Reviews/Novel Antibiotics/AMR causes over one million deaths annually
Amr Burden · NOVEL-AB-BURDEN-01 · v1
Borderline: within 4 points of the Established/Likely boundary, so a re-grade could land either side.
Two worldwide analyses counted the deaths each year in which a drug-resistant bacterial infection was the direct cause, and both landed above one million: about 1.27 million for 2019 [1] and 1.14 million for 2021 [9]. Regional studies in Africa [11], the Americas [10] and China [12] found death tolls on a matching scale, and a hospital study of bloodstream infections in Canada showed that resistant infections carry a higher risk of death than treatable ones [18]. The main reservation is that the two global counts come from the same research group using the same method, so they reinforce each other rather than acting as genuinely independent checks. The counts also depend on which comparison is used: the 'caused by' figure is the smaller one, and its margin of error reaches down toward the one-million mark. The panel treated the evidence as strong but held back from the very top because of those two limits.
Two global burden analyses estimate deaths directly attributable to bacterial resistance at 1·27 million in 2019 [1] and 1·14 million in 2021 [9], each alongside much larger associated totals of 4·95 million [1] and 4·71 million [9]. Regional analyses report burdens of the same order: 1·05 million associated deaths in the WHO African region [11], more than 600 thousand associated deaths in China [12], and 569,000 associated deaths in the Americas [10]. A 14,548-episode cohort of E. coli bloodstream infection in Ontario [18] shows higher adjusted odds of death for several resistant classes, supporting the causal direction at the patient level. Narrative and methodological pieces [13][15][17] restate figures or critique estimation rather than adding new counts. The contested point: The division is over independence and the counterfactual. Every global and regional count originates from one consortium using one five-component method [1][9][10][11][12], so they corroborate rather than independently replicate. The causal reading rests on the smaller attributable figures, whose uncertainty reaches the threshold — 0·911-1·71 million in 2019 [1] and a lower bound of 1·00 million in 2021 [9].
All 20 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Novel Antibiotics, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The global counts [1] and [9] and all regional counts [10][11][12] are GBD/IHME outputs sharing one five-component counterfactual methodology and overlapping authorship; this is methodological repetition, not independent replication, so the >1M figure rests on a single analytic framework.
The claim's verb 'causes' maps to the attributable counterfactual (1.27M [1], 1.14M [9]), not the far larger associated figures (4.95M, 4.71M) that use a no-infection counterfactual; conflating the two would overstate the causal count.
Where it remains uncertain. Every global and regional count comes from a single consortium and shared method [1][9][10][11][12], and the attributable uncertainty intervals reach down to about one million [1][9].
Aleatory
Posterior sampling variation in the modelled counts, e.g. the 2019 attributable interval 0·911-1·71 million [1] and the 2021 interval 1·00-1·28 million [9].
Epistemic
Counts for many low-income locations are modelled rather than observed, with microbiology and surveillance data acknowledged as scarce [1][11][13].
The load-bearing papers from a corpus of 20, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Central attributable estimate (1.27M) exceeds one million for 2019
“there were an estimated 4·95 million (3·62-6·57) deaths associated with bacterial AMR in 2019, including 1·27 million (95% UI 0·911-1·71) deaths attributable to bacterial AMR”
Panel auditInternal validity: medium · scope of inference: Global bacterial AMR deaths across 23 pathogens and 88 pathogen-drug combinations in 204 countries, 2019.Key limitations: UI lower bound below 1 million · modelled estimates for data-sparse locations · bacterial AMR only
Independent GBD round confirms >1M attributable deaths for 2021
“In 2021, we estimated 4·71 million (95% UI 4·23-5·19) deaths were associated with bacterial AMR, including 1·14 million (1·00-1·28) deaths attributable to bacterial AMR.”
Panel auditInternal validity: medium · scope of inference: Global bacterial AMR deaths 1990–2021 with forecasts to 2050, 204 countries.Key limitations: modelled estimates for data-sparse locations · bacterial AMR only · forecasts are scenario-based
Regional corroboration of attributable and associated AMR mortality
“We estimated 569,000 deaths (95% UI 406,000-771,000) associated with bacterial AMR and 141,000 deaths (99,900-196,000) attributable to bacterial AMR among the 35 countries in the WHO Region of the Americas in 2019.”
Panel auditInternal validity: medium · scope of inference: Bacterial AMR deaths in 35 countries of the WHO Region of the Americas, 2019.Key limitations: regional subset only · no global figure · modelled estimates
3 of 20 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
The corpus's definitive global burden estimates centrally exceed one million attributable deaths and are regionally corroborated, but single-consortium dependence and threshold-straddling uncertainty intervals cap confidence at the lower edge of Established.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Independent re-estimation of global burden. All global and regional counts share one consortium and method [1][9][10][11][12], so the >1M figure has not been reproduced independently. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A global AMR burden estimate built with a separate data pipeline and an alternative counterfactual, by a group outside the GBD/IHME network.
Open question: Does an estimation method independent of the GBD five-component model also place attributable AMR deaths above one million? Why it is still open: Every global and regional count in the corpus comes from one consortium and one method [1][9][10][11][12].
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19