Wake-Up Symptoms Linked to Delays in In-Hospital Stroke Diagnosis

A single-center retrospective study of 183 hospitalized patients evaluated for suspected in-hospital stroke found that symptoms first recognized after waking were associated with substantially longer diagnostic delays, while persistent and FAST-positive neurological symptoms were associated with the presence of true stroke. The median interval from last known well or symptom onset to an imaging-based final diagnosis was 295 minutes, and 77 patients, or 42.1%, were ultimately diagnosed with stroke. Among those cases, 70 were ischemic strokes, five were intracerebral hemorrhages and two were subarachnoid hemorrhages.

These associations offer clues for identifying high-risk cases, but the study did not test whether any particular triage system improves outcomes. The analysis, conducted at Fujita Health University Hospital, a 1,376-bed tertiary academic center serving the eastern part of the Nagoya metropolitan area in Japan, included patients hospitalized outside the neurology and neurosurgery departments who developed new neurological symptoms within 24 hours of symptom onset or their last known well. Researchers screened head computed tomography and magnetic resonance imaging orders placed from January 2023 through January 2026, then reviewed medical records to identify patients whose symptoms had prompted clinical suspicion of stroke.

The cohort included 110 men and 73 women and had a median age of 77 years. Because imaging was ordered at the discretion of attending physicians and the hospital had no standardized institutional imaging protocol for suspected in-hospital stroke, the study captured clinical practice rather than a controlled workflow. Wake-up onset was defined as neurological symptoms first recognized after a patient awakened or after a prolonged interval without neurological assessment.

When onset was unknown, investigators used the last documented time at which the patient was known to be neurologically unchanged. The median interval from last known well or symptom onset to symptom detection was 120 minutes, with an interquartile range of 41 to 507 minutes, while detection to initial imaging took a median of 69 minutes, with an interquartile range of 39 to 151 minutes. After adjustment for age, sex and other selected variables, wake-up symptom onset was associated with 3.34 times the diagnostic time, with a 95% confidence interval of 2.61 to 4.28.

The adjusted estimate from last known well to imaging-based final diagnosis was 810 minutes for patients with wake-up onset. Most of this delay occurred before symptoms were detected: the adjusted detection interval was 549 minutes, compared with a median of 73 minutes in the unadjusted comparison of patients with and without wake-up onset. The association was therefore concentrated in the interval before the hospital team recognized that a neurological emergency might be occurring.

Initial activation of the medical emergency team, or MET, showed the opposite association. MET comprised physicians and nurses from the intensive care unit and was available around the clock to assist with inpatient emergencies. Its initial activation was associated with a shorter time to diagnosis, with an exponentiated regression coefficient of 0.71 and a 95% confidence interval of 0.56 to 0.91.

The result does not establish that MET activation caused the improvement, because the study was observational and was not designed to evaluate the team’s effectiveness. It nevertheless suggests that early recognition of a neurological emergency may coincide with faster imaging, neurological consultation and diagnostic decision-making. Nurses detected the first symptoms in most cases: 45 of 53 patients with wake-up onset and 105 of 130 without it.

Physicians were the first to detect symptoms in eight patients with wake-up onset and 13 without it. The findings point to a workflow in which nursing staff frequently serve as the first link between a new neurological change and an urgent hospital response. Imaging was both a major part of the diagnostic process and a source of variation.

Non-contrast CT was the first examination for 153 patients, CT angiography for four and MRI for 26. Of the 154 patients who underwent non-contrast CT first, 120 required additional imaging, including MRI in 119 and CT angiography in one. Overall, MRI was ultimately required for 145 patients, or 79.2%, to establish the final diagnosis.

Additional imaging was required for 120 of the 154 patients who started with non-contrast CT, and for 33 of 53 patients with wake-up onset and 87 of 130 without it in the descriptive comparison. The median time from symptom detection to an imaging-based final diagnosis was 127 minutes for patients with wake-up onset and 156 minutes for those without it in the descriptive analysis. These findings describe the hospital’s existing practice; the study did not directly compare CT-first, MRI-first or other imaging strategies.

Persistent neurological symptoms were independently associated with true stroke, with an odds ratio of 1.44 and a 95% confidence interval of 1.24 to 1.68. Symptoms were classified as persistent unless they had completely resolved before the imaging needed to establish the final diagnosis. FAST-positive symptoms, defined in the study as the presence of motor and/or speech symptoms, were also associated with true stroke, with an odds ratio of 1.28 and a 95% confidence interval of 1.12 to 1.46.

In separate adjusted analyses, motor symptoms had an odds ratio of 2.61, with a 95% confidence interval of 1.35 to 5.04, while speech symptoms had an odds ratio of 2.40, with a 95% confidence interval of 1.12 to 5.13. These findings indicate associations rather than stand-alone diagnostic rules, particularly in a population in which more than half of suspected cases were not confirmed as stroke. The broader pattern reflects the diagnostic challenge posed by neurological deterioration in hospitalized patients.

The subjects were admitted for other medical or surgical problems and often had comorbidities, recent surgery, altered consciousness or medication effects that could complicate symptom assessment. Transient symptoms were more frequent among patients without a final stroke diagnosis: 40 of 106 non-stroke cases, or 37.7%, compared with five of 77 stroke cases, or 6.5%. Atrial fibrillation was more common in the stroke group than in the non-stroke group, at 29.9% versus 13.2%, while maintenance hemodialysis was more common among patients without stroke, at 21.7% versus 10.4%.

A history of stroke was reported in 19.5% of the true-stroke group and 34.0% of the non-stroke group. The cause of neurological symptoms remained unclear in some patients without a final diagnosis of stroke, so the non-stroke group should not be treated as a uniform category of confirmed mimics. Reperfusion treatment was received by 17 of the 70 patients with ischemic stroke, or 24.3%.

Three received intravenous thrombolysis alone, 12 underwent endovascular therapy alone and two received both. Among the 61 patients who did not receive intravenous thrombolysis, 56, or 91.8%, had at least one contraindication; exceeding the thrombolysis time window was the most common contraindication. The median National Institutes of Health Stroke Scale score was 6 among the 66 ischemic-stroke patients with persistent symptoms and an available score, with an interquartile range of 3 to 15.

The study did not evaluate functional outcomes, and it did not establish that reducing diagnostic delay would increase reperfusion or improve recovery. Although CT angiography and CT perfusion were used relatively infrequently in the cohort, the findings do not show that more vascular imaging would have changed treatment or outcomes. Several limitations constrain how the findings should be interpreted.

The study covered one hospital, used retrospective medical-record timestamps and had missing variables, creating potential for selection bias and limiting generalizability. Electronic timestamps may not precisely represent symptom onset or detection. Screening imaging orders also meant that patients whose symptoms were overlooked or not considered suggestive of stroke could have been missed.

The analysis was restricted to patients outside stroke-specialized departments, so its results may not apply to dedicated stroke units or neurological and neurosurgical wards. In addition, the cause of symptoms was unresolved in some non-stroke cases, and the observational design cannot determine whether wake-up onset, MET activation, symptom persistence or FAST findings directly caused changes in diagnosis time or stroke identification. The sample was also limited for the number of candidate variables considered in regression models, and the investigators used stepwise selection to reduce model overfitting.

Persistent motor or speech symptoms were associated with true stroke, but their predictive value should be interpreted within the wider clinical context because 106 of the 183 patients, or 57.9%, did not receive a final diagnosis of stroke. The results support further evaluation of continuous neurological observation, nurse-led recognition, standardized imaging and integrated in-hospital stroke-response pathways. Larger multicenter studies would be needed to determine whether such systems can shorten diagnosis safely, increase access to reperfusion treatment and improve functional outcomes for patients who develop stroke while already hospitalized.

Subject of Research: Diagnostic delay and clinical predictors of true stroke among patients evaluated for suspected in-hospital stroke

Article Title: Diagnostic delay and predictors of true stroke among patients with suspected in-hospital stroke: a single-center retrospective study

Article References: Tanaka, K., Nakamura, K., Mori, Y., Koge, J., Hashimoto, T., Fujiwara, E., Kuwahara, K., Iwata, M., Nakamura, T., Hirose, Y., & Matsumoto, S. (2026). Diagnostic delay and predictors of true stroke among patients with suspected in-hospital stroke: a single-center retrospective study. Frontiers in Neurology, 17, Article 1979991. https://doi.org/10.3389/fneur.2026.1979991

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1979991

Keywords: in-hospital stroke, diagnostic delay, wake-up symptoms, medical emergency team, FAST symptoms, brain imaging, stroke diagnosis, reperfusion therapy, retrospective study, hospitalized patients, Diagnostic, delay