Missed and delayed diagnosis: how diagnosis fails and how causation is proved

Diagnosis is the part of medicine most exposed to hindsight. Once the final answer is known — the cancer, the dissection, the abscess — every earlier encounter can be read as a missed opportunity, and the symptom that seemed ordinary at the time appears, in retrospect, to have been a signal. A fair review must resist that distortion in both directions. It must not treat every missed diagnosis as negligence, and it must not treat the difficulty of diagnosis as an excuse for a process that skipped steps any reasonable clinician would have taken.

Missed and delayed diagnosis is also where the legal element of causation does most of its work. Many diagnostic-error cases are decided not by whether the diagnosis was late, which is often conceded, but by whether diagnosing it earlier would probably have changed what happened to the patient. This chapter explains how diagnosis fails, which conditions dominate serious claims, how causation is argued in California, and what documentation and systems prevent harm.

What a diagnostic error is

The National Academies of Sciences, Engineering, and Medicine, in Improving Diagnosis in Health Care (2015), defined diagnostic error from the patient's perspective as the failure to establish an accurate and timely explanation of the patient's health problem, or to communicate that explanation to the patient. The definition has three parts that matter for review: accuracy, timeliness and communication. A correct diagnosis that is never communicated, or a correct diagnosis reached months after it reasonably could have been, is within the definition.

The definition is a patient-safety concept, not a legal one. A diagnostic error in the National Academies' sense is not necessarily negligence. The legal question is whether the clinician's process fell below what a reasonably careful clinician would have done in the same circumstances, and whether that departure caused injury. Some diagnoses are missed by careful clinicians because the disease presented atypically or the tests that would have revealed it were not reasonably indicated.

How the diagnostic process breaks

Diagnosis is a process, not a moment. It begins with the patient's account and ends with a plan that the patient understands and that someone follows through. Each step can fail:

Studies of closed malpractice claims locate most failures early in this sequence and in follow-up. In an analysis of 181 ambulatory claims involving harmful diagnostic errors, the most frequent breakdowns were failure to order an appropriate test (55%), failure to create a proper follow-up plan (45%), failure to obtain an adequate history or examination (42%) and incorrect interpretation of a test (37%). Most errors involved more than one breakdown and more than one contributing factor (Gandhi and colleagues, 2006).

Cognitive and system contributors

Two kinds of contributors are usually present together. Cognitive contributors include premature closure (accepting a diagnosis before it is verified), anchoring on an initial impression, attributing new symptoms to a known condition, and failure to revisit a working diagnosis when the course does not fit. In the large insurer database analysis described below, clinical-judgment factors contributed in about 86% of high-severity diagnostic-error cases.

System contributors include short visits for complex problems, results that route to the wrong inbox, referral processes without tracking, copied-forward documentation that hides what was actually assessed, fragmented records across organizations, and the absence of a mechanism for patients to report that they are not improving. The legal significance of system contributors is that they can shift some or all responsibility toward an institution, as explained in hospitals and institutions and institutional technology.

The conditions that dominate serious claims

A study of 11,592 diagnostic-error cases from an insurer-contributed closed-claims database (2006–2015) found that three groups — vascular events, infections and cancers — accounted for 74.1% of the 7,379 high-severity cases: cancers 37.8%, vascular events 22.8% and infections 13.5%. The most frequent single disease in each group was lung cancer, stroke and sepsis (Newman-Toker and colleagues, 2019). A later national estimate by the same research group, extrapolating from these disease groups, put serious harms from diagnostic error in the United States at roughly 795,000 per year, with a plausible range of about 598,000 to 1,023,000 (BMJ Quality & Safety, 2024). Estimates of this kind rest on modeling assumptions and should be cited with their ranges.

The conditions that recur in claims share a pattern: they are time-sensitive, they can present with common and nonspecific symptoms, and an effective intervention exists if they are recognized in time. A general list, for orientation only:

This chapter does not provide diagnostic criteria or treatment advice. The clinical standard for each condition is established in a given case through expert testimony and the guidelines in use at the time of the care — not the guidelines in use when the case is tried.

Reconstructing what was knowable, and when

The central evidentiary task in a diagnostic-error case is to reconstruct the clinical picture as it appeared to the clinician at each decision point, without the final diagnosis. A disciplined review does the following:

1. Builds a timeline of information, not just of visits: when each symptom was reported, when each result was available, when each message arrived and when each was seen. The records chapters explain how audit trails and timestamps establish this.
2. Identifies the decision points — the moments where a different reasonable choice was available: an order not placed, a referral not made, a result not acted on, a patient sent home.
3. Asks what a reasonable clinician would have done at each point given only what was then known, including the patient's own reports and any prior results.
4. Separates the clinician's conduct from the patient's. Missed appointments, declined tests and incomplete histories are relevant, but so is whether the clinician explained the stakes. A declined recommendation without explanation of its risks raises the informed-refusal issue described in informed consent and Truman v. Thomas.
5. Identifies system contributors that shaped the clinician's choices.

Hindsight works on experts as well as juries. A careful expert addresses it directly by explaining why a finding was or was not significant at the time, and by acknowledging what remained uncertain.

How causation is argued in California

Proving that a diagnosis should have been made earlier is not enough. The plaintiff must also prove, to a reasonable medical probability, that the delay caused the injury claimed. California does not permit recovery for the loss of a less-than-probable chance of a better outcome as a substitute for that proof: in Bromme v. Pavitt (1992) 5 Cal.App.4th 1487, the Court of Appeal held that a wrongful-death plaintiff alleging delayed cancer diagnosis had to show that earlier diagnosis would more likely than not have prevented the death, and that a reduced chance of survival below that threshold did not suffice. The elements chapter explains the framework.

In practice, causation in delayed-diagnosis cases is argued through a few recurring lines of evidence:

• Disease state at the earlier date. For cancers, experts estimate the stage at the time diagnosis should have occurred, using tumor size, growth rates and the time between imaging studies. These estimates rest on assumptions — growth is not uniform — and both sides should state them.
• Treatment available at that date. What intervention would have been offered, and would the patient have been a candidate for it?
• Outcome with timely treatment. Survival or recurrence data for the earlier stage, applied to this patient's circumstances. Population statistics describe groups; experts must explain why the statistic fits the individual.
• A different injury than death. Even where the ultimate outcome would probably have been the same, the delay may have caused separately compensable harm: more extensive surgery, additional treatment, a period of avoidable pain, or loss of function. Identifying the precise injury is often what decides whether a delayed-diagnosis case is viable.
• Time-critical conditions. For vascular emergencies and infections, the question is often measured in hours: whether treatment within a window would probably have prevented the injury. Evidence about when the condition began, which is frequently contested, decides these cases.

Jennings v. Palomar Pomerado Health Systems (2003) 114 Cal.App.4th 1108 is a reminder that an expert's causation conclusion must be supported by a reasoned explanation connecting the departure to the injury; an opinion that states the conclusion without the mechanism will not carry the case. The expert witnesses chapter explains what such an opinion should show.

Documentation that helps and documentation that hurts

Documentation cannot substitute for a sound diagnostic process, but it is the only contemporaneous evidence of one. Records that tend to withstand review:

• A brief differential for presentations where a dangerous condition is plausible, stating why it was considered less likely or how it was excluded.
• Specific safety-netting: what the patient should watch for, when to return or seek emergency care, and confirmation that the patient understood.
• A clear follow-up plan: who will act on pending results, when the patient will be seen again, and what will happen if the patient does not return.
• Documentation of informed refusal when a recommended test or referral is declined, including the risk explained.
• Reassessment when a patient returns with the same or worsening complaint, rather than a repeated diagnosis.

Records that tend to hurt are not only those with omissions. A copied-forward examination that contradicts the visit reason, a "normal" review of systems that conflicts with the nursing intake, or a templated plan that does not mention the presenting problem invites the inference that the documented assessment did not occur. The clinical documentation page discusses how those inconsistencies are interpreted. Ambient AI drafting adds a new version of the problem — a fluent note that does not faithfully reflect the encounter — discussed in clinical AI accountability.

Prevention at the level of the system

Individual vigilance reduces diagnostic error; systems reduce it more reliably. Practices and hospitals that take diagnostic safety seriously tend to adopt:

• Closed-loop result management, with defined ownership, coverage during absence, and tracking of results until acknowledged and acted on;
• Referral tracking that shows whether a referral was scheduled and completed, with outreach when it was not;
• Structured reassessment triggers for unscheduled return visits and for patients who are not improving as expected;
• Patient engagement: telling patients every result, inviting them to report when they are not improving, and making it easy to do so;
• Learning from cases: reviewing diagnostic delays through protected quality processes, as described in peer review and mortality review, and measuring whether changes work.

Federal patient-safety resources, including the Agency for Healthcare Research and Quality's diagnostic safety program and its patient-safety primers, describe these approaches in more detail.

Three fictional cases, read from both sides

Each case below is fictional and is used only to show how the same facts are argued in both directions.

1. Rectal bleeding attributed to hemorrhoids. A 46-year-old reports intermittent rectal bleeding at two visits eight months apart; both notes record hemorrhoids, with no examination documented at the second visit and no referral for endoscopy. Fourteen months after the first visit, colorectal cancer is diagnosed at an advanced stage. Plaintiff: repeated bleeding without examination or evaluation was a departure; at the first visit the cancer was probably at an earlier, curable stage. Defense: the presentation was consistent with hemorrhoids and the first examination supported that; the cancer's stage fourteen months earlier is speculative; the patient missed a scheduled follow-up. What decides it: whether the second visit required reassessment rather than repetition, the documented missed appointment and whether outreach followed, and expert estimates of the stage at each earlier date with their assumptions stated.

2. Headache in an urgent care clinic. A patient with a sudden, severe headache is treated for migraine and discharged with general instructions. A subarachnoid hemorrhage is diagnosed two days later after a second bleed. Plaintiff: the history of sudden onset required evaluation for hemorrhage or emergency-department referral. Defense: the documented history described gradual onset consistent with the patient's prior migraines; urgent care lacked imaging and the instructions advised emergency care for worsening. What decides it: the precise history as documented at triage and by the clinician, whether they conflict, whether the instructions were specific, and whether earlier treatment would probably have prevented the second bleed and its consequences.

3. Back pain with fever. A patient with diabetes returns three times in a week with worsening back pain; a temperature is recorded at the third visit but not addressed in the note. A spinal epidural abscess is diagnosed after the patient develops leg weakness. Plaintiff: returning visits, worsening pain and fever in a patient with risk factors required consideration of infection and urgent imaging; earlier treatment would probably have prevented the neurologic deficit. Defense: back pain is common and the earlier visits had no red flags; the fever was low-grade; the deficit may have begun abruptly. What decides it: whether the third visit's findings made the dangerous diagnosis reasonably recognizable, who saw the recorded temperature, and the timing of the neurologic decline relative to when imaging should have occurred.

Sources

• National Academies of Sciences, Engineering, and Medicine. Improving Diagnosis in Health Care. Washington, DC: The National Academies Press; 2015. Report page.
• Gandhi TK, et al. Missed and delayed diagnoses in the ambulatory setting: a study of closed malpractice claims. Ann Intern Med. 2006;145(7):488–496. PubMed 17015866.
• Newman-Toker DE, et al. Serious misdiagnosis-related harms in malpractice claims: the "Big Three." Diagnosis (Berl). 2019;6(3):227–240. PubMed 31535832.
• Newman-Toker DE, et al. Burden of serious harms from diagnostic error in the USA. BMJ Qual Saf. 2024;33(2):109–120. PubMed 37460118.
• Saber Tehrani AS, et al. 25-Year summary of US malpractice claims for diagnostic errors 1986–2010. BMJ Qual Saf. 2013. PubMed 23610443.
• Bromme v. Pavitt (1992) 5 Cal.App.4th 1487: opinion (Justia reproduction); Jennings v. Palomar Pomerado Health Systems, Inc. (2003) 114 Cal.App.4th 1108: opinion; Truman v. Thomas (1980) 27 Cal.3d 285: opinion (Justia reproduction).
• Agency for Healthcare Research and Quality, diagnostic safety and quality and the PSNet diagnostic errors primer.

General legal and clinical education; not individual legal or medical advice.