Refining Diagnosis-based Risk Adjustment Model with Prescription Information

Objective：Using risk adjustment to set personal premium reasonably can avoid the problem of risk selection and ensure the equity of access to care. By appropriate risk adjusters, a risk adjustment method can not only predict personal medical expense but also reasonably reflect medical need. Diagnosis–based adjusters and prescribed drugs adjusters have attracted the research attention for their rich clinical messages. Results of early studies in Taiwan demonstrated outstanding predictability and potentiality. Reviewing medical expense structure in Taiwan, one-third of outpatient medical expense is spent on prescription drugs, and chronic disease prescriptions account for the main portion in drug expenses. This study intends to refine diagnostic risk adjusters with prescription information to improve predictability of risk adjustment models in Taiwan.
Data and methods：With detailed enrollment data, medical expense data, diagnostic and prescription data of contracted institution in 2000 and 2001, this study constructed health-based risk assessment models. Using PCG adjusters, Taiwan’s outpatient and inpatient diagnosis-based adjusters, and the adjusters combining diagnostic and prescription data, this study constructed nine risk adjustment models and evaluated the predictions to medical expenses of individuals and specific subgroups.
Principal findings：More clinical information improves the predictability. In all models, the models with the TSAGs and the adjuster combining diagnostic and prescription data outperformed other methods, and the combined adjusters slightly outperform TASG adjusters, Compared with either diagnostic or prescription information, combined information improved the predictability of risk adjustment models. In particular, in specific chronic disease groups, the combined adjusters demonstrated a better predictive ratio.
Conclusion：Using prescription information to refine diagnosis-based risk by splitting it into refined cost group can improve the risk adjustment model not only in all models but also predictive ratio of specific chronic disease group, which helped to avoid risk selection. This study didn’t modify prescribed drug adjusters by local condition. Future research on modifying by local prescription habits and combining with secondary diagnostic information is suggested.