INTRODUCTION

Patient experience and satisfaction is intrinsically valued, as strong physician‐patient communication, empathy, and patient comfort require little justification. However, studies have also shown that patient satisfaction is associated with better health outcomes and greater compliance.[1, 2, 3] A systematic review of studies linking patient satisfaction to outcomes found that patient experience is positively associated with patient safety, clinical effectiveness, health outcomes, adherence, and lower resource utilization.[4] Of 378 associations studied between patient experience and health outcomes, there were 312 positive associations.[4] However, not all studies have shown a positive association between patient satisfaction and outcomes.

Nevertheless, hospitals now have to strive to improve patient satisfaction, as Centers for Medicare & Medicaid Services (CMS) has introduced Hospital Value‐Based Purchasing. CMS started to withhold Medicare Severity Diagnosis‐Related Groups payments, starting at 1.0% in 2013, 1.25% in 2014, and increasing to 2.0% in 2017. This money is redistributed based on performance on core quality measures, including patient satisfaction measured through the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.[5]

Various studies have evaluated interventions to improve patient satisfaction, but to our knowledge, no study published in a peer‐reviewed research journal has shown a significant improvement in HCAHPS scores.[6, 7, 8, 9, 10, 11, 12] Levinson et al. argue that physician communication skills should be taught during residency, and that individualized feedback is an effective way to allow physicians to track their progress over time and compared to their peers.[13] We thus aimed to evaluate an intervention to improve patient satisfaction designed by the Patient Affairs Department for Ronald Reagan University of California, Los Angeles (UCLA) Medical Center (RRUCLAMC) and the UCLA Department of Medicine.

METHODOLOGY

RESULTS

DISCUSSION

Our intervention, which included real‐time feedback to physicians on results of the patient survey, monthly recognition of physicians who stood out on this survey, and an educational conference, was associated with a clear improvement in patient satisfaction with physician‐patient communication and overall recommendation of the hospital. These results are significant because they demonstrate a cost‐effective intervention that can be applied to academic hospitals across the country with the use of nonmedically trained volunteers, such as the undergraduate volunteers involved in our program. The limited costs associated with the intervention were the time in managing the volunteers and movie package award ($20). To our knowledge, it is the first study published in a peer‐reviewed research journal that has demonstrated an intervention associated with significant improvements in HCAHPS scores, the standard by which CMS reimbursement will be affected.

The improvements associated with this intervention could be very valuable to hospitals and patient care. The positive correlation of higher patient satisfaction with improved outcomes suggests this intervention may have additional benefits.[4] Last, these improvements in patient satisfaction in the HCAHPS scores could minimize losses to hospital revenue, as hospitals with low patient‐satisfaction scores will be penalized.

There was a statistically significant improvement in adjusted scores for the question Did your physicians explain things understandably? with patients responding always to all 3 physician‐related HCAHPS questions and Would you recommend this hospital to friends and family. The results for the 2 other physician‐related questions (Did your doctor explain things understandably? and Did your doctor listen carefully?) did show a trend toward significance, with p values of <0.1, and a larger study may have been better powered to detect a statistically significant difference. The improvement in response to the adjusted scores for the question Did your physicians explain things understandably? was the primary driver in the improvement in the adjusted percentage of patients who responded always to all 3 physician‐related HCAHPS questions. This was likely because the IM cohort had the lowest score on this question, and so the feedback to the residents may have helped to address this area of weakness. The UCLA IM HCAHPS scores prior to 2012 have always been lower than other programs at UCLA. As a result, we do not believe the change was due to a regression to the mean.

We believe that the intervention had a positive effect on patient satisfaction for several reasons. The regular e‐mails with the results of the survey may have served as a reminder to residents that patient satisfaction was being monitored and linked to them. The immediate and individualized feedback also may have facilitated adjustments of clinical practice in real time. The residents were able to compare their own scores and comments to the anonymous results of their peers. The monthly department‐wide recognition for residents who excelled in patient communication may have created an incentive and competition among the residents. It is possible that there may be an element of the Hawthorne effect that explained the improvement in HCAHPS scores. However, all of the residents in the departments studied were already being measured through the ARC survey. The primary change was more frequent reporting of ARC survey results, and so we believe that perception of measurement alone was less likely driving the results. The findings from this study are similar to those from provider‐specific report cards, which have shown that outcomes can be improved by forcing greater accountability and competition among physicians.[19]

Brown et al. demonstrated that 2, 4‐hour physician communication workshops in their study had no impact on patient satisfaction, and so we believe that our 1‐hour workshop with only 50% attendance had minimal impact on the improved patient satisfaction scores in our study.[20] Our intervention also coincided with the implementation of the Accreditation Council for Graduate Medical Education (ACGME) work‐hour restrictions implemented in July 2011. These restrictions limited residents to 80 hours per week, intern duty periods were restricted to 16 hours and residents to 28 hours, and interns and residents required 8 to 10 hours free of duty between scheduled duty periods.[21] One of the biggest impacts of ACGME work‐hour restrictions was that interns were doing more day and night shifts rather than 28‐hour calls. However, these work‐hour restrictions were the same for all specialties and so were unlikely to explain the improved patient satisfaction associated with our intervention.

Our study has limitations. The study was a nonrandomized pre‐post study. We attempted to control for the differences in the cohorts with a multivariable regression analysis, but there may be unmeasured differences that we were unable to control for. Due to deidentification of the data, we could only control for patient health based on patient perceived health. In addition, the percentage of patients requiring ICU care in the IM cohort was higher in 2012 than in 2011. We did not identify differences in outcomes from analyses stratified by ICU or non‐ICU patients. In addition, patients who were excluded because of missing outcomes were more likely to be older and admitted through the ER. Further investigation would be needed to see if the findings of this study could be extended to other clinical situations.

In conclusion, our study found an intervention program that was associated with a significant improvement in patient satisfaction in the intervention cohort, even after adjusting for differences in the patient population, whereas there was no change in the control group. This intervention can serve as a model for academic hospitals to improve patient satisfaction, avoid revenue loss in the era of Hospital Value‐Based Purchasing, and to train the next generation of physicians on providing patient‐centered care.

INTRODUCTION

Patient experience and satisfaction is intrinsically valued, as strong physician‐patient communication, empathy, and patient comfort require little justification. However, studies have also shown that patient satisfaction is associated with better health outcomes and greater compliance.[1, 2, 3] A systematic review of studies linking patient satisfaction to outcomes found that patient experience is positively associated with patient safety, clinical effectiveness, health outcomes, adherence, and lower resource utilization.[4] Of 378 associations studied between patient experience and health outcomes, there were 312 positive associations.[4] However, not all studies have shown a positive association between patient satisfaction and outcomes.

Nevertheless, hospitals now have to strive to improve patient satisfaction, as Centers for Medicare & Medicaid Services (CMS) has introduced Hospital Value‐Based Purchasing. CMS started to withhold Medicare Severity Diagnosis‐Related Groups payments, starting at 1.0% in 2013, 1.25% in 2014, and increasing to 2.0% in 2017. This money is redistributed based on performance on core quality measures, including patient satisfaction measured through the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.[5]

Various studies have evaluated interventions to improve patient satisfaction, but to our knowledge, no study published in a peer‐reviewed research journal has shown a significant improvement in HCAHPS scores.[6, 7, 8, 9, 10, 11, 12] Levinson et al. argue that physician communication skills should be taught during residency, and that individualized feedback is an effective way to allow physicians to track their progress over time and compared to their peers.[13] We thus aimed to evaluate an intervention to improve patient satisfaction designed by the Patient Affairs Department for Ronald Reagan University of California, Los Angeles (UCLA) Medical Center (RRUCLAMC) and the UCLA Department of Medicine.

METHODOLOGY

RESULTS

DISCUSSION

Our intervention, which included real‐time feedback to physicians on results of the patient survey, monthly recognition of physicians who stood out on this survey, and an educational conference, was associated with a clear improvement in patient satisfaction with physician‐patient communication and overall recommendation of the hospital. These results are significant because they demonstrate a cost‐effective intervention that can be applied to academic hospitals across the country with the use of nonmedically trained volunteers, such as the undergraduate volunteers involved in our program. The limited costs associated with the intervention were the time in managing the volunteers and movie package award ($20). To our knowledge, it is the first study published in a peer‐reviewed research journal that has demonstrated an intervention associated with significant improvements in HCAHPS scores, the standard by which CMS reimbursement will be affected.

The improvements associated with this intervention could be very valuable to hospitals and patient care. The positive correlation of higher patient satisfaction with improved outcomes suggests this intervention may have additional benefits.[4] Last, these improvements in patient satisfaction in the HCAHPS scores could minimize losses to hospital revenue, as hospitals with low patient‐satisfaction scores will be penalized.

There was a statistically significant improvement in adjusted scores for the question Did your physicians explain things understandably? with patients responding always to all 3 physician‐related HCAHPS questions and Would you recommend this hospital to friends and family. The results for the 2 other physician‐related questions (Did your doctor explain things understandably? and Did your doctor listen carefully?) did show a trend toward significance, with p values of <0.1, and a larger study may have been better powered to detect a statistically significant difference. The improvement in response to the adjusted scores for the question Did your physicians explain things understandably? was the primary driver in the improvement in the adjusted percentage of patients who responded always to all 3 physician‐related HCAHPS questions. This was likely because the IM cohort had the lowest score on this question, and so the feedback to the residents may have helped to address this area of weakness. The UCLA IM HCAHPS scores prior to 2012 have always been lower than other programs at UCLA. As a result, we do not believe the change was due to a regression to the mean.

We believe that the intervention had a positive effect on patient satisfaction for several reasons. The regular e‐mails with the results of the survey may have served as a reminder to residents that patient satisfaction was being monitored and linked to them. The immediate and individualized feedback also may have facilitated adjustments of clinical practice in real time. The residents were able to compare their own scores and comments to the anonymous results of their peers. The monthly department‐wide recognition for residents who excelled in patient communication may have created an incentive and competition among the residents. It is possible that there may be an element of the Hawthorne effect that explained the improvement in HCAHPS scores. However, all of the residents in the departments studied were already being measured through the ARC survey. The primary change was more frequent reporting of ARC survey results, and so we believe that perception of measurement alone was less likely driving the results. The findings from this study are similar to those from provider‐specific report cards, which have shown that outcomes can be improved by forcing greater accountability and competition among physicians.[19]

Brown et al. demonstrated that 2, 4‐hour physician communication workshops in their study had no impact on patient satisfaction, and so we believe that our 1‐hour workshop with only 50% attendance had minimal impact on the improved patient satisfaction scores in our study.[20] Our intervention also coincided with the implementation of the Accreditation Council for Graduate Medical Education (ACGME) work‐hour restrictions implemented in July 2011. These restrictions limited residents to 80 hours per week, intern duty periods were restricted to 16 hours and residents to 28 hours, and interns and residents required 8 to 10 hours free of duty between scheduled duty periods.[21] One of the biggest impacts of ACGME work‐hour restrictions was that interns were doing more day and night shifts rather than 28‐hour calls. However, these work‐hour restrictions were the same for all specialties and so were unlikely to explain the improved patient satisfaction associated with our intervention.

Our study has limitations. The study was a nonrandomized pre‐post study. We attempted to control for the differences in the cohorts with a multivariable regression analysis, but there may be unmeasured differences that we were unable to control for. Due to deidentification of the data, we could only control for patient health based on patient perceived health. In addition, the percentage of patients requiring ICU care in the IM cohort was higher in 2012 than in 2011. We did not identify differences in outcomes from analyses stratified by ICU or non‐ICU patients. In addition, patients who were excluded because of missing outcomes were more likely to be older and admitted through the ER. Further investigation would be needed to see if the findings of this study could be extended to other clinical situations.

In conclusion, our study found an intervention program that was associated with a significant improvement in patient satisfaction in the intervention cohort, even after adjusting for differences in the patient population, whereas there was no change in the control group. This intervention can serve as a model for academic hospitals to improve patient satisfaction, avoid revenue loss in the era of Hospital Value‐Based Purchasing, and to train the next generation of physicians on providing patient‐centered care.

INTRODUCTION

Patient experience and satisfaction is intrinsically valued, as strong physician‐patient communication, empathy, and patient comfort require little justification. However, studies have also shown that patient satisfaction is associated with better health outcomes and greater compliance.[1, 2, 3] A systematic review of studies linking patient satisfaction to outcomes found that patient experience is positively associated with patient safety, clinical effectiveness, health outcomes, adherence, and lower resource utilization.[4] Of 378 associations studied between patient experience and health outcomes, there were 312 positive associations.[4] However, not all studies have shown a positive association between patient satisfaction and outcomes.

Nevertheless, hospitals now have to strive to improve patient satisfaction, as Centers for Medicare & Medicaid Services (CMS) has introduced Hospital Value‐Based Purchasing. CMS started to withhold Medicare Severity Diagnosis‐Related Groups payments, starting at 1.0% in 2013, 1.25% in 2014, and increasing to 2.0% in 2017. This money is redistributed based on performance on core quality measures, including patient satisfaction measured through the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.[5]

Various studies have evaluated interventions to improve patient satisfaction, but to our knowledge, no study published in a peer‐reviewed research journal has shown a significant improvement in HCAHPS scores.[6, 7, 8, 9, 10, 11, 12] Levinson et al. argue that physician communication skills should be taught during residency, and that individualized feedback is an effective way to allow physicians to track their progress over time and compared to their peers.[13] We thus aimed to evaluate an intervention to improve patient satisfaction designed by the Patient Affairs Department for Ronald Reagan University of California, Los Angeles (UCLA) Medical Center (RRUCLAMC) and the UCLA Department of Medicine.

METHODOLOGY

RESULTS

DISCUSSION

Our intervention, which included real‐time feedback to physicians on results of the patient survey, monthly recognition of physicians who stood out on this survey, and an educational conference, was associated with a clear improvement in patient satisfaction with physician‐patient communication and overall recommendation of the hospital. These results are significant because they demonstrate a cost‐effective intervention that can be applied to academic hospitals across the country with the use of nonmedically trained volunteers, such as the undergraduate volunteers involved in our program. The limited costs associated with the intervention were the time in managing the volunteers and movie package award ($20). To our knowledge, it is the first study published in a peer‐reviewed research journal that has demonstrated an intervention associated with significant improvements in HCAHPS scores, the standard by which CMS reimbursement will be affected.

The improvements associated with this intervention could be very valuable to hospitals and patient care. The positive correlation of higher patient satisfaction with improved outcomes suggests this intervention may have additional benefits.[4] Last, these improvements in patient satisfaction in the HCAHPS scores could minimize losses to hospital revenue, as hospitals with low patient‐satisfaction scores will be penalized.

There was a statistically significant improvement in adjusted scores for the question Did your physicians explain things understandably? with patients responding always to all 3 physician‐related HCAHPS questions and Would you recommend this hospital to friends and family. The results for the 2 other physician‐related questions (Did your doctor explain things understandably? and Did your doctor listen carefully?) did show a trend toward significance, with p values of <0.1, and a larger study may have been better powered to detect a statistically significant difference. The improvement in response to the adjusted scores for the question Did your physicians explain things understandably? was the primary driver in the improvement in the adjusted percentage of patients who responded always to all 3 physician‐related HCAHPS questions. This was likely because the IM cohort had the lowest score on this question, and so the feedback to the residents may have helped to address this area of weakness. The UCLA IM HCAHPS scores prior to 2012 have always been lower than other programs at UCLA. As a result, we do not believe the change was due to a regression to the mean.

We believe that the intervention had a positive effect on patient satisfaction for several reasons. The regular e‐mails with the results of the survey may have served as a reminder to residents that patient satisfaction was being monitored and linked to them. The immediate and individualized feedback also may have facilitated adjustments of clinical practice in real time. The residents were able to compare their own scores and comments to the anonymous results of their peers. The monthly department‐wide recognition for residents who excelled in patient communication may have created an incentive and competition among the residents. It is possible that there may be an element of the Hawthorne effect that explained the improvement in HCAHPS scores. However, all of the residents in the departments studied were already being measured through the ARC survey. The primary change was more frequent reporting of ARC survey results, and so we believe that perception of measurement alone was less likely driving the results. The findings from this study are similar to those from provider‐specific report cards, which have shown that outcomes can be improved by forcing greater accountability and competition among physicians.[19]

Brown et al. demonstrated that 2, 4‐hour physician communication workshops in their study had no impact on patient satisfaction, and so we believe that our 1‐hour workshop with only 50% attendance had minimal impact on the improved patient satisfaction scores in our study.[20] Our intervention also coincided with the implementation of the Accreditation Council for Graduate Medical Education (ACGME) work‐hour restrictions implemented in July 2011. These restrictions limited residents to 80 hours per week, intern duty periods were restricted to 16 hours and residents to 28 hours, and interns and residents required 8 to 10 hours free of duty between scheduled duty periods.[21] One of the biggest impacts of ACGME work‐hour restrictions was that interns were doing more day and night shifts rather than 28‐hour calls. However, these work‐hour restrictions were the same for all specialties and so were unlikely to explain the improved patient satisfaction associated with our intervention.

Our study has limitations. The study was a nonrandomized pre‐post study. We attempted to control for the differences in the cohorts with a multivariable regression analysis, but there may be unmeasured differences that we were unable to control for. Due to deidentification of the data, we could only control for patient health based on patient perceived health. In addition, the percentage of patients requiring ICU care in the IM cohort was higher in 2012 than in 2011. We did not identify differences in outcomes from analyses stratified by ICU or non‐ICU patients. In addition, patients who were excluded because of missing outcomes were more likely to be older and admitted through the ER. Further investigation would be needed to see if the findings of this study could be extended to other clinical situations.

In conclusion, our study found an intervention program that was associated with a significant improvement in patient satisfaction in the intervention cohort, even after adjusting for differences in the patient population, whereas there was no change in the control group. This intervention can serve as a model for academic hospitals to improve patient satisfaction, avoid revenue loss in the era of Hospital Value‐Based Purchasing, and to train the next generation of physicians on providing patient‐centered care.

A 51‐year‐old man presented to the emergency department after 1 day of progressive dyspnea and increasing confusion.

Acute dyspnea most commonly stems from a cardiac or pulmonary disorder such as heart failure, acute coronary syndrome, pneumonia, pulmonary embolism, or exacerbations of asthma or chronic obstructive pulmonary disease. Less frequent cardiopulmonary considerations include pericardial or pleural effusion, pneumothorax, aspiration, and upper airway obstruction. Dyspnea might also be the initial manifestation of profound anemia or metabolic acidosis.

The presence of confusion suggests either a severe presentation of any of the aforementioned possibilities (with confusion resulting from hypoxia, hypercapnia, or hypotension); a multiorgan illness such as sepsis, malignancy, thromboembolic disease, vasculitis, thyroid dysfunction, or toxic ingestion; or a metabolic derangement related to the underlying cause of dyspnea (for example, hypercalcemia or hyponatremia associated with lung cancer).

Twelve hours prior to presentation, he started to have visual hallucinations. He denied fever, chills, cough, chest discomfort, palpitations, weight gain, headache, neck pain, or weakness.

Visual hallucinations could result from a toxic‐metabolic encephalopathy, such as drug overdose or withdrawal, liver or kidney failure, or hypoxia. A structural brain abnormality may also manifest with visual hallucination. Acute onset at age 51 and the absence of auditory hallucinations argue against a neurodegenerative illness and a primary psychiatric disturbance, respectively.

Episodic hallucinations would support the possibility of seizures, monocular hallucinations would point to a retinal or ocular problem, and a description of yellow‐green hue would suggest a side effect of digoxin.

His past medical history was remarkable for diet‐controlled type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic low back pain. His medications included metoprolol tartrate 25 mg twice daily, omeprazole 40 mg daily, baclofen 15 mg twice daily, oxycodone 30 mg 3 times daily, and hydrocodone 10 mg/acetaminophen 325 mg, 2 tablets 3 times daily as needed for back pain. He was a smoker with a 30 pack‐year history. He had a history of alcohol and cocaine use, but denied any recent substance use. He had no known history of obstructive pulmonary disease.

The patient takes 3 medications well known to cause confusion and hallucinations (oxycodone, hydrocodone, and baclofen), especially when they accumulate due to excessive ingestion or impaired clearance. Although these medications may suppress ventilatory drive, dyspnea would not be a common presenting complaint. He has risk factors for ischemic heart disease and cardiomyopathy, and his smoking history raises the possibility of malignancy.

On exam, the patient's temperature was 94.4C, heart rate 128 beats per minute, respiration rate 28 breaths per minute, blood pressure 155/63 mm Hg, and oxygen saturation 100% while breathing ambient air. The patient was cachectic and appeared in moderate respiratory distress. His pupils were equal and reactive to light, and extraocular movements were intact. He did not have scleral icterus, or cervical or clavicular lymphadenopathy. His oropharynx was negative for erythema, edema, or exudate. His cardiovascular exam revealed a regular tachycardia without rubs or diastolic gallops. There was a 2/6 systolic murmur heard best at left sternal border, without radiation. He did not have jugular venous distention. His pulmonary exam was notable for tachypnea but with normal vesicular breath sounds throughout. He did not have stridor, wheezing, rhonchi, or rales. His abdomen had normal bowel tones and was soft without tenderness, distention, or organomegaly. His extremities were warm, revealed normal pulses, and no edema was present. His joints were cool to palpation, without effusion. On neurologic exam, he was oriented to person and place and able to answer yes/no questions, but unable to provide detailed history. His speech was fluent. His motor exam was without focal deficits. His skin was without any notable lesions.

The constellation of findings does not point to a specific toxidrome. The finding of warm extremities in a hypothermic patient suggests heat loss due to inappropriate peripheral vasodilation. In the absence of vasodilators or features of aortic insufficiency, sepsis becomes a leading consideration. Infection could result in hypothermia and altered sensorium, and accompanying lactic acidosis could trigger tachypnea.

Shortly after admission, he became more somnolent and developed progressive respiratory distress, requiring intubation. Arterial blood gas revealed a pH of 6.93, PaCO₂<20 mm Hg, PaO₂ 127 mm Hg, and HCO₃<5 mEq/L. Other laboratory results included a lactate of 4.1 mmol/L, blood urea nitrogen 49 mg/dL, creatinine 2.3 mg/dL (0.8 at 1 month prior), sodium level of 143 mmol/L, chloride of 106 mmol/L, and bicarbonate level of <5 mg/dL. His aspartate aminotransferase was 34 IU/L, alanine transaminase was 28 IU/L, total bilirubin was 0.6 mg/dL, International Normalized Ratio was 1.3. A complete blood count revealed a white blood cell count of 23,000/L, hemoglobin of 10.6 g/dL, and platelet count of 454,000/L. A urinalysis was unremarkable. Cultures of blood, urine, and sputum were collected. Head computed tomography was negative.

This patient has a combined anion gap and nongap metabolic acidosis, as well as respiratory alkalosis. Although his acute kidney failure could produce these 2 types of metabolic acidosis, the modest elevation of the serum creatinine is not commensurate with such profound acidosis. Similarly, sepsis without hypotension or more striking elevation in lactate levels would not account for the entirety of the acidosis. Severe diabetic ketoacidosis can result in profound metabolic acidosis, and marked hyperglycemia or hyperosmolarity could result in somnolence; however, his diabetes has been controlled without medication and there is no obvious precipitant for an episode of ketoacidosis.

Remaining causes of anion gap acidosis include ingestion of methanol, ethylene glycol, ethanol, or salicylates. A careful history of ingestions and medications from witnesses including any prehospital personnel might suggest a source of intoxication. Absent this information, the hypothermia favors an ingestion of an alcohol over salicylates, and the lack of urine crystals and the presence of prominent visual hallucinations would point more toward methanol poisoning than ethylene glycol. A serum osmolarity measurement would allow determination of the osmolar gap, which would be elevated in the setting of methanol or ethylene glycol poisoning. If he were this ill from ethanol, I would have expected to see evidence of hepatotoxicity.

I would administer sodium bicarbonate to reverse the acidosis and to promote renal clearance of salicylates, methanol, ethylene glycol, and their metabolites. Orogastric decontamination with activated charcoal should be considered. If the osmolar gap is elevated, I would also administer intravenous fomepizole to attempt to reverse methanol or ethylene glycol poisoning. I would not delay treatment while waiting for these serum levels to return.

Initial serologic toxicology performed in the emergency department revealed negative ethanol, salicylates, and ketones. His osmolar gap was 13 mOsm/kg. His acetaminophen level was 69 g/mL (normal <120 g/mL). A creatinine phosphokinase was 84 IU/L and myoglobin was 93 ng/mL. His subsequent serum toxicology screen was negative for methanol, ethylene glycol, isopropranol, and hippuric acid. Urine toxicology was positive for opiates, but negative for amphetamine, benzodiazepine, cannabinoid, and cocaine.

Serum and urine ketone assays typically involve the nitroprusside reaction and detect acetoacetate, but not ‐hydroxybutyrate, and can lead to negative test results early in diabetic or alcoholic ketoacidosis. However, the normal ethanol level argues against alcoholic ketoacidosis. Rare causes of elevated anion gap acidosis include toluene toxicity, acetaminophen poisoning, and ingestion of other alcohols. Toluene is metabolized to hippuric acid, and acetaminophen toxicity and associated glutathione depletion can lead to 5‐oxoproline accumulation, producing an anion gap. Patients who abuse alcohol are at risk for acetaminophen toxicity even at doses considered normal. However, this degree of encephalopathy would be unusual for acetaminophen toxicity unless liver failure had developed or unless there was another ingestion that might alter sensorium. Furthermore, the elevated osmolar gap is not a feature of acetaminophen poisoning. I would monitor liver enzyme tests and consider a serum ammonia level, but would not attribute the entire picture to acetaminophen.

The combination of elevated anion gap with an elevated osmolar gap narrows the diagnostic possibilities. Ingestion of several alcohols (ethanol, methanol, ethylene glycol, diethylene glycol) or toluene could produce these abnormalities. Of note, the osmolar gap is typically most markedly elevated early in methanol and ethylene glycol ingestions, and then as the parent compound is metabolized, the osmolar gap closes and the accumulation of metabolites produces the anion gap. Hallucinations are more common with methanol and toluene, and renal failure is more typical of ethylene glycol or toluene. The lack of oxalate crystalluria does not exclude ethylene glycol poisoning. Unfortunately, urine testing for oxalate crystals or fluorescein examination are neither sensitive nor specific enough to diagnosis ethylene glycol toxicity reliably. In most hospitals, assays used for serum testing for alcohols are insensitive, and require confirmation with gas chromatography performed at a specialty lab.

Additional history might reveal the likely culprit or culprits. Inhalant abuse including huffing would point to toluene or organic acid exposure. Solvent ingestion (eg, antifreeze, brake fluid) would suggest methanol or ethylene glycol. Absent this history, I remain suspicious for poisoning with methanol or ethylene glycol and would consider empiric treatment after urgent consultation with a medical toxicologist. A careful ophthalmologic exam might demonstrate characteristic features of methanol poisoning. Serum samples should be sent to a regional lab for analysis for alcohols and organic acids.

He was admitted to the intensive care unit, and empiric antibiotics started. He was empirically started on N‐acetylcysteine and sodium bicarbonate drips. However, his acidemia persisted and he required hemodialysis, which was initiated 12 hours after initial presentation. His acidemia and mental status quickly improved after hemodialysis. He was extubated on hospital day 2 and no longer required hemodialysis.

The differential diagnosis at this point consists of 3 main possibilities: ingestion of methanol, ethylene glycol, or inhalant abuse such as from toluene. The normal hippuric acid level points away from toluene, whereas serum levels can be misleading in the alcohol poisonings. Other discriminating features to consider include exposure history and unique clinical aspects. In this patient, an exposure history is lacking, but 4 clinical features stand out: visual hallucinations, acute kidney injury, mild lactic acidosis, and rapid improvement with hemodialysis. Both ethylene glycol and methanol toxicity may produce a mild lactic acidosis by increasing hepatic metabolism of pyruvate to lactate, and both are rapidly cleared by dialysis. Although it is tempting to place methanol at the top of the list of possibilities due to the report of visual hallucinations, the subjective visual complaints without objective exam corollaries (loss of visual acuity, abnormal pupillary reflexes, or optic disc hyperemia) are nonspecific and might be provoked by alcohol or an inhalant. Furthermore, the acute renal failure is much more typical of ethylene glycol, and thus I would consider ethylene glycol as being the more likely of the ingestions. Coingestion of multiple alcohols is a possibility, but it would be statistically less likely. Confirmation of ethylene glycol poisoning would consist of further insight into his exposures and measurement of levels using gas chromatography.

A urine sample from his emergency department presentation was sent to an outside lab for organic acid levels. Based on high clinical suspicion for 5‐oxoprolinemia (pyroglutamic acidemia) the patient was counseled to avoid any acetaminophen. His primary care provider was informed of this and acetaminophen was added as an adverse drug reaction. The patient left against medical advice soon after extubation. Following discharge, his 5‐oxoproline (pyroglutamic acid) level returned markedly elevated at greater than 10,000 mmol/mol creatinine (200 times the upper limit of normal).

Elevations in 5‐oxoproline levels in this patient most likely stem from glutathione depletion related to chronic acetaminophen use. Alcohol use and malnutrition may have heightened this patient's susceptibility. Despite the common occurrence of acetaminophen use in alcohol abusers or the malnourished, the rarity of severe 5‐oxoproline toxicity suggests unknown factors may be present in predisposed individuals, or under‐recognition. Although acetaminophen‐induced hepatotoxicity may occur along with 5‐oxoprolinemia, this does not always occur.

Several features led me away from this syndrome. First, its rarity lowered my pretest probability. Second, the lack of exposure history and details about the serum assays, specifically whether the measurements were confirmed by gas chromatography, reduced my confidence in eliminating more common ingestions. Third, several aspects proved to be less useful discriminating features: the mild elevation in osmolar gap, renal failure, and hallucinations, which in retrospect proved to be nonspecific.

The patient admitted that he had a longstanding use of acetaminophen in addition to using his girlfriend's acetaminophen‐hydrocodone. He had significant weight loss of over 50 pounds over the previous year, which he attributed to poor appetite. On further chart review, he had been admitted 3 times with a similar clinical presentation and recovered quickly with intensive and supportive care, with no etiology found at those times. He had 2 subsequent hospital admissions for altered mental status and respiratory failure, and his final hospitalization resulted in cardiac arrest and death.

DISCUSSION

5‐Oxoprolinemia is a rare, but potentially lethal cause of severe anion gap metabolic acidosis.[1, 2] The mechanism is thought to be impairment of glutathione metabolism, in the context of other predisposing factors. This can be a congenital error of metabolism, or can be acquired and exacerbated by acetaminophen use. Ingestion of acetaminophen leads to glutathione depletion, which in turn may precipitate accumulation of pyroglutamic acid and subsequent anion gap metabolic acidosis (Figure 1). Additional risk factors that may predispose patients to this condition include malnutrition, renal insufficiency, concurrent infection, and female gender.[1, 2, 3]

Figure 1

The diagnosis of 5‐oxoprolinemia is made via urine or serum organic acid analysis, testing routinely performed in pediatric populations when screening for congenital metabolic disorders. The pathophysiology suggests that obtaining a urine sample early in presentation, when acidosis is greatest, would lead to the highest 5‐oxoproline levels and best chance for diagnosis. Case patients have had normal levels prior to and in convalescent phases after the acute episode.[4] Given the long turnaround time for lab testing, presumptive diagnosis and treatment may be necessary.

Treatment of 5‐oxoprolinemia is primarily supportive, aimed at the metabolic acidosis. Fluid resuscitation and bicarbonate therapy are reasonable temporizing measures. Hemodialysis can clear 5‐oxoproline and may be indicated in severe acidosis.[5] Furthermore, the proposed pathophysiology suggests that administration of N‐acetylcysteine (NAC) may help to address the underlying process, but there are no trials to support a specific dosing regimen. However, given the fulminant presentation and common competing concern for acetaminophen toxicity, it is reasonable to initiate NAC aimed at treatment for possible acetaminophen overdose. Prevention of recurrence includes avoidance of acetaminophen, and counseling the patient to avoid acetaminophen in prescription combination medications and over‐the‐counter preparations.

Recent regulatory changes regarding acetaminophen/opioid combinations may reduce the incidence of 5‐oxoprolinemia. The US Food and Drug Administration has taken action to reduce adverse effects from acetaminophen exposure by limiting the amount of acetaminophen in opioid combination pills from 500 mg to a maximum of 325 mg per pill. This is aimed at preventing hepatotoxicity from ingestion of higher‐than‐recommended doses. However, clinicians should remember that 5‐oxoprolinemia can result from ingestion of acetaminophen at therapeutic levels.

Given its rare incidence, low clinical suspicion, and transient nature of confirmatory testing, it is likely this remains an underdiagnosed syndrome. In the case discussed, subsequent chart review demonstrated 5 previous admissions in multiple hospitals for severe transient anion gap acidosis. The likelihood that 5‐oxoprolinemia was missed in each of these cases supports a lack of awareness of this syndrome. In this patient, the discussant appropriately identified the possibility of 5‐oxoproline toxicity, but felt ethylene glycol ingestion was more likely. As this case underscores, a cornerstone in the management of suspected ingestions is empiric treatment for the most likely etiologies. Here, treatment for acetaminophen overdose and for methanol or ethylene glycol were warranted, and fortunately also addressed the rarer possibility of 5‐oxoproline toxicity.

The mnemonic MUDPILES is commonly used to identify possible causes of life‐threatening anion gap metabolic acidosis, as such heuristics have benefits in rapidly generating a differential diagnosis to guide initial evaluation. Given the fact that the traditional letter P (paraldehyde) in MUDPILES is no longer clinically utilized, some authors have suggested replacing this with pyroglutamic acid (a synonym of 5‐oxoproline). Such a change may help providers who have ruled out other causes of a high anion gap metabolic acidosis, facilitating diagnosis of this life‐threatening syndrome. In any case, clinicians must be mindful that simple memory aids may mislead clinicians, and a complete differential diagnosis may require more than a mnemonic.

TEACHING POINTS

1. Acetaminophen use, even at therapeutic levels, can lead to 5‐oxoprolinemia, a potentially lethal anion gap metabolic acidosis.
2. 5‐oxoprolinemia is likely related to glutathione depletion, worsened by acetaminophen, malnutrition, renal insufficiency, female gender, and infection. This implies theoretical benefit from administration of NAC for glutathione repletion.
3. Mnemonics can be useful, but have limitations by way of oversimplification. This case suggests that changing the letter P in MUDPILES from paraldehyde to pyroglutamic acid could reduce underdiagnosis.

Disclosure: Nothing to report.

A 51‐year‐old man presented to the emergency department after 1 day of progressive dyspnea and increasing confusion.

Acute dyspnea most commonly stems from a cardiac or pulmonary disorder such as heart failure, acute coronary syndrome, pneumonia, pulmonary embolism, or exacerbations of asthma or chronic obstructive pulmonary disease. Less frequent cardiopulmonary considerations include pericardial or pleural effusion, pneumothorax, aspiration, and upper airway obstruction. Dyspnea might also be the initial manifestation of profound anemia or metabolic acidosis.

The presence of confusion suggests either a severe presentation of any of the aforementioned possibilities (with confusion resulting from hypoxia, hypercapnia, or hypotension); a multiorgan illness such as sepsis, malignancy, thromboembolic disease, vasculitis, thyroid dysfunction, or toxic ingestion; or a metabolic derangement related to the underlying cause of dyspnea (for example, hypercalcemia or hyponatremia associated with lung cancer).

Twelve hours prior to presentation, he started to have visual hallucinations. He denied fever, chills, cough, chest discomfort, palpitations, weight gain, headache, neck pain, or weakness.

Visual hallucinations could result from a toxic‐metabolic encephalopathy, such as drug overdose or withdrawal, liver or kidney failure, or hypoxia. A structural brain abnormality may also manifest with visual hallucination. Acute onset at age 51 and the absence of auditory hallucinations argue against a neurodegenerative illness and a primary psychiatric disturbance, respectively.

Episodic hallucinations would support the possibility of seizures, monocular hallucinations would point to a retinal or ocular problem, and a description of yellow‐green hue would suggest a side effect of digoxin.

His past medical history was remarkable for diet‐controlled type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic low back pain. His medications included metoprolol tartrate 25 mg twice daily, omeprazole 40 mg daily, baclofen 15 mg twice daily, oxycodone 30 mg 3 times daily, and hydrocodone 10 mg/acetaminophen 325 mg, 2 tablets 3 times daily as needed for back pain. He was a smoker with a 30 pack‐year history. He had a history of alcohol and cocaine use, but denied any recent substance use. He had no known history of obstructive pulmonary disease.

The patient takes 3 medications well known to cause confusion and hallucinations (oxycodone, hydrocodone, and baclofen), especially when they accumulate due to excessive ingestion or impaired clearance. Although these medications may suppress ventilatory drive, dyspnea would not be a common presenting complaint. He has risk factors for ischemic heart disease and cardiomyopathy, and his smoking history raises the possibility of malignancy.

On exam, the patient's temperature was 94.4C, heart rate 128 beats per minute, respiration rate 28 breaths per minute, blood pressure 155/63 mm Hg, and oxygen saturation 100% while breathing ambient air. The patient was cachectic and appeared in moderate respiratory distress. His pupils were equal and reactive to light, and extraocular movements were intact. He did not have scleral icterus, or cervical or clavicular lymphadenopathy. His oropharynx was negative for erythema, edema, or exudate. His cardiovascular exam revealed a regular tachycardia without rubs or diastolic gallops. There was a 2/6 systolic murmur heard best at left sternal border, without radiation. He did not have jugular venous distention. His pulmonary exam was notable for tachypnea but with normal vesicular breath sounds throughout. He did not have stridor, wheezing, rhonchi, or rales. His abdomen had normal bowel tones and was soft without tenderness, distention, or organomegaly. His extremities were warm, revealed normal pulses, and no edema was present. His joints were cool to palpation, without effusion. On neurologic exam, he was oriented to person and place and able to answer yes/no questions, but unable to provide detailed history. His speech was fluent. His motor exam was without focal deficits. His skin was without any notable lesions.

The constellation of findings does not point to a specific toxidrome. The finding of warm extremities in a hypothermic patient suggests heat loss due to inappropriate peripheral vasodilation. In the absence of vasodilators or features of aortic insufficiency, sepsis becomes a leading consideration. Infection could result in hypothermia and altered sensorium, and accompanying lactic acidosis could trigger tachypnea.

Shortly after admission, he became more somnolent and developed progressive respiratory distress, requiring intubation. Arterial blood gas revealed a pH of 6.93, PaCO₂<20 mm Hg, PaO₂ 127 mm Hg, and HCO₃<5 mEq/L. Other laboratory results included a lactate of 4.1 mmol/L, blood urea nitrogen 49 mg/dL, creatinine 2.3 mg/dL (0.8 at 1 month prior), sodium level of 143 mmol/L, chloride of 106 mmol/L, and bicarbonate level of <5 mg/dL. His aspartate aminotransferase was 34 IU/L, alanine transaminase was 28 IU/L, total bilirubin was 0.6 mg/dL, International Normalized Ratio was 1.3. A complete blood count revealed a white blood cell count of 23,000/L, hemoglobin of 10.6 g/dL, and platelet count of 454,000/L. A urinalysis was unremarkable. Cultures of blood, urine, and sputum were collected. Head computed tomography was negative.

This patient has a combined anion gap and nongap metabolic acidosis, as well as respiratory alkalosis. Although his acute kidney failure could produce these 2 types of metabolic acidosis, the modest elevation of the serum creatinine is not commensurate with such profound acidosis. Similarly, sepsis without hypotension or more striking elevation in lactate levels would not account for the entirety of the acidosis. Severe diabetic ketoacidosis can result in profound metabolic acidosis, and marked hyperglycemia or hyperosmolarity could result in somnolence; however, his diabetes has been controlled without medication and there is no obvious precipitant for an episode of ketoacidosis.

Remaining causes of anion gap acidosis include ingestion of methanol, ethylene glycol, ethanol, or salicylates. A careful history of ingestions and medications from witnesses including any prehospital personnel might suggest a source of intoxication. Absent this information, the hypothermia favors an ingestion of an alcohol over salicylates, and the lack of urine crystals and the presence of prominent visual hallucinations would point more toward methanol poisoning than ethylene glycol. A serum osmolarity measurement would allow determination of the osmolar gap, which would be elevated in the setting of methanol or ethylene glycol poisoning. If he were this ill from ethanol, I would have expected to see evidence of hepatotoxicity.

I would administer sodium bicarbonate to reverse the acidosis and to promote renal clearance of salicylates, methanol, ethylene glycol, and their metabolites. Orogastric decontamination with activated charcoal should be considered. If the osmolar gap is elevated, I would also administer intravenous fomepizole to attempt to reverse methanol or ethylene glycol poisoning. I would not delay treatment while waiting for these serum levels to return.

Initial serologic toxicology performed in the emergency department revealed negative ethanol, salicylates, and ketones. His osmolar gap was 13 mOsm/kg. His acetaminophen level was 69 g/mL (normal <120 g/mL). A creatinine phosphokinase was 84 IU/L and myoglobin was 93 ng/mL. His subsequent serum toxicology screen was negative for methanol, ethylene glycol, isopropranol, and hippuric acid. Urine toxicology was positive for opiates, but negative for amphetamine, benzodiazepine, cannabinoid, and cocaine.

Serum and urine ketone assays typically involve the nitroprusside reaction and detect acetoacetate, but not ‐hydroxybutyrate, and can lead to negative test results early in diabetic or alcoholic ketoacidosis. However, the normal ethanol level argues against alcoholic ketoacidosis. Rare causes of elevated anion gap acidosis include toluene toxicity, acetaminophen poisoning, and ingestion of other alcohols. Toluene is metabolized to hippuric acid, and acetaminophen toxicity and associated glutathione depletion can lead to 5‐oxoproline accumulation, producing an anion gap. Patients who abuse alcohol are at risk for acetaminophen toxicity even at doses considered normal. However, this degree of encephalopathy would be unusual for acetaminophen toxicity unless liver failure had developed or unless there was another ingestion that might alter sensorium. Furthermore, the elevated osmolar gap is not a feature of acetaminophen poisoning. I would monitor liver enzyme tests and consider a serum ammonia level, but would not attribute the entire picture to acetaminophen.

The combination of elevated anion gap with an elevated osmolar gap narrows the diagnostic possibilities. Ingestion of several alcohols (ethanol, methanol, ethylene glycol, diethylene glycol) or toluene could produce these abnormalities. Of note, the osmolar gap is typically most markedly elevated early in methanol and ethylene glycol ingestions, and then as the parent compound is metabolized, the osmolar gap closes and the accumulation of metabolites produces the anion gap. Hallucinations are more common with methanol and toluene, and renal failure is more typical of ethylene glycol or toluene. The lack of oxalate crystalluria does not exclude ethylene glycol poisoning. Unfortunately, urine testing for oxalate crystals or fluorescein examination are neither sensitive nor specific enough to diagnosis ethylene glycol toxicity reliably. In most hospitals, assays used for serum testing for alcohols are insensitive, and require confirmation with gas chromatography performed at a specialty lab.

Additional history might reveal the likely culprit or culprits. Inhalant abuse including huffing would point to toluene or organic acid exposure. Solvent ingestion (eg, antifreeze, brake fluid) would suggest methanol or ethylene glycol. Absent this history, I remain suspicious for poisoning with methanol or ethylene glycol and would consider empiric treatment after urgent consultation with a medical toxicologist. A careful ophthalmologic exam might demonstrate characteristic features of methanol poisoning. Serum samples should be sent to a regional lab for analysis for alcohols and organic acids.

He was admitted to the intensive care unit, and empiric antibiotics started. He was empirically started on N‐acetylcysteine and sodium bicarbonate drips. However, his acidemia persisted and he required hemodialysis, which was initiated 12 hours after initial presentation. His acidemia and mental status quickly improved after hemodialysis. He was extubated on hospital day 2 and no longer required hemodialysis.

The differential diagnosis at this point consists of 3 main possibilities: ingestion of methanol, ethylene glycol, or inhalant abuse such as from toluene. The normal hippuric acid level points away from toluene, whereas serum levels can be misleading in the alcohol poisonings. Other discriminating features to consider include exposure history and unique clinical aspects. In this patient, an exposure history is lacking, but 4 clinical features stand out: visual hallucinations, acute kidney injury, mild lactic acidosis, and rapid improvement with hemodialysis. Both ethylene glycol and methanol toxicity may produce a mild lactic acidosis by increasing hepatic metabolism of pyruvate to lactate, and both are rapidly cleared by dialysis. Although it is tempting to place methanol at the top of the list of possibilities due to the report of visual hallucinations, the subjective visual complaints without objective exam corollaries (loss of visual acuity, abnormal pupillary reflexes, or optic disc hyperemia) are nonspecific and might be provoked by alcohol or an inhalant. Furthermore, the acute renal failure is much more typical of ethylene glycol, and thus I would consider ethylene glycol as being the more likely of the ingestions. Coingestion of multiple alcohols is a possibility, but it would be statistically less likely. Confirmation of ethylene glycol poisoning would consist of further insight into his exposures and measurement of levels using gas chromatography.

A urine sample from his emergency department presentation was sent to an outside lab for organic acid levels. Based on high clinical suspicion for 5‐oxoprolinemia (pyroglutamic acidemia) the patient was counseled to avoid any acetaminophen. His primary care provider was informed of this and acetaminophen was added as an adverse drug reaction. The patient left against medical advice soon after extubation. Following discharge, his 5‐oxoproline (pyroglutamic acid) level returned markedly elevated at greater than 10,000 mmol/mol creatinine (200 times the upper limit of normal).

Elevations in 5‐oxoproline levels in this patient most likely stem from glutathione depletion related to chronic acetaminophen use. Alcohol use and malnutrition may have heightened this patient's susceptibility. Despite the common occurrence of acetaminophen use in alcohol abusers or the malnourished, the rarity of severe 5‐oxoproline toxicity suggests unknown factors may be present in predisposed individuals, or under‐recognition. Although acetaminophen‐induced hepatotoxicity may occur along with 5‐oxoprolinemia, this does not always occur.

Several features led me away from this syndrome. First, its rarity lowered my pretest probability. Second, the lack of exposure history and details about the serum assays, specifically whether the measurements were confirmed by gas chromatography, reduced my confidence in eliminating more common ingestions. Third, several aspects proved to be less useful discriminating features: the mild elevation in osmolar gap, renal failure, and hallucinations, which in retrospect proved to be nonspecific.

The patient admitted that he had a longstanding use of acetaminophen in addition to using his girlfriend's acetaminophen‐hydrocodone. He had significant weight loss of over 50 pounds over the previous year, which he attributed to poor appetite. On further chart review, he had been admitted 3 times with a similar clinical presentation and recovered quickly with intensive and supportive care, with no etiology found at those times. He had 2 subsequent hospital admissions for altered mental status and respiratory failure, and his final hospitalization resulted in cardiac arrest and death.

DISCUSSION

5‐Oxoprolinemia is a rare, but potentially lethal cause of severe anion gap metabolic acidosis.[1, 2] The mechanism is thought to be impairment of glutathione metabolism, in the context of other predisposing factors. This can be a congenital error of metabolism, or can be acquired and exacerbated by acetaminophen use. Ingestion of acetaminophen leads to glutathione depletion, which in turn may precipitate accumulation of pyroglutamic acid and subsequent anion gap metabolic acidosis (Figure 1). Additional risk factors that may predispose patients to this condition include malnutrition, renal insufficiency, concurrent infection, and female gender.[1, 2, 3]

Figure 1

The diagnosis of 5‐oxoprolinemia is made via urine or serum organic acid analysis, testing routinely performed in pediatric populations when screening for congenital metabolic disorders. The pathophysiology suggests that obtaining a urine sample early in presentation, when acidosis is greatest, would lead to the highest 5‐oxoproline levels and best chance for diagnosis. Case patients have had normal levels prior to and in convalescent phases after the acute episode.[4] Given the long turnaround time for lab testing, presumptive diagnosis and treatment may be necessary.

Treatment of 5‐oxoprolinemia is primarily supportive, aimed at the metabolic acidosis. Fluid resuscitation and bicarbonate therapy are reasonable temporizing measures. Hemodialysis can clear 5‐oxoproline and may be indicated in severe acidosis.[5] Furthermore, the proposed pathophysiology suggests that administration of N‐acetylcysteine (NAC) may help to address the underlying process, but there are no trials to support a specific dosing regimen. However, given the fulminant presentation and common competing concern for acetaminophen toxicity, it is reasonable to initiate NAC aimed at treatment for possible acetaminophen overdose. Prevention of recurrence includes avoidance of acetaminophen, and counseling the patient to avoid acetaminophen in prescription combination medications and over‐the‐counter preparations.

Recent regulatory changes regarding acetaminophen/opioid combinations may reduce the incidence of 5‐oxoprolinemia. The US Food and Drug Administration has taken action to reduce adverse effects from acetaminophen exposure by limiting the amount of acetaminophen in opioid combination pills from 500 mg to a maximum of 325 mg per pill. This is aimed at preventing hepatotoxicity from ingestion of higher‐than‐recommended doses. However, clinicians should remember that 5‐oxoprolinemia can result from ingestion of acetaminophen at therapeutic levels.

Given its rare incidence, low clinical suspicion, and transient nature of confirmatory testing, it is likely this remains an underdiagnosed syndrome. In the case discussed, subsequent chart review demonstrated 5 previous admissions in multiple hospitals for severe transient anion gap acidosis. The likelihood that 5‐oxoprolinemia was missed in each of these cases supports a lack of awareness of this syndrome. In this patient, the discussant appropriately identified the possibility of 5‐oxoproline toxicity, but felt ethylene glycol ingestion was more likely. As this case underscores, a cornerstone in the management of suspected ingestions is empiric treatment for the most likely etiologies. Here, treatment for acetaminophen overdose and for methanol or ethylene glycol were warranted, and fortunately also addressed the rarer possibility of 5‐oxoproline toxicity.

The mnemonic MUDPILES is commonly used to identify possible causes of life‐threatening anion gap metabolic acidosis, as such heuristics have benefits in rapidly generating a differential diagnosis to guide initial evaluation. Given the fact that the traditional letter P (paraldehyde) in MUDPILES is no longer clinically utilized, some authors have suggested replacing this with pyroglutamic acid (a synonym of 5‐oxoproline). Such a change may help providers who have ruled out other causes of a high anion gap metabolic acidosis, facilitating diagnosis of this life‐threatening syndrome. In any case, clinicians must be mindful that simple memory aids may mislead clinicians, and a complete differential diagnosis may require more than a mnemonic.

TEACHING POINTS

1. Acetaminophen use, even at therapeutic levels, can lead to 5‐oxoprolinemia, a potentially lethal anion gap metabolic acidosis.
2. 5‐oxoprolinemia is likely related to glutathione depletion, worsened by acetaminophen, malnutrition, renal insufficiency, female gender, and infection. This implies theoretical benefit from administration of NAC for glutathione repletion.
3. Mnemonics can be useful, but have limitations by way of oversimplification. This case suggests that changing the letter P in MUDPILES from paraldehyde to pyroglutamic acid could reduce underdiagnosis.

Disclosure: Nothing to report.

A 51‐year‐old man presented to the emergency department after 1 day of progressive dyspnea and increasing confusion.

Acute dyspnea most commonly stems from a cardiac or pulmonary disorder such as heart failure, acute coronary syndrome, pneumonia, pulmonary embolism, or exacerbations of asthma or chronic obstructive pulmonary disease. Less frequent cardiopulmonary considerations include pericardial or pleural effusion, pneumothorax, aspiration, and upper airway obstruction. Dyspnea might also be the initial manifestation of profound anemia or metabolic acidosis.

The presence of confusion suggests either a severe presentation of any of the aforementioned possibilities (with confusion resulting from hypoxia, hypercapnia, or hypotension); a multiorgan illness such as sepsis, malignancy, thromboembolic disease, vasculitis, thyroid dysfunction, or toxic ingestion; or a metabolic derangement related to the underlying cause of dyspnea (for example, hypercalcemia or hyponatremia associated with lung cancer).

Twelve hours prior to presentation, he started to have visual hallucinations. He denied fever, chills, cough, chest discomfort, palpitations, weight gain, headache, neck pain, or weakness.

Visual hallucinations could result from a toxic‐metabolic encephalopathy, such as drug overdose or withdrawal, liver or kidney failure, or hypoxia. A structural brain abnormality may also manifest with visual hallucination. Acute onset at age 51 and the absence of auditory hallucinations argue against a neurodegenerative illness and a primary psychiatric disturbance, respectively.

Episodic hallucinations would support the possibility of seizures, monocular hallucinations would point to a retinal or ocular problem, and a description of yellow‐green hue would suggest a side effect of digoxin.

His past medical history was remarkable for diet‐controlled type 2 diabetes mellitus, hypertension, hyperlipidemia, and chronic low back pain. His medications included metoprolol tartrate 25 mg twice daily, omeprazole 40 mg daily, baclofen 15 mg twice daily, oxycodone 30 mg 3 times daily, and hydrocodone 10 mg/acetaminophen 325 mg, 2 tablets 3 times daily as needed for back pain. He was a smoker with a 30 pack‐year history. He had a history of alcohol and cocaine use, but denied any recent substance use. He had no known history of obstructive pulmonary disease.

The patient takes 3 medications well known to cause confusion and hallucinations (oxycodone, hydrocodone, and baclofen), especially when they accumulate due to excessive ingestion or impaired clearance. Although these medications may suppress ventilatory drive, dyspnea would not be a common presenting complaint. He has risk factors for ischemic heart disease and cardiomyopathy, and his smoking history raises the possibility of malignancy.

On exam, the patient's temperature was 94.4C, heart rate 128 beats per minute, respiration rate 28 breaths per minute, blood pressure 155/63 mm Hg, and oxygen saturation 100% while breathing ambient air. The patient was cachectic and appeared in moderate respiratory distress. His pupils were equal and reactive to light, and extraocular movements were intact. He did not have scleral icterus, or cervical or clavicular lymphadenopathy. His oropharynx was negative for erythema, edema, or exudate. His cardiovascular exam revealed a regular tachycardia without rubs or diastolic gallops. There was a 2/6 systolic murmur heard best at left sternal border, without radiation. He did not have jugular venous distention. His pulmonary exam was notable for tachypnea but with normal vesicular breath sounds throughout. He did not have stridor, wheezing, rhonchi, or rales. His abdomen had normal bowel tones and was soft without tenderness, distention, or organomegaly. His extremities were warm, revealed normal pulses, and no edema was present. His joints were cool to palpation, without effusion. On neurologic exam, he was oriented to person and place and able to answer yes/no questions, but unable to provide detailed history. His speech was fluent. His motor exam was without focal deficits. His skin was without any notable lesions.

The constellation of findings does not point to a specific toxidrome. The finding of warm extremities in a hypothermic patient suggests heat loss due to inappropriate peripheral vasodilation. In the absence of vasodilators or features of aortic insufficiency, sepsis becomes a leading consideration. Infection could result in hypothermia and altered sensorium, and accompanying lactic acidosis could trigger tachypnea.

Shortly after admission, he became more somnolent and developed progressive respiratory distress, requiring intubation. Arterial blood gas revealed a pH of 6.93, PaCO₂<20 mm Hg, PaO₂ 127 mm Hg, and HCO₃<5 mEq/L. Other laboratory results included a lactate of 4.1 mmol/L, blood urea nitrogen 49 mg/dL, creatinine 2.3 mg/dL (0.8 at 1 month prior), sodium level of 143 mmol/L, chloride of 106 mmol/L, and bicarbonate level of <5 mg/dL. His aspartate aminotransferase was 34 IU/L, alanine transaminase was 28 IU/L, total bilirubin was 0.6 mg/dL, International Normalized Ratio was 1.3. A complete blood count revealed a white blood cell count of 23,000/L, hemoglobin of 10.6 g/dL, and platelet count of 454,000/L. A urinalysis was unremarkable. Cultures of blood, urine, and sputum were collected. Head computed tomography was negative.

This patient has a combined anion gap and nongap metabolic acidosis, as well as respiratory alkalosis. Although his acute kidney failure could produce these 2 types of metabolic acidosis, the modest elevation of the serum creatinine is not commensurate with such profound acidosis. Similarly, sepsis without hypotension or more striking elevation in lactate levels would not account for the entirety of the acidosis. Severe diabetic ketoacidosis can result in profound metabolic acidosis, and marked hyperglycemia or hyperosmolarity could result in somnolence; however, his diabetes has been controlled without medication and there is no obvious precipitant for an episode of ketoacidosis.

Remaining causes of anion gap acidosis include ingestion of methanol, ethylene glycol, ethanol, or salicylates. A careful history of ingestions and medications from witnesses including any prehospital personnel might suggest a source of intoxication. Absent this information, the hypothermia favors an ingestion of an alcohol over salicylates, and the lack of urine crystals and the presence of prominent visual hallucinations would point more toward methanol poisoning than ethylene glycol. A serum osmolarity measurement would allow determination of the osmolar gap, which would be elevated in the setting of methanol or ethylene glycol poisoning. If he were this ill from ethanol, I would have expected to see evidence of hepatotoxicity.

I would administer sodium bicarbonate to reverse the acidosis and to promote renal clearance of salicylates, methanol, ethylene glycol, and their metabolites. Orogastric decontamination with activated charcoal should be considered. If the osmolar gap is elevated, I would also administer intravenous fomepizole to attempt to reverse methanol or ethylene glycol poisoning. I would not delay treatment while waiting for these serum levels to return.

Initial serologic toxicology performed in the emergency department revealed negative ethanol, salicylates, and ketones. His osmolar gap was 13 mOsm/kg. His acetaminophen level was 69 g/mL (normal <120 g/mL). A creatinine phosphokinase was 84 IU/L and myoglobin was 93 ng/mL. His subsequent serum toxicology screen was negative for methanol, ethylene glycol, isopropranol, and hippuric acid. Urine toxicology was positive for opiates, but negative for amphetamine, benzodiazepine, cannabinoid, and cocaine.

Serum and urine ketone assays typically involve the nitroprusside reaction and detect acetoacetate, but not ‐hydroxybutyrate, and can lead to negative test results early in diabetic or alcoholic ketoacidosis. However, the normal ethanol level argues against alcoholic ketoacidosis. Rare causes of elevated anion gap acidosis include toluene toxicity, acetaminophen poisoning, and ingestion of other alcohols. Toluene is metabolized to hippuric acid, and acetaminophen toxicity and associated glutathione depletion can lead to 5‐oxoproline accumulation, producing an anion gap. Patients who abuse alcohol are at risk for acetaminophen toxicity even at doses considered normal. However, this degree of encephalopathy would be unusual for acetaminophen toxicity unless liver failure had developed or unless there was another ingestion that might alter sensorium. Furthermore, the elevated osmolar gap is not a feature of acetaminophen poisoning. I would monitor liver enzyme tests and consider a serum ammonia level, but would not attribute the entire picture to acetaminophen.

The combination of elevated anion gap with an elevated osmolar gap narrows the diagnostic possibilities. Ingestion of several alcohols (ethanol, methanol, ethylene glycol, diethylene glycol) or toluene could produce these abnormalities. Of note, the osmolar gap is typically most markedly elevated early in methanol and ethylene glycol ingestions, and then as the parent compound is metabolized, the osmolar gap closes and the accumulation of metabolites produces the anion gap. Hallucinations are more common with methanol and toluene, and renal failure is more typical of ethylene glycol or toluene. The lack of oxalate crystalluria does not exclude ethylene glycol poisoning. Unfortunately, urine testing for oxalate crystals or fluorescein examination are neither sensitive nor specific enough to diagnosis ethylene glycol toxicity reliably. In most hospitals, assays used for serum testing for alcohols are insensitive, and require confirmation with gas chromatography performed at a specialty lab.

Additional history might reveal the likely culprit or culprits. Inhalant abuse including huffing would point to toluene or organic acid exposure. Solvent ingestion (eg, antifreeze, brake fluid) would suggest methanol or ethylene glycol. Absent this history, I remain suspicious for poisoning with methanol or ethylene glycol and would consider empiric treatment after urgent consultation with a medical toxicologist. A careful ophthalmologic exam might demonstrate characteristic features of methanol poisoning. Serum samples should be sent to a regional lab for analysis for alcohols and organic acids.

He was admitted to the intensive care unit, and empiric antibiotics started. He was empirically started on N‐acetylcysteine and sodium bicarbonate drips. However, his acidemia persisted and he required hemodialysis, which was initiated 12 hours after initial presentation. His acidemia and mental status quickly improved after hemodialysis. He was extubated on hospital day 2 and no longer required hemodialysis.

The differential diagnosis at this point consists of 3 main possibilities: ingestion of methanol, ethylene glycol, or inhalant abuse such as from toluene. The normal hippuric acid level points away from toluene, whereas serum levels can be misleading in the alcohol poisonings. Other discriminating features to consider include exposure history and unique clinical aspects. In this patient, an exposure history is lacking, but 4 clinical features stand out: visual hallucinations, acute kidney injury, mild lactic acidosis, and rapid improvement with hemodialysis. Both ethylene glycol and methanol toxicity may produce a mild lactic acidosis by increasing hepatic metabolism of pyruvate to lactate, and both are rapidly cleared by dialysis. Although it is tempting to place methanol at the top of the list of possibilities due to the report of visual hallucinations, the subjective visual complaints without objective exam corollaries (loss of visual acuity, abnormal pupillary reflexes, or optic disc hyperemia) are nonspecific and might be provoked by alcohol or an inhalant. Furthermore, the acute renal failure is much more typical of ethylene glycol, and thus I would consider ethylene glycol as being the more likely of the ingestions. Coingestion of multiple alcohols is a possibility, but it would be statistically less likely. Confirmation of ethylene glycol poisoning would consist of further insight into his exposures and measurement of levels using gas chromatography.

A urine sample from his emergency department presentation was sent to an outside lab for organic acid levels. Based on high clinical suspicion for 5‐oxoprolinemia (pyroglutamic acidemia) the patient was counseled to avoid any acetaminophen. His primary care provider was informed of this and acetaminophen was added as an adverse drug reaction. The patient left against medical advice soon after extubation. Following discharge, his 5‐oxoproline (pyroglutamic acid) level returned markedly elevated at greater than 10,000 mmol/mol creatinine (200 times the upper limit of normal).

Elevations in 5‐oxoproline levels in this patient most likely stem from glutathione depletion related to chronic acetaminophen use. Alcohol use and malnutrition may have heightened this patient's susceptibility. Despite the common occurrence of acetaminophen use in alcohol abusers or the malnourished, the rarity of severe 5‐oxoproline toxicity suggests unknown factors may be present in predisposed individuals, or under‐recognition. Although acetaminophen‐induced hepatotoxicity may occur along with 5‐oxoprolinemia, this does not always occur.

Several features led me away from this syndrome. First, its rarity lowered my pretest probability. Second, the lack of exposure history and details about the serum assays, specifically whether the measurements were confirmed by gas chromatography, reduced my confidence in eliminating more common ingestions. Third, several aspects proved to be less useful discriminating features: the mild elevation in osmolar gap, renal failure, and hallucinations, which in retrospect proved to be nonspecific.

The patient admitted that he had a longstanding use of acetaminophen in addition to using his girlfriend's acetaminophen‐hydrocodone. He had significant weight loss of over 50 pounds over the previous year, which he attributed to poor appetite. On further chart review, he had been admitted 3 times with a similar clinical presentation and recovered quickly with intensive and supportive care, with no etiology found at those times. He had 2 subsequent hospital admissions for altered mental status and respiratory failure, and his final hospitalization resulted in cardiac arrest and death.

DISCUSSION

5‐Oxoprolinemia is a rare, but potentially lethal cause of severe anion gap metabolic acidosis.[1, 2] The mechanism is thought to be impairment of glutathione metabolism, in the context of other predisposing factors. This can be a congenital error of metabolism, or can be acquired and exacerbated by acetaminophen use. Ingestion of acetaminophen leads to glutathione depletion, which in turn may precipitate accumulation of pyroglutamic acid and subsequent anion gap metabolic acidosis (Figure 1). Additional risk factors that may predispose patients to this condition include malnutrition, renal insufficiency, concurrent infection, and female gender.[1, 2, 3]

Figure 1

The diagnosis of 5‐oxoprolinemia is made via urine or serum organic acid analysis, testing routinely performed in pediatric populations when screening for congenital metabolic disorders. The pathophysiology suggests that obtaining a urine sample early in presentation, when acidosis is greatest, would lead to the highest 5‐oxoproline levels and best chance for diagnosis. Case patients have had normal levels prior to and in convalescent phases after the acute episode.[4] Given the long turnaround time for lab testing, presumptive diagnosis and treatment may be necessary.

Treatment of 5‐oxoprolinemia is primarily supportive, aimed at the metabolic acidosis. Fluid resuscitation and bicarbonate therapy are reasonable temporizing measures. Hemodialysis can clear 5‐oxoproline and may be indicated in severe acidosis.[5] Furthermore, the proposed pathophysiology suggests that administration of N‐acetylcysteine (NAC) may help to address the underlying process, but there are no trials to support a specific dosing regimen. However, given the fulminant presentation and common competing concern for acetaminophen toxicity, it is reasonable to initiate NAC aimed at treatment for possible acetaminophen overdose. Prevention of recurrence includes avoidance of acetaminophen, and counseling the patient to avoid acetaminophen in prescription combination medications and over‐the‐counter preparations.

Recent regulatory changes regarding acetaminophen/opioid combinations may reduce the incidence of 5‐oxoprolinemia. The US Food and Drug Administration has taken action to reduce adverse effects from acetaminophen exposure by limiting the amount of acetaminophen in opioid combination pills from 500 mg to a maximum of 325 mg per pill. This is aimed at preventing hepatotoxicity from ingestion of higher‐than‐recommended doses. However, clinicians should remember that 5‐oxoprolinemia can result from ingestion of acetaminophen at therapeutic levels.

Given its rare incidence, low clinical suspicion, and transient nature of confirmatory testing, it is likely this remains an underdiagnosed syndrome. In the case discussed, subsequent chart review demonstrated 5 previous admissions in multiple hospitals for severe transient anion gap acidosis. The likelihood that 5‐oxoprolinemia was missed in each of these cases supports a lack of awareness of this syndrome. In this patient, the discussant appropriately identified the possibility of 5‐oxoproline toxicity, but felt ethylene glycol ingestion was more likely. As this case underscores, a cornerstone in the management of suspected ingestions is empiric treatment for the most likely etiologies. Here, treatment for acetaminophen overdose and for methanol or ethylene glycol were warranted, and fortunately also addressed the rarer possibility of 5‐oxoproline toxicity.

The mnemonic MUDPILES is commonly used to identify possible causes of life‐threatening anion gap metabolic acidosis, as such heuristics have benefits in rapidly generating a differential diagnosis to guide initial evaluation. Given the fact that the traditional letter P (paraldehyde) in MUDPILES is no longer clinically utilized, some authors have suggested replacing this with pyroglutamic acid (a synonym of 5‐oxoproline). Such a change may help providers who have ruled out other causes of a high anion gap metabolic acidosis, facilitating diagnosis of this life‐threatening syndrome. In any case, clinicians must be mindful that simple memory aids may mislead clinicians, and a complete differential diagnosis may require more than a mnemonic.

TEACHING POINTS

1. Acetaminophen use, even at therapeutic levels, can lead to 5‐oxoprolinemia, a potentially lethal anion gap metabolic acidosis.
2. 5‐oxoprolinemia is likely related to glutathione depletion, worsened by acetaminophen, malnutrition, renal insufficiency, female gender, and infection. This implies theoretical benefit from administration of NAC for glutathione repletion.
3. Mnemonics can be useful, but have limitations by way of oversimplification. This case suggests that changing the letter P in MUDPILES from paraldehyde to pyroglutamic acid could reduce underdiagnosis.

Disclosure: Nothing to report.

Hospital medicine (HM), which is the fastest growing medical specialty in the United States, includes more than 40,000 healthcare providers.[1] Hospitalists include practitioners from a variety of medical specialties, including internal medicine and pediatrics, and professional backgrounds such as physicians, nurse practitioners. and physician assistants.[2, 3] Originally defined as specialists of inpatient medicine, hospitalists must diagnose and manage a wide variety of clinical conditions, coordinate transitions of care, provide perioperative management to surgical patients, and contribute to quality improvement and hospital administration.[4, 5]

With the evolution of the HM, the need for effective continuing medical education (CME) has become increasingly important. Courses make up the largest percentage of CME activity types,[6] which also include regularly scheduled lecture series, internet materials, and journal‐related CME. Successful CME courses require educational content that matches the learning needs of its participants.[7] In 2006, the Society for Hospital Medicine (SHM) developed core competencies in HM to guide educators in identifying professional practice gaps for useful CME.[8] However, knowing a population's characteristics and learning needs is a key first step to recognizing a practice gap.[9] Understanding these components is important to ensuring that competencies in the field of HM remain relevant to address existing practice gaps.[10] Currently, little is known about the demographic characteristics of participants in HM CME.

Research on the characteristics of effective clinical teachers in medicine has revealed the importance of establishing a positive learning climate, asking questions, diagnosing learners needs, giving feedback, utilizing established teaching frameworks, and developing a personalized philosophy of teaching.[11] Within CME, research has generally demonstrated that courses lead to improvements in lower level outcomes,[12] such as satisfaction and learning, yet higher level outcomes such as behavior change and impacts on patients are inconsistent.[13, 14, 15] Additionally, we have shown that participant reflection on CME is enhanced by presenters who have prior teaching experience and higher teaching effectiveness scores, by the use of audience participation and by incorporating relevant content.[16, 17] Despite the existence of research on CME in general, we are not aware of prior studies regarding characteristics of effective CME in the field of HM.

To better understand and improve the quality of HM CME, we sought to describe the characteristics of participants at a large, national HM CME course, and to identify associations between characteristics of presentations and CME teaching effectiveness (CMETE) scores using a previously validated instrument.

METHODS

RESULTS

There were 32 presentations during the MCP conference in 2014. A total of 277 (75.2%) out of 368 participants completed the survey. This yielded 7947 CMETE evaluations for analysis, with an average of 28.7 per person (median: 31, interquartile range: 2732, range: 632).

Demographic characteristics of course participants are listed in Table 1. Participants (number, %), described themselves as hospitalists (181, 70.4%), ABIM certified with HM focus (48, 18.8%), physicians with MD or MBBS degrees (181, 70.4%), nurse practitioners or physician assistants (52; 20.2%), and in practice 20 years (73, 28%). The majority of participants (148, 58.3%) worked in private practice, whereas only 63 (24.8%) worked in academic settings.

CMETE scores (mean [SD]) were significantly associated with the use of ARS (4.64 [0.16]) vs no ARS (4.49 [0.16]; P=0.01, Table 2, Figure 1), longer presentations (30 minutes: 4.67 [0.13] vs <30 minutes: 4.51 [0.18]; P=0.02), and larger number of slides (50: 4.66 [0.17] vs <50: 4.55 [0.17]; P=0.04). There were no significant associations between CMETE scores and use of clinical cases, defined goals, or summary slides.

Figure 1

The magnitude of score differences observed in this study are substantial when considered in terms of Cohen's effect sizes. For number of slides, the effect size is 0.65, for audience response the effect size is 0.94, and for presentation length the effect size is approximately 1. According to Cohen, effect sizes of 0.5 to 0.8 are moderate, and effect sizes >0.8 are large. Consequently, the effect sizes of our observed differences are moderate to large.[20, 21]

DISCUSSION

To our knowledge, this is the first study to measure associations between validated teaching effectiveness scores and characteristics of presentations in HM CME. We found that the use of ARS and longer presentations were associated with significantly higher CMETE scores. Our findings have implications for HM CME course directors and presenters as they attempt to develop methods to improve the quality of CME.

CME participants in our study crossed a wide range of ages and experience, which is consistent with national surveys of hospitalists.[22, 23] Interestingly, however, nearly 1 in 3 participants trained in a specialty other than internal medicine. Additionally, the professional degrees of participants were diverse, with 20% of participants having trained as nurse practitioners or physician assistants. These findings are at odds with an early national survey of inpatient practitioners,[22] but consistent with recent literature that the diversity of training backgrounds among hospitalists is increasing as the field of HM evolves.[24] Hospital medicine CME providers will need to be cognizant of these demographic changes as they work to identify practice gaps and apply appropriate educational methods.

The use of an ARS allows for increased participation and engagement among lecture attendees, which in turn promotes active learning.[25, 26, 27] The association of higher teaching scores with the use of ARS is consistent with previous research in other CME settings such as clinical round tables and medical grand rounds.[17, 28] As it pertains to HM specifically, our findings also build upon a recent study by Sehgal et al., which reported on the novel use of bedside CME to enhance interactive learning and discussion among hospitalists, and which was viewed favorably by course participants.[29]

The reasons why longer presentations in our study were linked to higher CMETE scores are not entirely clear, as previous CME research has failed to demonstrate this relationship.[18] One possibility is that course participants prefer learning from presentations that provide granular, content‐rich information. Another possibility may be that characteristics of effective presenters who gave longer presentations and that were not measured in this study, such as presenter experience and expertise, were responsible for the observed increase in CMETE scores. Yet another possibility is that effective presentations were longer due to the use of ARS, which was also associated with better CMETE scores. This explanation may be plausible because the ARS requires additional slides and provides opportunities for audience interaction, which may lengthen the duration of any given presentation.

This study has several limitations. This was a single CME conference sponsored by a large academic medical center, which may limit generalizability, especially to smaller conferences in community settings. However, the audience was large and diverse in terms of participants experiences, practice settings, professional backgrounds, and geographic locations. Furthermore, the demographic characteristics of hospitalists at our course appear very similar to a recently reported national cross‐section of hospitalist groups.[30] Second, this is a cross‐sectional study without a comparison group. Nonetheless, a systematic review showed that most published education research studies involved single‐group designs without comparison groups.[31] Last, the outcomes of the study include attitudes and objectively measured presenter behaviors such as the use of ARS, but not patient‐related outcomes. Nonetheless, evidence indicates that the majority of medical education research does not present outcomes beyond knowledge,[31] and it has been noted that behavior‐related outcomes strike the ideal balance between feasibility and rigor.[32, 33] Finally, the instrument used in this study to measure teaching effectiveness is supported by prior validity evidence.[16]

In summary, we found that hospital medicine CME presentations, which are longer and use audience responses, are associated with greater teaching effectiveness ratings by CME course participants. These findings build upon previous CME research and suggest that CME course directors and presenters should strive to incorporate opportunities that promote audience engagement and participation. Additionally, this study adds to the existing validity of evidence for the CMETE assessment tool. We believe that future research should explore potential associations between teacher effectiveness and patient‐related outcomes, and determine whether course content that reflects the SHM core competencies improves CME teaching effectiveness scores.

Hospital medicine (HM), which is the fastest growing medical specialty in the United States, includes more than 40,000 healthcare providers.[1] Hospitalists include practitioners from a variety of medical specialties, including internal medicine and pediatrics, and professional backgrounds such as physicians, nurse practitioners. and physician assistants.[2, 3] Originally defined as specialists of inpatient medicine, hospitalists must diagnose and manage a wide variety of clinical conditions, coordinate transitions of care, provide perioperative management to surgical patients, and contribute to quality improvement and hospital administration.[4, 5]

With the evolution of the HM, the need for effective continuing medical education (CME) has become increasingly important. Courses make up the largest percentage of CME activity types,[6] which also include regularly scheduled lecture series, internet materials, and journal‐related CME. Successful CME courses require educational content that matches the learning needs of its participants.[7] In 2006, the Society for Hospital Medicine (SHM) developed core competencies in HM to guide educators in identifying professional practice gaps for useful CME.[8] However, knowing a population's characteristics and learning needs is a key first step to recognizing a practice gap.[9] Understanding these components is important to ensuring that competencies in the field of HM remain relevant to address existing practice gaps.[10] Currently, little is known about the demographic characteristics of participants in HM CME.

Research on the characteristics of effective clinical teachers in medicine has revealed the importance of establishing a positive learning climate, asking questions, diagnosing learners needs, giving feedback, utilizing established teaching frameworks, and developing a personalized philosophy of teaching.[11] Within CME, research has generally demonstrated that courses lead to improvements in lower level outcomes,[12] such as satisfaction and learning, yet higher level outcomes such as behavior change and impacts on patients are inconsistent.[13, 14, 15] Additionally, we have shown that participant reflection on CME is enhanced by presenters who have prior teaching experience and higher teaching effectiveness scores, by the use of audience participation and by incorporating relevant content.[16, 17] Despite the existence of research on CME in general, we are not aware of prior studies regarding characteristics of effective CME in the field of HM.

To better understand and improve the quality of HM CME, we sought to describe the characteristics of participants at a large, national HM CME course, and to identify associations between characteristics of presentations and CME teaching effectiveness (CMETE) scores using a previously validated instrument.

METHODS

RESULTS

There were 32 presentations during the MCP conference in 2014. A total of 277 (75.2%) out of 368 participants completed the survey. This yielded 7947 CMETE evaluations for analysis, with an average of 28.7 per person (median: 31, interquartile range: 2732, range: 632).

Demographic characteristics of course participants are listed in Table 1. Participants (number, %), described themselves as hospitalists (181, 70.4%), ABIM certified with HM focus (48, 18.8%), physicians with MD or MBBS degrees (181, 70.4%), nurse practitioners or physician assistants (52; 20.2%), and in practice 20 years (73, 28%). The majority of participants (148, 58.3%) worked in private practice, whereas only 63 (24.8%) worked in academic settings.

CMETE scores (mean [SD]) were significantly associated with the use of ARS (4.64 [0.16]) vs no ARS (4.49 [0.16]; P=0.01, Table 2, Figure 1), longer presentations (30 minutes: 4.67 [0.13] vs <30 minutes: 4.51 [0.18]; P=0.02), and larger number of slides (50: 4.66 [0.17] vs <50: 4.55 [0.17]; P=0.04). There were no significant associations between CMETE scores and use of clinical cases, defined goals, or summary slides.

Figure 1

The magnitude of score differences observed in this study are substantial when considered in terms of Cohen's effect sizes. For number of slides, the effect size is 0.65, for audience response the effect size is 0.94, and for presentation length the effect size is approximately 1. According to Cohen, effect sizes of 0.5 to 0.8 are moderate, and effect sizes >0.8 are large. Consequently, the effect sizes of our observed differences are moderate to large.[20, 21]

DISCUSSION

To our knowledge, this is the first study to measure associations between validated teaching effectiveness scores and characteristics of presentations in HM CME. We found that the use of ARS and longer presentations were associated with significantly higher CMETE scores. Our findings have implications for HM CME course directors and presenters as they attempt to develop methods to improve the quality of CME.

CME participants in our study crossed a wide range of ages and experience, which is consistent with national surveys of hospitalists.[22, 23] Interestingly, however, nearly 1 in 3 participants trained in a specialty other than internal medicine. Additionally, the professional degrees of participants were diverse, with 20% of participants having trained as nurse practitioners or physician assistants. These findings are at odds with an early national survey of inpatient practitioners,[22] but consistent with recent literature that the diversity of training backgrounds among hospitalists is increasing as the field of HM evolves.[24] Hospital medicine CME providers will need to be cognizant of these demographic changes as they work to identify practice gaps and apply appropriate educational methods.

The use of an ARS allows for increased participation and engagement among lecture attendees, which in turn promotes active learning.[25, 26, 27] The association of higher teaching scores with the use of ARS is consistent with previous research in other CME settings such as clinical round tables and medical grand rounds.[17, 28] As it pertains to HM specifically, our findings also build upon a recent study by Sehgal et al., which reported on the novel use of bedside CME to enhance interactive learning and discussion among hospitalists, and which was viewed favorably by course participants.[29]

The reasons why longer presentations in our study were linked to higher CMETE scores are not entirely clear, as previous CME research has failed to demonstrate this relationship.[18] One possibility is that course participants prefer learning from presentations that provide granular, content‐rich information. Another possibility may be that characteristics of effective presenters who gave longer presentations and that were not measured in this study, such as presenter experience and expertise, were responsible for the observed increase in CMETE scores. Yet another possibility is that effective presentations were longer due to the use of ARS, which was also associated with better CMETE scores. This explanation may be plausible because the ARS requires additional slides and provides opportunities for audience interaction, which may lengthen the duration of any given presentation.

This study has several limitations. This was a single CME conference sponsored by a large academic medical center, which may limit generalizability, especially to smaller conferences in community settings. However, the audience was large and diverse in terms of participants experiences, practice settings, professional backgrounds, and geographic locations. Furthermore, the demographic characteristics of hospitalists at our course appear very similar to a recently reported national cross‐section of hospitalist groups.[30] Second, this is a cross‐sectional study without a comparison group. Nonetheless, a systematic review showed that most published education research studies involved single‐group designs without comparison groups.[31] Last, the outcomes of the study include attitudes and objectively measured presenter behaviors such as the use of ARS, but not patient‐related outcomes. Nonetheless, evidence indicates that the majority of medical education research does not present outcomes beyond knowledge,[31] and it has been noted that behavior‐related outcomes strike the ideal balance between feasibility and rigor.[32, 33] Finally, the instrument used in this study to measure teaching effectiveness is supported by prior validity evidence.[16]

In summary, we found that hospital medicine CME presentations, which are longer and use audience responses, are associated with greater teaching effectiveness ratings by CME course participants. These findings build upon previous CME research and suggest that CME course directors and presenters should strive to incorporate opportunities that promote audience engagement and participation. Additionally, this study adds to the existing validity of evidence for the CMETE assessment tool. We believe that future research should explore potential associations between teacher effectiveness and patient‐related outcomes, and determine whether course content that reflects the SHM core competencies improves CME teaching effectiveness scores.

Hospital medicine (HM), which is the fastest growing medical specialty in the United States, includes more than 40,000 healthcare providers.[1] Hospitalists include practitioners from a variety of medical specialties, including internal medicine and pediatrics, and professional backgrounds such as physicians, nurse practitioners. and physician assistants.[2, 3] Originally defined as specialists of inpatient medicine, hospitalists must diagnose and manage a wide variety of clinical conditions, coordinate transitions of care, provide perioperative management to surgical patients, and contribute to quality improvement and hospital administration.[4, 5]

With the evolution of the HM, the need for effective continuing medical education (CME) has become increasingly important. Courses make up the largest percentage of CME activity types,[6] which also include regularly scheduled lecture series, internet materials, and journal‐related CME. Successful CME courses require educational content that matches the learning needs of its participants.[7] In 2006, the Society for Hospital Medicine (SHM) developed core competencies in HM to guide educators in identifying professional practice gaps for useful CME.[8] However, knowing a population's characteristics and learning needs is a key first step to recognizing a practice gap.[9] Understanding these components is important to ensuring that competencies in the field of HM remain relevant to address existing practice gaps.[10] Currently, little is known about the demographic characteristics of participants in HM CME.

Research on the characteristics of effective clinical teachers in medicine has revealed the importance of establishing a positive learning climate, asking questions, diagnosing learners needs, giving feedback, utilizing established teaching frameworks, and developing a personalized philosophy of teaching.[11] Within CME, research has generally demonstrated that courses lead to improvements in lower level outcomes,[12] such as satisfaction and learning, yet higher level outcomes such as behavior change and impacts on patients are inconsistent.[13, 14, 15] Additionally, we have shown that participant reflection on CME is enhanced by presenters who have prior teaching experience and higher teaching effectiveness scores, by the use of audience participation and by incorporating relevant content.[16, 17] Despite the existence of research on CME in general, we are not aware of prior studies regarding characteristics of effective CME in the field of HM.

To better understand and improve the quality of HM CME, we sought to describe the characteristics of participants at a large, national HM CME course, and to identify associations between characteristics of presentations and CME teaching effectiveness (CMETE) scores using a previously validated instrument.

METHODS

RESULTS

There were 32 presentations during the MCP conference in 2014. A total of 277 (75.2%) out of 368 participants completed the survey. This yielded 7947 CMETE evaluations for analysis, with an average of 28.7 per person (median: 31, interquartile range: 2732, range: 632).

Demographic characteristics of course participants are listed in Table 1. Participants (number, %), described themselves as hospitalists (181, 70.4%), ABIM certified with HM focus (48, 18.8%), physicians with MD or MBBS degrees (181, 70.4%), nurse practitioners or physician assistants (52; 20.2%), and in practice 20 years (73, 28%). The majority of participants (148, 58.3%) worked in private practice, whereas only 63 (24.8%) worked in academic settings.

CMETE scores (mean [SD]) were significantly associated with the use of ARS (4.64 [0.16]) vs no ARS (4.49 [0.16]; P=0.01, Table 2, Figure 1), longer presentations (30 minutes: 4.67 [0.13] vs <30 minutes: 4.51 [0.18]; P=0.02), and larger number of slides (50: 4.66 [0.17] vs <50: 4.55 [0.17]; P=0.04). There were no significant associations between CMETE scores and use of clinical cases, defined goals, or summary slides.

Figure 1

The magnitude of score differences observed in this study are substantial when considered in terms of Cohen's effect sizes. For number of slides, the effect size is 0.65, for audience response the effect size is 0.94, and for presentation length the effect size is approximately 1. According to Cohen, effect sizes of 0.5 to 0.8 are moderate, and effect sizes >0.8 are large. Consequently, the effect sizes of our observed differences are moderate to large.[20, 21]

DISCUSSION

To our knowledge, this is the first study to measure associations between validated teaching effectiveness scores and characteristics of presentations in HM CME. We found that the use of ARS and longer presentations were associated with significantly higher CMETE scores. Our findings have implications for HM CME course directors and presenters as they attempt to develop methods to improve the quality of CME.

CME participants in our study crossed a wide range of ages and experience, which is consistent with national surveys of hospitalists.[22, 23] Interestingly, however, nearly 1 in 3 participants trained in a specialty other than internal medicine. Additionally, the professional degrees of participants were diverse, with 20% of participants having trained as nurse practitioners or physician assistants. These findings are at odds with an early national survey of inpatient practitioners,[22] but consistent with recent literature that the diversity of training backgrounds among hospitalists is increasing as the field of HM evolves.[24] Hospital medicine CME providers will need to be cognizant of these demographic changes as they work to identify practice gaps and apply appropriate educational methods.

The use of an ARS allows for increased participation and engagement among lecture attendees, which in turn promotes active learning.[25, 26, 27] The association of higher teaching scores with the use of ARS is consistent with previous research in other CME settings such as clinical round tables and medical grand rounds.[17, 28] As it pertains to HM specifically, our findings also build upon a recent study by Sehgal et al., which reported on the novel use of bedside CME to enhance interactive learning and discussion among hospitalists, and which was viewed favorably by course participants.[29]

The reasons why longer presentations in our study were linked to higher CMETE scores are not entirely clear, as previous CME research has failed to demonstrate this relationship.[18] One possibility is that course participants prefer learning from presentations that provide granular, content‐rich information. Another possibility may be that characteristics of effective presenters who gave longer presentations and that were not measured in this study, such as presenter experience and expertise, were responsible for the observed increase in CMETE scores. Yet another possibility is that effective presentations were longer due to the use of ARS, which was also associated with better CMETE scores. This explanation may be plausible because the ARS requires additional slides and provides opportunities for audience interaction, which may lengthen the duration of any given presentation.

This study has several limitations. This was a single CME conference sponsored by a large academic medical center, which may limit generalizability, especially to smaller conferences in community settings. However, the audience was large and diverse in terms of participants experiences, practice settings, professional backgrounds, and geographic locations. Furthermore, the demographic characteristics of hospitalists at our course appear very similar to a recently reported national cross‐section of hospitalist groups.[30] Second, this is a cross‐sectional study without a comparison group. Nonetheless, a systematic review showed that most published education research studies involved single‐group designs without comparison groups.[31] Last, the outcomes of the study include attitudes and objectively measured presenter behaviors such as the use of ARS, but not patient‐related outcomes. Nonetheless, evidence indicates that the majority of medical education research does not present outcomes beyond knowledge,[31] and it has been noted that behavior‐related outcomes strike the ideal balance between feasibility and rigor.[32, 33] Finally, the instrument used in this study to measure teaching effectiveness is supported by prior validity evidence.[16]

In summary, we found that hospital medicine CME presentations, which are longer and use audience responses, are associated with greater teaching effectiveness ratings by CME course participants. These findings build upon previous CME research and suggest that CME course directors and presenters should strive to incorporate opportunities that promote audience engagement and participation. Additionally, this study adds to the existing validity of evidence for the CMETE assessment tool. We believe that future research should explore potential associations between teacher effectiveness and patient‐related outcomes, and determine whether course content that reflects the SHM core competencies improves CME teaching effectiveness scores.