Using a Multidimensional Statistical Approach to Evaluate Diversity Trends Among U.S. Neurology Residents Between 2008–2020
Elijah Persad-Paisley1, Saba Paracha1, Selamawit Moges2, Sanaya Daruvala3, Ali Mahta4, Neishay Ayub5
1Warren Alpert Medical School of Brown University, 2Apella Technology Inc., 3Rhode Island Hospital, Department of Neurology, 4Brown University, 5Brown Neurology
Objective:

To develop a multidimensional statistical model to quantify diversity within neurology residency.

Background:

Studies have demonstrated the benefits of diversity in medicine. However, recruitment and retention within neurology consistently lags behind other medical specialties. While demographics of neurology graduates has been studied, no statistical models exist to objectively assess diversity among graduates entering into the neurological workforce.

Design/Methods:

Demographics of neurology residents between 2008–2020 across 3 categories (sex, medical school affiliation, race/ethnicity) was measured. Within each category, µ (the probability that two randomly individuals represent a different sex, medical school affiliation [US MD, DO, IMG, or Canadian], or race/ethnicity), and true diversity, D (the effective number of equally populated categories per variable), were calculated. All 3 categories were combined using the Sullivan composite diversity index (CDI) to produce Aw  (the proportion of attributes on which two random residents will differ). Linear regressions of D for each variable across time was calculated.

Results:

There were an average of 2,326 residents/year, with 36% female, 95% US MD graduate, and 70% White. The probability that two random residents represented a different: sex was 49.6%; medical school affiliation was 56.4%; and racial/ethnic group was 60.9%. This corresponded to an effective number of 1.98 sexes, 2.29 medical school types, and 2.56 races/ethnicities. True diversity D increased over time for medical school affiliation only (+0.016/year), while D remained constant for sex and race/ethnicity. Lastly, two random residents differed on about 55.6% of all attributes (Aw), suggesting relative heterogeneity.

Conclusions:

Both sexes are nearly equally represented across neurology residents, and neurology residency programs were represented by effectively two different medical school affiliations. However, despite the presence of seven races in the population, neurology residents were only as diverse as a cohort with 2.5 equally represented races. Leaders in neurology should consider additional strategies to improve the neurology residency recruitment.

10.1212/WNL.0000000000202819