Subscribe to eTOC

A Neurologist Takes on Medical Misdiagnoses at the Helm of the Society to Improve Diagnosis in Medicine — and Elsewhere



DR. DAVID NEWMAN-TOKER and colleagues published a prototype dashboard for stroke misdiagnosis that would allow hospitals to track and assess their performance. “Routine use of such dashboards could be transformational for institutions eager to start solving the diagnostic error problem.”

Patient-safety experts estimate that 40,000 to 80,000 hospital deaths each year can be attributed to diagnostic errors. Neurologist David Newman-Toker, MD, PhD, FAAN, the president of the Society to Improve Diagnostic Medicine, has devoted his career to research on strategies to bring that number down.

David Newman-Toker, MD, PhD, FAAN, was a neurology resident in Boston when he first became alarmed about the misdiagnosis of neurological conditions.

“People suffered permanent brain damage, strokes, paraplegia, and blindness that could have been prevented,” said Dr. Newman-Toker, director of the division of neuro-visual and vestibular disorders in the neurology department at Johns Hopkins University School of Medicine.

In 2011, Dr. Newman-Toker joined the Society to Improve Diagnosis in Medicine as a founding board member; in November, he became the fast-growing organization's second-ever president. It's his newest leadership role in nearly two decades focused on eliminating harm from diagnostic errors.

Dr. Newman-Toker directs the Johns Hopkins Medicine Armstrong Institute Center for Diagnostic Excellence, which is advancing the science of diagnostic safety and developing ways to improve diagnostic performance broadly. His own research focuses primarily on improving the diagnosis of stroke and other disorders affecting the brainstem and cranial nerves.

In doing so, he seeks to redress the preventable harm he first saw as a trainee.

“I said, ‘This can't happen, and I need to do something about it,’” he said. “So I devoted my career to it.”


The Society to Improve Diagnostic Medicine was incorporated in 2011, but its roots go back to the first Diagnostic Error in Medicine Conference three years earlier. That event brought together a relatively small group of individuals who recognized that health care's patient-safety movement was not focusing attention on diagnostic problems.

Indeed, To Err Is Human, the 1999 Institute of Medicine (IOM) report that exposed the huge safety shortcomings in American health care, did not identify the issue of diagnosis to be a problem.

“There were a few casual references, but it was virtually ignored in To Err is Human,” said Paul Epner, chief executive officer of the Society to Improve Diagnosis in Medicine.

One reason diagnostic problems had received little attention is that there was no standard definition of diagnostic errors and, thus, no way of counting how often they occurred.

The Society convinced the National Academies of Sciences, Engineering, and Medicine (the successor to the IOM) to convene a committee (NAM) to study diagnostic error — and raised nearly $1 million to help fund the work.

The NAM committee advanced the matter by creating this definition for diagnostic error: the failure to establish an accurate and timely explanation of a patient's health problem or communicate that explanation to the patient.

“The report said, ‘We can't tell you how often this occurred, but this is a huge problem,’” Epner said.

To be specific, the NAM committee said diagnostic errors contribute to about 10 percent of patient deaths; account for 6 to 17 percent of hospital adverse events; are the leading type of paid medical malpractice claims; and are almost twice as likely to have resulted in patient deaths than other claims and account for the highest proportion of total payments.

Beyond that, three high-profile patient-safety experts — Lucian Leape, MD; Don Berwick, MD; and David Bates, MD — have estimated that 40,000 to 80,000 hospital deaths each year can be attributed to diagnostic errors.


PAUL EPNER: “We estimate that were getting it right about 90 percent of the time — thats both the cognition side as well as all the systematic issues, which is kind of remarkable. But if a pilot lands 90 percent of their planes safely, thats not good enough. We know we can do better.”

Cognitive errors — physicians simply getting it wrong — are sometimes at fault; in some cases, that stems from barriers to physician/patient communication. Epner pointed out that there are some 10,000 known diseases and nearly 5,000 medical tests, but patients have fewer than 100 symptoms to describe how they feel — and about 10 minutes to convey a new complaint to a physician.

The bigger culprit is systematic issues, such as inaccurate medication lists in the electronic health record, patients' failure to pursue referrals, miscommunication between clinicians; lab or imaging reports that are not communicated to a physician or a patient and others.

“We estimate that we're getting it right about 90 percent of the time — that's both the cognition side as well as all the systematic issues, which is kind of remarkable,” Epner said. “But if a pilot lands 90 percent of their planes safely, that's not good enough. We know we can do better.”


Stroke presents a particular challenge because it can stem from many causes and show itself through a wide range of symptoms, said Louis Caplan, MD, FAAN, professor of neurology at Harvard Medical School and a senior member of the division of cerebrovascular diseases at Beth Israel Deaconess Medical Center in Boston.

Dr. Caplan, who will present a course on cognitive errors in diagnosis and management at the AAN 2019 Annual Meeting, said structural issues lead to cognitive errors in its diagnosis. For starters, physicians who evaluate patients for stroke feel pressure to act quickly and to use guidelines to identify inclusion or exclusion criteria that guide treatment decisions.

“So there are two big problems because of that: the absence of thinking and the absence of a systematic approach to diagnosis,” he said. “The systematic approach is to take a good history of how the stroke occurred, what the setting was, how quickly it developed, what the symptoms were, and what the risk factors were.”

Using that information, plus a physical examination, a physician should consider possible diagnoses before deciding which imaging studies would help make a diagnosis. “This is not what's happening,” he said. “People are going to the emergency department [ED] and [physicians] are following recipes. They're doing automatic pictures. They're not thinking of what's wrong with the patient.”

One result: roughly 9 percent of cerebrovascular events are missed on a patient's first visit to an emergency department, according to a meta-analysis published in Neurology in 2017 by Dr. Newman-Toker's team.

The rate of misdiagnosis increases if the patient's symptoms are mild, non-specific, or transient. Dizziness and vertigo are the symptoms often linked to missed stroke, so patients with stroke who go the ED complaining of those symptoms get the wrong diagnosis at least 40 percent of the time, according to Dr. Newman-Toker's research. So perhaps it is not surprising that an estimated 15,000 to 25,000 people with stroke who go to the ED with dizziness and vertigo each year suffer serious — and potentially preventable — harm from a diagnostic mistake, Dr. Newman-Toker said.

In many — perhaps most — cases, physicians involved in diagnostic errors never realize they occurred. An emergency physician who sends a patient home with the diagnosis of a benign ear condition (when it actually was an early stroke symptom) will usually not be notified if the patient is hospitalized with a major stroke a week later.


DR. LOUIS CAPLAN: “People are going to the emergency department [ED] and [physicians] are following recipes. Theyre doing automatic pictures. Theyre not thinking of whats wrong with the patient.”

“Once the patient gets into the hospital or is discharged, if there has been a mistake, the ED physician never hears about it,” Epner said. “There is no feedback loop.”

Without feedback, there is no opportunity for physicians to learn how to improve their performance. But that may be changing. Dr. Newman-Toker and his colleagues have developed a measure — a rate of excess harms after missed stroke — that they intend to submit to the National Quality Forum in 2019. That is the first step in the process of getting the Centers for Medicare & Medicaid Services to adopt the measure for its pay-for-quality program.

Earlier this year, researchers at the Johns Hopkins University School of Medicine and Kaiser Permanente published a prototype dashboard for stroke misdiagnosis that would allow hospitals to track and assess their performance. The rate of adverse events stemming from misdiagnosis are too small to allow for physician-level analysis, but the dashboards will contain process of care data that can show, for example, which tests are associated with diagnostic success for patients with certain symptoms.

“Routine use of such dashboards could be transformational for institutions eager to start solving the diagnostic error problem,” Dr. Newman-Toker wrote in a Johns Hopkins Medicine blog post earlier this year.


Dr. Newman-Toker chaired the Society's policy committee in 2017, when it worked to educate policymakers about diagnostic safety. Those efforts led to $2 million in the 2019 federal budget to support funding to address diagnostic errors.

“It's not a lot of money, but it was a huge lift to help them see the wisdom of that investment,” Epner said. “We're thrilled that they recognized the value of this work, and at least in my mind, created a down payment of $2 million.”

Meanwhile, more than 40 provider organizations, medical societies, and patient-advocacy organizations have formed the Coalition to Improve Diagnosis, pledging to identify and implement ways to improve diagnostic performance.

Johns Hopkins Medicine is a Coalition member and its first target is clear: reduce harms from missed strokes in half within five years. Dr. Newman-Toker's Center for Diagnostic Excellence is using a multi-pronged approach to address the problem. Their strategies include adapting a validated culture-change framework — the Comprehensive Unit-based Safety Program — to improve diagnoses by identifying diagnostic error problems, causes and solutions within a clinical unit; using screen-based educational simulations to train providers in bedside diagnostic techniques; deploying new technologies to improve early stroke recognition in emergency departments, and providing feedback about diagnostic accuracy and errors to providers, including information tracked from the diagnostic performance dashboard.


• Leape LL, Berwick DM, Bates DW. Counting deaths due to medical errors—Reply JAMA 2002; 288(19): 2405.
• Liberman AL, Newman-Toker DE. Symptom-disease pair analysis of diagnostic error (SPADE): A conceptual framework and methodological approach for unearthing misdiagnosis-related harms using big data BMJ Qual Saf 2018; Epub 2018 Jan 22.
    • Mane KK, Rubenstein KB, Nassery N, et al. Diagnostic performance dashboards: Tracking diagnostic errors using big data BMJ Qual Saf 2018. Epub 2018 Mar 17.
      • National Academies of Sciences, Engineering, and Medicine. 2015. Improving Diagnosis in Health Care Washington, DC: The National Academies Press.
        • Saber Tehrani AS, Lee H, Mathews SC, et al. 25-Year summary of US malpractice claims for diagnostic errors 1986-2010: an analysis from the National Practitioner Data Bank BMJ Qual Saf 2013;22(8):672–680.
        • Saber Tehrani AS, Kattah JD, Kerber KA, et al. Diagnosing Stroke in Acute Dizziness and Vertigo: Pitfalls and Pearls Stroke 2018. 2018;49(3): 788–795
        • Tarnutzer AA, Lee S, Robinson K, et al. ED misdiagnosis of cerebrovascular events in the era of modern neuroimaging: a meta-analysis Neurology 2017; 88 (15)1468–1477.