Clinical Context

All clinicians have missed a diagnosis at some point in their careers, and the current study cites expert opinion that from 10% to 15% of all diagnoses are incorrect. They also note that research into misdiagnosis is difficult, and retrospective reviews can be limited by poor documentation in the medical record, low interrater reliability, and hindsight bias. Still, a prospective study of 348 primary care visits found cases of misdiagnosis in 13% of all visits. Estimates in this range would result in 100 to 200 million cases of misdiagnosis in the United States each year. The goal of this clinical brief is to review the results of the current study related to clinical misdiagnosis.

The prevalence of serious harm related to misdiagnosis is even harder to capture, but the authors note previous research that has found rates of misdiagnosis-related harms of 0.22% among hospitalized patients and 0.81% of primary care patients. This translates into approximately 80,000 serious harms annually among hospitalized patients and approximately 4 million annual cases of harm among primary care patients.

Vascular events, cancer, and infections account for the majority of morbidity and mortality resulting from misdiagnosis. The current study assesses the rate of misdiagnoses in these domains, using previously published data.

Study Synopsis and Perspective

One in 10 people with dangerous symptoms from cancers, infections, or major vascular events are misdiagnosed or diagnosed too late, and more than half (53.9%) of those individuals will be permanently disabled or die as a result of the error, researchers have found.

David E. Newman-Toker, MD, PhD, director of the Armstrong Institute Center for Diagnostic Excellence at Johns Hopkins University School of Medicine in Baltimore, Maryland, and colleagues reported their findings online May 14 in Diagnosis.^[1]

They conducted a literature analysis that found that diagnostic rates have not declined appreciably during the last several decades, and that for some diseases they appear to be rising. The Society to Improve Diagnosis in Medicine funded the study.

Of the 9.7% of patients who are misdiagnosed with 1 of the top 15 conditions within the "Big Three" category (ie, vascular events, infections, and cancers), 53.9% will suffer permanent disability or death. Dr Newman-Toker told Medscape Medical News that numerous factors are behind misdiagnoses, but the overarching theme is lack of investment in the problem.

The money spent on researching diagnostic error is only about $7 million a year^[2]; for perspective, that is less than the $10 million spent each year on smallpox research, "a disease eradicated half a century ago," he explained.^[3] "This is the most underserved public health problem that we know of in all of medicine right now," he added.

The disease most often misdiagnosed (missed 62.1% of the time) among the 15 conditions was spinal abscess, an infection that can compress the spinal cord and cause paraplegia.

Reasons for Misdiagnosis Differ

The chances of being misdiagnosed were different among the Big Three and seemed to happen for different reasons.

With cancers, the diseases most likely to be missed are those with the least successful diagnostic screening programs, the authors suggest. The lowest misdiagnosis and combined diagnostic error-harm rates were seen with prostate cancer (2.4% and 1.2%, respectively), for which screening is frequent in the United States. The highest misdiagnosis and harm rates were seen for lung cancer (22.5% and 13.8%), for which screening remains below recommended levels.

For infections and major vascular events, the more uncommon diseases are more likely to be missed and result in the highest harm rates, such as endocarditis (25.5% and 13.4%), meningitis and encephalitis (25.6% and 14.3%), aortic aneurysm and dissection (27.9% and 16.8%), and spinal abscess (62.1% and 35.6%).

The researchers point out that myocardial infarction has the lowest diagnosis-related harm rate, at 1.2%, but that this rate comes "after a half century of focused efforts to automate electrocardiogram interpretation, develop and refine biomarkers (e.g. troponin), and create routine diagnostic protocols for chest pain or suspected acute coronary syndromes."

The other 14 conditions in the list need that kind of attention to help make a big dent in diagnostic error, he said.

Rates Have Not Declined; Some May Be Rising

The researchers note that Big Three misdiagnoses have not declined during the last several decades.

They point to an "alarming" trend highlighted by a study analyzing Medicare data from 2007 to 2014 that showed rates of missed diagnoses for stroke, subarachnoid hemorrhage, and aortic aneurysm rupture were rising.^[4]

Strokes are about as common as heart attacks, but diagnostic error rates "are about 5 to 10 times as high," Dr Newman-Toker said. "We've not had a concerted effort and push to make this a focal point of interventions."

Gordon Schiff, MD, associate director for the Center for Patient Safety Research at Brigham and Women's Hospital in Boston, Massachusetts, told Medscape Medical News that "we need more papers like this," but he said it should not be surprising that proper and timely diagnoses continue to confound medicine.

One problem in primary care is the lack of time physicians have with each patient to determine whether symptoms could be serious, he said. That is compounded if it is a new patient and the provider does not know whether a headache complaint is unusual, for instance.

"Almost any symptom could be cancer," he noted.

Dr Schiff added that there is also a lack of follow-up, either on the part of the physician or the patient, made worse by a fractured health system in which patients may see several providers without a foolproof way that all will be informed of what has happened to the patient in every step of care.

Tracking trends in diagnostic error is inherently difficult, he said, for reasons including that there is no set standard for at what point a diagnosis is delayed for each disease or whether a test would have certainly changed the outcome.

"It's hard to draw a conclusion about whether things are better or worse," he said in describing diagnostic trends over time.

Better training of physicians is also necessary, as they are caught in the middle of feeling they need to do more tests to avoid misdiagnoses and being told they need to do fewer tests to spend healthcare dollars more efficiently.

In some cases, testing may be the problem, Dr Schiff noted. If spinal abscess is so often missed, perhaps a better test would help diagnosis, he said.

Diagnostic "safety nets," such as automated calls that follow-up with a patient, are also important, although they do not guarantee that patients will answer the call or respond to it, he points out.

Ideally, he said, there would also be a national database in which patients and physicians could detail when signs and symptoms occurred, when a diagnosis was made, whether it was found correct, and if not, why not, and all that followed so learning could advance outside the arena of malpractice lawsuits.

"It's a matter of patients and doctors working together," he said.

Narrowing the Target

The authors hope this work narrows the target from a vast, seemingly unsolvable problem to a list of just 15 targeted conditions to focus on that could make a huge difference in reducing diagnostic error.

One way investment will help is in making subspecialty consults more available so providers can quickly reach specialists in their health system if they do not have them in their hospitals, Dr Newman-Toker said. Another is in reimbursing for telemedicine to scale expertise across the country.

"But in the long run we're going to have to invest in solutions in a much bigger way than we have been," he said.

Dr Schiff agrees and said the COVID-19 crisis is likely to present further obstacles to accurate diagnoses.

An example might be a provider asking a patient to get a colonoscopy after recent unexplained weight loss, only to have the procedure delayed as nonessential in the current pandemic.

"All my visits in the last 6 weeks have been by telephone," he said. "It doesn't allow me to examine their abdomen or feel their lymph nodes or see whether their body language conveys they have something serious."

Dr Newman-Toker and colleagues analyzed data from 28 published studies representing 91,755 patients.

To calculate harms, they used estimates from literature of the generic (disease-agnostic) rate of serious harms per diagnostic error and applied claims-based severity weights to calculate disease-specific rates.

"Results were validated via expert review and comparison to prior literature," the authors write.

The study was funded by the Society to Improve Diagnosis in Medicine through a grant from the Gordon and Betty Moore Foundation. The authors and Dr Schiff have disclosed no relevant financial relationships.

Diagnosis. Published online May 14, 2020.

Clinical Implications

• The overall rate of misdiagnosis is thought to be between 10% and 15%. Rates of serious harms related to misdiagnosis have been estimated to be 0.22% among hospitalized patients and 0.81% among primary care patients.
• The current study found the highest rates of misdiagnosis and harm associated with spinal abscess. Comparatively, misdiagnosis rates for myocardial infarction, sepsis, and breast and prostate cancer were much lower, with lower rates of harm.
• Implications for the Healthcare Team: The current study identifies commonly missed diagnoses that may be targeted for quality improvement efforts. The healthcare team can employ evidence-based guidelines in the evaluation of high-risk conditions.

Earn Credit