Lack of sex-related analysis and reporting in Cochrane Reviews: a cross-sectional study

Abstract

BACKGROUND: Sex-specific analysis and reporting may allow a better understanding of intervention effects and can support the decision-making process. Well-conducted systematic reviews (SRs), like those carried out by the Cochrane Collaboration, provide clinical responses transparently and stress gaps of knowledge. This study aimed to describe the extent to which sex is analysed and reported in a cross-section of Cochrane SRs of interventions, and assess the association with the gender of main authorships. METHODS: We searched SRs published during 2018 within the Cochrane Database of Systematic Reviews. An investigator appraised the sex-related analysis and reporting across sections of SRs and collected data on gender and country of affiliation of the review first and last authors, and a second checked for accuracy. We conducted descriptive statistics and bivariate logistic regression to explore the association between the gender of the authors and sex-related analysis and reporting. RESULTS: Six hundred and ten Cochrane SRs were identified. After removing those that met no eligibility criteria, 516 reviews of interventions were included. Fifty-six reviews included sex-related reporting in the abstract, 90 considered sex in their design, 380 provided sex-disaggregated descriptive data, 142 reported main outcomes or performed subgroup analyses by sex, and 76 discussed the potential impact of sex or the lack of such on the interpretations of findings. Women represented 53.1 and 42.2% of first and last authorships, respectively. Women authors (in first and last position) had a higher possibility to report sex in at least one of the review sections (OR 2.05; CI 95% 1.12-3.75, P=0.020) than having none. CONCLUSIONS: Sex consideration amongst Cochrane SRs was frequently missing. Structured guidance to sex-related analysis and reporting is needed to enhance the external validity of findings. Likewise, including gender diversity within the research workforce and relevant authorship positions may foster equity in the evidence generated.