Dr. James O’Malley has two papers accepted for publication in the Journal of Biostatistics and the journal of Statistics in Medicine

Paper 1: Adjusting for bias introduced by instrumental variable estimation in the Cox Proportional Hazards Model by Pablo Martinez-Camblor, Todd MacKenzie, Doug Staiger, Philip Goodney, and James O’Malley accepted for publication in Biostatistics (https://arxiv.org/abs/1711.02725). This is the first paper from PCORI statistical methods award led by James O’Malley. It address a major shortcoming in the literature by developing an instrumental variable method to account for unmeasured confounding (the curse of any observational study) when the outcome is a time-to-event variable. To date, no definitive method has existed. In the course of our research we made a major breakthrough by deriving a theoretical result which showed that adding an individual frailty to a control function procedure known as two-stage residual inclusion solves the problem. While nice in theory, there were doubts as to whether the method would work in practice – typically the addition of an individual level latent variable (or random effect) that is identified parametrically leads to a highly non-robust procedure. However, to our surprise the method proved to be quite robust to violations of the parametric assumption, which we think is due to the fact that the frailty is serving the purpose of freeing up the control function to take care of the unmeasured confounding problem, it does not itself seek to account for the unmeasured confounding problem. The new method is applied to a novel data set that compares the time to patient death under carotid endarterectomy versus carotid artery stenting using the vascular quality initiative registry data.

Paper 2: Analysis of the U.S. Patient Referral Network by Chuankai An, James O’Malley, Daniel Rockmore and Corey Stock, accepted for publication in Statistics in Medicine (https://arxiv.org/abs/1711.03245). In this paper we analyze the entire US Shared Patient Network and various subnetworks for the years 2009–2015. In these networks two physicians are linked if a patient encounters both of them within a specified time-interval, according to the data made available by the Centers for Medicare and Medicaid Services. We find power law distributions on most state-level data as well as a core-periphery structure. On a national and state level, we discover a so-called small-world structure as well as a “gravity law” of the type found in some large-scale economic networks. Some physicians play the role of hubs for interstate referral. Strong correlations between certain network statistics with healthcare system statistics at both the state and national levels are discovered. The patterns in the referral network evinced using several statistical analyses involving key metrics derived from the network illustrate the potential for using network analysis to provide new insights into the healthcare system and opportunities or mechanisms for catalyzing improvements. This paper is the outgrowth of a unique collaboration between Computer Science, Mathematics, Biomedical Data Science and The Dartmouth Institute and is an important stepping stone for future research involving physician networks.