Online Program

Return to main conference page

Thursday, January 11
Thu, Jan 11, 2:00 PM - 3:45 PM
Crystal Ballroom F
Use of CMS Data and Tools to Help Researchers

Identification of Outlying Prescribers to a Vulnerable Population (304242)

Molly Butler, Vanderbilt University 
Rameela Chandrasekhar, Vanderbilt University 
Michael Cull, Vanderbilt University 
Richard Epstein, Chapin Hall at the University of Chicago 
Kathy Gracey, Vanderbilt University 
*Ryan Jarrett, Vanderbilt University 
David Schlueter, Vanderbilt University 

Keywords: profiling, mixed-effects model, Bayesian hierarchical model, Shiny, Medicaid, geocoding

Risk standardization is an essential part of any cross-physician comparison due to differing case-mix. In this paper we emulate The Centers for Medicare and Medicaid Services’ (CMS) method of calculating "Risk-Standardized Mortality Rates" (RSMR) as a tool for identifying prescribers with potentially problematic prescribing habits to a vulnerable population. Our method combines prescription data with demographic and risk-assessment variables from administrative sources and socioeconomic variables obtained through the American Community Survey. Following the CMS’ methodology, we model the probability that each prescription will raise a red flag, according to predefined clinical criteria, using a mixed effects model to account for within-prescriber correlation. From the model results, we calculate a “Risk-Standardized Red Flag Rate” (RSRFR) for each prescriber and establish control limits for the identification of outlying prescribers with regard to high-risk prescribing patterns. These results are compared and contrasted to those of a Bayesian hierarchical model and displayed in an interactive funnel plot using the R package Shiny.