Background/Purpose: Up to half of RA patients receive virtual rheumatology care, which is associated with high satisfaction but potentially lower quality. We aimed to identify factors associated with virtual RA care utilization in a real-world cohort seen at the Veterans’ Health Administration, where telehealth services are established, well-integrated, and available across sites.

Methods: This retrospective cross-sectional study used national VA administrative claims to identify patients who a) met a validated claims-based definition of RA, b) had ≥1 rheumatologist visit for RA in 2022-2023 (study period). We compared demographics and treatment characteristics stratified by proportion of rheumatology care for RA received virtually (virtual proportion: 0, < 0.2, 0.2-0.5, 0.51-0.9, >0.9). To investigate factors associated with virtual proportion, we used zero-one inflated beta regression to account for the bounded nature of proportion data and excess zeros/ones. We estimated average marginal effects of age, sex, RA duration, log-transformed cumulative glucocorticoid (GC) dose and distance to rheumatologist’s VA site, number of unique DMARDs, all-cause in-person office visits, MTX and biologic use. We performed a sensitivity analysis limited to those with ≥2 rheumatologist RA visits.

Results: We analyzed 11,703 patients, of whom 4964 (42.4%) had a virtual rheumatology visit, 3895 (33.2%) had virtual proportion >0.2, and 1661 (14.2%) had virtual proportion >0.5. As virtual proportion increased, age and distance to rheumatologist increased, and proportion of male patients, MTX, biologic, and GC use, number of DMARDs, and rheumatology and in-office visits decreased. However, distance to rheumatologist and MTX use did not differ significantly for virtual proportion 0.2-0.5 vs 0 (p=0.17, 0.43 respectively), and biologic use, GC days’ supply, and GC dose did not differ significantly for virtual proportion >0.9 vs 0 (p=0.10, 0.72, 0.61 respectively). The average change in virtual proportion per incremental increase in each predictor was examined (Table 2). Virtual proportion was associated with an average increase of 0.14 percentage points per year of age (95% CI 0.06-0.22), and 4.2 points per unit increase in log-transformed site distance (approximately 2.7-fold increase in miles, 95% CI 3.3-5.1). Virtual proportion was associated with an average decrease of 2.3 percentage points per additional unique DMARD (95% CI 3.5 to -1.1) and 0.55 points per additional all-cause in- office visits (95% CI: -0.65 to -0.44). Sex, GC dose, RA duration, and MTX and biologic use did not show significant associations. Our sensitivity analysis showed similar findings.

Conclusion: In a real-world cohort of RA patients with access to telehealth, virtual RA care was common and associated with older age, female sex, greater distance from rheumatology site, fewer in-office visits, and lower DMARD and GC use. Initial findings suggest trends may vary across levels of utilization. Limitations include those associated with claims data (e.g. confounding by indication) and cross-sectional design limiting causal inferences. Further work is needed to identify factors affecting virtual care utilization for RA and their impact on clinical outcomes.

Table 1: Population characteristics stratified by proportion of rheumatology care for RA received virtually

Table 2: Variables significantly associated with virtual care proportion in a zero-inflated beta regression

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/prevalence-and-factors-associated-with-virtual-rheumatology-care-utilization-in-a-national-population-of-veterans-with-ra/