Development and internal validation of a clinical prediction model for 1-year recurrence after first-ever ischemic stroke

Objective: Ischemic stroke (IS) recurrence remains a major contributor to poor prognosis, particularly within the 1 year after the index event. This study aimed to develop and internally validate a clinical prediction model for estimating the risk of 1-year recurrence after first-ever IS using routinely available admission data. Methods: We conducted a retrospective cohort study including consecutive patients with first-ever IS admitted to a tertiary hospital in China between January and December 2021. The primary outcome was IS recurrence within 1 year after discharge. Missing predictor data were handled using multiple imputation by chained equations. A multivariable logistic regression model was developed in the full cohort using six predictors: National Institutes of Health Stroke Scale (NIHSS) score, age, admission systolic blood pressure, uric acid, apolipoprotein A1, and neutrophil percentage. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), bootstrap internal validation with 1,000 resamples, calibration intercept, calibration slope, Brier score, and decision curve analysis. Results: Among 738 eligible patients, 96 (13.0%) experienced IS recurrence within 1 year. Higher NIHSS score, older age, higher uric acid, and higher neutrophil percentage were associated with increased recurrence risk, whereas higher apolipoprotein A1 was associated with reduced risk. Admission systolic blood pressure showed a borderline association with recurrence risk. The model demonstrated moderate discrimination, with an apparent AUC of 0.764 (95% CI 0.710-0.819). Bootstrap internal validation yielded an optimism-corrected AUC of 0.750. The bootstrap-corrected calibration intercept, calibration slope, and Brier score were 0.0058, 0.9354, and 0.0981, respectively. Decision curve analysis showed greater net benefit than the treat-all and treat-none strategies across most threshold probabilities from 0.05 to 0.60. Conclusion: We developed and internally validated a six-predictor clinical prediction model for 1-year recurrence after first-ever IS using routinely available admission variables. The model showed moderate discrimination and acceptable calibration after bootstrap correction. It may serve as a preliminary tool for risk stratification, but external validation is required before routine clinical application.