Characteristics of randomized trials focusing on stroke due to intracerebral hemorrhage systematic review

Abstract

Background and Purpose—The absence of treatments for intracerebral hemorrhage with significant consistent benefit in randomized controlled trials (RCTs) could be because of lack of treatment efficacy or the design of RCTs. Methods—We searched the Cochrane Stroke Group Trials Register in December 2015 for completed and published RCTs reporting clinical outcome in adults with intracerebral hemorrhage. We collected data on publication year and language, study characteristics, and effect size. We regarded RCTs to be at lower risk of bias if they performed ≥2 of describing randomization, using blinding, or specifying the primary outcome. We registered this systematic review (PROSPERO, international prospective register of systematic reviews CRD42016051103). Results—We found 136 eligible RCTs: 57% were phase II, 76% were single center, 98% studied acute treatments, 49% involved drug interventions, 24% were placebo-controlled, the primary outcome was death or disability in 30%, and median sample size was 77 (interquartile range, 47–160). Forty-six percent explained randomization, 24% blinded treatment allocation, and 24% specified the primary outcome such that 38 (28%) were at lower risk of bias. RCTs at lower risk of bias were more likely to use multicenter recruitment (adjusted odds ratio, 6.95; 95% confidence interval, 2.2–21.5) and be published in English (adjusted odds ratio, 12.9, 95% confidence interval, 2.7–62.5). RCTs with larger sample sizes were independently more likely to be phase III/IV (P<0.01) and use multicenter recruitment (P<0.01). RCTs at lower risk of bias had smaller pooled treatment effects on death/disability (P=0.02). Conclusions—Intracerebral hemorrhage RCTs have often been at high risk of bias, and these RCTs have been characterized by small sample sizes and larger effect sizes.