Delayed Stroke Diagnosis: What the Timestamps Cannot Show

A delayed stroke diagnosis is the rare medico-legal claim in which the disputed interval is the easiest thing in the file to reconstruct. Triage time, order entry (the timestamped moment a clinician entered an order in the record), the radiology read, the time a drug reached the vein: all of it machine-generated, all of it in evidence, all of it agreeing with itself. None of it records whether anyone at the bedside was considering a stroke. That question is settled earlier, by a different document, and often by the absence of one. In a one-year, two-hospital review, 22 percent of ischemic strokes were missed at first emergency-department contact, and 40 percent of those missed carried no charted neurological examination containing the elements of the National Institutes of Health Stroke Scale (NIHSS, the bedside score for stroke severity: 0 is no detectable deficit, 42 the maximum, most treated strokes between 4 and 20), against 8 percent of the strokes diagnosed correctly.¹ The delay the timestamps describe is the last event in that sequence, not the first.

The Miss Comes Before the Clock

Stroke comes in two forms that share a name and little else: in an ischemic stroke a clot blocks an artery and the brain it feeds begins to die; in a hemorrhagic stroke a vessel ruptures and blood enters the brain. Everything below concerns the ischemic form, and the distinction is why the first study in any stroke pathway is a plain CT: the drug that dissolves the clot is catastrophic in a bleed. The intervals argued in these files concern tissue not yet dead but living on marginal flow, which a companion primer works through.

Posterior circulation strokes (the vertebral and basilar territory, supplying the brainstem and cerebellum and therefore balance, eye movement, swallowing and coordination) are missed at more than twice the rate of anterior circulation strokes (the carotid territory, supplying movement, sensation and language), and the mechanism sits in the scale. It omits gait and truncal ataxia (unsteadiness of walking and of sitting), dysphagia (difficulty swallowing) and cough, the deficits a brainstem or cerebellar stroke produces first, which is why posterior strokes score roughly half what anterior ones do. What they present with instead is a different list: difficulty walking in 41 percent, dizziness in 34, vision changes in 31, and dysmetria (clumsiness of a limb on reaching for a target) in 29.¹ A score taken alongside an examination of gait, swallowing and coordination means what it says; one taken without them does not yet mean anything.

A transient ischemic attack, a stroke syndrome that resolves within about an hour and leaves no injury on imaging, sits inside those rates without being separated from them. What it leaves behind is recurrence risk concentrated in the days just after, and the 2026 guideline recommends outright that dual antiplatelet therapy (two clot-inhibiting drugs together, usually aspirin with clopidogrel) begin within 24 hours and run 21 days.² A missed transient ischemic attack is a delayed-diagnosis claim with no treatment window in it, and the record question is whether that 24-hour decision was reached at all.

What makes a miss a case rather than an error is what it costs: a third of the missed strokes presented inside the three-hour window for intravenous thrombolysis (the clot-dissolving drug given into a vein), a further tenth between three and six hours from onset.¹ Part II of the primer works through what that window is worth once the stroke has been called, and the four ways the decision comes out.

Imaging does not rescue the miss where the miss happens. Among patients presenting with dizziness alone to a specialist neurological emergency service, a quarter had an acute ischemic stroke, and non-contrast computed tomography (a plain CT with no dye, whose job in the first hour is to exclude bleeding rather than find the infarct, the tissue already dead) caught about half of them, against nearly all on MRI.³ A normal head CT is not a negative stroke workup, and it is not an examination.

None of this interprets itself. The alternatives are few and checkable in minutes: low blood sugar, weakness after a seizure, migraine, an inner-ear disturbance. What the chart has to show is not that the clinician was right but that they were named and separated, by someone identified, at a time recorded.

Treatable Names Three Different Cases

Three things get called treatment in a stroke file, each with its own clock, imaging requirement and eligible population: a drug into a vein, the same drug late on the strength of a perfusion scan, and a catheter procedure that pulls the clot out. An analysis taking treatable as one category has answered a question no trial was designed to ask.

Intravenous thrombolysis is a drug, recommended outright within 4.5 hours of symptom onset or last known well (the last moment the patient was observed at baseline, which starts the clock when onset was unwitnessed). Where automated perfusion imaging (a scan mapping blood flow to separate tissue already dead from tissue still at risk) shows salvageable tissue, the 2026 American Heart Association and American Stroke Association guideline extends the drug past 4.5 hours, out to 24 in the largest occlusions, but only for patients who cannot receive thrombectomy.² That second window is absent from most delayed-diagnosis analysis; Part I of the primer carries the agents and the arc that moved it.

The decisive criterion inside the window is not the score but whether the deficit was disabling for this patient. The guideline states that the score alone does not suffice, gives leg weakness severe enough to prevent walking, which scores 2, as its example, and supplies a test: could the patient still perform basic activities of daily living (washing, dressing, walking, eating) or return to work if the deficit persisted.²

Mechanical thrombectomy is a catheter procedure: a device threaded from the groin or wrist into the brain to withdraw the clot. It applies only to large-vessel occlusion (LVO, a clot in one of the major arteries at the base of the brain rather than in their small branches), roughly 10 to 15 percent of acute stroke patients, and eligibility turns on the occluded segment, the severity score, how independent the patient was beforehand, and ASPECTS (the Alberta Stroke Program Early CT Score, a 10-point score docking a point for each region already injured on the plain CT, so a lower number means more brain lost). The segment is the line a reader can check: the procedure is recommended for the trunk of the middle cerebral artery (M1) and its dominant large branch (M2); past that point the guideline states affirmatively that trials show no benefit within 6 hours.² The posterior circulation carries its own recommendation and its own thresholds: for basilar artery occlusion the guideline recommends the procedure within 24 hours where the patient was independent beforehand, the severity score is 10 or above, and PC-ASPECTS (the same 10-point scoring applied to the posterior territory) is 6 or above.² A posterior circulation stroke can therefore be missed because its score is low and, on that same threshold, fall outside the recommendation that would have treated it, in a territory where the mean score at presentation was 3.8.¹ Every vessel-imaging report names the segment, and a causation theory that treats any visible occlusion as treatable fails at that line.

The Benchmarks Moved, and Each One Belongs to Its Edition

Edition currency matters for a narrower reason than the standard of care itself: an expert quoting a superseded edition has shown the check was never run, which goes to the opinion rather than the defendant. The editions differ on the points these cases turn on, and the window itself moved three times across them, so the standard is fixed by the date of care, not the document in force now.

The time targets moved too, and they measure different things. Forty-five minutes was never an acquisition target. It was the 2013 target for interpreting the scan, set alongside the 60-minute door-to-needle recommendation (arrival to the moment the drug is given) in the same edition.⁴ The 2026 guideline carries neither figure, only a recommendation that hospitals track the metrics, so a benchmark quoted from it for either interval comes from a document that does not contain one.²

Both figures are system targets with a compliance threshold, not per-patient deadlines, so a single case that exceeded either is not on the face of the recommendation a departure. Sixty minutes remains the only door-to-needle figure any edition ever set, and a population target does not excuse missing it: an expert calling 90 minutes unremarkable in 2019 must reckon with it.

Speed still carries weight: faster treatment tracks lower in-hospital mortality and a greater chance of going home, incrementally and without a threshold,⁵ which is why a fifteen-minute dispute is rarely worth litigating and a two-hour one usually is.

The interval that actually runs is neither of the ones usually pleaded. Among registry patients transferred between hospitals, mean door-in-door-out time (arrival at the first hospital to departure for the second) was 171 minutes: door-to-imaging (arrival to the first scan) took 18 of them and imaging-to-departure 153.⁶ Getting the patient into the scanner is rarely the bottleneck. Getting the patient out of the building is.

The Decisions the Chart Has to Document

From triage (the nurse's first assessment, which sets the priority level) onward, six decisions generate a record, each documented as it happens or not at all. A step that produced no document is not a gap in the file. It is the finding.

• The triage disposition and the stroke-code activation. The stroke code (a single page mobilizing the team in parallel rather than in sequence) generates an activation log with a time on it; where none appears, the triage note's record of the complaint is the record of the judgment.
• The neurological examination, and the time on it. Not a score alone. The scale has eleven items, among them gaze, visual fields, coordination, sensation, language and neglect (a failure to attend to one side of the world); a note recording strength alone does not contain them.
• What fixes last known well, and whether the sources agree. The clock may be set by an emergency-services run sheet (the ambulance crew's record of the call, usually the earliest timed document in the file), a triage note, a family account or a nursing entry, which disagree more often than a complaint assumes. A chart carrying one and no reconciliation has fixed it by default.
• Imaging, in the order it was obtained and with the edition-appropriate target named. Acquisition and interpretation live in separate fields, the imaging system holding one time and the report another, and a benchmark quoting one figure for both belongs to neither. CT angiography (a CT taken with dye in the arteries, which shows where a clot sits) identifies the occlusion, but the guideline states that non-contrast CT suffices for the thrombolysis decision in most patients and that multimodal imaging may delay it.²
• The transfer decision, and who owned it. Stroke centers are certified in tiers (a primary center gives the drug but cannot perform the catheter procedure; a thrombectomy-capable or comprehensive stroke center does both), and the tier fixes what the first hospital could do. The record has to show when transfer was raised, who accepted, when transport arrived, and whether the drug was given before departure. A telestroke consultation (neurological assessment by video link) leaves its own record and its own defendant.
• The stated reason treatment was or was not given. A quarter of registry patients arriving within two hours with no documented contraindication went untreated inside three hours, and a deficit called mild or improving is the most common reason given.⁷ What it is worth turns on the governing edition and on the year, since that proportion fell by roughly two thirds across the registry's own span, which Part II of the primer works through against the same data. These cases are most often lost on the conditions the guideline states are not contraindications: mild to moderate early ischemic changes on the CT, cerebral microbleeds (tiny old areas of bleeding visible only on MRI), prior antiplatelet treatment, and a disabling deficit persisting after a glucose derangement (a blood sugar too high or too low) is corrected. Blood pressure below 185/110 before the drug is given is a firm requirement.² A record naming a non-exclusion has recorded a clinical impression in the vocabulary of an exclusion.

Where the Record Fails

These claims fail in a few recurring ways, divided between what the chart did not record, the first three, and what the filing measured against the wrong thing, the last two.

• Dizziness triaged to a peripheral cause with no posterior-circulation examination charted. The complaint carrying the highest odds of a miss is worked as vestibular (an inner-ear cause), and the chart records the disposition without the examination.¹
• A low NIHSS entered as a mild stroke. The chart holds a number where it needed a statement of what this patient could no longer do.²
• A normal non-contrast CT read as excluding stroke. The study performed answers a different question from the one the chart treats it as answering.³
• A benchmark applied from the wrong edition. Forty-five minutes quoted as an acquisition target, or a mismatch requirement as an absolute in a case an earlier edition governed.
• The transfer clock measured at the wrong end. The pleading disputes the 18 minutes to imaging; the interval that consumed the window was the 153 to departure.⁶

The record failures are not separate failures. Each omission is the precondition for the next: under-triage produces no activation; no activation, no timed examination; no examination, a deficit that looks mild; a mild-looking deficit, no vessel imaging; no vessel imaging, an unnamed occlusion; and an unnamed occlusion produces the late transfer that closes the window. Only the last link leaves a timestamp, which is why the interval in the complaint is so often the one that was never the problem.

What the Evidence Will and Will Not Carry

Two bodies of evidence carry this subject, and each stops somewhere an opinion has to stop with it.

The recognition data is strong on direction and weak on rate. The review behind the record-level findings is retrospective and exposed to hindsight bias (the tendency, once the answer is known, to judge the earlier decision as more obviously wrong than it was).¹ What survives the discount is the record-level contrast: whether a charted examination and an activation log exist is not a matter of interpretation.

The late-window thrombectomy trials established a selection rule, not a duration. They enrolled on a mismatch between the size of the deficit and the tissue already lost, so an opinion reading them as authority for treating late, without the imaging that selected them, has read them backwards. That gate is itself contested: the 2026 guideline dropped the mismatch requirement on later trials showing benefit despite large established cores.

Reading a Delayed Stroke Diagnosis in Context

Several records that decide these cases are not produced by a routine chart request and must be asked for by name: the emergency-department tracking log (the board recording each patient's location and times); the medication administration record (the nursing log of every drug given, with times), where the actual needle time lives; and the hospital's own registry abstraction, stating its door-to-needle time before any claim existed.

A defensible review of a delayed stroke diagnosis, plaintiff or defense, works through a small set of specific questions.

• Which documents fix last known well, and do the run sheet, the triage note, the family account and the nursing entry agree with one another?
• Does the activation log carry a time, and does it fall before or after the CT? Where none appears, what did the triage note record as the complaint?
• Does any note carry the elements of the scale, gaze, fields, language, strength, coordination, sensation and neglect, timed and attributed to a named clinician, and where the complaint was dizziness or unsteadiness, does it also record gait tested on walking and swallowing, which the scale omits?
• Where thrombolysis was withheld because the deficit was called mild or improving, does the record state what this patient could no longer do, or only what it scored?
• Does the recorded reason for withholding the drug name something the governing edition treats as an exclusion, or something it expressly does not, and does a blood pressure above 185/110 appear in the nursing vital signs?
• What time was the CT angiogram acquired, what time was the drug given on the medication administration record, and which came first?
• Which edition was in force on the date of care, and does the benchmark in the expert report come from it?
• For a case argued on thrombectomy, does the vessel-imaging report name the occluded artery and its segment, and does any document state an ASPECTS or a core volume in milliliters (the measured volume of brain already dead)?

Placed in order, those elements usually settle which of the three cases this is, and that does more work than any interval in the file. A claim that reads strongly on the complaint dissolves against a chart carrying a timed examination, a disabling-deficit assessment and an exclusion the patient in fact had. The defense collapses against a nursing note moving last known well two hours earlier than the triage entry, or an activation log that begins after the CT.

The timestamps show how long it took. The chart shows whether anyone was looking. The document that answers the second question is the timed neurological examination, and in most of these files it is the one nobody thought to ask for.

References

Footnotes

1. Arch AE, Weisman DC, Coca S, Nystrom KV, Wira CR, Schindler JL. Missed ischemic stroke diagnosis in the emergency department by emergency medicine and neurology services. Stroke. 2016;47(3):668–673. doi:10.1161/STROKEAHA.115.010613 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

2. Prabhakaran S, Gonzalez NR, Zachrison KS, et al. 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke. 2026;57(8):e316–e436. doi:10.1161/STR.0000000000000513 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

3. Hu X, Liu S, Wu X, et al. Prevalence of stroke and diagnostic performance of emergency MRI in acute isolated dizziness. Ann Clin Transl Neurol. 2025;12(12):2514–2522. doi:10.1002/acn3.70195 ↩ ↩²

4. Jauch EC, Saver JL, Adams HP Jr, et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2013;44(3):870–947. doi:10.1161/STR.0b013e318284056a ↩

5. Saver JL, Fonarow GC, Smith EE, et al. Time to treatment with intravenous tissue plasminogen activator and outcome from acute ischemic stroke. JAMA. 2013;309(23):2480–2488. doi:10.1001/jama.2013.6959 ↩

6. Royan R, Stamm B, Giurcanu M, Messe SR, Jauch EC, Prabhakaran S. Emergency department process times and door-in-door-out times in interhospital transfers after acute ischemic stroke. JAMA Netw Open. 2024;7(9):e2431183. doi:10.1001/jamanetworkopen.2024.31183 ↩ ↩²

7. Messé SR, Khatri P, Reeves MJ, et al. Why are acute ischemic stroke patients not receiving IV tPA? Results from a national registry. Neurology. 2016;87(15):1565–1574. doi:10.1212/WNL.0000000000003198 ↩