Examining a Predictive Payer Model for Unplanned Health Care Utilization

Compared with claims data alone, merging claims data with Patient-reported Outcomes Measurement Information System (PROMIS) data only slightly improved the performance of a payer model for predicting unplanned health care utilization and cost, according to study findings published online ahead of print in Medical Care. 

Corresponding author Suzanne Kinsky, MPH, PhD, of the UPMC Center for High-Value Health Care in Pittsburgh, and coauthors conducted a retrospective analysis of patients insured by a private health plan and who were seen at 18 neurology clinics affiliated with the plan. As part of routine care, the neurology clinics collected data for the following PROMIS domains: anxiety, cognitive function abilities, depression, fatigue, pain interference, physical function, sleep disturbance, and ability to participate in social roles and activities. 

For the study, researchers looked at associations of covariates to health care utilization and cost using claims data alone, as well as claims data combined with the first PROMIS data collected between June 2018 and April 2019. 

Compared with claims data alone, area under the receiver operating characteristic curve values for unplanned health care utilization were slightly higher and Akaike information criterion for unplanned health care costs criteria values were similar for combined claims and PROMIS data, researchers reported. The combined claims and PROMIS model had slightly higher sensitivity and equivalent specificity with claims data alone in predicting the top 15% of unplanned care costs. 

Although combined claims and PROMIS data offered some predictive improvement, researchers concluded it was “likely not to an extent that indicates improved practical utility for payers.” 

Reference:

Kinsky S, Liang Q, Bellon J, et al. Predicting Unplanned Health Care Utilization and Cost: Comparing Patient-reported Outcomes Measurement Information System and Claims [published online ahead of print, 2021 Jun 29]. Med Care. 2021;10.1097/MLR.0000000000001601. doi:10.1097/MLR.0000000000001601