Developing a Data-Driven Method for Estimating Provider Penetration and Abusive Billing Practices

This study develops the framework for an abusive-billing detection system that can potentially be used for examining claims of Medicare home health agencies, skilled nursing facilities, hospital outpatient facilities and hospice facilities. This system, which utilizes Medicare claims and other data, is comprised of two components: (1) a single-linkage clustering algorithm that defines market areas based upon patient/provider travel patterns and (2) a logistic regression model that estimates a provider’s expected billing share given the market it is defined to operate in. The system estimates expected shares and compares these shares to the actual share that each provider claims. In this manner, the system identifies outlier-billing providers who, in turn, are envisioned to be correlated with abusive-billing providers. While evaluation of the system requires thorough investigation of providers that are identified to be outliers by the system (which has not been under taken yet), there are preliminary indications that the model can identify providers that excessively bill Medicare, given patient and provider attributes. The system has been developed by HCFA and is called the Service Area and Provider Penetration System (SAPPS).