Potential clinical and economic impacts of cutbacks in the President’s Emergency Plan for AIDS Relief program in South Africa: a modeling analysis

Abstract

Background:

Future US Congressional funding for the President’s Emergency Plan for AIDS Relief (PEPFAR) program is uncertain.

Objective:

To evaluate the clinical and economic impacts of abruptly scaling back PEPFAR funding ($460 million) from South Africa’s total HIV budget ($2.56 billion) in 2024.

Design:

Model-based analysis of 100%, 50%, and 0% PEPFAR funding with proportional decreases in HIV diagnosis rates (26.0, 24.3, 22.6/100 person-years [PY]), one-year treatment engagement (people with HIV [PWH] on/initiating antiretroviral therapy: 92.2%/80.4%, 87.1%/76.0%, 82.0%/71.5%), and primary prevention (4.0%, 2.2%, 0.5% reduction in incidence with no programming [1.24/100PY]).

Data sources:

Published HIV care continuum, PEPFAR funding estimates.

Target population:

South African adults (HIV prevalence: 16.2%; incidence: 0.32/100PY).

Time horizon:

Lifetime.

Perspective:

Healthcare sector.

Interventions:

PEPFAR funded 100% (PEPFAR_100%), 50% (PEPFAR_50%), 0% (PEPFAR_0%).

Outcome measures:

HIV infections, life expectancy, lifetime costs (2023 USD).

Results of base case analysis:

With current HIV programming (PEPFAR_100%), 1,190,000 new infections are projected over 10 years; life expectancy would be 61.42 years for PWH with lifetime costs of $11,180/PWH. Reduced PEPFAR funding (PEPFAR_50% and PEPFAR_0%) would add 286,000 and 565,000 new infections, respectively. PWH would lose 2.02 and 3.71 life-years with nominal lifetime cost reductions of $620/PWH and $1,140/PWH that would be offset at the population level by more PWH requiring treatment for infection.

Results of sensitivity analysis:

Countries with similar HIV prevalence and greater reliance on PEPFAR funding could experience disproportionately higher incremental infections and survival losses.

Limitation:

Budget fungibility and exact programmatic implications of reducing PEPFAR funding are unknown.

Conclusion:

Abrupt PEPFAR cutbacks would have immediate and long-term detrimental impacts on epidemiologic and clinical HIV outcomes in South Africa.

Primary funding source:

National Institutes of Health.

INTRODUCTION

South Africa has the world’s largest HIV epidemic, with 7.4 million adults with HIV and 152,000 new infections in 2023 (1). Recent strides have brought the nation close to meeting UNAIDS’ 95-95-95 targets: 94% of people with HIV (PWH) aware of their status, 80% of those aware on antiretroviral therapy (ART), and 89% of PWH on ART virally suppressed (1).

While this progress is largely attributable to domestic investments, international partners have played an important role, including the United States (US) President’s Emergency Plan for AIDS Relief (PEPFAR) program (2). With over $8 billion invested in South Africa and $110 billion worldwide since its inception in 2003 (3), PEPFAR supports over 20 million PWH in 55 countries (4). In 2022, $460 million from PEPFAR represented 18% of South Africa’s $2.56 billion HIV budget (5).

PEPFAR is a highly successful global health initiative and is recognized for its role in strengthening sustainable country-led systems, an important component of US foreign policy (6-8). PEPFAR’s future, however, is uncertain, as false claims about the misuse of PEPFAR funds and political shifts have led to waning bipartisan Congressional support (9-11). While the program secured short-term funding through March 2025 (12), long-term support remains unclear. Our objective is to assess the potential clinical and economic ramifications of curtailed PEPFAR support in South Africa to inform policy decisions by Congress and global health leaders and provide an evidence base to clinicians and the American public.

METHODS

Analytic overview

We use the CEPAC-International mathematical model to simulate 2024 HIV programming in South Africa with no cutbacks (PEPFAR_100%), partial cutbacks (PEPFAR_50%), and complete cutbacks (PEPFAR_0%) in PEPFAR funding. We model all South African adults (N=45,680,760) in 2024 over a lifetime: people with prevalent HIV (N=7,400,000) and people without HIV (N=38,280,760) including those who acquire HIV after 2024 (1,13). We assume that funding cutbacks are permanent in the base case, and limit cutbacks to five years in a secondary analysis. We estimate proportional decreases in HIV testing capacity, treatment engagement, and primary prevention with pre-exposure prophylaxis (PrEP) and voluntary medical male circumcision (VMMC), assuming fixed returns on PEPFAR investments (Appendix Methods, Table S1) (1,5). We assume non-PEPFAR funding would remain constant (4,14), and the availability of acute care for PWH would remain unchanged.

We report the HIV care continuum at 5 years and new HIV infections and HIV-related deaths at 10 years to demonstrate short-term epidemiologic impacts of PEPFAR cutbacks. We report lifetime outcomes for people with prevalent HIV (N=7,400,000) and the South African population (N=45,680,760) (1,13), including life expectancy and HIV-related costs in 2023 USD from a healthcare sector perspective (i.e., excluding nonmedical resource consumption, time, productivity; Appendix Methods) (15). We calculate decremental cost-effectiveness ratios (DCERs: ratios of the differences in 3%/year discounted lifetime costs and life-years) to quantify savings per year-of-life lost among people with prevalent HIV (16). While there are no generally agreed-upon thresholds by which to evaluate a DCER, empirical studies consistently find that decision makers’ minimum acceptable savings per year of life lost exceeds the maximum they are willing to pay per year of life gained (16,17).

Model structure

CEPAC-International model

The CEPAC-International model is a validated mathematical model of HIV diagnosis, treatment, and transmission (https://mpec.massgeneral.org/cepac-model/) (18,19). Simulated PWH receive user-specified characteristics (e.g., initial age, sex, CD4 count, HIV RNA) and experience CD4-dependent risks of opportunistic infections and HIV-related mortality. PWH have an initial probability of viral suppression on ART. Without suppressive ART, CD4 count declines monthly. With suppressive ART, PWH experience monthly increases in CD4 count and risks of virologic failure, care disengagement, and re-engagement. PWH face monthly risks of age- and sex-stratified non-HIV-related mortality and HIV RNA-dependent HIV transmission. The model generates HIV transmission rates representative of total community HIV RNA, which are used to project outcomes for one generation of incident HIV infection (Appendix Methods).

Model inputs

Modeled population

We model 7,400,000 adults with prevalent HIV using 2024 HIV care continuum data: 410,398 (5.6%) undiagnosed, 1,414,662 (19.1%) diagnosed but not on ART, and 5,574,940 (75.3%) diagnosed and on ART (Table 1) (1). We also model all 38,280,760 South African adults without HIV (1,13). Mean initial age is 30.3 years for undiagnosed PWH and people without HIV, and 39.2 years for diagnosed PWH (Table 1, Appendix Methods) (20). PWH on ART have a higher mean initial CD4 count (553/μL) reflecting prior HIV treatment compared with those who are undiagnosed (484/μL) or diagnosed but off ART (481/μL) (Appendix Methods) (20). Among PWH on ART at model start, 88.5% are virally suppressed (20).

Table 1:

Model inputs for an analysis of cutbacks in 2024 PEPFAR funding in South Africa

Variable	Base case value		Range	Source
2024 HIV care continuum, n (%)
Undiagnosed	410,398 (5.6%)		–*	(1)
Diagnosed, off ART	1,414,662 (19.1%)		–*	(1)
Diagnosed, on ART	5,574,940 (75.3%)		–*	(1)
Cohort characteristics: PWH	Undiagnosed	Diagnosed
Age, mean (SD), years	30.3 (10.0)†	39.2 (11.0)	20–45	(20)
Female sex, %	63.3	68.6	–*	(20)
CD4 count, mean (SD), cells/μL	484 (134)†	481 (319), off ART 553 (312), on ART	± 1 SD	(20,56)
Virologic suppression, %	0.0	0.0, off ART 88.5, on ART	–	(20,57,58)
Cohort characteristics: people without HIV at model start
Age, mean (SD), years	30.3 (10.0)		25–45	(20)
Female sex, %	63.3		–*	(20)
HIV natural history
HIV mortality monthly probability, range by OI, CD4, ART status	0.0001–0.0953		0.5–2.0x	(59,60)
HIV treatment
ART adherence, mean (SD), %
On ART	85.1 (19.5)		–	(61)
(Re)-initiating ART	66.8 (28.8)		–	(62)
HIV treatment, continued
48-week viral suppression, %, range by ART adherence‡
First-line: TDF/3TC/DTG	0.0–93.8		90–100	(63-66)
Second-line: ABC/FTC/ATV/r	0.0–84.2		75–90	(63-66)
HIV transmission and prevention
HIV transmission rate per 100PY, range by HIV RNA	0.0–9.0		0.5–2.0x	(28)
Status quo HIV incidence/100PY	0.32		0.2–0.5	(1)
Costs (2023 USD)
HIV diagnosis, per result
Negative	3.80		0.5–2.0x	(67)
Positive	5.90		0.5–2.0x	(67)
ART, monthly
TDF/3TC/DTG	3.60		0.5–2.0x	(68)
ABC/FTC/ATV/r	13.70		0.5–2.0x	(68)
OI prophylaxis, monthly	5.10		0.5–2.0x	(26)
CD4 count test	8.20		0.5–2.0x	(26)
HIV RNA test	31.10		0.5–2.0x	(26)
OI treatment, range by OI	208.00–725.00		0.5–2.0x	(30-32)
HIV care, monthly, range by CD4	14.00–137.00		0.5–2.0x	(30-32)

Abbreviations: PEPFAR, President’s Emergency Plan for AIDS Relief; ART, antiretroviral therapy; SD, standard deviation; OI, opportunistic infection; PY, person-years; TDF, tenofovir disoproxil fumarate; 3TC, lamivudine; DTG, dolutegravir; ABC, abacavir; FTC, emtricitabine; ATV/r, ritonavir-boosted atazanavir; USD, United States dollar.

^*Subgroup stratified outcomes are reported in sensitivity analyses in the Appendix.

^†Calibrated to published data at ART initiation: mean age of 32.9 years and CD4 count of 481 cells/μL. See Appendix Methods: HIV testing.

^‡48-week viral suppression rates are linearly interpolated between 10% and 85% adherence.

HIV natural history

HIV-related mortality is dependent on CD4 count, ART status, and present or prior opportunistic infection (Table 1). PWH on ART experience lower risks of opportunistic infection and mortality than PWH off ART. Non-HIV-related mortality is derived from age- and sex-stratified HIV-deleted lifetables (Appendix Table S2).

HIV testing

With PEPFAR_100%, we calculate an HIV diagnosis rate of 26.0/100 person-years (PY) based on 417,000 diagnoses estimated in 2023 (Table 2, Appendix Methods) (21,22). We calculate a cost of $102/diagnosis based on $42.4 million allocated for HIV testing, of which PEPFAR funds $4.8 million (Appendix Table S1) (5). Assuming a fixed $102/diagnosis, we estimate 394,000 diagnoses (24.3/100PY) with PEPFAR_50% (−$2.4 million) and 370,000 diagnoses (22.6/100PY) with PEPFAR_0% (−$4.8 million) (5).

Table 2:

Modeled assumptions of the relationship between cutbacks in 2024 PEPFAR funding and HIV programming in South Africa

Program	Model input	Base case value,* PEPFAR:			Sensitivity analyses,† PEPFAR:			Source
		100%	50%	0%		50%	0%
HIV testing	HIV diagnosis rate, per 100 person-years	26.0	24.3	22.6	Resilient	25.3	23.5	(5,21,22)
					Susceptible	21.5	18.5
Care & treatment	Care engagement at one year, %: PWH already on ART / (re)-initiating ART	92.2 / 80.4	87.1 / 76.0	82.0 / 71.5	Resilient	90.1 / 78.6	84.8 / 74.0	(5,22,24)
					Susceptible	78.6 / 68.5	68.7 / 59.9
Prevention	HIV incidence reduction with primary prevention services, %‡	4.0	2.2	0.5	Resilient	3.3	1.6	(5,29)
					Susceptible	1.8	0.2

Abbreviations: PEPFAR, President’s Emergency Plan for AIDS Relief; PWH, people with HIV; ART, antiretroviral therapy.

^*Base case inputs are estimated using a proportional (linear) relationship in cutbacks in PEPFAR funding and HIV programming.

^†Resilient scenario reflects ability of existing HIV programming to withstand cutbacks in up to 30% PEPFAR funding. Susceptible scenario reflects sensitivity of existing HIV programming to cutbacks with diminishing returns on investments (Appendix Methods).

^‡Relative to a projected HIV incidence of 1.24/100PY in the absence of any HIV programming (Appendix Methods).

HIV treatment

We model current South African HIV treatment guidelines: two ART regimens, CD4 and HIV RNA monitoring, and opportunistic infection prophylaxis (Appendix Methods) (23). We assume all PWH receiving or initiating ART are prescribed a dolutegravir-based regimen (20). Viral suppression probabilities are pooled from ART efficacy trials (Table 1).

Care engagement

We vary care engagement to represent decreased treatment capacity resulting from an abrupt $217 million PEPFAR cutback for HIV treatment (Appendix Table S1) (5). Of the $1.962 billion HIV treatment budget in South Africa (5), we calculate $307 million to be allocated for PWH who (re)-initiate ART based on 1,031,000 ART (re)-initiations in 2023, 80.4% care engagement at one year (829,000 PWH), and $371/person for the first year of treatment (Table 2, Appendix Methods) (22,24-26). We calculate the remaining $1.655 billion to be allocated for the 5,574,940 PWH already on ART based on 92.2% care engagement at one year (5,140,000 PWH) and $322/person per subsequent year of treatment (5,24,26,27). We assume that PEPFAR funding cutbacks for HIV treatment would be proportional to existing allocations for PWH (re)-initiating ART and PWH already on ART, and that costs for the first and subsequent years of treatment would not change. With PEPFAR_50% (−$109 million) and PEPFAR_0% (−$217 million), we estimate care engagement at one year to decrease from 829,000 (80.4%) to 783,000 (76.0%) and 737,000 (71.5%) PWH (re)-initiating ART, and from 5,140,000 (92.2%) to 4,856,000 (87.1%) and 4,572,000 (82.0%) PWH already on ART. We use IeDEA cohort data to derive adherence-stratified probabilities of care disengagement (Appendix Table S2) (24).

HIV transmissions and prevention

We project transmissions over five years from 7,400,000 people with prevalent HIV using HIV RNA-dependent transmission rates (0.0–9.0/100PY; Table 1) (28). From these transmissions, we estimate an HIV incidence of 0.37/100PY reflecting community levels of HIV RNA and the preventive impacts of HIV testing and ART (Appendix Methods). We next estimate the impact of primary prevention programming on HIV incidence; we conservatively assume that funding cutbacks would only affect PrEP and VMMC because these services rely on PEPFAR-supplied commodities (5). We assume behavior norms would not be affected by decreased HIV funding or programming (e.g., condom use, sexual practices). Based on current PrEP and VMMC uptake (400,000 and 173,000, respectively), we estimate these services would contribute a further 0.05/100PY reduction in HIV incidence (calculated as a 4.0% reduction in a theoretical incidence of 1.24/100PY projected in the absence of any HIV programming; Table 2, Appendix Methods) (1,22). We apply the 0.37/100PY incidence rate and 0.05/100PY incidence reduction to a cohort without HIV to reflect the risk of HIV acquisition with current programming (PEPFAR_100%). The resulting rate of 0.32/100PY matches published incidence estimates in South Africa (1). We calculate incidence reductions of 0.02/100PY (2.2% relative to a theoretical incidence of 1.24/100PY with no HIV programming) with PEPFAR_50% (−$57 million) and 0.01/100PY (0.5%) with PEPFAR_0% (−$113 million) (Appendix Methods) (5,29). We assume a fixed incidence over the model horizon to project outcomes to epidemiologic scale with primary transmissions only (i.e., one generation of transmissions from the cohort with prevalent HIV).

Costs

We model costs of ART ($3.60–13.70/month), CD4 monitoring ($8.20/test), HIV RNA monitoring ($31.10/test), opportunistic infection treatment ($208.00–725.00), and routine HIV care ($14.00–137.00/month) (Table 1, Appendix Methods) (26,30-32).

Sensitivity analyses

In one-way sensitivity analyses, we examine the impact of single input variation across plausible ranges (Table 1). In multiway sensitivity analyses, we examine the combined impact of varying cutbacks in HIV testing, treatment, and prevention.

Scenario analyses

Scenario analysis is a powerful means of examining the effects of simultaneous variation in several key parameters to simulate plausible alternatives from the base case. We model a “resilient” scenario where programming would remain unchanged with up to 30% cutbacks in PEPFAR funding, and a “susceptible” scenario where programming cutbacks would occur at a greater rate than funding cutbacks (Table 2, Appendix Methods). We model a “return to current status quo in 2029” scenario reflecting a hypothetical five-year lag to reallocate and/or identify other funds to fill the budget void created by PEPFAR cutbacks. We also model a “larger PEPFAR funding profile” scenario where PEPFAR funding in South Africa mirrors its support in Zimbabwe (HIV prevalence: 12.6%): 55% of testing, 45% of treatment, and 77% of prevention programming (Appendix Table S2) (33).

Role of the funding source:

The funding sources had no role in the design, analysis, or interpretation of the study. The content of this manuscript does not necessarily represent the views of the National Institutes of Health.

RESULTS

Base case results

Clinical outcomes: HIV infections and incidence

With status quo HIV programming in South Africa (PEPFAR_100%), 1,190,000 individuals are projected to acquire HIV by 10 years (incidence: 0.32/100PY), which would increase to 1,476,000 (0.40/100PY) with PEPFAR_50% and 1,755,000 (0.48/100PY) with PEPFAR_0% (Table 3).

Table 3:

Projected clinical and economic outcomes of no (PEPFAR_100%), partial (PEPFAR_50%), and complete (PEPFAR_0%) cutbacks in 2024 PEPFAR funding and base case, resilient, and susceptible HIV programming in South Africa

Strategy	HIV incidence	Population outcomes at 10 years, n (+X%)*		Per person outcomes
				HIV (n = 7.4 million)		Population (n = 45.7 million)
	Per 100PY	New HIV infections†	HIV-related deaths‡	Life- years§	Lifetime costs, 2023 USD§	Life- years§	Lifetime costs, 2023 USD§
Base case
PEPFAR_100%	0.32∥	1,190,000	1,585,000	22.71	11,180	37.26	2,550
PEPFAR_50%	0.40	1,476,000 (+24%)	1,900,000 (+20%)	20.69	10,560	36.63	2,580
PEPFAR_0%	0.48	1,755,000 (+47%)	2,186,000 (+38%)	19.00	10,040	36.05	2,620
Resilient scenario
PEPFAR_50%	0.36	1,310,000 (+10%)	1,726,000 (+9%)	21.79	10,900	36.97	2,560
PEPFAR_0%	0.44	1,602,000 (+35%)	2,040,000 (+29%)	19.84	10,300	36.35	2,600
Susceptible scenario
PEPFAR_50%	0.49	1,809,000 (+52%)	2,365,000 (+49%)	17.91	9,700	35.77	2,570
PEPFAR_0%	0.62	2,264,000 (+90%)	2,911,000 (+84%)	15.25	8,830	34.78	2,600

Abbreviations: PEPFAR, President’s Emergency Plan for AIDS Relief; PY, person-years; USD, United States dollar.

^*Projections are undiscounted and rounded to the nearest 1,000, followed by an “+X%” increase relative to PEPFAR_100%.

^†Projected increases in new HIV infections relative to PEPFAR_100% reflect greater primary transmissions from 7,400,000 people with prevalent HIV to 38,280,760 people without HIV over 10 years.

^‡HIV-related deaths are among people with prevalent and incident HIV over 10 years.

^§Life-years and HIV-related costs are undiscounted, projected from model start, and rounded to the nearest 0.01 year and $10.

^∥Estimated based on current HIV programming (Appendix Methods). Matches published estimates in South Africa.(1)

Clinical outcomes: HIV-related deaths and per-person life-years

South Africa would face 1,585,000 HIV-related deaths over 10 years with PEPFAR_100%; deaths would increase by 315,000 (+20%) with PEPFAR_50% and 601,000 (+38%) with PEPFAR_0% (Table 3). People with prevalent HIV would live 22.71 life-years from model start with PEPFAR_100%, 20.69 with PEPFAR_50%, and 19.00 with PEPFAR_0%. PWH already on ART would experience the greatest life expectancy losses (Appendix Table S3). Among all 45,680,760 adults in South Africa, life-years lived from model start would decrease from 37.26 to 36.63 and 36.05, respectively.

Clinical outcomes: HIV care continuum

The estimated 2024 HIV care continuum in South Africa is 94-80-89 (1). The projected HIV care continuum in 2029 would improve to 95-85-90 with PEPFAR_100% but worsen to 94-79-89 with PEPFAR_50% and 92-73-88 with PEPFAR_0% (Figure 1).

Figure 1: Projected HIV care continuum in 2029 with no cutbacks (PEPFAR_100%), partial cutbacks (PEPFAR_50%), and complete cutbacks (PEPFAR_0%) in 2024 PEPFAR funding.

The estimated 2024 HIV care cascade (black bars) is based on published data (1). The projected HIV care cascade in 2029 is shown for PEPFAR_100% (dark blue bars), PEPFAR_50% (blue bars), and PEPFAR_0% (light blue bars). Each set of bars represents PWH who are aware of their HIV status, receiving ART, and virally suppressed (HIV RNA <20 copies/mL), respectively. The percents shown are in respect to the conditional UNAIDS’ 95-95-95 epidemic targets (i.e., proportion of all PWH who are aware of their status, proportion of PWH aware of their status who are receiving ART, and proportion of PWH receiving ART who are virally suppressed). The estimated number of PWH alive in South Africa in 2024 is 7,400,000; the projected number of PWH alive in 2029 would be 6,913,000 with PEPFAR_100%, 6,950,000 with PEPFAR_50%, and 6,997,000 with PEPFAR_0%. The number of PWH associated with each bar is labeled below and rounded to the nearest 10,000. Abbreviations: PWH, people with HIV; ART, antiretroviral therapy; PEPFAR, President’s Emergency Plan for AIDS Relief.

Economic outcomes: people with prevalent HIV

Projected lifetime costs for people with prevalent HIV would decrease from $11,180/person with PEPFAR_100% to $10,560/person with PEPFAR_50% and $10,040/person with PEPFAR_0% due to lower life expectancy among PWH and reduced healthcare access (Table 3). However, HIV-related costs would increase for people with prevalent HIV in the shorter term due to increased expenditure for PWH off ART (Tables S4A-S4B).

Economic outcomes: South African population

Lifetime costs would increase for all 45,680,760 South African adults from $2,550/person with PEPFAR_100% to $2,580/person with PEPFAR_50% and $2,620/person with PEPFAR_0%, reflecting greater healthcare expenditures for a population with greater HIV-related morbidity (Table 3). Over 10 years, expenditures would increase by $880 million with PEPFAR_50% and $1.70 billion with PEPFAR_0% relative to current spending (Appendix Tables S4A-S4B). These increases reflect greater costs of acute care for an overall sicker population off ART despite fewer routine care costs for those on ART.

Sensitivity analyses

One-way sensitivity analysis

In one-way sensitivity analysis, additional infections at 10 years ranged 211,000–385,000 and 419,000–795,000 with PEPFAR_50% and PEPFAR_0% compared with PEPFAR_100% (Figure 2A). Per-person life-years lost ranged 1.28–2.72 and 2.17–4.93 for people with prevalent HIV (Appendix Figure S1) and 0.47–0.90 and 0.86–1.72 for the entire population (Figure 2B). Projected new infections were most sensitive to parameters that influenced community HIV RNA (e.g., care engagement, primary prevention, CD4 count) whereas life expectancy was most sensitive to parameters that influenced disease severity (e.g., age, CD4 count, HIV-related mortality) (Appendix Figures S2-S4). Due to life-years lost, lifetime costs decreased for PWH across all ranges examined (Appendix Figure S5); these costs were offset at the population level by treatment costs of new infections unless survival among PWH was very low (Appendix Figure S6). Costs of HIV care and opportunistic infection treatment had the greatest impacts on projected lifetime costs (Appendix Figures S7-S8).

Figure 2: One-way sensitivity analysis on additional HIV infections at 10 years (Panel A) and on life-years lost among the population (Panel B) with partial (PEPFAR_50%) and complete (PEPFAR_0%) cutbacks in 2024 PEPFAR funding compared with no cutbacks (PEPFAR_100%).

A tornado diagram displays the sensitivity of a key outcome measure (additional HIV infections in Panel A; and life years lost in Panel B) to variation in different model input parameter values. The y-axis shows input parameters for which one-way sensitivity analyses were performed, with the range examined in parentheses followed by the base case value. The x-axis shows additional HIV infections (Panel A) and life years lost (Panel B). The left and right black vertical lines represent the base case value with PEPFAR_50% and PEPFAR_0% compared with PEPFAR_100%. The horizontal bars (light blue: PEPFAR_50%, dark blue: PEPFAR_0%) represent the sensitivity of projected additional infections (Panel A) and projected life-years lost (Panel B) to the six most impactful parameters. Other parameters exerted little or no influence on the outcome including proportion of female sex, status quo rates of HIV diagnosis, and 48-week viral suppression on protease inhibitor-based ART. Abbreviations: PEPFAR, President’s Emergency Plan for AIDS Relief; SD, standard deviation; ART, antiretroviral therapy.

Multiway sensitivity analysis

Scaling back primary prevention programming alone would produce modest increases in new infections; these increases would compound when coupled with cutbacks in HIV treatment (Appendix Table S5). Cutbacks in treatment alone led to greater life expectancy losses than testing and primary prevention.

Scenario analyses

Resilient and susceptible programming scenarios

With greater resilience to PEPFAR cutbacks, fewer additional infections would be projected by 10 years than in the base case: 120,000 with PEPFAR_50% and 412,000 with PEPFAR_0% compared with PEPFAR_100% (Table 3). Projected life-years lost for PWH would also be lower: 0.92 and 2.88. With greater susceptibility to PEPFAR cutbacks, the projected impacts would be higher: 619,000 and 1,074,000 additional infections and 4.80 and 7.46 life-years lost for PWH. Population lifetime costs would increase in both scenarios.

Return to current status quo in 2029 scenario

In a scenario where the PEPFAR budget deficit is filled and HIV programming returns to its current status quo in 2029, the projected impacts of PEPFAR cutbacks would still be substantial. By 10 years, new infections would increase by 162,000 with PEPFAR_50% and 305,000 with PEPFAR_0% compared with PEPFAR_100% (Appendix Table S6). The costs of treating new infections would offset cost reductions associated with life-years lost, resulting in similar population-level lifetime costs ($2,540–2,550/person).

Larger PEPFAR funding profile scenario

In a scenario where PEPFAR funding in South Africa mirrors support in Zimbabwe (55% testing, 45% treatment, 77% prevention), new infections over 10 years would increase by 72% and 142% with PEPFAR_50% and PEPFAR_0% compared with PEPFAR_100%; PWH would lose 6.79 and 11.01 life-years (Appendix Table S7). While these life-year losses would lead to lower HIV treatment costs, lifetime costs for the population would decrease only modestly ($10–50/person) due to the cost of many more new infections.

Decremental cost-effectiveness ratios

Among people with prevalent HIV the projected DCER would be $220 saved per year-of-life lost for both PEPFAR_50% and PEPFAR_0% compared with PEPFAR_100% (Table S8). DCERs would be similar across scenario analyses ($220–240 saved per year-of-life lost) demonstrating that on a lifetime horizon, any potential cost savings would be a direct result of a loss in life-years.

DISCUSSION

With uncertain US Congressional support of PEPFAR, we developed a model-based analysis to determine the clinical and economic consequences of abrupt cutbacks in $460 million of PEPFAR HIV funding allocated to South Africa. We had three key findings.

First, abruptly scaling back PEPFAR would reverse progress towards the 95-95-95 goals and lead to substantial life expectancy losses. Life expectancy among PWH would decrease by 3.71 life-years–a major reduction compared to other health interventions, including universal ART eligibility (34-36). South Africa has faced prior programmatic disruptions, including in the early 2000s when political leadership delayed ART rollout due to misinformation about the cause of AIDS leading to an estimated 330,000 additional lives lost (37), and during the COVID-19 pandemic (38), which led to decreased HIV testing and ART initiations (25,39,40). Our analysis quantifies the impact of disruptions from scaling back PEPFAR, and adds to existing evidence demonstrating PEPFAR’s role in improving survival for PWH (35,41).

Second, withdrawing PEPFAR support would impact the health of the entire South African population, with a projected 565,000 additional HIV infections by 2034 and 1.20 life-years lost per person. Cutbacks in HIV care and treatment–which receive $217 million annually from PEPFAR–had the greatest impact on projected new infections. When combined with primary prevention cutbacks, new infections would compound due to a simultaneous greater force of infection and susceptibility to infection, as seen in other analyses,(5,29,39,42). These multiway sensitivity analyses demonstrate alternative budget re-balancing possibilities, reflecting the fungibility of current non-PEPFAR HIV funding. Future research is needed to inform optimal approaches to funding re-allocation. Curbing HIV incidence is critical for epidemic control, and PEPFAR continues to play a vital role in these efforts (6,43).

Third, cost reductions from defunded programs in 2024 would be offset by increased costs of treating a larger and sicker population with HIV. We projected that HIV-related costs would increase within one year relative to current spending with an excess of $1.70 billion by 2034. These additional costs would primarily be used to treat new opportunistic infections that would otherwise be averted by PEPFAR-funded programming, rather than treating new HIV infections. Furthermore, these extra costs would likely require diverting funds from other effective interventions without making progress toward HIV epidemic control. Our projections are consistent with a study that found that any reduction in HIV testing would increase care costs over time due to increased HIV infections (42). Population costs only decreased in conditions with substantial life expectancy losses. We projected $220 saved per year-of-life lost among PWH, which is substantially less than South Africa’s willingness to pay for health ($3,010–4,100/year-of-life) (13,34,44), and still below more affordable benchmarks for established HIV programs ($550/year-of-life) (45). These cost reductions would be marginal and directly related to shorter survival of PWH.

This study likely represents a best-case scenario for PEPFAR cutbacks compared to other settings. South Africa’s economic profile has changed since PEPFAR’s inception: from 2008–2022, the South African government increased total dollar HIV investments by approximately 15% year-after-year while maintaining funding for nearly three-quarters of all HIV expenditures (14). Investments from financial partners like PEPFAR have enabled overall healthcare system strengthening and decreased reliance on external funding (13,46). However, in-country and external HIV funding have recently flattened due to slowed GDP growth and the economic toll of the COVID-19 pandemic (47-49). A sudden $460 million deficit would therefore unlikely be readily replaced and would jeopardize access to highly successful HIV interventions, leading to substantial morbidity and mortality. PEPFAR contributes a greater share of HIV spending in other countries such as Zimbabwe (2023 per-capita GDP: $2,160 versus $6,020 in South Africa) (4,13), which would be even more vulnerable to abrupt PEPFAR cutbacks, with nearly 3-fold more HIV infections and life-years lost in a scenario analysis. Countries that rely more on PEPFAR support may be less equipped to withstand sudden HIV funding withdrawal. Geographic areas within South Africa and Uganda that transitioned away from PEPFAR funding have experienced decreased care engagement, more frequent commodity stockouts, and delays in implementing national treatment guidelines (50-52).

This analysis has several limitations. First, we do not model the impact of cutbacks on pregnant people and children, which would render our projections more pessimistic (53). Although we model population heterogeneity within South Africa, we do not explicitly model at-risk groups who may be disproportionately affected by PEPFAR cutbacks. Second, we assumed that PEPFAR funding cutbacks would be proportional to decreases in HIV programming, implying fixed care costs which may not reflect programmatic realities (e.g., additional costs for advanced HIV disease or to re-engage PWH). We examined alternative scenarios with non-linear returns on PEPFAR investments, which demonstrated substantial clinical and epidemiologic impacts even if current programming was resilient to cutbacks. While these scenarios are intended to illustrate the range of possible outcomes resulting from funding cutbacks, they do not quantify precision or the likelihood of occurring. Third, we assume that the programmatic effects of withdrawing PEPFAR funding would be permanent in the base case, which may overestimate our projections. However, refunding HIV infrastructure would take time and an immediate loss of services would have a lasting impact on PWH. Furthermore, we did not consider the impact of reduced US contributions to the Global Fund, which has consistently funded 2–5% of South Africa’s HIV budget (5). Even if cutbacks were limited to five years, new HIV infections could increase by 26%. Fourth, we assume a fixed HIV incidence, which may overestimate new infections that would be averted by future interventions. Although we do not account for the gradually decreasing incidence of new infections in South Africa, these trends would likely be negated by PEPFAR cutbacks. Furthermore, we only model the impact of PEPFAR cutbacks on primary transmissions, which may underestimate dynamic growth of new infections (i.e., second-order transmission, vertical transmission). Finally, we do not capture PEPFAR’s indirect impacts such as facilitating decreased healthcare costs, increasing employment, and supporting GDP growth (54,55).

In conclusion, this simulation modeling analysis demonstrates that abruptly scaling back PEPFAR funding would have a striking and deleterious impact on the progress South Africa has made towards HIV epidemic control. Any total cost reductions would be short-lived and at the expense of as many as an additional 565,000 new HIV infections and 601,000 HIV-related deaths in South Africa by 2034. These implications highlight the importance of PEPFAR as both a life-saving foreign aid program and a bridge to self-sustaining national HIV programs.