BACKGROUND: We investigate whether the National Institutes of Health Stroke Scale (NIHSS) at 24 hours could serve as a primary outcome in acute ischemic stroke trials, and whether combining 90-day modified Rankin Scale (mRS) and 24-hour NIHSS in a hierarchical outcome could enhance detection of treatment effect, using endovascular treatment (EVT) as an exemplary study intervention. METHODS: This was a post hoc analysis of pooled data from 7 randomized EVT trials. Twenty-four-hour NIHSS as a surrogate outcome for 90-day mRS was assessed in a causal mediation model. A 7-point ordinal NIHSS score was generated by grouping 24-hour NIHSS, including death as a separate category ("ordinal" NIHSS). EVT effect sizes and sample sizes required for detecting EVT benefit with 80% power were compared when using granular 24-hour NIHSS, ordinal 24-hour NIHSS, 90-day mRS, and a hierarchical outcome (win ratio) that combines 90-day mRS and 24-hour NIHSS. RESULTS: A total of 1720 patients were included. Twenty-four-hour NIHSS mediated the association between EVT and 90-day mRS and met criteria for a useful surrogate outcome. Effect sizes were highest and sample sizes required to detect EVT benefit smallest for the win ratio approach (228), followed by 90-day mRS (240) and ordinal 24-hour NIHSS (242). In patients with baseline NIHSS <
10 and ≥25, ordinal 24-hour NIHSS resulted in the highest effect size. CONCLUSIONS: Twenty-four-hour NIHSS is a useful surrogate outcome for 90-day mRS in patients with acute ischemic stroke undergoing EVT, with a similar EVT effect size compared with 90-day mRS. It could potentially enhance detection of EVT benefit in patients with very low or high baseline NIHSS. An ordered hierarchical outcome could improve detection of EVT treatment effect.