Warming carries 11% of Australia's Salmonella burden. By the 2050s it could double.

Enteric bacteria multiply faster in warmth. That is a laboratory fact old enough to be uninteresting, and it has a public health implication that has rarely been quantified at national scale: as countries warm, some fraction of their food poisoning becomes attributable to the climate rather than to handling.

A comparative risk assessment published in The Lancet Planetary Health on 23 December puts a number on that fraction for Australia, in what the authors describe as the first national assessment of temperature-attributable Salmonella and Campylobacter burden in the country [s1].

The estimate

Between 2003 and 2018, rising temperatures were attributed 11% of the Salmonella burden and 8% of the Campylobacter burden [s1].

Expressed as disability-adjusted life-years, the study reports 41.8 (SD 2.8) DALYs for Salmonella and 28.1 (SD 1.8) for Campylobacter [s1]. The abstract does not state the denominator — whether these are annual national totals, per-100,000 rates, or another unit — and the figures are too small to be national annual totals for a country of Australia's size. The percentages are the safer quantity to carry away; the DALY counts should not be interpreted without the full paper's units.

The highest burden fell in the tropical rainforest climate zone [s1].

The projection

Burden was projected to the 2030s and 2050s under two representative concentration pathways, RCP4.5 and RCP8.5, incorporating population growth and adaptation scenarios [s1].

By the 2050s, under RCP8.5 with medium population growth and no adaptation, Salmonella burden was projected to reach 100.6 (10.9) DALYs and Campylobacter 67.9 (7.4) [s1] — roughly a doubling on the 2003–2018 figures in the same units.

A 10% adaptation measure was projected to reduce these to 89.5 (8.3) and 61.8 (6.7) [s1]. That is a useful ratio to hold on to: a tenth of the exposure–response relationship removed buys back roughly a tenth of the projected increase, and no more.

How the estimate is built, and where it is soft

The method chain is worth stating, because each link introduces uncertainty that does not appear in the standard deviations quoted.

DALYs for both infections from 2003 to 2018 were obtained from the Australian Institute of Health and Welfare [s1]. A meta-regression model estimated the increase in infection risk per 1°C rise in temperature [s1]. Exposure distributions were calculated for each Köppen-Geiger climate zone and compared with a theoretical minimum risk exposure to derive the attributable burden [s1].

A comparative risk assessment of this design assumes that the temperature–infection relationship estimated from historical data holds under future conditions, that population distributions across climate zones evolve as modelled, and that food handling, refrigeration and reporting practices stay put unless explicitly varied in an adaptation scenario. None of those is guaranteed over seventy years. The adaptation scenarios are also stipulated rather than derived — a 10% reduction in the exposure–response coefficient is an assumption about what intervention could achieve, not a measured effect of any particular intervention.

The study was funded by the Australian Research Council Discovery Program [s1].

The question the second study asked

A common extension of the warming-and-bacteria argument holds that higher temperatures should also drive antimicrobial resistance, through faster bacterial growth, more horizontal gene transfer and more antibiotic use. It is plausible, it is frequently asserted, and a study published in Antibiotics on 16 December tested it directly in the same country and the same organism [s2].

Researchers analysed 1,012 cases of Salmonella bacteraemia in Queensland, using distributed lag non-linear models to test associations between deseasonalised temperature and resistance to ampicillin, ciprofloxacin, gentamicin and third-generation cephalosporins, adjusting for precipitation, seasonality and temporal trends [s2].

Resistance was common. Any-antibiotic resistance occurred in 25.5% of cases (95% CI 22.8–28.3), gentamicin resistance in 15.4% (13.2–17.8) and cephalosporin resistance in 15% (12.9–17.4), with variation across serotypes [s2].

Temperature explained none of it. After adjustment, no resistance outcome was significantly associated with temperature: gentamicin RR 1.23 per 1°C (95% CI 0.57–2.65, p=0.59), cephalosporins 1.19 (0.52–2.72, p=0.68), ciprofloxacin 1.88 (0.29–12.03, p=0.50), ampicillin 1.93 (0.28–13.17, p=0.50) [s2].

The authors also report, and then dismantle, a marginal finding: a temperature–precipitation interaction for cephalosporins identified using a generalised additive model at p=0.048, which did not survive multiple-testing correction, was not robust across model specifications (GLM p=0.058), and did not hold under cross-validation [s2]. Reporting a borderline result alongside the reasons not to believe it is the correct handling, and it is rarer than it should be.

The confidence intervals on those risk ratios are wide — ciprofloxacin's spans 0.29 to 12.03 [s2] — so this is an absence of evidence in a study with limited power, not strong evidence of absence. What it does establish is that the climate–resistance link cannot be assumed. The authors put it as the relationships not being universal, and depending on geographic, epidemiological and organism context [s2].

Why it matters

Read together, the two studies separate two claims that often travel as one. Warming appears to be increasing how much enteric infection Australia has [s1]. It does not appear, in this organism and this state, to be making that infection harder to treat [s2].

That distinction changes what adaptation would target. If the mechanism is incidence, the levers are cold chain, food safety regulation and outbreak surveillance in the warmest zones. If the mechanism were resistance, the levers would be stewardship. On current evidence, the first case is the one Australia has.

What to watch

Whether the full paper clarifies the DALY units, and whether the temperature–resistance null replicates in other organisms and jurisdictions with larger case numbers.

This article is informational and is not medical advice.

Sources

• [s1] Damtew YT, Varghese BM, Anikeeva O, et al. "Estimating non-optimal temperature-attributable burden of Salmonella and Campylobacter infections under various climate change, population, and adaptation scenarios in Australia: a comparative risk assessment modelling study." The Lancet Planetary Health, 9(12), published online 23 December 2025. https://doi.org/10.1016/j.lanplh.2025.101383
• [s2] Manchal N, Young MK, Castellanos ME, Adegboye OA. "No Evidence of Temperature-Driven Antimicrobial Resistance in Salmonella Bacteraemia in Queensland, Australia." Antibiotics, 14(12), published online 16 December 2025. https://doi.org/10.3390/antibiotics14121274