A 42‐year‐old woman with a history of mild asthma presented to the emergency department (ED) following 1 week of headache. She had been in her usual state of good health until 1 week prior to her presentation, when she noticed intermittent frontal headaches without neck stiffness or other neurologic symptoms. She then developed diffuse myalgias, fatigue, subjective fevers, and rigors for the 24 hours prior to presentation. On the morning of presentation, chest tightness, palpitations, and shortness of breath occurred. She used her albuterol metered‐dose inhaler without relief and went to the hospital.

Many of these features can be explained by a viral syndrome exacerbating underlying asthma or by a psychiatric condition such as anxiety or depression, but they may also be a harbinger of a systemic process, including infection, malignancy, or autoimmunity. Because the onset of headache is temporally distant from the other symptoms, I am more inclined to believe that it represents a primary intracranial process than I would if it were coincident with the onset of the other acute symptoms. If the fevers and rigors are verified, infection would be the initial concern. Failure to respond to her inhalers may either signify a severe asthma exacerbation or a nonbronchospastic cause of dyspnea.

She reported mild nausea, but denied photophobia, vomiting, abdominal pain, diarrhea, melena, or hematochezia. She did not have recent ill contacts, animal bites, or travel. Her medical history included asthma, diverticulitis, chronic right ankle pain, and obesity. She reported an allergic rash to amoxicillin. Her medications were sulindac and fluticasone/salmeterol, and albuterol metered‐dose inhalers. She worked as a preschool teacher and was married with 2 children. She denied any tobacco use and seldom drank alcoholic beverages. On exam, temperature was 36.7C, pulse was 107 beats per minute, blood pressure was 129/91 mm Hg, respiratory rate was 19 breaths per minute, and oxygen saturation was 98% while breathing ambient air. Her face and anterior neck were flushed and diaphoretic, and her sclerae were icteric. There was no nuchal rigidity. Her cardiac rhythm was regular without murmurs, lungs were clear to auscultation, and the abdomen was mildly tender to palpation in the epigastrium and right upper quadrant. The white blood cell (WBC) count was 9200/L, with 84% neutrophils, 3% lymphocytes, 6% monocytes, and 7% eosinophils. The hemoglobin was 14.8 g/dL and the platelet count was 166,000/L. Total serum bilirubin was 4.6 mg/dL, aspartate aminotransferase (AST) was 459 U/L (normal range, 8‐31), alanine aminotransferase (ALT) was 667 U/L (normal range, 7‐31), and alkaline phosphatase was 146 U/L (normal range, 39‐117). Serum electrolytes, creatinine, lactate, lipase, thyrotropin, coagulation studies, and cardiac enzymes were all normal. Urinalysis showed trace leukocyte esterase and bilirubin, as well as 3 WBCs and 2 red cells per high‐power field. Chest radiography and an electrocardiogram demonstrated no abnormalities.

The major findingwhich is critical to focusing problem‐solving in the face of a broad range of symptomsis her hepatitis. The common etiologies for hepatitis of this degree include viruses (hepatitis A and cytomegalovirus [CMV] should be considered given her work in preschool), toxins, autoimmunity, and vascular events. Liver disease in association with flushing raises the possibility of carcinoid syndrome with liver metastases. The lack of wheezing makes the bronchospasm of asthma or carcinoid less suitable explanations for her shortness of breath. Her eosinophilia is mild but probably is not accounted for alone by well‐controlled asthma in a person with no history of atopic disease. I would also ask her about any alternative and over‐the‐counter remedies. The paucity of lymphocytes raises the possibility of human immunodeficiency virus (HIV), Hodgkin's disease, or systemic lupus erythematosus. Although she does not have a documented fever or leukocytosis, she reported fevers and chills and is diaphoretic and tachycardic, so exclusion of biliary obstruction and cholangitis is the highest priority.

An abdominal ultrasound demonstrated hepatomegaly with moderate fatty infiltration and a normal gallbladder without pericholecystic fluid. The intrahepatic and extrahepatic biliary ducts were normal and the hepatic and portal veins were patent. Computed tomography of the abdomen showed slight thickening of the sigmoid colon wall. Ciprofloxacin and metronidazole were administered for possible diverticulitis. Over the first 48 hours of hospitalization her symptoms improved markedly. Her flushing resolved and she had no recorded fevers in the hospital. Serologies were negative for hepatitis A immunoglobulin M (IgM), hepatitis B surface antibody, hepatitis B surface antigen, and hepatitis C antibody. A monospot test was negative and the erythrocyte sedimentation rate was 11 mm/hour. Blood and urine cultures were negative. On the second hospital day the absolute eosinophil count rose to 855/L (15% of 5700 WBCs). On the fourth hospital day, the absolute eosinophil count was 1092/L, the total bilirubin was 1.9 mg/dL, and the AST and ALT were 174 U/L and 476 U/L, respectively. Antibiotics were stopped and she was discharged home.

Her prompt improvement suggests either a self‐limited condition or a response to the antibiotics. The rapid but incomplete resolution of her hepatitis is in keeping with a withdrawal of a toxin, relief of biliary obstruction, or a transient vascular event, and is less consistent with a viral hepatitis or an infiltrative process. With normal biliary system imaging, sterile blood cultures, and the absence of fever or leukocytosis, cholangitis is unlikely. Likewise, there is no suggestion of a vascular event, either obstructive or hemodynamic, that is impairing the liver.

A common cause of eosinophilia in hospitalized patients is medications, so it would be useful to monitor that count after the new antibiotics. At this point, I also wonder if the eosinophils are a feature of the underlying illness, as they were present to a modest degree on admission before any new medications were administered. The overlap of eosinophilia and hepatitis brings to mind a medication reaction (eg, to sulindac) or a hepatobiliary parasite, such as ascaris or clonorchis, for which she lacks a known exposure. Many patients experience flushing in the setting of fever or stress, but sustained flushing may suggest a systemic illness characterized by the release of vasoactive mediators such as carcinoid syndrome or mastocytosis. The latter might be considered more strongly if the eosinophilia is deemed to be primary (rather than reactive) after a thorough evaluation.

After 2 days at home, the patient had recurrence of subjective fevers, with chest, back, and abdominal pain, fatigue, loose stools, and rigors. She returned to the ED, where she was noted to have facial erythema and injected sclerae, but the remainder of her physical exam was normal. The total serum bilirubin was 1.1 mg/dL, AST was 156 U/L, ALT was 214 U/L, and alkaline phosphatase was 240 U/L. Serum lipase was normal. WBC count was 14,000/L, with 94% neutrophils, 3% lymphocytes, 2% monocytes, and 1% eosinophils. She was again treated empirically with ciprofloxacin and metronidazole. Endoscopic ultrasound was normal, with no evidence of gallbladder sludge or microlithiasis. Stool cultures, assay for Clostridium difficile, and examination for ova and parasites were negative. The 24‐hour urine demonstrated no elevation in 5‐hydroxyindoleacetic acid. An adrenocorticotropic hormone (ACTH) stimulation test was normal. HIV antibody was negative. Her symptoms improved within 2 days. The eosinophil count rose and peaked at 1541/L by the third hospital day, while the transaminase elevations resolved. Antibiotics were discontinued. A liver biopsy showed mixed macrovesicular and microvesicular fatty metamorphosis and steatohepatitis with eosinophils (Figures 1 and 2). She was discharged home on the sixth hospital day.

Figure 1

Figure 2

Her illness can now be characterized as relapsing inflammation, which given the frequency (over days) suggests either an indolent infectious focus that periodically causes systemic inflammation or reexposure to a toxic substance. The 2 most notable laboratory abnormalities, the hepatitis and the eosinophilia, persist but have differing trajectories. While the liver function tests have progressively normalized despite clinical relapses, the eosinophils have had a more fluctuating course characterized by increases during the hospitalization and higher levels during the second hospitalization. The absence of an infection, recurrent systemic inflammation, and eosinophilic hepatitis suggest a hypersensitivity reaction to a medication or other substance. She is most likely being reexposed at home, where her symptoms occur, and not in the hospital, where her symptoms resolve. Sulindac is a leading candidate, because nonsteroidal antiinflammatory drugs (NSAIDs) cause a number of hypersensitivity reactions and are frequently stopped when sick patients enter the hospital.

Seven days after discharge she developed acute onset of subjective fever, nausea, diffuse myalgias, and flushing, identical to the 2 prior episodes, and she again returned to the ED. Her temperature was 39.1C, heart rate was 120 beats per minute, blood pressure was 87/50 mm Hg, respiratory rate was 18 breaths per minute, and the oxygen saturation was 96% while breathing room air. She had diffuse flushing from her neck over her torso and was diaphoretic with injected sclera and conjunctiva. The WBC count was 11,400/L with 97% neutrophils and 3% lymphocytes. Total bilirubin was 0.8 mg/dL, AST was 134 U/L, ALT was 140 U/L, and alkaline phosphatase was 144 U/L. She was readmitted to the hospital. Following admission, she had no fevers, the flushing resolved, and AST and ALT levels decreased. The only treatment the patient received in the ED and during her hospital stay was acetaminophen as needed for pain or fever. The eosinophil count peaked at 1404/L by hospital day 4. Blood and urine cultures were negative. IgM antibodies to Epstein‐Barr virus were not detected, CMV DNA was not detected, and a rapid plasma reagin (RPR) test was nonreactive. Ferritin, ceruloplasmin, alpha‐1‐antitrypsin, and tryptase levels were normal. Antimitochondrial, antismooth muscle, antineutrophil cytoplasmic, and antinuclear antibodies were negative. There was no monoclonal band on serum protein electrophoresis. A blood smear for Borrelia detected no spirochetes.

A complete picture of the uncommon but classic flushing disorders, namely carcinoid, mastocytosis, and pheochromocytoma, has not emerged. The constellation of inflammation, mucosal and hepatic involvement, and eosinophilia are most consistent with a drug hypersensitivity reaction. Additionally, the recurrent inflammation is becoming more severe, as manifest by the fever and hemodynamic derangements, which suggests an increasing sensitization to the offending agent. I would review every drug she has received in the hospital, but given the recurrences after discharge her home medications are the most likely explanation. Of these, sulindac is the most likely culprit.

On further questioning, it was learned that the patient began taking sulindac 200 mg twice daily to treat her chronic ankle pain 6 weeks before the first admission. The medication had been stopped on each admission. She was instructed to discontinue sulindac. She has had no recurrences of symptoms and her hepatitis and eosinophilia have resolved.

DISCUSSION

This patient presented with recurrent skin findings, eosinophilia, hepatitis, and constitutional symptoms caused by hypersensitivity to sulindac. This drug‐induced hypersensitivity syndrome was originally described with anticonvulsant drugs (carbamazepine, phenytoin, and phenobarbitone) and named anticonvulsant hypersensitivity syndrome,1, 2 but has been observed with many other medications, including allopurinol, dapsone, minocycline, and nevirapine. The term drug rash with eosinophilia and systemic symptoms (DRESS) syndrome has been recently adopted to convey the cardinal features that characterize this disorder.3

DRESS syndrome is defined by rash, fever, and internal organ involvement.4 Also included in the diagnostic criteria are hematologic abnormalities (eosinophilia 1.5 10⁹/mm³ or the presence of atypical lymphocytes) and lymphadenopathy.3 The multiorgan involvement distinguishes DRESS from other cutaneous drug eruptions. In a review of the French Pharmacovigilance Database for all cases of DRESS over a 15‐year period, 73% to 100% of patients were reported to have dermatologic abnormalities, most frequently a maculopapular rash or erythroderma. Less common skin findings include vesicles, bullae, pustules, erythroderma, and purpuric lesions. Liver abnormalities were observed in more than 60% of patients and were the most frequent systemic finding.5 Eosinophilia was the most common hematologic abnormality, present in more than 50% of cases. As this case underscores, DRESS syndrome typically begins 3 to 8 weeks after initiation of the drug because it is a delayed type IV hypersensitivity reaction.6 Fever can occur within hours on rechallenge because of the presence of memory T cells.

The main treatment for DRESS syndrome is withdrawal of the offending drug. Systemic corticosteroids have been recommended in cases with life‐threatening pulmonary or cardiac involvement, but have not been shown to be helpful in reversing renal or hepatic disease. Mortality, usually from end‐organ damage, occurs in about 10% of cases. The most common drugs are phenobarbital, carbamazepine, and phenytoin, with incidences of 1 in 5000 to 1 in 10,000. NSAIDs and antibiotics also have been implicated frequently. Human herpesvirus type 6 coinfection and genetically inherited slow acetylation have been associated with DRESS, although causal links have yet to be established.7, 8

The initial challenge in caring for a patient with multiple symptoms, exam findings, and test abnormalities is the coherent framing of the key clinical features that require explanation. This process, called problem representation, allows clinicians to search among a bounded list of possible diagnoses (or solutions) rather than invoking a differential diagnosis for every single abnormality. In searching for the proper diagnosis, this patient's clinical course required frequent reframing as more data became available.

Initially, the problem was framed as a 42‐year‐old woman with hepatitis. As the flushing and eosinophilia, which initially appeared to be transient and possibly nonspecific, became more prominent, the problem representation was revised to a 42‐year‐old woman with hepatitis, eosinophilia, and flushing. Since this triad did not immediately invoke a single diagnosis for the treating clinicians or the discussant, the differential diagnosis of hepatitis and eosinophilia and the differential diagnosis of flushing were considered in parallel.

Hepatitis and eosinophilia can occur coincidentally in the setting of parasitic infections, particularly helminths (ascaris, strongyloidiasis, and toxocaris) and liver flukes (opisthorchis and clonorchis), which invade the hepatobiliary system and induce a reactive eosinophilia. Some neoplasms, such as lymphomas and leukemias, and myeloproliferative disorders, including hypereosinophilic syndrome and mastocytosis, may have neoplastic cellular invasion of the liver and induce eosinophilia. Systemic drug hypersensitivity reactions are typically characterized by eosinophilia, transaminase elevations,9 and hepatitis on histology.10 As with this patient, liver biopsy in drug‐induced hepatitis shows a mixed microvesicular and macrovesicular steatosis, often with eosinophils.10

The flushing prompted a thorough but negative workup for the classic flushing disorders. The discussant's attempts to unify the flushing with the other clinical features illustrate how framing can affect our reasoning. Hepatitis, eosinophilia, and flushing defied an obvious single explanation and led the discussant down parallel diagnostic reasoning pathways. Although flushing is 1 of the well‐described dermatologic manifestations in drug‐hypersensitivity reactions, framing the central features as hepatitis, eosinophilia, and rash would have more readily suggested a drug reaction as a unifying diagnosis.11

The tempo and periodicity of this patient's illness provided the final formulation of a 42‐year‐old woman with hepatitis, eosinophilia, and flushing (or rash) that occurs every few days at home and resolves in the hospital. This formulation and the increasingly severe presentation, suggesting sensitization, were highly suggestive of an exogenous cause of her illness.

This case highlights how easily medication side effects can be overlooked during an extensive evaluation and how vigilant medication reconciliation coupled with an increased understanding of the spectrum of drug reactions can lead to early detection and prevention of potentially serious effects. In the case of DRESS, recognizing an association between a rash and organ involvement is central to making the diagnosis. Eosinophilia that accompanies a rash can further aid in narrowing the differential diagnosis.

The case also serves as a reminder of how framing with the slightest imprecision (eg, flushing instead of rash) can derail or delay the diagnostic process, yet is indispensable in tackling a complicated case. Finally, a time‐honored lesson in diagnosis is highlighted yet again: the diagnosis can usually be flushed out from the history.1214

A 42‐year‐old woman with a history of mild asthma presented to the emergency department (ED) following 1 week of headache. She had been in her usual state of good health until 1 week prior to her presentation, when she noticed intermittent frontal headaches without neck stiffness or other neurologic symptoms. She then developed diffuse myalgias, fatigue, subjective fevers, and rigors for the 24 hours prior to presentation. On the morning of presentation, chest tightness, palpitations, and shortness of breath occurred. She used her albuterol metered‐dose inhaler without relief and went to the hospital.

Many of these features can be explained by a viral syndrome exacerbating underlying asthma or by a psychiatric condition such as anxiety or depression, but they may also be a harbinger of a systemic process, including infection, malignancy, or autoimmunity. Because the onset of headache is temporally distant from the other symptoms, I am more inclined to believe that it represents a primary intracranial process than I would if it were coincident with the onset of the other acute symptoms. If the fevers and rigors are verified, infection would be the initial concern. Failure to respond to her inhalers may either signify a severe asthma exacerbation or a nonbronchospastic cause of dyspnea.

She reported mild nausea, but denied photophobia, vomiting, abdominal pain, diarrhea, melena, or hematochezia. She did not have recent ill contacts, animal bites, or travel. Her medical history included asthma, diverticulitis, chronic right ankle pain, and obesity. She reported an allergic rash to amoxicillin. Her medications were sulindac and fluticasone/salmeterol, and albuterol metered‐dose inhalers. She worked as a preschool teacher and was married with 2 children. She denied any tobacco use and seldom drank alcoholic beverages. On exam, temperature was 36.7C, pulse was 107 beats per minute, blood pressure was 129/91 mm Hg, respiratory rate was 19 breaths per minute, and oxygen saturation was 98% while breathing ambient air. Her face and anterior neck were flushed and diaphoretic, and her sclerae were icteric. There was no nuchal rigidity. Her cardiac rhythm was regular without murmurs, lungs were clear to auscultation, and the abdomen was mildly tender to palpation in the epigastrium and right upper quadrant. The white blood cell (WBC) count was 9200/L, with 84% neutrophils, 3% lymphocytes, 6% monocytes, and 7% eosinophils. The hemoglobin was 14.8 g/dL and the platelet count was 166,000/L. Total serum bilirubin was 4.6 mg/dL, aspartate aminotransferase (AST) was 459 U/L (normal range, 8‐31), alanine aminotransferase (ALT) was 667 U/L (normal range, 7‐31), and alkaline phosphatase was 146 U/L (normal range, 39‐117). Serum electrolytes, creatinine, lactate, lipase, thyrotropin, coagulation studies, and cardiac enzymes were all normal. Urinalysis showed trace leukocyte esterase and bilirubin, as well as 3 WBCs and 2 red cells per high‐power field. Chest radiography and an electrocardiogram demonstrated no abnormalities.

The major findingwhich is critical to focusing problem‐solving in the face of a broad range of symptomsis her hepatitis. The common etiologies for hepatitis of this degree include viruses (hepatitis A and cytomegalovirus [CMV] should be considered given her work in preschool), toxins, autoimmunity, and vascular events. Liver disease in association with flushing raises the possibility of carcinoid syndrome with liver metastases. The lack of wheezing makes the bronchospasm of asthma or carcinoid less suitable explanations for her shortness of breath. Her eosinophilia is mild but probably is not accounted for alone by well‐controlled asthma in a person with no history of atopic disease. I would also ask her about any alternative and over‐the‐counter remedies. The paucity of lymphocytes raises the possibility of human immunodeficiency virus (HIV), Hodgkin's disease, or systemic lupus erythematosus. Although she does not have a documented fever or leukocytosis, she reported fevers and chills and is diaphoretic and tachycardic, so exclusion of biliary obstruction and cholangitis is the highest priority.

An abdominal ultrasound demonstrated hepatomegaly with moderate fatty infiltration and a normal gallbladder without pericholecystic fluid. The intrahepatic and extrahepatic biliary ducts were normal and the hepatic and portal veins were patent. Computed tomography of the abdomen showed slight thickening of the sigmoid colon wall. Ciprofloxacin and metronidazole were administered for possible diverticulitis. Over the first 48 hours of hospitalization her symptoms improved markedly. Her flushing resolved and she had no recorded fevers in the hospital. Serologies were negative for hepatitis A immunoglobulin M (IgM), hepatitis B surface antibody, hepatitis B surface antigen, and hepatitis C antibody. A monospot test was negative and the erythrocyte sedimentation rate was 11 mm/hour. Blood and urine cultures were negative. On the second hospital day the absolute eosinophil count rose to 855/L (15% of 5700 WBCs). On the fourth hospital day, the absolute eosinophil count was 1092/L, the total bilirubin was 1.9 mg/dL, and the AST and ALT were 174 U/L and 476 U/L, respectively. Antibiotics were stopped and she was discharged home.

Her prompt improvement suggests either a self‐limited condition or a response to the antibiotics. The rapid but incomplete resolution of her hepatitis is in keeping with a withdrawal of a toxin, relief of biliary obstruction, or a transient vascular event, and is less consistent with a viral hepatitis or an infiltrative process. With normal biliary system imaging, sterile blood cultures, and the absence of fever or leukocytosis, cholangitis is unlikely. Likewise, there is no suggestion of a vascular event, either obstructive or hemodynamic, that is impairing the liver.

A common cause of eosinophilia in hospitalized patients is medications, so it would be useful to monitor that count after the new antibiotics. At this point, I also wonder if the eosinophils are a feature of the underlying illness, as they were present to a modest degree on admission before any new medications were administered. The overlap of eosinophilia and hepatitis brings to mind a medication reaction (eg, to sulindac) or a hepatobiliary parasite, such as ascaris or clonorchis, for which she lacks a known exposure. Many patients experience flushing in the setting of fever or stress, but sustained flushing may suggest a systemic illness characterized by the release of vasoactive mediators such as carcinoid syndrome or mastocytosis. The latter might be considered more strongly if the eosinophilia is deemed to be primary (rather than reactive) after a thorough evaluation.

After 2 days at home, the patient had recurrence of subjective fevers, with chest, back, and abdominal pain, fatigue, loose stools, and rigors. She returned to the ED, where she was noted to have facial erythema and injected sclerae, but the remainder of her physical exam was normal. The total serum bilirubin was 1.1 mg/dL, AST was 156 U/L, ALT was 214 U/L, and alkaline phosphatase was 240 U/L. Serum lipase was normal. WBC count was 14,000/L, with 94% neutrophils, 3% lymphocytes, 2% monocytes, and 1% eosinophils. She was again treated empirically with ciprofloxacin and metronidazole. Endoscopic ultrasound was normal, with no evidence of gallbladder sludge or microlithiasis. Stool cultures, assay for Clostridium difficile, and examination for ova and parasites were negative. The 24‐hour urine demonstrated no elevation in 5‐hydroxyindoleacetic acid. An adrenocorticotropic hormone (ACTH) stimulation test was normal. HIV antibody was negative. Her symptoms improved within 2 days. The eosinophil count rose and peaked at 1541/L by the third hospital day, while the transaminase elevations resolved. Antibiotics were discontinued. A liver biopsy showed mixed macrovesicular and microvesicular fatty metamorphosis and steatohepatitis with eosinophils (Figures 1 and 2). She was discharged home on the sixth hospital day.

Figure 1

Figure 2

Her illness can now be characterized as relapsing inflammation, which given the frequency (over days) suggests either an indolent infectious focus that periodically causes systemic inflammation or reexposure to a toxic substance. The 2 most notable laboratory abnormalities, the hepatitis and the eosinophilia, persist but have differing trajectories. While the liver function tests have progressively normalized despite clinical relapses, the eosinophils have had a more fluctuating course characterized by increases during the hospitalization and higher levels during the second hospitalization. The absence of an infection, recurrent systemic inflammation, and eosinophilic hepatitis suggest a hypersensitivity reaction to a medication or other substance. She is most likely being reexposed at home, where her symptoms occur, and not in the hospital, where her symptoms resolve. Sulindac is a leading candidate, because nonsteroidal antiinflammatory drugs (NSAIDs) cause a number of hypersensitivity reactions and are frequently stopped when sick patients enter the hospital.

Seven days after discharge she developed acute onset of subjective fever, nausea, diffuse myalgias, and flushing, identical to the 2 prior episodes, and she again returned to the ED. Her temperature was 39.1C, heart rate was 120 beats per minute, blood pressure was 87/50 mm Hg, respiratory rate was 18 breaths per minute, and the oxygen saturation was 96% while breathing room air. She had diffuse flushing from her neck over her torso and was diaphoretic with injected sclera and conjunctiva. The WBC count was 11,400/L with 97% neutrophils and 3% lymphocytes. Total bilirubin was 0.8 mg/dL, AST was 134 U/L, ALT was 140 U/L, and alkaline phosphatase was 144 U/L. She was readmitted to the hospital. Following admission, she had no fevers, the flushing resolved, and AST and ALT levels decreased. The only treatment the patient received in the ED and during her hospital stay was acetaminophen as needed for pain or fever. The eosinophil count peaked at 1404/L by hospital day 4. Blood and urine cultures were negative. IgM antibodies to Epstein‐Barr virus were not detected, CMV DNA was not detected, and a rapid plasma reagin (RPR) test was nonreactive. Ferritin, ceruloplasmin, alpha‐1‐antitrypsin, and tryptase levels were normal. Antimitochondrial, antismooth muscle, antineutrophil cytoplasmic, and antinuclear antibodies were negative. There was no monoclonal band on serum protein electrophoresis. A blood smear for Borrelia detected no spirochetes.

A complete picture of the uncommon but classic flushing disorders, namely carcinoid, mastocytosis, and pheochromocytoma, has not emerged. The constellation of inflammation, mucosal and hepatic involvement, and eosinophilia are most consistent with a drug hypersensitivity reaction. Additionally, the recurrent inflammation is becoming more severe, as manifest by the fever and hemodynamic derangements, which suggests an increasing sensitization to the offending agent. I would review every drug she has received in the hospital, but given the recurrences after discharge her home medications are the most likely explanation. Of these, sulindac is the most likely culprit.

On further questioning, it was learned that the patient began taking sulindac 200 mg twice daily to treat her chronic ankle pain 6 weeks before the first admission. The medication had been stopped on each admission. She was instructed to discontinue sulindac. She has had no recurrences of symptoms and her hepatitis and eosinophilia have resolved.

DISCUSSION

This patient presented with recurrent skin findings, eosinophilia, hepatitis, and constitutional symptoms caused by hypersensitivity to sulindac. This drug‐induced hypersensitivity syndrome was originally described with anticonvulsant drugs (carbamazepine, phenytoin, and phenobarbitone) and named anticonvulsant hypersensitivity syndrome,1, 2 but has been observed with many other medications, including allopurinol, dapsone, minocycline, and nevirapine. The term drug rash with eosinophilia and systemic symptoms (DRESS) syndrome has been recently adopted to convey the cardinal features that characterize this disorder.3

DRESS syndrome is defined by rash, fever, and internal organ involvement.4 Also included in the diagnostic criteria are hematologic abnormalities (eosinophilia 1.5 10⁹/mm³ or the presence of atypical lymphocytes) and lymphadenopathy.3 The multiorgan involvement distinguishes DRESS from other cutaneous drug eruptions. In a review of the French Pharmacovigilance Database for all cases of DRESS over a 15‐year period, 73% to 100% of patients were reported to have dermatologic abnormalities, most frequently a maculopapular rash or erythroderma. Less common skin findings include vesicles, bullae, pustules, erythroderma, and purpuric lesions. Liver abnormalities were observed in more than 60% of patients and were the most frequent systemic finding.5 Eosinophilia was the most common hematologic abnormality, present in more than 50% of cases. As this case underscores, DRESS syndrome typically begins 3 to 8 weeks after initiation of the drug because it is a delayed type IV hypersensitivity reaction.6 Fever can occur within hours on rechallenge because of the presence of memory T cells.

The main treatment for DRESS syndrome is withdrawal of the offending drug. Systemic corticosteroids have been recommended in cases with life‐threatening pulmonary or cardiac involvement, but have not been shown to be helpful in reversing renal or hepatic disease. Mortality, usually from end‐organ damage, occurs in about 10% of cases. The most common drugs are phenobarbital, carbamazepine, and phenytoin, with incidences of 1 in 5000 to 1 in 10,000. NSAIDs and antibiotics also have been implicated frequently. Human herpesvirus type 6 coinfection and genetically inherited slow acetylation have been associated with DRESS, although causal links have yet to be established.7, 8

The initial challenge in caring for a patient with multiple symptoms, exam findings, and test abnormalities is the coherent framing of the key clinical features that require explanation. This process, called problem representation, allows clinicians to search among a bounded list of possible diagnoses (or solutions) rather than invoking a differential diagnosis for every single abnormality. In searching for the proper diagnosis, this patient's clinical course required frequent reframing as more data became available.

Initially, the problem was framed as a 42‐year‐old woman with hepatitis. As the flushing and eosinophilia, which initially appeared to be transient and possibly nonspecific, became more prominent, the problem representation was revised to a 42‐year‐old woman with hepatitis, eosinophilia, and flushing. Since this triad did not immediately invoke a single diagnosis for the treating clinicians or the discussant, the differential diagnosis of hepatitis and eosinophilia and the differential diagnosis of flushing were considered in parallel.

Hepatitis and eosinophilia can occur coincidentally in the setting of parasitic infections, particularly helminths (ascaris, strongyloidiasis, and toxocaris) and liver flukes (opisthorchis and clonorchis), which invade the hepatobiliary system and induce a reactive eosinophilia. Some neoplasms, such as lymphomas and leukemias, and myeloproliferative disorders, including hypereosinophilic syndrome and mastocytosis, may have neoplastic cellular invasion of the liver and induce eosinophilia. Systemic drug hypersensitivity reactions are typically characterized by eosinophilia, transaminase elevations,9 and hepatitis on histology.10 As with this patient, liver biopsy in drug‐induced hepatitis shows a mixed microvesicular and macrovesicular steatosis, often with eosinophils.10

The flushing prompted a thorough but negative workup for the classic flushing disorders. The discussant's attempts to unify the flushing with the other clinical features illustrate how framing can affect our reasoning. Hepatitis, eosinophilia, and flushing defied an obvious single explanation and led the discussant down parallel diagnostic reasoning pathways. Although flushing is 1 of the well‐described dermatologic manifestations in drug‐hypersensitivity reactions, framing the central features as hepatitis, eosinophilia, and rash would have more readily suggested a drug reaction as a unifying diagnosis.11

The tempo and periodicity of this patient's illness provided the final formulation of a 42‐year‐old woman with hepatitis, eosinophilia, and flushing (or rash) that occurs every few days at home and resolves in the hospital. This formulation and the increasingly severe presentation, suggesting sensitization, were highly suggestive of an exogenous cause of her illness.

This case highlights how easily medication side effects can be overlooked during an extensive evaluation and how vigilant medication reconciliation coupled with an increased understanding of the spectrum of drug reactions can lead to early detection and prevention of potentially serious effects. In the case of DRESS, recognizing an association between a rash and organ involvement is central to making the diagnosis. Eosinophilia that accompanies a rash can further aid in narrowing the differential diagnosis.

The case also serves as a reminder of how framing with the slightest imprecision (eg, flushing instead of rash) can derail or delay the diagnostic process, yet is indispensable in tackling a complicated case. Finally, a time‐honored lesson in diagnosis is highlighted yet again: the diagnosis can usually be flushed out from the history.1214

A 42‐year‐old woman with a history of mild asthma presented to the emergency department (ED) following 1 week of headache. She had been in her usual state of good health until 1 week prior to her presentation, when she noticed intermittent frontal headaches without neck stiffness or other neurologic symptoms. She then developed diffuse myalgias, fatigue, subjective fevers, and rigors for the 24 hours prior to presentation. On the morning of presentation, chest tightness, palpitations, and shortness of breath occurred. She used her albuterol metered‐dose inhaler without relief and went to the hospital.

Many of these features can be explained by a viral syndrome exacerbating underlying asthma or by a psychiatric condition such as anxiety or depression, but they may also be a harbinger of a systemic process, including infection, malignancy, or autoimmunity. Because the onset of headache is temporally distant from the other symptoms, I am more inclined to believe that it represents a primary intracranial process than I would if it were coincident with the onset of the other acute symptoms. If the fevers and rigors are verified, infection would be the initial concern. Failure to respond to her inhalers may either signify a severe asthma exacerbation or a nonbronchospastic cause of dyspnea.

She reported mild nausea, but denied photophobia, vomiting, abdominal pain, diarrhea, melena, or hematochezia. She did not have recent ill contacts, animal bites, or travel. Her medical history included asthma, diverticulitis, chronic right ankle pain, and obesity. She reported an allergic rash to amoxicillin. Her medications were sulindac and fluticasone/salmeterol, and albuterol metered‐dose inhalers. She worked as a preschool teacher and was married with 2 children. She denied any tobacco use and seldom drank alcoholic beverages. On exam, temperature was 36.7C, pulse was 107 beats per minute, blood pressure was 129/91 mm Hg, respiratory rate was 19 breaths per minute, and oxygen saturation was 98% while breathing ambient air. Her face and anterior neck were flushed and diaphoretic, and her sclerae were icteric. There was no nuchal rigidity. Her cardiac rhythm was regular without murmurs, lungs were clear to auscultation, and the abdomen was mildly tender to palpation in the epigastrium and right upper quadrant. The white blood cell (WBC) count was 9200/L, with 84% neutrophils, 3% lymphocytes, 6% monocytes, and 7% eosinophils. The hemoglobin was 14.8 g/dL and the platelet count was 166,000/L. Total serum bilirubin was 4.6 mg/dL, aspartate aminotransferase (AST) was 459 U/L (normal range, 8‐31), alanine aminotransferase (ALT) was 667 U/L (normal range, 7‐31), and alkaline phosphatase was 146 U/L (normal range, 39‐117). Serum electrolytes, creatinine, lactate, lipase, thyrotropin, coagulation studies, and cardiac enzymes were all normal. Urinalysis showed trace leukocyte esterase and bilirubin, as well as 3 WBCs and 2 red cells per high‐power field. Chest radiography and an electrocardiogram demonstrated no abnormalities.

The major findingwhich is critical to focusing problem‐solving in the face of a broad range of symptomsis her hepatitis. The common etiologies for hepatitis of this degree include viruses (hepatitis A and cytomegalovirus [CMV] should be considered given her work in preschool), toxins, autoimmunity, and vascular events. Liver disease in association with flushing raises the possibility of carcinoid syndrome with liver metastases. The lack of wheezing makes the bronchospasm of asthma or carcinoid less suitable explanations for her shortness of breath. Her eosinophilia is mild but probably is not accounted for alone by well‐controlled asthma in a person with no history of atopic disease. I would also ask her about any alternative and over‐the‐counter remedies. The paucity of lymphocytes raises the possibility of human immunodeficiency virus (HIV), Hodgkin's disease, or systemic lupus erythematosus. Although she does not have a documented fever or leukocytosis, she reported fevers and chills and is diaphoretic and tachycardic, so exclusion of biliary obstruction and cholangitis is the highest priority.

An abdominal ultrasound demonstrated hepatomegaly with moderate fatty infiltration and a normal gallbladder without pericholecystic fluid. The intrahepatic and extrahepatic biliary ducts were normal and the hepatic and portal veins were patent. Computed tomography of the abdomen showed slight thickening of the sigmoid colon wall. Ciprofloxacin and metronidazole were administered for possible diverticulitis. Over the first 48 hours of hospitalization her symptoms improved markedly. Her flushing resolved and she had no recorded fevers in the hospital. Serologies were negative for hepatitis A immunoglobulin M (IgM), hepatitis B surface antibody, hepatitis B surface antigen, and hepatitis C antibody. A monospot test was negative and the erythrocyte sedimentation rate was 11 mm/hour. Blood and urine cultures were negative. On the second hospital day the absolute eosinophil count rose to 855/L (15% of 5700 WBCs). On the fourth hospital day, the absolute eosinophil count was 1092/L, the total bilirubin was 1.9 mg/dL, and the AST and ALT were 174 U/L and 476 U/L, respectively. Antibiotics were stopped and she was discharged home.

Her prompt improvement suggests either a self‐limited condition or a response to the antibiotics. The rapid but incomplete resolution of her hepatitis is in keeping with a withdrawal of a toxin, relief of biliary obstruction, or a transient vascular event, and is less consistent with a viral hepatitis or an infiltrative process. With normal biliary system imaging, sterile blood cultures, and the absence of fever or leukocytosis, cholangitis is unlikely. Likewise, there is no suggestion of a vascular event, either obstructive or hemodynamic, that is impairing the liver.

A common cause of eosinophilia in hospitalized patients is medications, so it would be useful to monitor that count after the new antibiotics. At this point, I also wonder if the eosinophils are a feature of the underlying illness, as they were present to a modest degree on admission before any new medications were administered. The overlap of eosinophilia and hepatitis brings to mind a medication reaction (eg, to sulindac) or a hepatobiliary parasite, such as ascaris or clonorchis, for which she lacks a known exposure. Many patients experience flushing in the setting of fever or stress, but sustained flushing may suggest a systemic illness characterized by the release of vasoactive mediators such as carcinoid syndrome or mastocytosis. The latter might be considered more strongly if the eosinophilia is deemed to be primary (rather than reactive) after a thorough evaluation.

After 2 days at home, the patient had recurrence of subjective fevers, with chest, back, and abdominal pain, fatigue, loose stools, and rigors. She returned to the ED, where she was noted to have facial erythema and injected sclerae, but the remainder of her physical exam was normal. The total serum bilirubin was 1.1 mg/dL, AST was 156 U/L, ALT was 214 U/L, and alkaline phosphatase was 240 U/L. Serum lipase was normal. WBC count was 14,000/L, with 94% neutrophils, 3% lymphocytes, 2% monocytes, and 1% eosinophils. She was again treated empirically with ciprofloxacin and metronidazole. Endoscopic ultrasound was normal, with no evidence of gallbladder sludge or microlithiasis. Stool cultures, assay for Clostridium difficile, and examination for ova and parasites were negative. The 24‐hour urine demonstrated no elevation in 5‐hydroxyindoleacetic acid. An adrenocorticotropic hormone (ACTH) stimulation test was normal. HIV antibody was negative. Her symptoms improved within 2 days. The eosinophil count rose and peaked at 1541/L by the third hospital day, while the transaminase elevations resolved. Antibiotics were discontinued. A liver biopsy showed mixed macrovesicular and microvesicular fatty metamorphosis and steatohepatitis with eosinophils (Figures 1 and 2). She was discharged home on the sixth hospital day.

Figure 1

Figure 2

Her illness can now be characterized as relapsing inflammation, which given the frequency (over days) suggests either an indolent infectious focus that periodically causes systemic inflammation or reexposure to a toxic substance. The 2 most notable laboratory abnormalities, the hepatitis and the eosinophilia, persist but have differing trajectories. While the liver function tests have progressively normalized despite clinical relapses, the eosinophils have had a more fluctuating course characterized by increases during the hospitalization and higher levels during the second hospitalization. The absence of an infection, recurrent systemic inflammation, and eosinophilic hepatitis suggest a hypersensitivity reaction to a medication or other substance. She is most likely being reexposed at home, where her symptoms occur, and not in the hospital, where her symptoms resolve. Sulindac is a leading candidate, because nonsteroidal antiinflammatory drugs (NSAIDs) cause a number of hypersensitivity reactions and are frequently stopped when sick patients enter the hospital.

Seven days after discharge she developed acute onset of subjective fever, nausea, diffuse myalgias, and flushing, identical to the 2 prior episodes, and she again returned to the ED. Her temperature was 39.1C, heart rate was 120 beats per minute, blood pressure was 87/50 mm Hg, respiratory rate was 18 breaths per minute, and the oxygen saturation was 96% while breathing room air. She had diffuse flushing from her neck over her torso and was diaphoretic with injected sclera and conjunctiva. The WBC count was 11,400/L with 97% neutrophils and 3% lymphocytes. Total bilirubin was 0.8 mg/dL, AST was 134 U/L, ALT was 140 U/L, and alkaline phosphatase was 144 U/L. She was readmitted to the hospital. Following admission, she had no fevers, the flushing resolved, and AST and ALT levels decreased. The only treatment the patient received in the ED and during her hospital stay was acetaminophen as needed for pain or fever. The eosinophil count peaked at 1404/L by hospital day 4. Blood and urine cultures were negative. IgM antibodies to Epstein‐Barr virus were not detected, CMV DNA was not detected, and a rapid plasma reagin (RPR) test was nonreactive. Ferritin, ceruloplasmin, alpha‐1‐antitrypsin, and tryptase levels were normal. Antimitochondrial, antismooth muscle, antineutrophil cytoplasmic, and antinuclear antibodies were negative. There was no monoclonal band on serum protein electrophoresis. A blood smear for Borrelia detected no spirochetes.

A complete picture of the uncommon but classic flushing disorders, namely carcinoid, mastocytosis, and pheochromocytoma, has not emerged. The constellation of inflammation, mucosal and hepatic involvement, and eosinophilia are most consistent with a drug hypersensitivity reaction. Additionally, the recurrent inflammation is becoming more severe, as manifest by the fever and hemodynamic derangements, which suggests an increasing sensitization to the offending agent. I would review every drug she has received in the hospital, but given the recurrences after discharge her home medications are the most likely explanation. Of these, sulindac is the most likely culprit.

On further questioning, it was learned that the patient began taking sulindac 200 mg twice daily to treat her chronic ankle pain 6 weeks before the first admission. The medication had been stopped on each admission. She was instructed to discontinue sulindac. She has had no recurrences of symptoms and her hepatitis and eosinophilia have resolved.

DISCUSSION

This patient presented with recurrent skin findings, eosinophilia, hepatitis, and constitutional symptoms caused by hypersensitivity to sulindac. This drug‐induced hypersensitivity syndrome was originally described with anticonvulsant drugs (carbamazepine, phenytoin, and phenobarbitone) and named anticonvulsant hypersensitivity syndrome,1, 2 but has been observed with many other medications, including allopurinol, dapsone, minocycline, and nevirapine. The term drug rash with eosinophilia and systemic symptoms (DRESS) syndrome has been recently adopted to convey the cardinal features that characterize this disorder.3

DRESS syndrome is defined by rash, fever, and internal organ involvement.4 Also included in the diagnostic criteria are hematologic abnormalities (eosinophilia 1.5 10⁹/mm³ or the presence of atypical lymphocytes) and lymphadenopathy.3 The multiorgan involvement distinguishes DRESS from other cutaneous drug eruptions. In a review of the French Pharmacovigilance Database for all cases of DRESS over a 15‐year period, 73% to 100% of patients were reported to have dermatologic abnormalities, most frequently a maculopapular rash or erythroderma. Less common skin findings include vesicles, bullae, pustules, erythroderma, and purpuric lesions. Liver abnormalities were observed in more than 60% of patients and were the most frequent systemic finding.5 Eosinophilia was the most common hematologic abnormality, present in more than 50% of cases. As this case underscores, DRESS syndrome typically begins 3 to 8 weeks after initiation of the drug because it is a delayed type IV hypersensitivity reaction.6 Fever can occur within hours on rechallenge because of the presence of memory T cells.

The main treatment for DRESS syndrome is withdrawal of the offending drug. Systemic corticosteroids have been recommended in cases with life‐threatening pulmonary or cardiac involvement, but have not been shown to be helpful in reversing renal or hepatic disease. Mortality, usually from end‐organ damage, occurs in about 10% of cases. The most common drugs are phenobarbital, carbamazepine, and phenytoin, with incidences of 1 in 5000 to 1 in 10,000. NSAIDs and antibiotics also have been implicated frequently. Human herpesvirus type 6 coinfection and genetically inherited slow acetylation have been associated with DRESS, although causal links have yet to be established.7, 8

The initial challenge in caring for a patient with multiple symptoms, exam findings, and test abnormalities is the coherent framing of the key clinical features that require explanation. This process, called problem representation, allows clinicians to search among a bounded list of possible diagnoses (or solutions) rather than invoking a differential diagnosis for every single abnormality. In searching for the proper diagnosis, this patient's clinical course required frequent reframing as more data became available.

Initially, the problem was framed as a 42‐year‐old woman with hepatitis. As the flushing and eosinophilia, which initially appeared to be transient and possibly nonspecific, became more prominent, the problem representation was revised to a 42‐year‐old woman with hepatitis, eosinophilia, and flushing. Since this triad did not immediately invoke a single diagnosis for the treating clinicians or the discussant, the differential diagnosis of hepatitis and eosinophilia and the differential diagnosis of flushing were considered in parallel.

Hepatitis and eosinophilia can occur coincidentally in the setting of parasitic infections, particularly helminths (ascaris, strongyloidiasis, and toxocaris) and liver flukes (opisthorchis and clonorchis), which invade the hepatobiliary system and induce a reactive eosinophilia. Some neoplasms, such as lymphomas and leukemias, and myeloproliferative disorders, including hypereosinophilic syndrome and mastocytosis, may have neoplastic cellular invasion of the liver and induce eosinophilia. Systemic drug hypersensitivity reactions are typically characterized by eosinophilia, transaminase elevations,9 and hepatitis on histology.10 As with this patient, liver biopsy in drug‐induced hepatitis shows a mixed microvesicular and macrovesicular steatosis, often with eosinophils.10

The flushing prompted a thorough but negative workup for the classic flushing disorders. The discussant's attempts to unify the flushing with the other clinical features illustrate how framing can affect our reasoning. Hepatitis, eosinophilia, and flushing defied an obvious single explanation and led the discussant down parallel diagnostic reasoning pathways. Although flushing is 1 of the well‐described dermatologic manifestations in drug‐hypersensitivity reactions, framing the central features as hepatitis, eosinophilia, and rash would have more readily suggested a drug reaction as a unifying diagnosis.11

The tempo and periodicity of this patient's illness provided the final formulation of a 42‐year‐old woman with hepatitis, eosinophilia, and flushing (or rash) that occurs every few days at home and resolves in the hospital. This formulation and the increasingly severe presentation, suggesting sensitization, were highly suggestive of an exogenous cause of her illness.

This case highlights how easily medication side effects can be overlooked during an extensive evaluation and how vigilant medication reconciliation coupled with an increased understanding of the spectrum of drug reactions can lead to early detection and prevention of potentially serious effects. In the case of DRESS, recognizing an association between a rash and organ involvement is central to making the diagnosis. Eosinophilia that accompanies a rash can further aid in narrowing the differential diagnosis.

The case also serves as a reminder of how framing with the slightest imprecision (eg, flushing instead of rash) can derail or delay the diagnostic process, yet is indispensable in tackling a complicated case. Finally, a time‐honored lesson in diagnosis is highlighted yet again: the diagnosis can usually be flushed out from the history.1214

A 52‐year‐old man presented to the emergency department (ED) from a skilled nursing facility with a complaint of bilateral upper‐quadrant abdominal pain of 48 hours' duration. The pain was sharp, nonradiating, constant, and was associated with nausea, vomiting, and constipation. The patient denied any fever, back pain, dysuria, melena, or hematochezia. In the rehabilitation facility the patient had been initially evaluated for this pain. He was given laxatives and stool softeners for presumed constipation but these measures had not been effective. A computed tomography (CT) scan of the abdomen had only showed stool in the colon and he was sent to the ED for further evaluation.

Apart from severe degenerative joint disease in both his knees he was in good health. He was in the skilled nursing facility (SNF) for rehabilitation for bilateral knee replacement surgery done 9 days prior to this presentation. His postoperative course was unremarkable. He had been maintained on prophylaxis for venous thromboembolism with enoxaparin since postoperative day 1 at a daily dose of 40 mg subcutaneously, and was transferred to the SNF on postoperative day 6 on the same dose. His was receiving oxycodone and Tylenol for pain. He was on no other medications.

Vital signs on presentation revealed a temperature of 97.5F, a heart rate of 100 beats per minute, a respiratory rate of 16 breaths per minute, and a blood pressure of 136/69 mmHg. He was alert and oriented and in mild distress from the abdominal pain. Examination was normal except for tenderness in the upper quadrants of the abdomen though no rigidity or rebound tenderness were noted. Routine chemistries were normal except for sodium of 134 mg/dL. His white count, hemoglobin, hematocrit, and platelet levels were noted to be at 17.5K/L, 10 g/dL, 30%, and 345K/L, respectively, and were stable with regard to his discharge laboratory values. His serum eosinophil level was normal. A complete workup for hypercoagulable state and bleeding disorders including assays for antibodies associated with heparin‐induced thrombocytopenia were negative. He was admitted for further evaluation and treatment.

The patient had another CT scan of the abdomen (Figure 1), which when compared to the one done at the SNF 2 days prior showed markedly enlarged bilateral adrenal glands suggestive of bilateral acute adrenal hemorrhage. The enoxaparin was discontinued and empiric steroid replacement therapy was begun. A random cortisol level was normal but a cosyntropin stimulation test showed an absolute increase in cortisol level of only 0.8 g/dL at both 30 and 60 minutes after administration of 250 g of cosyntropin. An investigation was undertaken to determine if the patient had any prior risk factors for bleeding. There was no evidence of infection and a comprehensive evaluation for bleeding, and coagulation disorders was normal. The bilateral adrenal hemorrhage was attributed to the use of enoxaparin in the postoperative setting. Unfortunately, the patient subsequently developed a deep venous thrombosis in his lower extremity and an inferior vena cava (IVC) filter was placed before discharge. He was doing well 6 months later, and is still continued on glucocorticoid and mineralocorticoid replacement therapy and follows up with endocrinology as an outpatient.

Figure 1

Discussion

Bilateral adrenal hemorrhage is usually associated with massive sepsis from Gram‐negative organisms such as Neisseria meningitides, Pseudomonas aeroginosa, Escherichia coli, and Bacteroides fragilis. Rupert Waterhouse, in 1911, was the first person to describe a patient with severe meningococcal sepsis resulting in acute adrenal hemorrhage and collapse. This was also later described independently by Carl Friderichsen in 1918, and is now referred to as the Waterhouse‐Friderichsen syndrome. Other causes include antiphospholipid antibody syndrome, heparin‐associated thrombocytopenia (HIT), and severe physical stress. Bilateral adrenal hemorrhage can also spontaneously occur in the postoperative period, especially after cardiothoracic or orthopedic surgery. This phenomenon may be related to the frequent use of prophylactic anticoagulants after these types of procedures.

The first case report of bilateral adrenal hemorrhage secondary to use of anticoagulants was described in 1947, and the first case report of successful resuscitation after corticosteroid administration in a patient with bilateral adrenal hemorrhage secondary to anticoagulant use was described by Thorn in 1956.1 A review of the literature demonstrates multiple case reports of adrenal hemorrhage reported in the postoperative period, particularly after joint arthroplasty, and especially after knee replacement surgeries. Most of the recent cases have been associated with use of prophylactic low‐dose heparin or low‐molecular‐weight heparin at the time of adrenal hemorrhage. In a study of 157 case reports of individuals with bilateral hemorrhage (including 22 autopsies), 48 cases were associated with administration of anticoagulants, although the dose and effect were not specified.2 Amador et al.1 showed that out of 4325 autopsies performed from 1949 to 1962 in their institution, 30 cases were found of bilateral hemorrhage, of which 10 were receiving heparin at presumably prophylactic doses; 5 of these patients were also receiving dicumarol.

Mayo Clinic investigators performed a retrospective review of all cases of adrenal hemorrhage over a period of 25 years at their hospital, and found 141 cases of adrenal hemorrhage, of which 78 were bilateral and 63 were unilateral,3 and in 67 patients the condition was diagnosed at autopsy. In this study 14 patients had adrenal hemorrhage in the postoperative period in the absence of lupus anticoagulant or HIT; there was no specific mention in this study of the use of postoperative anticoagulants. Finally, a multicenter case control study was undertaken by Kovacs et al.4 to assess putative risk factors for development of bilateral massive adrenal hemorrhage. In the multivariate analysis, thrombocytopenia, exposure to heparin, and sepsis were found to be strongly associated with risk of hemorrhage. Of 23 patients with bilateral, massive adrenal hemorrhage, 16 had been exposed to heparin, and at least 6 were on exclusively subcutaneous heparin. The authors concluded that heparin exposure was a much bigger risk factor than other coagulopathies, and those exposed to heparin of any route or type for 4 to 6 days and those exposed for more than 6 days were about 17 and 34 times, respectively, more likely to develop bilateral hemorrhage than those who had less than 4 days or no exposure.

The clinical presentation of adrenal insufficiency due to bilateral adrenal hemorrhage is often nonspecific. Symptoms may include abdominal pain, back pain, fever, nausea, vomiting, weakness, obtundation, confusion, and hypotensionall of which are also common postoperative symptoms and can be missed or ignored.5 Rao et al.6 profiled the clinical presentation of 64 cases of bilateral hemorrhage and found the following: abdominal, flank, back, or chest pain (86%); anorexia, nausea, or vomiting (47%); psychiatric symptoms (42%); fever (66%); hypotension recognized before shock episode (19%); and abdominal rigidity or rebound (22%). Adrenal insufficiency becomes clinically evident once 90% of the gland is destroyed. About 50% of patients do not manifest typical laboratory abnormalities, so a high degree of suspicion is necessary to diagnose the condition.3 Also, the laboratory diagnosis of adrenal insufficiency using random cortisol levels is unreliable, as reference ranges in patients experiencing stress (as in the postoperative period) have not been well studied or established. In patients with bilateral hemorrhage postoperatively on prophylactic anticoagulants, the coagulation profile is usually within normal limits and there is typically no evidence of spontaneous bleeding elsewhere. In later stages, the typical laboratory findings of abnormal adrenal function such as hypokalemia, hyponatremia, declining cortisol levels, and an inappropriate response to adrenocorticotropic hormone stimulation test may be seen. A significant drop in hemoglobin secondary to hemorrhage may also be encountered in some patients secondary to the bleed.

CT is the most reliable and extensively used imaging modality for making the diagnosis, although magnetic resonance imaging (MRI) or ultrasound may also be utilized. Early in the course of adrenal hemorrhage, CT findings may be negative, and repeated imaging is appropriate when clinical suspicion is high. The presence of bilateral adrenal enlargement with increased signal attenuation suggests bilateral adrenal hemorrhage. MRI can both characterize adrenal hematomas, and estimate their age.7, 8

Postoperative adrenal hemorrhage and insufficiency is easily treatable and has excellent outcomes; survivors will need lifelong corticosteroid replacement (and usually mineralocorticoid replacement as well). In the Mayo Clinic study, survival was 100% with treatment vs. 17% without treatment. In comparison, sepsis‐induced or stress‐induced adrenal insufficiency has poor outcomes despite adequate treatment (9% survival with treatment vs. 6% survival without treatment).3 Death can occur within hours to days of symptoms if untreated. Treatment includes timely initiation of adrenal hormone replacement and reversal of coagulopathies.

Postoperative venous thromboembolism (VTE) prophylaxis with anticoagulants is the appropriate care in many cases, but, along with the postoperative state itself, also appears to be a risk factor for this unusual condition. Postoperative bilateral adrenal hemorrhage is rare and potentially fatal. Early identification and prompt initiation of steroid replacement therapy and reversal of coagulopathies can prove to be lifesaving. Making this diagnosis can be very challenging, as the clinical presentation and laboratory findings of adrenal hemorrhage are vague and nonspecific and mimic many nonlife threatening postoperative complications. Radiological diagnosis by CT may initially be normal and thus further confound the diagnosis. Hence, providers should remain vigilant for associated complications even with low‐dose prophylactic heparin or low‐molecular‐weight heparin in postoperative patients, and prompt, presumptive treatment with corticosteroids should be started while awaiting confirmation by imaging and laboratory testing.

A 52‐year‐old man presented to the emergency department (ED) from a skilled nursing facility with a complaint of bilateral upper‐quadrant abdominal pain of 48 hours' duration. The pain was sharp, nonradiating, constant, and was associated with nausea, vomiting, and constipation. The patient denied any fever, back pain, dysuria, melena, or hematochezia. In the rehabilitation facility the patient had been initially evaluated for this pain. He was given laxatives and stool softeners for presumed constipation but these measures had not been effective. A computed tomography (CT) scan of the abdomen had only showed stool in the colon and he was sent to the ED for further evaluation.

Apart from severe degenerative joint disease in both his knees he was in good health. He was in the skilled nursing facility (SNF) for rehabilitation for bilateral knee replacement surgery done 9 days prior to this presentation. His postoperative course was unremarkable. He had been maintained on prophylaxis for venous thromboembolism with enoxaparin since postoperative day 1 at a daily dose of 40 mg subcutaneously, and was transferred to the SNF on postoperative day 6 on the same dose. His was receiving oxycodone and Tylenol for pain. He was on no other medications.

Vital signs on presentation revealed a temperature of 97.5F, a heart rate of 100 beats per minute, a respiratory rate of 16 breaths per minute, and a blood pressure of 136/69 mmHg. He was alert and oriented and in mild distress from the abdominal pain. Examination was normal except for tenderness in the upper quadrants of the abdomen though no rigidity or rebound tenderness were noted. Routine chemistries were normal except for sodium of 134 mg/dL. His white count, hemoglobin, hematocrit, and platelet levels were noted to be at 17.5K/L, 10 g/dL, 30%, and 345K/L, respectively, and were stable with regard to his discharge laboratory values. His serum eosinophil level was normal. A complete workup for hypercoagulable state and bleeding disorders including assays for antibodies associated with heparin‐induced thrombocytopenia were negative. He was admitted for further evaluation and treatment.

The patient had another CT scan of the abdomen (Figure 1), which when compared to the one done at the SNF 2 days prior showed markedly enlarged bilateral adrenal glands suggestive of bilateral acute adrenal hemorrhage. The enoxaparin was discontinued and empiric steroid replacement therapy was begun. A random cortisol level was normal but a cosyntropin stimulation test showed an absolute increase in cortisol level of only 0.8 g/dL at both 30 and 60 minutes after administration of 250 g of cosyntropin. An investigation was undertaken to determine if the patient had any prior risk factors for bleeding. There was no evidence of infection and a comprehensive evaluation for bleeding, and coagulation disorders was normal. The bilateral adrenal hemorrhage was attributed to the use of enoxaparin in the postoperative setting. Unfortunately, the patient subsequently developed a deep venous thrombosis in his lower extremity and an inferior vena cava (IVC) filter was placed before discharge. He was doing well 6 months later, and is still continued on glucocorticoid and mineralocorticoid replacement therapy and follows up with endocrinology as an outpatient.

Figure 1

Discussion

Bilateral adrenal hemorrhage is usually associated with massive sepsis from Gram‐negative organisms such as Neisseria meningitides, Pseudomonas aeroginosa, Escherichia coli, and Bacteroides fragilis. Rupert Waterhouse, in 1911, was the first person to describe a patient with severe meningococcal sepsis resulting in acute adrenal hemorrhage and collapse. This was also later described independently by Carl Friderichsen in 1918, and is now referred to as the Waterhouse‐Friderichsen syndrome. Other causes include antiphospholipid antibody syndrome, heparin‐associated thrombocytopenia (HIT), and severe physical stress. Bilateral adrenal hemorrhage can also spontaneously occur in the postoperative period, especially after cardiothoracic or orthopedic surgery. This phenomenon may be related to the frequent use of prophylactic anticoagulants after these types of procedures.

The first case report of bilateral adrenal hemorrhage secondary to use of anticoagulants was described in 1947, and the first case report of successful resuscitation after corticosteroid administration in a patient with bilateral adrenal hemorrhage secondary to anticoagulant use was described by Thorn in 1956.1 A review of the literature demonstrates multiple case reports of adrenal hemorrhage reported in the postoperative period, particularly after joint arthroplasty, and especially after knee replacement surgeries. Most of the recent cases have been associated with use of prophylactic low‐dose heparin or low‐molecular‐weight heparin at the time of adrenal hemorrhage. In a study of 157 case reports of individuals with bilateral hemorrhage (including 22 autopsies), 48 cases were associated with administration of anticoagulants, although the dose and effect were not specified.2 Amador et al.1 showed that out of 4325 autopsies performed from 1949 to 1962 in their institution, 30 cases were found of bilateral hemorrhage, of which 10 were receiving heparin at presumably prophylactic doses; 5 of these patients were also receiving dicumarol.

Mayo Clinic investigators performed a retrospective review of all cases of adrenal hemorrhage over a period of 25 years at their hospital, and found 141 cases of adrenal hemorrhage, of which 78 were bilateral and 63 were unilateral,3 and in 67 patients the condition was diagnosed at autopsy. In this study 14 patients had adrenal hemorrhage in the postoperative period in the absence of lupus anticoagulant or HIT; there was no specific mention in this study of the use of postoperative anticoagulants. Finally, a multicenter case control study was undertaken by Kovacs et al.4 to assess putative risk factors for development of bilateral massive adrenal hemorrhage. In the multivariate analysis, thrombocytopenia, exposure to heparin, and sepsis were found to be strongly associated with risk of hemorrhage. Of 23 patients with bilateral, massive adrenal hemorrhage, 16 had been exposed to heparin, and at least 6 were on exclusively subcutaneous heparin. The authors concluded that heparin exposure was a much bigger risk factor than other coagulopathies, and those exposed to heparin of any route or type for 4 to 6 days and those exposed for more than 6 days were about 17 and 34 times, respectively, more likely to develop bilateral hemorrhage than those who had less than 4 days or no exposure.

The clinical presentation of adrenal insufficiency due to bilateral adrenal hemorrhage is often nonspecific. Symptoms may include abdominal pain, back pain, fever, nausea, vomiting, weakness, obtundation, confusion, and hypotensionall of which are also common postoperative symptoms and can be missed or ignored.5 Rao et al.6 profiled the clinical presentation of 64 cases of bilateral hemorrhage and found the following: abdominal, flank, back, or chest pain (86%); anorexia, nausea, or vomiting (47%); psychiatric symptoms (42%); fever (66%); hypotension recognized before shock episode (19%); and abdominal rigidity or rebound (22%). Adrenal insufficiency becomes clinically evident once 90% of the gland is destroyed. About 50% of patients do not manifest typical laboratory abnormalities, so a high degree of suspicion is necessary to diagnose the condition.3 Also, the laboratory diagnosis of adrenal insufficiency using random cortisol levels is unreliable, as reference ranges in patients experiencing stress (as in the postoperative period) have not been well studied or established. In patients with bilateral hemorrhage postoperatively on prophylactic anticoagulants, the coagulation profile is usually within normal limits and there is typically no evidence of spontaneous bleeding elsewhere. In later stages, the typical laboratory findings of abnormal adrenal function such as hypokalemia, hyponatremia, declining cortisol levels, and an inappropriate response to adrenocorticotropic hormone stimulation test may be seen. A significant drop in hemoglobin secondary to hemorrhage may also be encountered in some patients secondary to the bleed.

CT is the most reliable and extensively used imaging modality for making the diagnosis, although magnetic resonance imaging (MRI) or ultrasound may also be utilized. Early in the course of adrenal hemorrhage, CT findings may be negative, and repeated imaging is appropriate when clinical suspicion is high. The presence of bilateral adrenal enlargement with increased signal attenuation suggests bilateral adrenal hemorrhage. MRI can both characterize adrenal hematomas, and estimate their age.7, 8

Postoperative adrenal hemorrhage and insufficiency is easily treatable and has excellent outcomes; survivors will need lifelong corticosteroid replacement (and usually mineralocorticoid replacement as well). In the Mayo Clinic study, survival was 100% with treatment vs. 17% without treatment. In comparison, sepsis‐induced or stress‐induced adrenal insufficiency has poor outcomes despite adequate treatment (9% survival with treatment vs. 6% survival without treatment).3 Death can occur within hours to days of symptoms if untreated. Treatment includes timely initiation of adrenal hormone replacement and reversal of coagulopathies.

Postoperative venous thromboembolism (VTE) prophylaxis with anticoagulants is the appropriate care in many cases, but, along with the postoperative state itself, also appears to be a risk factor for this unusual condition. Postoperative bilateral adrenal hemorrhage is rare and potentially fatal. Early identification and prompt initiation of steroid replacement therapy and reversal of coagulopathies can prove to be lifesaving. Making this diagnosis can be very challenging, as the clinical presentation and laboratory findings of adrenal hemorrhage are vague and nonspecific and mimic many nonlife threatening postoperative complications. Radiological diagnosis by CT may initially be normal and thus further confound the diagnosis. Hence, providers should remain vigilant for associated complications even with low‐dose prophylactic heparin or low‐molecular‐weight heparin in postoperative patients, and prompt, presumptive treatment with corticosteroids should be started while awaiting confirmation by imaging and laboratory testing.

A 52‐year‐old man presented to the emergency department (ED) from a skilled nursing facility with a complaint of bilateral upper‐quadrant abdominal pain of 48 hours' duration. The pain was sharp, nonradiating, constant, and was associated with nausea, vomiting, and constipation. The patient denied any fever, back pain, dysuria, melena, or hematochezia. In the rehabilitation facility the patient had been initially evaluated for this pain. He was given laxatives and stool softeners for presumed constipation but these measures had not been effective. A computed tomography (CT) scan of the abdomen had only showed stool in the colon and he was sent to the ED for further evaluation.

Apart from severe degenerative joint disease in both his knees he was in good health. He was in the skilled nursing facility (SNF) for rehabilitation for bilateral knee replacement surgery done 9 days prior to this presentation. His postoperative course was unremarkable. He had been maintained on prophylaxis for venous thromboembolism with enoxaparin since postoperative day 1 at a daily dose of 40 mg subcutaneously, and was transferred to the SNF on postoperative day 6 on the same dose. His was receiving oxycodone and Tylenol for pain. He was on no other medications.

Vital signs on presentation revealed a temperature of 97.5F, a heart rate of 100 beats per minute, a respiratory rate of 16 breaths per minute, and a blood pressure of 136/69 mmHg. He was alert and oriented and in mild distress from the abdominal pain. Examination was normal except for tenderness in the upper quadrants of the abdomen though no rigidity or rebound tenderness were noted. Routine chemistries were normal except for sodium of 134 mg/dL. His white count, hemoglobin, hematocrit, and platelet levels were noted to be at 17.5K/L, 10 g/dL, 30%, and 345K/L, respectively, and were stable with regard to his discharge laboratory values. His serum eosinophil level was normal. A complete workup for hypercoagulable state and bleeding disorders including assays for antibodies associated with heparin‐induced thrombocytopenia were negative. He was admitted for further evaluation and treatment.

The patient had another CT scan of the abdomen (Figure 1), which when compared to the one done at the SNF 2 days prior showed markedly enlarged bilateral adrenal glands suggestive of bilateral acute adrenal hemorrhage. The enoxaparin was discontinued and empiric steroid replacement therapy was begun. A random cortisol level was normal but a cosyntropin stimulation test showed an absolute increase in cortisol level of only 0.8 g/dL at both 30 and 60 minutes after administration of 250 g of cosyntropin. An investigation was undertaken to determine if the patient had any prior risk factors for bleeding. There was no evidence of infection and a comprehensive evaluation for bleeding, and coagulation disorders was normal. The bilateral adrenal hemorrhage was attributed to the use of enoxaparin in the postoperative setting. Unfortunately, the patient subsequently developed a deep venous thrombosis in his lower extremity and an inferior vena cava (IVC) filter was placed before discharge. He was doing well 6 months later, and is still continued on glucocorticoid and mineralocorticoid replacement therapy and follows up with endocrinology as an outpatient.

Figure 1

Discussion

Bilateral adrenal hemorrhage is usually associated with massive sepsis from Gram‐negative organisms such as Neisseria meningitides, Pseudomonas aeroginosa, Escherichia coli, and Bacteroides fragilis. Rupert Waterhouse, in 1911, was the first person to describe a patient with severe meningococcal sepsis resulting in acute adrenal hemorrhage and collapse. This was also later described independently by Carl Friderichsen in 1918, and is now referred to as the Waterhouse‐Friderichsen syndrome. Other causes include antiphospholipid antibody syndrome, heparin‐associated thrombocytopenia (HIT), and severe physical stress. Bilateral adrenal hemorrhage can also spontaneously occur in the postoperative period, especially after cardiothoracic or orthopedic surgery. This phenomenon may be related to the frequent use of prophylactic anticoagulants after these types of procedures.

The first case report of bilateral adrenal hemorrhage secondary to use of anticoagulants was described in 1947, and the first case report of successful resuscitation after corticosteroid administration in a patient with bilateral adrenal hemorrhage secondary to anticoagulant use was described by Thorn in 1956.1 A review of the literature demonstrates multiple case reports of adrenal hemorrhage reported in the postoperative period, particularly after joint arthroplasty, and especially after knee replacement surgeries. Most of the recent cases have been associated with use of prophylactic low‐dose heparin or low‐molecular‐weight heparin at the time of adrenal hemorrhage. In a study of 157 case reports of individuals with bilateral hemorrhage (including 22 autopsies), 48 cases were associated with administration of anticoagulants, although the dose and effect were not specified.2 Amador et al.1 showed that out of 4325 autopsies performed from 1949 to 1962 in their institution, 30 cases were found of bilateral hemorrhage, of which 10 were receiving heparin at presumably prophylactic doses; 5 of these patients were also receiving dicumarol.

Mayo Clinic investigators performed a retrospective review of all cases of adrenal hemorrhage over a period of 25 years at their hospital, and found 141 cases of adrenal hemorrhage, of which 78 were bilateral and 63 were unilateral,3 and in 67 patients the condition was diagnosed at autopsy. In this study 14 patients had adrenal hemorrhage in the postoperative period in the absence of lupus anticoagulant or HIT; there was no specific mention in this study of the use of postoperative anticoagulants. Finally, a multicenter case control study was undertaken by Kovacs et al.4 to assess putative risk factors for development of bilateral massive adrenal hemorrhage. In the multivariate analysis, thrombocytopenia, exposure to heparin, and sepsis were found to be strongly associated with risk of hemorrhage. Of 23 patients with bilateral, massive adrenal hemorrhage, 16 had been exposed to heparin, and at least 6 were on exclusively subcutaneous heparin. The authors concluded that heparin exposure was a much bigger risk factor than other coagulopathies, and those exposed to heparin of any route or type for 4 to 6 days and those exposed for more than 6 days were about 17 and 34 times, respectively, more likely to develop bilateral hemorrhage than those who had less than 4 days or no exposure.

The clinical presentation of adrenal insufficiency due to bilateral adrenal hemorrhage is often nonspecific. Symptoms may include abdominal pain, back pain, fever, nausea, vomiting, weakness, obtundation, confusion, and hypotensionall of which are also common postoperative symptoms and can be missed or ignored.5 Rao et al.6 profiled the clinical presentation of 64 cases of bilateral hemorrhage and found the following: abdominal, flank, back, or chest pain (86%); anorexia, nausea, or vomiting (47%); psychiatric symptoms (42%); fever (66%); hypotension recognized before shock episode (19%); and abdominal rigidity or rebound (22%). Adrenal insufficiency becomes clinically evident once 90% of the gland is destroyed. About 50% of patients do not manifest typical laboratory abnormalities, so a high degree of suspicion is necessary to diagnose the condition.3 Also, the laboratory diagnosis of adrenal insufficiency using random cortisol levels is unreliable, as reference ranges in patients experiencing stress (as in the postoperative period) have not been well studied or established. In patients with bilateral hemorrhage postoperatively on prophylactic anticoagulants, the coagulation profile is usually within normal limits and there is typically no evidence of spontaneous bleeding elsewhere. In later stages, the typical laboratory findings of abnormal adrenal function such as hypokalemia, hyponatremia, declining cortisol levels, and an inappropriate response to adrenocorticotropic hormone stimulation test may be seen. A significant drop in hemoglobin secondary to hemorrhage may also be encountered in some patients secondary to the bleed.

CT is the most reliable and extensively used imaging modality for making the diagnosis, although magnetic resonance imaging (MRI) or ultrasound may also be utilized. Early in the course of adrenal hemorrhage, CT findings may be negative, and repeated imaging is appropriate when clinical suspicion is high. The presence of bilateral adrenal enlargement with increased signal attenuation suggests bilateral adrenal hemorrhage. MRI can both characterize adrenal hematomas, and estimate their age.7, 8

Postoperative adrenal hemorrhage and insufficiency is easily treatable and has excellent outcomes; survivors will need lifelong corticosteroid replacement (and usually mineralocorticoid replacement as well). In the Mayo Clinic study, survival was 100% with treatment vs. 17% without treatment. In comparison, sepsis‐induced or stress‐induced adrenal insufficiency has poor outcomes despite adequate treatment (9% survival with treatment vs. 6% survival without treatment).3 Death can occur within hours to days of symptoms if untreated. Treatment includes timely initiation of adrenal hormone replacement and reversal of coagulopathies.

Postoperative venous thromboembolism (VTE) prophylaxis with anticoagulants is the appropriate care in many cases, but, along with the postoperative state itself, also appears to be a risk factor for this unusual condition. Postoperative bilateral adrenal hemorrhage is rare and potentially fatal. Early identification and prompt initiation of steroid replacement therapy and reversal of coagulopathies can prove to be lifesaving. Making this diagnosis can be very challenging, as the clinical presentation and laboratory findings of adrenal hemorrhage are vague and nonspecific and mimic many nonlife threatening postoperative complications. Radiological diagnosis by CT may initially be normal and thus further confound the diagnosis. Hence, providers should remain vigilant for associated complications even with low‐dose prophylactic heparin or low‐molecular‐weight heparin in postoperative patients, and prompt, presumptive treatment with corticosteroids should be started while awaiting confirmation by imaging and laboratory testing.

Gastric outlet obstruction (GOO) is frequently a diagnostic dilemma as malignancies have surpassed benign diseases as etiologies of GOO.1, 2 A case is presented of a previously healthy patient with persistent vomiting who was sequentially diagnosed with peptic ulcer disease (PUD), pancreatitis, and cholecystitis. Unfortunately, the diagnostic workup was less straightforward than initially suspected, as she was eventually diagnosed with a malignant GOO.

To the best of our knowledge, this is the second case in which GOO was the presenting manifestation of a previously undiagnosed metastatic lobular breast carcinoma.1 Although gastrointestinal involvement may be a late manifestation of metastatic breast carcinoma that almost always follows after spread to other sites, this case illustrates GOO as a presenting manifestation.3 Perhaps the most salient teaching point, consistent with recent literature, is that endoscopic biopsies in the workup of malignant GOO are often misleading and delay diagnosis.4

Case Report

A 44‐year‐old female with a history of fibrocystic breast disease presented with 1 month of right upper quadrant abdominal pain and nonbilious emesis of undigested food occurring several hours postprandially. Gallstones were demonstrated on an abdominal sonogram, and esophagogastroduodenoscopy revealed a normal proximal esophagus, distal esophagitis, and a Schatzki ring. An 8‐mm duodenal bulb ulcer was biopsied with benign results, but no duodenal obstruction or narrowing was visualized. Therapy for presumptive Helicobacter pylori infection and PUD was initiated, but she was hospitalized shortly later with persistent vomiting and acute renal failure. An abdominal computed tomography (CT) scan showed evidence of pancreatitis but a normal biliary system and normal small and large bowels. There was no clinical jaundice, and hepatic function tests were normal. After medical therapy, an open cholecystectomy was performed because of dense adhesions of the large and small bowels to the liver and gallbladder. No gross abnormalities of the common bile duct or intestines were described.

With the persistence of intractable vomiting, abdominal and pelvic CT scans were repeated and revealed duodenal thickening and luminal narrowing. Follow‐up abdominal magnetic resonance imaging (Figure 1) demonstrated an infiltrating duodenal mass with a resultant high‐grade GOO. Histological examination from repeat endoscopic biopsies of the duodenal mass revealed signet ring cells infiltrating the lamina propria with positive estrogen and progesterone receptors. Upon presentation to the surgical service, a physical examination revealed a right breast mass, and a subsequent breast biopsy diagnosed lobular breast carcinoma with 10% estrogen receptor and 5% progesterone receptor and histology identical to that of the duodenal mass. Immunohistochemical assays confirmed primary breast carcinoma with duodenal metastasis. A lumpectomy was performed, and docetaxel was initiated for stage IV invasive lobular breast carcinoma. A gastrostomy and duodenal stents were placed for the GOO with resolution of the vomiting, and she received hospice and palliative care.

Figure 1

Discussion

Prior to the advent of histamine blockers, PUD was the most common cause of GOO.1 Currently, malignancy accounts for 60% of cases of GOO.13 To the best of our knowledge, this patient's presentation represents the second case in which the initial manifestation of an undiagnosed metastatic breast disease was GOO.1 The diagnosis was both challenging and unusual because of the patient's other organic diseases (gallbladder disease, pancreatic disease, and PUD), which distracted from the underlying diagnosis of malignant GOO.

GOO is a clinical syndrome of diverse etiologies and pathogenesis that is characterized by mechanical impediment of gastric emptying. GOO is divided into 2 well‐defined groups: benign and malignant causes.14 Benign causes include: PUD, gastric polyps, pyloric stenosis, congenital duodenal webs, gallstone obstruction, pancreatic pseudocysts, and bezoars. Gastric cancer is the most common malignant cause and is followed by duodenal carcinoma, pancreatic carcinoma, cholangiocarcinoma, and metastatic disease of the gastric outlet.

Because of the stomach's significant capacity to distend, malignant GOO is often undetected clinically until a high‐grade obstruction develops.13 In patients with a previous diagnosis of carcinoma, nausea and vomiting can be mistakenly attributed to radiation or chemotherapy. Characteristic symptoms include nonbilious emesis of undigested food, early satiety, epigastric fullness, and abdominal pain.13

This case presentation is consistent with recent literature, which reports a surprising lack of reliability in diagnosing malignant GOO by endoscopy.4 In 1 study, endoscopic biopsy for detection of malignant GOO was associated with poor sensitivity (37%) in comparison with surgical biopsy.4 The authors recommended that patients who have a high clinical suspicion of malignant GOO (older patients and those without PUD) and who have initially benign biopsies be considered for surgical exploration prior to medical therapy or undergo repeat endoscopy with jumbo biopsies.4

Alternatively, endoscopic ultrasound has been advocated for detecting early mucosal gastric or duodenal cancer in patients without ulcerous changes on endoscopy. Radiographic features associated with submucosal tumor infiltration include irregular narrowing and budding signs.5 As conventional CT is considered suboptimal for the detection of gastric or intestinal carcinomas, multidetector‐row CT with multiplanar reconstruction has enhanced the overall ability to detect early gastric cancer and advanced gastric cancers with a sensitivity of 96.2%.6

Conclusion

The preceding case demonstrates valuable teaching points encountered in diagnosing malignant GOO. In patients with intractable vomiting severe enough to produce renal failure, other organic causes should be considered before one proceeds directly to cholecystectomy. This case further confirms that endoscopic biopsy is often alarmingly inadequate for diagnosing malignant GOO.46 Thus, if worrisome symptoms persist, the provider should not be comforted by normal endoscopy and should escalate investigations accordingly. More clinical trials should evaluate the roles of endoscopic ultrasound and multidetector‐row CT in the early detection of gastric and duodenal carcinoma in patients with normal endoscopic biopsies, which may have led to a more timely diagnosis in this patient.5, 6

Gastric outlet obstruction (GOO) is frequently a diagnostic dilemma as malignancies have surpassed benign diseases as etiologies of GOO.1, 2 A case is presented of a previously healthy patient with persistent vomiting who was sequentially diagnosed with peptic ulcer disease (PUD), pancreatitis, and cholecystitis. Unfortunately, the diagnostic workup was less straightforward than initially suspected, as she was eventually diagnosed with a malignant GOO.

To the best of our knowledge, this is the second case in which GOO was the presenting manifestation of a previously undiagnosed metastatic lobular breast carcinoma.1 Although gastrointestinal involvement may be a late manifestation of metastatic breast carcinoma that almost always follows after spread to other sites, this case illustrates GOO as a presenting manifestation.3 Perhaps the most salient teaching point, consistent with recent literature, is that endoscopic biopsies in the workup of malignant GOO are often misleading and delay diagnosis.4

Case Report

A 44‐year‐old female with a history of fibrocystic breast disease presented with 1 month of right upper quadrant abdominal pain and nonbilious emesis of undigested food occurring several hours postprandially. Gallstones were demonstrated on an abdominal sonogram, and esophagogastroduodenoscopy revealed a normal proximal esophagus, distal esophagitis, and a Schatzki ring. An 8‐mm duodenal bulb ulcer was biopsied with benign results, but no duodenal obstruction or narrowing was visualized. Therapy for presumptive Helicobacter pylori infection and PUD was initiated, but she was hospitalized shortly later with persistent vomiting and acute renal failure. An abdominal computed tomography (CT) scan showed evidence of pancreatitis but a normal biliary system and normal small and large bowels. There was no clinical jaundice, and hepatic function tests were normal. After medical therapy, an open cholecystectomy was performed because of dense adhesions of the large and small bowels to the liver and gallbladder. No gross abnormalities of the common bile duct or intestines were described.

With the persistence of intractable vomiting, abdominal and pelvic CT scans were repeated and revealed duodenal thickening and luminal narrowing. Follow‐up abdominal magnetic resonance imaging (Figure 1) demonstrated an infiltrating duodenal mass with a resultant high‐grade GOO. Histological examination from repeat endoscopic biopsies of the duodenal mass revealed signet ring cells infiltrating the lamina propria with positive estrogen and progesterone receptors. Upon presentation to the surgical service, a physical examination revealed a right breast mass, and a subsequent breast biopsy diagnosed lobular breast carcinoma with 10% estrogen receptor and 5% progesterone receptor and histology identical to that of the duodenal mass. Immunohistochemical assays confirmed primary breast carcinoma with duodenal metastasis. A lumpectomy was performed, and docetaxel was initiated for stage IV invasive lobular breast carcinoma. A gastrostomy and duodenal stents were placed for the GOO with resolution of the vomiting, and she received hospice and palliative care.

Figure 1

Discussion

Prior to the advent of histamine blockers, PUD was the most common cause of GOO.1 Currently, malignancy accounts for 60% of cases of GOO.13 To the best of our knowledge, this patient's presentation represents the second case in which the initial manifestation of an undiagnosed metastatic breast disease was GOO.1 The diagnosis was both challenging and unusual because of the patient's other organic diseases (gallbladder disease, pancreatic disease, and PUD), which distracted from the underlying diagnosis of malignant GOO.

GOO is a clinical syndrome of diverse etiologies and pathogenesis that is characterized by mechanical impediment of gastric emptying. GOO is divided into 2 well‐defined groups: benign and malignant causes.14 Benign causes include: PUD, gastric polyps, pyloric stenosis, congenital duodenal webs, gallstone obstruction, pancreatic pseudocysts, and bezoars. Gastric cancer is the most common malignant cause and is followed by duodenal carcinoma, pancreatic carcinoma, cholangiocarcinoma, and metastatic disease of the gastric outlet.

Because of the stomach's significant capacity to distend, malignant GOO is often undetected clinically until a high‐grade obstruction develops.13 In patients with a previous diagnosis of carcinoma, nausea and vomiting can be mistakenly attributed to radiation or chemotherapy. Characteristic symptoms include nonbilious emesis of undigested food, early satiety, epigastric fullness, and abdominal pain.13

This case presentation is consistent with recent literature, which reports a surprising lack of reliability in diagnosing malignant GOO by endoscopy.4 In 1 study, endoscopic biopsy for detection of malignant GOO was associated with poor sensitivity (37%) in comparison with surgical biopsy.4 The authors recommended that patients who have a high clinical suspicion of malignant GOO (older patients and those without PUD) and who have initially benign biopsies be considered for surgical exploration prior to medical therapy or undergo repeat endoscopy with jumbo biopsies.4

Alternatively, endoscopic ultrasound has been advocated for detecting early mucosal gastric or duodenal cancer in patients without ulcerous changes on endoscopy. Radiographic features associated with submucosal tumor infiltration include irregular narrowing and budding signs.5 As conventional CT is considered suboptimal for the detection of gastric or intestinal carcinomas, multidetector‐row CT with multiplanar reconstruction has enhanced the overall ability to detect early gastric cancer and advanced gastric cancers with a sensitivity of 96.2%.6

Conclusion

The preceding case demonstrates valuable teaching points encountered in diagnosing malignant GOO. In patients with intractable vomiting severe enough to produce renal failure, other organic causes should be considered before one proceeds directly to cholecystectomy. This case further confirms that endoscopic biopsy is often alarmingly inadequate for diagnosing malignant GOO.46 Thus, if worrisome symptoms persist, the provider should not be comforted by normal endoscopy and should escalate investigations accordingly. More clinical trials should evaluate the roles of endoscopic ultrasound and multidetector‐row CT in the early detection of gastric and duodenal carcinoma in patients with normal endoscopic biopsies, which may have led to a more timely diagnosis in this patient.5, 6

Gastric outlet obstruction (GOO) is frequently a diagnostic dilemma as malignancies have surpassed benign diseases as etiologies of GOO.1, 2 A case is presented of a previously healthy patient with persistent vomiting who was sequentially diagnosed with peptic ulcer disease (PUD), pancreatitis, and cholecystitis. Unfortunately, the diagnostic workup was less straightforward than initially suspected, as she was eventually diagnosed with a malignant GOO.

To the best of our knowledge, this is the second case in which GOO was the presenting manifestation of a previously undiagnosed metastatic lobular breast carcinoma.1 Although gastrointestinal involvement may be a late manifestation of metastatic breast carcinoma that almost always follows after spread to other sites, this case illustrates GOO as a presenting manifestation.3 Perhaps the most salient teaching point, consistent with recent literature, is that endoscopic biopsies in the workup of malignant GOO are often misleading and delay diagnosis.4

Case Report

A 44‐year‐old female with a history of fibrocystic breast disease presented with 1 month of right upper quadrant abdominal pain and nonbilious emesis of undigested food occurring several hours postprandially. Gallstones were demonstrated on an abdominal sonogram, and esophagogastroduodenoscopy revealed a normal proximal esophagus, distal esophagitis, and a Schatzki ring. An 8‐mm duodenal bulb ulcer was biopsied with benign results, but no duodenal obstruction or narrowing was visualized. Therapy for presumptive Helicobacter pylori infection and PUD was initiated, but she was hospitalized shortly later with persistent vomiting and acute renal failure. An abdominal computed tomography (CT) scan showed evidence of pancreatitis but a normal biliary system and normal small and large bowels. There was no clinical jaundice, and hepatic function tests were normal. After medical therapy, an open cholecystectomy was performed because of dense adhesions of the large and small bowels to the liver and gallbladder. No gross abnormalities of the common bile duct or intestines were described.

With the persistence of intractable vomiting, abdominal and pelvic CT scans were repeated and revealed duodenal thickening and luminal narrowing. Follow‐up abdominal magnetic resonance imaging (Figure 1) demonstrated an infiltrating duodenal mass with a resultant high‐grade GOO. Histological examination from repeat endoscopic biopsies of the duodenal mass revealed signet ring cells infiltrating the lamina propria with positive estrogen and progesterone receptors. Upon presentation to the surgical service, a physical examination revealed a right breast mass, and a subsequent breast biopsy diagnosed lobular breast carcinoma with 10% estrogen receptor and 5% progesterone receptor and histology identical to that of the duodenal mass. Immunohistochemical assays confirmed primary breast carcinoma with duodenal metastasis. A lumpectomy was performed, and docetaxel was initiated for stage IV invasive lobular breast carcinoma. A gastrostomy and duodenal stents were placed for the GOO with resolution of the vomiting, and she received hospice and palliative care.

Figure 1

Discussion

Prior to the advent of histamine blockers, PUD was the most common cause of GOO.1 Currently, malignancy accounts for 60% of cases of GOO.13 To the best of our knowledge, this patient's presentation represents the second case in which the initial manifestation of an undiagnosed metastatic breast disease was GOO.1 The diagnosis was both challenging and unusual because of the patient's other organic diseases (gallbladder disease, pancreatic disease, and PUD), which distracted from the underlying diagnosis of malignant GOO.

GOO is a clinical syndrome of diverse etiologies and pathogenesis that is characterized by mechanical impediment of gastric emptying. GOO is divided into 2 well‐defined groups: benign and malignant causes.14 Benign causes include: PUD, gastric polyps, pyloric stenosis, congenital duodenal webs, gallstone obstruction, pancreatic pseudocysts, and bezoars. Gastric cancer is the most common malignant cause and is followed by duodenal carcinoma, pancreatic carcinoma, cholangiocarcinoma, and metastatic disease of the gastric outlet.

Because of the stomach's significant capacity to distend, malignant GOO is often undetected clinically until a high‐grade obstruction develops.13 In patients with a previous diagnosis of carcinoma, nausea and vomiting can be mistakenly attributed to radiation or chemotherapy. Characteristic symptoms include nonbilious emesis of undigested food, early satiety, epigastric fullness, and abdominal pain.13

This case presentation is consistent with recent literature, which reports a surprising lack of reliability in diagnosing malignant GOO by endoscopy.4 In 1 study, endoscopic biopsy for detection of malignant GOO was associated with poor sensitivity (37%) in comparison with surgical biopsy.4 The authors recommended that patients who have a high clinical suspicion of malignant GOO (older patients and those without PUD) and who have initially benign biopsies be considered for surgical exploration prior to medical therapy or undergo repeat endoscopy with jumbo biopsies.4

Alternatively, endoscopic ultrasound has been advocated for detecting early mucosal gastric or duodenal cancer in patients without ulcerous changes on endoscopy. Radiographic features associated with submucosal tumor infiltration include irregular narrowing and budding signs.5 As conventional CT is considered suboptimal for the detection of gastric or intestinal carcinomas, multidetector‐row CT with multiplanar reconstruction has enhanced the overall ability to detect early gastric cancer and advanced gastric cancers with a sensitivity of 96.2%.6

Conclusion

The preceding case demonstrates valuable teaching points encountered in diagnosing malignant GOO. In patients with intractable vomiting severe enough to produce renal failure, other organic causes should be considered before one proceeds directly to cholecystectomy. This case further confirms that endoscopic biopsy is often alarmingly inadequate for diagnosing malignant GOO.46 Thus, if worrisome symptoms persist, the provider should not be comforted by normal endoscopy and should escalate investigations accordingly. More clinical trials should evaluate the roles of endoscopic ultrasound and multidetector‐row CT in the early detection of gastric and duodenal carcinoma in patients with normal endoscopic biopsies, which may have led to a more timely diagnosis in this patient.5, 6

Patient satisfaction is an important issue for hospitals, as it may affect the decision to seek care at one institution over another, but it may soon have direct implications for hospital reimbursement with the recent proposals for Value Based Purchasing (VBP) models by the Centers for Medicare and Medicaid Services (CMS). Up to 5% of inpatient Medicare reimbursement would be linked to performance measures, 40% of which could come from percentile outcomes on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) alone.1 HCAHPS scores are now available for individual hospitals at the Hospital Compare website maintained by CMS (http://www.hospitalcompare.hhs.gov). Hospitalists will likely be held accountable by administrators for poor performance on the survey.

While the information garnered from the HCAHPS provides an external perception of hospital quality, the questions are broad and do not identify specific reasons for reduced satisfaction. Many have suggested that the incorporation of surveys already administered at hospitals may be required for successful HCAHPS administration in order to overcome the limitations inherent in its design.2

In order to identify explanatory factors for low HCAHPS scores, we decided to incorporate a technique known as a cognitive interview (CI). The CI is widely used as an evaluative tool for survey questions because of its ability to allow the interviewer to discern the processes that lead to responses.3 Up to this point, the focus of this CI method has been on the ability of the subject to comprehend and answer the questions.4 However, when a CI subject is answering a question, there is a large amount of information presented to the interviewer about the topic that is typically regarded as supplementary because of the focus on these specific issues.5 This study reports the supplemental information that may provide insight as to why patients answered as they did. Our goal was to gain further understanding of factors that may underlie HCAHPS responses.

Materials and Methods

Results

Discussion

Because of the limitations inherent in typical survey measures of investigation, we employed a CI technique. The objective was to generate hypotheses as to why patients may report reduced satisfaction with various aspects of hospitalization. The data set gleaned from this study was sizable, and held information pertinent to all parts of the hospital. The results reported in this article are focused on hospitalists.

The quantity of time hospitalists spend with patients may not be as important to the patients as the quality of interaction. Other studies have shown similar results, illustrating that the key factor in a patients' opinion of a visit with a physician was the perception of being taken seriously.10 The perception of being talked down to (as noted by Levy11) and the perception of not being responded to or focused on are the key negative factors for patients. By interacting with patients in a way that makes them feel valued, focused on, and responded to, physicians may improve patient satisfaction without requiring increased amounts of time spent with them.

Patient complaints that physicians do not respond to them may partially reflect inability to ask all of their questions. Hospitals may wish to consider methods to improve the ability of the patients to ask questions in the small amount of time available for physicians to talk with them (eg, placing notebooks by patients' bedsides and repeatedly encouraging them to write their questions down). This concept has also been endorsed as a means of improving the quality of care and reducing medical mistakes.12

Hospitalists and other physicians may positively impact the communication about medication issues by establishing a protocol for communicating major as well as minor side effects or even the lack of side effects of medication to patients. The challenge lies in determining the level of incidence that is significant enough (eg, 10%, 5%, or 1%) to warrant explaining minor side effects.

The findings of this study are limited by its small size, qualitative nature, nonrepresentative population, and loose approach to questioning. The patient responses were transcribed manually by the interviewer, leaving the possibility that some responses were not accurately recorded. Our survey instrument is not validated, and the results may not generalize to other settings. Instead of taking these findings as conclusive, we present them as suggestions for improvement that may be validated by further research.

Patient satisfaction is an important issue for hospitals, as it may affect the decision to seek care at one institution over another, but it may soon have direct implications for hospital reimbursement with the recent proposals for Value Based Purchasing (VBP) models by the Centers for Medicare and Medicaid Services (CMS). Up to 5% of inpatient Medicare reimbursement would be linked to performance measures, 40% of which could come from percentile outcomes on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) alone.1 HCAHPS scores are now available for individual hospitals at the Hospital Compare website maintained by CMS (http://www.hospitalcompare.hhs.gov). Hospitalists will likely be held accountable by administrators for poor performance on the survey.

While the information garnered from the HCAHPS provides an external perception of hospital quality, the questions are broad and do not identify specific reasons for reduced satisfaction. Many have suggested that the incorporation of surveys already administered at hospitals may be required for successful HCAHPS administration in order to overcome the limitations inherent in its design.2

In order to identify explanatory factors for low HCAHPS scores, we decided to incorporate a technique known as a cognitive interview (CI). The CI is widely used as an evaluative tool for survey questions because of its ability to allow the interviewer to discern the processes that lead to responses.3 Up to this point, the focus of this CI method has been on the ability of the subject to comprehend and answer the questions.4 However, when a CI subject is answering a question, there is a large amount of information presented to the interviewer about the topic that is typically regarded as supplementary because of the focus on these specific issues.5 This study reports the supplemental information that may provide insight as to why patients answered as they did. Our goal was to gain further understanding of factors that may underlie HCAHPS responses.

Materials and Methods

Results

Discussion

Because of the limitations inherent in typical survey measures of investigation, we employed a CI technique. The objective was to generate hypotheses as to why patients may report reduced satisfaction with various aspects of hospitalization. The data set gleaned from this study was sizable, and held information pertinent to all parts of the hospital. The results reported in this article are focused on hospitalists.

The quantity of time hospitalists spend with patients may not be as important to the patients as the quality of interaction. Other studies have shown similar results, illustrating that the key factor in a patients' opinion of a visit with a physician was the perception of being taken seriously.10 The perception of being talked down to (as noted by Levy11) and the perception of not being responded to or focused on are the key negative factors for patients. By interacting with patients in a way that makes them feel valued, focused on, and responded to, physicians may improve patient satisfaction without requiring increased amounts of time spent with them.

Patient complaints that physicians do not respond to them may partially reflect inability to ask all of their questions. Hospitals may wish to consider methods to improve the ability of the patients to ask questions in the small amount of time available for physicians to talk with them (eg, placing notebooks by patients' bedsides and repeatedly encouraging them to write their questions down). This concept has also been endorsed as a means of improving the quality of care and reducing medical mistakes.12

Hospitalists and other physicians may positively impact the communication about medication issues by establishing a protocol for communicating major as well as minor side effects or even the lack of side effects of medication to patients. The challenge lies in determining the level of incidence that is significant enough (eg, 10%, 5%, or 1%) to warrant explaining minor side effects.

The findings of this study are limited by its small size, qualitative nature, nonrepresentative population, and loose approach to questioning. The patient responses were transcribed manually by the interviewer, leaving the possibility that some responses were not accurately recorded. Our survey instrument is not validated, and the results may not generalize to other settings. Instead of taking these findings as conclusive, we present them as suggestions for improvement that may be validated by further research.

Patient satisfaction is an important issue for hospitals, as it may affect the decision to seek care at one institution over another, but it may soon have direct implications for hospital reimbursement with the recent proposals for Value Based Purchasing (VBP) models by the Centers for Medicare and Medicaid Services (CMS). Up to 5% of inpatient Medicare reimbursement would be linked to performance measures, 40% of which could come from percentile outcomes on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) alone.1 HCAHPS scores are now available for individual hospitals at the Hospital Compare website maintained by CMS (http://www.hospitalcompare.hhs.gov). Hospitalists will likely be held accountable by administrators for poor performance on the survey.

While the information garnered from the HCAHPS provides an external perception of hospital quality, the questions are broad and do not identify specific reasons for reduced satisfaction. Many have suggested that the incorporation of surveys already administered at hospitals may be required for successful HCAHPS administration in order to overcome the limitations inherent in its design.2

In order to identify explanatory factors for low HCAHPS scores, we decided to incorporate a technique known as a cognitive interview (CI). The CI is widely used as an evaluative tool for survey questions because of its ability to allow the interviewer to discern the processes that lead to responses.3 Up to this point, the focus of this CI method has been on the ability of the subject to comprehend and answer the questions.4 However, when a CI subject is answering a question, there is a large amount of information presented to the interviewer about the topic that is typically regarded as supplementary because of the focus on these specific issues.5 This study reports the supplemental information that may provide insight as to why patients answered as they did. Our goal was to gain further understanding of factors that may underlie HCAHPS responses.

Materials and Methods

Results

Discussion

Because of the limitations inherent in typical survey measures of investigation, we employed a CI technique. The objective was to generate hypotheses as to why patients may report reduced satisfaction with various aspects of hospitalization. The data set gleaned from this study was sizable, and held information pertinent to all parts of the hospital. The results reported in this article are focused on hospitalists.

The quantity of time hospitalists spend with patients may not be as important to the patients as the quality of interaction. Other studies have shown similar results, illustrating that the key factor in a patients' opinion of a visit with a physician was the perception of being taken seriously.10 The perception of being talked down to (as noted by Levy11) and the perception of not being responded to or focused on are the key negative factors for patients. By interacting with patients in a way that makes them feel valued, focused on, and responded to, physicians may improve patient satisfaction without requiring increased amounts of time spent with them.

Patient complaints that physicians do not respond to them may partially reflect inability to ask all of their questions. Hospitals may wish to consider methods to improve the ability of the patients to ask questions in the small amount of time available for physicians to talk with them (eg, placing notebooks by patients' bedsides and repeatedly encouraging them to write their questions down). This concept has also been endorsed as a means of improving the quality of care and reducing medical mistakes.12

Hospitalists and other physicians may positively impact the communication about medication issues by establishing a protocol for communicating major as well as minor side effects or even the lack of side effects of medication to patients. The challenge lies in determining the level of incidence that is significant enough (eg, 10%, 5%, or 1%) to warrant explaining minor side effects.

The findings of this study are limited by its small size, qualitative nature, nonrepresentative population, and loose approach to questioning. The patient responses were transcribed manually by the interviewer, leaving the possibility that some responses were not accurately recorded. Our survey instrument is not validated, and the results may not generalize to other settings. Instead of taking these findings as conclusive, we present them as suggestions for improvement that may be validated by further research.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

The management of patients on long‐term antithrombotic therapy (vitamin K antagonists [VKA] or antiplatelet agents) who may require temporary disruption for an invasive procedure is challenging. Management is controversial due to methodologically limited prospective data and varied consensus opinions. Yet periprocedural anticoagulation management is a commonly encountered clinical problem. It is estimated that there are 2.5 million patients on long‐term VKA therapy in North America1 and 41% of the U.S. population over age 40 years is on antiplatelet therapy.2 Further, the need for temporary disruption of these therapies for an invasive procedure is frequent. As an example, in 1 European study, approximately 15% of patients on long‐term VKA required a major surgical procedure in 4 years of follow‐up.3 The role of the hospitalist physician in managing these patients is increasing as hospitalists care for an increasing number of surgical patients and provide periprocedural consultation both in and out of the hospital. Therefore, it is imperative for the hospitalist physician to be proficient in making thoughtful and individualized recommendations on the appropriate management of periprocedural anticoagulants, drawing from the available literature and evidence‐based practice guidelines. Importantly, the Society of Hospital Medicine has cited perioperative management as an important core competency.4

The hospitalist physician is likely to encounter numerous periprocedural scenarios, including the management of antiplatelet agents, identifying low bleeding risk procedures wherein interruption of anticoagulants is unnecessary, and recognizing patients with a low short‐term thromboembolic risk where anticoagulants can be disrupted without the need for heparin or low molecular weight heparin (LMWH) in the periprocedural period (defined as bridging therapy). Further, all other clinical scenarios require both a careful individualized assessment of the patient's risk of periprocedural bleeding and thromboembolism and a thoughtful discussion with all involved parties. This discussion may involve the person performing the procedure, the anesthesiologist, and the patient. The purpose of this work is to explore these relevant areas through a review of the literature with a particular focus on the recently published 2008 American College of Chest Physicians (ACCP) evidence‐based clinical practice guidelines.

We reviewed medical literature from 1990 through May 2008 with the following key words: bridging, anticoagulation, perioperative, antiplatelet, heparin, and low molecular weight heparin. Individual studies were then independently reviewed by the authors. Studies that were felt relevant to a hospitalist physician were retrieved and reviewed. If there was uncertainty regarding applicability to a hospitalist setting, a second author's opinion was rendered. Additionally, we reviewed 1 author's personal reference list of articles relating to periprocedural anticoagulation that has been compiled over the past 10 years. This list and the reference lists of retrieved articles were also reviewed. Data were summarized to answer 4 clinically relevant questions:

• 1

  What is the optimal management of antiplatelet therapy in the periprocedural period?

• 2

  Are there very low‐bleeding risk procedures that do not require interruption of oral anticoagulation?

• 3

  Are there low thromboembolic risk populations who do not require periprocedural bridging?

• 4

  How do you manage patients who must discontinue anticoagulants but are at an increased thrombotic risk?

Clinical Question #1: What Is the Optimal Management of Antiplatelet Therapy in the Periprocedural Period?

The optimal management of oral antiplatelet therapy in the periprocedural period is not well studied. Most reviews, expert recommendations, and consensus statements either do not comment on periprocedural antiplatelet management or recommend the routine discontinuation of therapy at least 7 days prior to surgery.3, 5, 6 However, as the 2008 ACCP guidelines highlight, the recommendation to routinely discontinue antiplatelet therapy 7 days prior to the procedure is an oversimplification.1 In the era of both bare metal cardiac stents and drug‐eluting stents, the optimal management of these patients requires that 2 primary questions be asked: (1) Is this a low‐bleeding risk procedure whereby antiplatelet therapy can be continued? (2) Does the patient have a coronary stent whereby the continuation of antiplatelet therapy or delay of the intervention is necessary?

In the context of ongoing aspirin therapy, certain procedures have a low risk of significant hemorrhagic complications. These low bleeding risk procedures include cataract surgery, cutaneous surgery, oral surgery, and endoscopic procedures, including those with mucosal biopsies.710 Patients undergoing these procedures may safely continue low dose aspirin therapy, especially if they have a high‐risk indication for their aspirin such as recent myocardial infarction, stroke, or the presence of a coronary stent.5, 710 Whether these procedures can be safely performed in the setting of a thienopyridine or combination antiplatelet therapy is uncertain.

In the past several decades, the management of obstructive coronary artery disease has undergone a major evolution. Placement of coronary stents has become commonplace, and there are now several million patients with drug‐eluting stents.11 The major complication of these devices is stent thrombosis, which results in death or myocardial infarction in up to 64% of patients.12 Fortunately, dual antiplatelet therapy (aspirin and a thienopyridine such as clopidogrel) markedly reduces this risk.13 Current guidelines recommend using combination antiplatelet therapy for at least 4 to 6 weeks and ideally up to 12 months after placement of a bare metal stent and at least 12 months after placement of either a sirolimus‐ or paclitaxel‐eluting stent.1, 14 During this period of dual antiplatelet therapy, the premature discontinuation of the thienopyridine may be catastrophic. To guide clinicians in managing these patients in the periprocedural period, recent consensus guidelines recommend the following:1, 12

• 1

  In patients who are expected to need an invasive surgical procedure in the next 12 months, consideration should be given to avoiding drug‐eluting stents.

• 2

  Elective procedures which have an increased risk of bleeding should be deferred for at least 6 weeks after bare metal stent implantation and 12 months after drug‐eluting stent implantation.

• 3

  For patients undergoing a surgical procedure within 6 weeks of bare metal stent implantation and 12 months of drug‐eluting stent implantation, continuation of aspirin and clopidogrel is recommended. If bleeding risk prohibits this, then a cardiologist should be consulted.

• 4

  In patients with a drug‐eluting stent who need to undergo a procedure whereby the thienopyridine needs to be discontinued, aspirin should be continued if at all possible, and the thienopyridine should be resumed as soon as possible after the procedure. It may be reasonable to consider a loading dose of clopidogrel, up to 600 mg, in this setting, although prospective supportive data is lacking.1

It is important to recognize that delayed stent thrombosis is now reported well beyond 1 year after drug‐eluting stent implantation, and that there may not be a diminution in risk after the initial 12 months.1517 Until additional data is available, it seems prudent, if possible, to at least continue aspirin in the periprocedural period in these patients. If bleeding concerns obviate this, then antiplatelet therapy should be discontinued and resumed as soon as possible.

For patients on chronic antiplatelet therapy who do not have a cardiac stent and who are not undergoing a low‐bleeding‐risk procedure, the risks and benefits of the continuation or discontinuation of antiplatelet therapy in the periprocedural period are uncertain as absolute risks in the periprocedural period have not been well studied. Relative risks/benefits, however, can be estimated from prior studies. Aspirin leads to an approximate 25% relative risk reduction in cardiac or thrombotic event rates compared to placebo.14, 18 Although important, the absolute benefit of 1 week of therapy (vs. no therapy during the periprocedural period) is estimated to be small. The small absolute benefit of continued aspirin therapy may be offset by an increase in significant bleeding events. Although, not well studied, continued aspirin increases significant bleeding by 50% with absolute event rates varying by type of procedure.8 In some procedures, such as intracranial surgery or transurethral prostatectomy, this bleeding risk is prohibitive. For others, the risk may be modest and the decision to continue vs. discontinue aspirin therapy may be at the discretion of the person performing the procedure. In general, for most patients who do not have a coronary stent and have not had a recent (past 3 months) myocardial infarction or stroke, discontinuation of antiplatelet therapy 7 to 10 days prior to the procedure seems prudent. The primary exceptions are patients who are undergoing percutaneous coronary intervention or coronary artery bypass grafting. For these procedures continuing aspirin is recommended.1 Figure 1 outlines a proposed management strategy based upon available evidence and guidelines.

Figure 1

Clinical Question #4: How Do You Manage Patients Who Must Discontinue Anticoagulants But Are at an Increased Thrombotic Risk?

When anticoagulation must be held and the patient does not have a very low thromboembolic risk, a decision of whether or not to use bridging anticoagulation must be made. The current ACCP guideline gives grade 1C and 2C recommendations (evidence from observational studies, case series, or controlled trials with serious flaws) regarding for whom and how to implement bridging.1 The grade C designation is due to a lack of high‐quality randomized clinical trials. As such, the clinician must carefully consider an individual patient's estimated thromboembolic risk, procedurally‐related bleeding risk, patient‐related bleeding risk factors, and the patient's values regarding concerns of thromboembolism or bleeding. In these situations it is also imperative that the person performing the procedure is involved in the risk‐to‐benefit discussion.

When evaluating an individual patient's risk of thromboembolism, clinicians sometimes estimate the perioperative risk by prorating the annual incidence of thromboembolic complications to the few days that anticoagulation is withheld.67 Making this extrapolation discounts the effect of a potential increase in thromboembolic risk induced by surgery. As an example, an average patient with atrial fibrillation who has a 5% predicted annual stroke rate would be estimated to have a stroke risk of 0.05% if they are not anticoagulated for 4 days. However, studies have shown that the actual rate of perioperative thromboembolism is approximately 1%.1 With these limitations and uncertainties in mind, and until there is better prospective outcomes data, we must consider relative risks in the context of absolute event rate estimates when deciding a perioperative anticoagulant management plan. The estimated annual incidence of thrombosis without anticoagulation for various indications and the current guideline recommendations are presented in Table 2.

Table 2.Summary of Guidelines on Bridging Therapy

		Practice Guideline Preferred Management Recommendations
Indication for chronic anticoagulation	Estimated Annual Thrombotic Risk Without Anticoagulation	ACCP*1	ACC/AHA45, 46	British Haematologic Society70
Abbreviations: ACC, American College of Cardiology; ACCP, American College of Chest Physicians; A‐fib, atrial fibrillation; AHA, American Heart Association; CHADS2, CHFHtnage 75 yearsDMstroke/TIA (see Table 1); CHF, congestive heart failure; CVA, cerebrovascular accident; DM, diabetes mellitus; Htn, hypertension; N/A, not applicable; TIA, transient ischemic attack; VTE, venous thromboembolism. * ACCP recommends withholding full‐dose anticoagulation for 48‐72 hours postprocedure in patients at high risk of postoperative bleeding. Extrapolated from the British Committee for Standards in Haemotology. Risk factors: A‐fib, prior stroke or TIA, Htn, DM, CHF, age >75 years.
Dual prosthetic or older‐generation valve	>10%	Bridge	Bridge	Bridge
VTE within 3 months or severe thrombophilias		Bridge	N/A	Bridge
Pregnancy with prosthetic valve	Bridge	Bridge	N/A
Bileaflet valve in the mitral position	Bridge	Bridge	Prophylaxis
Valve with acute embolism <6 months	Bridge	N/A	Bridge
A‐fib valvular or CHADS2 score 5‐6	Bridge	Consider bridging	N/A
Recurrent venous thromboembolism	4‐10%	Bridge	N/A	N/A
VTE within 3‐12 months or active cancer		Bridge	N/A	Prophylaxis
Bileaflet aortic valve with additional risk factors	Bridge	Bridge	Prophylaxis
A‐fib CHADS2 score 3‐4	Bridge	Consider bridging	N/A
Bileaflet aortic valve without additional risk factors	<4%	Prophylaxis or no bridging	No bridging	Prophylaxis
VTE >12 months		Prophylaxis or no bridging	N/A	Prophylaxis
A‐fib CHADS2 score 0‐2 and no previous CVA/TIA	Prophylaxis or no bridging	No bridging	N/A

In addition to thromboembolic risk, we must also consider the bleeding risk associated with the procedure/surgery. Importantly, therapeutic heparin started early in the postoperative period is associated with major bleeding event rates as high as 10% to 20%.1, 50 Once a major bleeding event occurs, this will often lead to an extended interruption of anticoagulant therapy, placing the patient at a more prolonged risk of an associated thromboembolic event. For this reason, the resumption of full‐dose anticoagulation with LMWH/heparin should be delayed for at least 48 hours in most patients undergoing a surgery or procedure associated with an increased risk of bleeding. Examples of these higher‐bleeding‐risk procedures include major thoracic surgery, intracranial or spinal surgery, major vascular surgery, major orthopedic surgery, urologic surgery involving the bladder or prostrate, major oncologic surgery, reconstructive plastic surgery, colonoscopy with associated polypectomy, renal or prostate biopsies, and placement of a cardiac pacemaker/defibrillator.1, 5157

Taken together, these uncertainties surrounding thromboembolic and bleeding risk estimates imply that there are multiple options for periprocedural management. Several studies, many of which included patients with mechanical heart valves, have shown similar safety and efficacy between LMWH and intravenous (IV) unfractionated heparin.5864 Table 3 summarizes these studies. The ACCP recommends bridging with LMWH over IV unfractionated heparin due to equal efficacy and cost savings with LMWH.1 When bridging is used, careful attention must be given to the timing and dose of anticoagulation in both the preoperative and postoperative periods. Table 4 lists dosing of commonly used LMWHs in North America. When using LMWHs in the preprocedural setting it is important to note that unacceptably high levels of anticoagulation remain present when a patient is given a full once‐daily LMWH dose the morning prior to the procedure or when a full‐dose, twice‐daily LMWH dose is given the evening prior to the procedure.65, 66 For this reason, the ACCP recommends administering the last preoperative dose 24 hours before surgery and if full‐dose once‐daily LMWH is used, the dose should be decreased by one‐half on the day before the surgery in order to ensure that no residual anticoagulant effect remains at the time of surgery.

Table 3.Summary of Key Bridging Studies

AuthorReference/Study Type	Number of Patients	Patient Population	Type of Procedure	Bridging Strategy	Major Bleeds	Minor Bleeds	TE Rate
NOTE: Studies included are prospective cohort studies with at least 150 patients and registries with greater than 500 patients in which consecutive patients were followed for postintervention outcome assessment. Abbreviations: AC, anticoagulation; a‐fib, atrial fibrillation; bid, twice daily; DVT, deep venous thrombosis; IU, anti‐Xa activity in International Units; LMWH, low molecular weight heparin; POD, postoperative day; TE, thromboembolism; UFH, unfractionated heparin; VTE, venous thromboembolism.
Turpie and Douketis63/single arm cohort	174	66% aortic valve; 34% mitral or dual prosthetic valve	Not specified	Enoxaparin 1 mg/kg twice daily	2.3%	Not specified	None
Kovacs et al.61/single arm cohort	224	Prosthetic heart valves or a‐fib plus 1 major risk factor	67 surgical; 157 nonsurgical	Preoperative bridging with dalteparin 200 IU/kg daily; dose reduced to 100 IU/kg on preoperative day 1; restarted at 100 IU/kg on POD 1; dose reduced to 5000 IU daily if high risk for bleeding	6.7%; 8/15 occurred intraoperatively or <6 hours postoperatively; 2/15 occurred after 4 weeks	Not specified	3.6%; 6/8 episodes occurred after warfarin held secondary to bleeding; 2/8 thrombotic episodes judged to be due to cardioembolism
Douketis et al.59/prospective registry	650	A‐fib 58%; mechanical heart valve 33%	251 surgical; 399 nonsurgical	Dalteparin 100 IU/kg twice daily; held after high bleeding risk procedure and patients with poor hemostasis	0.92%	5.9%	0.6%
Spyropolous et al.62/prospective registry; 14 centers in United States and Canada	901	UFH: 40% mechanical valves, 33% a‐fib; LMWH: 24% mechanical valve, 40% a‐fib	394 surgical; 507 nonsurgical	LMWH mostly given twice daily 80%; UFH 20%	5.5% UFH; 3.3% LMWH	9.1% UFH; 12.0% LMWH	2.4% UFH; 0.9% LMWH
Dunn et al.66/prospective cohort	260	A‐fib 68% or prior DVT 37% (excluding prosthetic heart valves)	105 surgical; 145 nonsurgical	Enoxaparin 1.5 mg/kg daily	3.5% overall; minor surgery/procedures 0.9%; major surgery 28%	42%	1.9%; 1/5 events occurred after bleeding led to withdrawal of AC
Omran et al.77/prospective registry	779	Various indications	Major and minor procedures	All patients bridged with enoxaparin; moderate TE risk 1 mg/kg daily; high TE risk 1 mg/kg twice daily	0.5%; all in high‐risk group	5.9%	0
Garcia et al.71/prospective, observational cohort of 101 sites in United States	1024 patients with 1293 interruptions of AC	A‐fib 53%; VTE 14%; prosthetic valve 13%	Outpatient procedures only	At discretion of provider. Bridging performed in 8.3% of interruptions; 3% a‐fib, 10% VTE, and 29% mechanical valves	0.6%; 4/6 patients with major bleed received bridging	1.7%;10/17 patients with minor bleed received bridging	0.7%; no events in patients who were bridged
Wysokinski et al.64/prospective cohort	345 consecutive patients undergoing 386 procedures	100% nonvalvular a‐fib	Major and minor surgeries/procedures	Individualized in AC clinic; 52% of patients bridged	2.7%; no difference whether patient received bridging or not	3.0%; 10/11 occurred in bridged patients	1.1%; no difference in bridged vs. nonbridged patients

Table 4.Low Molecular Weight Heparin Dosing Regimens Evaluated in Periprocedural Management Studies

Low Molecular Weight Heparin	Subcutaneous Dose
Abbreviation: IU, anti‐Xa activity in International Units.
Dalteparin
Low dose (prophylaxis dose)	5,000 IU once daily
Full dose	100 IU/kg twice daily or 200 IU/kg once daily
Enoxaparin
Low dose (prophylaxis dose)	30 mg twice daily or 40mg daily
Full dose	1 mg/kg twice daily or 1.5 mg/kg once daily
Tinzaparin (full dose)	175 IU/kg once daily

In the postprocedural setting, timing and dose of anticoagulant is important, as major bleeding with the use of therapeutic anticoagulation can occur in up to 10% to 20% of cases. When restarting anticoagulation after the procedure, it is important to evaluate intraoperative hemostasis and to consider patient‐related factors that may further increase bleeding risk. These include advanced age, concomitant antiplatelet or nonsteroidal antiinflammatory medications, renal insufficiency, placement of spinal/epidural catheter, worsening liver disease, or the presence of other comorbid illnesses such as cancer.30, 67, 68 The ACCP recommends withholding full‐dose anticoagulation for at least 48 to 72 hours in patients who are felt to be at a high risk for postoperative bleeding.1 Figure 2 is a proposed management approach to the use of bridging anticoagulants that is consistent with the 2008 ACCP recommendations.

Figure 2

CONCLUSION

The evaluation and management of patients on long‐term antiplatelet or VKA therapy who require an invasive procedure or surgery is a common, complicated, and controversial area. Importantly, it is an area in which the hospitalist physician must be adept. Although there remain many unanswered clinical questions, an evolving literature base and recent practice guidelines can help guide management decisions.

The management of patients on long‐term antithrombotic therapy (vitamin K antagonists [VKA] or antiplatelet agents) who may require temporary disruption for an invasive procedure is challenging. Management is controversial due to methodologically limited prospective data and varied consensus opinions. Yet periprocedural anticoagulation management is a commonly encountered clinical problem. It is estimated that there are 2.5 million patients on long‐term VKA therapy in North America1 and 41% of the U.S. population over age 40 years is on antiplatelet therapy.2 Further, the need for temporary disruption of these therapies for an invasive procedure is frequent. As an example, in 1 European study, approximately 15% of patients on long‐term VKA required a major surgical procedure in 4 years of follow‐up.3 The role of the hospitalist physician in managing these patients is increasing as hospitalists care for an increasing number of surgical patients and provide periprocedural consultation both in and out of the hospital. Therefore, it is imperative for the hospitalist physician to be proficient in making thoughtful and individualized recommendations on the appropriate management of periprocedural anticoagulants, drawing from the available literature and evidence‐based practice guidelines. Importantly, the Society of Hospital Medicine has cited perioperative management as an important core competency.4

The hospitalist physician is likely to encounter numerous periprocedural scenarios, including the management of antiplatelet agents, identifying low bleeding risk procedures wherein interruption of anticoagulants is unnecessary, and recognizing patients with a low short‐term thromboembolic risk where anticoagulants can be disrupted without the need for heparin or low molecular weight heparin (LMWH) in the periprocedural period (defined as bridging therapy). Further, all other clinical scenarios require both a careful individualized assessment of the patient's risk of periprocedural bleeding and thromboembolism and a thoughtful discussion with all involved parties. This discussion may involve the person performing the procedure, the anesthesiologist, and the patient. The purpose of this work is to explore these relevant areas through a review of the literature with a particular focus on the recently published 2008 American College of Chest Physicians (ACCP) evidence‐based clinical practice guidelines.

We reviewed medical literature from 1990 through May 2008 with the following key words: bridging, anticoagulation, perioperative, antiplatelet, heparin, and low molecular weight heparin. Individual studies were then independently reviewed by the authors. Studies that were felt relevant to a hospitalist physician were retrieved and reviewed. If there was uncertainty regarding applicability to a hospitalist setting, a second author's opinion was rendered. Additionally, we reviewed 1 author's personal reference list of articles relating to periprocedural anticoagulation that has been compiled over the past 10 years. This list and the reference lists of retrieved articles were also reviewed. Data were summarized to answer 4 clinically relevant questions:

• 1

  What is the optimal management of antiplatelet therapy in the periprocedural period?

• 2

  Are there very low‐bleeding risk procedures that do not require interruption of oral anticoagulation?

• 3

  Are there low thromboembolic risk populations who do not require periprocedural bridging?

• 4

  How do you manage patients who must discontinue anticoagulants but are at an increased thrombotic risk?

Clinical Question #1: What Is the Optimal Management of Antiplatelet Therapy in the Periprocedural Period?

The optimal management of oral antiplatelet therapy in the periprocedural period is not well studied. Most reviews, expert recommendations, and consensus statements either do not comment on periprocedural antiplatelet management or recommend the routine discontinuation of therapy at least 7 days prior to surgery.3, 5, 6 However, as the 2008 ACCP guidelines highlight, the recommendation to routinely discontinue antiplatelet therapy 7 days prior to the procedure is an oversimplification.1 In the era of both bare metal cardiac stents and drug‐eluting stents, the optimal management of these patients requires that 2 primary questions be asked: (1) Is this a low‐bleeding risk procedure whereby antiplatelet therapy can be continued? (2) Does the patient have a coronary stent whereby the continuation of antiplatelet therapy or delay of the intervention is necessary?

In the context of ongoing aspirin therapy, certain procedures have a low risk of significant hemorrhagic complications. These low bleeding risk procedures include cataract surgery, cutaneous surgery, oral surgery, and endoscopic procedures, including those with mucosal biopsies.710 Patients undergoing these procedures may safely continue low dose aspirin therapy, especially if they have a high‐risk indication for their aspirin such as recent myocardial infarction, stroke, or the presence of a coronary stent.5, 710 Whether these procedures can be safely performed in the setting of a thienopyridine or combination antiplatelet therapy is uncertain.

In the past several decades, the management of obstructive coronary artery disease has undergone a major evolution. Placement of coronary stents has become commonplace, and there are now several million patients with drug‐eluting stents.11 The major complication of these devices is stent thrombosis, which results in death or myocardial infarction in up to 64% of patients.12 Fortunately, dual antiplatelet therapy (aspirin and a thienopyridine such as clopidogrel) markedly reduces this risk.13 Current guidelines recommend using combination antiplatelet therapy for at least 4 to 6 weeks and ideally up to 12 months after placement of a bare metal stent and at least 12 months after placement of either a sirolimus‐ or paclitaxel‐eluting stent.1, 14 During this period of dual antiplatelet therapy, the premature discontinuation of the thienopyridine may be catastrophic. To guide clinicians in managing these patients in the periprocedural period, recent consensus guidelines recommend the following:1, 12

• 1

  In patients who are expected to need an invasive surgical procedure in the next 12 months, consideration should be given to avoiding drug‐eluting stents.

• 2

  Elective procedures which have an increased risk of bleeding should be deferred for at least 6 weeks after bare metal stent implantation and 12 months after drug‐eluting stent implantation.

• 3

  For patients undergoing a surgical procedure within 6 weeks of bare metal stent implantation and 12 months of drug‐eluting stent implantation, continuation of aspirin and clopidogrel is recommended. If bleeding risk prohibits this, then a cardiologist should be consulted.

• 4

  In patients with a drug‐eluting stent who need to undergo a procedure whereby the thienopyridine needs to be discontinued, aspirin should be continued if at all possible, and the thienopyridine should be resumed as soon as possible after the procedure. It may be reasonable to consider a loading dose of clopidogrel, up to 600 mg, in this setting, although prospective supportive data is lacking.1

It is important to recognize that delayed stent thrombosis is now reported well beyond 1 year after drug‐eluting stent implantation, and that there may not be a diminution in risk after the initial 12 months.1517 Until additional data is available, it seems prudent, if possible, to at least continue aspirin in the periprocedural period in these patients. If bleeding concerns obviate this, then antiplatelet therapy should be discontinued and resumed as soon as possible.

For patients on chronic antiplatelet therapy who do not have a cardiac stent and who are not undergoing a low‐bleeding‐risk procedure, the risks and benefits of the continuation or discontinuation of antiplatelet therapy in the periprocedural period are uncertain as absolute risks in the periprocedural period have not been well studied. Relative risks/benefits, however, can be estimated from prior studies. Aspirin leads to an approximate 25% relative risk reduction in cardiac or thrombotic event rates compared to placebo.14, 18 Although important, the absolute benefit of 1 week of therapy (vs. no therapy during the periprocedural period) is estimated to be small. The small absolute benefit of continued aspirin therapy may be offset by an increase in significant bleeding events. Although, not well studied, continued aspirin increases significant bleeding by 50% with absolute event rates varying by type of procedure.8 In some procedures, such as intracranial surgery or transurethral prostatectomy, this bleeding risk is prohibitive. For others, the risk may be modest and the decision to continue vs. discontinue aspirin therapy may be at the discretion of the person performing the procedure. In general, for most patients who do not have a coronary stent and have not had a recent (past 3 months) myocardial infarction or stroke, discontinuation of antiplatelet therapy 7 to 10 days prior to the procedure seems prudent. The primary exceptions are patients who are undergoing percutaneous coronary intervention or coronary artery bypass grafting. For these procedures continuing aspirin is recommended.1 Figure 1 outlines a proposed management strategy based upon available evidence and guidelines.

Figure 1

Clinical Question #4: How Do You Manage Patients Who Must Discontinue Anticoagulants But Are at an Increased Thrombotic Risk?

When anticoagulation must be held and the patient does not have a very low thromboembolic risk, a decision of whether or not to use bridging anticoagulation must be made. The current ACCP guideline gives grade 1C and 2C recommendations (evidence from observational studies, case series, or controlled trials with serious flaws) regarding for whom and how to implement bridging.1 The grade C designation is due to a lack of high‐quality randomized clinical trials. As such, the clinician must carefully consider an individual patient's estimated thromboembolic risk, procedurally‐related bleeding risk, patient‐related bleeding risk factors, and the patient's values regarding concerns of thromboembolism or bleeding. In these situations it is also imperative that the person performing the procedure is involved in the risk‐to‐benefit discussion.

When evaluating an individual patient's risk of thromboembolism, clinicians sometimes estimate the perioperative risk by prorating the annual incidence of thromboembolic complications to the few days that anticoagulation is withheld.67 Making this extrapolation discounts the effect of a potential increase in thromboembolic risk induced by surgery. As an example, an average patient with atrial fibrillation who has a 5% predicted annual stroke rate would be estimated to have a stroke risk of 0.05% if they are not anticoagulated for 4 days. However, studies have shown that the actual rate of perioperative thromboembolism is approximately 1%.1 With these limitations and uncertainties in mind, and until there is better prospective outcomes data, we must consider relative risks in the context of absolute event rate estimates when deciding a perioperative anticoagulant management plan. The estimated annual incidence of thrombosis without anticoagulation for various indications and the current guideline recommendations are presented in Table 2.

Table 2.Summary of Guidelines on Bridging Therapy

		Practice Guideline Preferred Management Recommendations
Indication for chronic anticoagulation	Estimated Annual Thrombotic Risk Without Anticoagulation	ACCP*1	ACC/AHA45, 46	British Haematologic Society70
Abbreviations: ACC, American College of Cardiology; ACCP, American College of Chest Physicians; A‐fib, atrial fibrillation; AHA, American Heart Association; CHADS2, CHFHtnage 75 yearsDMstroke/TIA (see Table 1); CHF, congestive heart failure; CVA, cerebrovascular accident; DM, diabetes mellitus; Htn, hypertension; N/A, not applicable; TIA, transient ischemic attack; VTE, venous thromboembolism. * ACCP recommends withholding full‐dose anticoagulation for 48‐72 hours postprocedure in patients at high risk of postoperative bleeding. Extrapolated from the British Committee for Standards in Haemotology. Risk factors: A‐fib, prior stroke or TIA, Htn, DM, CHF, age >75 years.
Dual prosthetic or older‐generation valve	>10%	Bridge	Bridge	Bridge
VTE within 3 months or severe thrombophilias		Bridge	N/A	Bridge
Pregnancy with prosthetic valve	Bridge	Bridge	N/A
Bileaflet valve in the mitral position	Bridge	Bridge	Prophylaxis
Valve with acute embolism <6 months	Bridge	N/A	Bridge
A‐fib valvular or CHADS2 score 5‐6	Bridge	Consider bridging	N/A
Recurrent venous thromboembolism	4‐10%	Bridge	N/A	N/A
VTE within 3‐12 months or active cancer		Bridge	N/A	Prophylaxis
Bileaflet aortic valve with additional risk factors	Bridge	Bridge	Prophylaxis
A‐fib CHADS2 score 3‐4	Bridge	Consider bridging	N/A
Bileaflet aortic valve without additional risk factors	<4%	Prophylaxis or no bridging	No bridging	Prophylaxis
VTE >12 months		Prophylaxis or no bridging	N/A	Prophylaxis
A‐fib CHADS2 score 0‐2 and no previous CVA/TIA	Prophylaxis or no bridging	No bridging	N/A

In addition to thromboembolic risk, we must also consider the bleeding risk associated with the procedure/surgery. Importantly, therapeutic heparin started early in the postoperative period is associated with major bleeding event rates as high as 10% to 20%.1, 50 Once a major bleeding event occurs, this will often lead to an extended interruption of anticoagulant therapy, placing the patient at a more prolonged risk of an associated thromboembolic event. For this reason, the resumption of full‐dose anticoagulation with LMWH/heparin should be delayed for at least 48 hours in most patients undergoing a surgery or procedure associated with an increased risk of bleeding. Examples of these higher‐bleeding‐risk procedures include major thoracic surgery, intracranial or spinal surgery, major vascular surgery, major orthopedic surgery, urologic surgery involving the bladder or prostrate, major oncologic surgery, reconstructive plastic surgery, colonoscopy with associated polypectomy, renal or prostate biopsies, and placement of a cardiac pacemaker/defibrillator.1, 5157

Taken together, these uncertainties surrounding thromboembolic and bleeding risk estimates imply that there are multiple options for periprocedural management. Several studies, many of which included patients with mechanical heart valves, have shown similar safety and efficacy between LMWH and intravenous (IV) unfractionated heparin.5864 Table 3 summarizes these studies. The ACCP recommends bridging with LMWH over IV unfractionated heparin due to equal efficacy and cost savings with LMWH.1 When bridging is used, careful attention must be given to the timing and dose of anticoagulation in both the preoperative and postoperative periods. Table 4 lists dosing of commonly used LMWHs in North America. When using LMWHs in the preprocedural setting it is important to note that unacceptably high levels of anticoagulation remain present when a patient is given a full once‐daily LMWH dose the morning prior to the procedure or when a full‐dose, twice‐daily LMWH dose is given the evening prior to the procedure.65, 66 For this reason, the ACCP recommends administering the last preoperative dose 24 hours before surgery and if full‐dose once‐daily LMWH is used, the dose should be decreased by one‐half on the day before the surgery in order to ensure that no residual anticoagulant effect remains at the time of surgery.

Table 3.Summary of Key Bridging Studies

AuthorReference/Study Type	Number of Patients	Patient Population	Type of Procedure	Bridging Strategy	Major Bleeds	Minor Bleeds	TE Rate
NOTE: Studies included are prospective cohort studies with at least 150 patients and registries with greater than 500 patients in which consecutive patients were followed for postintervention outcome assessment. Abbreviations: AC, anticoagulation; a‐fib, atrial fibrillation; bid, twice daily; DVT, deep venous thrombosis; IU, anti‐Xa activity in International Units; LMWH, low molecular weight heparin; POD, postoperative day; TE, thromboembolism; UFH, unfractionated heparin; VTE, venous thromboembolism.
Turpie and Douketis63/single arm cohort	174	66% aortic valve; 34% mitral or dual prosthetic valve	Not specified	Enoxaparin 1 mg/kg twice daily	2.3%	Not specified	None
Kovacs et al.61/single arm cohort	224	Prosthetic heart valves or a‐fib plus 1 major risk factor	67 surgical; 157 nonsurgical	Preoperative bridging with dalteparin 200 IU/kg daily; dose reduced to 100 IU/kg on preoperative day 1; restarted at 100 IU/kg on POD 1; dose reduced to 5000 IU daily if high risk for bleeding	6.7%; 8/15 occurred intraoperatively or <6 hours postoperatively; 2/15 occurred after 4 weeks	Not specified	3.6%; 6/8 episodes occurred after warfarin held secondary to bleeding; 2/8 thrombotic episodes judged to be due to cardioembolism
Douketis et al.59/prospective registry	650	A‐fib 58%; mechanical heart valve 33%	251 surgical; 399 nonsurgical	Dalteparin 100 IU/kg twice daily; held after high bleeding risk procedure and patients with poor hemostasis	0.92%	5.9%	0.6%
Spyropolous et al.62/prospective registry; 14 centers in United States and Canada	901	UFH: 40% mechanical valves, 33% a‐fib; LMWH: 24% mechanical valve, 40% a‐fib	394 surgical; 507 nonsurgical	LMWH mostly given twice daily 80%; UFH 20%	5.5% UFH; 3.3% LMWH	9.1% UFH; 12.0% LMWH	2.4% UFH; 0.9% LMWH
Dunn et al.66/prospective cohort	260	A‐fib 68% or prior DVT 37% (excluding prosthetic heart valves)	105 surgical; 145 nonsurgical	Enoxaparin 1.5 mg/kg daily	3.5% overall; minor surgery/procedures 0.9%; major surgery 28%	42%	1.9%; 1/5 events occurred after bleeding led to withdrawal of AC
Omran et al.77/prospective registry	779	Various indications	Major and minor procedures	All patients bridged with enoxaparin; moderate TE risk 1 mg/kg daily; high TE risk 1 mg/kg twice daily	0.5%; all in high‐risk group	5.9%	0
Garcia et al.71/prospective, observational cohort of 101 sites in United States	1024 patients with 1293 interruptions of AC	A‐fib 53%; VTE 14%; prosthetic valve 13%	Outpatient procedures only	At discretion of provider. Bridging performed in 8.3% of interruptions; 3% a‐fib, 10% VTE, and 29% mechanical valves	0.6%; 4/6 patients with major bleed received bridging	1.7%;10/17 patients with minor bleed received bridging	0.7%; no events in patients who were bridged
Wysokinski et al.64/prospective cohort	345 consecutive patients undergoing 386 procedures	100% nonvalvular a‐fib	Major and minor surgeries/procedures	Individualized in AC clinic; 52% of patients bridged	2.7%; no difference whether patient received bridging or not	3.0%; 10/11 occurred in bridged patients	1.1%; no difference in bridged vs. nonbridged patients

Table 4.Low Molecular Weight Heparin Dosing Regimens Evaluated in Periprocedural Management Studies

Low Molecular Weight Heparin	Subcutaneous Dose
Abbreviation: IU, anti‐Xa activity in International Units.
Dalteparin
Low dose (prophylaxis dose)	5,000 IU once daily
Full dose	100 IU/kg twice daily or 200 IU/kg once daily
Enoxaparin
Low dose (prophylaxis dose)	30 mg twice daily or 40mg daily
Full dose	1 mg/kg twice daily or 1.5 mg/kg once daily
Tinzaparin (full dose)	175 IU/kg once daily

In the postprocedural setting, timing and dose of anticoagulant is important, as major bleeding with the use of therapeutic anticoagulation can occur in up to 10% to 20% of cases. When restarting anticoagulation after the procedure, it is important to evaluate intraoperative hemostasis and to consider patient‐related factors that may further increase bleeding risk. These include advanced age, concomitant antiplatelet or nonsteroidal antiinflammatory medications, renal insufficiency, placement of spinal/epidural catheter, worsening liver disease, or the presence of other comorbid illnesses such as cancer.30, 67, 68 The ACCP recommends withholding full‐dose anticoagulation for at least 48 to 72 hours in patients who are felt to be at a high risk for postoperative bleeding.1 Figure 2 is a proposed management approach to the use of bridging anticoagulants that is consistent with the 2008 ACCP recommendations.

Figure 2

CONCLUSION

The evaluation and management of patients on long‐term antiplatelet or VKA therapy who require an invasive procedure or surgery is a common, complicated, and controversial area. Importantly, it is an area in which the hospitalist physician must be adept. Although there remain many unanswered clinical questions, an evolving literature base and recent practice guidelines can help guide management decisions.

The management of patients on long‐term antithrombotic therapy (vitamin K antagonists [VKA] or antiplatelet agents) who may require temporary disruption for an invasive procedure is challenging. Management is controversial due to methodologically limited prospective data and varied consensus opinions. Yet periprocedural anticoagulation management is a commonly encountered clinical problem. It is estimated that there are 2.5 million patients on long‐term VKA therapy in North America1 and 41% of the U.S. population over age 40 years is on antiplatelet therapy.2 Further, the need for temporary disruption of these therapies for an invasive procedure is frequent. As an example, in 1 European study, approximately 15% of patients on long‐term VKA required a major surgical procedure in 4 years of follow‐up.3 The role of the hospitalist physician in managing these patients is increasing as hospitalists care for an increasing number of surgical patients and provide periprocedural consultation both in and out of the hospital. Therefore, it is imperative for the hospitalist physician to be proficient in making thoughtful and individualized recommendations on the appropriate management of periprocedural anticoagulants, drawing from the available literature and evidence‐based practice guidelines. Importantly, the Society of Hospital Medicine has cited perioperative management as an important core competency.4

The hospitalist physician is likely to encounter numerous periprocedural scenarios, including the management of antiplatelet agents, identifying low bleeding risk procedures wherein interruption of anticoagulants is unnecessary, and recognizing patients with a low short‐term thromboembolic risk where anticoagulants can be disrupted without the need for heparin or low molecular weight heparin (LMWH) in the periprocedural period (defined as bridging therapy). Further, all other clinical scenarios require both a careful individualized assessment of the patient's risk of periprocedural bleeding and thromboembolism and a thoughtful discussion with all involved parties. This discussion may involve the person performing the procedure, the anesthesiologist, and the patient. The purpose of this work is to explore these relevant areas through a review of the literature with a particular focus on the recently published 2008 American College of Chest Physicians (ACCP) evidence‐based clinical practice guidelines.

We reviewed medical literature from 1990 through May 2008 with the following key words: bridging, anticoagulation, perioperative, antiplatelet, heparin, and low molecular weight heparin. Individual studies were then independently reviewed by the authors. Studies that were felt relevant to a hospitalist physician were retrieved and reviewed. If there was uncertainty regarding applicability to a hospitalist setting, a second author's opinion was rendered. Additionally, we reviewed 1 author's personal reference list of articles relating to periprocedural anticoagulation that has been compiled over the past 10 years. This list and the reference lists of retrieved articles were also reviewed. Data were summarized to answer 4 clinically relevant questions:

• 1

  What is the optimal management of antiplatelet therapy in the periprocedural period?

• 2

  Are there very low‐bleeding risk procedures that do not require interruption of oral anticoagulation?

• 3

  Are there low thromboembolic risk populations who do not require periprocedural bridging?

• 4

  How do you manage patients who must discontinue anticoagulants but are at an increased thrombotic risk?

Clinical Question #1: What Is the Optimal Management of Antiplatelet Therapy in the Periprocedural Period?

The optimal management of oral antiplatelet therapy in the periprocedural period is not well studied. Most reviews, expert recommendations, and consensus statements either do not comment on periprocedural antiplatelet management or recommend the routine discontinuation of therapy at least 7 days prior to surgery.3, 5, 6 However, as the 2008 ACCP guidelines highlight, the recommendation to routinely discontinue antiplatelet therapy 7 days prior to the procedure is an oversimplification.1 In the era of both bare metal cardiac stents and drug‐eluting stents, the optimal management of these patients requires that 2 primary questions be asked: (1) Is this a low‐bleeding risk procedure whereby antiplatelet therapy can be continued? (2) Does the patient have a coronary stent whereby the continuation of antiplatelet therapy or delay of the intervention is necessary?

In the context of ongoing aspirin therapy, certain procedures have a low risk of significant hemorrhagic complications. These low bleeding risk procedures include cataract surgery, cutaneous surgery, oral surgery, and endoscopic procedures, including those with mucosal biopsies.710 Patients undergoing these procedures may safely continue low dose aspirin therapy, especially if they have a high‐risk indication for their aspirin such as recent myocardial infarction, stroke, or the presence of a coronary stent.5, 710 Whether these procedures can be safely performed in the setting of a thienopyridine or combination antiplatelet therapy is uncertain.

In the past several decades, the management of obstructive coronary artery disease has undergone a major evolution. Placement of coronary stents has become commonplace, and there are now several million patients with drug‐eluting stents.11 The major complication of these devices is stent thrombosis, which results in death or myocardial infarction in up to 64% of patients.12 Fortunately, dual antiplatelet therapy (aspirin and a thienopyridine such as clopidogrel) markedly reduces this risk.13 Current guidelines recommend using combination antiplatelet therapy for at least 4 to 6 weeks and ideally up to 12 months after placement of a bare metal stent and at least 12 months after placement of either a sirolimus‐ or paclitaxel‐eluting stent.1, 14 During this period of dual antiplatelet therapy, the premature discontinuation of the thienopyridine may be catastrophic. To guide clinicians in managing these patients in the periprocedural period, recent consensus guidelines recommend the following:1, 12

• 1

  In patients who are expected to need an invasive surgical procedure in the next 12 months, consideration should be given to avoiding drug‐eluting stents.

• 2

  Elective procedures which have an increased risk of bleeding should be deferred for at least 6 weeks after bare metal stent implantation and 12 months after drug‐eluting stent implantation.

• 3

  For patients undergoing a surgical procedure within 6 weeks of bare metal stent implantation and 12 months of drug‐eluting stent implantation, continuation of aspirin and clopidogrel is recommended. If bleeding risk prohibits this, then a cardiologist should be consulted.

• 4

  In patients with a drug‐eluting stent who need to undergo a procedure whereby the thienopyridine needs to be discontinued, aspirin should be continued if at all possible, and the thienopyridine should be resumed as soon as possible after the procedure. It may be reasonable to consider a loading dose of clopidogrel, up to 600 mg, in this setting, although prospective supportive data is lacking.1

It is important to recognize that delayed stent thrombosis is now reported well beyond 1 year after drug‐eluting stent implantation, and that there may not be a diminution in risk after the initial 12 months.1517 Until additional data is available, it seems prudent, if possible, to at least continue aspirin in the periprocedural period in these patients. If bleeding concerns obviate this, then antiplatelet therapy should be discontinued and resumed as soon as possible.

For patients on chronic antiplatelet therapy who do not have a cardiac stent and who are not undergoing a low‐bleeding‐risk procedure, the risks and benefits of the continuation or discontinuation of antiplatelet therapy in the periprocedural period are uncertain as absolute risks in the periprocedural period have not been well studied. Relative risks/benefits, however, can be estimated from prior studies. Aspirin leads to an approximate 25% relative risk reduction in cardiac or thrombotic event rates compared to placebo.14, 18 Although important, the absolute benefit of 1 week of therapy (vs. no therapy during the periprocedural period) is estimated to be small. The small absolute benefit of continued aspirin therapy may be offset by an increase in significant bleeding events. Although, not well studied, continued aspirin increases significant bleeding by 50% with absolute event rates varying by type of procedure.8 In some procedures, such as intracranial surgery or transurethral prostatectomy, this bleeding risk is prohibitive. For others, the risk may be modest and the decision to continue vs. discontinue aspirin therapy may be at the discretion of the person performing the procedure. In general, for most patients who do not have a coronary stent and have not had a recent (past 3 months) myocardial infarction or stroke, discontinuation of antiplatelet therapy 7 to 10 days prior to the procedure seems prudent. The primary exceptions are patients who are undergoing percutaneous coronary intervention or coronary artery bypass grafting. For these procedures continuing aspirin is recommended.1 Figure 1 outlines a proposed management strategy based upon available evidence and guidelines.

Figure 1

Clinical Question #4: How Do You Manage Patients Who Must Discontinue Anticoagulants But Are at an Increased Thrombotic Risk?

When anticoagulation must be held and the patient does not have a very low thromboembolic risk, a decision of whether or not to use bridging anticoagulation must be made. The current ACCP guideline gives grade 1C and 2C recommendations (evidence from observational studies, case series, or controlled trials with serious flaws) regarding for whom and how to implement bridging.1 The grade C designation is due to a lack of high‐quality randomized clinical trials. As such, the clinician must carefully consider an individual patient's estimated thromboembolic risk, procedurally‐related bleeding risk, patient‐related bleeding risk factors, and the patient's values regarding concerns of thromboembolism or bleeding. In these situations it is also imperative that the person performing the procedure is involved in the risk‐to‐benefit discussion.

When evaluating an individual patient's risk of thromboembolism, clinicians sometimes estimate the perioperative risk by prorating the annual incidence of thromboembolic complications to the few days that anticoagulation is withheld.67 Making this extrapolation discounts the effect of a potential increase in thromboembolic risk induced by surgery. As an example, an average patient with atrial fibrillation who has a 5% predicted annual stroke rate would be estimated to have a stroke risk of 0.05% if they are not anticoagulated for 4 days. However, studies have shown that the actual rate of perioperative thromboembolism is approximately 1%.1 With these limitations and uncertainties in mind, and until there is better prospective outcomes data, we must consider relative risks in the context of absolute event rate estimates when deciding a perioperative anticoagulant management plan. The estimated annual incidence of thrombosis without anticoagulation for various indications and the current guideline recommendations are presented in Table 2.

Table 2.Summary of Guidelines on Bridging Therapy

		Practice Guideline Preferred Management Recommendations
Indication for chronic anticoagulation	Estimated Annual Thrombotic Risk Without Anticoagulation	ACCP*1	ACC/AHA45, 46	British Haematologic Society70
Abbreviations: ACC, American College of Cardiology; ACCP, American College of Chest Physicians; A‐fib, atrial fibrillation; AHA, American Heart Association; CHADS2, CHFHtnage 75 yearsDMstroke/TIA (see Table 1); CHF, congestive heart failure; CVA, cerebrovascular accident; DM, diabetes mellitus; Htn, hypertension; N/A, not applicable; TIA, transient ischemic attack; VTE, venous thromboembolism. * ACCP recommends withholding full‐dose anticoagulation for 48‐72 hours postprocedure in patients at high risk of postoperative bleeding. Extrapolated from the British Committee for Standards in Haemotology. Risk factors: A‐fib, prior stroke or TIA, Htn, DM, CHF, age >75 years.
Dual prosthetic or older‐generation valve	>10%	Bridge	Bridge	Bridge
VTE within 3 months or severe thrombophilias		Bridge	N/A	Bridge
Pregnancy with prosthetic valve	Bridge	Bridge	N/A
Bileaflet valve in the mitral position	Bridge	Bridge	Prophylaxis
Valve with acute embolism <6 months	Bridge	N/A	Bridge
A‐fib valvular or CHADS2 score 5‐6	Bridge	Consider bridging	N/A
Recurrent venous thromboembolism	4‐10%	Bridge	N/A	N/A
VTE within 3‐12 months or active cancer		Bridge	N/A	Prophylaxis
Bileaflet aortic valve with additional risk factors	Bridge	Bridge	Prophylaxis
A‐fib CHADS2 score 3‐4	Bridge	Consider bridging	N/A
Bileaflet aortic valve without additional risk factors	<4%	Prophylaxis or no bridging	No bridging	Prophylaxis
VTE >12 months		Prophylaxis or no bridging	N/A	Prophylaxis
A‐fib CHADS2 score 0‐2 and no previous CVA/TIA	Prophylaxis or no bridging	No bridging	N/A

In addition to thromboembolic risk, we must also consider the bleeding risk associated with the procedure/surgery. Importantly, therapeutic heparin started early in the postoperative period is associated with major bleeding event rates as high as 10% to 20%.1, 50 Once a major bleeding event occurs, this will often lead to an extended interruption of anticoagulant therapy, placing the patient at a more prolonged risk of an associated thromboembolic event. For this reason, the resumption of full‐dose anticoagulation with LMWH/heparin should be delayed for at least 48 hours in most patients undergoing a surgery or procedure associated with an increased risk of bleeding. Examples of these higher‐bleeding‐risk procedures include major thoracic surgery, intracranial or spinal surgery, major vascular surgery, major orthopedic surgery, urologic surgery involving the bladder or prostrate, major oncologic surgery, reconstructive plastic surgery, colonoscopy with associated polypectomy, renal or prostate biopsies, and placement of a cardiac pacemaker/defibrillator.1, 5157

Taken together, these uncertainties surrounding thromboembolic and bleeding risk estimates imply that there are multiple options for periprocedural management. Several studies, many of which included patients with mechanical heart valves, have shown similar safety and efficacy between LMWH and intravenous (IV) unfractionated heparin.5864 Table 3 summarizes these studies. The ACCP recommends bridging with LMWH over IV unfractionated heparin due to equal efficacy and cost savings with LMWH.1 When bridging is used, careful attention must be given to the timing and dose of anticoagulation in both the preoperative and postoperative periods. Table 4 lists dosing of commonly used LMWHs in North America. When using LMWHs in the preprocedural setting it is important to note that unacceptably high levels of anticoagulation remain present when a patient is given a full once‐daily LMWH dose the morning prior to the procedure or when a full‐dose, twice‐daily LMWH dose is given the evening prior to the procedure.65, 66 For this reason, the ACCP recommends administering the last preoperative dose 24 hours before surgery and if full‐dose once‐daily LMWH is used, the dose should be decreased by one‐half on the day before the surgery in order to ensure that no residual anticoagulant effect remains at the time of surgery.

Table 3.Summary of Key Bridging Studies

AuthorReference/Study Type	Number of Patients	Patient Population	Type of Procedure	Bridging Strategy	Major Bleeds	Minor Bleeds	TE Rate
NOTE: Studies included are prospective cohort studies with at least 150 patients and registries with greater than 500 patients in which consecutive patients were followed for postintervention outcome assessment. Abbreviations: AC, anticoagulation; a‐fib, atrial fibrillation; bid, twice daily; DVT, deep venous thrombosis; IU, anti‐Xa activity in International Units; LMWH, low molecular weight heparin; POD, postoperative day; TE, thromboembolism; UFH, unfractionated heparin; VTE, venous thromboembolism.
Turpie and Douketis63/single arm cohort	174	66% aortic valve; 34% mitral or dual prosthetic valve	Not specified	Enoxaparin 1 mg/kg twice daily	2.3%	Not specified	None
Kovacs et al.61/single arm cohort	224	Prosthetic heart valves or a‐fib plus 1 major risk factor	67 surgical; 157 nonsurgical	Preoperative bridging with dalteparin 200 IU/kg daily; dose reduced to 100 IU/kg on preoperative day 1; restarted at 100 IU/kg on POD 1; dose reduced to 5000 IU daily if high risk for bleeding	6.7%; 8/15 occurred intraoperatively or <6 hours postoperatively; 2/15 occurred after 4 weeks	Not specified	3.6%; 6/8 episodes occurred after warfarin held secondary to bleeding; 2/8 thrombotic episodes judged to be due to cardioembolism
Douketis et al.59/prospective registry	650	A‐fib 58%; mechanical heart valve 33%	251 surgical; 399 nonsurgical	Dalteparin 100 IU/kg twice daily; held after high bleeding risk procedure and patients with poor hemostasis	0.92%	5.9%	0.6%
Spyropolous et al.62/prospective registry; 14 centers in United States and Canada	901	UFH: 40% mechanical valves, 33% a‐fib; LMWH: 24% mechanical valve, 40% a‐fib	394 surgical; 507 nonsurgical	LMWH mostly given twice daily 80%; UFH 20%	5.5% UFH; 3.3% LMWH	9.1% UFH; 12.0% LMWH	2.4% UFH; 0.9% LMWH
Dunn et al.66/prospective cohort	260	A‐fib 68% or prior DVT 37% (excluding prosthetic heart valves)	105 surgical; 145 nonsurgical	Enoxaparin 1.5 mg/kg daily	3.5% overall; minor surgery/procedures 0.9%; major surgery 28%	42%	1.9%; 1/5 events occurred after bleeding led to withdrawal of AC
Omran et al.77/prospective registry	779	Various indications	Major and minor procedures	All patients bridged with enoxaparin; moderate TE risk 1 mg/kg daily; high TE risk 1 mg/kg twice daily	0.5%; all in high‐risk group	5.9%	0
Garcia et al.71/prospective, observational cohort of 101 sites in United States	1024 patients with 1293 interruptions of AC	A‐fib 53%; VTE 14%; prosthetic valve 13%	Outpatient procedures only	At discretion of provider. Bridging performed in 8.3% of interruptions; 3% a‐fib, 10% VTE, and 29% mechanical valves	0.6%; 4/6 patients with major bleed received bridging	1.7%;10/17 patients with minor bleed received bridging	0.7%; no events in patients who were bridged
Wysokinski et al.64/prospective cohort	345 consecutive patients undergoing 386 procedures	100% nonvalvular a‐fib	Major and minor surgeries/procedures	Individualized in AC clinic; 52% of patients bridged	2.7%; no difference whether patient received bridging or not	3.0%; 10/11 occurred in bridged patients	1.1%; no difference in bridged vs. nonbridged patients

Table 4.Low Molecular Weight Heparin Dosing Regimens Evaluated in Periprocedural Management Studies

Low Molecular Weight Heparin	Subcutaneous Dose
Abbreviation: IU, anti‐Xa activity in International Units.
Dalteparin
Low dose (prophylaxis dose)	5,000 IU once daily
Full dose	100 IU/kg twice daily or 200 IU/kg once daily
Enoxaparin
Low dose (prophylaxis dose)	30 mg twice daily or 40mg daily
Full dose	1 mg/kg twice daily or 1.5 mg/kg once daily
Tinzaparin (full dose)	175 IU/kg once daily

In the postprocedural setting, timing and dose of anticoagulant is important, as major bleeding with the use of therapeutic anticoagulation can occur in up to 10% to 20% of cases. When restarting anticoagulation after the procedure, it is important to evaluate intraoperative hemostasis and to consider patient‐related factors that may further increase bleeding risk. These include advanced age, concomitant antiplatelet or nonsteroidal antiinflammatory medications, renal insufficiency, placement of spinal/epidural catheter, worsening liver disease, or the presence of other comorbid illnesses such as cancer.30, 67, 68 The ACCP recommends withholding full‐dose anticoagulation for at least 48 to 72 hours in patients who are felt to be at a high risk for postoperative bleeding.1 Figure 2 is a proposed management approach to the use of bridging anticoagulants that is consistent with the 2008 ACCP recommendations.

Figure 2

CONCLUSION

The evaluation and management of patients on long‐term antiplatelet or VKA therapy who require an invasive procedure or surgery is a common, complicated, and controversial area. Importantly, it is an area in which the hospitalist physician must be adept. Although there remain many unanswered clinical questions, an evolving literature base and recent practice guidelines can help guide management decisions.

In the late 1990s, hospitalist systems grew rapidly in an environment where cost containment was paramount, complexity of patients increased, and outpatient practices experienced increasing productivity and efficiency pressures.15 While the healthcare delivery environment has changed significantly since that time,68 hospitalists have continued to become more common. In fact, the field's present size of more than 25,000 has already exceeded early projections, and there are no signs of slackening demand.911

Growth has been attributed to primary care physicians' increasing focus on outpatient care, hospitals' response to financial pressures, and the need to facilitate improved communication among various hospital care providers.1216 Hospital leadership has played a similarly important role in fueling the growth of hospitalists, particularly since the vast majority of programs require and receive institutional (usually hospital) support.17 However, the factors that continue to influence leaders' decisions to utilize hospitalists and the current and future needs that hospitalists are fulfilling are unknown. Each of these factors is likely to impact growth of the field, as well as the clinical and organizational identity of hospitalists. In addition, an understanding of the market demand for hospitalists' competencies and the roles they play in the hospital may inform any changes in board certification and training for hospitalists.11, 1821

To gain a more complete understanding of a key part of the engine driving the growth of hospitalists, we performed a cross‐sectional survey of California hospital leaders who were involved with the funding or administration of their hospitalist groups. Our survey aimed to understand: (1) the prevalence of hospitalist groups in California hospitals, (2) hospital leaders' rationale for initiating the use of hospitalists, (3) the scope of clinical and nonclinical practice of hospitalists, and 4) hospital leaders' perspective on the need for further training and/or certification.

Materials and Methods

Results