Over 170 countries have antimicrobial resistance action plans and in 2024 only 10 percent of countries reported dedicated domestic funding

What is happening?

Bacteria that survive the antibiotics used against them turn treatable infections into untreatable ones. The global burden study published on 2024-09-16 estimates that in 2021 there were 1.14 million deaths attributable to bacterial antimicrobial resistance, with an uncertainty interval of 1.00 to 1.28 million, and 4.71 million deaths associated with it, with an interval of 4.23 to 5.19 million. The two figures answer different questions and are not interchangeable, and the difference between them decides what any target actually promises.

The world has an agreed number to hit. At the high-level meeting on 2024-09-26 the United Nations General Assembly adopted a political declaration setting a 10 percent reduction in deaths associated with bacterial resistance by 2030, measured against a 2019 baseline of 4.95 million associated deaths. In May 2026 the World Health Assembly adopted an updated global action plan running to 2036, which restates that goal.

The machinery beneath the goal is thinner than the goal. The adopted plan records that over 170 countries have developed multisectoral national action plans and, in the same document, that in 2024 only 10 percent of countries reported dedicated domestic funding for those plans, which it describes as significantly constraining implementation. The self-assessment reported in the 2024 declaration puts 52 percent of countries with a functioning multisectoral coordinating mechanism and 68 percent implementing their plan.

Whose problem is this?

Reaching the target requires action in every country, and the declaration and the plan both set global numbers without dividing them into national shares. No source opened here names a mechanism that obliges any government to fund the plan it has already written.

Where does this problem end?

What is the state now, and what should it be?

Now

Should be — the targets are stated by the bodies themselves, not chosen here.

The metric matters as much as the number. The target is set against deaths associated with resistance, which for 2019 the declaration puts at 4.95 million; the attributable count for the same year is 1.27 million. A statement of progress that switches between the two measures a different thing.

How big is it?

Between 4.23 and 5.19 million deaths a year were associated with bacterial antimicrobial resistance in 2021, around a central estimate of 4.71 million. That interval is the one the study published for that single definition.

The attributable count for the same year is smaller and is stated alongside it: 1.14 million, with an interval of 1.00 to 1.28 million. Which definition is in use moves the reported figure far more than the statistical uncertainty inside either one does, so the two are set out here side by side rather than merged.

Forward, the same study gives a reference scenario of 39.1 million cumulative deaths attributable to resistance between 2025 and 2050, with an interval of 33.0 to 46.0 million, and 169 million cumulative associated deaths, interval 145 to 196 million. For the single year 2050 it gives 1.91 million attributable and 8.22 million associated.

The shape of the rise is not what the totals suggest. Between 2022 and 2050 deaths are forecast to rise by 69.6 percent while disability-adjusted life years rise by 9.4 percent, and the all-age attributable rate of those life years is forecast to fall from 536 per 100,000 in 2022 to 496 per 100,000 in 2050. By 2050, 65.9 percent of attributable deaths are forecast to be in people aged 70 and older.

Under what conditions does it arise?

1. The burden is moving in two directions at once. Between 1990 and 2021 deaths from resistance fell by more than half among children under 5, and fell in every super-region for that age group, which the study attributes to infection prevention. Over the same period they rose by more than 80 percent among adults aged 70 and older. A single global total conceals both movements.

2. Restricting use and widening access pull against each other. Reducing antimicrobial use lowers the pressure that selects for resistance. Applied where access is already short, the same restriction removes treatment. The declaration of 2024 states that lack of access is responsible for more deaths than resistance is, and the antibiotic use report published in April 2025 reaches a compatible reading from the data side, finding use varying by a factor of ten between the highest-using and lowest-using countries and describing that spread as evidence of both overuse and low access.

3. The supply of new drugs is not arriving. The plan adopted in 2026 records that the global pipeline for new antimicrobials and diagnostics remains stagnant, attributing this to underinvestment and limited incentives for research and development, compounded by inequitable access to quality-assured products and weak stewardship across sectors. Its remedy names push mechanisms and pull incentives that separate revenue from sales volume.

4. Animal-sector use is now measured and is falling, while its share of the human burden is not measured at all. Between 2020 and 2022 antimicrobial use in animals fell by 5 percent globally, with reporting coverage reaching 71 percent of global animal biomass, and 71 percent of countries reported no use of antimicrobials as growth promoters in 2022. How much of the human burden that use produces is not quantified in any source opened here.

What has been tried?

What was found?

Why is it still unsolved?

Coordination failure — the target is global, the money is national, and nothing joins the two.

The first leg is that writing a plan and funding a plan are decided in different rooms. Over 170 countries have developed multisectoral national action plans, and in 2024 only 10 percent of countries reported dedicated domestic funding for them, a state the adopted plan itself describes as significantly constraining implementation. The intermediate self-assessments run in the same direction: 52 percent with a functioning coordinating mechanism, 68 percent implementing, and about a quarter with a costed and budgeted plan and effective monitoring. Adoption is cheap and near-universal, and every step after adoption thins out.

The second leg is that the target has no owner below the global level. The declaration sets one reduction against one worldwide baseline. No source opened here assigns any part of that reduction to any country, so no government can fall short of a share of its own, and the reporting the adopted plan provides for runs back to the same assembly of governments, in 2027, 2029 and 2031.

The third leg is that the largest measured lever is not the one the incentive debate is built around. Under a better care scenario, which it describes as improvements to health-care systems including access to existing antibiotics, the burden study estimates 92.0 million cumulative deaths averted between 2025 and 2050, and states in the same place that this total is inclusive of deaths attributable to or associated with resistance as well as deaths unrelated to it. Under a scenario in which a new pipeline of drugs against Gram-negative bacteria arrives, it estimates 11.08 million resistance deaths averted over the same period, which it puts at about a third of the cumulative attributable deaths in its reference scenario. Because the two totals are counted on different bases, one does not divide into the other; the comparison the study itself draws is that the better care scenario averts about five times what the drug scenario does.

The fourth leg is that the instrument has to move in two directions at once. Restricting antimicrobial use is what slows resistance, and the same restriction applied where access is already short removes treatment from people who need it. Member States stated in 2024 that the second harm is the larger one. A plan that only tightens use would fail against the statement its own signatories adopted, and a plan that only widens access would work against the resistance target, so the instrument carries both and the tension between them is resolved country by country, which returns to the first leg.

What observation would mean it is solved?

The condition is stated: deaths associated with bacterial antimicrobial resistance 10 percent below the 2019 baseline of 4.95 million by 2030. Three cautions attach to reading it.

The metric is associated deaths, and the attributable count is the one more often quoted. For 2019 the declaration puts the associated count at 4.95 million and the attributable count at 1.27 million, and the target is set against the first. A reported fall that has switched between the two measures is not a fall.

Deaths have not been re-estimated since the baseline was set, in any source opened here. The reference the World Health Organization carried on 2026-07-16 for its burden figures is still the study published in 2024 with data for 2021, and the plan adopted in 2026 names no interim measurement point. What has been re-measured is resistance prevalence rather than mortality, and that measurement recorded increases in 40 percent of the monitored combinations between 2018 and 2023.

Plan counts, surveillance counts and funding counts measure different things and none of them is the target. Reporting to the global surveillance system rose from 25 countries for 2016 to 104 for 2023, which is a real gain in visibility and not a reduction in deaths. Secondary conditions worth watching alongside the reduction target are the share of human antibiotic use from the Access group against the 70 percent goal, and the share of countries with funded plans against the 60 percent goal, because both are stated by the same declaration and both are measurable before 2030.

What is it connected to?

Fills with researchaccess to antimicrobials and diagnostics as a burden in its own right · health-system capacity and infection prevention · water, sanitation and hygiene · immunisation coverage · antimicrobial use in animal production · drug-resistant tuberculosis, reported at 3.2 percent of first-episode tuberculosis in 2024 · the economics of pharmaceutical research incentives. Relation type and evidence grade were not established for any of these.

What these sources do not say

• How much of the human burden comes from antimicrobial use in animals. No source opened here attributes a share of human resistance deaths to livestock use. What is quantified runs in the other direction, treating the sector as a bearer of cost and a falling user: cumulative losses in livestock production of USD 575 billion by 2050, a predicted 11 percent decline in livestock production in low-income countries by 2050, use down 5 percent globally between 2020 and 2022, and a valuation of USD 120 billion in global gross domestic product from a 30 percent cut in livestock use within five years.
• Any country-level death figure. The burden study states that estimates were produced for 204 countries and territories from 1990 to 2021, and those values sit in an appendix and in a separately published dataset. The hosts of both refused automated retrieval, so no country value was read for this document. The values exist and are published; they were not obtained here.
• A global death estimate more recent than 2021. The fact sheet as updated 2026-07-16 still carries the 2021 figures and lists the 2024 study as its first reference, and the plan adopted in 2026 sets no interim measurement point before 2030.
• What share of countries meets the appropriate-antibiotic-use target. The target is that at least 70 percent of human antibiotic use come from the Access group by 2030. The measured global share of use was 53 percent in 2022, with Watch antibiotics above 70 percent of use in nearly a third of countries, and the antibiotic use report of April 2025 records many countries not meeting the target without stating how many. The share of use is published; no source opened here gives the share of countries.
• Why only 10 percent of countries reported dedicated domestic funding. The adopted plan states the figure and its consequence for implementation and stops there. No source opened here carries a reason on the record from any government.
• The standalone economics study behind the USD 412 billion and USD 443 billion figures. It was reached only through three bodies that cite it, and the document itself was not located.
• How the two economic estimate sets relate. The 2017 scenario endpoints and the 2035 cost and productivity pair are carried side by side in the 2024 declaration without either being described as superseding the other, and no opened source reconciles them.

Every body named above is also the publisher of a source cited for it, so the positions set out here are the ones those bodies published themselves. Where a document has been superseded, the superseding document is cited alongside it rather than in place of it.

See the evidence

The burden study records its funders as a United Kingdom government health fund and a private research charity. Its publisher page refuses automated retrieval, so the copy cited above is the open-access deposit in PubMed Central.

People affected

Estimated range 4,230,000–5,190,000 As of 2021

Sensitivity The interval above is the statistical uncertainty the study published for one definition. The choice of definition moves the figure far more than that uncertainty does, and the two are therefore set out side by side rather than merged: deaths attributable to bacterial resistance in the same year are 1.14 million with an interval of 1.00 to 1.28 million, against 4.71 million associated with an interval of 4.23 to 5.19 million. This count excludes people who survive resistant infections, people harmed by lack of access to antimicrobials and diagnostics, which the 2024 United Nations political declaration states is responsible for more deaths than resistance itself, and resistance in non-bacterial pathogens. In the other direction the figure is a modelled estimate for 2021 rather than a count, and as of the World Health Organization fact sheet updated 2026-07-16 no later global estimate had displaced it, so it is five years old.

Regional breakdown The study states that estimates were produced for 204 countries and territories from 1990 to 2021, but those values sit in an appendix and in a separately published dataset record whose hosts returned HTTP 403 to automated retrieval, so no country value was read. The geographic figures that were read are super-region forecasts of cumulative deaths to 2050 rather than current burden, and super-regions are not administrative units.