1069 — Social Network Analysis to Quantify Changes in Coordination of Epilepsy Care in Veteran Health Administration from 2000-2013

Hamid H, VA Connecticut Health Care System, West Haven, CT; Fenton B, VA Connecticut Health Care System, West Haven, CT; Bates J, VA Connecticut Health Care System, West Haven, CT; Pugh MJ, South Texas Veterans Healthcare System; Cheung K, VA Connecticut Health Care System, West Haven, CT; Neal J, VA Connecticut Health Care System, West Haven, CT; Hareeri F, VA Connecticut Health Care System, West Haven, CT; Valente T, University of Southern California, CA; Brandt C, VA Connecticut Health Care System, West Haven, CT; Kerns R, VA Connecticut Health Care System, West Haven, CT

Objectives:
To assess healthcare service quality we quantified the coordination of services across Veterans Health Administration (VA) providers caring for Veterans with epilepsy (VWE), using social network analysis (SNA) before and after implementation of the VA Epilepsy Centers of Excellence (ECOE) in 2008.

Methods:
Data were extracted from the Corporate Data Warehouse. VWE were defined by ICD-9 Code 345.X and 780.39. All unique encounters and associated providers were identified from 2000 to 2013. Links between providers were defined as the number of VWE shared between two providers. We calculated centrality degree (SNA measure of connectivity) by summing the total number of patients shared by given provider. Degree was summed for providers caring for VWE in each VA facility. We calculated the change in degree for VA neurologists affiliated with ECOE and not affiliated with ECOEs (non-ECOE). Non-ECOEs were matched to ECOEs by facility size and level of complexity. A General Linear Model (GLM) was used to test the difference in degree between neurologists located at ECOE and non-ECOE VA facilities from 2000-2013.

Results:
From 2000-2013 the number of providers who encountered a VWE increased from 10,529 to 16,362. The increase in degree between ECOE and non-ECOE facilities was 850,029 and 368,395, respectively. The group (ECOE vs non-ECOE) X time (years 2000-2013) interaction was significant (F-value = 55.48, df = 1, p < .0001) in the GLM, supporting the hypothesis that centrality change over time was significantly greater for neurologists in ECOEs versus those in non-ECOEs.

Implications:
One tacit goal of the ECOEs was to develop an enhanced network for care of VWE. The demonstration of markedly greater increase in connectivity among providers affiliated with ECOEs compared to those not affiliated with ECOEs suggests that this system-wide intervention may have accomplished this goal.

Impacts:
This research lays the ground-work for targeting specific network interventions to improve collaboration across facilities and providers. It also suggests that data can be used creatively by VA as a learning organization to improve organization of care, and ultimately quality of care for Veterans.