View as PDF version

Patient Safety Tip of the Week

January 10, 2023

Diagnostic Errors in the ED



AHRQ recently published a study (Newman-Toker 2022) on diagnostic errors in the emergency department that has garnered substantial interest from the lay press. The study has been largely criticized in that some of the data came from European or Canadian emergency departments, rather than US emergency departments, and that it unfairly targets ED physicians. But the authors make it very clear that the diagnostic error rates found in the ED are very comparable to those found in hospital inpatient care or primary care.


We’ve long criticized the landmark report “To Err Is Human” in that it sensationalized statistics that garnered attention for a short period and ultimately did little to improve patient safety over the next decade. So, let’s cast aside the criticisms of the new AHRQ study, ignore the focus on statistics, and instead focus on the important lessons learned from the study.


The emergency department, by its very nature, is a high-risk venue. Interactions with patients are generally for a short period of time, often done under time pressure and multiple potential distractions. Most importantly, the clinician doing the ED evaluation may never see that patient again. Ensuring a proper handoff to the clinician who will next care for the patient is critical.  Our current systems for care coordination are not very good. We can’t tell you the number of times we hear a primary care physician say “I didn’t even know {my patient} had been in the ED”. Patients are often discharged from the ED before final imaging reports are available and we’ve all seen patients “fall through the cracks” when an important imaging finding is never communicated to someone who would ensure follow-up. Particularly in small rural hospitals, lack of specialty consultants may be a further problem. Add to all this the problem of overdiagnosis and impact of the “diagnostic cascade” that can itself lead to patient harm and add to healthcare expenditures.


The study was really a systematic review and meta-analysis based on an extensive review of the literature, encompassing 279 studies. An estimated 5.7% of all ED visits had at least one diagnostic error. Estimated preventable adverse event rates were 2.0% for any harm severity, 0.3% for any serious harms, and 0.2% for deaths. That translates to about 1 in 18 ED patients receiving an incorrect diagnosis, 1 in 50 suffering an adverse event, and 1 in 350 suffering permanent disability or death. Again, these rates are comparable to those seen in primary care and hospital inpatient care.


So those are the statistics. Yes, those are the ones garnering all the headlines in the lay press (New York Times, Boston Globe, CNN, and multiple others). And the statistics do little to help us improve patient safety. Rather, we must look at the real opportunities to improve patient safety. The more important questions to ask are “what are the serious conditions that are more likely to be missed?” and “what are the presenting symptoms that often lead to missed diagnoses?”. When seeing patients in any healthcare venue, the two most important questions we usually ask ourselves are “What could I be missing?” and “What’s the most serious thing I could be missing?”.


The top 15 clinical conditions associated with serious misdiagnosis-related harms, accounting for 68% of serious harms, were (in order of frequency):

·       stroke

·       myocardial infarction

·       aortic aneurysm and dissection

·       spinal cord compression and injury

·       venous thromboembolism

·       meningitis and encephalitis

·       sepsis

·       lung cancer

·       traumatic brain injury and traumatic intracranial hemorrhage

·       arterial thromboembolism

·       spinal and intracranial abscess

·       cardiac arrhythmia

·       pneumonia

·       gastrointestinal perforation and rupture

·       intestinal obstruction


The top five conditions noted above account for 39 percent of serious misdiagnosis-related harms.


But even knowing the clinical conditions most often missed is not enough. Patients don’t come to the ED with a complaint “possible spinal cord compression”. While it is nice to know the conditions most often missed, it would also be nice to know which diagnoses made were most often in error. For example, if it turns out a diagnosis of “vestibular neuritis” was erroneous 15% of the time, that might prompt ED clinicians to always think twice any time they consider such a diagnosis. And our focus should be on the presenting symptoms most often associated with missed diagnoses.


The AHRQ study did find that clinical presentation was important. For example, the average rate for missed stroke was 17% but it was 40% when the presenting symptom was dizziness/vertigo vs. 4% when the presenting symptom was weakness. The authors found that nonspecific, mild, transient, or “atypical” symptoms frequently contributed to missed diagnoses. In fact, atypical or non-specific symptoms were the strongest and most consistent predictors of increased risk for a missed diagnosis across diseases studied.


The authors provided a table with the most common “atypical” presenting symptoms and their related misdiagnosed diseases:


Abdominal pain

Myocardial infarction, aortic aneurysm/dissection, appendicitis, diverticulitis, ovarian disease, gallbladder pathology, cancer

Back pain

Spinal abscess or other spinal cord compression, myelitis, aortic aneurysm/dissection

Dyspnea/shortness of breath

Myocardial infarction, aortic aneurysm/dissection


Sepsis, aortic aneurysm/dissection (aortitis)


Stroke, (other diseases with headaches as a more “typical” presentation include subarachnoid hemorrhage, meningitis/encephalitis, raised intracranial pressure, and giant cell arteritis)


Myocardial infarction, aortic aneurysm/dissection, venous thromboembolism, stroke

Altered mental status/confusion

Stroke, sepsis



Fatigue/malaise/generalized weakness

Myocardial infarction, stroke, sepsis

Gait disturbance



Stroke, appendicitis, myocardial infarction


Atypical or non-specific symptoms were the strongest and most consistent predictors of increased risk for a missed diagnosis across diseases studied.


Patient age, gender, and race may impact error rates. For example, older age increases risk of missed appendicitis, while younger age increases risk of missed stroke 6.7-fold. The lay press (Medaris 2022) recently highlighted the latter problem, describing 3 patients in their 20’s whose strokes were misdiagnosed as being due to migraine or drugs.


Female sex and non-White race were often associated with important (20–30%) increases in misdiagnosis risk in the AHRQ study. There was also significant variation by hospital (for example, the rate of missed myocardial infarction varied from 0% to 29% across hospitals within a one study). Other notable predictors of misdiagnosis included care provided by less experienced clinicians, at non-teaching hospitals, with high ED discharge fraction, and during off hours. The diagnostic performance gap with academic (teaching) ED’s having lower false negative rates than community (non-teaching) ED’s was a fairly consistent finding, but it is unknown whether lower academic false-negative rates were achieved through greater overall diagnostic accuracy or by favoring overutilization, leading to arbitrarily greater admission fractions and resulting in higher false-positive rates.


The AHRQ study also cites previous work from malpractice claims on factors contributing to ED diagnostic error. Such factors include failure/delay in ordering diagnostic test; lack of/inadequate patient assessment with premature discharge; narrow diagnostic focus in patient assessment with failure to establish differential diagnosis; failure to appreciate and reconcile relevant signs, symptoms, or test results; failure/delay in obtaining a consultation or referral; misinterpretation of diagnostic test studies (e.g., X-rays); issues related to lack of health insurance; inadequate communication among providers regarding the patient’s condition; off-hours care (weekend, night shift, or holiday); inadequate history or physical examination.


Of course, we always need to balance the risk of misdiagnosis against the risk of overdiagnosis or unintentionally initiating the “diagnostic cascade”. For example, though stroke was often misdiagnosed when vertigo/dizziness was a presenting symptom, we certainly don’t want every patient with vertigo or dizziness to get a CT scan or MRI scan. We know that the diagnostic cascade not only adds to the cost of healthcare but may also lead to unnecessary procedures that, in turn, may lead to patient harm.


From the studies done on diagnostic error in the ED, it is also difficult to understand the likely role of fatigue, which we know increases error rates in a variety of venues. There is also one other consideration never assessed in these studies. We’ve often cited a study on shift workers in fields other than healthcare (Folkard 2003) which showed that the risk of incidents increased each consecutive day worked. For example, on average for night shifts risk was 6% higher on the second night, 17% higher on the third night, and 36% higher on the fourth night (for morning/day shifts the corresponding risks were 2%, 7% and 17%). There’s little reason to think that such trends might not also occur in healthcare, particularly in the ED where long consecutive-day shifts are common.


The AHRQ study wasn’t the only recent study on diagnostic errors in the ED. Baartmans et al. (Baartmans 2022) studied 23 serious error event reports of diagnostic events in emergency departments of Dutch general hospitals and identified human errors. Twenty-one reports contained a diagnostic error, in which the authors identified 73 human errors. These were mainly based on intended actions (n = 69) and could be classified as mistakes (n = 56) or violations (n = 13). Most human errors occurred during the assessment and testing phase of the diagnostic process.


Similar to the AHRQ report, most prevalent missed or delayed final diagnoses were cardiovascular (n = 11) or neurological (n = 5) conditions, such as aorta dissection, ruptured abdominal aortic aneurysm, subarachnoid hemorrhage, and spinal cord injury. They found that diagnostic errors occurred most often in the assessment and testing domains. Other human errors sometimes occurred during referral or consultation, history taking, the physical exam, follow-up, and at access to care or presentation. Most human errors involved a failure or delay in recognizing the urgency of the situation or putting too much weight on a competing or coexisting diagnosis (for example, the working diagnosis of a migraine disrupted the search for other causes of the symptoms of a patient with a subarachnoid hemorrhage). Other recurring themes were the failure or delay to consider a diagnosis, failed or delayed follow-up of (abnormal) test results, failure or delay in ordering needed tests, wrong test orders, and failure or delayed communication or follow-up of a consultation.


The authors note that hospitals seldom propose overarching interventions to support the diagnostic process, for example, evidence-based strategies to improve specific knowledge, reforming training methods, structural feedback and reflection on diagnostic discrepancies, implementation of diagnostic decision support systems to improve diagnostic calibration, or implementing team-based diagnosis. They also note that system-aimed interventions (e.g., lowering work pressure and crowding in the ED, improving patient safety culture, teamwork interventions, and cultural aspects) are also rarely proposed, while these types of recommendations may have a better chance of being effective in preventing similar cases and to improve diagnostic safety.


Diagnostic error remains our greatest patient safety challenge. It’s an area for which we’ve had the fewest solutions to date, and it applies to every healthcare venue, not just the ED. Sensationalizing the statistics is not likely to lead to any productive changes. Rather, focusing on the nature of the errors and factors contributing to them is much more likely to lead to productive changes. In that regard, the AHRQ study is a useful beginning.






Newman-Toker DE, Peterson SM, Badihian S, et al. Diagnostic Errors in the Emergency Department: A Systematic Review. Comparative Effectiveness Review No. 258. (Prepared by the Johns Hopkins University Evidence-based Practice Center under Contract No. 75Q80120D00003.) AHRQ Publication No. 22(23)-EHC043. Rockville, MD: Agency for Healthcare Research and Quality; December 2022



Medaris A. Doctors dismissed these 3 young people's symptoms as migraines or drug-related — but they were really having strokes. Insider 2022; Dec 26, 2022

(Medaris 2022)



Folkard S, Tucker P. Shift work, safety and productivity. Occupational Medicine 2003; 53: 95-101



Baartmans MC, Hooftman J, Zwaan L, et al. What Can We Learn From In-Depth Analysis of Human Errors Resulting in Diagnostic Errors in the Emergency Department: An Analysis of Serious Adverse Event Reports. Journal of Patient Safety 2022; 18(8): e1135-e1141






Print “PDF version










Tip of the Week Archive


What’s New in the Patient Safety World Archive