Predicting High-Cost Pediatric Patients: Derivation and Validation of a Population-Based Model

Lindsey Leininger, University of Illinois at Chicago
Brendan Saloner, Johns Hopkins University
Laura R. Wherry, University of California, Los Angeles

Targeting children at risk for elevated health care utilization is critical in implementing enhanced care coordination initiatives for pediatric populations. Existing algorithms, however, rely upon historical claims data that are unavailable for many patients, particularly low-income children who experience considerable insurance and provider disruption. Parent-reported health (PRH) measures could prove valuable in identifying likely high-risk children, yet there is little evidence on the relative predictive ability of different categories of PRH measures. Our objective was to test the incremental predictive capacity of a wide-ranging series of PRH measures to prospectively identify children in the top decile of health care expenditures. Using data from the 2001 – 2011 rounds of the Medical Expenditure Panel Survey, we examined the discriminative ability of a series of different PRH domains over a baseline model with sociodemographic characteristics. We then propose and validate a screening tool comprised of PRH measures that effectively classifies high-cost children.