Female scientists are underrepresented among authors of retracted medical articles

This is a cross-sectional study that analyses articles retracted for any reason published in the 15 journals with the highest impact factor within nine categories of knowledge. The main objective is to describe the gender distribution of the authors of these retracted articles. The sources used to identify retractions are robust and combine different databases, reducing the likelihood of omitting retracted articles. Similarly, the statistical analysis is adequate and consistent with the objectives set.

The study concludes that, in the publications analysed, the proportion of female authors is lower than that of male authors. This proportion is lower than the expected percentage of women in scientific authorship in general. In any case, it is likely that part of this difference reflects the underrepresentation of women in scientific literature in general and, consequently, also in retracted literature.

However, there are methodological limitations, many of which are acknowledged by the authors. For example, the selection of these nine clinical areas over other possible ones is not clearly justified, which leaves some uncertainty about the representativeness of the study. The assignment of authors' gender is complex: it is common not to know the full name, and in other cases, particularly with Asian names, names can be unisex, making classification difficult. These situations were excluded from the analysis, which could introduce a selection bias by including only researchers with complete information or unambiguous names.

Furthermore, the study does not incorporate other contextual variables that could help to refine the results, such as the geographical region of the authors. The results are limited to gender and category of knowledge of the journal in which the retracted article was published. External validity is limited to high-impact journals, as lower-impact publications are not included. Analysing the journal's quartile could have provided additional information of interest.

Despite the limitations described, the article is valuable in the sense that identifying gender in scientific publications remains a methodological challenge. The main contribution of the work lies in confirming the results of previous studies, but with a more robust methodology and a larger sample size.