Wednesday July 1 2009
Vegetarians are less likely to develop cancer than meat eaters, according to several newspapers. They have reported on a study which found that vegetarians are 45% less likely to develop cancer of the blood (such as leukaemias and lymphomas) and 12% less likely to develop cancer overall.
The findings come from the pooled results of two large studies, which looked at cancer rates and dietary habits in 61,566 people. Participants provided information on their diet at the start of the study and researchers followed them for up to 26 years to look at their development of cancer. Of 20 cancers examined, the risk of stomach, bladder and blood cancers was reduced in vegetarians, while eating fish but no meat decreased the risk of ovarian cancer.
However, the incidence of these four cancers across the whole sample was low (particularly for stomach and bladder cancer), which decreases the reliability of the risk figure calculated and the clinical relevance for the general public. The study has some other limitations, which mean its conclusion that “being a vegetarian decreases your risk of cancer” must be made with great caution if based solely on findings of this study.
Where did the story come from?
T J Kay of the University of Oxford and colleagues of other institutions in the UK and New Zealand carried out this research. The study was funded by Cancer Research UK. The principal author has declared that he is a member of the Vegetarian Society. The study was published in the peer-reviewed British Journal of Cancer.
What kind of scientific study was this?
This study examined cancer incidence among vegetarians, an area that has not previously been examined in depth. To do this, the authors pooled the results of two UK cohort studies: the Oxford Vegetarian Study and EPIC-Oxford cohort.
The Oxford Vegetarian Study recruited 11,140 participants from throughout the UK between 1980 and 1984. Vegetarians were recruited through media advertisements and told they could also invite their non-vegetarian friends and relatives to participate. At enrolment, participants completed a food frequency questionnaire and provided information on smoking status, alcohol use, exercise habits, social class, weight, height and reproductive status.
The EPIC-Oxford cohort recruited participants from the UK through GP practices and a mailed invitation, which specifically targeted vegetarians and vegans. A questionnaire was mailed directly to all members of the Vegetarian Society, Vegan Society and all surviving participants in the Oxford Vegetarian Study. Respondents could also recruit friends and relatives.
A total of 7,423 participants were recruited through GP practices and 58,042 through the postal method. The questionnaire included a food frequency questionnaire and collected the same additional lifestyle and health information as the Oxford Vegetarian Study.
Participants of both studies were followed up to the end of 2006 through records from the National Health Service Central Register, which provides information on diagnoses of cancer and all deaths. Participants who were originally in the Oxford Vegetarian Study and were later included in the EPIC-Oxford cohort contributed follow-up data to the Oxford Vegetarian Study until the date that they transferred.
Participants were excluded if they were not aged between 20 and 89 at the time of recruitment, if they had a malignancy (cancer) before the study or if they had no information for one or more of the factors, such as age, sex, smoking and dietary group. This left a total of 61,566 participants across both studies (15,571 men and 45,995 women). Of these, 2,842 contributed data to both studies.
The researchers calculated the risk of 20 cancers and an overall risk of cancer according to dietary categories. They also adjusted for other possible confounding risk factors. The dietary categories were: ‘meat eaters’, ‘fish eaters’ (who did not eat any meat), ‘vegetarian’ (who ate neither meat nor fish) or ‘unknown’ if this was not clear.
What were the results of the study?
One-third of participants were vegetarian and 75% were women. The overall sample contained a low number of current smokers. There were additional differences in other factors, such as BMI, alcohol use and reproductive status, between people of the different dietary categories.
The significant findings of the studies were:
- Being vegetarian decreased the risk of stomach cancer compared to being a meat eater (relative risk [RR] 0.36, 95% confidence interval [CI] 0.16 to 0.78). There was no significant difference in risk between fish eaters and meat eaters.
- Being a fish eater decreased the risk of ovarian cancer compared to being a meat eater (RR 0.37, 95% CI 0.18 to 0.77). There was no significant difference in risk between vegetarians and meat eaters.
- Being a vegetarian decreased the risk of bladder cancer compared to being a meat eater (RR 0.47, 95% CI 0.25 to 0.89). There was no significant difference in risk between fish eaters and meat eaters.
- Being a vegetarian decreased the risk of blood cancers compared to being a meat eater (RR 0.55, 95% CI 0.39 to 0.78). There was no significant difference in risk between fish eaters and meat eaters.
- Compared to eating meat, being a vegetarian or eating fish but no meat significantly decreased the risk of any malignancy overall (RR 0.88 and 0.82, respectively).
What interpretations did the researchers draw from these results?
The authors concluded that incidence of some cancers may be lower in vegetarians and fish eaters than in meat eaters.
What does the NHS Knowledge Service make of this study?
The pooled results of these two large cohort studies have demonstrated that being a vegetarian reduces the risk of certain cancers and cancer overall. However, there are some limitations to the design of this study which must be considered:
- This study has combined the results of two large cohort studies which assessed diet and then looked at cancer outcomes after several years of follow-up. However, the authors do not appear to have conducted a systematic review of other research in this area. This means that we cannot be sure that that they have examined other relevant trials that might potentially have different results to their own.
- Diet was assessed only once at the beginning of the study. It is not known for how long this dietary pattern had already existed at the time of enrolment (for example, a person could have been vegetarian for weeks or years) or whether this dietary pattern continued during follow-up. Additionally, the self-completed dietary questionnaires, which simply asked whether participants ever ate meat, fish, dairy or eggs, may have led to participants being wrongly categorised into different dietary groups.
- The study examined the risk of a number of cancers, not all of which were found to be significantly linked to diet. While vegetarianism significantly decreased the risk of four types of cancer, these were rare during the follow-up. There were only 49 cases of stomach cancer, 85 of bladder cancer, 140 of ovarian cancer and 257 of blood cancer in the total study group. This means that the absolute risk of this cancer for people of any dietary group is fairly low. Also, calculating a risk reduction by dietary group with such small numbers in each category means that the calculated risk figures may not be precise.
- Statistical adjustments were made to take into account the influence of several lifestyle factors, such as smoking. Again, these were assessed only once and were unlikely to have remained the same throughout follow-up. Each cancer also has a variety of other risk factors, including genetic, medical and lifestyle factors. These were not adjusted for in the risk analyses.
- It is difficult to know when the cancers seen actually developed. While the study found that there was a reduction in overall risk of cancer from being a vegetarian, this was no longer significant once the authors excluded people who had been diagnosed with cancer in the two years following recruitment (i.e. those who they considered may have already been developing cancer when the questionnaire was completed).
- The study’s participants are not necessarily representative of the general population. For example, one-third of participants were vegetarian, 75% of them were women and smoking rates were lower than in the general population.