A new paper by Dunneram et al. (University of Oxford) has reported a higher average statistical risk for colon cancer observed in the vegan group compared to omnivores. This is based on a meta-analysis that, the authors report, includes 8849 vegans in nine cohorts - in the USA, UK, Taiwan, and India.
Where did they find all these cohort studies with so many vegans?
Well, actually I am not so sure it was a good idea to combine these nine studies - because only two of these studies were really designed to recruit vegans, assess vegan status, and compare health outcomes in vegans compared to omnivores. These two studies are well-known to most vegans with a bit of an interest in nutrition science.
- EPIC-Oxford (England)
- Adventist Health Study-2 (AHS-2; USA and Canada)
The relevant graphic for the "vegans and colon cancer" result can be found in their (Dunneram et al.'s) Supplementary Figure 5. I modified this figure by taking out the "poultry eaters", "pescatarians", and "vegetarians" (presumably they mean ovo-lacto-vegetarians) and by adding the text in red. See the figure below. Click on the figure to enlarge it.
Original figure description by Dunneram et al.:
"
Supplementary Figure 5. Study-specific and pooled multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals comparing the risk of colorectal cancer in poultry eaters, pescatarians and vegetarians to that in meat eaters.
All models used age as the underlying time variable and are stratified by region or method of recruitment and sex, and adjusted for living with partner, educational status, ethnic group, height, body mass index, cigarette smoking, physical activity, alcohol intake, history of diabetes, parity, and ever use of hormone replacement therapy. Confidence intervals extending beyond the axis are marked with arrows.
*Ref, number of cases in the reference group of meat eaters.
Abbreviations: AHS-2, Adventist Health Study-2; EPIC, European Prospective Investigation into Cancer and Nutrition; MWS, Million Women Study; NIH-AARP, National Institutes of Health-AARP Diet and Health Study; OVS, Oxford Vegetarian Study; TCHS, Tzu Chi Health Study; UKWCS, UK Women’s Cohort Study; UKB, UK Biobank.
Supplementary Figure 5. Study-specific and pooled multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals comparing the risk of colorectal cancer in poultry eaters, pescatarians and vegetarians to that in meat eaters.
All models used age as the underlying time variable and are stratified by region or method of recruitment and sex, and adjusted for living with partner, educational status, ethnic group, height, body mass index, cigarette smoking, physical activity, alcohol intake, history of diabetes, parity, and ever use of hormone replacement therapy. Confidence intervals extending beyond the axis are marked with arrows.
*Ref, number of cases in the reference group of meat eaters.
Abbreviations: AHS-2, Adventist Health Study-2; EPIC, European Prospective Investigation into Cancer and Nutrition; MWS, Million Women Study; NIH-AARP, National Institutes of Health-AARP Diet and Health Study; OVS, Oxford Vegetarian Study; TCHS, Tzu Chi Health Study; UKWCS, UK Women’s Cohort Study; UKB, UK Biobank.
"
The figure shows that vegans in AHS-2 basically had more or less the same colon (to be more precise: colorectal) cancer risk than omnivores. This might suggest vegans have some benefit (no red meat etc.) but maybe also have some disadvantage (like lower calcium intake). Higher calcium intake is known (thought) to be protective.
The AHS-2 study result is also an indication that the fact that vegans do not eat animal products is not a clear risk factor for colon cancer. Still, some animal products (dairy) are rich in calcium, which seems protective.
The figure shows that, in EPIC-Oxford, the vegans had a higher risk of colon cancer than the omnivores. Possible reasons could be "reverse causation" (some people who were ill became vegan, then took part in the study) but also true dietary reasons, like lower consumption of calcium, vitamin D, selenium - or coffee - in vegans.
The rest of the figure is - in my crazy opinion - not that relevant because these other studies were not meant to assess vegans at all.
A more reasonable combination of studies would be to just combine EPIC-Oxford and AHS-2. AHS-2 contributed the largest number of vegans and also the largest number of colon cancer cases in vegans. Then we have EPIC-Oxford, and the rest of the studies contribute only a handful of vegans with colon cancer cases - and I am not so convinced that these people were vegans. If we just look at the top of the figure, we could assume that combining AHS-2 and EPIC-Oxford results would lead to an average somewhere between the two black dots, leaning towards the left (AHS-2) because AHS-2 has more participants.
More interesting details for the "vegans and colon cancer" aspect can be found in Dunneram et al.'s Supplementary Figure 3. I modified it by cutting out all cancer types related to anything but the colon and by adding the red dots (for statistically significant results) and the orange dots (for statistically non-significant results). Click on the figure to enlarge it.
The figure shows that the results were clearest for rectal cancer. I would guess this was - like the main result above - also mainly driven by EPIC-Oxford results.
Note: This result regarding vegans is only a minor result of Dunneram et al.'s paper.
Side note: I did not see a mention of which outcome parameters and which participant groups (e.g., colon cancer risk in vegans vs. omnivores) were pre-defined (pre-specified) before the statistical analysis was done. Nor do I see a mention of "methods used to assess risk of bias in the individual studies" (PRISMA-IPD checklist). Nor do I see a mention of "how the [..] variables to be collected were chosen" (PRISMA-IPD checklist) - of course, getting more vegans into the analysis is a good reason to include all the studies, and colon cancer is one of the most common cancers, but why did they decide that all the studies apart from AHS-2 and EPIC-Oxford provided reliable enough data on vegans to permit them being included in the analysis for vegans? ... These are not important aspects for real-life vegans, because EPIC-Oxford did show an increase risk, and there must be reasons for this association which may or may not be causal.
Side note: One peculiarity in EPIC-Oxford is that the UK is pretty much the only country where white flour is routinely fortified with calcium, which makes it surpising that calcium intake in vegans (who are not "whole foods vegans") would be low.
So what does it all mean? What should vegans eat?
I think for real-world vegans - particularly outside of the UK - this meta-analysis does not change anything.
In a paper in 2022, I wrote:
"While it has been hypothesized that the lack of a protective effect of vegetarian and vegan diets against colon cancer risk observed in the EPIC-Oxford study may be a consequence of low selenium status among British vegetarians/vegans (Tsilidis et al. 2021; Sobiecki 2017), multiple other reasons are possible such as a low vitamin D status (Crowe et al. 2011), low calcium intake (Tong et al. 2020; Molina-Montes et al. 2021), and/or low zinc intake (Li et al. 2022)."
Possible colon cancer risk factors to avoid:
- Generally less healthy diet - which might be the case in some EPIC-Oxford vegans. Their average fibre intake was estimated (in 2003) to be about 26-28 g/day (Davey et al. 2003, Table 4).
- Low calcium intake
- Low vitamin D status (sunshine and/or intake)
- Low zinc intake - maybe due to low intake of legumes, nuts, seeds, whole grains
- Low selenium intake - appears to be the case in some European countries like the UK and Germany -- therefore, the recommendation to take a supplement like VEG 1 (Vegan Society) or eat some Brazil nuts.
- Low coffee intake - does it justify becoming a coffee drinker? I don't know.
- Smoking
- Excess alcohol intake
- Lack of physical activity/exercise
- Overweight/obesity
Also see:
Reference:
Yashvee Dunneram, Jia Yi Lee, Cody Z Watling, Izabella Lawson, Mahboubeh Parsaeian, Gary E Fraser, Fayth M Butler, Dorairaj Prabhakaran, Krithiga Shridhar, Dimple Kondal, Viswanathan Mohan, Mohammed K Ali, K M Venkat Narayan, Nikhil Tandon, Tammy Y N Tong, Ruth C Travis, Tina H T Chiu, Ming-Nan Lin, Chin-Lon Lin, Hsin-Chou Yang, Yu-Jen Liang, Darren C Greenwood, Gillian K Reeves, Keren Papier, Sarah Floud 1, Rashmi Sinha, Linda M Liao, Erikka Loftfield, Janet E Cade, Timothy J Key, Aurora Perez-Cornago: Vegetarian diets and cancer risk: pooled analysis of 1.8 million women and men in nine prospective studies on three continents. Br J Cancer . 2026 Feb 27. doi: 10.1038/s41416-025-03327-4. Online ahead of print. https://pubmed.ncbi.nlm.nih.gov/41748939/
