Methods of Monitoring Outcomes of Medical Providers

• Activity Number: 161 - Practical and Methodological Issues Related to Profiling Hospitals in the United States
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): John Kalbfleisch* and Zhi He and Lu Xia and Yanming Li
• Companies: University of Michigan and University of Michigan and University of Michigan and University of Michigan
• Keywords: fixed effects; random effects ; hierarchical models; empirical null

Abstract

We consider methods for monitoring patient outcomes of health care providers including methods based on fixed effects, random effects and the empirical null. Fundamental to this discussion is the nature of the between facility variation and the extent to which it can be attributed to the quality of the care delivered. We note that the usual methods of fixed effects and random effects are essentially based on an assumption that all the inter facility variation is due to quality of care, and within this framework, fixed effects provide a better tool for identifying outlying providers. More generally however, the variation between providers is often primarily due to factors not under the providers' control and profiling methods such as those based on the empirical null or funnel plots are more appropriate. We also, however, consider situations in which a predetermined proportion of the total inter-provider variation is due to quality and the implications for profiling in this case. The methods and concepts are illustrated in monitoring patient outcomes among dialysis facilities in the United States.