Exploring the Use of Social Network Analysis On Physician Networks Created from Medicare Data through Studying the Use of Minimally Invasive Breast Biopsy Among Physicians: Descriptions, Regressions, and Network Models

Social Network Analysis (SNA) has been applied in a variety of scientific fields. In particular, SNA has been utilized with Medicare data to study the structure of physician networks. We utilized SNA to elucidate the structure of physician networks derived from Medicare data, to gain a better understanding of variation in treatment patterns ascertained from Medicare data. This was in service of a fuller exploration of utility of SNA use on networks derived from Medicare data. With the main context being the use of minimally invasive breast biopsy among physicians in Texas, our aims were to assess whether network structure, as identified by SNA, and measures derived from these networks, are useful as supplements to standard statistical models such as hierarchical regression. Further, we aimed to build social network models, specifically exponential random graph models and network autocorrelation models, in exploring the structure of physician networks and in determining if social relationships affect the outcome of interest. Our results reveal that network analysis has potential in not only controlling for variation in regression models but in also highlighting the importance of physician relationships and network positions in health outcomes and health outcomes research. However, a better understanding of the interpretation of these physician networks derived from Medicare data is needed in order to fully tap into the potential of network analysis in health services research.