Abstract
<title>Abstract</title> <p>Purpose. Randomized trials of endovascular therapy (EVT) for stroke due to medium-vessel occlusion (MeVO) are discordant and did not share a common dichotomous primary endpoint. We reconciled them by converting each arm's full 7-level modified Rankin Scale (mRS) distribution into expected utility. Methods. We performed a trial-specific, non-pooled utility-weighted decision analysis of ORIENTAL-MeVO, DISTAL, and ESCAPE-MeVO. Arm-level mRS counts were derived deterministically from the published intention-to-treat distributions and verified against reported anchor counts. The utility-weighted mRS difference (EVT minus control) was estimated by Dirichlet probabilistic sensitivity analysis with common random numbers across utility mappings. Symptomatic intracranial hemorrhage (sICH) and mortality were parallel safety outcomes. Results. The direction of the point estimate differed across trials. In ORIENTAL-MeVO the utility-weighted mRS difference was positive (+ 0.054; 95% credible interval [CrI] − 0.007 to + 0.115; probability of positive net benefit 0.96), whereas in DISTAL (− 0.014; 95% CrI − 0.071 to + 0.042; probability 0.31) and ESCAPE-MeVO (− 0.034; 95% CrI − 0.092 to + 0.024; probability 0.13) the point estimates were negative. All three credible intervals crossed zero. Direction was stable across utility mappings and aligned with baseline severity and thrombolysis exposure. sICH was more frequent with EVT. Conclusion. The net functional benefit of MeVO EVT is not uniform; the most favorable point estimate occurred in the higher-severity, lower-thrombolysis trial population. A favorable dichotomous result may overstate benefit once the full disability distribution and safety are considered, supporting severity- and selection-based decision-making.</p>