Sample Size Estimates for Clinical Trials of Vasospasm in Subarachnoid Hemorrhage

Publisher Website
Google Scholar
Abstract
Background and Purpose— Clinical trials for prevention of vasospasm after aneurysmal subarachnoid hemorrhage (SAH) seldom have improved overall outcome; one reason may be inadequate sample size. We used data from the tirilizad trials and the Columbia University subarachnoid hemorrhage outcomes project to estimate sample sizes for clinical trials for reduction of vasospasm after SAH, assuming trials must show effect on 90-day patient-centered outcome. Methods— Sample size calculations were based on different definitions of vasospasm, enrichment strategies, sensitivity of short- and long-term outcome instruments for reflecting vasospasm-related morbidity, different event rates of vasospasm, calculation of effect size of vasospasm on outcome instruments, and different treatment effect sizes. Sensitivity analysis was performed for variable event rates of vasospasm for a given treatment effect size. Sample size tables were constructed for different rates of vasospasm and outcome instruments for a given treatment effect size. Results— Vasospasm occurred in 12% to 30% of patients. Symptomatic deterioration and infarction from vasospasm exhibited the strongest relationship to mortality and morbidity after SAH. Enriching for vasospasm by selection of patients with thick SAH slightly decreased sample sizes. Assuming β=0.80, α=0.05 (2-tailed) and treatment effect size of 50%, total sample size exceeds 5000 patients to demonstrate efficacy on 3-month patient-centered outcome (modified Rankin Scale). Conclusions— Clinical trials targeting vasospasm and using traditional patient-centered outcome require very high sample sizes and will therefore be costly, time-consuming, and impractical. This will hinder development of new treatment strategies.