Primary Care Centrality in Robust Specialist Networks Results in Lower Emergency Department Utilization: A Network Analysis of Physician Networks in Texas Medicaid

Importance: Primary care coordinates care across a complex healthcare ecosystem that includes the spectrum of service delivery and the socio-ecologic infrastructure of communities and populations. To inform best practices in designing physician networks, understanding these complex interactions and their impacts on patient health outcomes is essential.

Objective: To apply social network analysis to assess the impact of primary and specialty physician networks on emergency department (ED) visits for patients with ambulatory care sensitive conditions (ACSCs).

Design, Setting, and Participants: Cross sectional network analyses of primary care physicians (PCP) and specialty physicians caring for 135,165 18-64, non-dual eligible Medicaid beneficiaries with at least one ACSC in Texas in 2009 using Medicaid Analytic eXtract (MAX) files; 42,493 patients were continuously enrolled in 2009 with a continuity PCP (31% of total sample). These patients were assigned to 10,665 PCPs connected to 11,709 specialist physicians. Physician specialty was identified based on the National Plan and Provider Enumeration System database specialty taxonomy. Patients were assigned to a continuity PCP if they had at least two ACSC related visits to the PCP and the PCP accounted for at least half of the patient’s total PCP visits in 2009. A community detection algorithm was used to adjust the centrality measure (i.e. rate of patients returning after specialty consult) such that the majority of the PCPs were not assigned the same community. A negative binomial regression analyzed the impact of PCP i’s network characteristics on the number of ACSC ED visits from patients in PCP i’s panel.

Results: Overall more PCPs, compared to specialists, were females (22% versus 20% specialists, p<0.01), practiced in rural settings (14% versus 9% specialists, p<0.01), and Health Professional Shortage Areas (HPSAs) (7% versus 3%, p<0.01). PCPs whose continuity patients did not visit a specialist were associated with 86% lower patient panel ACSCs ED rates, compared to PCPs whose patients saw specialists. Among non-solo PCPs, those with a higher number of specialist collaborators and those with a high degree of centrality had lower patient panel ED rates. ED rates were negatively associated with the number of specialist collaborators if the PCP had an eigenvector centrality of at least 0.49 and negatively associated with the eigenvector centrality of a PCP if the number of specialist collaborators was at least 113. Specifically, a PCP with ten specialist collaborators and a low number of patients flowing between the specialist and the PCP has 184% higher ED rates in his or her patient panel than a PCP with 5 specialist collaborators and the same patients flow volume.

Conclusions and Relevance: Texas Medicaid PCPs with the lowest rates of ED utilization included those providing solo care and those with an adequate specialist network combined with high degree of centrality in their network. To obtain the best patient outcomes, the right fit between care coordination, primary care centrality, and adequate capacity of specialty networks is needed. Our approach provides a blueprint for assessing the impact of social network effects of other provider specialties and other patient populations.