Excess HIV Infections and Costs Associated With Reductions in HIV Prevention Services in the US

Key Points

Question

What are the likely changes in HIV transmission and health care costs if policy changes result in decreased pre-exposure prophylaxis (PrEP) use in the US?

Findings

This economic evaluation estimated that if PrEP coverage declined modestly—approximately 3% annually—8618 new infections would fail to be averted in a decade, and the estimated lifetime medical costs of these infections would be $3.6 billion.

Meaning

These findings suggest that changes in policies, especially those that increase out-of-pocket costs of PrEP, risk reversing progress toward ending the HIV epidemic, accruing avertable HIV infections, and incurring additional costs for medical care from avoidable infections.

This economic evaluation uses previously published ecological data on the associations between coverage for pre-exposure prophylaxis (PrEP) and new HIV diagnoses to estimate excess infections that would occur with diminished public insurance programs and policies that limit out of pocket costs for preventive care for HIV prevention and PrEP.

Abstract

Importance

Pre-exposure prophylaxis (PrEP) is a proven effective intervention to reduce the risk for HIV infection. Critically, changes in policies that lead to increased out-of-pocket PrEP costs or that decrease access to proximate PrEP locations could reduce PrEP coverage, resulting in excess HIV infections and costs.

Objective

To estimate how decreases in PrEP coverage that would be likely results of federal policy changes may be associated with new HIV infections and their costs.

Design, Setting, and Participants

In this economic evaluation, US population-based data sources were used to describe population-level PrEP use and new diagnoses under different hypothetical changes in PrEP coverage. Estimations of excess HIV infections under different policy changes were conducted using parameters from a previously published ecological model of the association between PrEP coverage and new HIV infections. Data on PrEP prescriptions from January 1, 2012, to December 31, 2022, and estimates from a previously reported original clinical model, which described decreases in HIV diagnoses associated with increases in population PrEP use, were analyzed. Data were analyzed from February 25 to May 23, 2025.

Exposure

Change in PrEP coverage.

Main Outcomes and Measures

Estimated change in new HIV infections under different assumed reductions in PrEP coverage, costs of treatment for avoidable HIV infections, and net costs of avoidable infections after accounting for costs of PrEP medications. We also estimated increases in new HIV diagnoses associated with hypothesized levels of decreasing PrEP coverage, and the costs of treating infections not averted.

Results

In 2012, there were 9565 PrEP users in the US; they were predominately male (5857 [61.2%]), and 7109 (74.3%) were aged 25 to 54 years. By race and ethnicity, 1235 PrEP users (12.9%) were Hispanic, 1857 (19.4%) were non-Hispanic Black, and 5404 (56.5%) were non-Hispanic White. Based on analyses of data from a census of US PrEP users including 17 333 732 person-years of time using PrEP, an absolute 3.3% annual reduction in PrEP coverage during the next 10 years (eg, 2023 to 2033) would result in 8618 avoidable HIV infections, with lifetime medical costs of $3.6 billion (discounted) for treatment.

Conclusions and Relevance

In this economic evaluation estimating effects of the possible health care policy changes on HIV transmission, findings suggest that even modest reductions in PrEP coverage would result in thousands of avoidable HIV infections and billions of dollars of increases in net health care costs.

Introduction

Antiretroviral pre-exposure prophylaxis (PrEP) to prevent HIV acquisition is a mainstay of HIV epidemic control in the US.¹ PrEP, when taken as directed, reduces the risk of acquiring HIV by as much as 99%.² However, progress to achieve high levels of PrEP use by people who might benefit from it has been slow: in 2022, only 36% of those with indications for PrEP were taking it.^3,4 PrEP uptake is driven by multiple factors, including perceived risk of HIV infection,⁵ availability of low-cost or no-cost PrEP, access to associated PrEP services,⁶ proximity of knowledgeable clinicians and clinics that prescribe PrEP,^7,8,9 and perceived efficacy of PrEP.^5,10

Evidence from clinical trials has documented the efficacy of PrEP for reducing HIV infections in people who take it,² and in ecological studies, states with high PrEP coverage among people with indications experienced correspondingly lower rates of new HIV infections.¹¹ Although these associations were ecological, they were directionally consistent with the known individual efficacy of PrEP. A recent study that included several of the current investigators used ecological data to demonstrate that increase in PrEP coverage in the US by every 5 per 100 persons with indications is associated with a subsequent 3.4% decline in HIV diagnosis rate, after adjusting for state-specific rates of viral suppression.¹²

Despite this demonstrated association between increases in PrEP coverage and decreases in new HIV diagnoses, changes in PrEP coverage during the past decade have been gradual, and there are substantial differences in the extent of PrEP coverage among US states.¹² At the individual level, PrEP use is 4 times more likely among patients with health insurance,¹³ and on a jurisdictional level, equitable PrEP use is substantially higher in states with Medicaid expansion or PrEP drug assistance programs and highest in states with both.⁶ This suggests that policies related to access to health care and affordability of PrEP likely influence the extent of PrEP coverage and, in turn, risks for new HIV infections in the population.

In early 2025, amid substantial cuts to US Health and Human Services public health staff and programs, there was a discussion about reducing public funding for PrEP programs (ie, increasing out-of-pocket costs and copayments).¹³ Public insurance programs have historically supported the medical costs of nearly 20% of PrEP users, including an estimated 12% of PrEP users who have historically relied on Medicaid to pay for PrEP.¹⁴ In July 2025, legislation was passed that is estimated to result in 600 000 to more than 2 million people losing health insurance coverage in the US.¹⁵ Thus, people currently insured through some public programs are likely to lose coverage for PrEP, an intervention that is cost-saving when used in populations with indications for treatment.^16,17 Beyond insurance vulnerabilities to supporting PrEP care, cuts to staffing and programs at the Centers for Disease Control and Prevention in the first quarter of 2025,¹⁸ cuts targeting HIV-focused research, including PrEP research funded by the National Institutes of Health during the same period,¹⁹ and cuts to nearly all global prevention funding—including to USAID^20,21—were early indications that public sector support for HIV prevention programs and research was under attack. The recent passage of the 2025 US budget promises substantial cuts to public support of prevention services through Medicaid.¹⁵ Given the important contributions of (1) public-sector supports for PrEP programs, (2) federal guidelines limiting out-of-pocket costs for PrEP services, and (3) public support for community clinics that provide HIV prevention services, including PrEP services, such reductions in public financial support would likely decrease PrEP use. For example, declines in PrEP use might occur as a result of discontinuation or reduction of Medicaid coverage or PrEP drug assistance programs or as a result of reduction or elimination of Affordable Care Act provisions that minimize out-of-pocket costs for PrEP, associated laboratory testing, and other components of the PrEP regimen beyond the medication.^6,10 Importantly, 88% of PrEP users in 2022 were benefiting from low-cost or no-cost PrEP through these programs.²² Declines in PrEP use might also occur due to changes in clinician behavior²³; disinvestment in PrEP promotion campaigns; closing of publicly supported PrEP provision sites, which would lead to increased commute times to obtain PrEP²⁴; or discontinuation of education programs that encourage primary care clinicians to routinely offer PrEP to patients.²³

To estimate the possible effects on HIV incidence and costs associated with withdrawing public funding and/or support for PrEP programs, we used previously published ecological data on the associations between PrEP coverage and new HIV diagnoses to estimate excess infections that would occur with diminished public support for HIV prevention and PrEP (ie, infections not averted because of decreased PrEP uptake).^11,12 We also assessed published estimates of PrEP and treatment costs to estimate net costs of HIV treatment because of infections not averted under scenarios of decreased PrEP use.²⁵

Methods

In this economic evaluation, we estimated the association between future changes in PrEP coverage (ie, the proportion of people with an indication for PrEP who are taking it²⁶) and subsequent new HIV diagnoses by using parameters from a recent model of the clinical association between PrEP coverage and new diagnoses.¹² Sullivan et al¹¹ previously reported methods to evaluate the relative change in HIV diagnosis rates per year for the period from January 1, 2012, to December 31 2022, using Poisson log-linear models from the generalized linear mixed models family, with the yearly new HIV diagnoses as the outcome and the log of the yearly corresponding state population denominators as an offset. We used year as an independent continuous variable to generate the estimated annual percent change in HIV diagnosis rates and 95% CIs.²² The same modeling method was used to generate an estimated decade percent change outcome (ie, an estimated cumulative percent change during the 10-year study period). The model did not examine other predictive factors associated with change in new diagnoses but controlled for state- and year-specific viral suppression.¹² Our data did not include any data at the individual level nor was it a clinical investigation as defined in the federal regulations; the investigators determined the analysis to be non–human participant research using the Non–Human Subjects Research Determination Electronic Form provided for this purpose by the Emory University Institutional Review Board. We followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guidelines for economic evaluation.

The associations of PrEP coverage with changes in HIV diagnosis rates were previously assessed by some of our investigators,¹² including PrEP coverage as a continuous predictive factor in Poisson log-linear models. The estimated percentage change in HIV diagnosis rates for an increase of 1 per 100 persons with indications and for other levels of increase in PrEP coverage was reported.¹² Analyses of the association of PrEP coverage with change in HIV rates were adjusted for contemporary state-level viral suppression rates. Models incorporated normally distributed random effects to account for between-state variation and when year was not a fixed effect, and included random effects for time to account for within-state variation. To assess whether it was appropriate to analyze the log-linear trend in HIV diagnoses without inflection points and for data visualization purposes, we used Joinpoint regression analysis.²⁷ We found no inflection points and used one slope coefficient for the entire period.

For the present analysis, we evaluated the potential impact of policy changes that would result in decreasing PrEP use in the upcoming years. We parameterized the model using the previously reported Poisson log-linear model of the national-level ecological association between PrEP coverage and new HIV diagnoses. To present the counterfactual scenarios of changes in new HIV diagnoses in future years, we used data from the original clinical model¹² and reversed the sign of the negative parameter estimate (eg, the coefficient representing declines in HIV diagnoses associated with increasing PrEP use) to obtain the estimated increases in new HIV diagnosis rates associated with several hypothesized levels of decreasing PrEP coverage. The comparator scenario was derived from the original model (observed data from 2012 to 2022): we calculated the predicted percent increase in HIV diagnosis rates for a 1-unit (1 per 100 persons with indication) decrease in PrEP coverage compared with the original model.¹¹ The Figure illustrates how hypothetical year-to-year changes in PrEP coverage would be estimated to affect new diagnosis rates over the study period.

Figure. Estimated Yearly New HIV Diagnosis Rates at Observed Levels of Pre-Exposure Prophylaxis (PrEP) Use and at Counterfactual Levels of Decrease in PrEP Coverage.

Based on clinical evidence from observed national new HIV diagnosis rates between 2012 and 2022 (excluding COVID-19 years 2020 and 2021), the year-by-year variation of increased new HIV diagnosis rates at different levels of potential absolute annual decreases in PrEP coverage is depicted. The differences shown in each year are the estimated effect of PrEP according to the assumptions described. New HIV diagnosis rates at existing levels of PrEP use (no changes related to policy) are based on the historical HIV diagnosis rate in 2012 (15.2 per 100 000 population) through 2022 (13.3 per 100 000 population).

We applied these estimates to 3 possible future scenarios of PrEP coverage. Scenario 1 reflected an assumption that the increases in PrEP coverage observed during the past decade would be reversed over the following decade. We calculated the degree of decrease in PrEP coverage that would result in the reversal of the estimated annual percent change in our original estimation, with the mean calculated across a decade (2012-2022) and adjusted for viral suppression rates at the state level: a 3.3 per 100 mean absolute annual decrease in PrEP coverage approximated a reversal of gains from the prior decade (eg, a 2.3% annual increase in HIV diagnosis rates during the prior decade). We then applied these percentages to the new HIV diagnosis rate and PrEP coverage, reported in 2022 as baseline values. We calculated the projected number of persons not receiving PrEP during the next decade given the reversal of prior decade gains in PrEP coverage and the resulting number of HIV infections not averted. Hypothetical scenarios 2 and 3 followed the same approach, but at 2 alternative levels of absolute annual decreases in PrEP coverage: 10% absolute annual decreases for scenario 2 and 2% absolute annual decreases for scenario 3. No significance testing was performed; the models represent population-based data.

Cost analyses were performed based on our counterfactual infection case predictions. To estimate net costs due to decreases in PrEP investment during the 10-year analytic horizon, we applied current estimates of HIV-related treatment costs to estimate costs incurred because of HIV infections not averted by PrEP.²⁸ To calculate net incremental costs, we subtracted the savings realized from providing PrEP to fewer people. We used undiscounted annual HIV-related medical costs per person living with HIV²⁸ to estimate the increase in 10-year HIV medical costs associated with unprevented HIV infections. The 10-year horizon was chosen because there can be a considerable lag among HIV infection, diagnosis, and treatment (or adverse outcomes for untreated infections). Discounted analyses indicated the present value, reflecting the time value of money; undiscounted analyses represented the total projected cash flows associated with the scenario. The corresponding net costs for a decade were calculated by subtracting estimated medical costs for PrEP for people who were assumed to not be taking PrEP under an adverse policy scenario. Similarly, we used lifetime HIV-related medical costs per person living with HIV²⁸ (2019 US $420 285, discounted at 3%; or 2019 US $1 079 999, undiscounted) to estimate the increase in lifetime HIV medical costs associated with unprevented onward infections. The health care sector perspective is used in these analyses.

Data were analyzed from February 25 to May 23, 2025. All analyses were conducted using SAS software, version 9.4 (SAS Institute Inc) and the Joinpoint Regression Program, version 5.0.2 (National Cancer Institute). The Poisson log-linear models were performed using the GIMMIX procedure in SAS.

Results

In 2012, there were 9565 PrEP users in the US; they were predominately male (5857 [61.2%] compared with 3663 [38.3%] female and 45 [4.7%] missing). By age, 1234 (12.9%) were younger than 24 years; 2750 (28.8%), 25 to 34 years; 2324 (24.3%), 35 to 44 years; 2035 (21.3%), 45 to 54 years of age; 1200 (12.5%), 55 years or older; and 45 (0.5%) missing. By race and ethnicity, 1235 PrEP users (12.9%) were Hispanic, 1857 (19.4%) were non-Hispanic Black, and 5404 (56.5%) were non-Hispanic White. Other race and ethnicity classifications were suppressed due to small or unstable numbers. Assuming that negative changes in PrEP programs reduced PrEP coverage yearly by 3.3 per 100 individuals with indications during the next decade (eg, effects of discontinuing interventions to increase awareness of PrEP, increasing out-of-pocket costs of PrEP, rolling back the expansion of health coverage and drug assistance programs, and/or decreasing retention in PrEP care), HIV diagnosis rates were estimated to increase by a mean of 2.3% (95% CI, 2.2%-2.4%) annually. This modest decrease in PrEP coverage would be expected to erase all the reductions in HIV transmissions achieved during the past decade. In this (base-case) scenario 1, 8618 new infections (95% CI, 7548-9691) would fail to be averted in a decade because of lowered PrEP uptake; the estimated lifetime medical costs of these infections would be $3 622 016 130 (discounted) and $9 307 431 382 (undiscounted). The net costs during a 10-year period in this scenario would be $1 430 864 206 (discounted) and $1 922 961 843 (undiscounted). We also evaluated a scenario of larger cuts to PrEP programs, with estimates of yearly decrease in PrEP coverage of 10 per 100 individuals with indications for PrEP. This scenario 2 resulted in 26 873 HIV infections (95% CI, 23 467-30 307) that failed to be averted in a decade with correspondingly higher lifetime medical costs of $11 294 318 805 (discounted) and $29 022 813 127 (undiscounted) (Table).

Table. Estimated Effects of Reduced PrEP Coverage on HIV Diagnoses and Costs During the 10-Year Study Period^a.

Scenario	Annual change in PrEP coverage, %b	Annual HIV infections not averted, No. (95% CI)	Increase in lifetime HIV medical costs, US $c		Increase in HIV medical costs for 10 y, US $d		People not receiving PrEP, No.	Decrease in PrEP medical costs for 10 y, US $e		Net costs for 10 y, US $
			Discounted	Undiscounted	Discounted	Undiscounted		Discounted	Undiscounted	Discounted	Undiscounted
1: Base casef	−3.3	8618 (7548-9691)	3 622 016 130	9 307 431 382	1 856 448 093	2 494 911 000	51 251	469 768 562	631 329 666	1 386 679 531	1 863 581 334
2: High reduction in PrEP coverage	−10	26 873 (23 467-30 307)	11 294 318 805	29 022 813 127	5 788 852 357	7 779 733 500	156 105	1 430 864 206	1 922 961 843	4 357 988 151	5 856 771 657
3: Low reduction in PrEP coverage	−2	5226 (4580-5874)	2 196 409 410	5 644 074 774	1 125 759 774	1 512 927 000	31 221	286 172 841	384 592 369	839 586 933	1 128 334 631

Abbreviation: PrEP, pre-exposure prophylaxis.

^a Discounted analyses applied a 3% per year discount rate to reflect that money spent or saved in the future is valued less. Costs are in 2019 USD.

^b Per 100 mean absolute annual decrease in PrEP coverage.

^c Assumes $1 079 999 (undiscounted) in lifetime HIV-related medical costs per person living with HIV.

^d Assumes $28 950 (undiscounted) in annual HIV-related medical costs per person living with HIV per year.

^e Assumes $1231.84 (undiscounted) in PrEP-related medical costs per person per year; this includes the cost of PrEP drugs (with an assumed increasing proportion using generic drugs), ancillary services, and outreach and/or PrEP navigation.

^f Approximates a reversal of gains in the prior decade.

Discussion

Nearly 38 000 new HIV infections occurred in the US in 2022.²⁹ At present, no vaccine or cure is imminent, so PrEP is a necessary mainstay of public health prevention programs. HIV prevention via PrEP improves the health of US residents and can save money in terms of medical costs from the payer perspective.²⁵ Moreover, although not considered in this study, cost estimates from the societal perspective would likely show even more cost savings, because most HIV infections occur in younger people.¹² Conversely, investigators including members of our group have previously shown that growth in PrEP coverage is significantly associated with substantial reductions in new diagnoses in the US.¹²

According to our analyses, any program or policy change that reverses the gains in PrEP coverage that have been made during the past decade will result in increased new HIV infections and substantial increased costs to the health care system and society. Encouragingly, the 2025 decision in Kennedy v Braidwood Management upheld the legality of the procedure used to appoint members of the US Preventive Services Task Force; an adverse decision would have created substantial threats to PrEP and other preventive services.³⁰ However, recently enacted budget cuts to Medicaid and Affordable Care Act marketplace funding are estimated by the congressional budget office to remove as many as 12 million US people from insurance coverage.³¹

Conversely, by maintaining or expanding current effective programs, the US could avert billions of dollars in medical costs associated with infections that would not be averted following disinvestment in HIV prevention. We focus herein on recent and possible threats to maintaining PrEP coverage; conversely, programs that increase investments in PrEP programs offer the opportunity to increase PrEP coverage, which has been reported to be cost-effective in prior analyses.^13,25,32,33 Increases in PrEP coverage would also likely prevent onward (secondary) HIV transmissions.

Our cost estimates are conservative, in that the costs included in our calculations only address the lifetime medical costs of those people who are estimated to acquire new HIV infections because of less PrEP coverage. We do not include the costs of secondary HIV infections that would likely occur from individuals acquiring HIV in our model in the interval between their HIV infection and their diagnosis and effective treatment.³⁴ Including such onward transmissions would significantly increase our estimates of costs associated with failing to sustain current support for PrEP programs. Our analyses assume that no other programmatic changes will be made, for example, reductions to the Ryan White Care Act or other programs that support people engaged in HIV care. If such other changes occur—which seems likely given recently enacted legislation that will decrease the number of people on the Medicaid rolls¹⁵—our data would reflect a minimum estimate of impacts on new infections.

Consideration of policies that would reduce access to PrEP could not come at a worse moment. After sustained national investments, we now have not only a daily pill option for PrEP² but also injectable PrEP that can be administered every 2 months.³⁵ Further, injectable PrEP that only requires 2 shots a year was approved by the US Food and Drug Administration in June 2025,³⁶ and there is potential for an annual injection.³⁷ Injectable cabotegravir shows strong evidence for HIV prevention in women.³⁸ Each of these advances makes it easier to achieve greater population-level coverage of PrEP and associated reductions in HIV and to realize the payoff from sustained US national investments in HIV prevention and care. Disinvestment in PrEP programs will slow the current progress in reducing new infections; disinvestment in PrEP research structures will also destabilize a present (and desperately needed) acceleration in HIV prevention technologies.

Limitations

Our analysis has some important limitations. We rely on a quasi-experimental approach to estimate the associations between levels of PrEP coverage and subsequent HIV infections, and we assume that decreasing PrEP coverage will result in mathematically inverse effects on new HIV diagnoses. Further, our assumption that PrEP use would decline at the annual 3.3% base-case rate that characterized the rate of increases during the prior decade is a simplistic assumption, made in the absence of primary data about the effects of adverse policy changes on PrEP use. The potential withdrawal of federal supports for PrEP programs and cost-saving policies are simply unprecedented, so there are not appropriate models to estimate the potential size of the effects. For this reason, we estimated the effects of a range of possible reductions. If (mainly) federal programs and policies resulted in a 3.3% mean growth of PrEP users prospectively, we believe it is reasonable to predict that removing those supports will result in a corresponding decrease in PrEP use.

The population dynamics of viral suppression and transmission risk are likely not as deterministic as our model would assume, in that the behaviors of individuals are not independent of their PrEP status and adherence. Testing these assumptions in other types of models, such as agent-based models³⁹ that could parameterize heterogeneity in behaviors, would help provide additional context to our findings. Systems of PrEP delivery are complicated, and interventions and many other factors can influence trends in new HIV diagnoses, but the observed associations between population levels of PrEP coverage and new HIV diagnoses are based on clinical data, are robust, and are consistent with our understanding of the mechanisms of action and magnitude effect of PrEP on reducing HIV transmission. Although ecological, these associations meet many of the criteria for causality, including biological plausibility, dose-response associations (biological gradient), temporality, coherence, and consistency.⁴⁰ Further, other factors shape the realized effect of PrEP, for which we did not account in our analyses. For example, we did not evaluate the extent to which those at highest risk for infection are the people receiving PrEP or the effects of adherence, viral resistance to active components of PrEP medications, or adjunctive use of other prevention modalities. It is also possible that PrEP coverage could have a threshold effect on new HIV diagnoses, and if so, our assumption that the trajectory of new diagnoses would be the inverse of the prevention trajectory might not hold true. These limitations are characteristic of analyses of clinical evidence, and despite the limitations, we believe that the association between the extent of PrEP coverage and population-level new HIV infections is distinguishable.

Conclusions

In this economic evaluation of PrEP coverage and associated effects on new HIV infections, we estimated that even modest decreases in coverage of PrEP would have substantial negative effects on health, manifested as increases in new HIV infections and associated health care costs. Such changes in coverage would be the expected results of disinvestment in HIV prevention activities or policy changes that discourage PrEP use by taking away existing health care coverage, increasing out-of-pocket costs for PrEP, or increasing barriers to HIV screening, which is the gateway to PrEP services. We are in a time of increasing PrEP choices and declining new HIV diagnoses in the US; any obstacles to sustained HIV prevention services, including access to HIV screening and PrEP, will burden our country by future increases in health and financial costs.

Supplement.

Data Sharing Statement