How Diagnostic Errors Happen — And How to Protect Yourself
In 2015, the National Academy of Medicine — formerly the Institute of Medicine, the body whose 1999 report on medical errors first brought the issue to national attention — published a landmark report titled "Improving Diagnosis in Health Care." Its central finding was striking: most people will experience at least one meaningful diagnostic error in their lifetime, often at a moment when accurate diagnosis matters most.
The report estimated that diagnostic errors contribute to approximately 10% of patient deaths and are a factor in 6% to 17% of adverse events in hospitals. It identified diagnostic error as a largely unaddressed patient safety problem — one that had received far less attention than medication errors and surgical complications despite being at least as consequential.
The causes it identified were not primarily about incompetent physicians. They were about the cognitive patterns that all physicians share, the systemic structures within which medicine is practiced, and the predictable ways in which those factors combine to produce errors — even in the hands of skilled, well-intentioned clinicians.
Understanding those causes is not an academic exercise. It is the foundation for the practical steps patients can take to protect themselves.
The Cognitive Factors: How the Mind Makes Mistakes
Diagnostic reasoning is one of the most cognitively demanding tasks in medicine. A physician evaluating a patient must simultaneously hold in mind the patient's symptoms, their history, the relevant anatomy and physiology, the range of conditions that could explain the presentation, and the probability of each — while conducting a physical examination, managing the time pressure of the appointment, and attending to the patient as a person. The cognitive load is substantial, and the human mind manages it through shortcuts that are usually helpful and occasionally catastrophic.
Anchoring bias is the tendency to fix on an initial impression and weight subsequent information in light of it rather than reconsidering it from the ground up. When a physician forms an early hypothesis about what a patient has, that hypothesis becomes the lens through which subsequent findings are interpreted. Information that fits the hypothesis is registered; information that doesn't fit tends to be minimized or reframed rather than treated as evidence against the initial conclusion.
Anchoring bias is one of the most documented sources of diagnostic error in the literature. It is the mechanism by which a patient who presents with fatigue, weight loss, and low mood gets labeled with depression early in the encounter — and then every subsequent finding, including worsening physical symptoms, is interpreted as consistent with depression rather than as a potential signal that the diagnosis is wrong.
Premature closure is the tendency to stop gathering information once a diagnosis has been reached. In a busy clinical environment, arriving at an explanation is a cognitive relief — it allows the physician to move from diagnostic uncertainty to a treatment plan, from complexity to clarity. The pressure to close is real and understandable. But closing prematurely — before all relevant information has been considered, or before the diagnosis has been adequately tested — is one of the most common mechanisms behind missed diagnoses.
Premature closure is often the step that follows anchoring bias: the initial hypothesis anchors the reasoning, and the mind closes on that hypothesis before exploring alternative explanations fully.
Confirmation bias is the tendency to seek information that confirms an existing hypothesis rather than information that might disconfirm it. A physician who has concluded a patient has a particular condition will naturally be alert to findings that are consistent with it — and may be less alert to findings that point in a different direction. This is not a character flaw; it is a feature of how human cognition works. But in a diagnostic context, it means that important information can be systematically underweighted simply because it doesn't fit the leading hypothesis.
Availability bias affects which diagnoses a physician considers. Conditions that come easily to mind — because they are common, because the physician recently saw a case, because they are well-publicized — are overweighted relative to conditions that are rarer or less salient. A physician who saw three cases of influenza in the preceding week may be more likely to attribute fever and body aches to influenza and less likely to consider less common diagnoses that could present similarly.
Framing effects describe how the way information is presented shapes how it is processed. A patient who arrives with a chief complaint of "anxiety" frames the diagnostic encounter differently than a patient who arrives with "heart palpitations" — even if the underlying condition is the same. A referral letter that says "evaluate this patient for depression" frames the specialist's evaluation before they have seen the patient. The framing of a case shapes the cognitive path through it in ways that can be difficult to override even when the physician is aware of the effect.
The Systemic Factors: How the Healthcare System Creates Conditions for Error
Cognitive biases operate within a system, and the system matters. The structural features of modern healthcare create conditions that amplify the cognitive vulnerabilities above and introduce additional sources of error.
Time pressure is the most pervasive. As explored elsewhere, the standard primary care appointment provides 15 to 20 minutes for the physician to hear the presenting complaint, take a history, examine the patient, order or review tests, discuss findings, and document the encounter. Careful diagnostic reasoning — the kind that considers multiple hypotheses, questions the leading explanation, and reviews the full picture before concluding — takes time that the appointment structure doesn't reliably provide. Under time pressure, the shortcuts that produce cognitive errors become more likely.
Fragmented care means that no single physician has the full picture. A patient who has seen multiple providers across multiple health systems, who has had tests done at different institutions, and whose records are distributed across platforms that don't communicate has a medical history that no single physician has ever seen in its entirety. Diagnoses are made from partial information, connections between findings are missed, and the pattern that would be visible in a comprehensive review remains invisible in any individual encounter.
Specialist silos produce a related problem. Each specialist evaluates the patient through the lens of their specialty. The cardiologist looks for cardiac causes. The gastroenterologist looks for gastrointestinal ones. Conditions that cross specialty boundaries, or that require integrating findings from multiple domains, may fall into the gaps between specialists who are each doing their job without adequate awareness of what the others have found.
Handoffs and transitions are recognized high-risk moments in healthcare. When care is transferred from one physician to another — at the end of a shift, at discharge, between primary care and specialist — information is lost and assumptions are inherited. A working diagnosis that was appropriately provisional when it was first noted may be treated as established by the next physician who inherits it. The question marks disappear in the handoff.
Electronic health record design has introduced new sources of error while solving old ones. Copy-forward functionality perpetuates errors through successive notes. Templates that structure clinical documentation may cause physicians to document what the template expects rather than what the clinical situation requires. Alert fatigue — the tendency to dismiss the many automated warnings that EHR systems generate — can cause important signals to be ignored alongside the noise.
What the National Academy of Medicine Report Recommended
The 2015 report was notable not just for its diagnosis of the problem but for its insistence that diagnostic error must be treated as a system problem, not an individual failure. Its recommendations addressed the need for better education of physicians in diagnostic reasoning, improved communication between patients and clinicians, better use of diagnostic feedback to help physicians learn from errors, and greater engagement of patients as active participants in the diagnostic process.
That last recommendation — patient engagement — is the one most directly actionable by patients themselves. The report explicitly identified patients and their families as an underutilized resource in diagnostic accuracy: people who know their own history, know when something doesn't fit, and can ask the questions that surface the information that changes the clinical picture.
The report also called for a culture shift: away from treating diagnostic error as a shameful failure and toward treating it as a predictable consequence of a complex cognitive and systemic process — one that can be mitigated through better design, better communication, and better use of the full range of people and tools involved in care.
What Patients Can Do: Practical Protection Against Diagnostic Error
The following steps are grounded in the research on what actually produces diagnostic errors and what reduces them. They are not about distrusting physicians — they are about being an active participant in a process that benefits from active participation.
Don't accept a diagnosis you don't understand. A diagnosis that has been explained and is comprehensible — that accounts for the symptoms, that follows logically from the findings, that the patient can describe in plain language — is a diagnosis that is more likely to be correct than one that is accepted without understanding. When a diagnosis doesn't make sense to the patient, that dissonance is worth exploring rather than dismissing.
Ask what else it could be. This is the question that directly addresses premature closure and anchoring bias. Asking a physician "what other diagnoses did you consider, and what makes you more confident in this one than those?" invites a discussion of the differential that would otherwise remain implicit. A physician who can answer this question clearly and specifically is one who has done the cognitive work to arrive at the diagnosis. A physician who dismisses the question may not have.
Request a complete differential diagnosis. A differential diagnosis is the list of conditions that could explain a patient's presentation, ranked by likelihood. In medical training, developing and working through a differential is the foundation of diagnostic reasoning. In the rushed reality of clinical practice, the differential is often implicit rather than explicit — the physician has a leading hypothesis and may not have articulated the alternatives. Asking specifically what the differential includes makes the implicit explicit and creates an opportunity to discuss whether it's complete.
Ask about findings that don't fit. If there are symptoms, results, or aspects of the clinical picture that the proposed diagnosis doesn't fully account for, those deserve attention rather than dismissal. Asking directly — "this symptom doesn't seem to fit this diagnosis; what do you think explains it?" — surfaces the question that premature closure might otherwise bury.
Provide a complete, organized history. The history is the most important input into diagnostic reasoning, and it is the input most subject to distortion by cognitive bias. A patient who provides a clear, organized, chronological account of their symptoms — including the ones that seem unrelated — gives the physician more to work with and makes it harder to overlook findings that should change the clinical picture. Written timelines, symptom journals, and organized records all serve this function.
Don't let prior diagnoses foreclose new thinking. Once a diagnosis is in the medical record, it tends to persist — and to shape how subsequent physicians interpret new information. Patients who have a prior diagnosis and are developing new symptoms should be aware that the prior diagnosis may be causing the current symptoms to be interpreted in its light. Asking whether the current presentation could represent something new, or a change in the underlying condition, keeps the diagnostic question open.
Get a second opinion for serious conditions. The evidence on second opinions in serious diagnoses is clear: meaningful proportions of patients receive different or refined diagnoses when another expert reviews the case. For conditions with significant treatment implications — cancer, rare diseases, neurological conditions, conditions that are not improving with treatment — the standard of due diligence includes seeking an independent evaluation.
Consider a medical navigator for complex situations. For patients whose situations involve multiple providers, a long diagnostic history, or a diagnosis that doesn't seem to fit, a physician-led medical navigator who can review the full record — comprehensively, without the assumptions of any prior conclusion — provides the kind of fresh-eyes analysis that individual appointments can't deliver. The navigator's job is to ask the questions that time pressure, anchoring bias, and fragmented records have prevented from being asked, and to connect the findings that haven't yet been connected.
The System Works Better When Patients Engage It
The National Academy of Medicine report was explicit on this point: patients who are engaged — who ask questions, who provide organized histories, who seek second opinions, who challenge diagnoses that don't fully explain their symptoms — produce better diagnostic outcomes than patients who are passive.
This is not because patients know more than physicians. It is because patients know things that physicians don't — about their own bodies, their own histories, the ways in which their current situation doesn't match the explanation they've been given. When that knowledge is brought into the diagnostic process, it changes the information available, and better information produces better diagnoses.
The cognitive biases that produce diagnostic errors are a feature of human cognition, not a character flaw. The systemic pressures that amplify them are features of a healthcare system under strain. Neither can be eliminated. Both can be mitigated — by physicians who practice with awareness of their own cognitive vulnerabilities, and by patients who understand the process well enough to participate in it actively.
That understanding is protective. And it is available to any patient willing to ask the right questions.
———————
If you or a loved one is concerned that a diagnosis may be incomplete or incorrect, Pilot Rock Medical Navigators can help review the full picture with fresh eyes. Book a free 15-minute introductory call to discuss your situation. Learn how Pilot Rock can help →