select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A study published in the Lancet Regional Health – Europe looks at food processing and the risk of type 2 diabetes.
Prof Janet Cade, Professor of Nutritional Epidemiology and Public Health, University of Leeds, said:
“This is another analysis looking at UPF but based on data which was not collected, as the authors acknowledge, to assess level of processing. Whilst they have done their best to characterise the limited food frequency and other data which they had available, the data differed by country and was collected before the NOVA system had been created. So inevitably using the NOVA classification required a lot of assumption.
“However, the results do show that not all food groups are equal; so focussing on overall results could be misleading. Savoury snacks and animal-based products were foods associated with a higher risk. Breads, biscuits and breakfast cereals and plant-based alternatives were associated with lower risk. What this analysis doesn’t do is tell us what types of bread or cereal may be associated with a lower risk. It also cannot unpick which aspects of savoury snacks and animal products is driving the association with higher risk.”
Dr Nerys Astbury, Associate Professor – Diet & Obesity, Nuffield Department of Primary Health Care Sciences, University of Oxford, said:
“This new study exploring the association between ultra-processed foods (UPFs) and subsequent incidence of type 2 diabetes in a large prospective study for several different counties, re-affirms previous studies which have reported associations between the risk of developing a range of health conditions and the proportion of ultra-processed foods consumed in the diet.
“It is important to note that the groups with the lowest consumption of UPFs were younger, employed at the time diet was assessed and less likely to have disease (which are also linked with lower risk of type 2 diabetes). They also reported lower sugar, fat and salt(sodium) intakes – and their diets were lower in energy density than the diets reported by the people consuming the highest proportion of UPFs. It is well established that diets high in these nutrients are associated with higher risk of developing non-communicable diseases such as obesity and hypertension.
“Whilst these new findings add to the body of evidence showing that many health problems are linked with the greater consumption of UPFs, there is no evidence to show that these health conditions are caused by them.
“Furthermore, there is no evidence that the effects seen are independently associated with the degree of processing or additive content per se.”
Dr Gavin Stewart, Reader in Interdisciplinary Evidence Synthesis, Newcastle University, said:
“People who consume ultra processed food probably are at increased risk of diabetes as this observational study finds, but it is not possible to disentangle the effects of poor diet and degree of food processing without robust randomised studies. The problem is that ultra processed foods are not consumed in isolation. They are usually part of a high fat, high sugar diet and may also be associated with life-style factors that have deleterious health impacts. The authors attempt to disentangle these factors where they can with sensitivity analyses testing modelling assumptions and make appropriately nuanced conclusions but without better (randomised) evidence we don’t understand the causal links. The finding that different kinds of ultra processed foods may have different effects is interesting, but some of these differences may be false positives or reflect spurious association rather than providing mechanistic insights. These findings should inform future research, but additional data is needed to draw recommendations for policy or practice.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:
“This is at least the second major research paper published this month that looks beyond the simple division of foods into ultra-processed foods (UPFs) and other categories. These papers (and a few others before them, but only very few) divide UPFs into a number of subcategories.
“This new paper found associations of higher consumption of some types of UPF with the risk of type 2 diabetes, particularly for savoury snacks, processed meats and other ultra-processed animal products, ready meals, and sugar-sweetened and artificially-sweetened drinks. For some other types of UPF – bread, biscuits and breakfast cereals, sweets and desserts, and plant-based alternatives – the association went the other way, with higher consumption being associated with lower type 2 diabetes risk. For yet other types of UPF, there was no evidence of association either way. And it’s important to remember, as the research paper and the press release make clear, that an observational study of this kind can’t on its own give evidence that any of these associations demonstrate cause and effect.
“The other September paper1 looked at associations between UPF consumption and cardiovascular disease (CVD, that is, heart attacks, strokes, etc.) Among its conclusions were that “The diverse nutritional composition within these products warrants the need to deconstruct the ultra-processed food classification for a nuanced understanding of their impact on cardiovascular health.” That is, those researchers had found that associations between consumption of different types of UPF and CVD weren’t all the same by any means, just as the researchers of the new diabetes paper found for associations between UPF consumption and type 2 diabetes risk.
“It’s not straightforward to compare the results of this CVD study with the results on diabetes in the new paper, partly because their divisions of UPFs into different types are not the same. But, like the new paper on type 2 diabetes, the CVD paper found evidence of associations between CVD risk and high consumption of processed meat products, and of sugar-sweetened drinks. Unlike the direction of association in the new diabetes study, the CVD study found some evidence that higher consumption of savoury snacks was associated with slightly lower risk of CVD. The diabetes paper also reports an association between higher consumption of bread and cold cereals, and of yoghurt and dairy desserts, and lower CVD risk.
“Again unlike the new diabetes paper, the CVD researchers found no clear association either way between consumption of ready meals and CVD risk – though, as with all these categorisations, what the two studies actually meant by ready meals might have differed. (For instance, the new diabetes paper had a category for plant-based alternatives, but the CVD paper did not – perhaps the CVD paper counted these in with ready meals.) As with the new paper, the CVD researchers cautioned against concluding that these associations were ones of cause and effect.
“Some of the differences between these two studies may be because the types of UPF are defined differently. Some differences may well be because the statistical associations, that were found, do not relate to what people were eating in a cause-and-effect way anyway. But the two papers, and other similar findings, do convince me that simply saying that the entire group of UPFs can have bad health consequences is just too simple. All UPFs are not equal.
“There is an obvious question as to why the differences between different types of UPF arise. The new paper does address this to some extent, pointing out that there may be explanations in terms of the high content in some UPFs of ingredients (such as saturated fat, sugar and salt), that have been associated with poor health going back to before the categorisation based on processing was invented. It also mentions there are other possible explanations that relate more directly to the processing or to the physical characteristics of many UPFs.
“The research paper does address the question of content of fats, sugar and salt to some extent, as well as looking at an overall measure of diet quality (adherence to a Mediterranean diet). In statistical analyses that make adjustments for these things, the researchers found that an association between UPF consumption and diabetes risk did not disappear, so that there seems to be something going on beyond these measures of food content and quality. This does not, however, mean that the processing of UPFs is definitely part of the problem, rather than only what the foods contain. It remains possible that some other aspect of their ingredients than their content of fats, sugar and salt is what’s behind the observed association – or indeed that the associations aren’t causal at all.
“This new study uses more sophisticated statistical analyses than studies of this type usually do. I think the statistical analyses are generally of high quality, and the extra information from the additional statistical analyses is helpful.
“However, the study does suffer, inevitably, from issues that arise in all studies of its type, and that cause difficulties in interpretation.
“The study is observational, which is the overall reason why we can’t conclude what is causing what. The differences in diabetes risk, associated with consumption of different food types, might be caused by the food types or might be caused by other differences between the people that happen to have different diets. As is standard in observational studies, the researchers made statistical adjustments to allow for other differences on which they had data, but one can never be even approximately sure that all relevant factors have been adjusted for, or even that one has data on everything relevant.
“Most of the long-term cohort studies of associations between UPFs and chronic diseases, like this one, rely on measures of food intake that use data collected before the notion of UPFs had been developed and published. That’s simply because chronic diseases generally take a long time to develop, so the participants have to be followed up for a long time, in order to get enough diagnoses of the health issues to be able to analyse the data effectively. As a result, the categorisation of people’s diets on the basis of how much the food is processed may not be accurate. Also, many such studies, including this one, measure people’s diets only when they first enter the study, which may be many years or even decades before. People’s diets may change over time, and also the composition of processed food that the eat may well have changed too, but the study cannot take this int account.
“Finally, though it’s good that this study (like some others, but certainly not all) did take into account some aspects of the composition of the food (saturated fat, sugar, salt), and of an overall diet quality measure, that doesn’t deal with other characteristics of the food, of its marketing, or of many other aspects. So, though it is a good study in many respects, and does throw a little more light on what might be behind the association between UPF consumption and poor health, the new study leaves a great number of questions completely unanswered.”
1 Mendoza et al., https://www.thelancet.com/journals/lanam/article/PIIS2667-193X(24)00186-8/fulltext
‘Food consumption by degree of food processing and risk of type 2 diabetes mellitus: a prospective cohort analysis of the European Prospective Investigation into Cancer and Nutrition (EPIC)’ by Samuel J. Dicken et al. was published in The Lancet Regional Health – Europe at 00:01 UK time on Monday 16 September 2024.
DOI: 10.1016/j.lanepe.2024.101043
Declared interests
Prof Janet Cade: “I am an employee of the University of Leeds.
I am Director and shareholder of Dietary Assessment Ltd., a spinout company from the University of Leeds, which has produced a software tool for dietary assessment called myfood24.
I chair the Advisory Committee of the British Nutrition Foundation: https://www.nutrition.org.uk/our-work/who-we-are/committees/advisory-committee
I receive no financial remuneration for doing that role. (BNF weblink: https://www.nutrition.org.uk/our-work/who-we-work-with/).
I lead a WHO Collaborating Centre in Nutritional Epidemiology.”
Dr Nerys Astbury: “No conflicts.”
Dr Gavin Stewart: “No conflict of interest to declare.”
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. My quote above is in my capacity as an independent professional statistician.”