This week a study published in the BMJ (British Medical Journal) has been fuelling some very scary headlines. You might have seen some:
As is usually the case, the newspapers have sensationalised the results!
I’m going to explain what the study really says and what it means for you.
Firstly, what are “ultra-processed foods”?
Processed foods often get a bad name and we associate all processed foods with high sugar/salt/fat. But that is not really the case. Food processing includes almost everything we do to our food, whether that is at home or in a factory on a large scale. It
can simply be cooking the food and therefore killing bacteria and making it safe to consume. It can be drying or freezing the food so that it lasts longer. It can be adding ingredients or heat to change to product (think of cheese!). Processed food is not bad.
Of course, some foods are processed more than others. So a scale was developed to categorise them.
Group 1: Unprocessed or minimally processed foods. “Minimally processed foods are natural foods altered by processes such as removal of inedible or unwanted parts, drying, crushing, grinding, fractioning, filtering, roasting, boiling, pasteurisation, refrigeration, freezing, placing in containers, vacuum packaging, or non-alcoholic fermentation. None of these processes adds substances such as salt, sugar, oils or fats to the original food.” This group includes mostly “fresh” products, like milk, veggies, fruit, grains, meat and fish, herbs and spices.
Group 2: Processed culinary ingredients. These are mostly products we use to add to our cooking, like salt, sugar, butter and oil. According to their classification, they should only contain around 2 ingredients and are using with foods in group 1.
Group 3: Processed foods. “These are relatively simple products made by adding sugar, oil, salt or other group 2 substances to group 1 foods. Most processed foods have two or three ingredients.” This group includes canned fruit, veggies and fish, cheeses, bread cured and smoked meats.
Group 4: Ultra-processed food and drink products. These are deemed to be industrial formulations, usually with 5 or more ingredients, which go through several industrial processes. These include, carbonated drinks, confectionary, margarine, powdered food supplements, pre-prepared foods/meals and mass produced bread and cakes.
Disclaimer: I hate this scale. It feels like it is separating so-called “good” and “bad” foods. Please always remember that food is just food, there is a place in the diet for ALL foods, whether processed or not! Also, the scale is quite vague and therefore can easily lead to misclassification of foods.
What did the study do?
The research was carried out on a group of French individuals (104,980 of ‘em), with an average age of 42 years. The data on food consumption in this study was self-reported, i.e. the participants filled in what they ate and sent it in to the researchers, and to be included participants had to fill in what they ate on at least 2 days over 2 years (but most participants did it 5 times).
The first problems with the study arise here:
100,000 people is a lot, but there are 7 billion people on earth. This group of French people is unlikely to be representative of the entire world population, and therefore the results CANNOT be generalised to all populations.
The participants volunteered to take part. That means the group is likely to be biased- when studies use people who volunteer the group tends to be mostly female (here it is 78% female), middle-class (this study doesn’t note many socioeconomic) and generally healthy to start with (again this study doesn’t really say). This means that the results, again, WILL NOT apply to the whole population.
When people self-report information they lie! It is really difficult first of all to remember everything you ate over a 24 hour period, so you may miss things out. But often people don’t write down foods that are “unhealthy”, or they say they had less than they actually ate. Equally, people sometimes over-report “healthy” foods. This means we are not analysing true food intakes= biased results.
Looking at what people eat over x2 24 hour periods does not give a good overview of their daily food consumption. Most participants provided 5 days of data, but again that is unlikely enough to understand their habitual food consumption.
This study used food intake data to essentially multiply exposure to ultra-processed foods over the lifetime of participants. I.e. if you reported eating bacon for breakfast on 3 of the days you provided data for, the research team will assume you eat bacon around 3 days a week and multiply that by your adults years. But that is probably not accurate right? (If they do that it looks like there is a greater association than their truly is!)
What the study found?
The study in question focused only on the consumption of ultra-processed foods according to the scale above. They found that sugary products (26%), drinks (20%), starchy foods and breakfast cereals (16%) and ultra-processed fruit and veggies (15%) contributed most to ultra-processed food consumption in their population. They state that contribution of these foods to the diet is similar for men and women, but those with the highest consumption tend to be younger, smokers, less educated, less family history of cancer, have a lower physical activity level, higher intakes of energy (kcals), fats, carbohydrates and salt and lower alcohol consumption.
When they followed up with their participants there were 2,228 cases of cancer- 739 breast cancer, 281 prostate cancer and 153 colorectal cancer. Of these, 108 were found to have died.
According to their results, ultra-processed food intake was associated with increased risks of overall cancer and post-menopausal breast cancer, but not premenopausal breast cancer, prostate cancer or colorectal cancer.
Issues with the data:
The researchers do not state what they counted as sugary products or starchy foods or ultra-processed fruit and veggies or anything else. As mentioned above, the scale used to classify processed foods is vague so who is to say that the researchers used it appropriately?
They state that those who consume ultra-processed foods “tend to be…” But as we have discussed their sample is not representative of all sections of the population. For example, the average age of their sample is 42, and when they say those who consume ultra-processed foods tend to be younger, they mean 36. Not really what we consider “young adults” is it? (not offence intended!). So stating that younger individuals consume more ultra-processed foods is not necessarily true- they don’t actually know!
In their sample, only 3 forms of cancer were reported- breast, colorectal and prostate. Of course there are lots more, they just didn’t happen to affect this group of people. That does not mean that other cancers are not (or are) associated with ultra-processed food consumption. Nor does it mean that if you consume ultra-processed foods you will end up developing breast cancer!
The researchers played around with the data in this study A LOT. They used modelling and hazard ratios, and basically make it seem like they played until they got a significant result! What has been reported in the newspapers are the significant findings- “a 10% increase in the proportion of ultra-processed foods in the diet was associated with significant increases of 12% in the risk of overall cancer and 11% in the risk of breast cancer”. But, the researchers tested lots of other stats that showed no association and were not significant. For example, they could not find any association between prostate cancer and ultra-processed foods, pre-menopausal breast cancer and ultra-processed foods or colorectal cancer and ultra-processed foods. But “ researchers fail to find link between colorectal cancer and ultra-processed foods” isn’t an attention grabbing headline, is it?
They actually do not mention any associations between consumption and mortality (death) from cancer, although they found that 108 of their participants died from cancer.
What the study doesn’t take into account:
They make assumptions throughout that processed foods are high in sugar, salt, fat and energy and therefore are likely to be associated with illness. However, as I’ve mentioned processed foods have a valuable place in the diet. By processing food; we ensure it is safe and microbe free, it ensures we have a variety of foods all year around, it allows us to try foods not native to our country without travelling abroad, it often makes food easier and faster to prepare, it makes new food products and it often makes products cheaper, as they are produced on a large scale, making food accessible for all.
There is an assumption that recent increases in illness must be associated with an increased consumption of processed foods. However, since the introduction of processed foods into the diet there have been other changes that may play a role. For example, our lifestyles have become more hectic, we are more sedentary, we can diagnose illness (especially cancers) much earlier (so it often appears there is more of it), our air is more polluted….the list goes on!
As is the trouble with all studies on humans, we cannot control for every outside variable that could affect the data. Here they have included plenty of information about their participants, but it isn’t the whole story. We don’t know about their income, working hours, where they live, what type of physical activity they do, what other foods they have consumed… and again, the list goes on! Us humans are complex, we come into contact with multiple things every day that could impact data such as this.
These results show an association, not causation. That is to say, that ultra-processed foods and breast cancer are association, but one does not definitely cause the other. There may be another factor at play which is not being studied, such as those mentioned in the above point. E.g. breast cancer risk in women increases greatly following the menopause, because our hormone levels change dramatically. That means that as women get older their risk increases. The researchers here state that post-menopausal breast cancer risk is increased by consumption of ultra-processed foods, but this may just be because of the increasing time since menopause (and likely has nothing to do with diet).
Take home message:
Processed foods are very useful and should be included as part of a varied diet. This study does not prove that processed foods are dangerous or cancer-causing at all. In fact, I don’t believe it proves anything (and honestly I’m a bit surprised it was published!), they noted a pattern and it grabbed journalists. Please always remember that a journalist, not a scientist, has written what you read in the newspaper, and therefore is likely to have been misinterpreted the data! I find it quite frustrating reading these articles because they create fear and encourage people to go against dietary recommendations and professional advice. Understanding scientific literature is difficult and it takes a lot of practice to fully understand research language and data. As far as I know, journalism courses don’t cover that, so please make sure your information comes from informed sources.
If you see articles referring to health/nutrition and you aren’t sure whether they are true, send then to me and I’ll write another “behind the headlines” piece. Don’t be afraid to question articles!
If you want to read the study I’ve been referring to, get it here.
Lots of love,
Little O x