← Articles
Read in another language
Health

Superbugs and Antibiotic Resistance: A Slow Global Crisis

Jayden

Analyzes global supply chains, industrial policy, and technology issues.

Published

Key points

  • WHO's GLASS surveillance found that one in six laboratory-confirmed bacterial infections was resistant in 2023, and that resistance rose in more than 40% of the drug-bug combinations it tracks between 2018 and 2023.
  • Measured surveillance and modeled forecasts are different things: GRAM's 39.1 million cumulative deaths to 2050 and the O'Neill Review's 10 million a year are projections, and the O'Neill figure was a worst-case, no-action scenario rather than a base case.
  • The map has holes - only 104 countries reported to GLASS in 2023 and nearly half of the world's countries did not submit usable data, so the true burden is likely underestimated rather than absent.
  • The pipeline is shrinking: WHO counted 90 antibacterial agents in clinical development in 2025, down from 97 in 2023, of which only 15 were judged genuinely innovative and just five were effective against at least one "critical" pathogen.
  • September 2024 produced the first numerical political target - a 10% cut in resistance-associated deaths by 2030 - but it is a political commitment rather than binding law, and GRAM's own model says about 92 million projected deaths could be averted with tools that already exist.

In October 2025, the World Health Organization reported that one in six laboratory-confirmed bacterial infections worldwide was already resistant to the antibiotics meant to treat it [source: World Health Organization, 2025]. That single statistic captures a quiet emergency. Antimicrobial resistance (AMR) does not arrive as a sudden outbreak with a name and a map. It advances slowly, infection by infection, as the drugs that have underpinned modern medicine for eighty years gradually stop working. A year earlier, in September 2024, world leaders gathered at the United Nations and did something they had never done before: they agreed to a hard number, pledging to cut deaths linked to bacterial resistance by 10% by 2030 [source: United Nations Environment Programme, 2024]. The gap between that promise and the surveillance data is what makes this the moment to pay attention.

This article separates what we actually measure from what we model, treats the fierce debate over farm antibiotics with care, and lays out the competing ideas for fixing a drug pipeline that has nearly run dry.

In this article

  • What resistance is, and why it is not the flu
  • The numbers: measured versus modeled
  • Why now: a decade of warnings meets a policy moment
  • The One Health tangle: hospitals, farms, and the environment
  • The broken pipeline: why new antibiotics are not coming
  • What would actually move the needle
  • Conclusion: what to watch

What resistance is, and why it is not the flu

Antibiotics work by killing bacteria or stopping them from multiplying. But bacteria reproduce in minutes and swap genetic material readily, so any survivor of a drug can pass on the traits that saved it. Use an antibiotic and you inevitably select for the bugs it cannot kill. This is evolution on fast-forward, and it is why resistance is often called a "slow pandemic": there is no single virus to sequence, only a rising tide of ordinary infections — urinary tract infections, pneumonia, bloodstream infections, gonorrhoea — that become harder, more expensive, and sometimes impossible to cure.

The consequence is not just about exotic diseases. Routine surgery, chemotherapy, childbirth, and organ transplants all rely on antibiotics working in the background to prevent and treat infection. When they fail, the risk of everyday medicine climbs. That is why the WHO frames AMR not as one disease but as a threat to the entire foundation of modern care [source: World Health Organization, 2025].

The numbers: measured versus modeled

Any honest account of AMR has to keep two kinds of figures apart — what surveillance systems actually count, and what statistical models project into the future. They tell related but different stories, and conflating them is the most common mistake in AMR coverage.

What we measure

The measured picture is already sobering. WHO's Global Antibiotic Resistance Surveillance System (GLASS) found that across eight common bacterial pathogens and 22 antibiotics, one in six confirmed infections in 2023 was resistant, and that resistance rose in more than 40% of the drug–bug combinations it tracks between 2018 and 2023, at an average of 5–15% per year [source: World Health Organization, 2025]. The burden is uneven: resistance affected roughly one in three infections in WHO's South-East Asia and Eastern Mediterranean regions, one in five in Africa, and one in seven in the Americas [source: World Health Organization, 2025].

A concrete example comes from the United States. The Centers for Disease Control and Prevention reported in 2025 that infections from NDM-producing carbapenem-resistant Enterobacterales — a group so hard to treat that clinicians nicknamed them "nightmare bacteria" — rose more than 460% between 2019 and 2023 [source: US Centers for Disease Control and Prevention, 2025]. The NDM enzyme disables nearly all available antibiotics, leaving few or no options. For baseline scale, the CDC's threats reporting attributes more than 2.8 million resistant infections and over 35,000 deaths a year in the US alone [source: US Centers for Disease Control and Prevention, 2019].

One caveat sits under all of this: the map has holes. Only 104 countries reported to GLASS in 2023, and nearly half of the world's countries did not submit usable data, with coverage weakest in Africa and parts of the Western Pacific [source: World Health Organization, 2025]. Where surveillance is thin, the true burden is likely underestimated, not absent.

What we model

The headline forecasts come from statistical modeling, and they deserve both attention and skepticism. The most rigorous is the Global Research on Antimicrobial Resistance (GRAM) study, published in The Lancet in September 2024. It estimated that bacterial resistance directly caused more than a million deaths every year between 1990 and 2021, and forecast 39 million cumulative deaths directly attributable to AMR between 2025 and 2050 — roughly three every minute [source: The Lancet, 2024]. Direct annual deaths are projected to reach 1.91 million by 2050, and deaths in which resistance plays a contributing role could climb to 8.22 million a year [source: The Lancet, 2024].

These are estimates with wide uncertainty bands, and the study is careful to distinguish deaths caused by resistance from the larger count associated with it. It also carries a crucial and more hopeful finding: under a scenario of better care for severe infections and better access to the right antibiotics, about 92 million of those projected deaths could be averted [source: The Lancet, 2024]. In other words, the model does not describe a fixed destiny.

It is worth remembering how a famous earlier number has been misused. The 2016 O'Neill Review, commissioned by the UK government, estimated AMR could cause up to 10 million deaths a year by 2050 [source: Review on Antimicrobial Resistance, 2016]. That figure has been repeated for a decade, often stripped of the context that it was a worst-case, no-action scenario rather than a base projection — and researchers have cautioned that AMR burden estimates carry large uncertainty [source: PLOS Medicine, 2016]. The lesson is not that the threat is overblown, but that the scary single number and the measured reality are different things and should be labeled as such.

One measured trend cuts against the gloom: between 1990 and 2021, AMR deaths in children under five fell by about half, thanks to vaccines, cleaner water, and better infection control [source: The Lancet, 2024]. Over the same period deaths in people over 70 rose by more than 80% — a reminder that the burden is shifting toward older, aging populations rather than growing uniformly.

Why now: a decade of warnings meets a policy moment

AMR has been on expert radar since Alexander Fleming himself warned about resistance in 1945. What makes 2024–2026 different is that the warnings collided with political action and worsening data at the same time. The September 2024 UN declaration was the second-ever high-level meeting on AMR and the first to set a concrete target — a 10% cut in resistance-associated deaths by 2030 — backed by a pledge of at least US$100 million in catalytic funding and a goal of having 60% of countries operate funded national action plans by 2030 [source: United Nations Environment Programme, 2024]. Then, through 2025, the WHO's surveillance and pipeline reports arrived to show how far there is to go.

Whether the declaration amounts to more than words is the open question. It is a political commitment, not binding law, and public-health analysts note that targets without sustained financing and enforcement have a poor track record. That tension — ambitious goals against thin follow-through — is the honest frame for "why now."

The One Health tangle: hospitals, farms, and the environment

Resistance does not respect the boundary between human and animal medicine. The dominant framework, endorsed by the UN's "Quadripartite" of the WHO, the World Organisation for Animal Health, the Food and Agriculture Organization, and the UN Environment Programme, is called One Health: the recognition that human, animal, and environmental resistance are linked [source: United Nations Environment Programme, 2024].

Livestock account for a large share of global antibiotic consumption — an estimated 99,502 tonnes in food-producing animals in 2020, projected to rise about 8% by 2030 under business as usual [source: PLOS Global Public Health, 2023]. The mechanism by which this breeds resistance is not in dispute: using antibiotics for growth promotion or routine disease prevention applies constant selective pressure, favoring resistant bacteria on farms.

Here, though, layering matters. The mechanism is established; the magnitude of the human toll from farm use specifically is genuinely contested. Resistance genes do move between animal, human, and environmental reservoirs, but scientists have found it hard to quantify how much human resistance is attributable to agriculture versus overuse in human medicine itself. Advocates for tighter farm rules point to the European Union's 2006 ban on antibiotic growth promoters as a model; critics counter that human clinical misuse — prescriptions for viral illnesses, incomplete courses, over-the-counter access in many countries — remains a major independent driver that farm reform alone will not fix. Both can be true. The responsible reading is that agriculture is one real pressure among several, not a single villain.

The broken pipeline: why new antibiotics are not coming

Even perfect stewardship would only buy time; the world also needs new drugs. Here the news is bleak. In its 2025 analysis, the WHO counted 90 antibacterial agents in clinical development, down from 97 in 2023 [source: World Health Organization, 2025]. Of those, only 15 were judged genuinely innovative, and just five were effective against at least one of the pathogens WHO classes as "critical" [source: World Health Organization, 2025]. For the most dangerous Gram-negative bacteria, the cupboard is nearly bare.

The reason is economic, not scientific. A new antibiotic is, commercially, a bad product: to slow resistance, doctors are told to use it as little as possible, so the drug that society needs most is the one it wants to sell least. Several large pharmaceutical companies have exited antibiotics entirely, and small biotechs that reach the market have gone bankrupt despite approvals. The pipeline is thin because the incentives are inverted.

Pull versus push

Two families of fixes are on the table, and the debate between them is unresolved. "Push" incentives fund research up front — grants and partnerships such as CARB-X and GARDP that de-risk early development. "Pull" incentives reward a successful drug at the finish line. The most-watched pull model is the United Kingdom's "Netflix-style" subscription, made a permanent NHS commissioning route in 2024, which pays a developer a fixed annual fee based on a drug's value to the health system rather than the volume sold — deliberately delinking revenue from prescriptions [source: NICE, 2024]. The United States has repeatedly proposed a similar scheme, the PASTEUR Act, but it has stalled in Congress for years [source: amr.solutions, 2024].

Most experts argue both push and pull are needed, and Europe is weighing "transferable exclusivity vouchers" as another pull mechanism [source: World Health Organization, 2025]. But none is a settled success. Critics question how much such schemes cost, whether they reward genuine innovation or minor tweaks, and — crucially — how new antibiotics would ever reach patients in the lower-income countries that carry the heaviest burden. Subsidizing invention in wealthy markets does not automatically solve global access.

What would actually move the needle

There is no single fix, but the evidence points to a portfolio. Better surveillance so that every country can see resistance in near real time, which WHO wants achieved by 2030 [source: World Health Organization, 2025]. Smarter use of existing drugs — WHO's AWaRe framework urges countries to make "Access" antibiotics at least 70% of consumption and rein in the broader-spectrum "Watch" drugs that drive resistance fastest [source: World Health Organization, 2025]. Basic infection prevention — clean water, sanitation, vaccination, and hospital hygiene — which is precisely what drove the halving of young-child deaths. Rapid diagnostics so clinicians prescribe the right drug instead of guessing. And a financing model that makes discovering new antibiotics viable without encouraging their overuse.

None of these is glamorous, and that is part of the problem: AMR competes for attention and money against threats with sharper edges. But the GRAM model's most important message is that a large share of the projected deaths is avoidable with tools that already exist [source: The Lancet, 2024].

Conclusion: what to watch

Antimicrobial resistance is the rare crisis where the science is clear, the trajectory is measurable, and the solutions are largely known — yet progress is slow because the incentives and attention are misaligned. Over the next few years, a handful of signals will show whether the 2024 promises translate into results: whether resistance-associated deaths bend toward the UN's 10% reduction target by 2030; whether GLASS surveillance finally covers the countries now in the dark; whether the drug pipeline produces anything effective against critical Gram-negative pathogens; and whether pull incentives spread beyond one or two countries into a system that also serves the poorest patients. Superbugs are winning slowly. The open question is whether the response can speed up.

Charts

Countries reporting to WHO's GLASS surveillance system

Countries reporting to WHO's GLASS surveillance system2016 25countries, 2023 104countries25countries2016104countries2023
Measured reporting coverage, not resistance level. Coverage has widened, but roughly 48% of countries still did not report usable data in 2023, with the largest gaps in Africa, parts of the Americas and the Western Pacific - which means the global burden is likely underestimated, not absent.World Health Organization (GLASS 2025) (opens in a new tab)

Antibacterial agents in clinical development

Antibacterial agents in clinical development2023 97agents, Feb 2025 90agents97agents202390agentsFeb 2025
WHO's two pipeline snapshots. The count of drugs in clinical development fell between them - a measured inventory of programmes, not a forecast of approvals.World Health Organization (antibacterial pipeline report 2025) (opens in a new tab)

The 2025 pipeline, broken down

The 2025 pipeline, broken downTraditional 50agents, Non-traditional 40agents, Judged innovative 15agents, Effective vs a "critical" pathogen 5agents50agentsTraditional40agentsNon-traditional15agentsJudged innovative5agentsEffective vs a "critical" pathogen
Traditional (50) plus non-traditional (40) together make up the 90 agents. The last two bars are WHO's assessment of subsets of that same 90 - not additional programmes - so they must not be added to the first two.World Health Organization (antibacterial pipeline report 2025) (opens in a new tab)

AMR-associated deaths per year, GRAM model

AMR-associated deaths per year, GRAM model2021 (estimated) 4.71million per year, 2050 (forecast) 8.22million per year4.71million per year2021 (estimated)8.22million per year2050 (forecast)
Modeled, not counted, and carrying wide uncertainty bands. "Associated" deaths - those in which resistance plays a contributing role - are a broader measure than deaths directly attributable to resistance, which the same model forecasts at 1.91 million a year in 2050. The two measures should never be used interchangeably.GRAM Project / The Lancet (2024) (opens in a new tab)

Antimicrobial use in food-producing animals

Antimicrobial use in food-producing animals2020 (estimate) 99,502tonnes, 2030 (projected, business as usual) 107,472tonnes99,502tonnes2020 (estimate)107,472tonnes2030 (projected, business as usual)
An estimate and a business-as-usual projection, not a measured inventory. The mechanism by which farm use selects for resistant bacteria is established; the share of human resistant infections attributable to agriculture specifically is contested and is deliberately not quantified here.Mulchandani et al., PLOS Global Public Health (2023) (opens in a new tab)

Timeline

  1. Alexander Fleming warns that bacteria can become resistant to penicillin - resistance has been on the expert radar ever since.

  2. The European Union bans antibiotic growth promoters in livestock; enforcement of comparable rules varies elsewhere.

  3. The O'Neill Review, commissioned by the UK government, publishes a worst-case, no-action scenario of up to 10 million deaths a year by 2050.

    Review on Antimicrobial Resistance (opens in a new tab)
  4. de Kraker et al. caution in PLOS Medicine that AMR burden estimates carry large uncertainty and that the 10 million figure is one scenario, not a base projection.

    PLOS Medicine (opens in a new tab)
  5. The US CDC's AR Threats Report attributes more than 2.8 million resistant infections and over 35,000 deaths a year in the United States.

    US Centers for Disease Control and Prevention (opens in a new tab)
  6. The UK pilots a "Netflix-style" subscription that pays antibiotic developers a fixed annual fee based on value to the health system rather than volume sold.

    NICE / NHS England (opens in a new tab)
  7. The subscription model becomes a permanent NHS commissioning route.

    NICE / NHS England (opens in a new tab)
  8. The first procurement under the permanent UK model takes place.

    NICE / NHS England (opens in a new tab)
  9. The GRAM study publishes in The Lancet, forecasting 39.1 million cumulative deaths directly attributable to AMR between 2025 and 2050, and about 92 million avertable under a better-care scenario.

    GRAM Project / The Lancet (opens in a new tab)
  10. UN member states adopt a political declaration targeting a 10% reduction in bacterial AMR-associated deaths by 2030, with at least US$100 million in catalytic funding and 60% of countries running funded national action plans by 2030.

    United Nations Environment Programme (opens in a new tab)
  11. The CDC reports that NDM-producing carbapenem-resistant Enterobacterales infections rose more than 460% in the United States between 2019 and 2023.

    US Centers for Disease Control and Prevention (opens in a new tab)
  12. WHO reports the clinical antibacterial pipeline has fallen to 90 agents, of which 15 are judged innovative and five effective against a "critical" pathogen.

    World Health Organization (opens in a new tab)
  13. WHO's GLASS report finds one in six laboratory-confirmed bacterial infections was resistant in 2023, across eight pathogens and 22 antibiotics.

    World Health Organization (opens in a new tab)

Analysis

One in six is a floor, not a ceiling

The GLASS headline counts only what surveillance systems actually saw. With 104 reporting countries in 2023 and nearly half of the world's countries absent from the dataset - most conspicuously in Africa, parts of the Americas and the Western Pacific - the thin parts of the map are where resistance is least likely to be recorded, not where it is least likely to exist.

Attributable and associated are not the same number

GRAM reports two mortality measures and they are routinely conflated. Deaths directly attributable to resistance are forecast at 1.91 million a year in 2050; deaths in which resistance plays a contributing role are forecast at 8.22 million. Quoting the larger figure as if it were the smaller one inflates the burden; quoting only the smaller one understates the clinical reality. Both need their label.

The 10 million figure was a scenario, not a projection

The O'Neill Review's up-to-10-million-deaths-a-year estimate has been repeated for a decade, usually stripped of the context that it described a worst case in which nothing was done. Researchers writing in PLOS Medicine cautioned that AMR burden estimates carry large uncertainty. The threat is not overblown - but a scenario and a measurement should never be cited in the same breath.

The burden is aging, not simply growing

Between 1990 and 2021 AMR deaths in children under five fell by about half, while deaths in people over 70 rose by more than 80%. Vaccines, cleaner water and infection control did the first; demographic aging and hospital-intensive medicine are driving the second. A single global total hides two trends pointing in opposite directions.

The thin pipeline is an economics problem, not a science problem

A new antibiotic is commercially perverse: to preserve its usefulness, prescribers are told to use it as little as possible, so the drug society needs most is the one it wants to sell least. Large pharmaceutical companies have exited the field and small developers have gone bankrupt after approval. That is why WHO's count fell from 97 to 90, and why only five of those 90 work against a critical pathogen.

On farms, the mechanism is established and the magnitude is not

Sub-therapeutic use for growth promotion and routine prophylaxis applies constant selective pressure - that part is not in dispute. What remains genuinely contested is how much human resistance is attributable to agriculture as against misuse in human medicine itself. Responsible coverage names agriculture as one real pressure among several and resists putting a clean percentage on it.

Comparison

Every headline AMR number, sorted by what kind of claim it is
FigureType of claimSource
1 in 6 confirmed bacterial infections resistant (2023)Measured - laboratory surveillance, limited to reporting countriesWHO GLASS, 2025
Resistance rose in more than 40% of tracked drug-bug combinations, 2018-2023, at 5-15% a yearMeasured - a range, not a point valueWHO GLASS, 2025
NDM-CRE infections up more than 460%, 2019-2023 (US)Measured - national surveillance, a floor stated as "more than"US CDC, 2025
More than 1 million direct deaths a year, 1990-2021Modeled from observed data - historical estimate, not a countGRAM / The Lancet, 2024
39.1 million cumulative direct deaths, 2025-2050Modeled forecast - wide uncertainty bandsGRAM / The Lancet, 2024
Up to 10 million deaths a year by 2050Modeled worst-case scenario assuming no action - not a base projectionO'Neill Review, 2016
10% reduction in AMR-associated deaths by 2030Political target - a commitment, not binding lawUN political declaration, 2024
Resistance by WHO region in 2023 - reported as ratios, exactly as WHO states them
RegionShare of confirmed infections that were resistantBasis
South-East Asia and Eastern MediterraneanAbout 1 in 3Measured, 2023 GLASS reporting
AfricaAbout 1 in 5Measured, 2023 GLASS reporting - coverage weakest here
AmericasAbout 1 in 7Measured, 2023 GLASS reporting
Global1 in 6Measured across 8 pathogens and 22 antibiotics
Push and pull incentives: what each is trying to fix, and what is unresolved
ApproachExampleStatusOpen question
Push - fund the research up frontCARB-X, GARDP grants and partnershipsOperatingDe-risks early science but does not fix the revenue problem after approval
Pull - reward the finished drugUK "Netflix-style" NHS subscription, delinking payment from volumeVerified policy - permanent commissioning route since 2024, first procurement the same yearWhether one national scheme is large enough to change global developer economics
Pull - proposed legislationUS PASTEUR ActRepeatedly introduced, stalled in Congress for yearsWhether it passes at all, and at what cost to public budgets
Pull - regulatory rewardEurope's transferable exclusivity vouchersUnder discussionWhether it rewards genuine innovation or minor tweaks, and who pays through delayed generic entry

Process

  1. See it

    Surveillance good enough that every country can track resistance in near real time - the coverage WHO wants in place by 2030, and the gap behind the 104-country GLASS dataset.

  2. Use existing drugs better

    WHO's AWaRe framework urges countries to make "Access" antibiotics at least 70% of consumption and to rein in the broader-spectrum "Watch" drugs that drive resistance fastest.

  3. Prevent the infection

    Clean water, sanitation, vaccination and hospital hygiene - precisely the combination credited with halving AMR deaths in children under five between 1990 and 2021.

  4. Diagnose before prescribing

    Rapid diagnostics so clinicians give the right drug rather than guessing, which is where much avoidable human-medicine overuse originates.

  5. Pay for discovery without paying per prescription

    A financing model - subscription-style pull incentives alongside push funding - that makes new antibiotics viable without rewarding volume.

  6. Get the drugs to where the burden is

    Access in lower-income countries, which carry the heaviest burden and which subsidising invention in wealthy markets does not automatically reach.

Sources

  1. World Health Organization — Global antibiotic resistance surveillance report 2025 / "WHO warns of widespread resistance to common antibiotics worldwide" (2025-10-13).View source (opens in a new tab)
  2. World Health Organization — Analysis of antibacterial agents in clinical and preclinical development: overview and analysis 2025 (2025-10-02).View source (opens in a new tab)
  3. The Lancet (GRAM Project, IHME) — Global burden of bacterial antimicrobial resistance 1990–2021: a systematic analysis with forecasts to 2050 (2024-09).View source (opens in a new tab)
  4. United Nations Environment Programme — World leaders commit to decisive action on antimicrobial resistance (UNGA Political Declaration) (2024-09-26).View source (opens in a new tab)
  5. US Centers for Disease Control and Prevention — CDC report finds sharp rise in dangerous drug-resistant bacteria (NDM-CRE) (2025-09).View source (opens in a new tab)
  6. US Centers for Disease Control and Prevention — Antimicrobial Resistance Facts and Stats / 2019 AR Threats Report.View source (opens in a new tab)
  7. Review on Antimicrobial Resistance (Jim O'Neill, chair) — Antimicrobial Resistance: Tackling a crisis for the health and wealth of nations (2016).View source (opens in a new tab)
  8. de Kraker et al., PLOS Medicine — Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050? (2016).View source (opens in a new tab)
  9. Mulchandani et al., PLOS Global Public Health — Global trends in antimicrobial use in food-producing animals: 2020 to 2030 (2023).View source (opens in a new tab)
  10. NICE / NHS England — A new model for evaluating and purchasing antimicrobials in the UK (subscription "Netflix" model) (2024).View source (opens in a new tab)
  11. AMR.Solutions — Incentives: PASTEUR Act and pull-incentive tracker (2024).View source (opens in a new tab)

Tags

  • #antimicrobial-resistance
  • #superbugs
  • #antibiotics
  • #amr
  • #one-health
  • #public-health