Enhancing generalizability of stroke clinical trial results: Illustrations from upper-limb motor recovery trials.

Abstract

Broadening eligibility criteria has been a focus to increase the generalizability of trial findings. Using upper-limb motor trials conducted early post-stroke as the illustrative domain, we sought to (1) investigate whether the published aim and conclusion statements adequately reflect the generalizability of findings and (2) explore internal validity and feasibility as constraints to achieving generalizability. We systematically applied a conceptual model of a trial sampling process to published literature from systematic review and prospective cross-sectional data. The eligibility criteria reported and used to exclude patients were classified by consensus as related to safety, internal validity, feasibility, or a combination thereof. Categorical data were reported as counts/proportions, and continuous data were reported as median (interquartile range (IQR)). Thirty trials (n = 1638 participants) were included in the published literature and 1013 patients in the prospective data set. Thirty-seven percent of trials did not describe their target population in the aim and conclusion, and 80% did not report all trial screening data. Eligibility criteria related to internal validity were the most common type reported and applied to exclude patients across both data sets. In the prospective data set, 70% of patients were excluded for more than one reason. Key information to support the generalizability of trial findings was insufficiently reported in published upper-limb motor research conducted early post-stroke. Broadening eligibility criteria alone is unlikely to sufficiently improve trial inclusivity due to internal validity constraints. Trials could achieve inclusivity through targeting multiple sub-populations, that in combination, produce clinically relevant results that are applicable to a broader population.