Racial and Ethnic Bias Through Language Application in Letters of Recommendation
Clinical question: Are there racial, ethnic, and underrepresented in medicine (UIM) differences in letters of recommendation and standardized letters of recommendation?
Background: Letters of recommendation are becoming an increasingly important element in the evaluation of applicants at all stages of academic promotion. As traditional components of assessment such as Step 1 and clerkship grading move toward “pass/fail,” narrative comments, including letters of recommendation, are becoming increasingly consequential. Studies have demonstrated gender bias in letters of recommendation, specifically with adjectives to describe men and women in the domains of agentic (assertiveness), communal (interpersonal sensitivity), grindstone (dedication), and standout (outstanding) language. Standardized letters of recommendation are intended to be more objective, however, they have also been found to demonstrate gender bias.
Study design: Systematic review with data abstraction
Setting: Original, peer-reviewed, English-language publications assessing racial, ethnic, or UIM status differences in letters of recommendation from the following electronic databases: MEDLINE via PubMed, SCOPUS, Education Resources Information Center (ERIC), PsycInfo, Embase, and Cochrane Database of Systematic Review
Synopsis: 23 studies were included, which looked at 19,012 applications and 41,925 letters of recommendation, with 82.6% of studies assessing the letters for residency and 17.4% assessing them for fellowship. 17 of 23 studies looked at linguistic differences, with 15 (88.2%) reporting statistically significant differences based on race or ethnicity. 7 of 17 studies showed fewer agentic terms used for Black and Latino applicants. One study noted more communal terms used in letters for Hispanic, Latino, and Black applicants, which were often framed negatively. White applicants were more often described as “exceptional,” “best,” or “outstanding.” 8.7% of all studies demonstrated that UIM applicants were more likely to have doubt-raising language compared to non-UIM applicants. Standardized letters of recommendation were found to have fewer linguistic differences among applicants compared to narrative or unstructured letters of recommendation.
Bottom line: Bias exists in academic medicine letters of recommendation based on ethnicity, race, and UIM status, specifically in regard to language application. The use of standardized letters should be considered as they may decrease racial and ethnic bias compared to traditional letters of recommendation.
Citation: Deshpande SR, et al. Racial and ethnic bias in letters of recommendation in academic medicine: a systematic review. Acad Med. 2024;99(9):1032-1037. doi:10.1097/ACM.0000000000005688.
Dr. Medina is an academic hospitalist and assistant clinical professor at the University of California Davis School of Medicine and the internal medicine clerkship director and vice president of SHM’s Sacramento chapter, both in Sacramento, Calif.