Antimicrobial resistance (AMR) crisis

What this driver is

Antimicrobial resistance is a continuing One Health selection mechanism: bacteria, fungi, and other microbes that survive antibiotics, antifungals, and related drugs, then spread in clinics, farms, water, and households. It is not a single outbreak, not a unique Factrail death total, and not access to essential medicines. Access is whether a quality drug arrives. This node is whether the drug still works.

The legal-political core is the WHO Global Action Plan adopted as WHA68.7 (26 May 2015), the UN General Assembly high-level meeting political declaration of 21 September 2016, and the second high-level meeting declaration of 26 September 2024 (A/RES/79/2). Quadripartite FAO–UNEP–WHO–WOAH coordination is the institutional neighbourhood. A modelled “10 million deaths by 2050” line from the 2016 O’Neill review is a named scenario, not a measured series on this page.

This driver is not COVID-19 pandemic, not infectious-disease transmission as a standing pressure, and not pandemic preparedness. Those companions can move with AMR. They do not replace it. It is not ukraine-war.

It applies worldwide from at least the 2015 GAP onward, with a longer microbiological history. Veterinary growth-promotion, hospital empiric therapy, and poor diagnostics are channels, not competing slugs.

Mechanism

AMR moves welfare through named channels.

First, failed first-line therapy. Third-generation cephalosporin-resistant E. coli bloodstream infection is the SDG 3.d.2 human-health lens. When empiric treatment fails, sepsis lengthens, last-resort carbapenems are used, and last-resort drugs select further. A resistance percentage is not a body-count.

Second, agricultural and environmental selection. Routine farm use, untreated effluent, and aquaculture select the same gene families that later appear in human isolates. One Health is a mechanism, not a slogan. WOAH ANIMUSE and FAO InFARM are neighbouring official systems. They are not this meter.

Third, missing diagnostics and delayed stewardship. GLASS participation grew; testing coverage still differs sharply by income. A high reported resistance share in a thin-testing country can be a surveillance artefact. The 2018 World point in the SDG series jumps; this page treats that year as a coverage-mix observation, not a proven doubling of biology.

Fourth, pipeline failure and access contradiction. Reserve antibiotics are scarce in some low-income settings while Watch drugs are overused in others. The 2024 UNGA text set a target that at least 70% of human antibiotic use should be WHO Access-group. Missing Access is not a licence to invent a unique R&D coefficient.

Fifth, political declaration without funded national action plans. The 2024 declaration’s catalytic-funding and 60%-of-countries-funded-plans language is a dated commitment. A commitment is not an implemented plan.

Timing is years for gene spread and hours for a failed empiric dose. A later UNGA meeting in 2029 would be a new fact.

Evidence base

The best evidence is WHA and UNGA instruments plus GLASS/SDG surveillance.

WHA68.7 adopted the Global Action Plan on 26 May 2015 (https://apps.who.int/gb/ebwha/pdf_files/WHA68/A68_R7-en.pdf (opens in a new tab)). The 2016 high-level meeting political declaration is A/RES/71/3 (21 September 2016). The 2024 meeting on 26 September adopted A/RES/79/2, including a target to reduce bacterial-AMR-associated deaths 10% by 2030 against a 2019 baseline of 4.95 million associated deaths named in WHO’s readout (https://www.who.int/news/item/26-09-2024-world-leaders-commit-to-decisive-action-on-antimicrobial-resistance (opens in a new tab)). That 4.95 million figure is the 2019 GRAM/Lancet associated-death estimate (Murray et al., Lancet, 2022). Attributable deaths in the same paper were about 1.27 million. Associated is not attributable. This page names both.

The meter is the World series for SDG 3.d.2 — percentage of E. coli bloodstream infections resistant to third-generation cephalosporins — as republished by Our World in Data from WHO GLASS (https://ourworldindata.org/grapher/share-of-e-coli-bloodstream-infections-due-to-antimicrobial-resistant-bacteria.csv (opens in a new tab)): 35.8% (2016), 32.4% (2017), 69.2% (2018), 36.6% (2019), 47.4% (2020), 39.21% (2021), 44.7% (2022). Values on this driver are that percentage divided by 100. The 2018 spike is disclosed as a likely reporting-mix break, not a proven biological doubling. GLASS 2025 later estimated 44.8% of E. coli BSI resistant to 3GC in its modelled 2023-era assessment; that is a neighbouring official figure, not copied onto the 2022 meter.

Evidence is strongest for a signed UNGA declaration and a named SDG indicator. It is weaker for any unique 2050 death forecast. It is weakest where a political target is asked to prove a national plan was funded.

Effect on welfare indicators

This driver, when it actually runs, is a first-order channel into infection survival and a second-order channel into whether hospitals can keep using cheap first-line drugs.

• Under-five mortality rate: direction higher deaths when neonatal and paediatric sepsis meet resistant Gram-negatives. Horizon: days to years. Confidence: medium where blood-culture coverage exists; low where the only number is a global modelled total. The series is lower-is-better, so strengthening AMR is welfare-negative. No unique neonatal AMR integer is invented here.
• Life expectancy at birth: direction lower period LE if resistant bloodstream infection becomes a large adult and child mortality term. Horizon: years. Confidence: low for a unique global coefficient; the GRAM associated/attributable split shows why. Strengthening AMR weakens a higher-is-better series.
• Universal health coverage service coverage index: direction lower effective coverage when first-line drugs fail and Reserve drugs are absent. Horizon: years.

A PISA path does not follow. Missing coefficients stay missing.

Interactions and contested points

AMR interacts with access to essential medicines, infectious-disease transmission, health-system capacity, and international health cooperation. Reinforcement is obvious when a thin laboratory, a Watch-heavy prescription, and farm effluent hit the same catchment. Offsetting is the claim that a 2024 UNGA target or an Access-70% slogan cancels 2019 GRAM deaths. A target is not a reversal.

The strongest competing view is that SDG 3.d.2 World percentages are not a global incidence series because participating countries and testing density change year to year — especially the 2018 World point. That objection is serious. The competing defence of this node is only this: the indicator is the UN’s official human-health AMR SDG, the transformation is disclosed, and the 2018 break is named rather than smoothed away.

A second competing view is that O’Neill’s 10 million-by-2050 figure, or GRAM’s 39 million cumulative to 2050 updates, should be the meter. Those are modelled forecasts. They are not this driver’s time series.

Established: WHA68.7 (2015); UNGA 2016 and 2024 declarations; GLASS/SDG 3.d.2 World percentages 2016–2022 as published; GRAM 2019 associated 4.95 million vs attributable ~1.27 million as a named split. Not established: a unique Factrail death total, a smoothed 2018, or that A/RES/79/2 funded 60% of national plans. A 2029 UNGA review would be a new fact.