What’s in a Number? It Matters in Health Policy Conversations

Ask a multistakeholder panel of health care leaders what percentage of overall health care expenditures is attributed to prescription medicines, and you’ll likely hear a range of responses, from a low of 10% to a high of nearly 30%.

These considerable differences exemplify how policy discussions about drug spending in the U.S. can often be misinformed. These varied estimates of drug spending aren’t necessarily wrong. The estimates are each targeted for a specific research purpose, none of which are directly comparable. But they do need to be put in the proper context to ensure that they aren’t misinterpreted or used inappropriately, especially by policymakers.

A new study published in Value in Health takes a deeper dive on the data sources and methods behind a wide range of prescription medicine spending estimates cited by the Centers for Medicare and Medicaid Services (CMS), AHIP, Kaiser Family Foundation, the BlueCross BlueShield Association and the Massachusetts Health Policy Commission. The study found important differences in the underlying data and approaches, which account for the variation in the estimates.

Among the five estimates evaluated in this study, there are significant differences in how prescription medicines costs and total health care costs are measured. For example, some estimates of drug spending excluded spending on nonretail medicines or did not account for rebates. One estimate of total health care spending included the Centers for Disease Control and Prevention and National Institutes of Health budgets. Another key driver of differences is variation in populations, ranging from only the privately insured in Massachusetts, to all privately insured individuals, to all individuals in the U.S.

The analysis found that the wide variation in the estimates is driven primarily by methodologic differences. When the estimates were adjusted using a standardized approach for measuring and defining drug costs, total health care costs and populations, the range across the estimates was narrowed from 18.8 percentage points to 4.0 percentage points.

Bringing a wide range of estimates to the table “isn’t conducive to a collaborative and policy discussion,” notes the study. In fact, “none of the source estimates are ideally suited for a broad policy discussion about U.S. health care spending. The estimates are each targeted for a specific research purpose, none of which are directly comparable, and none of which completely represent the share of U.S. health care spending attributable to prescription drugs.”

And as policymakers and other health care stakeholders “discuss issues relating to U.S. health care spending and maximizing the value of the health care dollar, it is imperative that this discussion is informed by relevant and accurate data.” For truly constructive policy conversations, stakeholders should mutually agree upon an estimation methodology that is appropriate for the policy question at hand.

View the full analysis, conducted by researchers from the IQVIA Institute for Human Data Science and the National Pharmaceutical Council, in Value in Health.