Abstract 73: Using Network Science Community Detection Methods To Identify Insurance-based Disparities In Stroke Center Access In California

Abstract

Introduction: Ischemic stroke (IS) patients are frequently transferred between hospitals. Our objective was to determine whether patient insurance status contributes to variation in access to stroke center care among transferred patients with IS. Methods: We compiled California data on every nonfederal hospital admission from 2010-17 and used ICD-9, ICD-10, and DRG codes to identify IS patients transferred from an initial emergency department to another hospital. Transfers were categorized based on whether or not the patient was ultimately discharged from a stroke center hospital (primary or comprehensive). Patient insurance status was categorized as private, Medicare, Medicaid or self/uninsured. Clusters of closely connected hospitals via transfer frequency were identified using network science community detection methods. Within each cluster, we examined the degree of disparity in stroke center access by quantifying the difference between the insurance groups with the highest and lowest proportion of transfers discharged from a stroke center. Results: We identified 10,049 IS transfers during the study period (private 5,297 [53%]; Medicare 3,328 [33%]; Medicaid 904 [9%]; self/uninsured 520 [5%]). Stroke center access varied by patient insurance (overall 87%, private 89%, Medicare 87%, Medicaid 82%, self 72%). There were 14 clusters of closely connected hospitals via transfers. In the highest performing cluster, 100% of transferred patients in each insurance category were discharged from a stroke center (delta 0). The lowest performing cluster was also the largest (n=2,364 transfers); in this cluster 69% of transfers were discharged from a stroke center, ranging from 32% of self-pay transfers to 81% of privately insured transfers (highest delta among all clusters: 49%). Conclusions: These findings demonstrate that current care patterns differ by insurance status. Further research is needed to determine interventions to address this disparity.