Estimation of the Share of Net Expenditures on Insulin Captured by US Manufacturers, Wholesalers, Pharmacy Benefit Managers, Pharmacies, and Health Plans From 2014 to 2018

Key Points Question How are net expenditures on insulin distributed across manufacturers, wholesalers, pharmacies, pharmacy benefit managers, and health plans in the pharmaceutical distribution system? Findings In this cross-sectional study of the distribution of insulin expenditures between 2014 and 2018, the list price of insulin increased, and the net price received by manufacturers decreased. The share of insulin expenditures flowing to manufacturers decreased, and the share flowing to distribution system intermediaries increased, particularly among pharmacies and pharmacy benefit managers. Meaning Results of this cross-sectional study suggest that commercial practices of all distribution system participants influence insulin costs, and that policies to control insulin costs should consider all participants, including pharmacy benefit managers and pharmacies, who share responsibility for rising insulin costs.

Notes: X indicates that the product data was used in year NDC: National Drug Code 1 Product strength is 200 units per mL. 2 Product strength is 500 units per mL. 3 Product strength is 300 units per mL.

Data inputs and their sources
Our analysis relies on data from multiple sources, including Drug Channels Institute, a consulting firm specializing in the economics of the pharmaceutical distribution industry that publishes annual reports with industry data gathered from many other sources. eTable2 lists those inputs, the sources from which we took them, and those sources' source: Methods to allocate gross-to-net reductions across pharmaceutical supply chain entities In this section we describe the data sources and methods used to allocate total gross-to-net reductions or gross-to-net bubble into the amounts represented by flows 2-6 in Figure 1 of the manuscript.
We begin with the total gross-to-net reductions reported by Drug Channels Institute's (DCI) annual Economic Report on U.S. Pharmacies and Pharmacy Benefit Managers in the first row of eTable3. 3 We decompose the total gross-to-net reduction into its constituent parts, using data published by DCI. 3 These data are not available to us in every year, so we use additional sources and methods, as detailed below, to estimate values in missing years.
Following the DCI methodology, we decompose the gross-to-net reduction into five components: (a) spending on patient assistance and copayment support, (b) Medicaid rebates, (c) Medicare Part D rebates, (d) commercial payer rebates, and (e) fees and discounts from manufacturers to drug channel participants, including discounts to providers under the 340B pricing program.

(a) Patient Assistance and Copayment Support
To calculate the total amount spent on patient assistance and copayment support for years 2014 to 2019, we used Exhibit 114 from the 2020 DCI Report ("Manufacturer Spending on Copay Offset Programs from 2014 to 2019") which estimates the value for 2019. 3 These values are reported in row A of eTable3.
For the remaining components, the data needed to decompose the gross-to-net reduction were only available from DCI in some years, so we augmented with data from other sources as follows: Using the values in eTable3, we calculated the share of the total gross-to-net reduction that each component represented in each year and applied that share to the product-specific gross-to-net reductions for insulins, calculated from SSR data in each year to calculate the magnitude of the flows 2-6 in Exhibit 1 of the main manuscript. To calculate the shares, we relied on estimates from a common source (DCI) for both the denominator and the numerator of the estimate. But while we rely on DCI estimates of the aggregate gross-to-net reduction to calculate the shares in eTable3, we apply those shares to product-specific estimates of the gross-to-net reductions from SSR; DCI does not provide gross-to-net reduction estimates at the product level.
The values for manufacturer spending on copay offset programs (row A) correspond to flow 3, total rebates (row E) correspond to flow 2, and fees and discounts from manufacturers to drug channel participants (row G) correspond to the sum of flows 4, 5 and 6 in Exhibit 1 of the manuscript.
To estimate how the amounts in row G are allocated into individual shares for flows 4, 5 and 6 (fees and discounts to wholesalers, PBMs and pharmacies, respectively), we used the gross profit margins reported in the financial statements of the three largest wholesalers for each year, and adjusted the model parameters so that the average gross profit margin of the wholesalers in our model match the observed margin of the wholesalers in each year. We assumed that the remaining share of discounts and fees is distributed evenly among pharmacies and PBMs.

Sensitivity analyses
Because our analysis relies on parameter estimates from a wide range of sources with varying precision, we explored the impact of changes in those parameters, to understand how uncertainty in those estimates may affect our results. eTable7 presents the results of varying key parameters in 8 scenarios: Scenario 1a,b: Vary the share of gross-to-net reductions going to rebates (see row E of eTable3 above) by +/-10%. To fully allocate the gross-to-net reduction across rebates, copay assistance programs and other fees and discounts, we simultaneously changed the remaining share going to the last two categories, as reflected in eTable5, so that the shares sum to 100% in each year. Scenario 5a,b: Assume that, after wholesaler discounts, remaining fees and discounts are split 25%/75% (Scenario 5a) or 75%/25% (Scenario 5b) between PBMs and pharmacies (see eTable6).
Scenario 7a,b: Vary gross margin of health plans by +/-10%. In 2017-18, wholesalers and PBMs make slightly more when the biosimilar is excluded, suggesting that the biosimilar is slightly less profitable than average insulin. Minimal effect on other participants.