select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
A conference abstract, presented at the European Association for the Study of Diabetes Annual Meeting 2022, looks at tea consumption and risk of type 2 diabetes.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“At this time of year, a lot of scientific and medical conferences take place. The reports and findings discussed in them cover research that might be very important or might be of dubious quality – you just can’t tell at this stage. That’s because the full details of the work are not available, and (usually) the report would not have been through anything like full peer review before the work is presented at the conference.
“This new study is no exception. With what we’ve been told so far, the findings need to be taken with a very large pinch of salt (or cup of tea). They could, just possibly, be important, but we don’t know anywhere like enough details to tell, and there are many likely limitations.
“The quote in the press release from the lead author says that the results suggest that drinking four cups of tea a day might potentially reduce people’s risk of developing type 2 diabetes. My emphases, but the words “suggest” and “potentially” are crucial here. Tea drinking would only be useful for reducing diabetes risk if the tea drinking causes reductions in risk, that is, if the risk is reduced if you drink the tea and not if you don’t – and this study simply can’t show whether it does this or not. I’ll explain that in more detail later. Even if there is a cause-and-effect association, it might depend on what (if anything) people stop drinking, if they drink more tea, and again this study can’t tell us about that.
“Some of the limitations of the research findings are mentioned in the last paragraph of the press release, but there are others. First, the researchers’ own cohort study, and all of the studies they included in their review and meta-analysis of previous work, are observational. People weren’t assigned by the researchers to drink different quantities of tea. They just did what they would have done anyway, and their tea consumption and other factors were observed and recorded, together with whether they later developed type 2 diabetes. So people who drank different amounts of tea would have differed, on average, in many other ways too. Those other factors may have been the cause of any observed differences in diabetes risk, rather than the tea consumption itself. The researchers on the new study say that they made some statistical adjustments to allow for other factors, but we don’t have details on what they adjusted for, or how they did the adjustment statistically, so I can’t tell how adequate they might be. Even if they did the best conceivable statistical job on the adjustments, there will always be relevant factors that can’t be adjusted for because the data weren’t available. So it’s never possible to be sure what’s causing what, the basis of a single observational study.
“In fact the researchers’ own cohort study didn’t, in any case, find any evidence at all of an association between tea drinking and type 2 diabetes risk. On its own, it provides absolutely no basis for any advice about drinking tea. Their assertion that there is an association (causal or not) is based on a formal meta-analysis of 19 different cohort studies of tea drinking and type 2 diabetes, where the results from those studies are put together statistically to give an overall picture of what might be going on. It’s that meta-analysis that found evidence of an association, but only in people who drank more than four cups a day, compared to those who drank no tea. But all these studies that were included were also observational, so every one has the same problems with cause and effect.
“The trouble with meta-analysis findings is that the devil is always in the detail, and we don’t have the detail. Which studies were included? What was their quality? Which people, from which countries, were studied? What adjustments, if any, did the researchers make for other factors? What statistical methods were used to do the meta-analysis? We just don’t know. It’s true that combining the results of many observational studies can sometimes give a clearer indication of what might be causing what than you’d get from a single study – but here, there’s simply not enough information to make any judgement about that.
“The researchers point out that the tea consumption figures come from the participants’ subjective measures of how much tea they drank – and in their own cohort study, those figures seem only to have been collected at the start of participation, so wouldn’t have taken into account any changes in tea consumption or other aspects of diet (or anything else) during the twelve years of follow-up. There may well be similar issues about the other studies in their meta-analysis, but we just can’t tell.
“I’d add that people tend to drink tea in different ways in different cultures. In China, people are very much less likely to put sugar or milk in their tea than in the UK, for example, so it’s far from clear how the results of the researchers’ own cohort study would apply in the UK context anyway. And we haven’t been told which countries or cultures provided the data used in the meta-analysis that they did.”
Dr Baptiste Laurent, Lecturer in Medical Statistics, University College London, said:
“A study to be presented next week at the European Association for the Study of Diabetes claims that drinking plenty of tea may reduce the risk of developing type 2 diabetes.
“Being only a conference abstract, it is difficult to assess the quality of this research.
“They conducted two studies. One of them (a cohort of 5,000 adults in China) did not find an association between tea consumption and risk of diabetes.
“Their second study, a meta-analysis of 19 observational studies, found “each cup of tea consumed per day reducing the risk of developing T2D by around 1%”.
“However, all the studies included were observational, and therefore no cause-effect conclusions can be drawn. The association could simply be due to other factors, such as those drinking more tea having a healthier lifestyle. It does not seem that the authors tried to control for confounders, which is usually difficult in meta-analysis.
“Overall, this apparent association could be explained by other factors, such as different diet, and we should be very cautious before concluding that drinking tea could reduce the risk of type 2 diabetes, particularly as the study has not yet been published.”
Prof Naveed Sattar, Professor of Metabolic Medicine, University of Glasgow, said:
“Given the nature of this study, it cannot prove tea prevents diabetes per se. Rather it could be that people who drink more tea avoid or less often drink more harmful sugary drinks or equivalent or that the have other health behaviours that leads them to have lower risks of type 2 diabetes.
“There is no good trial evidence whatsoever that the chemicals in tea prevent diabetes, so I suspect its more about tea being healthier (less calorific) than many alternative drinks or tea drinkers leading healthier lives more generally.”
Prof Matt Sydes, Professor of Clinical Trials & Methodology, MRC Clinical Trials Unit (CTU), said:
“This is large, observational data. It’s not a randomised controlled trial so there’s plenty of room for data to be misunderstood. Importantly, everyone drinks fluids. If there is an effect here (and that’s a big if), it might be not about the tea they drink, but about what they don’t drink because they are drinking tea at those moments. One can’t tell at the moment. It seems unlikely that a large randomised controlled trial could be done to disambiguate.”
Dr Jonathan Cook, Centre for Statistics in Medicine, University of Oxford, said:
“It is hard to judge if this conference abstract accurately reflects the science as we don’t have the full paper, they are just going to present results at a conference so the associated paper it seems has not yet been published. Therefore, there is reason to be a bit sceptical at this point, we really need to have the full details to assess it properly.
“It’s a fair attempt to look at this but not cutting edge fairly standard approaches. There is one large study (though perhaps still not large enough to look at a question like this in this way), and a review of other studies which have looked at this (some must be very large given the combined size of these is over 1 million people).
“It appears to add to existing evidence but with one big caveat. The large cohort study doesn’t seem to have been set up to look at the impact of tea drinking as such, and therefore seems to lack relevant information. This may well also be true of other studies in the review they have conducted. This matters in that the data you want to control for is often not available. My view is the conclusion of the abstract overstates what they have observed and can justify at this point.
“Only limited adjustment for confounding has been done. I suspect this is all they have data for but there could readily be other factors driving the association.
“To my mind, there are no real world implications at this point. It is suggestive, and the apparent dose response where you need 4+ a day to get a substantial health benefit really needs further exploration and research to back it up. This study alone isn’t enough.
“There may be other reasons for this correlation, consumption could be related to other factors like general health and mobility etc. They have adjusted for “age, sex and physical activity” which is useful but almost certainly not sufficient to cover all potential factors that influence development of type 2 diabetes and could through up a spurious relationship. The press release does acknowledge the need for further research at the very end.”
Dr Duane Mellor, Registered Dietitian and Senior Teaching Fellow, Aston Medical School, Aston University, said:
“It is important to note this is a conference proceeding, so it has not been subject to the detailed peer review of a full paper and many details are not available.
“Although this reports on a very large number of people, it uses data from a new analysis combined with data published previously. It is interesting that it does not show any association when looking at different types of tea or where people lived were looked at separately. Although mathematically it suggested a small reduction in risk associated with type 2 diabetes per cup of tea, the data itself only showed a significant reduced risk in people drinking more than 4 cups per day. This perhaps shows that there are other factors confounding which have not been accounted for which happen to have given a significant result. It is plausible that people who drink tea may not be drinking other drinks which are higher in energy or contain free sugar or alcohol. It is also interesting that similar studies have shown a reduced risk associated with coffee drinking and risk of type 2 diabetes previously.
“Important take home messages is that lifestyle is important in managing risk of developing type 2 diabetes, that includes choosing low calorie drinks including mainly water as well as unsweetened tea and coffee as you drinks of choice as part of a healthy lifestyle.”
Prof Aedin Cassidy, Chair in Nutrition & Preventative Medicine and Director for Interdisciplinary Research, Institute for Global Food Security, Queen’s University Belfast, said:
“The study is currently a conference abstract so the research hasn’t yet been peer reviewed and published in a high quality journal. The presented results will need this review process to confirm their robustness and validity over the next few months. if validated through peer review these data will add to our knowledge base that tea is an important part of a healthy diet and the simple addition of a few cups a day (with the greatest benefits seen when 4 cups a day or more are drunk) may reduce the risk of developing type 2 diabetes. The data are a pooled analysis of previous studies which together suggest the greatest benefit is seen when at least 4 cups a day are drunk. Further work is needed to determine why tea may reduce risk and it is likely that key constituents in tea like the flavonoids hold the key to explaining these benefits as they have been shown to influence blood glucose and insulin levels and reduce inflammation.”
Abstract title: “Tea consumption and the risk of type 2 diabetes: a cohort study and updated systematic review and dose-response meta-analysis”, by Xiaying Li et al; Presented at the European Association for the Study of Diabetes Annual Meeting 2022
There is no paper.
Declared interests
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.”
Dr Baptiste Laurent: “No conflict of interest.”
Prof Matt Sydes: “No conflicts to report.”
Dr Duane Mellor: “No conflicts of interest.”
Dr Naveed Sattar: “I have consulted for many companies that make diabetes and cardiovascular drugs. I have also been involved in multiple trials of lifestyle for prevention and remission of diabetes.”
Prof Aedin Cassidy: “I co-chaired an international conference on tea and health in April ( and was paid for my time in organising the session by a US PR agency who were sponsored by the tea industry).”
Dr Jonathan Cook: “I have no competing interests to declare.”