Journal Logo

Special Report

Special Report

The Persistent Challenge of Missed Strokes

Shaw, Gina

doi: 10.1097/01.EEM.0000488815.49329.37
    Stroke or not? This patient had a strong clinical history for stroke-like onset, but actually had a glioblastoma. The T2W axial section, above, showed edema primarily in the right middle cerebral artery territory, but with additional involvement of the medial temporal lobe, thalamus, and periatrial regions.

    More than one in five patients presenting to two hospitals with certified stroke centers were initially misdiagnosed, according to a new study published in Stroke. (2016;47[3]:668).

    A retrospective chart review of 465 patients at one academic medical center and one community hospital found that 103 ischemic strokes — 22 percent — were missed. A stroke was considered “missed” if emergency physicians consulted neurology for a possible stroke diagnosis and the neurology consultant felt that the patient did not have a stroke and admitted the patient to a medicine service. The missed diagnoses had clear consequences: Thirty-three percent of misdiagnosed patients were readmitted at 60 days post-discharge compared with 17 percent of accurately diagnosed patients (p=0.012).

    A larger proportion of strokes was missed at the community hospital (48 of 185; 26%) than at the academic hospital (55 of 280; 20%, p=0.11), but they also missed fewer strokes that presented within a clear time window for intervention; 28 of 55 (51%) missed strokes arrived at the academic hospital in time for consideration of lytics or endovascular therapy compared with 17 of 48 (35%) at the community hospital (p<0.001). Overall, 33 percent of the missed-stroke cases presented within the three-hour time window for tPA eligibility, while an additional 11 percent presented between three and six hours of symptom onset for consideration of endovascular therapy.

    A neurologic consult contributed to accurate diagnoses. Virtually all of the accurate diagnoses (213 of 225; 95%) were seen by neurology in the ED (p<0.001). But more than one-third of missed ischemic strokes at the academic medical center (20 of 55; 35%) had a neurology consult in the emergency department, and still missed early diagnosis. Forty percent of missed-stroke patients did not have neurological examinations with elements of the NIHSS compared with eight percent of the accurately diagnosed stroke patients (p<0.001). Patients with nausea and vomiting, dizziness, and prior strokes were more likely to be misdiagnosed, while patients with focal weakness, vision changes, gaze preference, and dysarthria were more likely to be correctly diagnosed.

    One of the strongest risk factors for misdiagnosis found in the study, not surprisingly, was posterior-circulation strokes, which were nearly three times more likely than anterior strokes to be missed, with 37 percent of posterior strokes initially misdiagnosed compared with 16 percent of anterior strokes (p<0.001). Posterior circulation strokes represent about 20 percent of all ischemic strokes.

    “The idea that we miss strokes is not novel, but this study confirms the hypothesis in both an academic and community setting and illustrates the type of strokes that are missed,” said lead author Joseph Schindler, MD, an associate professor of neurology and neurosurgery and the clinical director of the Yale-New Haven Stroke Center. “Stroke is a complex disease, and patients present with a varied presentations. Some of these presentations are less recognized in the emergency [department]. Unfortunately, unlike other diseases, there is no blood test or biomarker that can quickly test to rule in the diagnosis similar to myocardial infarction. Physicians have to rely on their clinical skills, and most identification tools are biased toward recognizing strokes that occur in the anterior hemisphere, causing symptoms such as language dysfunction or impairments in strength.”

    The National Institutes of Health Stroke Scale (NIHSS) is strongly weighted toward deficits caused by anterior circulation strokes while deficits from posterior circulation strokes receive fewer points. This is well illustrated by the new study's findings. The mean NIHSS scores were dramatically lower in missed-stroke patients than in accurately diagnosed patients — 4.9 vs. 7.3 at the academic center (p=0.042) and 2.8 vs. 6.4 at the community hospital (p=0.076). Overall, the mean NIHSS was 8.2 for anterior strokes versus 3.8 in posterior strokes (p<0.001).

    “The most common atypical stroke symptom leading to misdiagnosis is dizziness and vertigo, followed by symptoms such as nausea or vomiting, trouble walking, and headaches. All are common symptoms of posterior strokes, but none of them really conjures up an image of stroke for most people,” said David Newman-Toker, MD, PhD, an associate professor of neurology at the Johns Hopkins University School of Medicine and a leading expert on diagnostic errors. “This is true whether you're an emergency physician or a neurologist, although neurologists are probably on average better trained to detect posterior strokes. The truth is that sometimes diagnosing stroke requires expertise that isn't available on site, even in a stroke center.”

    Operating on False Information

    Dr. Newman-Toker and his colleagues published a paper a number of years ago about a simple 10-question true-or-false questionnaire about the implications of specific clinical findings for diagnosis in patients with dizziness. (Acta Otolaryngol 2008;128(5):601.) A sample question: “A patient with several isolated bouts of vertigo lasting 5-10 minutes is most likely to have BPPV [benign paroxysmal positional vertigo]. True or False?” (In fact, the paper noted, “Vertigo caused by BPPV typically lasts 10-40 seconds, and almost never longer than two minutes. Five to 10 minutes is a common duration reported in patients experiencing TIAs in the posterior circulation.”

    When a sample of 14 emergency physicians (a mix of residents and attendings) and 14 primary care physicians (all attendings) at two different university teaching hospitals took the quiz, their answers were “statistically significantly worse than chance,” Dr. Newman-Toker recalled. The emergency physicians had a median score of 30 percent, while the primary care physicians had a median score of 29 percent, and not a single question was answered by the two groups at rates significantly above what you would expect from guessing alone. In fact, six of the 10 questions were answered correctly at rates significantly worse than a coin flip (8-26%, p=0.00002-0.02).

    “That's because they were operating with misinformation. They weren't guessing; they ‘knew’ the right answer; they had just been taught the wrong thing,” Dr. Newman-Toker said. “The misconceptions in their answers paralleled those found in commonly used medical and neurologic textbooks. When we gave the same questionnaire to a group of neurology residents, they got about half of them right, which supports the idea that there could be a fair number of neurologists out there unable to distinguish stroke from benign causes of dizziness or vertigo.”

    In training, the emphasis is on patients with anterior strokes who usually present with one-sided weakness or obvious language or mental disturbances, Dr. Newman-Toker noted. “When a patient presents with sudden weakness on one side, stroke is the most common cause, so these patients are rarely missed. By contrast, only about three percent of dizzy patients are actually having a stroke, and the most common cause is benign inner ear disease. Residents are not usually taught how to distinguish subtle eye-movement signs of stroke from more common vestibular diseases in their neurology training, and certainly not in emergency training, so the default is often to mistakenly assume the problem is from an inner ear disease. This puts stroke patients experiencing dizziness or vertigo at particularly high risk of being missed.”

    Fine-Tuning Diagnosis

    For patients with acute, severe vertigo or dizziness known as the acute vestibular syndrome (AVS), the HINTS exam, a three-step bedside test developed by Dr. Newman-Toker and colleagues (Stroke 2009;40[11]:3504) and previously reported in EMN (2014;36[10]:1;, has been found to be highly sensitive (96-100%) and highly specific (85-98%) for differentiating stroke. (Acad Emerg Med 2013;20[10]:986; J Neurol 2011;258[5]:855; Ann Emerg Med 2014;64(3):265.)

    The three components are:

    • HI for head impulse testing, which is usually abnormal with ear causes but normal in stroke.
    • N for nystagmus, to remind the clinician to look for direction-changing or vertical nystagmus.
    • TS for test of skew, to look for vertical misalignment of the eyes, also usually a sign of stroke.

    A step-by-step video guide to HINTS can be found at

    Emergency physicians and neurologists alike are operating from the wrong paradigm when differentiating stroke from more benign causes of AVS, said Jonathan Edlow, MD, a professor of emergency medicine at Beth Israel Deaconess Medical Center and Harvard Medical School in Boston. “We tend to focus on the patient's description of their ‘type’ of dizziness — the symptom quality paradigm — but patients do not reliably and consistently distinguish one type of dizziness from another, and their descriptions do not correlate well with a list of possible diagnoses.”

    He suggested a “timing and triggers” approach to these diagnoses that is more consistent with the way other chief complaints are addressed. With the mnemonic ATTEST, it combines HINTS with several other items:

    • A for associated symptoms: Are there co-chief complaints or abnormal vital signs that suggest a given diagnosis or group of diagnoses, such as dizziness plus fever, cough, and sputum production?
    • TT for timing and triggers: Using history, define the timing and triggers category as above, just as you would for a patient with chest pain.
    • ES for exam signs: Use the physical examination, including a neurologic examination, testing the gait, and the HINTS exam, to clarify the diagnosis.
    • T for testing: Diagnostic ambiguity will persist in many cases where other tests will be necessary, such as to rule out a pulmonary embolism or a cerebellar stroke.

    Clinicians should also learn to look beyond the predictive risk of patient demographics and other features, Dr. Newman-Toker added, to factors such as age, smoking status, and blood pressure. “The risk prediction approach gets you about two-thirds of the way there, but it misses about one-third of the posterior strokes we're looking for. Young patients with symptoms like headache, dizziness, or vision problems are more likely to have these symptoms dismissed where these same symptoms in an older person would usually prompt consideration of stroke. In one of our studies, patients 18-45 years old were seven times more likely to have their stroke missed than patients over the age of 75.” (Acad Emerg Med 2013;20[10]:987; Diagnosis 2014;1[2]:155.)

    “It is true that if you took 100 40-year-olds with sudden-onset weakness, fewer would have stroke — as opposed to complex migraine or conversion reaction or post-seizure weakness — compared with 100 70-year-olds, but one is not going to make a stroke diagnosis if it's not even on the differential diagnosis,” Dr. Edlow said.

    Structured examination pathways and checklists also have the potential to improve diagnostic accuracy in more unusual stroke presentations, but these should be symptom-based checklists for dizziness, nausea, and headache. “It can't be a ‘stroke checklist,’ or it obviously won't work. It may not occur to the clinician to even use it,” Dr. Newman-Toker said. “Symptom-oriented checklists can help diagnose patients where the idea of a stroke has not occurred to the provider. But reminding them may not be enough if they don't know how to follow up to look for or rule out stroke. For example, some clinicians think that a negative CT scan rules out a stroke, but it actually misses more than 80 percent of acute ischemic strokes.”

    Misdiagnosis of stroke will never be completely eradicated, and it's probably not a goal to shoot for, Dr. Newman-Toker said. “There is always some inherent uncertainty, and some presentations are so nonspecific that if you were to evaluate every single patient, you would do more harm than good. You wouldn't want to work up every patient with back pain for stroke. But that doesn't mean we can't do a lot better than we're doing right now with posterior strokes.”

    Dr. Schindler said his study should serve as a wake-up call for hospitals to be sensitive to this issue, review their own cases, and have neurologists and emergency physicians collaborate to improve stroke systems of care. “Most hospitals can improve this metric by better communication between services and by lowering the threshold by which practitioners should evaluate stroke in the setting of nontraditional symptoms.”

    Dr. Newman-Toker also proposed using new technology to bring specialized diagnostic expertise to institutions that do not have subspecialists on site. “Re-educating all general clinicians in a narrowly focused area like eye movement diagnosis is probably not the best solution. Instead, portable devices that measure eye movements are now available that can be paired with computer-based decision support or linked directly to experts via telemedicine. This sort of technology will probably get you the most diagnostic bang for the buck in the long run.”

    Wolters Kluwer Health, Inc. All rights reserved.