Digital Health Readiness and Medicare Primary Care Spending: Nationwide County-Level Observational Analysis

Abstract Background Digital health technologies are increasingly promoted as mechanisms to improve care coordination, enhance access, and reduce health care costs. However, whether community-level digitalization is associated with lower primary care spending in Medicare-participating safety net settings remains unclear, particularly in federally qualified health centers (FQHCs) and rural health clinics (RHCs). Objective This study examined the association between county-level digitalization and Medicare primary care spending for FQHC and RHC services across US counties. Specifically, we assessed whether higher digitalization was associated with lower spending, whether observed associations reflected persistent differences between counties or changes within counties over time, and how digitalization interacted with health care access and socioeconomic conditions. Methods We conducted a county-level observational study of 2993 US counties from 2017 to 2023. Digitalization was measured using the Digital Health Index (DHI) digitalization subindex, and the outcome was geographically adjusted for per capita Medicare spending on FQHC and RHC services. A hybrid within-between panel model was used to distinguish between-county from within-county associations. Additional analyses included threshold models, digitalization trajectory analyses, and interaction models. Using 2023 cross-sectional data, multivariable regression and Extreme Gradient Boosting models with Shapley additive explanations were applied to evaluate the relative contributions of digitalization, health care access, and socioeconomic conditions to spending variation. Results Higher county-level digitalization was associated with lower Medicare primary care spending. In longitudinal analyses, the association was driven primarily by persistent between-county differences (β=–67.62, 95% CI −73.51 to −61.74; P <.001), whereas within-county changes in digitalization were not significantly associated with spending (β=–2.72, 95% CI −9.10 to 3.67; P =.40). Compared with counties in the lowest digitalization tertile, spending was substantially lower among counties in the highest tertile (β=–165.05, 95% CI −179.74 to −150.36; P <.001). Counties characterized by early and sustained digital development also demonstrated significantly lower spending (β=–123.34, 95% CI −138.24 to −108.43; P <.001). In a separate 2023 cross-sectional analysis of the complete DHI, health care access exhibited the strongest inverse association with spending (β=–71.13, 95% CI −81.20 to −61.07; P <.001), followed by digitalization (β=–48.48, 95% CI −59.75 to −37.21; P <.001), whereas socioeconomic conditions were positively associated with spending (β=22.01, 95% CI 13.12-30.89; P <.001). Among the 3 DHI dimensions examined, Shapley additive explanations analyses identified health care access and digitalization as the most informative predictors of county-level spending variation. Conclusions Higher county-level digitalization was associated with lower Medicare primary care spending in FQHC and RHC settings. However, the association reflected long-standing structural differences across counties rather than short-term changes in digital capacity. These findings suggest that digitalization functions as a component of broader health system capacity and may contribute to improved primary care efficiency when supported by sustained investment, adequate health care access, and favorable local conditions.