Published in the February 2011 issue of Today’s Hospitalist
ASK ROBERT L. TROWBRIDGE, MD, a patient safety expert and hospitalist at Maine Medical Center in Portland, Maine, about diagnostic errors, and he’ll tell you that they’re "the sleeping dog of the patient safety movement." While these errors are common, Dr. Trowbridge explains, they’re often left unchallenged because "nobody is quite sure what will happen when we start poking" at the problem.
"As clinicians, we all recognize that we’ve made a lot of diagnostic errors in our careers," he says, "but we wall these off in our own minds rather than discuss them out in the open."
To face this unknown, Dr. Trowbridge and his colleagues have launched an anonymous, self-reporting system when diagnostic errors occur. The goal is to delve into the errors’ root causes and figure out how to prevent them from recurring.
According to the literature, more than 8% of inpatients are subject to a major diagnostic error. More than one-quarter of all reported errors involve some diagnostic error, and as many as 11% of reported errors result in patient death.
Despite those data, Dr. Trowbridge says, diagnostic errors have been ignored until recently. In part, that’s because researchers have paid more attention to clear causes of patient harm like medication errors and hand hygiene. He talked to Today’s Hospitalist about his center’s new reporting system, which the adult inpatient medicine service has piloted since last fall.
How does the reporting system work?
We wanted to make it easy to report an error in real time by the people best suited to recognize such errors: physicians.
Our clinical workstations all have a "Diagnostic Error Reporting" icon. Clicking on it takes you to a database that asks for four pieces of information: the patient’s medical record number; the type of error, whether it was a delayed, wrong or missed diagnosis; a brief description of the error, such as "iron deficiency anemia for six months; missed colon cancer"; and any patient harm that resulted. This can be done anonymously and very quickly at the point of care. That makes it more likely that a physician will report an error, either his or her own or a colleague’s.
The big thing with diagnostic errors is that they are hard to recognize, and physicians are not very good at reporting errors. Right now, all these diagnostic errors are sitting in individual clinicians’ minds and are not being made available to improve individual and institutional performance.
How many reports do you receive?
We’ve been getting about two a week, and we’re starting to review them to sort out the causes retrospectively. Our physician reviewers were nominated by medical group leadership and trained to identify the causes of diagnostic errors. We go through the chart and ask, for example: What was the context of care? An example might be that the patient was in the ICU with 17 consultants, and the ultimate error cause was miscommunication among them.
We also look to see what diagnoses are frequently made that are incorrect. If the patient had pneumonia but we labeled him as having congestive heart failure, then we record both pneumonia as the missed diagnosis and congestive heart failure as the wrong diagnosis.
We look for institutional trends and plan to report findings back to the divisions and the institution and say, for example, that we are missing pneumonia a lot or that we are wrongly labeling people with congestive heart failure. To design effective, local interventions, we need to know what we are good at and what we aren’t good at, in terms of diagnosis.
How many reports do you need to identify trends?
You can have an action plan based on a single error. Did the lab report confuse people? Did we have to examine people in a room without any privacy? At least half the time, a systems issue contributes to the diagnostic error, and if it happened once, it’s likely to happen again.
How can doctors be sure the information won’t end up in the hands of a malpractice lawyer?
This is a performance improvement project, which is not meant for risk management or to mitigate liability for the institution. After we analyze the cases, we remove patient and physician identifiers, and we don’t keep medical record numbers. We will not have a giant database with all our mistakes in it.
From what you know so far, are these errors mainly gaps in knowledge?
Knowledge is a relatively small player in all this, and diagnostic errors are more a matter of nuance. It’s not that I didn’t know the disease existed; it’s that I didn’t put everything together to say this person has it. It is more about synthesizing information and recognizing the biases that we bring to the table.
Why are diagnostic errors so hard to discuss?
If I prescribe the wrong medication or the wrong dose, I can justify that by saying that we all make mistakes. But when it’s a critique of your thinking as a diagnostician, that cuts to the core of who you are as a physician.
We also have a hard time recognizing affective bias. We want to be dispassionate when we approach the diagnostic process, but we are not. Some patients push our buttons, and it’s the same if we are really tired post-call or we’ve just gotten into a fight with our spouse: We do something a bit differently without necessarily recognizing that we do.
How do you start to fix something like cognitive errors?
One way is provider education, letting clinicians know that diagnostic errors are common and that we need to be careful. Another is clinical decision-support at the point of care. Right now, a lot of clinical decision-support doesn’t work well, but new systems in their infancy could augment diagnosis, helping physicians recognize patterns.
A computerized aid, for example, could recognize that the combination of left ventricular hypertrophy on echocardiography and low voltage on the ECG should prompt the clinician to consider amyloidosis.
We also need to stress the value of the clinical exam because the history and physical are so important. A lot of times, diagnostic errors are due to a failure to look or a failure to ask.
We may need to slow down sometimes and perhaps take a "diagnostic time-out." Step away and ask, how do we know this? We should also be more aware of how certain patients make us feel “and while we teach students to think horses, not zebras, when they hear hoof beats, we also have to remind them to think about zebras.
I think the key is to be more cognizant of how we think as physicians and admit when we make a mistake. A huge service for residents, fellows and students is having attendings say, "I messed up and this is why. You guys are at risk of the same thing, so be careful."
Deborah Gesensway is a freelance writer based in Toronto who reports on U.S. health care.