Distinguishing Clinical from Statistical Significances in Contemporary Comparative Effectiveness Research

Abstract

Objective: To determine the prevalence of clinical significance reporting in contemporary comparative effectiveness research (CER). Background: In CER, a statistically significant difference between study groups may or may not be clinically significant. Misinterpreting statistically significant results could lead to inappropriate recommendations that increase healthcare costs and treatment toxicity. Methods: CER studies from 2022 issues of Annals of Surgery, Journal of the American Medical Association, Journal of Clinical Oncology, Journal of Surgical Research, and Journal of the American College of Surgeons were systematically reviewed by two different investigators. The primary outcome of interest was whether authors specified what they considered to be a clinically significant difference in the Methods. Results: Of 307 reviewed studies, 162 were clinical trials and 145 were observational studies. Authors specified what they considered to be a clinically significant difference in 26 studies (8.5%). Clinical significance was defined using clinically validated standards in 25 studies and subjectively in 1 study. Seven studies (2.3%) recommended a change in clinical decision-making, all with primary outcomes achieving statistical significance. Five (71.4%) of these studies did not have clinical significance defined in their methods. In randomized controlled trials with statistically significant results, sample size was inversely correlated with effect size (r=−0.30, P=0.038). Conclusion: In contemporary CER, most authors do not specify what they consider to be a clinically significant difference in study outcome. Most studies recommending a change in clinical-decision making did so based on statistical significance alone, and clinical significance was usually defined with clinically validated standards.