Probability of Being Above a Percentile Cut-Point as a Metric for Facility Profiling and Payment

• Activity Number: 635
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section for Statistical Programmers and Analysts
• Author(s): Michael Shwartz*+ and James F. Burgess, Jr and Erol A. Pekoz and Cindy L. Christiansen and Boris Kader and Amy Rosen and Dan Berlowitz
• Companies: Boston University and Center for Organization, Leadership and Management Research and School of Management, Boston University and School of Public Health, Boston University and Center for Health Quality, Outcomes and Economic Research and Center for Organization, Leadership and Management Research and Center for Health Quality, Outcomes and Economic Research
• Keywords: profiling ; performance measurement ; composite measures

Abstract

In pay-for-performance and profiling, high performing facilities are usually identified because they are above some percentile cut-point (e.g., in the top 20 percent) or because the lower level of their 95% confidence (or credible) interval is above the population average. The first approach results in an unrealistic distinction between facilities just above the cut-point and those just below; the second approach ignores the often significant overlap in confidence intervals between facilities whose lower CI level just misses the population average and facilities whose lower level just covers the average. We propose, as an alternative basis for ranking and rewarding facilities, a continuous measure: the probability a facility is above some policy-identified cut-point. These probabilities can be easily calculated when Markov Chain Monte Carlo methods are used to estimate performance. We illustrate the impact of this approach on pay-for-performance payment using a composite measure of performance calculated from data on 28 quality indicators from 112 VA nursing homes. We also use the metric to evaluate the impact of risk adjustment on payment.