Addressing Modifiable Healthcare Spending Through Dynamic Modeling Approaches

This study seeks to improve the ability to use routinely-collected claims data to predict high spending and intervene to prevent it by (1) using group-based trajectory modeling to classify and predict dynamic health care spending patterns over multiple years, (2) isolating and examining patterns of modifiable and non-modifiable spending, and (3) pinpointing specific health care services, events and other potentially-modifiable factors that precede significant spending increases. The work is expected to produce new tools that improve cost predictions and cost-containment efforts, which could be helpful to policymakers and third-party payers.