Rural Background, Training Exposure, and Rural Practice Location Among US Physicians
Abstract
The physician shortage has limited health care access in the rural US. Rural recruitment efforts typically follow 2 strategies: recruiting students from rural backgrounds1–3 and providing exposure to rural settings during undergraduate medical education (UME) and graduate medical education (GME) training.4–8 Despite mounting evidence that rural practice can be influenced by both rural background and training at different stages, prior studies have only examined these influences separately.1,3,5,8 Our cross-sectional study accounts for the interdependence between these variables and compares the magnitude of their associations within a single model.
We analyzed the AAMC’s 2022 National Sample Survey of Physicians (NSSP), in which 5917 active US physicians provided their demographic and professional information, including rural background (a binary variable), and the percentage of time spent in rural areas during their medical school clinical rotation, residency and fellowship, and patient care (continuous variables). NSSP data were collected following a 2-step method, and postsurvey weights were created to adjust for differential nonresponse using the American Medical Association’s Physician Professional Data 2022.9 Participants provided written and electronic informed consent at the beginning of the survey. The American Institutes for Research institutional review board deemed this study exempt from review because it is a secondary analysis of deidentified data. We followed the STROBE reporting guideline.
Statistical analysis was performed from January 3, 2024, to January 27, 2026. We conducted an ordinary least-squares (OLS) multiple regression using rural origin, rural UME, and GME exposure to estimate rural patient care time. We then developed a structural equation model (SEM) to account for interdependent associations among rural origin, rural exposure during UME and GME, and rural practice (eMethods in Supplement 1). In both models, we included physicians’ gender, age, and whether they were primary care physicians (PCPs).2,3
All analyses adjusted postsurvey weights and were performed using Stata, version SE18. We set significance levels at a 2-sided P < .05.
The analytical sample (N = 5778; mean [SD] age, 53.8 (10.8) years) included 2149 women (37.2%), 3629 men and individuals of other gender identities (62.8%), 1780 PCPs (30.8%), and 751 physicians (13.0%) who grew up in rural areas. The mean (SD) time spent in rural areas was 8.3% (14.8%) during UME, 6.2% (17.9%) during GME, and 11.6% (29.4%) during current patient care.
The OLS results show that after adjusting for age, gender, and PCP status, rural origin physicians spent 23.5 percentage points (pp) (95% CI, 21.4-25.6 pp; P < .001) more of their practice time in rural areas than nonrural origin physicians. Completing all GME training in rural areas was associated with increased rural practice time by 42.0 pp (95% CI, 37.7-46.3 pp; P < .001). Rural UME showed no significant association with rural practice.
SEM analysis revealed indirect associations that the OLS model did not capture, while the direct associations aligned with the OLS estimates. Holding covariates constant, rural origin had a total of 27.1 pp (95% CI, 21.8-32.4 pp; P < .001) association with rural practice time (direct, 23.5 pp; indirect, 3.6 pp) (Figure 1). Rural GME had the largest total association, entirely direct. Rural UME had a total association of 22.3 pp (95% CI, 12.3-32.4 pp) with rural practice time (direct, 1.4 pp; indirect, 20.9 pp; P < .001), due mostly to the indirect association through increased rural GME time.
Our analyses suggest that the associations of rural UME and origin may be underestimated in OLS-based studies, which capture only direct associations. Second, the total association of rural UME with rural practice is largely mediated by the continuation of rural training during GME, cautioning against relying solely on rural UME to increase rural physician supply. Third, we confirm that rural GME is significantly associated with increased time spent in rural practice, even among physicians without rural origins, offering a strategy for increasing rural physician supply amid a dearth of physicians with rural origins.
This study has limitations. First, as an observational study based on cross-sectional survey data, although sequential associations are assumed among rural origin, UME, and GME training, and practice, we can make no causal inferences. Second, self-reported estimates of training and practice time in rural areas may be subject to inaccuracy, as people tend to overestimate the rurality of places relative to standardized scales (eg, Rural-Urban Commuting Area classification).10 Nevertheless, applying a consistent measurement framework across UME, GME, and practice offset individual tendencies, thereby providing insight into the relative associations of rural origin, rural UME, and rural GME training.
Authors:
Hu X, Patterson DG, Conrad SS, Luo Q, Dill MJ
Journal/Publisher:
JAMA Network Open
Edition:
Aug 2026.
Link to Article
Access the article here: JAMA Network Open
Citation:
Hu X, Patterson DG, Conrad SS, Luo Q, Dill MJ. Rural Background, Training Exposure, and Rural Practice Location Among US Physicians. JAMA Network Open. Aug 2026