Background: Gender disparities in scientific authorship are well documented, yet little is known about gender representation among authors of retracted publications.

Methods: We analyzed 878 retracted publications from 131 high-impact medical journals across nine clinical disciplines (anesthesiology, dermatology, general internal medicine, gynecology/obstetrics, neurology, oncology, pediatrics, psychiatry, and radiology). Gender was inferred using Gender API for all, first, and last authors. Two analytic samples were constructed based on prediction confidence thresholds (≥60% and ≥70%). We examined gender distribution across authorship positions, number of retractions per author, and disciplinary representation. Wilcoxon rank-sum and chi-squared tests were used to assess group differences. Gender proportions were compared with publication benchmarks from 2008-2017, restricting retraction data to the same period for comparability.

Results: Among 4,136 authors, 3,909 had full first names, and gender could be assigned to 3,865 (98.9%). In the sample with prediction confidence ≥60% (n = 3,743), 863 (23.1%) were identified as women. They accounted for 16.5% (123/747) of first and 12.7% (87/687) of last authors. They had significantly fewer retractions per author and were less likely to have >5 retractions (all authors: 3 women [8.1%] vs 34 men [91.9%], p < 0.001). Across most disciplines, their representation was below publication benchmarks. Dermatology (retractions = 80.0%, publications = 48.9-51.8%) and radiology (retractions = 40.0%, publications = 31.0-36.8%) were exceptions among first authors, while pediatrics (retractions = 50.0%, publications = 37.0%-42.6%) was an exception among last authors, though all based on small numbers.

Conclusions: Women are markedly underrepresented among authors of retracted publications, particularly in cases involving multiple retractions. Further research is needed to clarify underlying mechanisms.