The World Health Organization has updated its estimate of what unsafe food costs the world, and the numbers are larger than the last time it counted. In a paper published in The Lancet Global Health (online June 3, 2026; August 2026 issue), WHO’s Foodborne Disease Burden Epidemiology Reference Group estimates that 42 foodborne hazards caused 866 million illnesses, 1.52 million deaths and 57.1 million disability-adjusted life-years (DALYs) in 2021. A DALY is one year of healthy life lost to illness, disability or early death.
I have spent more than thirty years representing people sickened by food in the United States. This paper is the wider picture those cases sit inside, and parts of it surprised me.
The numbers at a glance
Illnesses: 866 million (95% uncertainty interval 680 million to 1.09 billion). Against a 2021 world population of about 7.9 billion, that is roughly one person in nine.
Deaths: 1.52 million (783,000 to 2.51 million).
Healthy life lost: 57.1 million DALYs. WHO’s 2021 figures for tuberculosis, malaria and HIV are 60 million, 52 million and 40 million.
Lost productivity: US$310 billion, or US$647 billion adjusted for purchasing power. That is 0.52% of global GDP and 1.16% of GDP in low-income countries.
Most of the illnesses are infections. Most of the deaths are not.
Diarrheal pathogens account for 666 million of the 866 million illnesses. Chemical hazards account for 6.26 million illnesses, which is 0.7% of the total, and for 1.12 million deaths, which is 73%. Inorganic arsenic in food is tied to an estimated 641,000 deaths (42%) and lead to 466,000 (31%), largely through heart disease and stroke.
This is the first time WHO has counted cardiovascular disease from dietary metals, and the authors say it largely explains why the death estimate is so far above the 420,000 WHO reported for 2010. They also advise against comparing the two editions directly: the hazard list grew from 31 to 42, and the populations, models and inputs changed.
The authors attach their own cautions, and they belong in any honest summary. Most of the arsenic burden comes from exposure in India and China. The uncertainty interval for lead runs below zero at its lower end. Dietary metals are a small share of the world’s cardiovascular disease. And the authors write that the metals finding “must not divert efforts to address microbial hazards.”
The bugs I see in my practice
Set the metals aside and the diarrheal and invasive pathogens still account for about 404,000 deaths a year (296,000 and 108,000). The ones my clients know by name:
Non-typhoidal Salmonella: 23.5 million illnesses and 55,100 deaths. It is the leading microbial hazard by DALYs (3.58 million), as it was in the 2010 estimates.
Campylobacter: 148 million illnesses and 35,400 deaths.
Norovirus: 54.8 million illnesses and 14,100 deaths.
Shiga toxin-producing E. coli: 51.9 million illnesses and 20,500 deaths.
Hepatitis A: 20 million illnesses and 11,900 deaths.
Cyclospora cayetanensis: 10.2 million illnesses and 14,600 deaths. Cyclospora is new to the WHO estimates in this edition, and the uncertainty is wide; the interval for illnesses dips below zero.
Listeria monocytogenes: 22,800 illnesses and 5,280 deaths. That is nearly one death for every four illnesses, the pattern I have seen in every Listeria outbreak I have worked on.
Children under five carry the heaviest load
Children younger than five get sick from food at 2.7 times the rate of everyone else and lose healthy life at 4.3 times the rate. Their death rate from diarrheal hazards is 17.5 per 100,000, against 2.47 per 100,000 for people five and older, about seven times higher. The leading causes of burden in this age group are methylmercury, non-typhoidal Salmonella and rotavirus.
Where the burden falls
The highest rates are in two African subregions (1,261 and 1,150 DALYs per 100,000 people), driven by diarrheal disease, and in one South-East Asian subregion (1,264), driven by chemical hazards.
Wealthy countries are lower, not exempt. In the paper’s high-income Americas cluster (AMR A), the rate is 195 DALYs per 100,000, and 144 of that is chemical; arsenic alone is 116. Among infections in that cluster, Toxoplasma gondii (21.3) and non-typhoidal Salmonella (15.4) lead, with Listeria at 2.9 and Shiga toxin-producing E. coli at 1.8. High-income countries account for just under US$108 billion of the productivity loss.
What I take from it
There is good news. The total burden fell between 2000 and 2021, most visibly in Africa. The authors say the trend should be read cautiously and point to better food safety systems, water, sanitation and medical care as likely reasons.
The count is still an underestimate, by the authors’ own account. Hazards such as astrovirus were left out, three bacterial toxins counted in 2010 were dropped, and the dollar figure covers lost productivity only, not medical bills.
For the infections, the paper lists the tools: cooking, pasteurization, better irrigation practices, slaughter and meat inspection, and vaccination of animals and people. Nothing on that list is new. I have said for years that I would like to be put out of business. A world in which some 400,000 people a year die of foodborne infections is a long way from that.
The metals are a different problem. Arsenic and lead cannot be cooked or washed out; the authors say removing a chemical from food after contamination is unlikely, so prevention has to happen at the source, in soil, water and industrial activity, or through dietary advice about the foods that contribute most.
WHO has posted country-level estimates and trends in its 2026 edition dashboard and its Global Health Observatory. A count is not a fix. But governments set priorities by what they measure, and this is the best measurement we have had.
Source
Lake RJ, Devleesschauwer B, Majowicz SE, et al., on behalf of the WHO Foodborne Disease Burden Epidemiology Reference Group for 2021–25. WHO estimates of the global, regional, and national burden of 42 foodborne infectious and chemical hazards, 2000–21: an updated data synthesis. Lancet Glob Health 2026; 14: 103994. https://doi.org/10.1016/S2214-109X(26)00156-7. Open access (CC BY 3.0 IGO).
Recent Comments