Document resource
Background Studies have been conducted to explore the potential predictive indicators of unfavorable outcomes in patients with acute ischemic stroke (AIS) caused by large vessel occlusion (LVO). However, few studies have proposed a comprehensive predictive model combined with clinical baseline data and ancillary examination before surgery.Method In a retrospective study, we collected data on 823 patients with AIS-LVO who had undergone endovascular therapy (EVT); 562 patients who achieved successful revascularization with complete clinical and prognostic information were incorporated into the study. Those patients with a 90-day modified Rankin Scale (mRS) score of 0–2 were defined as having a favorable outcome, while a score of 3–6 represented an unfavorable outcome or futile reperfusion. To build up a predictive model, we applied multivariate logistic regression stepwise backward selection to decide which factors are supposed to be the components of the predictive model. Final model validity was testified by the variance inflation factor test and the Hosmer-Lemeshow (HL) goodness of fit test. The ultimate efficacy was supported by an area under the curve (AUC) value in both training groups and validation groups.Results 562 patients were enrolled in our study and divided into the training group and verification group in a ratio of 7:3. Factors of baseline data with P<0.1 in univariate logistic regression analysis were enrolled as the potential risk variables to conduct stepwise backward selection. The model was constructed by eight variables; higher mRS score (adjusted OR (aOR) 93.64, 95% CI 12.05 to 727.82, P<0.01), age >80 years (aOR 91.11, 95% CI 1.36 to 6116.36, P<0.05), National Institutes of Health Stroke Scale (NIHSS) >14 (aOR 0.15, 95% CI 0.02 to 0.99, P<0.05), operation history (aOR 8.13, 95% CI 1.32 to 50.20, P<0.05), creatinine (aOR 1.10, 95% CI 1.04 to 1.17, P<0.01), and neutrophil count (aOR 1.07, 95% CI 1.01 to 1.13, P<0.05) were associated with poor outcomes.Conclusion We established an estimation model for invalid reperfusion in AIS-LVO patients and constructed the nomogram for individualized predictions. The AUC of the training group and validation group were both 0.96, with excellent HL and decision curve analysis, presenting excellent clinical prediction efficiency and application potential.