Spin in Randomized Controlled Trials in Obstetrics and Gynecology: A Systematic Review

Abstract

Objectives: The objective of this study was to evaluate the extent, type, and severity of spin in randomized controlled trials (RCTs) in obstetrics and gynecology.

Data sources: The top five highest impact journals in obstetrics and gynecology were systematically searched for RCTs with non-significant primary outcomes published between January 1, 2019, and December 31, 2020.

Methods: Study selection and data extraction assessment were conducted independently and in duplicate. The extent, type, and severity of spin was identified and reported with previously established methodology, and risk of bias was assessed with the Cochrane Risk-of-Bias 2 Tool independently and in duplicate. Fisher's exact tests were used to evaluate the association between study characteristics, risk of bias, and spin.

Results: We identified 1475 publications, of which 59 met our inclusion criteria. Articles evaluated interventions in obstetrics (n = 37, 63%) and gynecology (n = 22, 37%). Spin was not detected in 28 (47%) of the articles: Three (5%) had one, 10 (17%) had two, and 18 (31%) had greater than two occurrences of spin. Compared with articles where no spin was detected, spin was associated with the Cochrane Risk-of-Bias domain pertaining to missing data (p < 0.05). No association was observed with the journal, funding source, number of authors, types of interventions, and whether the study involved gynecology or obstetrics.

Conclusions: Spin was detected in nearly half of 1:1 parallel two-arm RCTs in obstetrics and gynecology, highlighting the need for caution in the interpretation of RCT findings, particularly when the primary outcome is nonsignificant.

Keywords: RCTs; gynecology; interpretive bias; methodology; obstetrics; reporting.

FIG. 1.

PRISMA flow diagram for study selection, randomized trials in obstetrics and gynecology literature, January 2019 to December 2020. 1475 publications were identified, of which 96 (6.5%) were duplicates. Seventy-nine (5.3%) titles and abstracts passed initial title and abstract screening, and full-text articles were retrieved. Six (0.41%) articles were excluded, as they had statistically significant primary outcomes. Fourteen (0.95%) articles were excluded, as they did not have 1:1 parallel, two-arm study designs. A total of 59 (4.0%) studies met the inclusion criteria. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses.