Objective:

To assess if letters of recommendation (LORs) from Neurologists for US Neurology applicants show racial or gender biases.

Background:

LORs play a significant role in the decision to interview and rank residency applicants. Neurology does not use standardized LORs which are believed to reduce bias but are less personal. Other specialties that use narrative LORs have reported gender and racial bias; presence of bias in LORs have not been explored in our specialty.

Design/Methods:

A validated text analysis software program, Linguistic Inquiry and Word Count (LIWC), characterized the content of LORs written for every American medical school graduate applying to a single, academic neurology residency program in the 2021-22 application year based on predefined categories. Multivariable analysis was performed to assess the relationship of self-identified race, gender, and letter writer position with frequency of word categories used, word count, authenticity, and tone.

Results:

434 LORs were analyzed, 238 (54.8%) of which were written for women and 72 (16.6%) written for underrepresented minorities (URM) in medicine. A significant difference was seen in the grindstone category of descriptors (i.e. hardworking, tenacious, thorough, etc.) used for males vs females (.46 vs .53, p=.04). A significant difference was seen in the authenticity rating for males vs females (7.04 vs 5.95, p=.02). There was no word category difference between LORs written for applicants who identified as Caucasian, Asian, African American, Hispanic. A significant difference was seen in the word count of letters written by assistant, associate, and professors (145 vs 105 vs 103, p=.02). 

Conclusions:

LORs written for neurology residency applicants had a statistically significant difference for more grindstone words and less authentic tone used for female vs male applicants with no significant racial difference seen. Further studies are needed to ensure the current narrative model of Neurology LORs provides for an equitable evaluation of applicants.