Maybe it’s a misread mammogram or a stroke dismissed as an inner ear problem. Most Americans will experience at least one diagnostic error in their lifetime.
That’s the conclusion of a recent Institute of Medicine report, and the challenge for David Newman-Toker. The professor of neurology and otolaryngology–head and neck surgery is director of the new Center for Diagnostic Excellence, established with a $5 million gift from C. Michael Armstrong. It is part of the Armstrong Institute for Patient Safety and Quality.
Dome spoke recently with Newman-Toker. Excerpts are below.
Q: What are diagnostic errors?
A: The Institute of Medicine defines diagnostic error as “the failure to establish an accurate and timely explanation of the patient’s health problem(s) or communicate that explanation to the patient.” The communication piece is a big change from prior definitions. It doesn’t do any good to write down the correct diagnosis if the patient never gets it.
The best measures available suggest 12 million Americans suffer a diagnostic error every year, and up to one-third of these suffer serious permanent harms, including disability or death.
Q: Are certain misdiagnoses very common?
A: Cancer, infections and vascular events, like heart attacks and strokes, account for at least one-third of diagnostic errors and probably more than half of the harms from diagnostic errors.
Q: Why are diagnostic errors so pervasive?
A: Misdiagnosis is incredibly frequent because medicine is incredibly hard. There’s uncertainty and complexity and incomplete information all the time. But we can do better, and we will.
Q: How does your background inform your understanding of diagnostic errors?
A: I’m a neuro-ophthalmologist and neuro-otologist. We deal with stuff like unexplained vision loss, headaches, optic nerve problems, dizziness, vertigo and related diseases. Neuro-ophthalmologists and neuro-otologists have developed pretty extensive skills in bedside diagnosis. We can learn a surprising amount from looking carefully at people’s eyes and eye movements. We can tell the difference between a stroke and an ear problem more accurately than an MRI can in the
first 72 hours because the anatomy lags behind the physiology.
What got me interested in diagnostic error, aside from my general interest in diagnosis, was that during my residency, I saw many instances of misdiagnoses that harmed patients. One case was a woman in her 50s. English wasn’t her first language, and nobody drilled down to what she was actually saying. They kept hearing chest pain and trouble walking the stairs, which they put together as a cardiac story. After a series of workups that didn’t reveal any heart problems, she continued to come back to the hospital, where she was dismissed as overly concerned with “benign” symptoms.
The important detail was that she struggled going down stairs, but not up. And when you asked her specifically about the chest pain, even though she pointed to her chest, she said it wrapped around to her midback. Those are classic symptoms of spinal cord compression. She finally had emergency surgery, but it was too late. She wound up paralyzed from the waist down.
Q: How do language, culture and race differences stand in the way of correct diagnosis?
A: Some diagnostic challenges are purely related to language barriers and cultural differences. These can be solved by bringing in people with language and cultural skills training. The tougher pieces, though, are the biological disease differences across races and genders. Women and minorities are more likely to be misdiagnosed because our data are based on white men.
For example, women’s heart attack symptoms are different from men’s. But the science and literature about chest pain diagnosis is about men, and that is basically what is taught in medical school. So when women come to the hospital with back pain and nausea, rather than classic symptoms of chest pain traveling down the left arm, they can be misdiagnosed.
Q: How can diagnosis be improved?
A: Teamwork is critical. I just wrote a paper about this with a colleague who is a therapist in North Carolina. It’s called “Diagnosis Is a Team Sport,” and it appeared in Diagnosis earlier this year.
Physicians would send the therapist patients, asking her to treat their benign paroxysmal positional vertigo (BPPV). My colleague had to tell physicians that the patients didn’t have BPPV—they’d had strokes. It was complicated because she’s a physical therapist, and they are physicians.
We spent a lot of time breaking down those walls and getting the doctors to recognize her expertise. Now, a lot of them send her patients with instructions to evaluate and treat for dizziness, acknowledging that she may be better at diagnosing the problem.
Nurses represent a really special and important case because they are with the patient most of the time. They have the opportunity to see early signs of diseases like sepsis and pulmonary embolus. But because there are sociocultural barriers blocking nurses from assisting in medical diagnosis, we are wasting that information, and the patients are suffering. Nurses need to be integrated into the diagnostic team.
Q: The center set an ambitious goal of reducing harms from stroke misdiagnosis by 50 percent within five years for all emergency departments in the health system. How will you do that?
A: Seven years ago, we published our method for using expert analysis of eye movements to tell inner ear disease from stroke. It takes two minutes, and the results are more accurate than an MRI scan. But the expertise takes five to 10 years to develop. It’s unrealistic to think every emergency physician and primary care doctor is going to learn how to do it.
So we use a device that looks like a set of swim goggles. It measures head and eye movements, and software interprets the results. We’re doing a National Institutes of Health clinical trial to see if combining the goggles with decision support will lead to better diagnoses. Meanwhile, we’re launching a “tele-dizzy” consultation that will use the goggles for real-time access to our eye movement experts via telemedicine, reducing the risks of missed strokes.
Q: Will there be tools like these stroke goggles for other illnesses?
A: Absolutely. Take sepsis, for instance. Suchi Saria is a computer scientist who is developing computer-based tools that take mountains of data that humans couldn’t possibly sift through and send out warning signals when patients are deteriorating in the intensive care unit. And this is just the beginning.
Q: What are the center’s long-term goals?
A: The center, which is the first of its kind in the world, is designed to convene transdisciplinary teams for studying misdiagnosis, the same way the Armstrong Institute convenes such teams for issues related to patient safety.
The center provides a space for clinicians, researchers, engineers and data experts to work together and be thoughtful about how we tackle both the technical challenges and the sociocultural challenges of misdiagnosis. Together, we will solve this problem.