select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to study looking at dairy products and cancer incidence in Chinese adults

A study published in BMC Medicine looks at dairy consumption and cancer risk in Chinese adults.


Dr David Nunan, Departmental Lecturer and Senior Research Fellow at the Centre for Evidence Based Medicine, University of Oxford, said:

“Strengths of this study are that it is large and prospectively captured incidence of new cancers using methods that are generally well accepted for this type of research question and study design.”

“However, as this is an observational study we cannot be certain other factors and not dairy food intake may be responsible for, and certainly influencing to some degree, the observed associations with cancer incidence.”

“Even if we believe the observed associations to be real and true, it is important to place the findings of this study in context of how much the risk there is to an individual. Taking the findings for breast cancer as an example, there was 22% increase in the risk of breast cancer in those who stated they regularly ate dairy foods compared with those who stated rarely or never eating them. However, the important question to ask is what is the risk of breast cancer anyway? The people in this study where Chinese adults. The annual risk of breast cancer in this population is around 0.03%. This means that for every 1,000 people, 3 will develop breast cancer in 1 year (1000 multiplied by 0.003 [0.03% as a fraction]). If these 1,000 people increased their daily consumption of dairy foods to 80 grams – the equivalent to 6-7 tablespoons of milk, 4 to 5 tablespoons of butter or cheese – the risk may increase by 22% meaning 4 people in every 1,000 may now develop breast cancer (baseline risk 0.03% multiplied by 0.22 = 0.0066%. Add 0.0066% to the baseline risk of 0.03% = 0.0366%, rounded up to 0.04%; 1,000 multiplied by 0.004 = 4). It’s important to remember that for those 1,000 people who increase their dairy consumption, 996 will not develop breast cancer and would not have done so regardless.

“Where these new findings sit in relation to other similar studies is also important to recognise. Existing evidence that has systematically pulled together these types of studies but conducted across different countries have not shown the same findings of an increased risk in breast cancer with higher dairy consumption. In fact, they tend to demonstrate a protective effect of dairy consumption and incidence of breast cancer. Previous studies in Chinese participants have also not demonstrated an increased risk. The conflicting findings by Kakkoura et al will need careful consideration given their potential implications.”


Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This is an interesting study and one that was carried out well in statistical terms, in my view.  Like any study of its type, though, there are aspects that need care in interpretation.

“One important point is that it’s an observational study.  The researchers did not try to change what the participants ate and drank – they just recorded it at several points during the study, and followed the participants up for about 11 years on average, recording any cancer diagnoses.  An issue is that there were many differences between the people that consumed different amounts of dairy products, apart from their difference in dairy consumption.  For instance, of those who never or rarely consumed dairy products, fewer than a third lived in urban areas, but of regular dairy consumers (at least once a week), 83% lived in urban areas.  Regular consumers were considerably more likely to be well educated than those who never or rarely consumed dairy products, and there were other differences too.  So if, as the researchers found, a greater proportion of the regular consumers than of the never or rare consumers had a cancer diagnosis, that could have been because of their different dairy consumption, or it could have been (in part or entirely) because of the different places they lived, or their different education levels, or any of the other factors on which the groups differed.  The researchers knew this, of course, and accounted for the differences that they knew about by making statistical adjustments or looking at subgroups of the participants separately.  But one can never be sure that all the relevant factors have been adjusted for.  That’s why the researchers rightly say that these results can’t establish whether the associations between dairy consumption and the risks of some cancers, that they found, are there because the dairy consumption differences change the cancer risks in a cause-and-effect way.  They might, or they might not.

“Since a single observational study of this kind can rarely, if ever, establish cause and effect, knowledge of cause and effect has to be built up by looking across several studies, and taking into account other things, like understanding of how different levels of dairy consumption might work inside the body.  The researchers do give reasons, in terms of the biology, why they believe it is plausible that dairy consumption might well have a causal effect on cancer risk.  But they can’t put their results together with other observational studies to build up a clearer picture, because this is the first one of any substantial size in a Chinese population.  There have been several large studies, but generally in Western populations.  Chinese populations differ in several relevant ways: dairy consumption is much lower there than in the West (with, in this study, over two-thirds of participants consuming either no dairy products at all, or consuming them less than once a month); cancer rates are different (including the balance between different types of cancer); and a genetically-determined inability to digest dairy products occurs in a large majority of Chinese adults.  So it would make no sense simply to put together the results of this new study with similar studies in Western populations.  Indeed a major motivation for doing the study was to investigate the ways in which associations between dairy consumption and cancers might be different in China from in the West.

“So, all in all, despite its size and the care with which it was generally carried out, this new research can’t establish cause and effect.  The researchers make that explicit, and call for further research to validate their findings (or otherwise) and to provide more evidence on cause and effect.

“Some of the findings about associations between dairy consumption and cancers in this Chinese study do look different from the overall pattern in studies in Western populations.  The study found that, after the statistical adjustments, regular consumers of dairy products did have a higher risk of all cancers taken together, of liver cancer, and of breast cancers (in females), compared to people who never or only rarely consumed dairy products.  But they found no clear evidence of associations between dairy consumption and the risk of cancers of any other site, including bowel and prostate cancers.  Studies in Western populations have indicated an increase in the risk of prostate cancers and a decrease in the risk of bowel cancers associated with higher dairy consumption, but no clear associations with cancers of other sites (including liver and breast).  Issues about cause and effect apply to the Western observational studies too, of course, though there have been more of them so the patterns are arguably clearer.

“On the face of it, at least, though, there are differences in patterns of association between the West and China.  That would indicate that these new results don’t apply to the UK population, for instance.  And it’s not even clear how far they might apply to people of Chinese ancestry living in the West, because there are many aspects of lives in the West, of people of whatever ethnicity, that are different from lives in China.  Indeed, all broad results from large epidemiological studies like this apply to whole populations or large groups of people, and provide results on averages across those populations.  Individuals can be very different from the average for the population from which they come.  So I don’t think anyone should decide to change their individual diet solely because of the results of this new study.”


Prof Ian Givens, Professor of Food Chain Nutrition, Director, Institute of Food, Nutrition and Health, University of Reading, said:

“The latest reports from the World Cancer Research Fund (WCRF; that collates and regularity updates evidence) concluded that for the association of dairy foods and premenopausal breast cancer there was some limited evidence of a reduced risk although for milk there were no associations seen (i.e. no increased or decreased risk). I don’t think there were any conclusions for postmenopausal breast cancer or for dairy and liver cancer.

“So the current study needs to be seen the light of the wider evidence and no doubt WCRF will provide this in due course.

“The current study is however potentially very important for Chinese people if it can be confirmed that dairy products affect the risk of breast and/or liver cancer differently in Chinese subjects to those in Western Societies, especially as dairy consumption in China is much lower than in most Western diets.

“Interesting to note the suggestive link with IGF-1 (which milk proteins do stimulate) as this has been linked with prostate cancer yet there was no increased risk of this for men in the current study.

“As always it needs to be kept in mind that this type of study can only establish associations with disease risk, not cause.”


Dr Duane Mellor, Registered Dietitian and Senior Teaching Fellow, Aston Medical School, Aston University, said:

“Although this study looked at half a million people for over a decade, it is important to note that compared to Western populations dairy consumption is lower, and this represents relatively recent changes in dietary habits.  This could explain why these associations have not been seen in European or North American population studies.  Also only 20% of the half a million Chinese participants were regular (weekly) consumers of dairy, which could skew the data.  Also those in the study who consume dairy were more likely to live in cities and have other health conditions including cardiovascular disease and diabetes – although some of these factors were considered in the analysis, not all of these covariates were, which could influence the findings.

“Although the paper suggests a 12% increased relative risk for female breast cancer, this does not equate to 12 more cases per 100 individuals – in absolute terms this would be more like 1 or 2 cases per thousand people.

“Although the paper suggested a biological link through insulin like growth factors, it is important as the paper acknowledges that this paper does not provide evidence of a causal link, especially as the mechanism suggested by the authors could be reflected in the differences in rates of diabetes in the study population at the start of the study, which has also been associated with increased risk of breast cancer.

“In my view this study alone does not provide strong evidence that reducing dairy intake would reduce cancer risk.  Dairy products can be useful sources of nutrients including calcium, vitamin B12 and protein and if people wish to reduce their intake or exclude them from their diet completely it is important they carefully consider what foods they are using to replace them and make sure they are still consuming enough of these essential nutrients as part of a varied and balanced diet.”



‘Dairy consumption and risks of total and sitespecific cancers in Chinese adults: an 11year prospective study of 0.5 million people’ by Maria G. Kakkoura et al. was published in BMC Medicine at 01:00 UK time on Friday 6 May 2022.

DOI: /10.1186/s12916-022-02330-3



Declared interests

Dr David Nunan: “No conflicts 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.”

Prof Ian Givens: “No interests to declare.”

Dr Duane Mellor: “I do not have any DoI for this one.”

in this section

filter RoundUps by year

search by tag