Social Network Analysis of Cancer Provider Collaboration

Abstract

Early data suggests that cancer providers who are more tightly connected have better clinical outcomes and lower costs. Many current methods to describe and measure provider connectivity are relatively simple, yet not easy to implement at a single institution as they are based on single payor claims data. The secondary use of routinely collected data from clinical environments is one opportunity to evaluate provider connectivity across an entire institution. We hypothesized that we could quantitatively model provider connectivity in outpatient clinics through a social network analysis of routinely collected health data. We demonstrate that social network analysis can be used to assess provider connectivity and that we can detect differences between cancer populations as expected from clinical experience. Additionally, we developed a temporal approach to model relationships such that we can represent the dynamics of clinical care.