J Clin Outcomes Manage
Development of a risk assessment tool for predicting pediatric health services utilization
Snyder AM, Willey C, McKenna M, Foley P, Coleman R
Abstract Objective: To determine the utility and predictive value of a survey tool for prospective pediatric health care utilization and risk stratification. Design: Prospective cohort study. Methods: We analyzed 872 questionnaires from the general population of parents with children younger than age 18 years and linked responses to utilization and claims data. We examined the association of demographic variables, health status indicators, family history variables, and health utilization measures as predictors of health care costs in the subsequent year. Significant variables were entered into a logistic model to determine which combination best predicted children in top decile of health care costs. Results: Three variables (number of physician visits in the previous 12 months, whether the child has breathing problems, and whether the child takes medications regularly) correctly detected 60% of children with claims costs in the highest decile of the sample (sensitivity) and correctly excluded 70% of children in the lower 90% cost group (specificity). The positive predictive value was only 19%, but this low value results from the small number of children with high health care costs rather than the robustness of the statistical model. Conclusion: The algorithm can serve as first step in a process to determine if a child is at risk and needs further assessment for case management intervention. The 3-question model for a pediatric risk assessment screening tool is unique and not previously reported in the literature.
Search the Turner White index to find abstracts of articles published in JCOM.