Abstract

Background

Estimates of drug spending are often central to the public policy debate on how to manage healthcare spending in the United States. Nevertheless, common estimates of prescription drug spending vary substantially by source, which can inhibit productive policy dialogue.

Objectives

To review publicly reported estimates of drug spending and uncover the underlying methodological inputs that drive the substantial variation in estimates of prescription drug spending.

Methods

We systematically evaluated 5 estimates of drug spending to identify differences in the underlying methodological inputs and approaches. To uniformly assess and compare estimates, we developed a model to identify the inputs of 3 primary components associated with each estimate: numerator (How is drug cost measured?), denominator (How is healthcare cost measured?), and population (What group of individuals is included in the measurement?). We then applied standardized methodological inputs to each estimate to assess whether variation among estimates could be reconciled. We then conducted a sensitivity analysis to address important limitations.

Results

We found that the 18.8 percentage point range in the publicly reported estimates is predominately attributed to methodological differences. Reconciling estimates using a standardized methodological approach reduces this range to 4.0 percentage points.

Conclusions

Because variation in estimates of drug spending is primarily driven by methodological differences, stakeholders should seek to establish a mutually agreed upon methodological approach that is appropriate for the policy question at hand to provide a sound basis for health spending policy discussions.