select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
An observational study published in Gut looks at sugar-sweetened beverage intake in adulthood and adolescence and risk of early-onset colorectal cancer among American women.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This is generally a well-conducted study, but it needs careful interpretation and the conclusions have to be tentative. Like the great majority of studies of associations between diet and disease, it’s observational, because that’s usually the only practicable approach available. You couldn’t, for instance, take a group of young people, choose half of them at random to drink lots of sugary drinks and forbid the other half from drinking them, and then follow them up for many years to see who gets bowel cancer. So, instead, the researchers did something that sounds roughly similar – they took a large existing group of people (US registered nurses) who had been recruited originally in 1989, recorded what they ate and drank on several occasions, followed them up for many years, and recorded when any of them were diagnosed with bowel cancer under the age of 50. Then they looked at whether the risk of bowel cancer at these relatively early ages was different in women who drank different amounts of sugar-sweetened drinks – and they found evidence that the risk was higher in women who reported that they drank more sugary drinks. The snag here is that there were plenty of differences between women who drank few sugary drinks and women who drank a lot, apart from their consumption of sugary drinks. For example, the women who drank the largest amounts of sugary drinks tended to smoke more (if they smoked), to be less active physically, to drink less alcohol, to eat more meat, and to eat less fibre, than women who drank the smallest amounts. It’s possible that these other differences are, at least in part, the cause of the differences in early onset bowel cancer risk, rather than the sugary drink consumption. It’s also possible that these other differences might mask the effect of drinking sugary drinks, or make it look smaller than it actually is.
“It’s possible to make statistical adjustments to try to allow for other differences between groups in an observational study, and the researchers did that for many other factors that are known or suspected to be related to bowel cancer – and on the whole those adjustments made the differences in risk between the sugary drink consumption groups larger rather than smaller. That’s consistent with the possibility that the other differences between the groups were to some extent masking the association between sugary drinks and early onset bowel cancers – but there are other possibilities. Maybe the researchers did not adjust for some important factor – you can never be sure these adjustments have dealt with everything, and you can’t adjust for factors on which you have no data. So we just can’t be sure whether the observed association between sugary drinks and bowel cancer under the age of 50 is one of cause and effect. The researchers describe a possible biological mechanism by which the sugary drinks might have a causal effect on cancer risk, but just because this is possible, that doesn’t mean it actually occurs. Normally one would build up a clearer picture of what causes what by, for instance, looking at a range of different observational studies of the same thing and seeing whether their findings are consistent. But that can’t be done, yet, for sugary drink consumption and early onset bowel cancer risk, because this is the first major study. The issue of whether the sugary drinks actually cause increased risk is important, because if the association is not one of cause and effect, you won’t be able to change people’s bowel cancer risk by getting them to drink fewer sugary drinks. There is, however, better evidence about other reasons for reducing consumption of drinks sweetened with sugar, so just because this particular study can’t establish cause and effect, that certainly doesn’t mean that all advice about reducing sugary drinks is inappropriate.
“How big are the risks anyway? An important point is that bowel cancers diagnosed under the age of 50 are not all that common. Bowel cancer overall is one of the most common cancers in countries like the US and UK, but most bowel cancers arise in people over 50, usually well over 50. According to Cancer Research UK*, only about 7 in every 100 bowel cancers in women in the UK occur before the age of 50, and only about 5 in every 100 for men. However, because bowel cancer is, across all ages, quite common, there were on average about 2,500 new bowel cancers each year in people (men and women) aged under 50 in the UK, over the years 2015-2017, looking for ways to reduce the risk is clearly worthwhile (though we can’t really be sure how the results in US women from this new study would carry over to the UK). In the new study, there were only 109 diagnoses of bowel cancer under the age of 50 across the more than 95,000 women involved. The strongest evidence of an association between sugary drink consumption and early-onset bowel cancer was in comparing those who drank more than two servings a day with those who drank fewer than one serving a week. (A serving is quite large – 12 US fluid ounces, or about 350 ml or over half a UK pint.) For 10,000 women, like those in this study, who drank fewer than one serving a week, on average fewer than one each year would have a bowel cancer diagnosis under the age of 50. Over the average follow-up period in this study, about 12 of them would have a bowel cancer diagnosis under the age of 50. (The press release says that the follow-up was 24 years, but that is the maximum for all the women, and the average was about 14 years.) If instead all these 10,000 women drank 2 or more servings every day, and if the increased risk is entirely caused by the sugary drink consumption (and, remember, we don’t know that that’s the case), that number would go up from about 12 in every 10,000 women to about 26 – though there’s still a lot of statistical uncertainty and that number could plausibly be anywhere between 13 (a very small increase) to about 50 (a fairly large increase). But going from just one serving a week, or less, to at least two a day (so more than a UK pint a day) is a pretty big increase in consumption. Another way to look at it is to imagine 10,000 women like those in this study, drinking the range of amounts of sugary drink that the nurses in the study drank. On average, the number who would have of a bowel cancer diagnosis under the age of 50 over the follow-up period would be about 11. If every one of them drank an extra 350ml serving of sugary drink every day, and if the sugary drink was the sole cause of the difference in risk, that number could go up from 11 to 13, though again there’s statistical variability and it could go up to about 15 or not increase at all.
“Just a word of caution, though, about the mention in the press release about a bigger association of sugary drink consumption in adolescence with early-onset bowel cancer risk – a 32% increase per serving of sugary drink, compared to a 16% increase on the basis of the dietary information the women gave during this study. That sounds quite a big difference. It’s true that that is the finding in the research report, though there’s a lot of statistical uncertainty and the range of plausible values around that 32% goes from zero to 75%. However, this is based on asking the participants to recall their diet as teenagers, a minimum of 16 years later and in some cases, it appears, over 30 years later. I’m dubious about how accurate their recall would be. The researchers quote two studies that indicate that recall might be reasonably accurate, but they looked at recall over a considerable shorter period of time. Also, the other studies looked at correlation between the actual diet in adolescence and the recalled diet later, and correlation isn’t a measure of how close the measures are, just of how commonly high or low values of one are associated with high or low values of the other – and in any case the correlations are not particularly high. My reading of the results of these previous studies is that people tended to give higher values than the true level for high consumptions, and lower values of low consumptions. That wouldn’t affect the correlation measures, but it would affect the size of any association in terms of how much a risk increases with each extra serving. So maybe the association with sugary drink consumption in adolescence isn’t really any different from the association with sugary drink consumption in adulthood, or maybe it is – we can’t tell.”
Dr Adam Jacobs, Director of Biostatistics, Premier Research, said:
“This looks on the face of it like a large study, as it included 95,464 women. But the conclusions depend mainly not on the total sample size, but on the number of people who developed early onset colorectal cancer (EO-CRC), and that was only 109 participants. Their main conclusions seem to be based on the risk of drinking two or more servings of sugary drinks per day (there was no significant increase for less frequent categories compared with the least frequent category), and there were only 16 participants who consumed two or more servings of sugary drinks per day in adulthood and who developed early onset colorectal cancer. I think it is not sensible to try to draw firm conclusions from just 16 events.
“The results for sugary drink consumption in adolescence are even less convincing: only six participants developed early onset colorectal cancer in the >=2 servings of sugary drinks per day group. It is also worth noting that the consumption in adolescence was not measured prospectively: participants were asked to recall their consumption in adolescence years or potentially decades after the fact, which is potentially subject to all sorts of biases.
“They did get nominally statistically significant results in some (but not all of their analyses), but they are only just significant. Quite apart from the small sample size, there are many different ways you could analyse data like this, and the authors do not mention any kind of prospective registration of their study. That means that there is no way of knowing whether this was their primary analysis or whether they tried a dozen other analysis methods first before they found one that gave a significant result.
“The authors made good attempts to adjust for various confounders. However, many of the confounders are lifestyle factors, such as red meat consumption, which are difficult to measure accurately. Any inaccuracies in measurement of confounding variables could easily bias the results, especially when the associations are so marginal in the first place.
“It’s also worth putting the results into context by looking at the absolute risk. Even if we believe that the risk of early onset colorectal cancer was increased in the >=2 servings per day group, there were only 16 cases from 138,469 person years of observation. That’s a very low absolute risk. If we accept that sugary drinks double the risk of early onset colorectal cancer, and we can’t know that from this study because it can’t prove cause and effect, that means that over a 10 year follow-up period, you’d only expect to see one extra case of early onset colorectal cancer in about 1700 people.
“Finally, it’s worth noting that this study only included women, so we don’t know if the results would also apply to men.”
Dr Carmen Piernas, University Research Lecturer and nutrition scientist, Nuffield Department of Primary Care Health Sciences, University of Oxford, said:
“This is an epidemiological observational study looking at the associations between sugary drinks consumption and risk of developing colorectal cancer among American women. The research question is interesting and relevant, and their methods are sound, as they tried to adjust for the typical confounders which could have explained some of this increased risk, such as dietary quality and body mass index.
“However, the main findings are limited by the lack of power in this analysis. The main finding shows that adults drinking two or more daily servings of sugary drinks had a higher relative risk of developing colorectal cancer than those drinking less than one sugary drink a week. But this conclusion is based on just 16 cases of colorectal cancer in 138,469 person-years among the group drinking two or more sugary drinks. The absolute risk is very low. A related analysis reporting the increase in risk with each additional sugary drink serving per day is only borderline significant. The second finding showed that those drinking two or more sugary drinks at the age of 13-18 years also had a higher relative risk than those drinking less than one sugary drink a week. There is again a very high level of uncertainty to this finding because the analysis is based on only six cases of cancer found in this group. This is too small to draw any strong conclusions.
“Overall, these findings should be considered as preliminary and exploratory until larger studies are done in other populations (e.g. including men and women, and with a broader range of socioeconomic status, with varying dietary patterns). Meanwhile, there is already strong evidence that consuming sugary drinks increases risk of weight gain and diabetes. Everyone should aim to minimise their consumption – this analysis does not change that advice.”
Dr Ian Johnson, Nutrition researcher and Emeritus Fellow, Quadram Institute Bioscience, said:
“This carefully conducted observational study has identified a statistically significant relationship between consumption of sugar-sweetened beverages (SSBs) and the risk of early onset colorectal cancer (CRC) in a large cohort of female American nurses. Colorectal cancer is primarily a disease of later life but the incidence among younger age-groups in their forties and fifties has been rising in the United States and other industrialised nations in recent years. In this new study, participants consuming more than two servings of SSBs per day experienced about twice the risk or early onset colorectal cancer compared to those consuming less than one serving per week. Replacing sugary beverages with artificially sweetened drinks, coffee or milk was associated with a lower risk.
“The statistical uncertainty of these results was relatively high, reflecting the fact that early onset colorectal cancer is still fairly uncommon, but the size of the increased risk was large compared to the effects of other, better-known dietary risk-factors, such as processed meat intake, and a low consumption of dietary fibre. The work needs to be taken very seriously and further studies should be carried out to confirm or refute the findings. As with any observational investigation, this study does not conclusively prove a causal effect.
“Nevertheless, the authors point to several plausible mechanisms, including recent laboratory studies from other scientists showing that the sugars used to sweeten SSBs favour the development of colorectal cancer in an animal model. Excessive consumption of sugar-sweetened beverages is already strongly implicated as a risk-factor for excess weight gain and type-2 diabetes in human populations. Current dietary advice in the UK recommends that their consumption should be minimised in both children and adults. This new study provides further support for this advice.”
Dr Duane Mellor, Registered Dietitian and Senior Teaching Fellow, Aston Medical School, Aston University, said:
“This is a very large study of women in the US over 25 years, it looked at dietary intake measured in over 95 thousand women in the Nurse’s Health Study II. Every four years it asked women what they ate on average over the past year. This study suggested an association between consuming sugar sweetened drinks and being diagnosed with bowel cancer before the age of 50. The way this is presented was by what is called a risk ratio, which was found to be nearly 2.2 fold higher for women who drank more than two sugary drinks a day compared to those having less than one a week. However, when looking at number of cases per year for each person in the study, it was 0.000116 cases per person per year for high consumers of sugary drinks compared to 0.000084 cases per person per year for those who rarely drink sugary drinks. This suggests although there is a statistical significant association with risk, the overall number of cases of bowel cancer and risk in this middle aged population was low. This is perhaps because bowel cancer is more common in individuals over the age of 50.
“Part way through the study some of the women who at this time would have been at least 30 years old were asked to recall how many sugary drinks they drank as teenagers. This was also associated with an increased risk of bowel cancer before the age of 50. This would have required women to remember what they drank decades ago!
“The study compares those drinking sugary drinks less than weekly with those having more than two drinks per day, the data also suggested those women who had up to one drink per day did not seem to be at higher risk when directly compared as a group to those who drank them rarely, but when the data was combined the statistics suggest an increased risk per drink per day. This is likely to be an effect of the mathematical modelling, but as we already know related to the health of our teeth and the extra energy associated with these drinks can negatively impact your health.
“The researchers also showed that having drinks with milk, artificially sweetened drinks or coffee instead of sugary drinks appeared to be associated with a lower risk of bowel cancer before the age of 50, however having water or tea rather than sugary drinks was not. This is perhaps an effect of relatively small numbers of cases of bowel cancer in this group, and does not mean that replacing sugary drinks with water makes no difference, as most health messages would recommend choosing water over other drinks for adults. The researchers did control for the key factors which influence risk of bowel cancer, and it is important to remember that not smoking, being physical activity and increased intakes of fibre rich foods including fruit and vegetables can reduce your risk for developing bowel cancer. So, although it is possible that reducing intakes of sugary drink intake may in a small way moderate risk of bowel cancer, perhaps it is likely to have little effect if done without improving lifestyle and health including overall diet.
“Although the authors suggest the increase in bowel cancer may be linked to increased sugary drink consumption, this may not be quite the case in the UK as following the introduction of the ‘Sugar Tax’ (Sugar Sweetened Beverage Industry Levy) this has led to a reduction in sugar content of most soft drinks available in the UK. Most drinks have gone from around 30-35g (6-7 teaspoons of sugar) per can down to 15-20g (3-4 teaspoons) per can, with increasing numbers of sugar-free (artificially sweetened) drinks becoming available. So, it will be interesting to see if these increases noted in the US are also happening in the UK over the next decade, to see if the theory that sugary drinks are associated with increased risk of bowel cancer continues to be seen following the introduction of the ‘sugar tax’.”
‘Sugar-sweetened beverage intake in adulthood and adolescence and risk of early-onset colorectal cancer among women’ by Jinhee Hur et al. was published in Gut at 23:30 UK time on Thursday 6 May 2021.
DOI: 10.1136/gutjnl-2020-323450
Declared interests
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. However, my quote above is in my capacity as an independent professional statistician.”
Dr Adam Jacobs: “I have no competing interests in terms of my professional affiliations or other financial considerations, but I should mention that I wrote a blogpost about sugar tax a few years ago: http://www.statsguy.co.uk/sugar-tax/“
Dr Carmen Piernas: “No conflicts.”
Dr Ian Johnson: “Ian Johnson was a member of the SACN Carbohydrates Working Group from 2009 to 2015.”
Dr Duane Mellor: “I was part of International Sweetener Association roundtable which developed an expert consensus on low-calorie sweeteners.”