On Monitoring Outcomes of Medical Providers

• Activity Number: 462
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: International Chinese Statistical Association
• Author(s): John David Kalbfleisch*+ and Robert A Wolfe
• Companies: University of Michigan and University of Michigan
• Keywords: fixed effects ; random effects ; empirical null ; flagging ; confounding

Abstract

An issue of substantial importance is the monitoring and improvement of health care outcomes for providers such as hospitals, dialysis units or surgical wards. Statistical tools are needed to aid centers in instituting and evaluating quality improvement programs and also to aid overseers and payers in identifying and addressing sub-standard performance. In the latter case, one aim is to identify facilities whose outcomes are outside of normal expectations; such facilities are flagged and perhaps audited for potential difficulties or censured in some way. Methods in use are based on models with fixed or random effects. We assess the merits of these approaches when the patient outcomes of interest arise from a linear model. We argue that methods based on fixed effects are better suited for identifying providers with extreme outcomes and avoid confounding issues that can arise in the random effects models. Finally, we consider methods for flagging based on the Z-scores arising from the fixed effects model, but which account in a robust way for the intrinsic variation between facilities. We provide an illustration in monitoring survival outcomes for dialysis facilities in the US.