Using Predicted Therapy Visits in the Medicare Home Health Prospective Payment System

Abstract

A criticism of Medicare’s home health prospective payment system is its partial reliance on cost-based reimbursement of therapy services provided by home health agencies (HHAs) to Medicare fee-for-service (FFS) beneficiaries, potentially overincentivizing the provision of therapy services. Using Medicare FFS home health claims and assessment data, we estimated a model to predict therapy use as a proxy for clinical need and replace actual therapy use with the prediction in the home health payment system. We estimated a $1.178 billion (95% confidence interval, $1.171-$1.184) decrease in home health payments relative to levels under the current system. The majority of the decrease was due to the model predicting fewer high therapy episodes than actually occurred. It may therefore be more appropriate to predict both therapy and nontherapy use, requiring an overhaul of the current system.