Lost in Transcription
http://www.100md.com
《新英格兰医药杂志》
In this Journal feature, information about a real patient is presented in stages (boldface type) to an expert clinician, who responds to the information, sharing his or her reasoning with the reader (regular type). The authors' commentary follows.
Urinary urgency and fever developed in a 55-year-old, bedridden woman with multiple sclerosis, a long-term indwelling Foley catheter, and multiple prior urinary tract infections. The patient had recently been transferred from an assisted-living facility to a skilled-nursing facility because of progressive disability. On the day of presentation, she reported urinary urgency and dysuria, despite a normally functioning urethral catheter; her temperature was 38.5°C. The nursing staff obtained a complete blood count and sent specimens of her blood and urine for culture. Oral levofloxacin was administered empirically for the treatment of a presumed urinary tract infection.
Patients with long-term indwelling urinary catheters frequently have pyuria and bacteriuria, which may or may not represent infection in the urinary tract. Such patients are often treated with antibiotics, especially if fever or changes in mental status occur. There is some justification for empirical therapy in this setting, since urinary tract infections are common in patients with indwelling catheters and may present atypically. This patient has urinary urgency and dysuria, symptoms that are suggestive of cystitis, but fever may indicate pyelonephritis, appendicitis, or other infectious or inflammatory conditions, including Crohn's disease, which occurs with increased frequency among patients with multiple sclerosis. At this stage, I would inquire about past genitourinary symptoms related to the patient's multiple sclerosis.
Multiple sclerosis had been diagnosed 16 years earlier and followed a relapsing–remitting course. The patient became bedridden and required placement in an assisted-living facility 6 years before presentation. Placement of an intrathecal baclofen pump resulted in symptomatic improvement in spasticity. Her history included gastroesophageal reflux disease, depression, and a hysterectomy that had been performed 20 years earlier. She reported allergies to penicillin and sulfa-containing agents. Her long-term medications included ranitidine for reflux symptoms, rofecoxib for pain control, baclofen (administered with a pump) for spasticity, gabapentin for neuropathic pain, and weekly methotrexate as corticosteroid-sparing therapy for her progressive multiple sclerosis. She had had mouth pain and fatigue for several days before presentation, in addition to her genitourinary problems, but she reported no other symptoms.
The most important new information is the medication list. None of the patient's prescribed medications are likely to cause fever and dysuria. Methotrexate places the patient at increased risk for infection, as does the presence of an intrathecal pump. However, another possibility is an error involving one of her medications. Her recent relocation to a new facility increases the risk of error because of the staff's unfamiliarity with the patient, her medical conditions, and her medications. For instance, the staff at the new facility may not have had experience with medications administered by the intrathecal route or with this particular intrathecal pump. Inadvertent underdosing might have resulted in baclofen withdrawal, which could account for the patient's fever and dysuria.
Laboratory evaluation at the nursing home revealed a white-cell count of 500 per cubic millimeter, a platelet count of 10,000 per cubic millimeter, and a hematocrit of 28%. Blood counts obtained 10 weeks previously were normal. The patient was transferred to the emergency department of a tertiary care hospital for further evaluation.
The patient has pancytopenia and meets the criteria for febrile neutropenia. Potential causes of acute bone marrow failure include acute leukemia, megaloblastic anemia, and toxic effects of heavy metals or radiation. Another possibility is a medication reaction. For example, even low-dose methotrexate can cause myelosuppression, especially in the presence of predisposing factors such as reduced renal function, hypoalbuminemia, folate deficiency, and polymorphisms in the methylenetetrahydrofolate reductase gene, and interactions between methotrexate and certain other drugs can also result in myelosuppression. An interaction with trimethoprim–sulfamethoxazole, which is often used in urinary tract infections (although it should not have been used in this case because of the documented sulfa allergy), would be particularly hazardous. In addition, daily, instead of weekly, administration of methotrexate is a common error.
On examination, the patient was thin, somnolent, diaphoretic, and ill-appearing, lying on her side with obvious leg contractures. She was arousable but had difficulty speaking because of mouth pain — a change from previously documented examinations. Her temperature was 35.7°C; blood pressure, 90/60 mm Hg (prior documented systolic pressures were 100 to 120 mm Hg); respiratory rate, 20 breaths per minute; and heart rate, 122 beats per minute. She had notable bitemporal wasting and dry oral mucosa, with desquamation of her tongue, palate, and buccal surfaces. There was no airway compromise. Coarse rhonchi were present in the upper airway on pulmonary examination. The cardiac examination was notable only for tachycardia. The abdomen was normal except for a well-healed hysterectomy scar; the spleen was not palpable. Muscle strength was symmetrically reduced (4/5) in the arms. The leg muscles were atrophic, with symmetric spasticity, contractures, and reduced strength (2/5). There were no ecchymoses, petechiae, or decubitus ulcers; no lymphadenopathy was found.
Cachexia and pancytopenia could reflect cancer, but we would need to invoke an additional diagnosis to explain the severe mucositis. Pemphigus may occur as a paraneoplastic syndrome characterized by severe oral lesions, but this disorder is uncommon and is even less likely in the absence of conjunctival involvement or other skin findings. Other causes of mucositis, such as malabsorption, Crohn's disease, and vitamin deficiencies, seem unlikely in this case. My working diagnosis is methotrexate-induced toxicity, probably due to an error in drug administration. Regardless of the underlying diagnosis, my concern is that the patient has hypovolemia and may be in early septic shock.
I would also be concerned about the possibility of adrenal insufficiency, particularly adrenal suppression after corticosteroid therapy, and would want to know when she was last treated with corticosteroids. I would obtain a random cortisol measurement, and I would have a very low threshold for adding stress doses of corticosteroids until adrenal insufficiency was ruled out. Additional tests should include measurements of electrolytes, urea nitrogen, creatinine, and serum methotrexate; a complete blood count; coagulation studies; and blood and urine cultures. A chest radiograph should also be obtained.
Repeated laboratory testing revealed a white-cell count of 330 per cubic millimeter, with an absolute neutrophil count of 50 per cubic millimeter; the platelet count was 7000 per cubic millimeter, and the hematocrit was 34%. The sodium level was 143 mmol per liter; potassium, 3.3 mmol per liter; chloride, 107 mmol per liter; bicarbonate, 24 mmol per liter; urea nitrogen, 45 mg per deciliter (16 mmol per liter); serum creatinine, 1.0 mg per deciliter (88.4 μmol per liter), increased from 0.5 mg per deciliter (44.2 μmol per liter) 10 weeks earlier; and glucose, 162 mg per deciliter (9 mmol per liter). The partial-thromboplastin time, international normalized ratio, and fibrinogen and lactate dehydrogenase levels were normal. Serum cortisol was not measured. The urine was trace-positive for leukocyte esterase and negative for nitrites, with 0 to 5 white cells, more than 30 red cells, and occasional bacteria per high-power field. Gram's staining revealed occasional white cells in the urine but no organisms.
Methotrexate-induced bone marrow toxicity remains the most likely diagnosis, with secondary sepsis presumably due to infection from the urine or skin. I would communicate directly with the personnel at the nursing home to determine whether an increased dose of methotrexate had been given to the patient, intentionally or inadvertently, during the previous weeks. I would administer broad-spectrum antibiotics, provide aggressive fluid resuscitation, and admit the patient to the intensive care unit (ICU). Although there is no evidence of bleeding, a platelet count of 7000 per cubic millimeter warrants prophylactic platelet transfusion.
The patient was admitted to the ICU and treated intravenously with normal saline and piperacillin–tazobactam for presumed urosepsis. Her chest radiograph showed no abnormalities. Several sets of blood cultures showed no growth. A culture of the urine revealed fewer than 1000 colony-forming units of Escherichia coli, with sensitivity to numerous antibiotics. The serum methotrexate level was 40 nmol per liter.
Orally administered methotrexate is rapidly absorbed and then either excreted in urine or taken up by the tissues. Thus, the serum level of methotrexate does not correlate well with its effect on the bone marrow. Nonetheless, this patient's serum methotrexate level exceeds the estimated threshold of 10 mmol per liter of plasma required to inhibit DNA synthesis in bone marrow. I believe the clinical signs and symptoms are all consistent with methotrexate-induced toxicity, probably related to a medication error. Leucovorin may limit any ongoing toxic effects by restoring folate levels in affected tissues, but its use is routinely recommended in oncologic treatment protocols only until the serum methotrexate level is below 50 nmol per liter. There is no harm in giving leucovorin, but I suspect that it would be of little benefit at this point. I would also consider administering granulocyte colony-stimulating factor (G-CSF). Although it might have little effect if the bone marrow is acellular, the patient is at high risk for death from overwhelming infection while she remains neutropenic. G-CSF may shorten the duration of her neutropenia and improve her chances of survival. I would obtain an emergency hematologic consultation for advice regarding these management options.
Methotrexate is effectively excreted by functioning kidneys, with an elimination half-life of about 6 hours, but serum levels may remain elevated for longer periods if there is renal dysfunction or redistribution of the drug from tissue and third spaces such as ascites and pleural fluid. With this patient's acute decline in renal function, it is difficult to predict when the methotrexate level will fall below the toxic threshold and allow DNA synthesis to resume. Standard hemodialysis has generally not been effective in reducing serum methotrexate levels. Greater success has been reported with intermittent high-flux dialysis in small case series. However, dialysis in these cases was discontinued once the plasma methotrexate level fell below 300 nmol per liter, and this patient's level is much lower. While awaiting advice from the hematologist, I would alkalinize the urine to a pH of 7 to 8 with intravenously administered sodium bicarbonate, in an effort to reduce methotrexate precipitation within the renal tubules and limit tubular damage.
An emergency hematologic consultation was obtained, and treatment was initiated with leucovorin and G-CSF. The patient's records were carefully reviewed. Six years earlier, a specialist in rehabilitation medicine had prescribed methotrexate at a dose of 7.5 mg per week and after 2 years had increased the dose to 15 mg per week. During the patient's first visit with her outpatient primary care physician 3 years before admission, her methotrexate dose was erroneously recorded in the electronic record as 10 mg per day; however, she continued to receive the 15-mg weekly dose from her pharmacy, which filled the prescription on the basis of a standing order from the rehabilitation medicine service. Seventeen days before admission, at the time of the patient's transfer to the skilled-nursing facility, the primary care physician transcribed the patient's medications directly from the electronic record; the incorrect dose of 10 mg per day was transcribed and subsequently administered to the patient.
The clinicians providing her care will have to inform the patient and her family of the cause of her current illness. Physicians have an ethical duty to disclose errors to patients — certainly those that involve harm, such as the error in this case. If I were involved in this case, I would notify the primary care physician of the error and ask whether he or she would like to be involved in disclosing the error to the patient, but I would not allow this step to delay a frank discussion with the patient and her family.
The primary team, hematologist, and outpatient primary care physician met with six family members at the patient's bedside that evening to disclose the error and the grave prognosis. The family was concerned about the error but directed the providers to deal immediately with the clinical problems that the error had caused.
At this point, a bone marrow biopsy might provide useful prognostic information. If the cellularity of the bone marrow is intact, with only megaloblastic changes, treatment with leucovorin would be expected to result in clinical improvement within roughly 1 week. If the marrow is "empty," a return to normal peripheral-blood counts might take several weeks.
A bone marrow biopsy on the third hospital day revealed arrested erythropoiesis, with markedly decreased numbers of myeloid precursors and megakaryocytes (Figure 1). Repeated blood analysis revealed a white-cell count of 180 per cubic millimeter, with an absolute neutrophil count of 10 per cubic millimeter. Platelets were transfused, and the platelet count increased from 7000 to 63,000 per cubic millimeter. The hematocrit fell to 25% after fluid resuscitation but subsequently remained stable.
Figure 1. Bone Marrow Biopsy on the Third Hospital Day, Showing Maturing Erythroid Cells and Markedly Decreased Numbers of Myeloid Precursors and Megakaryocytes (Hematoxylin and Eosin).
The patient's white-cell count began to rise on the 7th hospital day; the absolute neutrophil count exceeded 500 per cubic millimeter, and treatment with G-CSF was stopped on the 9th day. Her voiding symptoms gradually resolved, and her mucositis improved. She was tolerating oral fluids by the 22nd day and was discharged to her nursing facility. Methotrexate therapy was not resumed. In the 8 months since discharge, the patient has had less frequent urinary infections and no progression of her multiple sclerosis, even though she is not receiving immunosuppressive therapy.
Commentary
Drug-related adverse events have consistently been identified as a primary cause of medical injuries.1,2 In a comprehensive study of such events at two teaching hospitals in Boston, medication errors that caused harm or had the clear potential to do so occurred in roughly 1 of every 15 hospitalized patients.3 Over 40% of drug-related adverse events that were classified as life-threatening or serious were judged to be preventable. Almost two thirds of the errors resulting in preventable events were made during ordering or transcription, as occurred in the case under discussion.
Classes of medications frequently implicated in drug-related adverse events include cardiovascular drugs, anticoagulants, analgesics, sedatives, antibiotics, and chemotherapeutic agents, such as methotrexate.4,5 As the discussant noted, the erroneous administration of methotrexate on a daily basis rather than on a weekly basis has been reported previously,5,6 and it has recently been confirmed as the most common error involving methotrexate that has been reported to the Food and Drug Administration.7 Errors in the weekly dosing schedule can result from confusion on the part of the patient or from an error in transcription. The latter occurred in this case; the primary physician carried forward a previous error in the electronic medical record, which itself may have been due to incomplete, incorrect, or illegible outside records. Although the dispensing pharmacist may intercept such errors, either by recognizing that the dose is incorrect or through information-system alerts for nonstandard doses, this opportunity was missed in the case under discussion.
Broad implementation of computerized medical records is expected to reduce errors.8 However, computerization may also be associated with errors. In some cases, aspects of computerization (e.g., poorly designed order screens or incorrect default doses) may actually introduce errors. A more likely scenario, however, is that computerization will facilitate the transfer of preexisting errors in the traditional medical-record system.9 In this case, the initial error in the documentation of the methotrexate dose may have been perpetuated in the electronic record because electronic systems make information (correct or incorrect) easily transferable, without the need to review old charts.
After initially considering a broad differential diagnosis, the discussant in this case quickly focused on the possibility of medical error, given the number and nature of the patient's medications and her recent transfer to a new health care facility. The patient's urinary urgency and hematuria probably resulted from methotrexate-induced leukopenia, which confers a predisposition to infection, and thrombocytopenia, respectively.
Although consideration of intriguing but uncommon diagnoses, such as paraneoplastic pemphigus in this case, is a core part of clinical training and provides the substrate for many clinical case conferences, the likelier circumstance of medical error is frequently not considered, despite the fact that more than 5% of hospitalized patients have documented iatrogenic injuries, many resulting from errors.1,2 The well-documented prevalence of medical errors, which has prompted the leaders of health care systems to focus on patient safety and problems of quality control,1 suggests that clinicians should routinely consider error as a possible explanation for any finding not readily explained. One never knows when a key clinical detail may have been lost in transcription.
Dr. Shojania is the recipient of a Canada Research Chair in Patient Safety and Quality Improvement. Dr. Amory is the recipient of a grant (1K23 HD45386-10A1) from the National Institute of Child Health and Human Development and research funding from Schering and GlaxoSmithKline. Dr. Saint is the recipient of a Career Development Award from the Health Services Research and Development Program of the Department of Veterans Affairs and a Patient Safety Developmental Center grant (P20-HS11540) from the Agency for Healthcare Research and Quality. No other potential conflict of interest relevant to this article was reported.
We are indebted to Dr. Virginia Broudy for reviewing the manuscript.
Source Information
From the Department of Internal Medicine, University of Washington, Seattle (R.M.K., J.K.A.); the Clinical Epidemiology Program, Ottawa Health Research Institute, and the Department of Medicine, University of Ottawa — both in Ottawa (K.G.S.); and the Ann Arbor Veterans Affairs Health Services Research and Development Center of Excellence and University of Michigan Medical School — both in Ann Arbor (S.S.).
Address reprint requests to Dr. Kalus at the Division of General Internal Medicine, University of Washington, Box 359702, Seattle, WA 98104, or at rkalus@u.washington.edu.
References
Kohn LT, Corrigan JM, Donaldson MS, eds. To err is human: building a safer health system. Washington, DC: National Academy Press, 2000.
Shojania KG, Wald H, Gross R. Understanding medical error and improving patient safety in the inpatient setting. Med Clin North Am 2002;86:847-867.
Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events: implications for prevention. JAMA 1995;274:29-34.
Lesar TS, Lomaestro BM, Pohl H. Medication-prescribing errors in a teaching hospital: a 9-year experience. Arch Intern Med 1997;157:1569-1576.
Grove ML. Methotrexate and bone marrow suppression: drug errors may be implicated in death. BMJ 2003;326:1145-1146.
Lomaestro BM, Lesar TS, Hager TP. Errors in prescribing methotrexate. JAMA 1992;268:2031-2032.
Moore TJ, Walsh CS, Cohen MR. Reported medication errors associated with methotrexate. Am J Health Syst Pharm 2004;61:1380-1384.
Kaushal R, Shojania KG, Bates DW. Effects of computerized physician order entry and clinical decision support systems on medication safety: a systematic review. Arch Intern Med 2003;163:1409-1416.
Koppel R, Metlay JP, Cohen A, et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005;293:1197-1203.(Robert M. Kalus, M.D., Ka)
Urinary urgency and fever developed in a 55-year-old, bedridden woman with multiple sclerosis, a long-term indwelling Foley catheter, and multiple prior urinary tract infections. The patient had recently been transferred from an assisted-living facility to a skilled-nursing facility because of progressive disability. On the day of presentation, she reported urinary urgency and dysuria, despite a normally functioning urethral catheter; her temperature was 38.5°C. The nursing staff obtained a complete blood count and sent specimens of her blood and urine for culture. Oral levofloxacin was administered empirically for the treatment of a presumed urinary tract infection.
Patients with long-term indwelling urinary catheters frequently have pyuria and bacteriuria, which may or may not represent infection in the urinary tract. Such patients are often treated with antibiotics, especially if fever or changes in mental status occur. There is some justification for empirical therapy in this setting, since urinary tract infections are common in patients with indwelling catheters and may present atypically. This patient has urinary urgency and dysuria, symptoms that are suggestive of cystitis, but fever may indicate pyelonephritis, appendicitis, or other infectious or inflammatory conditions, including Crohn's disease, which occurs with increased frequency among patients with multiple sclerosis. At this stage, I would inquire about past genitourinary symptoms related to the patient's multiple sclerosis.
Multiple sclerosis had been diagnosed 16 years earlier and followed a relapsing–remitting course. The patient became bedridden and required placement in an assisted-living facility 6 years before presentation. Placement of an intrathecal baclofen pump resulted in symptomatic improvement in spasticity. Her history included gastroesophageal reflux disease, depression, and a hysterectomy that had been performed 20 years earlier. She reported allergies to penicillin and sulfa-containing agents. Her long-term medications included ranitidine for reflux symptoms, rofecoxib for pain control, baclofen (administered with a pump) for spasticity, gabapentin for neuropathic pain, and weekly methotrexate as corticosteroid-sparing therapy for her progressive multiple sclerosis. She had had mouth pain and fatigue for several days before presentation, in addition to her genitourinary problems, but she reported no other symptoms.
The most important new information is the medication list. None of the patient's prescribed medications are likely to cause fever and dysuria. Methotrexate places the patient at increased risk for infection, as does the presence of an intrathecal pump. However, another possibility is an error involving one of her medications. Her recent relocation to a new facility increases the risk of error because of the staff's unfamiliarity with the patient, her medical conditions, and her medications. For instance, the staff at the new facility may not have had experience with medications administered by the intrathecal route or with this particular intrathecal pump. Inadvertent underdosing might have resulted in baclofen withdrawal, which could account for the patient's fever and dysuria.
Laboratory evaluation at the nursing home revealed a white-cell count of 500 per cubic millimeter, a platelet count of 10,000 per cubic millimeter, and a hematocrit of 28%. Blood counts obtained 10 weeks previously were normal. The patient was transferred to the emergency department of a tertiary care hospital for further evaluation.
The patient has pancytopenia and meets the criteria for febrile neutropenia. Potential causes of acute bone marrow failure include acute leukemia, megaloblastic anemia, and toxic effects of heavy metals or radiation. Another possibility is a medication reaction. For example, even low-dose methotrexate can cause myelosuppression, especially in the presence of predisposing factors such as reduced renal function, hypoalbuminemia, folate deficiency, and polymorphisms in the methylenetetrahydrofolate reductase gene, and interactions between methotrexate and certain other drugs can also result in myelosuppression. An interaction with trimethoprim–sulfamethoxazole, which is often used in urinary tract infections (although it should not have been used in this case because of the documented sulfa allergy), would be particularly hazardous. In addition, daily, instead of weekly, administration of methotrexate is a common error.
On examination, the patient was thin, somnolent, diaphoretic, and ill-appearing, lying on her side with obvious leg contractures. She was arousable but had difficulty speaking because of mouth pain — a change from previously documented examinations. Her temperature was 35.7°C; blood pressure, 90/60 mm Hg (prior documented systolic pressures were 100 to 120 mm Hg); respiratory rate, 20 breaths per minute; and heart rate, 122 beats per minute. She had notable bitemporal wasting and dry oral mucosa, with desquamation of her tongue, palate, and buccal surfaces. There was no airway compromise. Coarse rhonchi were present in the upper airway on pulmonary examination. The cardiac examination was notable only for tachycardia. The abdomen was normal except for a well-healed hysterectomy scar; the spleen was not palpable. Muscle strength was symmetrically reduced (4/5) in the arms. The leg muscles were atrophic, with symmetric spasticity, contractures, and reduced strength (2/5). There were no ecchymoses, petechiae, or decubitus ulcers; no lymphadenopathy was found.
Cachexia and pancytopenia could reflect cancer, but we would need to invoke an additional diagnosis to explain the severe mucositis. Pemphigus may occur as a paraneoplastic syndrome characterized by severe oral lesions, but this disorder is uncommon and is even less likely in the absence of conjunctival involvement or other skin findings. Other causes of mucositis, such as malabsorption, Crohn's disease, and vitamin deficiencies, seem unlikely in this case. My working diagnosis is methotrexate-induced toxicity, probably due to an error in drug administration. Regardless of the underlying diagnosis, my concern is that the patient has hypovolemia and may be in early septic shock.
I would also be concerned about the possibility of adrenal insufficiency, particularly adrenal suppression after corticosteroid therapy, and would want to know when she was last treated with corticosteroids. I would obtain a random cortisol measurement, and I would have a very low threshold for adding stress doses of corticosteroids until adrenal insufficiency was ruled out. Additional tests should include measurements of electrolytes, urea nitrogen, creatinine, and serum methotrexate; a complete blood count; coagulation studies; and blood and urine cultures. A chest radiograph should also be obtained.
Repeated laboratory testing revealed a white-cell count of 330 per cubic millimeter, with an absolute neutrophil count of 50 per cubic millimeter; the platelet count was 7000 per cubic millimeter, and the hematocrit was 34%. The sodium level was 143 mmol per liter; potassium, 3.3 mmol per liter; chloride, 107 mmol per liter; bicarbonate, 24 mmol per liter; urea nitrogen, 45 mg per deciliter (16 mmol per liter); serum creatinine, 1.0 mg per deciliter (88.4 μmol per liter), increased from 0.5 mg per deciliter (44.2 μmol per liter) 10 weeks earlier; and glucose, 162 mg per deciliter (9 mmol per liter). The partial-thromboplastin time, international normalized ratio, and fibrinogen and lactate dehydrogenase levels were normal. Serum cortisol was not measured. The urine was trace-positive for leukocyte esterase and negative for nitrites, with 0 to 5 white cells, more than 30 red cells, and occasional bacteria per high-power field. Gram's staining revealed occasional white cells in the urine but no organisms.
Methotrexate-induced bone marrow toxicity remains the most likely diagnosis, with secondary sepsis presumably due to infection from the urine or skin. I would communicate directly with the personnel at the nursing home to determine whether an increased dose of methotrexate had been given to the patient, intentionally or inadvertently, during the previous weeks. I would administer broad-spectrum antibiotics, provide aggressive fluid resuscitation, and admit the patient to the intensive care unit (ICU). Although there is no evidence of bleeding, a platelet count of 7000 per cubic millimeter warrants prophylactic platelet transfusion.
The patient was admitted to the ICU and treated intravenously with normal saline and piperacillin–tazobactam for presumed urosepsis. Her chest radiograph showed no abnormalities. Several sets of blood cultures showed no growth. A culture of the urine revealed fewer than 1000 colony-forming units of Escherichia coli, with sensitivity to numerous antibiotics. The serum methotrexate level was 40 nmol per liter.
Orally administered methotrexate is rapidly absorbed and then either excreted in urine or taken up by the tissues. Thus, the serum level of methotrexate does not correlate well with its effect on the bone marrow. Nonetheless, this patient's serum methotrexate level exceeds the estimated threshold of 10 mmol per liter of plasma required to inhibit DNA synthesis in bone marrow. I believe the clinical signs and symptoms are all consistent with methotrexate-induced toxicity, probably related to a medication error. Leucovorin may limit any ongoing toxic effects by restoring folate levels in affected tissues, but its use is routinely recommended in oncologic treatment protocols only until the serum methotrexate level is below 50 nmol per liter. There is no harm in giving leucovorin, but I suspect that it would be of little benefit at this point. I would also consider administering granulocyte colony-stimulating factor (G-CSF). Although it might have little effect if the bone marrow is acellular, the patient is at high risk for death from overwhelming infection while she remains neutropenic. G-CSF may shorten the duration of her neutropenia and improve her chances of survival. I would obtain an emergency hematologic consultation for advice regarding these management options.
Methotrexate is effectively excreted by functioning kidneys, with an elimination half-life of about 6 hours, but serum levels may remain elevated for longer periods if there is renal dysfunction or redistribution of the drug from tissue and third spaces such as ascites and pleural fluid. With this patient's acute decline in renal function, it is difficult to predict when the methotrexate level will fall below the toxic threshold and allow DNA synthesis to resume. Standard hemodialysis has generally not been effective in reducing serum methotrexate levels. Greater success has been reported with intermittent high-flux dialysis in small case series. However, dialysis in these cases was discontinued once the plasma methotrexate level fell below 300 nmol per liter, and this patient's level is much lower. While awaiting advice from the hematologist, I would alkalinize the urine to a pH of 7 to 8 with intravenously administered sodium bicarbonate, in an effort to reduce methotrexate precipitation within the renal tubules and limit tubular damage.
An emergency hematologic consultation was obtained, and treatment was initiated with leucovorin and G-CSF. The patient's records were carefully reviewed. Six years earlier, a specialist in rehabilitation medicine had prescribed methotrexate at a dose of 7.5 mg per week and after 2 years had increased the dose to 15 mg per week. During the patient's first visit with her outpatient primary care physician 3 years before admission, her methotrexate dose was erroneously recorded in the electronic record as 10 mg per day; however, she continued to receive the 15-mg weekly dose from her pharmacy, which filled the prescription on the basis of a standing order from the rehabilitation medicine service. Seventeen days before admission, at the time of the patient's transfer to the skilled-nursing facility, the primary care physician transcribed the patient's medications directly from the electronic record; the incorrect dose of 10 mg per day was transcribed and subsequently administered to the patient.
The clinicians providing her care will have to inform the patient and her family of the cause of her current illness. Physicians have an ethical duty to disclose errors to patients — certainly those that involve harm, such as the error in this case. If I were involved in this case, I would notify the primary care physician of the error and ask whether he or she would like to be involved in disclosing the error to the patient, but I would not allow this step to delay a frank discussion with the patient and her family.
The primary team, hematologist, and outpatient primary care physician met with six family members at the patient's bedside that evening to disclose the error and the grave prognosis. The family was concerned about the error but directed the providers to deal immediately with the clinical problems that the error had caused.
At this point, a bone marrow biopsy might provide useful prognostic information. If the cellularity of the bone marrow is intact, with only megaloblastic changes, treatment with leucovorin would be expected to result in clinical improvement within roughly 1 week. If the marrow is "empty," a return to normal peripheral-blood counts might take several weeks.
A bone marrow biopsy on the third hospital day revealed arrested erythropoiesis, with markedly decreased numbers of myeloid precursors and megakaryocytes (Figure 1). Repeated blood analysis revealed a white-cell count of 180 per cubic millimeter, with an absolute neutrophil count of 10 per cubic millimeter. Platelets were transfused, and the platelet count increased from 7000 to 63,000 per cubic millimeter. The hematocrit fell to 25% after fluid resuscitation but subsequently remained stable.
Figure 1. Bone Marrow Biopsy on the Third Hospital Day, Showing Maturing Erythroid Cells and Markedly Decreased Numbers of Myeloid Precursors and Megakaryocytes (Hematoxylin and Eosin).
The patient's white-cell count began to rise on the 7th hospital day; the absolute neutrophil count exceeded 500 per cubic millimeter, and treatment with G-CSF was stopped on the 9th day. Her voiding symptoms gradually resolved, and her mucositis improved. She was tolerating oral fluids by the 22nd day and was discharged to her nursing facility. Methotrexate therapy was not resumed. In the 8 months since discharge, the patient has had less frequent urinary infections and no progression of her multiple sclerosis, even though she is not receiving immunosuppressive therapy.
Commentary
Drug-related adverse events have consistently been identified as a primary cause of medical injuries.1,2 In a comprehensive study of such events at two teaching hospitals in Boston, medication errors that caused harm or had the clear potential to do so occurred in roughly 1 of every 15 hospitalized patients.3 Over 40% of drug-related adverse events that were classified as life-threatening or serious were judged to be preventable. Almost two thirds of the errors resulting in preventable events were made during ordering or transcription, as occurred in the case under discussion.
Classes of medications frequently implicated in drug-related adverse events include cardiovascular drugs, anticoagulants, analgesics, sedatives, antibiotics, and chemotherapeutic agents, such as methotrexate.4,5 As the discussant noted, the erroneous administration of methotrexate on a daily basis rather than on a weekly basis has been reported previously,5,6 and it has recently been confirmed as the most common error involving methotrexate that has been reported to the Food and Drug Administration.7 Errors in the weekly dosing schedule can result from confusion on the part of the patient or from an error in transcription. The latter occurred in this case; the primary physician carried forward a previous error in the electronic medical record, which itself may have been due to incomplete, incorrect, or illegible outside records. Although the dispensing pharmacist may intercept such errors, either by recognizing that the dose is incorrect or through information-system alerts for nonstandard doses, this opportunity was missed in the case under discussion.
Broad implementation of computerized medical records is expected to reduce errors.8 However, computerization may also be associated with errors. In some cases, aspects of computerization (e.g., poorly designed order screens or incorrect default doses) may actually introduce errors. A more likely scenario, however, is that computerization will facilitate the transfer of preexisting errors in the traditional medical-record system.9 In this case, the initial error in the documentation of the methotrexate dose may have been perpetuated in the electronic record because electronic systems make information (correct or incorrect) easily transferable, without the need to review old charts.
After initially considering a broad differential diagnosis, the discussant in this case quickly focused on the possibility of medical error, given the number and nature of the patient's medications and her recent transfer to a new health care facility. The patient's urinary urgency and hematuria probably resulted from methotrexate-induced leukopenia, which confers a predisposition to infection, and thrombocytopenia, respectively.
Although consideration of intriguing but uncommon diagnoses, such as paraneoplastic pemphigus in this case, is a core part of clinical training and provides the substrate for many clinical case conferences, the likelier circumstance of medical error is frequently not considered, despite the fact that more than 5% of hospitalized patients have documented iatrogenic injuries, many resulting from errors.1,2 The well-documented prevalence of medical errors, which has prompted the leaders of health care systems to focus on patient safety and problems of quality control,1 suggests that clinicians should routinely consider error as a possible explanation for any finding not readily explained. One never knows when a key clinical detail may have been lost in transcription.
Dr. Shojania is the recipient of a Canada Research Chair in Patient Safety and Quality Improvement. Dr. Amory is the recipient of a grant (1K23 HD45386-10A1) from the National Institute of Child Health and Human Development and research funding from Schering and GlaxoSmithKline. Dr. Saint is the recipient of a Career Development Award from the Health Services Research and Development Program of the Department of Veterans Affairs and a Patient Safety Developmental Center grant (P20-HS11540) from the Agency for Healthcare Research and Quality. No other potential conflict of interest relevant to this article was reported.
We are indebted to Dr. Virginia Broudy for reviewing the manuscript.
Source Information
From the Department of Internal Medicine, University of Washington, Seattle (R.M.K., J.K.A.); the Clinical Epidemiology Program, Ottawa Health Research Institute, and the Department of Medicine, University of Ottawa — both in Ottawa (K.G.S.); and the Ann Arbor Veterans Affairs Health Services Research and Development Center of Excellence and University of Michigan Medical School — both in Ann Arbor (S.S.).
Address reprint requests to Dr. Kalus at the Division of General Internal Medicine, University of Washington, Box 359702, Seattle, WA 98104, or at rkalus@u.washington.edu.
References
Kohn LT, Corrigan JM, Donaldson MS, eds. To err is human: building a safer health system. Washington, DC: National Academy Press, 2000.
Shojania KG, Wald H, Gross R. Understanding medical error and improving patient safety in the inpatient setting. Med Clin North Am 2002;86:847-867.
Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events: implications for prevention. JAMA 1995;274:29-34.
Lesar TS, Lomaestro BM, Pohl H. Medication-prescribing errors in a teaching hospital: a 9-year experience. Arch Intern Med 1997;157:1569-1576.
Grove ML. Methotrexate and bone marrow suppression: drug errors may be implicated in death. BMJ 2003;326:1145-1146.
Lomaestro BM, Lesar TS, Hager TP. Errors in prescribing methotrexate. JAMA 1992;268:2031-2032.
Moore TJ, Walsh CS, Cohen MR. Reported medication errors associated with methotrexate. Am J Health Syst Pharm 2004;61:1380-1384.
Kaushal R, Shojania KG, Bates DW. Effects of computerized physician order entry and clinical decision support systems on medication safety: a systematic review. Arch Intern Med 2003;163:1409-1416.
Koppel R, Metlay JP, Cohen A, et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005;293:1197-1203.(Robert M. Kalus, M.D., Ka)