Clinical prediction of 72-hour early neurological deterioration and exploratory 90-day outcomes after endovascular or medical therapy in large-artery stenosis stroke: a single-center retrospective cohort study

Early neurological deterioration (END) after acute ischemic stroke is clinically important and is associated with worse functional outcomes. Patients with intracranial or extracranial large-artery stenosis may be vulnerable to early worsening because stenotic disease can be associated with impaired perfusion, unstable collateral compensation, and recurrent ischemic injury. However, clinical prediction of END in this population remains uncertain, and the 90-day outcome profile of patients with END treated by stenosis-directed endovascular therapy or medical therapy requires cautious evaluation. To evaluate clinical predictors of 72-hour END in acute ischemic stroke attributed to intracranial or extracranial large-artery stenosis and to describe exploratory 90-day functional and safety outcomes among patients with END managed with endovascular therapy or medical therapy. This single-center retrospective cohort study analyzed 200 de-identified patient records from Anqing First People’s Hospital of Anhui Medical University. END was defined as an increase of at least 4 points in the National Institutes of Health Stroke Scale within 72 h. A multivariable clinical logistic-regression model was developed to estimate END risk using available demographic, vascular, neurological, and timing variables. Model performance was assessed using discrimination, calibration, Brier score, and bootstrap optimism correction. Among patients with END, 90-day modified Rankin Scale outcomes were analyzed using the full ordinal distribution. Because no patient in the END cohort achieved mRS ≤ 2, binary favorable-outcome modelling was not performed. Propensity-score diagnostics were used to evaluate measured covariate balance between endovascular and medical-therapy groups. Safety outcomes included symptomatic intracranial hemorrhage, mortality, and recurrent ischemia. END occurred in 126 of 200 patients (63.0%). Among patients with END, 83 underwent an endovascular procedure and 43 received medical therapy. Endovascular procedures consisted of balloon angioplasty, stenting, or balloon angioplasty plus stenting. The clinical END prediction model had an apparent AUC of 0.690 and an optimism-corrected AUC of 0.607, indicating limited-to-fair internally corrected discrimination. The apparent Brier score was 0.210 and the optimism-corrected Brier score was 0.240. In the END cohort, no patient had mRS ≤ 2 at 90 days. In the adjusted ordinal mRS model, endovascular therapy was not associated with a statistically detectable shift toward worse or better 90-day mRS distribution compared with medical therapy (OR for worse mRS, 0.97; 95% CI, 0.49–1.93; p = 0.933). Safety outcomes were numerically similar between groups, with wide confidence intervals. In this single-center cohort of large-artery stenosis-related ischemic stroke, END was frequent and associated with poor 90-day functional status. A clinical prediction model showed limited-to-fair internally corrected discrimination and requires external validation before clinical use. Among patients with END, exploratory adjusted analyses did not show a clear association between endovascular therapy and ordinal 90-day mRS outcome compared with medical therapy. The treatment comparison should be interpreted cautiously because of retrospective treatment selection, limited sample size, incomplete procedural timing data, and potential residual confounding.