Omolola Adepoju; Lu Tran; Rosemary Agwuncha; Minji Chae; Jason Franco-Castano; Tracy Angelocci; Winston Liaw
Corresponding Author: Omolola E Adepoju, PhD, MPH; Department of Health Systems and Population Health Sciences - UH College of Medicine.
Email: oadepoju@uh.edu
Section: Original Research
Published: www.jabfm.org/content/35/3/457
PURPOSE: Increased telemedicine implementation may promote primary care access. However, gaps in telemedicine uptake may perpetuate existing disparities in primary care access. This study assessed provider and patient-level factors associated with telemedicine utilization in community-based family practices clinics.
METHODS: This retrospective study utilized electronic medical records (EMR) data from a large federally qualified health center (FQHC). A three-level mixed-effects logistic regression model explored predictors of telemedicine utilization, with provider and patient as random effects.
RESULTS: The analytic sample included 37,428 unique patients with 106,567 primary care encounters with 42 family medicine providers, between January 1, 2020-November 30, 2020. 57% of the sample identified as Hispanic, 28% as non-Hispanic White and 11% were non-Hispanic Black. Compared to Hispanics, non-Hispanic White patients had 61% higher odds of a telemedicine visit (OR=1.61; p<0.001), and non-Hispanic Black patients had 32% higher odds of a telemedicine visit (OR=1.36; p<0.001). The odds of telemedicine use were lower for those who were uninsured (OR=0.81; p<0.001). Those residing in metropolitan areas (OR=1.25; p=0.004) or medically underserved areas (OR=1.19; p<0.001) had greater odds of a telemedicine appointment. Commute time exhibited a dose response relationship with telemedicine use. Provider characteristics were not significantly associated with telemedicine use. Conditional on the fixed-effects covariates, provider and patient random effects comprised approximately 39% of the total residual variance.
CONCLUSIONS: While provider characteristics were not associated with telemedicine use, greater focus on patient characteristics specific to the population served is necessary.