Best (but oft-forgotten) practices: sensitivity analyses in randomized controlled trials. Issue 1 (16th December 2015)

• Record Type: Journal Article
• Authors: de Souza, Russell J; Eisen, Rebecca B; Perera, Stefan; Bantoto, Bianca; Bawor, Monica; Dennis, Brittany B; Samaan, Zainab; Thabane, Lehana
• Is Part Of: American journal of clinical nutrition. Volume 103:Issue 1(2016)
• Page Start: 5
• Page End: 17
• Publication Date: 2015-12-16
• Subjects: analysis -- commentary -- sensitivity -- statistics -- tutorial; Diet therapy -- Periodicals; Nutrition -- Periodicals; Dietetics -- Periodicals; 613.205
• DOI: 10.3945/ajcn.115.121848
• Languages: English
• ISSNs: 0002-9165

Abstract

ABSTRACT: A randomized controlled clinical trial is the best way to minimize bias in ascertaining treatment effects, but the credibility of the results of a trial depends on the validity of the methods used to analyze the data, and the conditions under which such methods produce valid answers. A sensitivity analysis is a method to determine the robustness of trial findings by examining the extent to which results are affected by changes in methods, models, values of unmeasured variables, or assumptions. The goal of a sensitivity analysis is to identify results that are most dependent on questionable or unsupported assumptions. In this article, we briefly review the current use of sensitivity analyses in a random sample of published nutrition literature and provide a guide on the use of sensitivity analyses in randomized trials as to when to consider them, what to consider when planning them, and different methods of implementing them. We propose an 8-step strategy for improving the approach to conducting and reporting sensitivity analyses in nutrition-based trials.