BST 290 Seminar Series
Tuesday, March 3rd, 2015, at 4:10pm, Location: MSB 2112 (Math Dept, 2nd floor)
Refreshments at 3:30pm in MSB 2112
Speaker: Rebecca Hubbard (University of Pennsylvania)
Title: Sources of variability and error in ranking and classification of healthcare providers
Abstract: Assessing and reporting the comparative performance of health care providers is one approach to meaningful quality measurement that is associated with improvements in provider performance. Reliable performance measures are needed for this activity. While prior work has discussed necessary prerequisites to achieve precise estimation of provider performance, there has been little systematic evaluation of the performance of provider profiling approaches for other objectives including classification of providers with respect to fixed benchmarks and ranking of providers. In this talk I will discuss the relative importance of three sources of error, within-provider variability, between-provider variability, and outcome misclassification, in classification and ranking of providers. Using simulation studies, I will demonstrate the relative importance of each of these sources of error in profiling contexts motivated by data from two examples, one on patient satisfaction with behavioral health care and one on the use of Medicare claims to quantify mammography facility performance. I will illustrate the feasibility of provider profiling in each of these settings, one characterized by low between-provider variability and the other by high within-provider variability and conclude with some general considerations on prerequisites for meaningful provider profiling.