Screening, extraction and analysis

Three researchers (SM, AP and DG) used keywords across five databases and uploaded the articles from PubMed, PsycINFO and Web of Science on Covidence, whereas articles from Nepali databases (NepMed and NepJol) were downloaded in Excel for screening because they were unable to be uploaded in Covidence. SM, AP and DG screened the titles and abstracts for inclusion based on our criteria, followed by a detailed abstract and full-text review. Any discrepancy was discussed and resolved by a senior researcher (BAK). SM, AP, DG and MS then coded the articles in an extraction sheet, which recorded information on gender and ethnicity of first, last and corresponding authors, funding type, study type, journal of publication, list of authors, ethnicity and gender.

To determine caste/ethnicity among Nepali authors, researchers of Nepali origin (DG, MS, AP and PS) identified ethnicity by surname because caste/ethnicity in Nepal is typically indicated by last name. Though determining gender without self-report methods risks inaccuracies, we followed the practice of coding gender by first name in authorship lists, which is standard practice in other studies which have analysed authorship representation.^{8 9} Because of the limited information available from authorship name lists, we have limited the use of ‘gender’ to the terms ‘women’ and ‘men’, while acknowledging this will not accurately capture all authors’ gender identities.

To identify patterns of difference in authorship, we conducted within group comparisons for one sample binomial comparisons for gender biases within the specific group (ie, Nepali and non-Nepali authored publications are analysed separately). Significant results demonstrate a deviation from 50% split by gender. In addition, between group comparisons are Mantel-Haenszel ORs with 95% CIs. Odds represent the ratio of the odds of an event in one demographic group compared with the odds of an event in a comparison demographic group, for example, women’s authorship compared to men’s authorship among Nepalis referenced against the ratio for non-Nepali authors based on gender. Analyses were conducted using SPSS Statistics V.27.0. ³⁰