We analyzed data on insurance coverage, type of employer, and payments for prescription medications from the 2019 Medical Expenditure Panel Survey (MEPS) Full-Year Consolidated Data File and Prescribed Medicines File. The MEPS, described in detail elsewhere,10 is a nationally representative survey that collects detailed information on health care coverage, health care use, and expenditures by and for the civilian, non-institutionalized US population.

MEPS queries respondents about prescription fills (including refills) and obtains information on the amounts paid and sources of payment from their pharmacies (including mail order pharmacies). The sources of payments include those made by patients OOP, as well as insurance payments from both private and public insurers. We identified all filled prescriptions, which MEPS categorized according to the Multum Lexicon Therapeutic Classification Scheme.11 We further subcategorized insulins according to duration of effect/rapidity of onset and examined individual insulin types.

We determined mean payments per prescription for each drug category, as well as for the insulin subcategories. Because MEPS data represents payments to pharmacies, it does not account for rebates that manufacturers often pay to insurers. Hence, we also report figures adjusted downward based on a published estimate of that rebates averaged 21% overall, including 22% for Medicare Part D, 51% for Medicaid, 12% for all private insurers, and 21% for “other” federal and state payers.12 We assumed that no rebates are paid to the Veterans Health Administration (VHA). For a sensitivity analysis, we calculated rebates assuming 27% rebates for all non-VHA payers, based on a detailed study of rebates in Colorado in 2019.13

Figure 1 displays the conceptual model of the flow of funds for outpatient retail prescription drug purchases underlying our analysis. Table 1 summarizes the data sources and calculations used in our analysis.

Conceptual model of flows of payments from individuals, employers, private insurers, and government to pharmacies for outpatient retail prescription drugs.

Our estimates encompass direct as well as indirect taxpayer spending in 2019 for all outpatient retail drugs (i.e., excluding those administered in hospitals, clinics, or physicians’ offices, which are not reported in the MEPS data) combined and for insulins. To calculate direct spending by government health programs, we summed payments to pharmacies by Medicare, Medicaid, the Veterans Health Administration, and smaller federal, state, and local government programs.

We assessed several types of indirect government payments for medications, i.e., taxpayer-funded payments that flow through private insurers. First, to calculate expenditures for civilian government workers (and their dependents) covered by employer-based private insurance, we identified such respondents based on the MEPS variable indicating employer type. We summed private insurers’ payments for drugs on their behalf and deflated these figures by the percentage of premiums paid by workers themselves: 11% for federal workers and their families (based on US Office of Personnel Management figures on the proportion of family and individual coverage, and the weighted average of federal employees’ share of private coverage) and 21% for state and local government employees and their families (based on figures from the Agency for Healthcare Research and Quality).14,15

We used similar methods to estimate the federal government’s 88% share of drug costs paid by the Department of Defense’s TRICARE program on behalf of US-based civilians, e.g., dependents of military personnel and military retirees.16

To estimate indirect tax expenditures due to tax subsidies for workers with employment-based coverage, we calculated tax subsidies’ share of premiums paid to private insurers using previously published methods.17,18 We first obtained the Office of Management and Budget’s (OMB) 2019 estimates of the value of the federal income tax and payroll tax subsidies (which the OMB labels “tax expenditures”) to health care and health insurance.19 While state/local governments that impose income taxes also indirectly subsidize employment-based insurance through taxes, no published estimates of these subsidies are available. Hence, we estimated state and local income tax subsidies to health insurance premiums in 2019 by multiplying the value of the federal income tax subsidy by the ratio of (local + state) income tax receipts to federal income tax receipts. We calculated this ratio using data from the Census Bureau’s quarterly surveys of state and local taxes.20 Together, the value of federal and state tax subsidies is equivalent to 24.52% of expenditures for employer-sponsored private coverage.

To avoid double-counting tax subsidies received by government employees, we adjusted the tax subsidy estimates downward based on the share of employer-sponsored coverage paid by state/local and federal government employers reported in the National Health Expenditure Accounts.21

Finally, we used CMS figures on Advanced Payment Tax Credits for private coverage purchased through the Affordable Care Act (ACA) Marketplace Health Insurance Exchanges to estimate the proportion (76.6%) of exchange coverage (and hence payments for prescription drugs) attributable to public funds.22

To account for prescription drug rebates, we adjusted all figures downward by rebate amounts, as noted above.

All analyses used Stata version 17 procedures that account for the survey’s complex design and MEPS-provided weights to derive nationally representative estimates. The authors’ IRB does not consider analyses of de-identified, public-use data to be human subjects research.