The prescription drug components of the Consumer Price Index (CPI) and the Producer Price Index (PPI) measure recent and historical drug price inflation, and provide the basis for projecting future price trends and health care expenditures. They also serve as important inputs in the construction of the National Health Accounts. However, there are concerns about the adequacy of these price indexes in light of recent changes in the structure of the market for prescription drugs. To begin with, there has been a large growth in the reliance on generics as Medicare Part D and private insurance have increased the financial incentives to adopt generics. In 2016, it was estimated that unbranded generics accounted for 85 percent of drug prescriptions, compared with 50 percent in 2005.1 With an unprecedented number of major drugs coming off patent in recent years, mismeasurement in the timing and the extent of the transition to generics could have large consequences.
Furthermore, the BLS’s ability to obtain price quotes for the CPI program has been affected by the shift to prescription drug insurance coverage by public and private third-party payers (TPPs). These electronic transactions require information on the identity of the TPP, the beneficiary’s identification number, and specific provisions of the benefit plan, such the copayment rate, information that the pharmacy cannot provide to BLS field staff. As a result, BLS must rely disproportionately on cash transactions that comprise only about seven percent of the retail market, raising concerns about the extent to which they are representative of the overall drug market.2
In “An evaluation of the CPI Indexes for prescription drugs” (PDF), Barry Bosworth, John Bieler, Michael Kleinrock, Eric Koepcke, and Ernst R. Berndt compare the measures of prescription drug prices reported in the CPI and PPI programs with indexes computed from an alternative set of price data compiled by the IQVIA Institute for Human Data Science (IQVIA). There are some fundamental differences in that the CPI aims to measure only the out-of-pocket costs to consumers and the BLS does not have full access to retail transactions that are embedded in prescription drug plans. The analysis is thereby limited to a comparison with the cash transactions of the IQVIA data. The PPI is restricted to drugs produced within the United States, excluding imports from both other countries and Puerto Rico, a distinction that is not identified in the IQVIA data. Thus, the analysis is largely limited to a comparison of the CPI data with that of IQVIA.
They find the increase in the IQVIA index substantially exceeds that of the CPI, and the difference remains when the comparison is restricted to a matched set of prescription drugs. In the analysis of the component indexes, we find that the discrepancy between the IQVIA and CPI is heavily concentrated in two therapeutic categories of cardiovascular-renal and skin/mucous membranes. The differences are surprising because the two data sets appear to contain a similar distribution of drug sales.
Read the full report here.
This research was supported by the National Institute of Aging of the National Institutes of Health under grant number R01 AG043560 to the National Bureau of Economic Research. The analysis was conducted with restricted access to Bureau of Labor Statistics (BLS) data, but the views expressed do not necessarily reflect those of the BLS. Any opinions expressed herein are those of the authors and do not necessarily represent the views of the National Institutes of Health, the Bureau of Labor Statistics, IQVIA Institute for Human Data Science, the National Bureau of Economic Research, or the Brookings Institution. All results have been reviewed to ensure that no confidential information is disclosed.
- A recent overview of the major issues involved in the construction of price indexes for pharmaceuticals is provided by Aizcorbe and Nestoriak (2012). See also Berndt and others (2000). A discussion of the evolution of the drug market in response to passage of the 1984 Waxman-Hatch Act is available in Berndt and Aitkin (2011).