Coordination of Specialty Care Within and Across Health Systems: Patient Experience in the Veterans Health Administration

OBJECTIVE: To compare patients' experience of specialty care coordination within a single health system to that across organizational boundaries, and whether coordination varies by illness burden and across coordination domains. STUDY SETTING AND DESIGN: National cross-sectional survey of patients referred from Veterans Health Administration (VA) primary care to VA specialty care ("within-system") or VA-paid community specialty care ("cross-system"). The primary outcome was patient-reported overall coordination. Illness burden was assessed using the Veterans RAND 12-item Health Survey. Secondary outcomes included 10 different domains of coordination. DATA SOURCES AND ANALYTIC SAMPLE: Data were collected between September 2022 and March 2025 using the Coordination of Specialty Care-Patient Survey and linked to VA administrative data. Eligible participants were Veterans with a recent first visit to cardiology, endocrinology, or pulmonary medicine and no prior visit to the same specialty within 18 months. The analytic sample included 1388 respondents (40% response rate). We used mixed-effects logistic regression models with primary care facility as a random effect. PRINCIPAL FINDINGS: In adjusted analyses, within-system patients were 5.3 percentage points more likely than cross-system patients to report well-coordinated care [average marginal effect (AME) 0.053, 95% CI: 0.052, 0.053]. Higher illness burden was associated with worse coordination (AME -0.145, 95% CI: -0.147, -0.144), and this did not differ by care setting. Among 10 coordination domains, only Specialist Knowledge of Patient History differed consistently, favoring within-system care. CONCLUSIONS: Patients receiving within-system specialty care reported better overall coordination than those receiving cross-system care despite similar scores across most domains. Coordination was worse among those with higher illness burden, identifying a higher-risk group in both settings. The divergence between overall coordination and individual domains highlights a challenge in cross-system comparisons: differences in patient experience may arise from processes not represented in available data.