乌鸦传媒

Medical errors can vary greatly, but there are several types of typical diagnostic "pitfalls" that clinicians should especially watch out for, said Gordon Schiff, MD, associate director of the Brigham and Women's Center for Patient Safety Research and Practice in Boston.

Speaking at the annual meeting of the , he said think of a pitfall as a diagnostic error waiting to happen -- "vulnerability to errors that I think we all agree exist. And the question is, are there certain situations or patterns where these vulnerabilities are seen repeatedly, and we can better understand them by collecting them and learning from them?"

Old, Yet New Approach

It turns out that this is not a new concept, he added, displaying a discussing how what may initially be diagnosed as hemorrhoids may turn out later to be rectal cancer. "This idea about trying to collect these pitfalls and trying to understand them is both a new way of thinking as well as an old approach," he said.

To develop a list of common diagnostic pitfalls, Schiff and colleagues analyzed a variety of data, including information from 5 years of closed malpractice claims, patient safety incident reports, ambulatory morbidity and mortality reports, and findings from physician focus groups conducted in six 1-hour sessions.

Focus group participants included neurologists, gastroenterologists, dermatologists, pulmonologists, rheumatologists, and otolaryngologists. The researchers looked both at where the diagnosis went wrong as well as the contributing factors that made the diagnosis harder than usual. They ended up analyzing 836 diagnostic cases.

The Top 10 missed diagnoses, in order of frequency from highest to lowest, were:

• Colorectal cancer

• Lung cancer

• Breast cancer

• Myocardial infarction

• Prostate cancer

• Stroke

• Sepsis

• Bladder cancer

• Pulmonary embolism

• Brain hemorrhage

Errors in the Process

In terms of where things went wrong in the process, failure to order a needed test was the most frequent issue, followed by failure to consider the correct diagnosis, failure to or delay in follow-up for an abnormal test result, failure to heavily weigh a critical piece of the patient's history, and failure or delay in ordering a referral.

"Not unsurprisingly, tests are where a lot of the things went wrong in these cases -- a failure to order an immediate test, [like when] somebody needed a colonoscopy with rectal bleeding and then didn't get one," Schiff said. "Or, failure to consider the correct diagnosis. In many cases, both of those applied, so someone didn't consider colon cancer, so they didn't order a colonoscopy, or they didn't consider sepsis so they didn't order blood cultures."

Another specialty-specific example is a woman who is referred to a specialist for a breast lump. She is given a screening mammography and receives a false-negative result, and that's not followed up on because the specialist didn't know the woman had a specific complaint.

"Collecting these all together, hopefully, is helpful to get a picture of the kinds of things that go wrong," Schiff said.

Overdiagnosis is also an issue, he continued, such as with a patient diagnosed with a stroke when they actually had Bell's palsy. Another patient was said to have had a seizure disorder when they didn't, and several people were erroneously diagnosed with multiple sclerosis.

Schiff went on to discuss more pitfalls, including the following:

• Disease A is mistaken for Disease B. Diseases are often misdiagnosed or mistaken for one another, such as aortic dissection being misdiagnosed as myocardial infarction, or bipolar disorder misdiagnosed as depression

• Misinterpretation of test results. False positives or false negatives may occur, but clinicians aren't thinking of the test's limitations -- for instance, a negative COVID-19 test may occur early or late in the course of the disease

• Failure to recognize atypical presentation, signs, or symptoms, such as sepsis occurring in an elderly patient who is afebrile or hypothermic

• Failure to recognize the urgency of diagnosis -- for instance, in patients with pericardial tamponade or tension pneumothorax

• Perils of intermittent symptoms, in which someone looks fine when they see you but they're actually in a "lucid interval" of an epidural hematoma

• Failure to consider a drug or environmental cause

• Confusion about response to empiric treatment. "Somebody is on steroids or pain medicine, so their their symptoms are masked and they look better, when in fact, they have something more serious that is going to be overlooked," Schiff said

• Failure to appreciate the limitations of the physical exam

• Chronic disease is presumed to account for new symptoms

How Can Pitfalls Be Used?

What should be done with this information? "Why not have a new section for each disease in Harrison's or UpToDate" listing diagnostic pitfalls?, Schiff suggested.

Although Harrison's added a chapter on the issue of diagnostic error, the researchers have not been entirely successful with this idea, he said, adding that "I think this has the potential of engaging our fellow clinicians more than other things they don't get as excited about, like cognitive quality improvement, cognitive biases, or industrial CQI [continuous quality improvement] models."

In addition, "it lends itself to less defensive clinical responses if people see that this is a recurring pitfall: 'Yes, I missed the dissecting aneurysm, but I'm not the only one who's done this.' It doesn't mean that it was okay, but it begins to allow people to learn from their mistakes, to feel less singled out and less defensive," Schiff said.

In addition, it can "be a way to shine a light on what the recurring pitfalls in an institution," which might promote organizational change. For instance, with the example of a breast lump, "we've actually designed 'forcing functions' to prevent a woman from being sent away with a breast lump just after her normal mammogram," so they will have an ultrasound or other further testing, he noted.

Schiff said he would like to see various specialties each come forth with their own pitfalls using the framework he and his colleagues have developed. He also sees the framework as a teaching tool, "as a way of teaching about diagnosis and worrying about particular symptoms and what are some of the things that go wrong? Hopefully we could see sections in medical textbooks so people could be warned about these when they read up on a disease or even in real time."

Clinical decision support is another possibility for the framework, he said, "but we say this very cautiously -- we don't want doctors bombarded with a bunch of other warnings that they're ignoring, but in theory, the computer should have a context-aware knowledge of the patient's history and symptoms," he said. "And the physician could query to see what are some of the pitfalls in the symptom or the diagnosis they're considering."

Comment