Global health aid is on track for $38.4bn this year, a level last seen in 2009

The Institute for Health Metrics and Evaluation has tracked development assistance for health every year for more than a decade, and the exercise is usually a matter of shifting shares between donors. This year's edition, published in The Lancet, records something different: a contraction that returns global health aid to where it stood in 2009 [s1].

The accounting

Development assistance for health peaked at US$80.3 billion in 2021, the height of the pandemic response, and had already fallen to $49.6 billion by 2024 [s1]. For 2025, incorporating budget cuts announced by major donors — particularly reductions in US bilateral aid — the authors' preliminary estimate is $38.4 billion [s1].

The forecast does not show a rebound. Under current policies, the authors project continued stagnation to 2030, reaching $36.2 billion [s1]. Their sensitivity analyses put the 2025 figure between $36.8 billion in a pessimistic scenario and $40.0 billion in an optimistic one, depending on the scale of US cuts, and the 2030 figure between $34.5 billion and $37.8 billion on the same basis [s1].

These are estimates, not audited totals. They are built from the OECD's Creditor Reporting System, online databases from agencies including the Global Fund and Gavi, and financial reports from private philanthropies and non-governmental organisations, with disbursements categorised by source, agency, health focus area and recipient country using keyword tagging methods developed across 15 years of the same report series [s1]. The 2025 number in particular is a preliminary estimate incorporating announced rather than executed cuts [s1].

That distinction matters and cuts both ways. Announced cuts can be reversed, softened, or partially backfilled. They can also turn out to have been understatements.

What is shrinking, and what is not

The report names the agencies contracting their disbursements for major infectious diseases and childhood vaccines, including the UK's Foreign, Commonwealth & Development Office, the US Agency for International Development, and Agence Française de Développement [s1].

One category held. Because the major multilateral development banks were protected from the cuts, the World Bank's relative share of total development assistance for health has increased [s1].

That shift is worth dwelling on, because a bank is not a grantmaker. A larger World Bank share of a smaller total means more of what remains arrives as lending against government balance sheets rather than as grants — which changes who carries the fiscal risk of a health programme, and which countries can access it at all.

What it means for one disease

Aggregate financing figures are hard to feel. A companion modelling study in The Lancet Global Health, published in the same month, attempts the translation for tuberculosis [s2].

The authors calibrated a deterministic model of Mycobacterium tuberculosis transmission, progression and care against epidemiological indicators in 79 countries representing 91% of global tuberculosis incidence and 90% of global tuberculosis mortality in 2023 [s2]. They projected three scenarios: 2024 funding levels continuing; USAID funding terminated from 2025; and further reductions through the Global Fund in line with announced donor decisions on top of that termination [s2].

Termination of USAID funding alone, in their model, leads to 1.4 million (95% uncertainty interval 1.1 to 1.7) excess episodes of symptomatic tuberculosis and 537,700 (451,900 to 662,300) excess deaths by 2035 [s2].

Further announced reductions add to that baseline. Relative to the USAID-termination scenario, the model attributes an additional 2.8 million (2.1 to 3.7) episodes and 1.0 million (0.8 to 1.3) deaths to further US reductions; 257,600 (192,500 to 332,900) episodes and 90,500 (72,400 to 112,800) deaths to France; 206,000 (153,900 to 266,100) and 72,400 (57,900 to 90,100) to the UK; and 124,700 (93,200 to 161,000) and 43,800 (35,000 to 54,500) to Germany [s2].

How much weight those numbers carry

Less than their precision suggests, and the modelling paper is explicit about why. Its central assumption is that tuberculosis treatment initiation rates fall in proportion to budget reductions in each scenario [s2]. That is a defensible simplifying assumption and it is almost certainly wrong in detail. Programmes triage; governments partially substitute; some services degrade faster than money is withdrawn and others survive on inertia for a year or two.

The model also runs to 2035, which means most of the projected burden accrues years after any of these decisions could be reversed. Its output is a statement about a trajectory, not a forecast of what will happen.

What can be said with more confidence is directional: the financing measured in the Lancet accounting has been withdrawn, tuberculosis programmes in many settings rely heavily on international donor funding [s2], and reductions of the size recorded will produce excess disease unless something replaces them.

The proposed replacement

Both papers land on the same policy conclusion, and both hedge it. The financing study warns that major reductions from historically leading donors threaten to widen health disparities unless mitigated by increased domestic resource mobilisation or alternative financing mechanisms, and calls for greater efficiency, strategic reprioritisation and strengthened fiscal resilience in recipient countries [s1]. The tuberculosis model concludes that expanded support from domestic and international donors is essential to address immediate gaps in prevention, diagnosis and treatment services [s2].

"Domestic resource mobilisation" is doing heavy lifting in that sentence. The countries most dependent on development assistance for health are, by construction, those with the least domestic fiscal room, and many are simultaneously managing debt service costs. The financing study forecasts stagnation through 2030 precisely because it does not assume that substitution materialises [s1].

What to watch

Three things. Whether the 2025 estimate of $38.4 billion holds when executed disbursements are counted rather than announcements [s1]. Whether the World Bank's rising share translates into actual health lending uptake or simply a larger slice of a shrinking pie [s1]. And whether facility-level data — treatment initiations, testing volumes, stockouts — begin to show the pattern these models assume, because until they do, the estimates above remain projections rather than measurements.

The financing study was funded by the Gates Foundation; the tuberculosis modelling study reports no funding [s1][s2].

Sources

• [s1] Tracking development assistance for health, 1990-2030: historical trends, recent cuts, and outlook — The Lancet, July 2025 (Crossref record created 16 July 2025). https://doi.org/10.1016/S0140-6736(25)01240-1
• [s2] The potential impact of reductions in international donor funding on tuberculosis in low-income and middle-income countries: a modelling study — The Lancet Global Health, published online July 2025 (September 2025 issue). https://doi.org/10.1016/S2214-109X(25)00232-3