Long-term restricted mean survival time survival of anterior strokes treated with endovascular thrombectomy

IntroductionEndovascular thrombectomy (EVT) has revolutionized the treatment of patients with large vessel occlusion (LVO) stroke. Three-month functional independence ranges between 33 and 60%, and mortality between 12 and 25%. Few studies have presented outcomes beyond five years and their reported odds or hazard ratios are difficult to apply clinically. Median survival is also challenging to interpret correctly. Restricted mean survival time (RMST) is a clinically intuitive metric that gives survival times up to specified cutoff points. We aimed to determine RMST at one, two and five years after EVT for anterior circulation stroke patients and evaluate survival impact of various risk factors. MethodsThis cohort study examined survival of patients treated with EVT for anterior circulation LVO stroke from 2011-2024 from the New Zealand national stroke database. Unadjusted RMST was determined for all age brackets. Cox regression adjusted survival curves were used to determine RMST and survival probabilities at one, two and five years for significant covariates. ResultsThere were 1457 anterior circulation LVO patients treated with EVT included for analysis (median follow-up 8.9 years [95% CI 7.57 - inf]). At five years, 415 (28%) patients had died. Patients aged 90 years or older had a RMST of 1.64 years, and a five-year survival probability of 33% compared with a 97% survival probability for those aged 15-39 years. Five-year RMST survival differences were; independent level of function versus dependent (0.77 years), low comorbidity versus high (0.71 years), mild-moderate NIHSS versus higher (0.49 years), smaller infarct cores versus large (0.48 years), and good reperfusion versus (0.67 years). DiscussionRMST provided clinically useful survival estimates and subgroup survival differences after EVT, especially in those of advanced age. RMST is easy to communicate, captures survival nuances, and can provide absolute differences that are not possible when outcomes are measured using median survival alone.