September 22, 2026

expert reaction to UPF consumption linked to higher risk of multiple chronic diseases

A study published in Family Medicine and Community Health investigates the link between UPF consumption and risk of multiple chronic disease. 

Dr Adam Jacobs, Executive Director and Strategic Consultant, Biostatistics, Ergomed, said:

“The paper by Liu et al is a meta-analysis of 51 studies of the association of ultra-processed food (UPF) consumption with 8 different health outcomes, including over 8 million participants. As such, it is an impressively large study. 

However, it is important to note that while a large study given give good statistical precision, what it cannot do is compensate for biases or confounding in the individual studies included in the analysis. All it does is give you a larger sample of biased or confounded data, and if the individual studies on which the meta-analysis is based give misleading estimates of the effects of UPF because of bias or confounding, then the results of the meta-analysis will be just as misleading.

The meta-analysis is based on prospective cohort studies. While these are the strongest types of non-randomised study, they are still non-randomised, and you are left with the perennial question of whether any differences in UPF consumption groups are due to direct effects of UPF consumption, or due to people with different characteristics choosing to consume different amounts of UPF in the first place.

Here, there is one particular confounder that I think is particularly worrisome, namely socioeconomic status. People who consume low amounts of UPF are generally more affluent than those who consume more UPF, and we know that poverty is one of the strongest predictors of poor health. So it is possible that UPF consumption is simply a marker for poverty, rather than being harmful per se.

It is therefore important to look at whether the individual studies adjusted for socioeconomic status. There was some attempt to do this, but it seems to me unlikely to be adequate. While most (but not all) of the studies adjusted for education level, only 19 of the 51 studies adjusted for any kind of economic measure, and often this was a relatively crude measure such as an area-level deprivation score, which is unlikely to capture all the nuances of socioeconomic status with sufficient precision. I am therefore not convinced that confounding by socioeconomic status can be ruled out, and Liu at al did not investigate this in any way in their analysis.

It is worth noting that Liu et al assessed the quality of their evidence using the GRADE assessment tool, and assessed only one of their outcomes (obesity and overweight) as having moderate certainty: the other 7 were all rated as either low or very low certainty.

There are other possible confounding variables to consider, to do with lifestyle. For example, people doing shift work or with particularly stressful lives may eat more UPF. Few studies adjusted for such factors.

A further problem with meta-analyses is that they may be subject to publication bias. Studies with positive and newsworthy results are more likely to be published than studies that find no effect. So while Liu et al found 51 studies that mostly found UPF consumption to be associated with poor outcomes, it is possible that many other studies had been done that found no such association, but were not published, or possibly published in some form (e.g. a PhD thesis) that was less likely to be found in the literature databases that Liu et al searched. Techniques do exist to look for a high risk of publication bias, but Liu et al did not use them. They state that this was not possible because fewer than 10 studies were included for each outcome, but with 51 studies overall I think it should have been possible to do some examination of the risk of publication bias.

Overall, because of the risk of residual confounding and publication bias, I do not find Liu et al’s conclusions compelling.”

Prof Gunter Kuhnle, Professor of Nutrition and Food Science, University of Reading, said:

“This is a meta-analysis of studies that have been conducted previously. While these kinds of studies are meant to provide the highest level of evidence by pooling research, the results can only be as good as the underlying research – and this is the problem with this study.

Measuring food intake is notoriously difficult, as people are often not very good at remembering what they ate. Estimating UPF intake from these data is even more difficult for several reasons, e.g.

a) the definition of UPF is ambiguous, especially for more complex foods. 

b) most dietary assessment methods do not collect sufficient data to categorise foods reliably; for example, artisan bread is not UPF, while supermarket bought bread generally is – but many assessment methods simply record ‘bread’

c) the definition of UPF includes the ‘purpose’ of food production, i.e. food with the self-same composition and manufacturing method is only UPF when it is sold for profit – otherwise, it is just processed. This distinction is not normally recorded in research.

This means that observational studies that investigate associations between UPF intake and health rely on very unreliable data. Moreover, some foods categorised as UPF are known to have an adverse effect on health, e.g. sugar-sweetened beverages or processed meat – it is therefore not surprising that a dietary pattern that includes such foods will show adverse effects with health. 

Randomised controlled trials provide a much more nuanced picture – some trials show that adherence to dietary recommendations is more important than whether food is processed or ultra processed ^[1]. It is therefore important to encourage the general public to follow current dietary recommendation, but there is little evidence that avoiding UPF is beneficial to health”

[1] https://www.nature.com/articles/s41591-025-03842-0

‘Ultra-processed food consumption and risk of multiple chronic diseases among 8,819,894 adults from 51 prospective cohorts: a dose-response meta-analysis’ by Xiao Liu et al. was published in Family Medicine and Community Health at 23:30 UK time on Tuesday 22^ndSeptember. 

DOI: 10.1136/fmch-2026-003925

Declared interests

Dr Adam Jacobs: “No conflicts of interest to declare.”

Prof Gunter Kuhnle: “I am a former member of the UK Committee on Toxicity of Chemicals in Food, Consumer Products, and the Environment; a current member of the Advisory Committee on Novel Foods and Processes; the Director of the Chemical Analysis Facility at the University of Reading, which provides analytical services to academic and commercial clients; have received research funding (2010–20) from Mars for work on flavanols; and have received consultancy payments from RSM UK and EQT, paid to the University of Reading.  As a member of the EFSA ANS panel (2018-2019) and the UK’s COT (2019-2025) I was involved in the evaluation of food additives.”