly sicker, their diseases under a high‐powered lens, and their options diminished. This freedom of choice is integral to the patient‐physician partnership. Yet this freedom of choice is largely hindered by the employer's choice in the health plan for their employees or an individual's ability to pay for a health plan. These represent some of the many barriers to choice facing patients in the American model of health insurance.

As the hospitalist system grows to become the accepted mode of hospital care, more patients need to be informed about the transition of care to another physician and what steps are taken to ensure appropriate continuity of care. Transfers of patients from PCPs to hospitalists must be voluntary, with the decision left to patient care preferences.48 Educating patients in the outpatient setting about the hospitalist model, its benefits, risks, and alternatives, is necessary for them to make informed decisions about hospital care. This will require the collaboration of PCPs and hospitalists together. The continued success of the model depends on the nurturance of the partnership between the PCP, the hospitalist, and the patient.

Meltzer and Herthko14 have proposed that patients pay a premium for the option to choose a PCP that is not mandated to transfer their care to a hospitalist, in order to offset cost savings with the hospitalist system. Yet Meltzer and Herthko's14 study suggests that many patients could not afford to pay this premium and, in effect, patient autonomy would be preserved for the affluent. This raises the oft‐neglected professional ethic of justice for low‐income patients. Alexander and Lantos49 were resigned to see this infringement on patient autonomy as an inevitable consequence of balancing the desires of patients with the drive to lower cost and improve outcomes. If the hospitalist model grows to be the predominant mode of care, it is unclear if patient choice can survive. Investigators need to test whether the advantages of hospitalist care can coexist with voluntary programs. If it proves that they indeed cannot, then the hospitalist system will need to respond to concerned patients with honest answers and find pragmatic solutions to diminishing patient choice.

Conflict of Interest

The hospitalist system's main benefit of cost‐savings prompted Pantilat et al.6 to wonder whether hospitalists would face a conflict of interest between what is best for the patient and what financial incentives and utilization review encourage or require them to do. The financial support provided by many hospitals to meet the operating expenses of hospitalist programs is often associated with explicit or implicit incentives to reduce the length of hospital stay and costs.50 With hospitals employing hospitalists and increasingly pressuring them to decrease length of stay and discharge patients quickly, patients may have no advocate to protect them from discharge planners. Many hospitalists supplement their income by supervising discharge planners, and a dispute would put the hospitalist in the uncomfortable position of advocating for his patient against his employer and colleagues. While conflicts of interests occur in many managed care arrangements, they may be more acute in hospitalist systems. A weakened patient‐physician relationship may put the patients' best interest inferior to the employer's interests. Hospitalists do not immediately deal with adverse consequences of premature discharges in the outpatient setting and virtually no malpractice case law considers the obligations and practices of hospitalists in these settings.51

The SHM identified a core competency of hospitalists to

Ethical issues concerning conflict of interest remain unanswered, largely because no information about organizational features such as explicit incentives for reductions in length of stay is available to researchers or to patients. This is the wrong approach and only feeds the fear that hospitalists may weigh patients' best interest with financial incentives. Abbo and Volandes53 have argued that ambivalence to cost considerations is hazardous. If the hospitalist model cannot be forthright with the active considerations of costs in daily clinical practice, it is unlikely to truly make strides at cost savings, and may even raise the cost of care in the long run.

Jonsen et al.54 provide ethical standards for considering costs in clinical decisions. First, a physician's first priority should be to provide patient‐centered care that focuses on medical indications and patient preferences. Second, quality care does not mean all available care; quality care reflects what is not only diagnostically sound and technically correct, but also appropriate. Third, conflicts of interest are most vulnerable when there is a failing of the patient‐physician relationship. Health care organizations should expect physicians to argue for policies that provide all services that have a reasonable likelihood of benefiting the patient. Fourth, patient and physician autonomy and freedom of choice should be maximized within the limits of the system. Persons should be fully informed of the constraints of the system before choosing it. Plans need to disclose any financial incentive arrangements that exist between the plan and the physician. And incentive arrangements should be based on quality of care rather than on underutilization of care services. Fifth, the system should reflect principles of just distribution, ensuring that all who have a fair claim to service should receive it without discrimination. Last, capitation plans should share risks among physicians, not patients, while incentives are provided for improvements in access, prevention, and patient satisfaction.

Conflicts of interest have been a concern for as long as physicians have been paid for services. Fears about interference into the doctor‐patient relationship, whether they are from government or business, continue to stall real efforts to lower skyrocketing medical costs. The hospitalist model rebuts conflict of interest claims with improved outcomes, efficiency, and quality of care in the many reviews cited above. These arguments do prove that the hospitalist model's emphasis on medically indicated and appropriate care does address Jonsen et al.'s54 first and second standards. Yet, as Jonsen et al.54 point out, without strongly emphasizing the patient‐physician relationship and patient autonomy, it leaves itself vulnerable to creating conflicts of interest. Hospitalist systems need to be forthright about their explicit or implicit incentive structures and disclose this information to patients in a timely manner for them to make informed decisions. These incentives should be linked to quality of care and patient satisfaction, not cost savings. Last, hospitalist training programs should make ethical cost considerations a core competency of their curriculum.

Conclusions

Hospitalism was founded on the premise that it could improve the quality and reduce the cost of hospital care. Many randomized studies have all but definitively proven this original assertion. It is now time for the model to prove that these gains are not to the detriment of the patient‐physician relationship. Hospitalism must define itself as the steward of this relationship, valuing it as much as it values outcomes and costs. This is of particular concern in the United States as Medicare Part A (payment for inpatient care) is scheduled to go bankrupt in 2019, leading to potentially reasonable fears of hospital‐motivated cost containment.57

Investigators must find an outcome that encompasses the complexity of the patient‐physician relationship, and methods to improve it must be studied and improved upon. Preserving the patient‐physician relationship is a systemic issue, and full‐time hospitalists may be in the best position to implement systemic reforms to improve communication and continuity of care. Pham's56 case study of a hospitalist piecing together disparate parts of the patient's story illustrates this point. This should include more investigation into the prevalence of use and success of methods aimed at protecting the patient‐physician relationship at critical points in the handover of care. When proven successful, The SHM should propose new standards and safeguards to insure that these methods become standard practice in patient care. This effort, led by Snow et al.,57 is currently underway.

A hospitalist model that does not emphasize mitigating the effects of the diminishing patient‐physician relationship leaves itself exposed to further infringements on autonomy and choice. It is unclear whether patient autonomy and choice can coexist in a successful hospitalist system. The consequences of these unanswered ethical questions need to be explored. The professions of primary care need to be more proactive in educating patients about choice of care in hospitals, and hospitalists need to provide that choice, allowing voluntary programs in hospital care when feasible.

When combined, a wounded patient‐physician relationship and impaired patient autonomy leave the hospitalist model vulnerable to claims of financial and fiduciary conflict of interest. These concerns need not be inherent to the hospitalist systems, but hospitalists will need to be forthright and honest about incentives structures, and link them to quality of care and patient satisfaction, not to efficiency and cost savings.

It is indeed time for hospitalism to move onaway from proving its founding premise, and toward addressing these lingering ethical issues. Hospitalism's continued growth and success depends on it.

Hospitalizations often impose a tremendous financial burden on patients and their families, adding to the stress and long‐term impact from medical illnesses. It is widely underappreciated that physicians can play an important role in substantially reducing patients' out‐of‐pocket expenses by participating in hospital‐based case review and utilization management. These topics are not a focus of most formal training curricula and unfortunately are typically viewed by medical staff as intrusive, time consuming, or only in terms of enhancing the facility's profitability. In reality, with strict rules governing insurance benefits the facility's interests are typically aligned with those of the patient.

One of the greatest impacts on a patient's financial liability is whether an admission is classified in observation vs. inpatient status, and is subject to much confusion. It is a common misperception that these are time‐based designations. Instead, they revolve around stringent medical necessity guidelines that examine the severity of the illness and the intensity of services provided.1 Inpatient stays may be brief, even a single day, if justified by medical need (although these short durations are closely scrutinized by the payors) or if involving a short list of procedures automatically triggering that status (ie, defibrillator placement).2 Conversely, observation status, although usually up to only 48 hours, can extend longer if inpatient criteria are never met and are then apt to generate large bills.

The key concept for the financial liability of patients in observation status is that their billing structure revolves around being categorized as outpatients, even though they stay overnight and are physically housed and cared for in the expensive hospital setting.3 This nonintuitive classification can culminate in unexpectedly high charges for which the patient is liable (Table 1): medications at inflated hospital pharmacy prices, especially when expensive antibiotics or immunosuppressive agents are administered (since outpatient prescriptions are not often covered by policies); ancillary services, radiology or laboratory tests with a high patient share of cost; and an hourly room charge that can easily exceed $30 per hour. The latter can be especially burdensome, as most insurance plans only cover the first 48 hours of observation. During that period the patient would be liable for just their copayment, but afterward they could be billed for the full amount. Hospitalizations well beyond the 48 hours can thus present tremendous hardships to those patients who never meet the stringent criteria for categorization as inpatients, and whose status thus must remain outpatient‐observation. Keeping patients over a weekend for procedures that are not available at the facility until the following Monday can put these individuals beyond the 48 hour observation interval and cause unintentional rapidly escalating out‐of‐pocket expenses. Other strategies to reduce the patient's financial liability include allowing patients to take their own medications from home (with pharmacy supervision and verification, per hospital guidelines), and limiting evaluations to just the admitting diagnosis (ie, pursuing other issues after discharge). In addition, an observation stay can never be ordered ahead of time for an outpatient procedure, as that type of admission is reserved for those individuals who unexpectedly need further care at the conclusion of the recovery period (typically 4 to 6 hours). Thus, the not uncommon practice of doing a patient a favor by letting them stay overnight after an outpatient procedure thereby can be a great disservice by dramatically increasing patient liability. One can well imagine how these scenarios lead to lay press exposs of the patient receiving a bill for a $25 aspirin and a night's stay 4 times more expensive than a luxury hotel. This is not to say that going home is the best or safest plan for a particular patient, but rather that the hospital is often an unnecessarily expensive (and in that sense inappropriate) location when there are alternatives. It is up to the individual hospital to determine how to handle rapidly escalating charges related to the admission status and the timeliness of a discharge. Many centers in effect write off highly select bills that are considered either uncollectible (ie, from indigent patients) or the fault of the facility's inefficiencies. So as not to have inconsistent billing policies across different insurers and patients, however, facilities are obligated to have uniform protocols for attempting to collect chargesa scenario that can be quite harsh for those individuals with significant and discoverable monetary resources.

Working with the physician for a timely discharge, hospital case managers and social workers are likely to arrive at creative solutions in the patient's best financial interest (ie, taxicab coupons and inexpensive hotels). As many patients simply do not have the resources to cope with unplanned overnight charges, it behooves the physician to make every effort to start outpatient procedures early in the day so as to minimize the chance of logistic problems triggering a potentially expensive overnight hospital stay.

Compare the observation patient's liability to that of the typically much‐preferred status of inpatient (Table 1) in which all expenses are rolled into one diagnosis‐related group (DRG) prospective payment.3 In the case of Medicare, the patient's bill would be the inpatient deductible, and this might be covered in its entirety by a supplemental policy. One absolutely cannot, however, simply avoid using the observation status and instead make all admissions inpatients; this would cause unnecessary resource utilization and expose the hospital to denial of payment for the entire episode of care. To prevent this situation, there are nationally‐recognized guidelines that strictly define when a hospitalization warrants an inpatient level of care. Integral to the individual qualifying for their policy's inpatient benefit, however, is that the chart must reflect not just the severity of illness but also intensity of services ordered by the physician.1 Similarly, changing a patient's status (ie, from observation to inpatient) must follow rigorous guidelines wherein the justification and timing are fully described in the body of the chart to an extent that would withstand audit.

Consider the example (Table 2) of a patient with a leg fracture admitted for pain due to edema and early compartment syndrome: a scenario appropriate for inpatient status, liability of just the $1092 Medicare deductible, and eligibility for postdischarge skilled nursing facility care. Had the charting erroneously only indicated pain and need for a new cast, then observation status would have yielded a bill for $3426, plus out of pocket nursing home expenses of over $150/day.

Not only does the physician need to accurately chart the reasons for admission, but it is also extremely helpful to specifically document why the patient is not amenable to outpatient therapy. Examples include a clearly articulated history of failed attempts at home or emergency room treatment, or the need for close monitoring (ie, telemetry). In this regard case managers also provide a fresh set of eyes to evaluate the clarity and completeness of medical charting. What seems like obvious decision‐making to a physician may require expanded detailed notes to satisfy a third‐party review.

The work design of the case managers and utilization review team varies between facilities. Ideally, cases are reviewed upon admission (or within the first 24 hours), and then periodically thereafter. Many medical centers have this process computerized, wherein inpatient criteria are available online and status issues can be tracked daily. This nearly real‐time information serves as the basis for interacting with the attending physician, and is necessary because the chart documentation may not be amended after discharge. Having a robust database for all admissions is also immensely helpful in those hospitals which employ a Physician Advisor (PA) as a liaison and educator to the medical staff. This newly and now nationally recognized PA position serves an important role in educating the providers not just about these patient advocacy topics, but also other issues such as length‐of‐stay. Interestingly, having the infrastructure of a criteria‐driven database to follow the intensity of inpatient services on a daily basis gives case managers an objective perspective of when a patient requires less care and is ready for transfer to a lower acuity facility or discharge home. Physician participation is important when the patient thus runs out of intensities, since there will need to be early coordination of efforts for home health or skilled nursing care, durable medical equipment supplies, or outpatient infusions. It is important that physicians not view these activities as an inappropriate rush for discharge. In our experience most patients are in fact much happier to be out of the hospital and receiving home or skilled nursing care. Those in need of physical or occupational therapy may in fact have superior care in facilities dedicated to those activities. In addition, unnecessarily prolonged hospitalizations carry their own risks, such as hospital‐acquired infections and deep venous thromboses. An additional motivator for discharge is that, just as there are insurance plan limits for outpatient benefits, there can also be caps for inpatient services. Physicians thus have a role in preserving the limited and precious number of covered inpatient days of care, beyond which time the patient would be financially totally responsible. For example, most states limit the number of inpatient days covered by Medicaid. In Florida there is a cap of only 45 days per year (unless the patient is pediatric or within the first year of a transplant4). Similarly, there have been patients and families shocked and ill‐prepared to discover that all their Medicare hospital benefits were exhausted5: a not well‐publicized possibility, as in the setting of expensive intensive care units, transplantation, or chemotherapy. Timely discharges and careful resource utilization by physicians thus not only help the hospital but also are important for the patient.

In summary, physicians need to be aware that there can be tremendous financial hardship to patients caused by inappropriate or unnecessarily long observation stays, especially in cases where an inpatient designation would have been justified by appropriate documentation. Case managers, although employed by the facility, can thus assist physicians in this regard and together play an important role as patient advocates.

Hospitalizations often impose a tremendous financial burden on patients and their families, adding to the stress and long‐term impact from medical illnesses. It is widely underappreciated that physicians can play an important role in substantially reducing patients' out‐of‐pocket expenses by participating in hospital‐based case review and utilization management. These topics are not a focus of most formal training curricula and unfortunately are typically viewed by medical staff as intrusive, time consuming, or only in terms of enhancing the facility's profitability. In reality, with strict rules governing insurance benefits the facility's interests are typically aligned with those of the patient.

One of the greatest impacts on a patient's financial liability is whether an admission is classified in observation vs. inpatient status, and is subject to much confusion. It is a common misperception that these are time‐based designations. Instead, they revolve around stringent medical necessity guidelines that examine the severity of the illness and the intensity of services provided.1 Inpatient stays may be brief, even a single day, if justified by medical need (although these short durations are closely scrutinized by the payors) or if involving a short list of procedures automatically triggering that status (ie, defibrillator placement).2 Conversely, observation status, although usually up to only 48 hours, can extend longer if inpatient criteria are never met and are then apt to generate large bills.

The key concept for the financial liability of patients in observation status is that their billing structure revolves around being categorized as outpatients, even though they stay overnight and are physically housed and cared for in the expensive hospital setting.3 This nonintuitive classification can culminate in unexpectedly high charges for which the patient is liable (Table 1): medications at inflated hospital pharmacy prices, especially when expensive antibiotics or immunosuppressive agents are administered (since outpatient prescriptions are not often covered by policies); ancillary services, radiology or laboratory tests with a high patient share of cost; and an hourly room charge that can easily exceed $30 per hour. The latter can be especially burdensome, as most insurance plans only cover the first 48 hours of observation. During that period the patient would be liable for just their copayment, but afterward they could be billed for the full amount. Hospitalizations well beyond the 48 hours can thus present tremendous hardships to those patients who never meet the stringent criteria for categorization as inpatients, and whose status thus must remain outpatient‐observation. Keeping patients over a weekend for procedures that are not available at the facility until the following Monday can put these individuals beyond the 48 hour observation interval and cause unintentional rapidly escalating out‐of‐pocket expenses. Other strategies to reduce the patient's financial liability include allowing patients to take their own medications from home (with pharmacy supervision and verification, per hospital guidelines), and limiting evaluations to just the admitting diagnosis (ie, pursuing other issues after discharge). In addition, an observation stay can never be ordered ahead of time for an outpatient procedure, as that type of admission is reserved for those individuals who unexpectedly need further care at the conclusion of the recovery period (typically 4 to 6 hours). Thus, the not uncommon practice of doing a patient a favor by letting them stay overnight after an outpatient procedure thereby can be a great disservice by dramatically increasing patient liability. One can well imagine how these scenarios lead to lay press exposs of the patient receiving a bill for a $25 aspirin and a night's stay 4 times more expensive than a luxury hotel. This is not to say that going home is the best or safest plan for a particular patient, but rather that the hospital is often an unnecessarily expensive (and in that sense inappropriate) location when there are alternatives. It is up to the individual hospital to determine how to handle rapidly escalating charges related to the admission status and the timeliness of a discharge. Many centers in effect write off highly select bills that are considered either uncollectible (ie, from indigent patients) or the fault of the facility's inefficiencies. So as not to have inconsistent billing policies across different insurers and patients, however, facilities are obligated to have uniform protocols for attempting to collect chargesa scenario that can be quite harsh for those individuals with significant and discoverable monetary resources.

Working with the physician for a timely discharge, hospital case managers and social workers are likely to arrive at creative solutions in the patient's best financial interest (ie, taxicab coupons and inexpensive hotels). As many patients simply do not have the resources to cope with unplanned overnight charges, it behooves the physician to make every effort to start outpatient procedures early in the day so as to minimize the chance of logistic problems triggering a potentially expensive overnight hospital stay.

Compare the observation patient's liability to that of the typically much‐preferred status of inpatient (Table 1) in which all expenses are rolled into one diagnosis‐related group (DRG) prospective payment.3 In the case of Medicare, the patient's bill would be the inpatient deductible, and this might be covered in its entirety by a supplemental policy. One absolutely cannot, however, simply avoid using the observation status and instead make all admissions inpatients; this would cause unnecessary resource utilization and expose the hospital to denial of payment for the entire episode of care. To prevent this situation, there are nationally‐recognized guidelines that strictly define when a hospitalization warrants an inpatient level of care. Integral to the individual qualifying for their policy's inpatient benefit, however, is that the chart must reflect not just the severity of illness but also intensity of services ordered by the physician.1 Similarly, changing a patient's status (ie, from observation to inpatient) must follow rigorous guidelines wherein the justification and timing are fully described in the body of the chart to an extent that would withstand audit.

Consider the example (Table 2) of a patient with a leg fracture admitted for pain due to edema and early compartment syndrome: a scenario appropriate for inpatient status, liability of just the $1092 Medicare deductible, and eligibility for postdischarge skilled nursing facility care. Had the charting erroneously only indicated pain and need for a new cast, then observation status would have yielded a bill for $3426, plus out of pocket nursing home expenses of over $150/day.

Not only does the physician need to accurately chart the reasons for admission, but it is also extremely helpful to specifically document why the patient is not amenable to outpatient therapy. Examples include a clearly articulated history of failed attempts at home or emergency room treatment, or the need for close monitoring (ie, telemetry). In this regard case managers also provide a fresh set of eyes to evaluate the clarity and completeness of medical charting. What seems like obvious decision‐making to a physician may require expanded detailed notes to satisfy a third‐party review.

The work design of the case managers and utilization review team varies between facilities. Ideally, cases are reviewed upon admission (or within the first 24 hours), and then periodically thereafter. Many medical centers have this process computerized, wherein inpatient criteria are available online and status issues can be tracked daily. This nearly real‐time information serves as the basis for interacting with the attending physician, and is necessary because the chart documentation may not be amended after discharge. Having a robust database for all admissions is also immensely helpful in those hospitals which employ a Physician Advisor (PA) as a liaison and educator to the medical staff. This newly and now nationally recognized PA position serves an important role in educating the providers not just about these patient advocacy topics, but also other issues such as length‐of‐stay. Interestingly, having the infrastructure of a criteria‐driven database to follow the intensity of inpatient services on a daily basis gives case managers an objective perspective of when a patient requires less care and is ready for transfer to a lower acuity facility or discharge home. Physician participation is important when the patient thus runs out of intensities, since there will need to be early coordination of efforts for home health or skilled nursing care, durable medical equipment supplies, or outpatient infusions. It is important that physicians not view these activities as an inappropriate rush for discharge. In our experience most patients are in fact much happier to be out of the hospital and receiving home or skilled nursing care. Those in need of physical or occupational therapy may in fact have superior care in facilities dedicated to those activities. In addition, unnecessarily prolonged hospitalizations carry their own risks, such as hospital‐acquired infections and deep venous thromboses. An additional motivator for discharge is that, just as there are insurance plan limits for outpatient benefits, there can also be caps for inpatient services. Physicians thus have a role in preserving the limited and precious number of covered inpatient days of care, beyond which time the patient would be financially totally responsible. For example, most states limit the number of inpatient days covered by Medicaid. In Florida there is a cap of only 45 days per year (unless the patient is pediatric or within the first year of a transplant4). Similarly, there have been patients and families shocked and ill‐prepared to discover that all their Medicare hospital benefits were exhausted5: a not well‐publicized possibility, as in the setting of expensive intensive care units, transplantation, or chemotherapy. Timely discharges and careful resource utilization by physicians thus not only help the hospital but also are important for the patient.

In summary, physicians need to be aware that there can be tremendous financial hardship to patients caused by inappropriate or unnecessarily long observation stays, especially in cases where an inpatient designation would have been justified by appropriate documentation. Case managers, although employed by the facility, can thus assist physicians in this regard and together play an important role as patient advocates.

Hospitalizations often impose a tremendous financial burden on patients and their families, adding to the stress and long‐term impact from medical illnesses. It is widely underappreciated that physicians can play an important role in substantially reducing patients' out‐of‐pocket expenses by participating in hospital‐based case review and utilization management. These topics are not a focus of most formal training curricula and unfortunately are typically viewed by medical staff as intrusive, time consuming, or only in terms of enhancing the facility's profitability. In reality, with strict rules governing insurance benefits the facility's interests are typically aligned with those of the patient.

One of the greatest impacts on a patient's financial liability is whether an admission is classified in observation vs. inpatient status, and is subject to much confusion. It is a common misperception that these are time‐based designations. Instead, they revolve around stringent medical necessity guidelines that examine the severity of the illness and the intensity of services provided.1 Inpatient stays may be brief, even a single day, if justified by medical need (although these short durations are closely scrutinized by the payors) or if involving a short list of procedures automatically triggering that status (ie, defibrillator placement).2 Conversely, observation status, although usually up to only 48 hours, can extend longer if inpatient criteria are never met and are then apt to generate large bills.

The key concept for the financial liability of patients in observation status is that their billing structure revolves around being categorized as outpatients, even though they stay overnight and are physically housed and cared for in the expensive hospital setting.3 This nonintuitive classification can culminate in unexpectedly high charges for which the patient is liable (Table 1): medications at inflated hospital pharmacy prices, especially when expensive antibiotics or immunosuppressive agents are administered (since outpatient prescriptions are not often covered by policies); ancillary services, radiology or laboratory tests with a high patient share of cost; and an hourly room charge that can easily exceed $30 per hour. The latter can be especially burdensome, as most insurance plans only cover the first 48 hours of observation. During that period the patient would be liable for just their copayment, but afterward they could be billed for the full amount. Hospitalizations well beyond the 48 hours can thus present tremendous hardships to those patients who never meet the stringent criteria for categorization as inpatients, and whose status thus must remain outpatient‐observation. Keeping patients over a weekend for procedures that are not available at the facility until the following Monday can put these individuals beyond the 48 hour observation interval and cause unintentional rapidly escalating out‐of‐pocket expenses. Other strategies to reduce the patient's financial liability include allowing patients to take their own medications from home (with pharmacy supervision and verification, per hospital guidelines), and limiting evaluations to just the admitting diagnosis (ie, pursuing other issues after discharge). In addition, an observation stay can never be ordered ahead of time for an outpatient procedure, as that type of admission is reserved for those individuals who unexpectedly need further care at the conclusion of the recovery period (typically 4 to 6 hours). Thus, the not uncommon practice of doing a patient a favor by letting them stay overnight after an outpatient procedure thereby can be a great disservice by dramatically increasing patient liability. One can well imagine how these scenarios lead to lay press exposs of the patient receiving a bill for a $25 aspirin and a night's stay 4 times more expensive than a luxury hotel. This is not to say that going home is the best or safest plan for a particular patient, but rather that the hospital is often an unnecessarily expensive (and in that sense inappropriate) location when there are alternatives. It is up to the individual hospital to determine how to handle rapidly escalating charges related to the admission status and the timeliness of a discharge. Many centers in effect write off highly select bills that are considered either uncollectible (ie, from indigent patients) or the fault of the facility's inefficiencies. So as not to have inconsistent billing policies across different insurers and patients, however, facilities are obligated to have uniform protocols for attempting to collect chargesa scenario that can be quite harsh for those individuals with significant and discoverable monetary resources.

Working with the physician for a timely discharge, hospital case managers and social workers are likely to arrive at creative solutions in the patient's best financial interest (ie, taxicab coupons and inexpensive hotels). As many patients simply do not have the resources to cope with unplanned overnight charges, it behooves the physician to make every effort to start outpatient procedures early in the day so as to minimize the chance of logistic problems triggering a potentially expensive overnight hospital stay.

Compare the observation patient's liability to that of the typically much‐preferred status of inpatient (Table 1) in which all expenses are rolled into one diagnosis‐related group (DRG) prospective payment.3 In the case of Medicare, the patient's bill would be the inpatient deductible, and this might be covered in its entirety by a supplemental policy. One absolutely cannot, however, simply avoid using the observation status and instead make all admissions inpatients; this would cause unnecessary resource utilization and expose the hospital to denial of payment for the entire episode of care. To prevent this situation, there are nationally‐recognized guidelines that strictly define when a hospitalization warrants an inpatient level of care. Integral to the individual qualifying for their policy's inpatient benefit, however, is that the chart must reflect not just the severity of illness but also intensity of services ordered by the physician.1 Similarly, changing a patient's status (ie, from observation to inpatient) must follow rigorous guidelines wherein the justification and timing are fully described in the body of the chart to an extent that would withstand audit.

Consider the example (Table 2) of a patient with a leg fracture admitted for pain due to edema and early compartment syndrome: a scenario appropriate for inpatient status, liability of just the $1092 Medicare deductible, and eligibility for postdischarge skilled nursing facility care. Had the charting erroneously only indicated pain and need for a new cast, then observation status would have yielded a bill for $3426, plus out of pocket nursing home expenses of over $150/day.

Not only does the physician need to accurately chart the reasons for admission, but it is also extremely helpful to specifically document why the patient is not amenable to outpatient therapy. Examples include a clearly articulated history of failed attempts at home or emergency room treatment, or the need for close monitoring (ie, telemetry). In this regard case managers also provide a fresh set of eyes to evaluate the clarity and completeness of medical charting. What seems like obvious decision‐making to a physician may require expanded detailed notes to satisfy a third‐party review.

The work design of the case managers and utilization review team varies between facilities. Ideally, cases are reviewed upon admission (or within the first 24 hours), and then periodically thereafter. Many medical centers have this process computerized, wherein inpatient criteria are available online and status issues can be tracked daily. This nearly real‐time information serves as the basis for interacting with the attending physician, and is necessary because the chart documentation may not be amended after discharge. Having a robust database for all admissions is also immensely helpful in those hospitals which employ a Physician Advisor (PA) as a liaison and educator to the medical staff. This newly and now nationally recognized PA position serves an important role in educating the providers not just about these patient advocacy topics, but also other issues such as length‐of‐stay. Interestingly, having the infrastructure of a criteria‐driven database to follow the intensity of inpatient services on a daily basis gives case managers an objective perspective of when a patient requires less care and is ready for transfer to a lower acuity facility or discharge home. Physician participation is important when the patient thus runs out of intensities, since there will need to be early coordination of efforts for home health or skilled nursing care, durable medical equipment supplies, or outpatient infusions. It is important that physicians not view these activities as an inappropriate rush for discharge. In our experience most patients are in fact much happier to be out of the hospital and receiving home or skilled nursing care. Those in need of physical or occupational therapy may in fact have superior care in facilities dedicated to those activities. In addition, unnecessarily prolonged hospitalizations carry their own risks, such as hospital‐acquired infections and deep venous thromboses. An additional motivator for discharge is that, just as there are insurance plan limits for outpatient benefits, there can also be caps for inpatient services. Physicians thus have a role in preserving the limited and precious number of covered inpatient days of care, beyond which time the patient would be financially totally responsible. For example, most states limit the number of inpatient days covered by Medicaid. In Florida there is a cap of only 45 days per year (unless the patient is pediatric or within the first year of a transplant4). Similarly, there have been patients and families shocked and ill‐prepared to discover that all their Medicare hospital benefits were exhausted5: a not well‐publicized possibility, as in the setting of expensive intensive care units, transplantation, or chemotherapy. Timely discharges and careful resource utilization by physicians thus not only help the hospital but also are important for the patient.

In summary, physicians need to be aware that there can be tremendous financial hardship to patients caused by inappropriate or unnecessarily long observation stays, especially in cases where an inpatient designation would have been justified by appropriate documentation. Case managers, although employed by the facility, can thus assist physicians in this regard and together play an important role as patient advocates.

An 83‐year‐old man with a 7‐year history of myelofibrosis presented to the hospital with progressive weakness and fatigue, which resulted in him tripping and falling onto his left hand and arm 1 day prior to admission. His past medical history was significant for transfusion‐dependent anemia and hypertension. His current treatment regimen for myelofibrosis included thalidomide and darbopoetin alfa.

Physical examination revealed a pale and edematous man who was holding his injured arm to his chest, but in no distress. He had massive hepatosplenomegaly (Figure 1) and pitting edema of the lower extremities that extended to his abdomen.

Figure 1

Laboratory studies showed a white blood count of 5000, hematocrit of 29%, and platelets of 218,000. The peripheral blood smear (Figure 2) showed marked anisocytosis, poikilocytosis, and teardrop cells (Figure 2; arrow).

Figure 2

Imaging of the left arm and hand was significant for a third metacarpal fracture and first phalanx fracture. Of note, these x‐rays also revealed numerous round lucencies within the osseous structures of the left hand, wrist, and forearm (Figure 3; arrow).

Figure 3

The patient's hospital course was uncomplicated and included casting of the left arm, treatment of his lower extremity edema, and transfusion for a slowly declining hematocrit. He was discharged home after several days but died 1 month later.

Primary myelofibrosis is a myeloproliferative disease that consists of 2 phases. The first phase is the growth and proliferation of abnormal bone marrow stem cells, which leads to ineffective erythropoiesis. This is followed by reactive myelofibrosis and extramedullary hematopoiesis.1 These 2 phases of the disease can lead to a constellation of findings, as illustrated in these images. The median length of survival from diagnosis is 3 to 5 years, with the main causes of death being infection, hemorrhage, cardiac failure, and leukemic transformation.1 Presenting signs, symptoms, and laboratory results may include cachexia, splenomegaly, anemia, an increased or decreased white blood cell count and/or platelet count, and an increase in lactate dehydrogenase. Radiographically, the most common findings are marked splenomegaly and osteosclerosis.2

Osteosclerotic lesions are found in 30% to 70% of patients with myelofibrosis and are a result of marrow fibrosis, which leads to the appearance of diffuse, patchy increases in bone density.2 Osteolytic lesions, as seen in this case, are much less common. They appear in the literature in case reports, but are not considered to be a typical finding. They are usually painful and have been reported as a poor prognostic indicator.3, 4

An 83‐year‐old man with a 7‐year history of myelofibrosis presented to the hospital with progressive weakness and fatigue, which resulted in him tripping and falling onto his left hand and arm 1 day prior to admission. His past medical history was significant for transfusion‐dependent anemia and hypertension. His current treatment regimen for myelofibrosis included thalidomide and darbopoetin alfa.

Physical examination revealed a pale and edematous man who was holding his injured arm to his chest, but in no distress. He had massive hepatosplenomegaly (Figure 1) and pitting edema of the lower extremities that extended to his abdomen.

Figure 1

Laboratory studies showed a white blood count of 5000, hematocrit of 29%, and platelets of 218,000. The peripheral blood smear (Figure 2) showed marked anisocytosis, poikilocytosis, and teardrop cells (Figure 2; arrow).

Figure 2

Imaging of the left arm and hand was significant for a third metacarpal fracture and first phalanx fracture. Of note, these x‐rays also revealed numerous round lucencies within the osseous structures of the left hand, wrist, and forearm (Figure 3; arrow).

Figure 3

The patient's hospital course was uncomplicated and included casting of the left arm, treatment of his lower extremity edema, and transfusion for a slowly declining hematocrit. He was discharged home after several days but died 1 month later.

Primary myelofibrosis is a myeloproliferative disease that consists of 2 phases. The first phase is the growth and proliferation of abnormal bone marrow stem cells, which leads to ineffective erythropoiesis. This is followed by reactive myelofibrosis and extramedullary hematopoiesis.1 These 2 phases of the disease can lead to a constellation of findings, as illustrated in these images. The median length of survival from diagnosis is 3 to 5 years, with the main causes of death being infection, hemorrhage, cardiac failure, and leukemic transformation.1 Presenting signs, symptoms, and laboratory results may include cachexia, splenomegaly, anemia, an increased or decreased white blood cell count and/or platelet count, and an increase in lactate dehydrogenase. Radiographically, the most common findings are marked splenomegaly and osteosclerosis.2

Osteosclerotic lesions are found in 30% to 70% of patients with myelofibrosis and are a result of marrow fibrosis, which leads to the appearance of diffuse, patchy increases in bone density.2 Osteolytic lesions, as seen in this case, are much less common. They appear in the literature in case reports, but are not considered to be a typical finding. They are usually painful and have been reported as a poor prognostic indicator.3, 4

An 83‐year‐old man with a 7‐year history of myelofibrosis presented to the hospital with progressive weakness and fatigue, which resulted in him tripping and falling onto his left hand and arm 1 day prior to admission. His past medical history was significant for transfusion‐dependent anemia and hypertension. His current treatment regimen for myelofibrosis included thalidomide and darbopoetin alfa.

Physical examination revealed a pale and edematous man who was holding his injured arm to his chest, but in no distress. He had massive hepatosplenomegaly (Figure 1) and pitting edema of the lower extremities that extended to his abdomen.

Figure 1

Laboratory studies showed a white blood count of 5000, hematocrit of 29%, and platelets of 218,000. The peripheral blood smear (Figure 2) showed marked anisocytosis, poikilocytosis, and teardrop cells (Figure 2; arrow).

Figure 2

Imaging of the left arm and hand was significant for a third metacarpal fracture and first phalanx fracture. Of note, these x‐rays also revealed numerous round lucencies within the osseous structures of the left hand, wrist, and forearm (Figure 3; arrow).

Figure 3

The patient's hospital course was uncomplicated and included casting of the left arm, treatment of his lower extremity edema, and transfusion for a slowly declining hematocrit. He was discharged home after several days but died 1 month later.

Primary myelofibrosis is a myeloproliferative disease that consists of 2 phases. The first phase is the growth and proliferation of abnormal bone marrow stem cells, which leads to ineffective erythropoiesis. This is followed by reactive myelofibrosis and extramedullary hematopoiesis.1 These 2 phases of the disease can lead to a constellation of findings, as illustrated in these images. The median length of survival from diagnosis is 3 to 5 years, with the main causes of death being infection, hemorrhage, cardiac failure, and leukemic transformation.1 Presenting signs, symptoms, and laboratory results may include cachexia, splenomegaly, anemia, an increased or decreased white blood cell count and/or platelet count, and an increase in lactate dehydrogenase. Radiographically, the most common findings are marked splenomegaly and osteosclerosis.2

Osteosclerotic lesions are found in 30% to 70% of patients with myelofibrosis and are a result of marrow fibrosis, which leads to the appearance of diffuse, patchy increases in bone density.2 Osteolytic lesions, as seen in this case, are much less common. They appear in the literature in case reports, but are not considered to be a typical finding. They are usually painful and have been reported as a poor prognostic indicator.3, 4

A 61‐year‐old healthy man presented with recurrent right wrist pain. The patient underwent unsuccessful carpal tunnel surgery and pathology revealed granulomatous inflammation. With worsening pain and new nodular inflammation, prednisone and azathioprine were prescribed for presumed sarcoidosis. Subsequently, right arm ulceration developed (Figure 1), and wound and blood cultures revealed Sporothrix schenkii. Immunosuppressive medications were stopped, but the ulceration progressed and ultimately involved the entire arm (Figure 2). New lower‐extremity fluid collections seen on the magnetic resonance imaging (MRI) MRI (Figures 3, 4) prompted several surgical debridements. Multiple abscesses formed in all extremities despite amphotericin and itraconazole therapy. The patient was eventually discharged with ongoing amphotericin and plans for surveillance imaging and repeated debridements.

Figure 1

Figure 2

Figure 3

Figure 4

Sporothrix schenckii is a dimorphic fungus often associated with cutaneous infections of the extremities in gardeners. These infections are definitively treated with oral azole medications or topical potassium iodide. Disseminated sporotrichosis is almost exclusively seen in immunosuppressed patients including those with human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS)1 and hematologic malignancies.2 Iatrogenic immunosuppression from chemotherapy or corticosteroids can also place a patient at risk for disseminated disease. Disseminated sporotrichosis is extremely difficult to manage with significant morbidity and mortality; fatal cases are commonly reported in the literature.3

A 61‐year‐old healthy man presented with recurrent right wrist pain. The patient underwent unsuccessful carpal tunnel surgery and pathology revealed granulomatous inflammation. With worsening pain and new nodular inflammation, prednisone and azathioprine were prescribed for presumed sarcoidosis. Subsequently, right arm ulceration developed (Figure 1), and wound and blood cultures revealed Sporothrix schenkii. Immunosuppressive medications were stopped, but the ulceration progressed and ultimately involved the entire arm (Figure 2). New lower‐extremity fluid collections seen on the magnetic resonance imaging (MRI) MRI (Figures 3, 4) prompted several surgical debridements. Multiple abscesses formed in all extremities despite amphotericin and itraconazole therapy. The patient was eventually discharged with ongoing amphotericin and plans for surveillance imaging and repeated debridements.

Figure 1

Figure 2

Figure 3

Figure 4

Sporothrix schenckii is a dimorphic fungus often associated with cutaneous infections of the extremities in gardeners. These infections are definitively treated with oral azole medications or topical potassium iodide. Disseminated sporotrichosis is almost exclusively seen in immunosuppressed patients including those with human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS)1 and hematologic malignancies.2 Iatrogenic immunosuppression from chemotherapy or corticosteroids can also place a patient at risk for disseminated disease. Disseminated sporotrichosis is extremely difficult to manage with significant morbidity and mortality; fatal cases are commonly reported in the literature.3

A 61‐year‐old healthy man presented with recurrent right wrist pain. The patient underwent unsuccessful carpal tunnel surgery and pathology revealed granulomatous inflammation. With worsening pain and new nodular inflammation, prednisone and azathioprine were prescribed for presumed sarcoidosis. Subsequently, right arm ulceration developed (Figure 1), and wound and blood cultures revealed Sporothrix schenkii. Immunosuppressive medications were stopped, but the ulceration progressed and ultimately involved the entire arm (Figure 2). New lower‐extremity fluid collections seen on the magnetic resonance imaging (MRI) MRI (Figures 3, 4) prompted several surgical debridements. Multiple abscesses formed in all extremities despite amphotericin and itraconazole therapy. The patient was eventually discharged with ongoing amphotericin and plans for surveillance imaging and repeated debridements.

Figure 1

Figure 2

Figure 3

Figure 4

Sporothrix schenckii is a dimorphic fungus often associated with cutaneous infections of the extremities in gardeners. These infections are definitively treated with oral azole medications or topical potassium iodide. Disseminated sporotrichosis is almost exclusively seen in immunosuppressed patients including those with human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS)1 and hematologic malignancies.2 Iatrogenic immunosuppression from chemotherapy or corticosteroids can also place a patient at risk for disseminated disease. Disseminated sporotrichosis is extremely difficult to manage with significant morbidity and mortality; fatal cases are commonly reported in the literature.3

Hospitalized patients with sepsis or severe nosocomial infections are frequently treated empirically with broad‐spectrum antibiotics. Cefepime hydrochloride, a fourth‐generation cephalosporin, is a common antibiotic of first choice. Its proconvulsant properties are well described in the literature,1, 2 but its importance as a potential cause of change in mental status is probably underestimated. We report a case of change in mental status related to nonconvulsive status epilepticus (NCSE) caused by the use of cefepime in an elderly, hospitalized patient. Our goal is to raise awareness about this uncommon and still underrecognized complication.

Case Report

A 72‐year‐old woman with stage III chronic kidney disease secondary to hypertension with a stable creatinine of 1.5 mg/dL (glomerular filtration rate (GFR) estimated by the modification of diet in renal disease (MDRD) at 36 mL/minute/1.73 m²) was admitted to the hospital for worsening of her chronic back pain. She had a past medical history significant for hyperlipidemia, asthma, and peripheral vascular disease, with breast cancer in remission since 1989. She had no history of seizures or cerebrovascular disease. Her medications were ibuprofen, oxycontin, cilastazol, acetaminophen/oxycodone, and an albuterol/empratropium inhaler. Her physical examination was remarkable only for decreased strength in the right lower extremity. Magnetic resonance imaging (MRI) of the lumbosacral spine showed signs consistent with an inflammatory process at the level of L4‐L5. A computed tomography (CT)‐guided biopsy was performed and confirmed a diagnosis of osteomyelitis on biopsy. Cultures from the biopsy grew Pseudomonas aeruginosa and treatment with intravenous cefepime at a dose of 1 g every 12 hours was initiated. Over the next 3 days, the patient had a gradual worsening of her mental status, leading to pronounced somnolence with occasional episodes of agitation during which she had no focal motor deficits. Her mental status declined to the point of unresponsiveness to simple verbal commands. She had not received any new medications other than cefepime. Her creatinine level was stable throughout this time period at 1.6 mg/dL. No other abnormalities were found on laboratory evaluation or on CT and MRI scans of the brain. An electroencephalogram (EEG) was markedly abnormal due to a generalized background slowing and disorganization with frequent bilateral paroxysmal epileptiform discharges, confirming the clinical diagnosis of subclinical generalized status epilepticus. Given that there were no other intrinsic neurological or metabolic reasons for this mental status change, and given that cefepime was the only new medication added before the patient started deteriorating, cefepime was discontinued and treatment for seizures was started with intravenous benzodiazepines. Over the next 2 days, her mental status returned to normal. She was soon discharged to a rehabilitation center.

Discussion

Beta‐lactam antibiotics have been described to induce seizures due to their direct and/or indirect inhibition of the gamma‐aminobutyric acid (GABA) system.1, 3 Previous experiments have shown a dose‐dependent effect on seizures, and suggest that the cephalosporin with the most pronounced proconvulsant effect is cefazolin.1, 3

Cefepime has been associated with neurological side effects such as headache, confusion, hallucinations, agitation, myoclonus, ataxia, seizures, and coma. Another underrecognized but critical side effect is NSCE. This is defined as seizure activity for more than 30 minutes, with cognitive and behavioral changes, but without convulsive clinical manifestations. This complication has been reported in the literature, but it is probably underrecognized.3‐7 The tendency for cefepime to produce more subclinical activity than the other cephalosporins is not well understood.

Cefepime is mainly eliminated though renal excretion (85%) and displays linear pharmacokinetic properties, thus its dose needs to be adjusted according to renal function. Consequently, in the case of renal dysfunction, accumulation of the drug is proportional to the degree of renal impairment. For NSCE, the most important risk factor is renal impairment, although cases in patients with normal kidney function have been described.36 Age, preexisting central nervous system (CNS) disease, sepsis, and cardiopulmonary bypass have also been reported as possible risk factors for NCSE.1

Cefepime can accumulate in the cerebrospinal fluid (CSF) in the setting of renal dysfunction, decreased protein‐binding capacity (as is sometimes seen in the elderly), and increased blood‐brain permeability in the setting of CNS infections. Accumulation of the drug in the CSF can lead to blockade of the GABA‐A receptor through a mechanism of competitive antagonism1, 8

The onset of NSCE varies between 1 and 16 days after initiation of cefepime therapy.3‐5, 7 It is frequently confused with delirium, since hospitalized patients treated with broad‐spectrum antibiotics such as cefepime frequently have other comorbidities and risk factors for delirium.

This can delay the diagnosis of NSCE due to a lack of awareness of this critical complication in the setting of renal dysfunction. In order to quantify the likelihood that the NSCE was related to cefepime and not to other causes, we calculated a Naranjo adverse drug events probability score, which consists of 9 questions on the relationship between the adverse event and the incriminated drug.9 Each answer is scored from 1 to +2 points. This score was designed to quantify the strength of the association between any adverse event and a pharmacological agent.

In our patient, the Naranjo score was 7 points, suggesting that the diagnosis of cefepime‐induced NCSE was probable.

The diagnosis of NCSE is made through a combination of a high index of clinical suspicion, specific findings on EEG, and improvement with withdrawal of the drug. Fatal outcomes have been reported.5, 6 Early and prompt recognition of the condition is crucial for the prevention of its morbidity and mortality.

The mainstay of treatment is prompt withdrawal of antibiotics and symptomatic treatment with benzodiazepines or barbiturates. Very severe cases with refractory seizures have been treated with hemodialysis. Phenytoin should be avoided as a treatment of this condition due to its lack of GABA‐agonist activity.

Conclusion

Cefepime can cause NCSE, predominantly in patients with renal dysfunction. Its frequency is probably underestimated in hospitalized patients with multiple comorbid conditions. Hospitalists should be aware of this unusual but critical relationship, especially in patients with renal failure. A high level of clinical suspicion and an emergency EEG are essential to obtain a prompt and accurate diagnosis.

Hospitalized patients with sepsis or severe nosocomial infections are frequently treated empirically with broad‐spectrum antibiotics. Cefepime hydrochloride, a fourth‐generation cephalosporin, is a common antibiotic of first choice. Its proconvulsant properties are well described in the literature,1, 2 but its importance as a potential cause of change in mental status is probably underestimated. We report a case of change in mental status related to nonconvulsive status epilepticus (NCSE) caused by the use of cefepime in an elderly, hospitalized patient. Our goal is to raise awareness about this uncommon and still underrecognized complication.

Case Report

A 72‐year‐old woman with stage III chronic kidney disease secondary to hypertension with a stable creatinine of 1.5 mg/dL (glomerular filtration rate (GFR) estimated by the modification of diet in renal disease (MDRD) at 36 mL/minute/1.73 m²) was admitted to the hospital for worsening of her chronic back pain. She had a past medical history significant for hyperlipidemia, asthma, and peripheral vascular disease, with breast cancer in remission since 1989. She had no history of seizures or cerebrovascular disease. Her medications were ibuprofen, oxycontin, cilastazol, acetaminophen/oxycodone, and an albuterol/empratropium inhaler. Her physical examination was remarkable only for decreased strength in the right lower extremity. Magnetic resonance imaging (MRI) of the lumbosacral spine showed signs consistent with an inflammatory process at the level of L4‐L5. A computed tomography (CT)‐guided biopsy was performed and confirmed a diagnosis of osteomyelitis on biopsy. Cultures from the biopsy grew Pseudomonas aeruginosa and treatment with intravenous cefepime at a dose of 1 g every 12 hours was initiated. Over the next 3 days, the patient had a gradual worsening of her mental status, leading to pronounced somnolence with occasional episodes of agitation during which she had no focal motor deficits. Her mental status declined to the point of unresponsiveness to simple verbal commands. She had not received any new medications other than cefepime. Her creatinine level was stable throughout this time period at 1.6 mg/dL. No other abnormalities were found on laboratory evaluation or on CT and MRI scans of the brain. An electroencephalogram (EEG) was markedly abnormal due to a generalized background slowing and disorganization with frequent bilateral paroxysmal epileptiform discharges, confirming the clinical diagnosis of subclinical generalized status epilepticus. Given that there were no other intrinsic neurological or metabolic reasons for this mental status change, and given that cefepime was the only new medication added before the patient started deteriorating, cefepime was discontinued and treatment for seizures was started with intravenous benzodiazepines. Over the next 2 days, her mental status returned to normal. She was soon discharged to a rehabilitation center.

Discussion

Beta‐lactam antibiotics have been described to induce seizures due to their direct and/or indirect inhibition of the gamma‐aminobutyric acid (GABA) system.1, 3 Previous experiments have shown a dose‐dependent effect on seizures, and suggest that the cephalosporin with the most pronounced proconvulsant effect is cefazolin.1, 3

Cefepime has been associated with neurological side effects such as headache, confusion, hallucinations, agitation, myoclonus, ataxia, seizures, and coma. Another underrecognized but critical side effect is NSCE. This is defined as seizure activity for more than 30 minutes, with cognitive and behavioral changes, but without convulsive clinical manifestations. This complication has been reported in the literature, but it is probably underrecognized.3‐7 The tendency for cefepime to produce more subclinical activity than the other cephalosporins is not well understood.

Cefepime is mainly eliminated though renal excretion (85%) and displays linear pharmacokinetic properties, thus its dose needs to be adjusted according to renal function. Consequently, in the case of renal dysfunction, accumulation of the drug is proportional to the degree of renal impairment. For NSCE, the most important risk factor is renal impairment, although cases in patients with normal kidney function have been described.36 Age, preexisting central nervous system (CNS) disease, sepsis, and cardiopulmonary bypass have also been reported as possible risk factors for NCSE.1

Cefepime can accumulate in the cerebrospinal fluid (CSF) in the setting of renal dysfunction, decreased protein‐binding capacity (as is sometimes seen in the elderly), and increased blood‐brain permeability in the setting of CNS infections. Accumulation of the drug in the CSF can lead to blockade of the GABA‐A receptor through a mechanism of competitive antagonism1, 8

The onset of NSCE varies between 1 and 16 days after initiation of cefepime therapy.3‐5, 7 It is frequently confused with delirium, since hospitalized patients treated with broad‐spectrum antibiotics such as cefepime frequently have other comorbidities and risk factors for delirium.

This can delay the diagnosis of NSCE due to a lack of awareness of this critical complication in the setting of renal dysfunction. In order to quantify the likelihood that the NSCE was related to cefepime and not to other causes, we calculated a Naranjo adverse drug events probability score, which consists of 9 questions on the relationship between the adverse event and the incriminated drug.9 Each answer is scored from 1 to +2 points. This score was designed to quantify the strength of the association between any adverse event and a pharmacological agent.

In our patient, the Naranjo score was 7 points, suggesting that the diagnosis of cefepime‐induced NCSE was probable.

The diagnosis of NCSE is made through a combination of a high index of clinical suspicion, specific findings on EEG, and improvement with withdrawal of the drug. Fatal outcomes have been reported.5, 6 Early and prompt recognition of the condition is crucial for the prevention of its morbidity and mortality.

The mainstay of treatment is prompt withdrawal of antibiotics and symptomatic treatment with benzodiazepines or barbiturates. Very severe cases with refractory seizures have been treated with hemodialysis. Phenytoin should be avoided as a treatment of this condition due to its lack of GABA‐agonist activity.

Conclusion

Cefepime can cause NCSE, predominantly in patients with renal dysfunction. Its frequency is probably underestimated in hospitalized patients with multiple comorbid conditions. Hospitalists should be aware of this unusual but critical relationship, especially in patients with renal failure. A high level of clinical suspicion and an emergency EEG are essential to obtain a prompt and accurate diagnosis.

Hospitalized patients with sepsis or severe nosocomial infections are frequently treated empirically with broad‐spectrum antibiotics. Cefepime hydrochloride, a fourth‐generation cephalosporin, is a common antibiotic of first choice. Its proconvulsant properties are well described in the literature,1, 2 but its importance as a potential cause of change in mental status is probably underestimated. We report a case of change in mental status related to nonconvulsive status epilepticus (NCSE) caused by the use of cefepime in an elderly, hospitalized patient. Our goal is to raise awareness about this uncommon and still underrecognized complication.

Case Report

A 72‐year‐old woman with stage III chronic kidney disease secondary to hypertension with a stable creatinine of 1.5 mg/dL (glomerular filtration rate (GFR) estimated by the modification of diet in renal disease (MDRD) at 36 mL/minute/1.73 m²) was admitted to the hospital for worsening of her chronic back pain. She had a past medical history significant for hyperlipidemia, asthma, and peripheral vascular disease, with breast cancer in remission since 1989. She had no history of seizures or cerebrovascular disease. Her medications were ibuprofen, oxycontin, cilastazol, acetaminophen/oxycodone, and an albuterol/empratropium inhaler. Her physical examination was remarkable only for decreased strength in the right lower extremity. Magnetic resonance imaging (MRI) of the lumbosacral spine showed signs consistent with an inflammatory process at the level of L4‐L5. A computed tomography (CT)‐guided biopsy was performed and confirmed a diagnosis of osteomyelitis on biopsy. Cultures from the biopsy grew Pseudomonas aeruginosa and treatment with intravenous cefepime at a dose of 1 g every 12 hours was initiated. Over the next 3 days, the patient had a gradual worsening of her mental status, leading to pronounced somnolence with occasional episodes of agitation during which she had no focal motor deficits. Her mental status declined to the point of unresponsiveness to simple verbal commands. She had not received any new medications other than cefepime. Her creatinine level was stable throughout this time period at 1.6 mg/dL. No other abnormalities were found on laboratory evaluation or on CT and MRI scans of the brain. An electroencephalogram (EEG) was markedly abnormal due to a generalized background slowing and disorganization with frequent bilateral paroxysmal epileptiform discharges, confirming the clinical diagnosis of subclinical generalized status epilepticus. Given that there were no other intrinsic neurological or metabolic reasons for this mental status change, and given that cefepime was the only new medication added before the patient started deteriorating, cefepime was discontinued and treatment for seizures was started with intravenous benzodiazepines. Over the next 2 days, her mental status returned to normal. She was soon discharged to a rehabilitation center.

Discussion

Beta‐lactam antibiotics have been described to induce seizures due to their direct and/or indirect inhibition of the gamma‐aminobutyric acid (GABA) system.1, 3 Previous experiments have shown a dose‐dependent effect on seizures, and suggest that the cephalosporin with the most pronounced proconvulsant effect is cefazolin.1, 3

Cefepime has been associated with neurological side effects such as headache, confusion, hallucinations, agitation, myoclonus, ataxia, seizures, and coma. Another underrecognized but critical side effect is NSCE. This is defined as seizure activity for more than 30 minutes, with cognitive and behavioral changes, but without convulsive clinical manifestations. This complication has been reported in the literature, but it is probably underrecognized.3‐7 The tendency for cefepime to produce more subclinical activity than the other cephalosporins is not well understood.

Cefepime is mainly eliminated though renal excretion (85%) and displays linear pharmacokinetic properties, thus its dose needs to be adjusted according to renal function. Consequently, in the case of renal dysfunction, accumulation of the drug is proportional to the degree of renal impairment. For NSCE, the most important risk factor is renal impairment, although cases in patients with normal kidney function have been described.36 Age, preexisting central nervous system (CNS) disease, sepsis, and cardiopulmonary bypass have also been reported as possible risk factors for NCSE.1

Cefepime can accumulate in the cerebrospinal fluid (CSF) in the setting of renal dysfunction, decreased protein‐binding capacity (as is sometimes seen in the elderly), and increased blood‐brain permeability in the setting of CNS infections. Accumulation of the drug in the CSF can lead to blockade of the GABA‐A receptor through a mechanism of competitive antagonism1, 8

The onset of NSCE varies between 1 and 16 days after initiation of cefepime therapy.3‐5, 7 It is frequently confused with delirium, since hospitalized patients treated with broad‐spectrum antibiotics such as cefepime frequently have other comorbidities and risk factors for delirium.

This can delay the diagnosis of NSCE due to a lack of awareness of this critical complication in the setting of renal dysfunction. In order to quantify the likelihood that the NSCE was related to cefepime and not to other causes, we calculated a Naranjo adverse drug events probability score, which consists of 9 questions on the relationship between the adverse event and the incriminated drug.9 Each answer is scored from 1 to +2 points. This score was designed to quantify the strength of the association between any adverse event and a pharmacological agent.

In our patient, the Naranjo score was 7 points, suggesting that the diagnosis of cefepime‐induced NCSE was probable.

The diagnosis of NCSE is made through a combination of a high index of clinical suspicion, specific findings on EEG, and improvement with withdrawal of the drug. Fatal outcomes have been reported.5, 6 Early and prompt recognition of the condition is crucial for the prevention of its morbidity and mortality.

The mainstay of treatment is prompt withdrawal of antibiotics and symptomatic treatment with benzodiazepines or barbiturates. Very severe cases with refractory seizures have been treated with hemodialysis. Phenytoin should be avoided as a treatment of this condition due to its lack of GABA‐agonist activity.

Conclusion

Cefepime can cause NCSE, predominantly in patients with renal dysfunction. Its frequency is probably underestimated in hospitalized patients with multiple comorbid conditions. Hospitalists should be aware of this unusual but critical relationship, especially in patients with renal failure. A high level of clinical suspicion and an emergency EEG are essential to obtain a prompt and accurate diagnosis.

Rapid onset visual loss with limb involvement in a postpartum female invites a number of differential diagnoses, including infectious encephalomyelitis, vasculitis, an acute demyelinating disorder, and arterial/venous sinus thrombosis. We report an interesting, but initially puzzling, case of rapidly progressive optic neuritis and transverse myelitis in a postpartum Ghanaian female, later diagnosed as neuromyelitis optica (NMO).

Case Report

A 33‐year‐old Ghanaian female presented with bilateral eye pain worsened by eye movement, rapid visual loss, bifrontal headache, and right leg weakness for 48 hours. She was 8 weeks postpartum, with labor having been induced at 35 weeks due to hypertension; she was discharged on labetolol and no other medications. There was no other relevant past medical or family history, or illicit drug use.

On admission, her blood pressure was 148/99 mm Hg, pulse rate 92 beats per minute, temperature 37.8C, oxygen saturation 97% on room air, respiratory rate 14 breaths per minute, and Glasgow coma scale 15/15. Cranial nerve examination showed absent light perception in the right eye with a relative afferent pupillary defect, 20/30 Snellen visual acuity in the left eye, bilateral optic disc edema (worse on the right) and grade 1 hypertensive retinopathy, and slight facial asymmetry but otherwise grossly intact cranial nerves. Neurological examination of the limbs showed 5/5 power in upper limbs, but flexion was weaker (3/5) than extension (4/5) in lower limbs. There was hyperreflexia in the right arm and right leg with positive Babinski sign. Sensation, proprioception, and coordination were normal in both upper and lower limbs; a Romberg test was negative.

Other systems examinations were unremarkable. Complete blood count, standard electrolytes, erythrocyte sedimentation rate, C‐reactive protein, urinalysis, chest x‐ray, and an urgent head computed tomography (CT) with contrast were all noncontributory.

To cover a broad range of possible infective processes, ceftriaxone and acyclovir were started. In view of the rapid progression of her presentation she was transferred urgently to a specialty neurology hospital. There she was started on intravenous methylprednisolone to cover possible inflammatory causes, such as multiple sclerosis (MS) and vasculitis. However, her lower limb weakness progressed and she developed urinary retention. Lumbar puncture was performed; the opening pressure was 160 mm H₂O. Cerebrospinal fluid (CSF) analysis demonstrated elevated protein level (121 mg/dL), 16/mm³ lymphocytosis, normal glucose, and negative gram stain and culture. Conventional CSF polymerase chain reaction for herpes simplex, Epstein‐Barr, varicella‐zoster, cytomegalovirus, John Cunningham virus (to rule out progressive multifocal leukoencephalopathy), human herpesvirus 6 viruses, and toxoplasma were negative. Antinuclear antibody, anticytoplasmic antibody, antiganglioside antibodies (GM1, GQ1b), thrombophilia screen (activated protein C resistance ratio, lupus anticoagulant and anticardiolipin antibodies, antithrombin III, protein C, and S deficiency), B12, folate, ferritin, and human immunodeficiency virus (HIV) screen were also negative. Brain magnetic resonance imaging (MRI), including venography, was normal. However, T2‐weighted MRI images of the orbit showed signal hyperintensity of the left optic chiasm and optic nerves, and T2‐weighted images of the spinal cord showed hyperintensity within the upper 4 and lower 4 thoracic levels. The diagnosis of NMO was entertained.

Following intravenous steroids, the patient was treated with a tapering course of oral steroids. She began to make a slow recovery with the aid of intense physiotherapy. Having walked in at her initial admission, she returned home needing crutches 4 weeks later and the visual acuity in her right eye remained 20/120 at best. Her requirement for intermittent urinary self‐catheterization and rectal enemas decreased with time. She continued to be rehabilitated at home with input from the community physiotherapy team and remains in remission at 12 months.

Discussion

First described by Eugene Devic in 1894, NMO is a rare, chronic inflammatory, demyelinating disease of the central nervous system.1 It is more common among Asians and Africans, with a female predominance (females comprise over two‐thirds of patients and more than 80% of those with the relapsing form of the disease).2 The mean age of onset is 35 to 47 years.2 The prevalence and incidence of NMO has not been established, partly because the disease is underrecognized and often confused with MS.

Rapid onset visual loss with limb involvement in a postpartum female invites a number of differential diagnoses, including infectious encephalomyelitis, vasculitis, an acute demyelinating disorder, and arterial/venous sinus thrombosis. We report an interesting, but initially puzzling, case of rapidly progressive optic neuritis and transverse myelitis in a postpartum Ghanaian female, later diagnosed as neuromyelitis optica (NMO).

Case Report

A 33‐year‐old Ghanaian female presented with bilateral eye pain worsened by eye movement, rapid visual loss, bifrontal headache, and right leg weakness for 48 hours. She was 8 weeks postpartum, with labor having been induced at 35 weeks due to hypertension; she was discharged on labetolol and no other medications. There was no other relevant past medical or family history, or illicit drug use.

On admission, her blood pressure was 148/99 mm Hg, pulse rate 92 beats per minute, temperature 37.8C, oxygen saturation 97% on room air, respiratory rate 14 breaths per minute, and Glasgow coma scale 15/15. Cranial nerve examination showed absent light perception in the right eye with a relative afferent pupillary defect, 20/30 Snellen visual acuity in the left eye, bilateral optic disc edema (worse on the right) and grade 1 hypertensive retinopathy, and slight facial asymmetry but otherwise grossly intact cranial nerves. Neurological examination of the limbs showed 5/5 power in upper limbs, but flexion was weaker (3/5) than extension (4/5) in lower limbs. There was hyperreflexia in the right arm and right leg with positive Babinski sign. Sensation, proprioception, and coordination were normal in both upper and lower limbs; a Romberg test was negative.

Other systems examinations were unremarkable. Complete blood count, standard electrolytes, erythrocyte sedimentation rate, C‐reactive protein, urinalysis, chest x‐ray, and an urgent head computed tomography (CT) with contrast were all noncontributory.

To cover a broad range of possible infective processes, ceftriaxone and acyclovir were started. In view of the rapid progression of her presentation she was transferred urgently to a specialty neurology hospital. There she was started on intravenous methylprednisolone to cover possible inflammatory causes, such as multiple sclerosis (MS) and vasculitis. However, her lower limb weakness progressed and she developed urinary retention. Lumbar puncture was performed; the opening pressure was 160 mm H₂O. Cerebrospinal fluid (CSF) analysis demonstrated elevated protein level (121 mg/dL), 16/mm³ lymphocytosis, normal glucose, and negative gram stain and culture. Conventional CSF polymerase chain reaction for herpes simplex, Epstein‐Barr, varicella‐zoster, cytomegalovirus, John Cunningham virus (to rule out progressive multifocal leukoencephalopathy), human herpesvirus 6 viruses, and toxoplasma were negative. Antinuclear antibody, anticytoplasmic antibody, antiganglioside antibodies (GM1, GQ1b), thrombophilia screen (activated protein C resistance ratio, lupus anticoagulant and anticardiolipin antibodies, antithrombin III, protein C, and S deficiency), B12, folate, ferritin, and human immunodeficiency virus (HIV) screen were also negative. Brain magnetic resonance imaging (MRI), including venography, was normal. However, T2‐weighted MRI images of the orbit showed signal hyperintensity of the left optic chiasm and optic nerves, and T2‐weighted images of the spinal cord showed hyperintensity within the upper 4 and lower 4 thoracic levels. The diagnosis of NMO was entertained.

Following intravenous steroids, the patient was treated with a tapering course of oral steroids. She began to make a slow recovery with the aid of intense physiotherapy. Having walked in at her initial admission, she returned home needing crutches 4 weeks later and the visual acuity in her right eye remained 20/120 at best. Her requirement for intermittent urinary self‐catheterization and rectal enemas decreased with time. She continued to be rehabilitated at home with input from the community physiotherapy team and remains in remission at 12 months.

Discussion

First described by Eugene Devic in 1894, NMO is a rare, chronic inflammatory, demyelinating disease of the central nervous system.1 It is more common among Asians and Africans, with a female predominance (females comprise over two‐thirds of patients and more than 80% of those with the relapsing form of the disease).2 The mean age of onset is 35 to 47 years.2 The prevalence and incidence of NMO has not been established, partly because the disease is underrecognized and often confused with MS.

Rapid onset visual loss with limb involvement in a postpartum female invites a number of differential diagnoses, including infectious encephalomyelitis, vasculitis, an acute demyelinating disorder, and arterial/venous sinus thrombosis. We report an interesting, but initially puzzling, case of rapidly progressive optic neuritis and transverse myelitis in a postpartum Ghanaian female, later diagnosed as neuromyelitis optica (NMO).

Case Report

A 33‐year‐old Ghanaian female presented with bilateral eye pain worsened by eye movement, rapid visual loss, bifrontal headache, and right leg weakness for 48 hours. She was 8 weeks postpartum, with labor having been induced at 35 weeks due to hypertension; she was discharged on labetolol and no other medications. There was no other relevant past medical or family history, or illicit drug use.

On admission, her blood pressure was 148/99 mm Hg, pulse rate 92 beats per minute, temperature 37.8C, oxygen saturation 97% on room air, respiratory rate 14 breaths per minute, and Glasgow coma scale 15/15. Cranial nerve examination showed absent light perception in the right eye with a relative afferent pupillary defect, 20/30 Snellen visual acuity in the left eye, bilateral optic disc edema (worse on the right) and grade 1 hypertensive retinopathy, and slight facial asymmetry but otherwise grossly intact cranial nerves. Neurological examination of the limbs showed 5/5 power in upper limbs, but flexion was weaker (3/5) than extension (4/5) in lower limbs. There was hyperreflexia in the right arm and right leg with positive Babinski sign. Sensation, proprioception, and coordination were normal in both upper and lower limbs; a Romberg test was negative.

Other systems examinations were unremarkable. Complete blood count, standard electrolytes, erythrocyte sedimentation rate, C‐reactive protein, urinalysis, chest x‐ray, and an urgent head computed tomography (CT) with contrast were all noncontributory.

To cover a broad range of possible infective processes, ceftriaxone and acyclovir were started. In view of the rapid progression of her presentation she was transferred urgently to a specialty neurology hospital. There she was started on intravenous methylprednisolone to cover possible inflammatory causes, such as multiple sclerosis (MS) and vasculitis. However, her lower limb weakness progressed and she developed urinary retention. Lumbar puncture was performed; the opening pressure was 160 mm H₂O. Cerebrospinal fluid (CSF) analysis demonstrated elevated protein level (121 mg/dL), 16/mm³ lymphocytosis, normal glucose, and negative gram stain and culture. Conventional CSF polymerase chain reaction for herpes simplex, Epstein‐Barr, varicella‐zoster, cytomegalovirus, John Cunningham virus (to rule out progressive multifocal leukoencephalopathy), human herpesvirus 6 viruses, and toxoplasma were negative. Antinuclear antibody, anticytoplasmic antibody, antiganglioside antibodies (GM1, GQ1b), thrombophilia screen (activated protein C resistance ratio, lupus anticoagulant and anticardiolipin antibodies, antithrombin III, protein C, and S deficiency), B12, folate, ferritin, and human immunodeficiency virus (HIV) screen were also negative. Brain magnetic resonance imaging (MRI), including venography, was normal. However, T2‐weighted MRI images of the orbit showed signal hyperintensity of the left optic chiasm and optic nerves, and T2‐weighted images of the spinal cord showed hyperintensity within the upper 4 and lower 4 thoracic levels. The diagnosis of NMO was entertained.

Following intravenous steroids, the patient was treated with a tapering course of oral steroids. She began to make a slow recovery with the aid of intense physiotherapy. Having walked in at her initial admission, she returned home needing crutches 4 weeks later and the visual acuity in her right eye remained 20/120 at best. Her requirement for intermittent urinary self‐catheterization and rectal enemas decreased with time. She continued to be rehabilitated at home with input from the community physiotherapy team and remains in remission at 12 months.

Discussion

First described by Eugene Devic in 1894, NMO is a rare, chronic inflammatory, demyelinating disease of the central nervous system.1 It is more common among Asians and Africans, with a female predominance (females comprise over two‐thirds of patients and more than 80% of those with the relapsing form of the disease).2 The mean age of onset is 35 to 47 years.2 The prevalence and incidence of NMO has not been established, partly because the disease is underrecognized and often confused with MS.

The first observation units were implemented more than 40 years ago with the goal of reducing the number and duration of inpatient stays. Since then, observation units (OUs) have evolved as a safe alternative to hospitalization14 for the delivery of finite periods of care, typically less than 24 hours.58 Observation services allow for time to determine the need for hospitalization in cases that are unclear after their initial evaluation and treatment.9 Observation status is an administrative classification related to reimbursement that can be applied to patients whose diagnosis, treatment, stabilization, and discharge can reasonably be expected within 24 hours.10, 11 The site of care for observation is dependent in part upon existing facility structures; some institutions utilize virtual OUs within the emergency department (ED) or hospital ward, while others have dedicated, geographically distinct OUs, which may function as an extension of either the ED or inpatient settings.9

OUs have been instrumental in providing care to adult patients with chest pain, asthma, and acute infections.1218 Recently, there has been an increase in the number of publications from pediatric OUs in the United States and abroad. Observation may be a preferred model of care for select pediatric patients, as hospitalized children often experience brief stays.1921 Previous reviews on this model of care have combined adult and pediatric literature and have included research from countries with healthcare structures that differ considerably from the United States.2224 To date, no systematic review has summarized the pediatric OU literature with a focus on the US healthcare system.

As payers and hospitals seek cost‐effective alternatives to traditional inpatient care, geographically distinct OUs may become integral to the future of healthcare delivery for children. This systematic review provides a descriptive overview of the structure and function of pediatric OUs in the United States. We also scrutinize the outcome measures presented in the included publications and propose future directions for research to improve both observation unit care, as well as the care delivered to patients under observation status within general inpatient or ED settings.

Methods

Results

A systematic search of the electronic databases identified 222 unique citations (Figure 1). A total of 107 abstracts were evaluated. We identified 48 articles for full‐text review, of which 18 met inclusion criteria. Hand search of references yielded 24 additional articles, of which 3 met inclusion criteria. Interrater agreement for selected articles was high at 98% (kappa = 0.85).

Figure 1

Discussion

In the 40 years since the first studies of pediatric OUs, several US health systems have extended observation services to children. This model of care may be expanding, as suggested by an increase in the number of publications in the past 10 years. However, the number of centers within the US reporting on their OU experience remains small. Our systematic review identified a recurrent theme related to OUsthe opportunity to improve operational processes of care compared with the traditional inpatient alternative. We have identified the need to standardize OU outcomes and propose measures for future OU research.

Conclusions

OUs have emerged to provide treatment at the intersection of outpatient and inpatient care during a time of dramatic change in both emergency and hospital medicine. As hospitalists expand their scope of practice to include observation care, opportunities will arise to collaborate with ED physicians and share their growing expertise in quality and efficiency of hospital care delivery to improve observation services for children. OUs have been established with laudable goalsto reduce inpatient admissions, increase patient safety, improve efficiency, and control costs. The current evidence is not adequate to determine if this model of healthcare delivery achieves these goals for children. Through synthesis of existing data, we have identified a need for standard reporting for OU outcomes and propose consistent measures for future observation care research. Only through prospective evaluation of comparable outcomes can we appraise the performance of pediatric OUs across institutions.