How Many Lives Does US Foreign Aid Save?

The efficacy and impact of US assistance is of considerable interest at the moment. Below, we present some estimates for lives saved by US assistance worldwide, with illustrative estimates by recipient country. Our core estimates are for deaths prevented from HIV/AIDS, vaccine-preventable illnesses covered by Gavi, TB, malaria, and emergency/humanitarian relief. We suggest the number of lives saved per year may range between 2.3 to 5.6 million with our preferred number resting on gross estimates of 3.3 million. We briefly discuss some other lifesaving interventions financed by US assistance in a later section.

Methodology

The appendix at the end of this post walks through our approach to estimating lives saved for each category of US aid that we consider. At the broadest level though, here are a few things to keep in mind as you read:

We distinguish conceptually between two types of estimates of lives saved. Where possible, we examine both of the questions “how many lives were saved by US financing?’ (gross lives saved) and ‘how many would have been lost absent US financing?” (net lives saved). The number of people who would be dead if the US hadn’t provided assistance isn’t necessarily all of the lives saved by that assistance because of effects including ‘crowding out’ and “crowding in”—international investments displacing or spurring more domestic support for lifesaving interventions—and market effects like generating lower prices. If the US pays for 1,000 vaccine doses that save 3 lives, we’re counting those 3 lives in the “gross” calculation. But perhaps 300 of those vaccinations would’ve happened without US aid, and so the “net” calculation only attributes 1 life saved to US aid.

In general, our “gross” estimates of lives saved are modeled based on evidence of the efficacy of the underlying interventions funded by US foreign aid programs, scaled up by the number of people reached and/or dollars spent. Our “net” numbers are more direct empirical observations (for the most part) based on mortality comparisons of countries or time periods where aid was or was not present—allowing that other actors may have filled some of those gaps.

We used the best available evidence, but there is considerable uncertainty in some of these numbers. That’s particularly true around net lives saved, where there are no clean natural experiments to allow for reliable quantification for some sectors, but also regarding the modeled estimates of gross lives saved which face data challenges around disease burdens, location-specific treatment efficacy and actual levels of treatment provided. In the case of humanitarian assistance, where there was no modelled estimate, we created our own which includes some thinly evidenced assumptions detailed below.

We don’t look at the benefits of aid beyond directly saving lives. Most US assistance isn’t focused on simply and directly saving lives. Even within health, resources are directed to reducing the debilitating effect of diseases that usually don’t kill you. Polio, for example, permanently paralyzes about one in 200 of those infected, and “only” kills about 5-10 percent of those paralyzed. US assistance for polio vaccination through the decades has prevented ten to twenty cases of paralysis for every life saved. Polio also points up an issue that it is now almost eradicated, so that the marginal effect of vaccination programs appears small even if the global benefit of full eradication would be huge (in fact we do not present estimates for polio as part of this analysis despite the US contributing $4.5 billion to the Global Polio Eradication Initiative over the past 35 years preventing an estimated 2.4 million paralytic cases in Initiative countries). Or take a case of indirect impact: children of educated mothers are less likely to die young, as are children in more wealthy countries. US assistance for education or infrastructure has a considerable long-term effect on health.

We focused on the interventions that had the largest direct impact in terms of lives saved, based on the available evidence. We did not include every possible lifesaving intervention funded by US foreign assistance into our calculations. Some lifesaving interventions that are not in our core estimates include:

• Water and sanitation: In 2014, about 780 million people lacked access to clean water. The link through water access to diarrhea was largely responsible for a Global Burden of Disease estimate of 337,000 annual deaths from inadequate water and sanitation. Between 2008-2019, USAID’s assistance resulted in 53.7 million people gaining access to sustainable drinking water. Assuming that reduced the death burden by the same percentage as it reduced the percentage without clean water access, it would equal about 23,200 lives saved per year.
• Family planning: The Guttmacher Institute estimates that US assistance provides modern family planning services to about 11.7 million women worldwide, and that were those women using traditional methods, this would result in 8,340 maternal deaths a year because of complications from unwanted pregnancies.
• Nutrition: Some of the lives saved by US nutrition assistance should be caught in our attempts to measure the impact of humanitarian relief. At the same time, malnutrition and assistance efforts to reduce it are far more widespread than simply in emergency situations. In countries with high rates of malnutrition, modeled estimates suggest a cost per life saved of nutrition interventions at between USD 1,600-4,300 per child death and stillbirth averted. Given US nutrition specific disbursements of $309m in 2020 that suggests the potential for 71,000 to 193,000 averted deaths.

These exclusions are reasons to believe our estimates are at the low end. Note also we focus purely on assistance.

Finally, non-assistance activities of the US government have considerable impacts on global health both positive (e.g., a. research and development, surveillance, support for peacekeeping operations) and negative (e.g., weak regulation of weapons production and export, greenhouse gas emissions, subsidies for military assistance).

A full discussion of our methodology and data is available at the bottom of this piece. Table 1 summarizes our topline numbers, and the basic methodology behind each calculation. We provide a narrative description of the underlying methodology for each figure in the appendix, and full calculations can be found at this link.

Country-level estimates

US foreign aid targets many of the world’s poorest countries; that’s especially true of American aid in the life-saving areas highlighted here like assistance for combating HIV/AIDS, malaria, and TB, and of course emergency humanitarian relief.

As described above, our global estimates of lives saved are mostly grounded in a calculation of the efficacy of the underlying services that foreign aid funds in each sector (e.g., lives saved by antiretrovirals), with some courageous extrapolation based on funding levels.

Here we apply that same approach to individual countries. We start from the cost-effectiveness parameter (i.e. dollars per life saved) reported for each sector above, and apply it to US aid allocations on a country-by-country, sector-by-sector basis. While at the global level we partially corroborate our estimates for HIV/AIDS, malaria, and vaccines by comparing actual mortality data in countries that get more or less aid, that’s not possible for the country-specific estimates, so we’re relying entirely on dollar values and global assumptions here.

This is obviously fairly crude and simple. In reality, a dollar humanitarian aid to Darfur may save more lives than a dollar in Ukraine; a dollar for malaria control will save more lives in a resource-constrained country with high malaria burden like the DRC than a country with a lower burden and more resources like Kenya.

Turning to the results:

• Overall lives saved are concentrated in East and Southern Africa, plus Nigeria, in line with the geography of the HIV/AIDS epidemic and malaria burdens.
• The specific countries where we estimate the largest number of lives saved are Nigeria and South Africa, both at around 150,000 lives saved per year from US foreign aid, almost entirely due to PEPFAR support for HIV/AIDS in South Africa’s case, and largely so in Nigeria, albeit with significant numbers due to vaccines, tuberculosis control, and malaria control in that case.
• Outside of Africa, we see the highest numbers in Indonesia, India, and Pakistan, based on allocations for TB controls (and for vaccines in Pakistan’s case), as well as in Ukraine (mostly due to TB and HIV/AIDS control).

What Would Happen if the US Ended Its Support Today?

Our net estimates may suggest the likely long-term impact were the US to abandon lifesaving support, but gross impacts are more likely to capture the more immediate effects. It is worth repeating that US health assistance is concentrated on poorer countries in Africa least able to ramp up spending. And while other donors remain more generous than the US in terms of the percentage of their economy that goes to foreign aid, most are flatlining or reducing budgets. Add to that the fact the US remains biggest donor in absolute terms, it is implausible that domestic or other foreign resources could even partially fill the gap left by US funding in the short term. Sadly, that is what we are seeing play out during a 90-day pause on foreign assistance that began on January 24th and the shuttering of USAID, involving a set of waivers for lifesaving assistance that are at best partially effective.

The multilateral agencies that the US supports including Gavi and the Global Fund continue to operate. But most bilateral assistance and food aid through the World Food Programme (WFP) appears to have been at least considerably disrupted. Give or take, our estimates for each core intervention suggest that on an average day when US foreign assistance is working as it normally does, it saves a little more than 9,000 lives, of which a little more than 6,000 are through bilateral programs or involving in-kind support from the US. While there is surely some resilience in the system that will have prevented that level of mortality, it appears inevitable that the cost in terms of deaths will continue to climb because of the disorder and confusion within USAID and the number of service providers who have already been forced to close down.

Appendix 1. Country-level estimates

Table 2. Estimates of annual lives saved by US foreign aid, by country and program area

Appendix 2. Description of the methodology for the lives saved calculation in each sector

HIV/AIDS

• Our gross estimate of the lives saved by PEPFAR is computed as follows:
  • Start from the number of people receiving US-funded antiretroviral therapy (ART) around the world, which the State Department estimates at 20.6 million people.
  • What would happen without these drugs? Available evidence suggests that the median survival time for untreated HIV/AIDS is about 12.5 years, based on evidence from around the world, including East Africa (Glaubius et al 2021). That implies that roughly 8% of people on PEPFAR-funded ART would die each year without the drugs.
  • That yields a total number of lives saved per annum of about 1.6 million.
• Our “net” estimate allows for the fact that the realistic alternative to US aid is not nothing, and compares mortality outcomes in PEPFAR focus countries to actual outcomes in comparison countries over time.
  • Specifically, our calculation draws on differences-in-differences estimates of PEPFAR’s impact on adult mortality, based on the results in Gaumer et al (2024).
  • Gaumer et al estimate that in the most recent five-year period in their data, PEPFAR reduced all-cause mortality by roughly 3 deaths annually per 1,000 people.
  • Given total PEPFAR spending in these “high-intensity” countries of about $38 billion over 15 years, this implies a cost per death averted of around $900.
  • Applying that rate to all US aid for HIV/AIDS implies nearly 3 million deaths averted per annum.
  • In their own back-of-the-envelope calculations based on Gaumer et al, Piper at al 2025 argue that this is likely an overestimate, as the regression coefficients may be driven by smaller countries with higher HIV prevalence, and they adjust the estimates downward accordingly (closer to 1.4 million lives a year). This seems reasonable, though any approach here is fairly ad hoc, and we present the raw estimates for transparency, though feel they should be taken with a considerable grain of salt.
• Note that our net estimate of lives saved from PEPFAR compared to realistic counterfactuals is considerably higher than our gross estimate (2.9 compared to 1.6 million), which implicitly assumes everyone on PEPFAR-funded ARVs would simply go without them otherwise. One possibility is that PEPFAR saves more lives than even these direct benefits, through reduction of mother-to-child transmission, voluntary male circumcision, and other services.But another possibility, of course, is that the diff-in-diff estimates reported by Gaumer et al are unrealistically high.

Vaccine-preventable infections

• Our gross estimate of the lives saved by Vaccines funded by the US is computed as follows:
  • Carter et. al. 2024 estimate that vaccines are expected to avert 51.5 million deaths between 2021 and 2030 of which approximately 1m per annum are in upper-middle income countries largely outside Gavi's remit.
  • Dykstra el al. estimate that Gavi is responsible for procuring about 50 percent of total vaccine coverage and the US is responsible for about 24 percent of Gavi funding.
  • This suggests a US ‘share’ of about 500,000 lives saved each year.
• Our net estimate of the lives saved by Vaccines funded by the US is computed as follows:
  • Shastry and Tortrice use differential timing of rollout across countries and vaccines to estimate that Gavi-supported vaccine programs saved around 1.5 million lives 2000-2019.
  • Given the US share of Gavi spending this amounts to about 19,360 lives a year.
• Note our core estimate does not include the impact of support for Covid vaccines. The US supplied 687 million Covid vaccine doses through June 2023, that out of about 14,000 million doses administered. Vaccines as a whole have prevented about 2.5 million deaths from Covid, pro rata that suggests US foreign assistance can take (gross) credit for about 123,000 lives saved over four years.

Malaria

• As with the case of PEPFAR and HIV/AIDS, our ‘net’ estimate of lives saved by US aid for malaria through the President’s Malaria Initiative (PMI) is higher than the gross estimate, which attributes all reductions in malaria deaths since 2000 to foreign aid. Once again, this may reflect positive spillovers from malaria aid on the broader health system, or unrealistically high econometric estimates of PMI’s effects.
• Our gross estimate is calculated, following the approach outlined in Winskill et al 2017 and adopted in the World Malaria Report, of calculating the reduction in malaria deaths since 2000, and attributing that to foreign aid donors in proportion to their contribution to total malaria aid.
  • If the malaria death rate from 2000 were applied to the 2023 population of people at risk for malaria, an additional 600,000 people would have died of the disease.
  • Including both PMI bilateral programs and the US contribution to the Global Fund, the US accounted for nearly half (47.8%) of total foreign aid for malaria in 2022.
  • Applying this funding share to deaths averted yields a figure ofaround 300,000 malaria deaths averted by US foreign aid.
• Our net estimate compares mortality rates in PMI and non-PMI countries over time, drawing on the difference-in-difference estimates in Jakubowski et al 2017.
  • They estimate a reduction of 4.5 child deaths per 1,000 person years due to PMI aid.
  • Applied to the 136 million under-5 population in PMI countries, this implies approximately 600,000 child deaths averted each year due to PMI.
  • Assuming malaria aid through the Global Fund is similarly effective, and allowing for the US’s 42% stake in the Global Fund roughly double this number to nearly 1.2 million deaths averted per year.
• As somewhat of a robustness check, the WHO suggests that 282m nets were distributed in 2022 and a Cochraine analysis estimates 5.6 lives are saved for every 1,000 children protected by nets—if all had been correctly used, that suggests 1.6m deaths averted.
• As a robustness check on vaccine preventable and malaria estimates, in 2014, the Institute of Health Metrics and Evaluation tried to estimate the number of child deaths prevented by donor support for child health programs (including Gavi and bed nets) using analysis of financing and overall declines in child mortality. They calculated that for the average low income country, a child death was prevented for each $4,205 in spending, and for a lower middle income country $6,496. Given US spending on child health programs, the IHME estimate was that the US prevented 367,000 child deaths in 2014. We don’t have the data to replicate and extend their modeling, but assuming a cost in 2022 USD of about $6,500 and using IHME data on US child health and malaria spending suggests a total of over 400,000 child deaths may have been prevented in 2023.

Tuberculosis

• For tuberculosis, we know of no empirical estimate of the impact of US foreign aid relative to a realistic counterfactual, and present exclusively a ‘gross’ estimate.
• The WHO estimates that TB treatment and provision of antiretrovirals to HIV-positive people with TB averted an estimated 44 million deaths between 2010 and 2022 including TB/ART combination treatment preventing 9.2 million. To avoid any risk of double-counting, we assume 36 million deaths averted from TB treatment alone.
• The WHO also estimates that roughly 20% of TB control is funded internationally, with 51% of aid coming from the US.
• Combining these figures yields a total of 300,000 tuberculosis deaths averted due to US foreign aid.
• As a robustness check, the estimated cost per death averted of international TB assistance is around $2,000, this exercise suggests a cost per life saved of $2,150.

Emergency/Humanitarian Relief

• Our estimates for humanitarian assistance are perhaps the most uncertain: the literature we found on the health impact of assistance is limited. Available global estimates are limited to numbers of people ‘in need’ of humanitarian assistance, and ‘reached’, and we found no estimate of net or gross lives saved. We present two alternative estimates to highlight the degree of uncertainty here, based on completely different modeling strategies, and reaching widely divergent results. We would characterize both as “gross” measures, as they model the impact of assistance per se, without trying to account for realistic counterfactuals to US aid specifically.
• The first approach is based on observed mortality rates in a handful of high-profile humanitarian emergencies:
  • We estimate lives saved by humanitarian assistance by comparing mortality outcomes in crisis areas with limited or humanitarian support to mortality outcomes in areas of more considerable support (usually measured in terms of humanitarian workers in proportion to local population), using Checchi et al 2023 (Northeast Nigeria), Degomme and Guha-Sapir 2010 (Darfur) and Heudtlass et al 2016 (multi-country). These suggest (i.a.) crude death rates can be as low as background (non-crisis) death rates amongst refugees (largely in camps) and 2.5 times higher amongst internally displaced. Three approaches using the available data suggest a potential fall in annualized death rates from 22.4 per 100,000 to 5.8 per 100,000 assuming a year of support, with a very high range of both variability and uncertainty.
  • We use the 2023 number from UNOCHA for the population reached by humanitarian assistance of 128 million and we (arbitrarily) assume ‘reached’ is the equivalent of six months high-quality access to humanitarian assistance. This number may appear high given that less than seven million were in refugee camps, or about one in twenty of those ‘reached’. On the other hand, humanitarian response is concentrated in areas with the highest crisis death rates. The assumption suggests the humanitarian system may be responsible for saving approximately 1.8 million lives a year.
  • UNOCHA suggests the US is responsible for about 29.2 percent of global humanitarian funding. This would imply the US may be responsible for saving about 549,000 lives a year through the humanitarian system.
• Our second approach starts from RCT estimates of the impact of food aid on child nutrition, and applies those to the well-studied empirical relationship between nutrition and child mortality.
  • Experimental results in Isanaka et al 2009 find that 500 calories of food aid over multiple months lead to a durable increase in weight-for-height Z-scores of 0.22 standard deviations in Niger.
  • In a paper modeling the impacts of US food aid reform, Nikulkov et al 2016 link these increases in WHZ scores to reduced child mortality, using a formula we reproduce in the appendix spreadsheet.
  • We apply this formula to baseline nutrition levels typically associated with each phase of the IPC’s classification of food insecurity, and the current number of people in each phase. These numbers are on the same order of magnitude as the total number of people plausibly reached by US humanitarian assistance, but we assume US aid targets those in the highest IPC phases first.
  • The resulting numbers are quite modest: we estimate that US humanitarian and food aid, were it all distributed as food and if the existing parameters linking food to nutrition and nutrition to mortality all held, would save something on the order of 9,000 lives globally.
  • In reality, we expect this is a considerable underestimate, both because humanitarian assistance provides considerably more than food and because the mapping from malnutrition to mortality is much steeper in many crisis situations. Existing data from non-crisis contexts may suggest a child whose weight is 2 standard deviations below average has only a moderately elevated risk of death. But a child with similar signs of malnutrition in a crisis situation where the nutrition situation is deteriorating rapidly, and various public health crises are unfolding in parallel, may face a much higher risk of death. In short, it’s very different to be chronically hungry (stable trajectory) than to be suddenly starving (downward trajectory), even if your weight-for-height score is the same.