Activity Number:
|
182
- SPEED: Data Challenge
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 31, 2017 : 10:30 AM to 11:15 AM
|
Sponsor:
|
Government Statistics Section
|
Abstract #325369
|
|
Title:
|
Identification of Potentially Problematic Prescribing to a Vulnerable Population
|
Author(s):
|
Ryan Jarrett* and Richard Epstein and Michael Cull and David Schlueter and Kathy Gracey and Molly Butler and Rameela Chandrasekhar
|
Companies:
|
and Chapin Hall at the University of Chicago and Vanderbilt University and Vanderbilt University and Vanderbilt University and Vanderbilt University and Vanderbilt University
|
Keywords:
|
profiling ;
mixed-effects model ;
Bayesian hierarchical model ;
Shiny ;
Medicaid ;
geocoding
|
Abstract:
|
The Centers for Medicare and Medicaid Services have developed "Risk-Standardized Mortality Rates" to allow for the comparison of hospitals for purposes of reimbursement and quality improvement. We emulate this method to identify prescribers with potentially risky prescribing habits to a vulnerable population. Our method merges administrative data with prescription data that indicates whether or not each prescription has been flagged as potentially inadvisable. Prescriptions are flagged on the basis of predefined criteria relating to dosage or combination with other medications. These data were geocoded and merged with US Census Bureau data to obtain socioeconomic variables. We fit multilevel models to account for within-prescriber correlation, and compare the results of a frequentist mixed-effects model with those of a hierarchical Bayesian model. The results are displayed in an interactive funnel plot using the RStudio package Shiny that allows for the presentation of personalized information regarding each prescriber. This format facilitates inter-prescriber comparisons as well as conversations with prescribers identified as outliers with regard to high-risk prescribing patterns.
|
Authors who are presenting talks have a * after their name.