Risk of bias in randomized controlled trials of psychological treatments for bulimia nervosa and binge eating

Author(s)

Trevizo, Juliana Pagotto ^[1] ; Hay, Phillipa ^{[2, 3]} ; Swardfager, Walter ^{[4, 5]} ; Cogo-Moreira, Hugo ^[1]

Affiliation

^[1] Univ Fed Sao Paulo, Dept Psychiat, BR-04039032 Sao Paulo - Brazil

^[2] Western Sydney Univ, Sch Med, Sydney, NSW - Australia

^[3] Western Sydney Univ, Translatinal Inst Hlth Res, Sydney, NSW - Australia

^[4] Univ Toronto, Dept Pharmacol & Toxicol, Toronto, ON - Canada

^[5] Sunnybrook Res Inst, Hurvitz Brain Sci Program, Toronto, ON - Canada

• Document type: Journal article
• Source: ANNALS OF EPIDEMIOLOGY; v. 28, n. 9, p. 625-628, SEP 2018.
• Web of Science Citations: 0

Abstract

Purpose: In the context of Cochrane systematic reviews/meta-analyses of randomized clinical trials, risk of bias (RoB) is assessed using categorical indicators (low, unclear, or high RoB). This study sought to evaluate the indicators of the Cochrane RoB tool available for construct validity as applied to randomized clinical trials of psychological treatments for bulimia nervosa and binge eating. Methods: Bayesian confirmatory factor analysis was used to test the construct validity of the measurement model underlying the set of five categorical items, and the reliability of these indicators to measure RoB. Results: In 48 primary randomized clinical trials, the model showed good fit indices and factor loadings higher than 0.4. Conclusions: The results support the construct validity of the Cochrane RoB tool and the reliability of three of five items in this health intervention context. (C) 2018 Elsevier Inc. All rights reserved. (AU)

• FAPESP's process: 16/12703-4 - The bias risk assessment model proposed by the Cochrane shows up well-adjusted when applied to clinical trials of psychological treatments for bulimia and binge?
• Grantee: Juliana Pagotto Trevizo
• Support Opportunities: Scholarships in Brazil - Scientific Initiation