A study published in The BMJ looks at maternal consumption of ultra-processed foods and risk of overweight or obesity in offspring.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“Though this is a carefully conducted study and brings together data from a fairly large number of participants, it’s very important not to over-interpret its findings. The researchers’ conclusions are, rightly, rather carefully expressed. They mention that further study is needed to confirm the findings, and also to understand what might be the underlying causes and mechanisms behind the associations that they found between the mothers’ consumption of ultra-processed food during the child-rearing period and the risk of obesity in their children.
“This is an observational study, and like pretty well all observational studies, doubts remain about whether the higher levels of obesity in children, whose mothers ate larger amounts of ultra-processed foods, are actually caused in some way by the mothers’ diet. That’s because there are many other differences between the families where mothers ate different quantities of ultra-processed foods than the differences in the mothers’ diets. Some of these other differences could be, in whole or in part, the actual causes of the differences in levels of obesity in the children. The researchers recognise this issue, of course. They made statistical adjustments to allow for a number of factors that could have been involved in the patterns of cause and effect, and even after those adjustments, the pattern of association between the mothers’ consumption of ultra-processed foods and the level of obesity in their children still remained. But one can never be sure that all the relevant factors have been adjusted for, so that some doubt about what is causing the observed associations must remain, as the researchers acknowledge.
“So it’s certainly possible that there is some way in which the mother’s diet can be linked to levels of child obesity in a cause-and-effect way, so that changing the diet of families could lead to fewer obese children. But that’s not the only possibility. Hence the call for more study of what’s going on. That said, this is far from being the first study that has found associations between the amount of ultra-processed food people eat and various aspects of ill health. We can’t say that all those associations are ones of cause and effect, but it’s certainly possible that some of them are.
“However, if the aim is to reduce the level of child obesity, or indeed other health issues, by improving diet, it would be important to know a lot more about how eating a lot of ultra-processed food might actually lead to obesity, so that appropriate interventions could be planned. But those mechanisms and links between ultra-processed food consumption and child obesity aren’t well understood, according to the researchers, and they could involve both biological and environmental factors.
“One problem in understanding those links further might be that the category of foods that are described as ultra-processed is rather wide, and includes a diverse range of foods. In this new research, the level of association between the mothers’ consumption of different types of ultra-processed food and their children’s chance of being obese did vary between different types of food. (See Figure 2 in the research paper.) For example, high consumption of ultra-processed breads and breakfast foods, by the mothers during the child-rearing period had a stronger statistical association with the risk of obesity or overweight in their children than did, for example, consumption of frozen and ready-to-eat meals, though there remains some statistical uncertainly about these comparisons. So maybe just advising people to lay off the ultra-processed foods, without going into more detail, might not be the right approach.
“Another related point is that the definition of an ultra-processed food has been criticised as being not clear enough. It is not a question of all food processing being bad. Many foods that have been processed in some way would not be defined as ultra-processed anyway. But further than that, a study* published earlier this year involved asking food and nutrition specialists in France to assign foods to the different NOVA groups, NOVA being the set of classifications of foods that is usually used to define ultra-processed food. The researchers found that, though assignments of foods to the different groups was more consistent between the specialists on some foods rather than others, the overall level of consistency between the experts was described as “low”, even when they were given detailed information on the ingredients. The researchers concluded that “These results suggest current NOVA criteria do not allow for robust and functional food assignments.” So how to provide good advice to consumers, when even experts don’t always agree on how to categorise particular foods, may not be easy.”
* Braesco et al. (March 2022) “Ultra-processed foods: how functional is the NOVA system?”, European Journal of Clinical Nutrition, https://www.nature.com/articles/s41430-022-01099-1
Sarah Coe, Nutrition Scientist, British Nutrition Foundation, said:
“This observational study looked at a large number of mother and child pairs from well-established, ongoing studies and found interesting results that add to the growing number of studies looking at the effects of ultra-processed foods on health. However, it’s important to remember that it can’t be said from an observational study alone that eating more ultra-processed foods during pregnancy might cause an increased risk of overweight or obesity in children.
“The results also don’t tell us why eating more ultra-processed foods during pregnancy might cause an increased risk of overweight or obesity in children, so more research is needed to explore the mechanisms behind this, which the authors do acknowledge. There are normally many risk factors involved in why overweight and obesity develops. Because of the demographics of the study population (predominantly white and from similar socio-economic backgrounds in the United States), it would be of interest to look at this potential relationship in more diverse ethnic and socio-economic groups.
“Many foods that would be classified as ultra-processed are less healthy options, such as crisps, cakes, sweets, chocolate and sugary drinks, which health professionals already recommend that many of us need to cut down on. However, ultra-processed foods can also include foods like sliced wholemeal bread and vegetable-based pasta sauces, which can be a useful part of a healthy, balanced diet. The researchers in this study did assess the risk of childhood overweight and obesity by different subtypes of ultra-processed foods, which would be of interest to explore further in future research.
“We know that the quality of a mother’s diet can affect her baby’s health in both in early and later life. Evidence from studies in both humans and animal models shows that if a baby is exposed to too little or too much nutrition during fetal life, they are more likely to develop conditions such as type 2 diabetes, heart disease and osteoporosis later in life. However, we don’t yet know whether altering the mother’s nutrient intake during pregnancy can influence her baby’s risk of these conditions as this is difficult to study.
“Pregnancy is a good time to target health behaviour changes like having a healthier diet. We are still understanding any links between eating ultra-processed foods and health, so more research is needed before any dietary recommendations can be made. Pregnant women in vulnerable groups, such as ethnic minority groups, women living with obesity and women from lower socio-economic groups, may also have more barriers to reducing their intake of foods that would be classified as ultra-processed, so these may need addressing while supporting them to make healthier food choices.”
Hannah Whittaker, Paediatric & Maternal Health Dietitian, Bump2baby Nutrition LTD, and BDA (British Dietetic Association) media spokesperson, said:
“From the study, it is apparent that there may be some limitations, including self-reporting of anthropometry and food intake. We are aware that there can be variance in taking anthropometric measurements due to human error and also that food diaries are fraught with underreporting. This may have affected the outcomes, however, the authors do comment on this.
“As this study was conducted in the USA, results may not be replicated if the same research was completed within the UK due to differences in eating habits. The outcome may also not be transferable to certain ethnic groups again due to multiple factors, including different BMI parameters. BMI can also be a poor measurement for younger children, therefore of those children who were classed as obese at a younger age it may have been beneficial to use growth charts to plot weight and height to ensure that comparable.
“We are aware that the eating environment that a child is exposed to at home can affect their own health, for example, if parents eat a diet high in ultra-processed food and are overweight then the child is more likely to do so and also become overweight. We are also aware that a mothers diet within pregnancy can also have a positive or negative effect on birth outcomes.
“This study does give rise to an area for further research on this topic and could influence policy to ensure that certain population groups are targeted with education. This study has produced some interesting findings with regards to maternal and peri pregnancy dietary intake and risk of being overweight or obese in childhood. The authors have considered multiple variables that may have affected data outcomes and found that comparing diets that were highest in ultra-processed foods with those that were lowest that there was a 26% increase in risk of being overweight or obese as a child.
“Dietitians play an integral role in educating women in pregnancy to adopt healthy eating principles for better outcomes for both mother and baby. This evidence shows to strengthen the importance of this education.”
Dr Hilda Mulrooney, Associate Professor of Nutrition, Kingston University, said:
“In my view the press release accurately reflects the science and it also includes the limitations which is important in this sort of study. The study is based on a large group of individuals and the inclusion and exclusion criteria are clear. Methods used are well described and validated data gathering tools are used. In addition, there is recognition of and allowance made for potential confounding factors. It is not clear that the sample was representative of the American population (it is suggested that it was not), and of course, it is unlikely to be representative of the UK population. However, it is an observational study and it is not possible to demonstrate causality from this type of study something which is acknowledged by the authors. It is useful to highlight potential relationships, which can point towards more detailed work that should be done. There are weaknesses in the study, although in many respects it is a strong study. There is a lack of clarity in relation to ethnicity; the text suggests mothers were defined as white or others, which is very non-specific and is important since different ethnicities have an important role in relation to the risk of obesity, and indeed the specific cut-off points used to define obesity. There is a lot of disagreement about the NOVA classification system for ultra-processed foods and even whether UPF themselves have an impact other than that they are often high in calories, fat, salt and/or sugar. The NOVA system is not agreed on in the literature, and suggests that the level of processing rather than the nutritional quality of the food, is more important for health. this is not universally agreed in nutritional science, which relates the risk of ill-health to intakes of nutrients. They also use self-reported weights and heights which given that they are interested in obesity as a primary outcome is a weakness; again it is acknowledged by the authors but nonetheless, it is not ideal practice. They suggest that financial approaches to ensure that healthy food can be afforded are part of the solution; I would not disagree with that as a principle, but could not see a relationship between household income and obesity or UPF intake (apologies if I missed that). So the basis for this, from this study, was unclear although the general principle is not one I would disagree with.
“We have known for a long time that maternal diet is an important risk factor for children and that healthy eating in pregnancy and during childhood is not just important for adequate nutrition for children but for role modelling of the behaviours you want to embed. If you want children to eat healthy foods, you eat healthy foods yourself – you model it so it is their norm. That also shapes their environment, and they make choices based on that environment. We know that during pregnancy a variety of risk exposures (e.g. diet, smoking) can have longer cross-generational effects on the foetus potentially through epigenetics. What was noteworthy here in my opinion was that maternal consumption of UPF during pregnancy was not influential in terms of obesity risk of the child only during child rearing but it was then independent of what the child was actually eating. So how to square that circle is very unclear, and in fact the authors didn’t really have much to offer in terms of potential mechanisms there. they mentioned various possible effects of different additives but much of that is cell or animal rather than human-based, and it is not clear that those specific components were in the foods consumed by participants in this study.
“The implications remain that maternal nutrition is important for mother and child health and wellbeing. That hasn’t changed. It is surprising that diet in pregnancy appeared not to be a factor whereas diet during childrearing was but independently of child eating behaviours. That in itself leaves me wondering about the extent to which there is truly an effect here apart from statistical manipulation. Nonetheless. the importance of healthy eating during maternity is extremely important, but not only in relation to food processing but to nutrient intake too.”
Dr Duane Mellor, Registered Dietitian and Senior Teaching Fellow, Aston Medical School, Aston University, said:
“This is an interesting paper linking data from nearly 20000 mothers and their children from within a large US study. It attempted to assess dietary intake using a food frequency questionnaire, asking questions about what participants ate over the past year, the data used in this study covered what the authors described in the abstract as the child rearing period, but details suggest this was only a year which included part of the pregnancy, which in the methods is called the peri-pregnancy period. So, it covered a variable amount of time before their child was conceived and some of the pregnancy. It also should be noted that this methodology of assessing food intake may not be that precise, as looking at the average intakes of fat, carbohydrates and protein between the quintile of women who ate the most ultra processed foods and those who ate the least suggested very little difference, however the reported energy intake was nearly 1000kcal more in the group consuming the most ultra-processed foods. This suggests either they were getting the extra energy from another source e.g. alcohol which was not reported, or more likely there was an error in estimating either nutrient or calorie intake. This along with what although developed by consensus, the potentially subjective description of ultra-processed foods can introduce bias and errors. As something made in the home from ingredients may not be considered to be ultra-processed, but when the identical product is bought in a shop it is not, but after all a cake is still a cake, shop bought or home baked!
“This study is interesting as it compares mother and child data to see if mother’s eating habits are associated with the child’s body weight and risk of being overweight or obese. The one year of food intake data around pregnancy was then compared with the weight change in the child over an average of 4 years. Although the analysis suggests an associated risk, it is important that this data as it is observational does not suggest evidence of causality. It is also important to note, that maternal food intake at other times in the child’s life was not included in the analysis and at no point was the food intake of the other parent considered (only their income).
“Therefore, although this does not change the important public health messages for families to all eat a healthy and varied diet, based on vegetables, fruits, pulses, seeds and wholegrain with moderate amounts of meat (or alternatives) and dairy. It does not show any more than an association between intake of maternal intake ultra processed foods around pregnancy and child weight gain and more detail of the shared family diet is needed to see if it is an effect of diet during the time around pregnancy or the family diet in general which is most important with respect to healthy weight gain and growth.”
Prof Gunter Kuhnle, Professor of Nutrition and Food Science, University of Reading, said:
“The term “ultra-processed food” is based on the NOVA definition that is incredibly vague and rejected by many nutritionists and food scientists for exactly that reason. It is based not only on ingredients and processing methods, but also the intention of the producer – the self-same food is only “ultra-processed” if it is used to generate profit, otherwise it is only processed. The problem with such a definition is that it attributes health effects not to the physical composition of foods but – like in homoeopathy – involves also meta-physical aspects.
“Research into ultra-processed foods almost always show that consumption is associated with adverse health outcomes – and this is not surprising. Firstly, ultra-processed foods most commonly consumed are often “hyper palatable” – and it has been shown in controlled intervention studies that this can result in overconsumption. However, hyper palatability is not restricted to ultra-processed foods. Secondly, ultra-processed foods are often cheaper – in particular as they have a longer shelf-life and can make better use of resources – and therefore are often consumed by a population group with overall lower health.
“The results of the study by Wang and colleagues is therefore hardly surprising. They found in observational studies that children of mothers who consume foods classed as ultra-processed are more likely to be overweight or obese. The problem with the study is one that plagues most research into ultra-processed foods: the dietary data used are based on food-frequency questionnaires that were never designed to capture the consumption of ultra-processed foods and were never validated for that use. Moreover, food production methods have changed during the duration of the study. The ambiguity of the method has been recognised by the authors who used a voting system to decide which food is ultra-processed – and they decided to class for example “bacon” as ultra-processed even though not all types of bacon would fall into this category.
“It is interesting to look at the study population: those with the highest intake of ultra-processed food have the overall lowest diet quality, most likely to smoke, have a lower income and lower educational attainment – it is therefore very likely that socio-economic factors can explain at least some of the results.
“The results show that “ultra-processed breads and breakfast goods” are the main contributor to the observed effect. From a nutritionist’s view, this finding is highly problematic as it combines two very different types of food. Highly processed and oft very sugary breakfast cereals are likely to increase the risk of weight gain, and this is something that is well known. However, the definition used classes virtually all bread bought in supermarkets as “ultra-processed”, and for many people it would be impossible to avoid – but this is also not necessary as there is no reason to suspect this bread to have an adverse effect on health.
“In my opinion, such studies detract from the real problem by focusing on a fashionable but wrong definition. Many of the foods classed as “ultra-processed” can be part of a healthy diet and there is no need to avoid them. As they often have a long shelf-life and make much better use of resources – for example fish fingers – they are more affordable and can reduce food waste. Instead of demonising foods and making people feel guilty for not being able to afford more expensive foods, it would be better to understand the physiological reasons and find ways to mitigate those. Ultimately, this would also require to address existing health disparities.”
‘Maternal consumption of ultra-processed foods and subsequent risk of offspring overweight or obesity: results from three prospective cohort studies’ by Yiqing Wang et al. was published in The BMJ at 23:30 UK time Wednesday 5 October 2022.
DOI: 10.1136/ bmj-2022-071767
Prof Gunter Kuhnle: “None to declare.”
Dr Duane Mellor: “No conflicts of interest.”
Sarah Coe: “Funding to support the British Nutrition Foundation’s charitable aims and objectives comes from a range of sources including membership, donations and project grants from food producers and manufacturers, retailers and food service companies, contracts with government departments; conferences, publications and training; overseas projects; funding from grant providing bodies, trusts and other charities. Further information about the British Nutrition Foundation’s activities and funding can be found at http://www.nutrition.org.uk/aboutbnf/”
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.”
For all other experts, no reply to our request for DOIs was received.