The Association of the 340B Program with Affordable Care Act (ACA) Premiums: A Longitudinal Analysis from 2018 to 2022

Abstract

We tested whether the 340B program impacts Affordable Care Act (ACA) premiums. Data from 2018 to 2022 was used to establish a baseline for silver benchmark premiums and other key measures, including: the number of active 340B sites per 10,000 people in a county (Hospital Site Density, or HSD), ACA benchmark plan premiums for every county, and measures likely to influence insurance premiums including per capita income, unemployment, and hospital market power. We used a multivariate fixed effects regression that included state and year effects, the explanatory variables, and the county-level HSD for each year. The impact of 340B on ACA premiums was illustrated using 2022 data. We estimate that a 1-unit change in 340B HSD was associated with a 1.1% (95% CI 0.73-1.15) change in the benchmark ACA premium. In 2022, the mean county 340B HSD was 1.63, and the mean Silver Benchmark Plan monthly premium was $500. After adjusting for per capita income, hospital concentration index, and other factors, the mean 340B HSD accounted for 1.8% (95% CI: 1.3-2.1) of the average Silver Benchmark Plan monthly premium, or $8.90 (95% CI $6.50-$10.25) per month, implying a cost in additional subsidies of over $106 per year per subsidized ACA enrollee. The results support that the 340B program is associated with a financially meaningful component of ACA premiums.

PRECIS: We find that higher levels of 340B activity, as measured by the presence of 340B hospital sites, is associated with higher ACA benchmark premiums after controlling for other factors at the county level.

KEY TAKEAWAYS: We use variation in county-level 340B activity to estimate the association of 340B concentration on ACA benchmark premiums. We find that:

• 340B activity is associated with 1.8% of ACA benchmark premiums and over $2 billion per year in higher ACA subsidies per year.

Introduction

The 340B Drug Pricing Program (340B) was created by Congress in 1992 to ameliorate the unintended consequence of the Medicaid best price rule, which reduced the amount of charity care donations to hospitals. Under the terms of the program, drug manufacturers participating in Medicaid and Medicare Part B are required to sell discounted outpatient drugs to 340B covered entities (CEs), including the majority of not-for-profit hospitals and thousands of federal grantees (eg, community health centers, hemophilia treatment centers, etc. According to HRSA, the purpose of the program is to “stretch scarce federal resources as far as possible.” ¹ 340B has grown rapidly over the past decade after significant policy and regulatory changes since the enactment of the Affordable Care Act. Specifically, the number of 340B CEs and child sites has grown steadily as new categories of eligibility and rules regarding existing categories have expanded. There are now thousands of CEs, contract pharmacies, and child sites, often in higher-income areas full of well-insured patients.^2,3 CEs may sell or give away 340B outpatient drugs to patients of the CE either directly, via 340B “child sites” (outpatient clinics, for example), or through contract pharmacy arrangements. CEs earn profits on the spread between the reimbursed price of each drug and the 340B acquisition costs. This gap can be anywhere between roughly 23% and 99%. ⁴ CEs purchased over $66 billion in discounted 340B drugs in 2023, ⁵ and those drugs generated over $124 billion in sales. ⁶

The 340B program ensures that manufacturers provide discounted outpatient drugs to covered entities before the point of sale of a prescription, while various insurers pay for 340B prescriptions at higher (eg, list) prices through reimbursements from Medicare, Medicaid, or through taxpayer-subsidized insurance programs after the point of sale. 340B entities buying medicines at low prices and selling them at higher prices may have financial implications for Affordable Care Act (ACA) plans. Proponents describe 340B as “costless” to taxpayers,^7,8 and the Congressional Budget Office (CBO) does not account for the program in the federal budget. However, given its size and scope, 340B has the potential to introduce direct and indirect costs across the healthcare ecosystem. The link between 340B and ACA premiums may be driven by any number of factors that have previously been associated with the 340B program, such as increased horizontal or vertical provider consolidation, increased hospital price markups, and misaligned incentives for covered entities on site of care or choice of therapy. For example, the acquisition of physicians or physician practices, that is, hospital-physician consolidation, was associated with 340B hospital eligibility. ⁹ Specifically, 340B hospital eligibility was associated with 230% more hematologist-oncologists and 900% more ophthalmologists practicing in the 340B facility compared to ineligible hospitals. Eligible 340B hospitals have incentives to acquire community-based clinics or physician practices, for example, outpatient oncology clinics, because they can buy more drugs at the 340B discounted price and charge for higher-priced services in the hospital outpatient setting. Additionally, 340B covered entity participation has financial consequences for employers and employees. A recent whitepaper modeled this scenario for self-insured employers and demonstrated that lost health plan rebates from 340B program participation contribute to higher commercial employer and worker costs, ¹⁰ which may impact premiums. A previous analysis showed that 340B accounted for $7.8B in healthcare costs for commercial employers and workers from forgone manufacturer rebates, resulting in $1.8 billion in lost federal and state tax revenue. ¹¹ Eligible ACA Marketplace enrollees receive subsidies in the form of premium tax credits, funded by taxpayers, to reduce enrollee health insurance costs. However, given the structure of ACA subsidies, if 340B raises ACA premiums, then taxpayers bear a dollar-for-dollar direct cost for subsidized enrollees. Higher employer costs under 340B correspond to a decrease in taxable employer income and, thus, lower tax revenue for federal and state governments, which may affect their ability to subsidize future ACA premiums. With the number of subsidized enrollees at nearly 20 million in 2024, small increases in benchmark premiums produce significant budgetary costs. ¹² Further, about 1 in 5 ACA enrollees are not receiving subsidies for ACA premiums, leaving such enrollees financially responsible for the potential impacts of the 340B program. ¹³ The objective of this study was to test the association of the 340B program with Affordable Care Act (ACA) premiums and explore the potential financial impacts of subsidized and unsubsidized ACA premiums by county and income level.

Data and Methods

We collected data from the Health Resources and Services Administration’s (HRSA) Office of Pharmacy Affairs (OPAIS), which captures information about every active covered entity and their active child sites (e.g., an infusion center owned by a 340B hospital will have its own 340B identifier). We tabulated the number of active 340B hospital entities and the number of active subsites, considering an entity “active” if it is active for any part of the year. Over time, the number of active sites has increased along with the growth in overall 340B inventory purchases by hospitals, as summarized in Figure 1.

Figure 1.

340B Hospital Sites, Purchases and Reimbursement, 2017 to 2023.

Hospitals become eligible for 340B for various reasons, but 2 important criteria are their not-for-profit status and maintaining a Medicare Disproportionate Share (DSH) Percentage of at least 11.75%. The DSH percentage depends in part on the share of inpatients covered under Medicaid. While 340B eligibility is uniform across the states, other differences indirectly impact the level of 340B activity in each state. For example, some states (e.g., New York) forbid for-profit hospital ownership, thus increasing the likely share of 340B hospitals all things equal; similarly, states have widely varying Medicaid eligibility rules, resulting in different shares of the population covered by Medicaid and, by extension, different DSH percentages. While every state has experienced significant growth in 340B activity in recent years, growth has differed significantly across states and within states, and growth has differed across counties. Shifts in income, unemployment rates, insurance coverage rates, and hospital concentration at the local level all drive differences in ACA benchmark premiums. While the specific level of 340B spending and profit in every county is unknown, we use the number of 340B hospital sites as a proxy for 340B activity at the county level, adjusting for population. We calculated the number of active 340B hospitals and hospital child sites per 10 000 people in each county and called this the hospital site density (HSD).

Similarly, ACA benchmark premiums vary according to local economic, regulatory, and market structure conditions. Every state regulates health plans a little differently; hospital and insurer consolidation, labor market conditions, and broader economic forces change across counties, states, and years. States with higher overall income levels have a deeper tax base that they may rely on to increase benefits or spending, while some states, under pressure from high unemployment, may feel pressure in the opposite direction. States with similar income distributions may have different eligibility criteria and thus may find different shares of their populations in the Medicaid program, creating pressure on spending. Furthermore, states may have chosen to expand or change their programs over time. We test whether growth in 340B activity impacts benchmark premiums after controlling for such factors (e.g., per capita income, unemployment, and hospital market concentration) that are likely to influence insurance premiums.

We estimate a model of ACA benchmark premiums for a 40-year-old nonsmoking male. We conducted a longitudinal analysis composed of silver plan benchmark data for every county in every year, from 2018 to 2022. These years provide consistent data for most marketplaces across states and for the federal marketplace. Benchmark premiums for issuers in states using the federal marketplace were taken from each year’s Medical Individual Market file of the HealthCare.gov Qualified Health Plan Landscape dataset. All data for non-federal marketplace states were taken from relevant public state exchange data available online. The dependent variable in our regression is the natural logarithm of the ACA benchmark premium. Using the natural logarithm approach allows us to interpret percentage changes in ACA benchmark premiums based on 1-unit changes in our independent variables, which include HSD, Herfindahl-Hirschman Index (HHI), and other variables.

Next, we collected data for each county and year for per capita income, the unemployment rate, and the coverage rate for the under-sixty-five population from the Bureau of Labor Statistics. We used Medicare Cost Report data (as collated by the RAND Corporation Hospital Data Sets, available at www.hospitaldatasets.org) to calculate a county-level HHI for each year based on inpatient revenue market shares. We acknowledge that there are many potential denominators for hospital HHI calculations, all of which are highly correlated. The HHI was transformed to a 0 to 1 scale for ease of interpretation. We included state-level fixed effects and year effects to account for factors that we do not measure directly (e.g., the broader trend in 340B utilization). We only included counties with at least 50,000 people in our analysis in order to measure the impact of HHI, as less populous counties generally have only 1 hospital or 1 very dominant hospital. We also excluded Alaska and Hawaii given their unique geographic differences at the county level. Our regression model includes data representing 4,027 county-years, between 824 and 925 counties per year that met our population threshold and for which there was no missing data. We used this data to estimate the association between HSD and ACA benchmark premiums in 2022.

Results

The mean county-level 340B densities and benchmark premiums for 2018 to 2022 in our dataset are shown in Table 1. 340B density increased significantly over the period, reflecting the overall growth in the 340B program. The total study sample from 2018 to 2022 was 4,027 county-years. In 2022, 2,198 (72%) of all counties and 839 (82%) of all counties with at least 50,000 people had at least 1 340B hospital site. The benchmark premium declined over the period, though in the ensuing years, benchmark premiums have increased each year. While our analysis focuses on the benchmark silver plan premium, ACA metal tier prices are highly correlated. We calculated that the correlation coefficient for the average state ACA gold plan and silver benchmark plan premiums was 0.91.

Table 1.

Mean and Standard Deviation for County-Level 340B Hospital Site Density and ACA Benchmark Premium, 2018 to 2022.

Year	Silver benchmark premium		340B hospital site density
	Mean	SD	Mean	SD
2018	$553	$131	1.33	2.1
2019	$555	$141	1.43	2.2
2020	$535	$132	1.49	2.2
2021	$512	$109	1.53	2.2
2022	$500	$109	1.63	2.6

Source. Health Capital Group analysis of ACA benchmark data (see endnote for full details of sources); hospital site density was calculated from the Health Services Research Administration OPAIS Database.

Note. n = 3043; SD: standard deviation.

We find statistically significant impacts across most of the key variables in the regressions, with 340B HSD associated with significantly higher ACA silver benchmark premiums (Table 2). We estimate that a 1-unit change in HSD was associated with a 1.1% (95% CI 0.73-1.15) change in the benchmark ACA premium. In 2022, the mean county 340B HSD was 1.63, and the mean Silver Benchmark Plan monthly premium was $500. After adjusting for per capita income, hospital concentration index, and other factors, the mean 340B HSD accounted for 1.8% (1.3-2.1) of the average Silver Benchmark Plan monthly premium, or $8.90 ($6.50-$10.25) per month, implying a cost in additional subsidies of over $106 per year per subsidized ACA enrollee.

Table 2.

Fixed Effects Regression Model Results. The Dependent Variable is ln (ACA Benchmark Premium).

Variable	Coefficient	t-Statistic
340B Hospital Site Density (HSD)	0.011	5.8
Herfindahl-Hirschman Index (HHI)	0.107	11.5
Personal Income per Capita	3.4E−7	1.8
Under 65 Insurance Coverage Rate	−0.005	−5.5
Unemployment Rate	−0.007	−3.3

Source. Health Capital Group analysis.

Note. R-squared = 0.53; n = 4,027; includes state and year effects.

The results for the other variables are all statistically significant and run in the expected directions. The coefficient on the normalized HHI index is 0.107, and the mean HHI index in our sample is 0.72 with a standard deviation of 0.28. If we apply a 0.53 (±1 SD) difference in the HHI index between a “less concentrated” and “more concentrated” county, the coefficient implies a 5.7% (0.53 × 0.107) higher premium in the more concentrated county. Since the mean ACA benchmark premium is $500, more concentrated markets would have a $28.36 higher premium than less concentrated markets. We find that higher unemployment rates are associated with slightly lower premiums (a 1-point increase in unemployment implies a 0.7% ($3.50 per month) reduction in the local ACA benchmark premium); a 1-point increase in the insurance coverage rate implies a 0.5% reduction in the ACA benchmark premium; and a $10,000 mean increase in personal income levels is associated with a 0.3% increase in the ACA benchmark premium.

Discussion

In 2024, total ACA subsidies were $124 billion (thanks in part to enhanced subsidies which are scheduled to expire after 2025), ¹⁴ with almost 20 million enrollees receiving some amount of subsidy. ¹⁵ ACA subsidies are calculated based on the difference between the benchmark plan cost and an enrollee’s income, so any increase in the benchmark premium results in a dollar-for-dollar increase in subsidies for subsidized enrollees. The 1.8% increase in benchmark premiums associated with 340B activity is borne entirely by taxpayers for subsidized enrollees. Applying the 1.8% estimate to the subsidy base implies that roughly $2.2 billion of the subsidies were associated with overall 340B activity. There are also roughly 1.7 million enrollees in ACA plans who receive no subsidies. Unsubsidized enrollees bear the full cost of higher premiums.

Our primary finding that higher 340B activity is associated with increased ACA benchmark premiums is consistent with recent research that examined per-enrollee total Medicaid costs (including all types of health spending). Although this study is non-peer-reviewed research, year effects and other variables were controlled via a multivariate regression analysis, which showed that Medicaid costs increased faster in states with higher 340B activity, holding other factors constant. ¹⁶ Commercial insurance healthcare prices have also been investigated at 340B hospitals. Researchers found that the prices that large 340B hospitals charge commercially insured patients are, on average, 7.5% higher than those of the same-sized non-340B hospitals, and these 340B hospitals bill 24% more for outpatient procedures. ¹⁷ Although this study found correlation and not causation between 340B status and commercial insurance prices, the increased 340B hospital price markups for patients observed in this study may be another reason for our observed relationship between 340B growth and increased ACA premiums. Similarly, emerging non-peer-reviewed evidence has shown that employer-based insurance premiums increase with 340B activity at the state level after controlling for other factors. ¹⁸ Another set of white papers found that 340B program participation increased drug costs for employer-sponsored and state and local government health plans.^10,19 Specifically, researchers found that employers lose out on drug rebates when prescriptions are filled as 340B eligible, resulting in annual increases of $6.6 billion for all employer-sponsored plans and $1.0 billion for state and local government plans. ²⁵ Although the financial models in these studies relied on several estimates and key assumptions, comprehensive public sources (e.g., Medical Expenditure Panel Survey (MEPS)) and robust proprietary pharmacy and medical claims data were used as model inputs.

The relationship we observed between 340B hospital consolidation and benchmark premiums is aligned with findings by other researchers.^9,20 Recent literature has shown that 340B affects many facets of the healthcare delivery ecosystem.^21,22 For instance, 340B participation has contributed to both provider consolidation ⁹ and hospital consolidation, ²³ which in turn has been shown to raise prices for many healthcare services. ²⁴ Recent research has shown that 340B CE hospitals are more likely to be buyers of other hospitals compared to the national average (70.1% vs 58.7%). ²⁵ Additionally, access to 340B prices impacts decisions made around the site of care for oncology patients (which was shown to increase Medicare spending by over $1100 per patient). ²⁶ One study found that 340B hospitals were reimbursed 6.59 times more than the price they acquired outpatient drugs when compared to independent physician practices. ²⁷ These indirect effects provide ample reason to posit a link between 340B growth generally and overall per-enrollee spending and premiums across all payers, including ACA plans.

Applying our findings nationally is somewhat limited by the fact that we excluded counties with less than 50,000 people in our analysis to sufficiently capture the impact of HHI, as less populous counties generally have only 1 hospital or 1 very dominant hospital. As a sensitivity test, we lowered the population cutoff to 40,000, adding 733 county-years and resulting in a slightly lower estimate for the HSD variable of 0.9% (95% CI 0.7-1.1) vs. 1.1 (95% CI: 0.73 - 1.15). A more significant limitation is that we did not differentiate 340B hospitals and child sites by system affiliation or 340B qualification criteria, and our analysis does not capture differences in local payer mix or underlying health risk factors, which may influence pricing dynamics and insurance premiums. Our analysis also does not provide guidance for the specific causes of the association between 340B activity and benchmark premiums, and further studies are needed to understand the relationship between the 340B program and premiums. Our state-specific and year-specific controls minimize the chance that this is a case of 2 variables simply increasing together even if they are unrelated. In fact, the benchmark premium declined over the period even as hospital site density increased sharply. The observed decline in benchmark premiums may be due to a variety of factors, including but not limited to insurer competition, insurer variation among counties, and unemployment rates, of which employment rates are captured in our analysis. ²⁸

Conclusion

The 340B program has grown in recent years and is intended to help lower-income Americans. Our results suggest that an unintended consequence of 340B is an association with a proportion of ACA benchmark premiums and a commensurate increase in taxpayer-funded ACA subsidies estimated at over $2 billion per year. This evidence adds to recent findings showing similar effects for per-enrollee Medicaid spending and employer-based premiums. Federal policymakers should be aware of this connection so that any changes to the 340B program can be incorporated into the broader healthcare budgeting framework while ensuring the program functions as intended for low-income and uninsured patients.