Gender and geographical bias in the editorial decision-making process of biomedical journals: a case-control study

Objectives: To assess whether the gender (primary) and geographical affiliation (post-hoc) of the first and/or last authors are associated with publication decisions after peer review.

Design: Case-control study.

Setting: Biomedical journals.

Participants: Original peer-reviewed manuscripts submitted between 1 January 2012 and 31 December 2019.

Main outcome measure: Manuscripts accepted (cases) and rejected for publication (controls).

Results: Of 6213 included manuscripts, 5294 (85.2%) first and 5479 (88.1%) last authors' gender were identified; 2511 (47.4%) and 1793 (32.7%) were women, respectively. The proportion of women first and last authors was 48.4% (n=1314) and 32.2% (n=885) among cases and 46.4% (n=1197) and 33.2% (n=908) among controls. After adjustment, the association between the first author's gender and acceptance for publication remained non-significant 1.04 (0.92 to 1.17). Acceptance for publication was lower for first authors affiliated to Asia 0.58 (0.46 to 0.73), Africa 0.75 (0.41 to 1.36) and South America 0.68 (0.40 to 1.16) compared with Europe, and for first author affiliated to upper-middle country-income 0.66 (0.47 to 0.95) and lower-middle/low 0.69 (0.46 to 1.03) compared with high country-income group. It was significantly higher when both first and last authors were affiliated to different countries from same geographical and income groups 1.35 (1.03 to 1.77), different countries and geographical but same income groups 1.50 (1.14 to 1.96) or different countries, geographical and income groups 1.78 (1.27 to 2.50) compared with authors from similar countries. The study funding was independently associated with the acceptance for publication (when compared with no funding, 1.40; 1.04 to 1.89 for funding by association & foundations, 2.76; 1.87 to 4.10 for international organisations, 1.30; 1.04 to 1.62 for non-profit & associations & foundations). The reviewers' recommendations of the original submitted version were significantly associated with the outcome (unadjusted 5.36; 4.98 to 5.78 for acceptance compared with rejection). Gender of the first author was not associated with reviewers' recommendations (adjusted 0.96, 0.87 to 1.06).

Conclusions: We did not identify evidence of gender bias during the editorial decision-making process for papers sent out to peer review. However, the under-representation in manuscripts accepted for publication of first authors affiliated to Asia, Africa or South America and those affiliated to upper/lower-middle and low country-income group, indicates poor representation of global scientists' opinion and supports growing demands for improving equity, diversity and inclusion in biomedical research. The more diverse the countries and incomes of the first and last authors, the greater the chances of the publication being accepted.