Sentiment analysis of letters of recommendation for a U.S. pain medicine fellowship from 2020 to 2023

Christopher P. Cheng,Vikram Vasan,Alopi M. Patel,Paul R. Shekane
DOI: https://doi.org/10.1111/papr.13416
IF: 3.079
2024-10-06
Pain Practice
Abstract:Objectives Letters of recommendation (LORs) are an important part of pain medicine fellowship applications that may be subject to implicit bias by the letter's author. This study evaluated letters of recommendation for applications to pain medicine fellowships in the United States to characterize biases and differences among applicants over four application cycles. Methods This was a retrospective single‐site cohort study. De‐identified LORs were collected from 2020 to 2023 from one institution. The Valence Aware Dictionary and sEntiment Reasoner (VADER) natural language processing package scored positive LOR sentiment. In addition, the deep learning tool, Empath, scored LORs for 15 sentiments. Wilcoxon rank‐sum and one‐way ANOVA tests compared scores between applicant demographics: gender, race, medical school type, residency specialty, and chief resident status, as well as letter writers' academic position. Results Nine hundred and sixty‐four applications were studied over four application cycles. Program directors wrote fewer words (p = 0.020) and less positively (p
clinical neurology,anesthesiology
What problem does this paper attempt to address?