About one in ten infection deaths worldwide may be attributable to obesity

Adult obesity has been linked to specific infections, but evidence across the full spectrum of infectious diseases has remained scarce — a gap the authors of a new study state plainly as their starting point [s1]. What has been missing is the denominator: whether the association holds across infectious disease generally, or only across the handful of conditions anyone has studied.

A study published in The Lancet on February 9 attempts the full sweep. It examined the incidence, hospitalisation, and mortality of 925 bacterial, viral, parasitic, and fungal infectious diseases against body-mass index, then modelled what share of infection deaths worldwide might be attributable to obesity [s1].

The design

The authors pooled two Finnish cohort studies and repeated the analysis in the UK Biobank as an independent population [s1]. BMI was measured at baseline — 1998 to 2002 in the Finnish studies, 2006 to 2010 in the UK Biobank — and participants were sorted into healthy weight (18.5–24.9 kg/m²), overweight (25.0–29.9), or obesity, subdivided into class I (30.0–34.9), class II (35.0–39.9), and class III (≥40.0) [s1].

Participants were then followed through national hospitalisation and mortality registries. Anyone with a recent history of infection-related hospitalisation at baseline was excluded [s1].

The Finnish cohorts contributed 67,766 adults (mean age 42.1 years, SD 10.8; 49,516 [73.1%] female, 18,250 [26.9%] male). The UK Biobank contributed 479,498 adults (mean age 57.0, SD 8.1; 261,084 [54.4%] female, 218,414 [45.6%] male) [s1]. Over follow-up there were 8,230 incident infection cases in the Finnish cohorts and 81,945 in the UK Biobank [s1].

The association

Compared with people of healthy weight, those with class III obesity had roughly three times the risk across every outcome tested [s1]:

• Infection-related hospital admission: hazard ratio 2.75 (95% CI, 2.24–3.37) in the Finnish cohorts, 3.07 (2.95–3.19) in the UK Biobank
• Infection-related death: 3.06 (1.25–7.49) Finnish, 3.54 (3.15–3.98) UK Biobank
• Either outcome: 2.69 (2.19–3.30) Finnish, 3.07 (2.95–3.19) UK Biobank

Across obesity classes I to III combined, the pooled hazard ratio for fatal or non-fatal severe infection was 1.7 (1.7–1.8) [s1].

The consistency is the part worth dwelling on. The authors report that the association held across three different measures of adiposity — BMI, waist circumference, and waist-to-height ratio — across demographic and clinical subgroups, and across a wide range of infections: non-fatal and fatal, acute and chronic, bacterial and viral including subtypes, and parasitic and fungal [s1]. A finding that survives that many cuts is less likely to be an artefact of one measurement choice.

The global number, and what it is

Applying those hazard ratios to obesity prevalence estimates from the Global Burden of Diseases, Injuries, and Risk Factors Study database, the authors estimated the population attributable fraction of infection-related deaths due to obesity at three time points: 8.6% (6.6–11.1) in 2018, before the pandemic; 15.0% (12.8–17.4) in 2021, during it; and 10.8% (8.6–13.6) in 2023, after it [s1].

The headline reading — approximately one in ten infection-related deaths worldwide might be attributable to obesity — is the authors' own interpretation [s1]. It should be read as a modelled projection rather than an observation. The hazard ratios come from two European populations; the prevalence data come from everywhere; the multiplication assumes the risk relationship transfers. Whether it does across regions with different pathogen mixes, different health systems, and different age structures is not established by this study.

The pandemic-period spike to 15.0% is consistent with what was observed clinically about obesity and COVID-19 severity, and it is also the point at which the modelling assumption is doing the most work.

The causal question

This is observational. Confounding by indication does not apply, but confounding generally does: people with class III obesity differ from people of healthy weight in ways registries capture imperfectly. The exclusion of participants with recent infection-related hospitalisation at baseline reduces reverse causation but does not eliminate it — chronic conditions that both cause weight gain and predispose to infection are not fully removed by that filter.

An accompanying comment in the same issue, by Sørensen and Osler, is titled "Mutually reinforced burdens of obesity and infections" [s2] — a framing that points at the bidirectionality this design cannot resolve.

Why it matters regardless

Infection risk has not been part of how obesity is usually discussed clinically. The framing has been cardiometabolic — diabetes, cardiovascular disease, fatty liver. If this association is real at the magnitude reported, it adds a category of harm that current risk communication largely omits, and it does so at a moment when highly effective weight-loss pharmacotherapy is becoming widely available. Whether reducing weight reduces infection risk is a separate question this study does not answer.

The work was funded by the Wellcome Trust, the Medical Research Council, and the Research Council of Finland [s1].

What to watch

Whether the risk estimates replicate in non-European cohorts, and whether any trial or target-trial emulation of weight-loss treatment reports infection outcomes.

This article describes observational research and is not medical advice.

Sources

• [s1] Nyberg ST, Frank P, Ahmadi-Abhari S, et al. Adult obesity and risk of severe infections: a multicohort study with global burden estimates. The Lancet, published online 2026-02-09.
• [s2] Sørensen TIA, Osler M. Mutually reinforced burdens of obesity and infections. The Lancet, published online 2026-02-09.