62 Network Analysis To Understand Regional Patient Flow

Abstract

Consolidation of health care systems in the United States has created integrated enterprises with large geographical reach and complex interacting components. Specialty services vary among sites, and patients often need to travel between facilities for different aspects of care. Network science is an established method employed to investigate these complex systems, and can be used to identify bottlenecks, opportunities to increase value and patient-centeredness in health care. The COVID-19 pandemic changed care demand patterns unexpectedly. We wanted to investigate if network analysis would allow us to better understand these changes and challenges - where our patients receive hospital-based care, what services they used and how far they travelled. We focused this analysis on a multispecialty, tertiary care academic center emergency department (“AC_ED”), with 70,000 patient visits annually. We extracted patient location information from electronic health records, including originating location and level of care, to create a network representation of all care pathways that passed through our ED. The volume of transfers between nodes and the distance travelled were encoded as weighted/colored edge attributes, with edges between nodes in closer proximity being darker. Nodes include: communities within our service area, EDs, and academic center units, and an outcome of mortality. They are sized/colored by betweenness centrality, reflecting the importance of the node in the integrity of the network. The figure shows the overall network structure was similar for pre- and post-pandemic onset with some changes in details. AC_ED receives patients from many home locations and referring hospitals. A large proportion of visits come from the local area reflected by M16 and M12. There are many patients who travel far to access emergency care at AC_ED, bypassing local EDs though the average distance travelled to access care reduced from 114 to 85 miles. During the pandemic there was more traffic to the AC_ED from the local area (M16), fewer connections to surrounding hospitals and disproportionately reduced visits from distant areas (O and OT). Low ED volumes and restricted outpatient clinic availability during the pandemic time frame likely affected this. Inter-hospital transfer volumes declined overall, several sites transferred very few patients to AC_ED post-pandemic start, and other sites increased their transfer rate (eg, CAH13). Looking at hospital systems through the lens of network science can reveal changes in patterns of referrals, allows for identification of unexpected results by presenting data visually and can assist identifying crucial components of a health care system. Application of this methodology to other variables has the potential to identify new areas of improvement to increase value, outcomes and services to improve patient-oriented care.