Mardi | 2019-02-26
Salle des thèses 16h30 – 18h00
Health status is a key determinant of health care utilization. However, many measures of health status, including precise personalized physiologically based illness severity scores, are imperfect markers of health ability. This complicates many analysis of health care utilization and leaves open questions as to the attributable role of health status. We suggest a method for uncovering the underlying health ability of a child at the time of admission to the pediatric intensive care unit, which consequently provides different results in an analysis of specialized pediatric health care utilization. To do so, we first refine an economic model of pediatric critical care utilization to explicitly account for a child’s underlying health ability. Our theoretical model coincides empirically with the accelerated life regression, a risk adjustment specification common to both clinical health and quality analysis. We extend the accelerated life model using a novel empirical strategy that utilizes the trauma status of a pediatric intensive care unit patient as an instrumental variable. Our results, reported as confidence sets revealing both data and identification uncertainty, suggest that once underlying health ability is accounted for, variations in health utilization attributable to health status change with a clinically and economically relevant magnitude. Furthermore we find that results are sensitive to the choice of health status measures or categorization, which has implications for most types of common risk adjustment. Our theoretical and empirical results also suggest that the clinical uncertainty surrounding underlying health ability is a possible contributor to geographic variation in health utilization.