Pooling of Animal Experimental Data Reveals Influence of Study Design and Publication Bias

Abstract

Background and Purpose— The extensive neuroprotective literature describing the efficacy of candidate drugs in focal ischemia has yet to lead to the development of effective stroke treatments. Ideally, the choice of drugs taken forward to clinical trial should be based on an unbiased assessment of all available data. Such an assessment might include not only the efficacy of a drug but also the in vivo characteristics and limits—in terms of time window, dose, species, and model of ischemia used—to that efficacy. To our knowledge, such assessments have not been made. Nicotinamide is a candidate neuroprotective drug with efficacy in experimental stroke, but the limits to and characteristics of that efficacy have not been fully described. Methods— Systematic review and modified meta-analysis of studies of experimental stroke describing the efficacy of nicotinamide. The search strategy ensured ascertainment of studies published in full and those published in abstract only. DerSimonian and Laird random effects meta-analysis was used to account for heterogeneity between studies. Results— Nicotinamide improved outcome by 0.287 (95% confidence interval 0.227 to 0.347); it was more effective in temporary ischemia models, after intravenous administration, in animals without comorbidities, and in studies published in full rather than in abstract. Studies scoring highly on a quality measure gave more precise estimates of the global effect. Conclusions— Meta-analysis provides an effective technique for the aggregation of data from experimental stroke studies. We propose new standards for reporting such studies and a systematic approach to aggregating data from the neuroprotective literature.