Comparing alternative methods of targeting potential high-cost individuals for chronic condition management.

Abstract

To compare the performance of predictive models based on health care claims, health risk assessment (HRA), or both in prospectively identifying high-cost individuals with chronic conditions. Participants included 8056 employees who completed an HRA and were enrolled in a health plan for at least 6 months during the 36-month study period. Two-stage multivariate regression identified predictors of claims cost in four age-by-gender groups. All models predicted costs effectively, but the combined model performed better (R2 = 0.198 to 0.309) than either the claims-based (R2 = 0.168 to 0.263) or HRA-based model (R2 = 0.125 to 0.205). An HRA-based predictive model appears to be a reasonable alternative to claims-based predictive models for identifying individuals for chronic condition management outreach. Where practical, a combined model offers advantages of both approaches and meaningfully increases accuracy.