In a microbiology lab, a paper disc soaked in antibiotic is dropped onto a lawn of bacteria. If the drug works, a clear ring opens around it — the zone of inhibition. Across the world, those rings are closing. This is what the arithmetic of that closing looks like.
Deaths directly attributable to bacterial antimicrobial resistance — people who would likely have lived had the infecting bug still been susceptible to the drug. Resistance played a role in 4.71 million more. GRAM / IHME, The Lancet, 2024 (modelled estimate).
01 · The rate today
The WHO's 2025 surveillance report pooled more than 23 million laboratory-confirmed bacterial infections from 104 countries — about 70% of the world's population. In 2023, one in six of those infections was resistant to the antibiotic used to treat it. But the median hides an enormous spread.
Source: WHO, Global antibiotic resistance surveillance report 2025 (data year 2023). Regional figures are reported medians expressed as ratios; percentages shown are the direct conversion (1 in 6 ≈ 17%). 3GC = third-generation cephalosporins, the standard first-line treatment for these bloodstream infections.
02 · Geography
Where health systems are weakest, resistance is highest — and diagnostics are scarcest, so the true figure is likely worse than the one measured. Below, the four WHO regions for which a pooled resistance share was reported in 2025.
Source: WHO, Global antibiotic resistance surveillance report 2025. Map is hand-simplified and illustrative; region shapes are schematic. Grey regions are not "low resistance" — WHO did not publish a single pooled share for them in this summary.
03 · The trajectory
Between 1990 and 2021 the global death toll barely moved — better hospitals and vaccines offsetting rising resistance. The forecast is what changes: by 2050, deaths directly attributable to resistance are projected to rise about 68%, and deaths in which resistance plays a role to reach 8.2 million a year.
Source: GRAM Project / IHME, The Lancet, Sept 2024. Values are modelled estimates for the three published reference years; intermediate years are not plotted, so the connecting lines indicate direction, not year-by-year path. Cumulatively, the study projects 39 million deaths directly attributable to resistance between 2025 and 2050, and 169 million associated deaths.
04 · Who, and what
The single most striking finding of the 2024 analysis is a reversal. Vaccines, clean water and better paediatric care cut resistant-infection deaths in young children by more than half. Over the same period, deaths among people over 70 nearly doubled — an ageing world, spending more time in hospitals, where the hardest organisms live.
CHANGE IN ATTRIBUTABLE DEATHS, 1990 → 2021
ATTRIBUTABLE DEATHS BY ORGANISM GROUP
Source: GRAM Project / IHME, The Lancet, Sept 2024. MRSA = methicillin-resistant Staphylococcus aureus; carbapenems are a last-line class for gram-negative infections. All values modelled.
05 · The driver
Resistance is evolution on a deadline, and the deadline is set by exposure. Human antibiotic consumption rose by nearly half in under two decades — and the far larger tonnage goes into livestock, where antimicrobials are routinely used across whole herds and flocks.
HUMAN USE — DEFINED DAILY DOSES PER 1,000 PEOPLE PER DAY
ANIMAL USE — TONNES OF ANTIMICROBIALS, FOOD-PRODUCING ANIMALS
Sources: GRAM Project, The Lancet Planetary Health, 2021 (human consumption, 204 countries, spatial model); Mulchandani et al., PLOS Global Public Health, 2023 (livestock). *2030 livestock figure is a projection from current trends, not a measurement.
06 · The other side
The last genuinely new class of antibiotic to reach patients was discovered in 1987; almost everything since is a chemical variation on an older theme, which resistance tends to defeat faster. As of February 2025, the WHO counted 90 antibacterials in clinical development worldwide — down from 97 two years earlier. Filter for what's actually new, and for what works on the organisms WHO calls critical, and the number collapses.
Source: WHO, Analysis of antibacterial agents in clinical and preclinical development: overview and analysis 2025 (Oct 2025). The remaining 40 candidates are non-traditional approaches — bacteriophages, antibodies, microbiome therapies. "Innovative" is WHO's own test: no known cross-resistance, a new target, a new mode of action, or a new class.
07 · The bill
The most extensive modelling to date — a joint effort by the World Organisation for Animal Health, IHME, the Center for Global Development and RAND Europe — put numbers on the second-order damage: hospitals, herds, harvests, and the food supply that depends on them.
Cumulative deaths, 2025–2050
39M
Directly attributable to bacterial resistance — plus 169 million deaths in which it plays a part.
Annual GDP loss by 2050
$1.7T
Under a high-resistance scenario. Extra health-care spending rises from $66 billion a year today to $159 billion.
Food supply put at risk
2 billion
People whose food supply could be jeopardised as resistant infections cut into livestock production.
Source: EcoAMR series (WOAH, IHME/GRAM, Center for Global Development, RAND Europe), 2024. All figures are scenario projections, not measurements. The same analysis estimates a return of $28 for every $1 invested in acting now.
Why it matters
Antibiotics are the quiet floor under modern medicine: caesareans, chemotherapy, transplants, hip replacements, neonatal intensive care. All of it assumes an infection can be stopped. The same 2024 forecast that projects 39 million deaths also models the alternative — and the alternative is enormous. Better infection care and reliable access to the right existing drugs would avert an estimated 92 million deaths by 2050. New antibiotics aimed at gram-negative bacteria would avert about 11 million more.
Like this? Get a new data story in your inbox every morning.