By Andy Skean
Senior Editor

 

Getting the diagnosis right is among the most fundamental of patient expectations. However, according to a 2015 report by the Institute of Medicine (IOM), how often physicians get it wrong is not known, because there are few accepted methods for measuring the scope of the problem.

 

Answering the Call

For a study published in JAMA Internal Medicine, my colleagues and I answered the IOM’s call to develop tools to measure diagnostic error, by considering the scenario of patients presenting to the ED with life-threatening cardiovascular emergencies. Specifically, we considered ruptured abdominal aortic aneurysm, acute myocardial infarction, stroke, aortic dissection, and subarachnoid hemorrhage, and asked what proportion of ED visits attributable to these conditions (or their prodromes) end in discharge home without a diagnosis.

The tricky part is that one can’t usually know with certainty whether an emergency was present during a visit when it was not diagnosed. If a patient is discharged from the ED with a diagnosis of lumbago and comes back 2 days later with a ruptured abdominal aortic aneurysm, was the first visit a miss? Since the answer is unknowable, we took a population-based approach, using Medicare claims to identify all patients hospitalized for the five above-noted emergencies, and all their ED visits in the preceding year. Using the same patients’ actual ED visit history, we estimated how many would be expected to visit the ED for reasons unrelated to the cardiovascular emergency on any given day. We then counted the number of ED discharges observed in the days immediately preceding hospital admission, and reasoned that the excess visits—the difference between the number of ED discharges observed and the number expected—represented cases in which the acute pathology was present but unrecognized. To arrive at the false negative rate, we divided the number of excess discharges (unrecognized emergencies) by the sum of excess discharges plus index hospitalizations (recognized emergencies).

 

Cardiovascular Emergencies Infrequently Missed

We found that these acute emergencies are missed in fewer than one in 20 ED visits in which they are present. Specifically, we estimated that the false negative rate ranged from 2.3% (ruptured abdominal aortic aneurysm) to 4.5% (aortic dissection) and remained relatively stable across the 2007-2014 study period. While low overall, we also found that the probability of non-recognition depends to some extent on the patient. In a multivariable analysis, we demonstrated that patients who are younger than 65 years, female, poor (dually eligible for Medicare and Medicaid), or have chronic medical conditions were at increased risk.

Should our findings change practice?  The finding that physicians make more mistakes among some types of patients than others should certainly be cause for reflection and vigilance. But the study was not designed to answer clinical questions such as whether to test or admit more patients. Our purpose was to provide policy makers and quality measurers with an example of how insurance claims can be used to monitor how often important diagnoses are not being made when they might have been. We’ve shown that, on average, physicians have been doing a good job diagnosing these problems among Medicare patients when they come to the ED. But the story might change as the pressure to avoid hospitalization grows, so having an easy way to spot an uptick might be helpful as an early warning system that too many corners are being cut.

 

More Opportunity Outside the ED?

In a study published shortly after ours, researchers approached a similar question by interviewing patients during their acute myocardial infarction hospitalization. When asked whether they had recently sought care for similar symptoms (anywhere, not just the ED), 22% of men and 30% of women said yes. Contrasted to our results, this suggests that missed opportunities for recognizing acute coronary syndrome may be an order-of-magnitude more plentiful in primary care offices and urgent care centers than in emergency departments. Given that there are many more visits to primary care than to the ED, perhaps this should not be surprising; yet, to date, there has been little scrutiny of diagnostic accuracy in non-ED outpatient settings. Future research might address this gap by applying our method to claims for office visits.

Author