Framework for assessing pharmacy value

Excess spending and poor quality in the US healthcare system has led to proliferation of performance-based payment models. These models have the potential to enhance value by creating a meritocratic system whereby providers delivering the best patient care are rewarded, while providers failing to provide such care are given incentives to improve. However, early experience suggests that unless these systems are appropriately designed, payments can be withheld from high performers, bonuses paid to low performers, and health disparities can be worsened. Performance-based payments are new to community pharmacies, and opportunity exists to strengthen pharmacy value measurement and potentially avoid problems observed with other performance-based payment models.

This article describes the process by which a framework to assess community pharmacy value was developed, then applies the framework to produce a draft composite pharmacy performance measure. The pharmacy value framework addresses potential shortcomings of existing community pharmacy performance measures through four key principles: 1) theory-based quality and spending measures, 2) scoring which accounts for measure reliability, 3) full risk-adjustment, and 4) a value matrix to identify high and low value pharmacies. Based on these principles, a draft community pharmacy composite performance measure was developed, and was successful in dividing community pharmacies into high, typical, and low value categories.

By using this framework to develop future composite measures, payers may find closer alignment between performance-based payments and actual pharmacy performance. This early work is intended to encourage further research into the establishment of a scientifically firm foundation for pharmacy performance measurement. More testing is needed to determine reliability, validity, and comparative superiority of any composite measure derived from this framework before it is used to support performance-based pharmacy payment models.