Background: While there are now equal numbers of women and men graduating from medical school, disparities in female representation within academic medicine persist. Gender bias has been cited as one of the main drivers of gender disparity in academic medicine and continues to be a significant barrier for women in the workplace. Grand rounds is an important avenue of exposure and dissemination of scholarly achievements, and is linked to academic merit and promotion. However, studies have shown discrepancies between the number of female and male grand rounds speakers. Another important discrepancy between men and women is the use of professional titles during introductions, which in professional settings convey expertise and competence. Failure to acknowledge a presenter’s title can influence the perceived expertise and authority of the individual. We sought to assess the prevalence and type of gender specific differences in introductions for male and female speakers utilizing natural language processing techniques.
Methods: A retrospective observational study of forms of address used during University of Colorado internal medicine grand rounds was conducted utilizing the internal medicine grand rounds presentations stored in a publicly accessible video archive. The following information was collected from each introduction: name of the speaker and introducer, gender of the speaker and introducer, length of introduction in minutes, and home institution (internal vs external). To extract information from the introductions, Natural Language Processing (NLP) techniques were applied including keyword extraction to identify gendered language in unstructured text content, phrase extraction to identify strings of words that have meaning independent of the meaning ascribed to the separate words included in the string, and sentiment analysis to categorize content as positive or negative and/or masculine or feminine. NLP was also used to analyze key gendered words and phrases utilized, gender of introducer, and gender of the presenter. Participants in the grand rounds were unaware that a retrospective analysis of the instruction was considered.
Results: One hundred and seventy two grand rounds held between 12/4/2013 to 10/2/2019 were reviewed. There were 114 male speakers and 66 female speakers (p=0.0003). Most of the introducers were male (Male = 152, Female=16, p<.0001). Twenty-one women speakers were invited outside speakers in comparison to 43 men (p=0.006). Professional titles are more likely to be used for male speakers than female (63% male vs 50% female, p=0.398). When males introduce female speakers, they are more likely to use the speaker’s full name in comparison to males introducing male speakers (75% vs. 56%, p=0.031). Analysis of term and phrase frequency showed several significant differences between introductions of men versus women. For example, the word “outstanding” was used in the introductions to male speakers significantly more often than for women (16 men vs. 7 women, p=0.0347).
Conclusions: Disparities for women exist in formal introductions at University of Colorado internal medicine grand rounds. There were significantly more male speakers than female and that most of the introducers were men. Several positive words (such as “outstanding”) were used significantly more frequently for men than women. Women were also more likely to be introduced by their full name when introduced by men at internal medicine ground rounds compared to men introduced by men or women introduced by women.