Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 322 - Novel Statistical Methods for Current Health Policy Issues
Type: Contributed
Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
Sponsor: Health Policy Statistics Section
Abstract #312193
Title: Predictive Model for Avoidable Hospitalization Event
Author(s): Fei Han* and Ian Stockwell and Morgan Henderson
Companies: The Hilltop Institute at UMBC and The Hilltop Institute at UMBC and The Hilltop Institute at UMBC
Keywords: Predictive model; Avoidable hospitalization; Healthcare
Abstract:

Researchers from The Hilltop Institute at UMBC, sponsored by the Maryland Department of Health, developed a risk prediction model (Pre-AHâ„¢) that helps health care practices identify patients who are at a high risk of having an avoidable hospitalization (AH) or emergency department (ED) visit (AH). The model deployed in October 2019 and produces AH scores monthly for approximately 210,000 Medicare beneficiaries in the state of Maryland.

The tool uses discrete time survival model to predict the patient-level probability of incurring an AH event in the next month using diagnostic, care utilization, procedure, demographic, and environmental covariates from Medicare claims data. The discrete time survival model generates interpretable output while accounting for time-variant covariates. Model validation is measured by a concentration curve and shows high prediction power. For example, in September 2019, the top 10% predicted riskiest patients account for approximately 50% of the patients who had AH, and the top 20% predicted riskiest patients account for approximately two-thirds of the patients who had AH.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program