The EPIC – Oxford study
The EPIC-Oxford study and its contribution to the study of vegetarian health
By Paul Appleby, retired senior statistician
In the early 1990s, scientists at Oxford University began to study the impact of diet on health. This is the largest UK research programme of its kind, involving 65,000 people. Known as the EPICOxford study, it tracks the health of participants and studies the differences in health outcomes of four groups: vegetarians, vegans, pescetarians (fish-eaters) and meat-eaters. The many papers to have come out of the study have found often striking and sometimes surprising differences in risk between the four types of diet. Risks of heart disease and certain types of cancer have been found to be markedly lower in vegetarians and vegans, but positive health effects are also found in relation to a diverse range of other conditions from diabetes to cataracts. In this article, Fellow of the Vegetarian Society and retired senior statistician of the EPIC-Oxford team, Paul Appleby outlines the main findings of the study so far.
Much of the last quarter century of my professional life as a statistician was devoted to studies of the long-term health of vegetarians and vegans. Chief among these was the EPIC-Oxford study. Although cancer was a key focus of our research, it was to be much more than a study of the associations between diet and cancer. The EPIC-Oxford study has produced some of the most compelling findings into the effects of vegetarian and other meat-free diets on the risks of other significant health conditions such as heart disease, diabetes, hypertension and even cataracts.
EPIC-Oxford study design
Participants in the EPIC-Oxford substudy were recruited from 1993 to 1999. GP surgeries took part in the recruitment, and large-scale mail-outs were aimed at recruiting as many vegetarians and vegans as possible. The Vegetarian Society played a large part in helping us to reach and encourage vegetarians to join the study; perhaps if you were a member of the Vegetarian Society in the mid-1990s you will remember the call for volunteer participants, and may have even taken part.
For the purposes of the study, participants were classified into one of four groups based on their dietary habits: meat-eaters, fish-eaters (who ate fish but not meat), vegetarians and vegans.
The table below gives the approximate number of participants recruited to the study by gender and diet group, showing that about one-third were either vegetarian or vegan.
When people were recruited to the study we gathered as much detail as possible about their medical history and lifestyle in terms of smoking, alcohol consumption, physical activity and other factors. One of the most important things in a study of diet and health is to be able to ‘control’ for other factors that affect health. When studying risks for rarer diseases and conditions it was also necessary to combine the diet groups in order to produce clearer results due to the small sample sizes; for example, vegetarians and vegans were frequently combined as “vegetarians”.
At recruitment we recorded ‘baseline characteristics’ such as height, weight and lifestyle including smoking and alcohol drinking and most participants filled out a detailed food frequency questionnaire. Subsequently, many participants provided a blood sample and completed up to three follow-up questionnaires and two 7-day food diaries over the course of the study.
Data from the third follow-up questionnaire were used to compare nutrient intakes by diet group in a paper published in 2016.
The results showed that, on average, all four diet groups easily met recommended protein intakes and that vegetarian diets were generally rich in dietary fibre, vitamin C, folate, omega-6 essential fat, vitamin E and magnesium. We found that vegans had the highest dietary fibre intake and were the only group to, on average, achieve the goal of keeping saturated fat intake below the recommended maximum of 10% of dietary energy. However, vegans had the lowest average calcium intake, with a sizeable proportion consuming less than the recommended 700 mg per day, and average vitamin B12 intake from food in vegans was well below the recommended 1.5 mcg per day. Although half the vegans were taking supplements containing vitamin B12, the data suggest that many vegans risked being deficient in vitamin B12.
Overweight and obesity
Being overweight is a significant cause of ill-health and mortality. In the UK, nearly a quarter of adults and about 10% of children are classified as obese, based on their body mass index (BMI) which is calculated by dividing your weight in kilograms by the square of your height in metres. A healthy BMI is generally considered to be in the range 20-25 kg/m2, with a BMI of 25-30 kg/m2 classified as ‘overweight’, and a BMI of 30 kg/m2 and over classified as ‘obese’. Obesity is rare among EPIC-Oxford participants, but there are noticeable differences in average BMI between the diet groups, with the vegetarians and fish-eaters being slimmer than the meat-eaters (by about 1 kg/m2) and the vegans slimmest of all. Shown below are the results for average BMI by age group, with men on the left and women on the right of the graph.
Cholesterol and blood pressure
A cross-sectional analysis documenting the level of fats present in the blood (lipid concentrations) was published in the European Journal of Clinical Nutrition in 2014. Within a sample of 424 meat-eaters, 425 fish-eaters, 423 vegetarians and 422 vegans, the study found that serum concentrations of both total and non-HDL cholesterol (the bad sort) were lower in vegetarians than in meat-eaters, and lower still in vegans, with smaller differences in HDL cholesterol (the good sort). A small proportion of the observed differences in serum lipid concentrations were explained by differences in BMI, but much of the variation was most likely due to diet, including differences in saturated fat intake between the diet groups.
A healthy blood pressure is key to health and high blood pressure (hypertension) is linked to a range of health problems. A paper published in 2002 found small but statistically significant differences in both systolic and diastolic blood pressure between the diet groups, with meat-eaters having the highest and vegans the lowest average values, reflecting the differences in BMI.
Coronary heart disease and stroke
Being overweight and having raised cholesterol are known risk factors for diseases such as cardiovascular disease and some cancers, but it is only by looking at the incidence of these and other diseases in relation to diet group that we are able to study comparative disease risk in vegetarians and vegans.
A study published in 2019 looked at cardiovascular disease risk in relation to diet group. In the 18 years of followup there were 2,820 cases of ischaemic (coronary) heart disease (caused by a narrowing of the arteries of the heart) and 1,072 cases of stroke (including 519 ischaemic strokes and 300 haemorrhagic strokes). After adjusting for factors such as smoking and exercise, vegetarians (including vegans) had a 22% lower rate of ischaemic heart disease than meateaters. This difference was equivalent to 10 fewer cases of ischaemic heart disease in vegetarians than in meat-eaters per 1000 population over 10 years.
Interestingly and by contrast, vegetarians and vegans had a 20% higher rate of stroke than meat-eaters, equivalent to 3 more cases of stroke per 1000 population over 10 years, mostly due to a higher rate of haemorrhagic stroke; the reason for this is not known, but there is some evidence that, contrary to its effect on ischaemic heart disease, low blood cholesterol may increase the risk of haemorrhagic stroke.
Cancer incidence in British vegetarians was investigated in a 2014 paper published in the American Journal of Clinical Nutrition. For this analysis, data from EPIC-Oxford and an earlier, but quite separate study, called the Oxford Vegetarian Study, were combined in order to study relative cancer incidence in 32,491 meat-eaters, 8,612 fish-eaters and 20,544 vegetarians, including 2,246 vegans. After an average follow-up of 14.9 years there were 4,998 incident cancers.
The findings showed that there was a 12% lower risk of cancer overall (i.e. all malignant cancers combined) in vegetarians and vegans, with significantly lower risks for stomach cancer, bladder cancer and haematological (blood) cancers, such as leukaemia.
The graph below shows relative risks for vegetarians and vegans (combined) compared with meat-eaters for specific types of cancer. In this instance, the relative risk (RR; shown by the black squares and diamond in the graph) compares the risk of cancer in the combined vegetarian and vegan diet group with the risk in the meat-eater diet group.
Values below 1.0 suggest that vegetarians and vegans have a lower risk of being diagnosed with the specified cancer than meat-eaters, with values above 1.0 suggesting that vegetarians and vegans have a higher risk of being diagnosed with that cancer compared to meateaters.
The horizontal lines through the black squares show the 95% confidence interval (95% CI), a measure of the precision of the relative risk; the shorter the line the greater the precision.
The 95% CI is important because the true risk, of which the calculated RR is only an estimate, almost certainly lies within
the 95% CI. So, if the 95% CI does not bisect (i.e. cross) the vertical axis, the RR is considered to be ‘statistically significant’, meaning that it is likely to indicate a real association. For example, the RR for stomach cancer is 0.37 with a 95% CI of 0.19 to 0.69, implying that (because 0.69 is less than 1.00) vegetarians have a statistically significant lower risk of stomach cancer than meateaters.
In contrast, the RR for cancer of the oesophagus for vegetarians is 1.07, which on the face of it, suggests that vegetarians and vegans have a slightly increased risk of developing this type of cancer, compared to meat-eaters. However, with a 95% CI of 0.58 to 1.95 the horizontal line for this cancer does bisect the vertical line, implying that there is no difference in the incidence of this cancer between vegetarians and meat-eaters.
It was also possible to examine risks in vegans alone compared with meateaters for several common cancer sites. Although there was no significant difference in risk for cancers of the colorectum, breast and prostate, the risk of cancer overall (i.e. all malignant cancers combined) was 19% lower (95% CI: 2% to 34% lower) in vegans compared with meat-eaters.
A 2019 paper looked at diabetes risk by diet group. Over an average 17.6 years of follow-up, there were 1,224 incident cases of diabetes. When compared with regular meat-eaters (defined as participants eating at least 50 grams of meat per day), other diet groups including low meat-eaters (less than 50 grams of meat per day), fish-eaters, and vegetarians (including vegans), all had a lower risk of developing diabetes.
More specifically, the risk of developing diabetes for the combined vegetarian and vegan diet groups were 37% (95% CI: 26% to 46%) lower, while for the vegan diet group alone the risk was 47% (95% CI: 21% to 64%) lower.
It is important to note that being overweight is a major risk factor for diabetes, and much of the reduction in risk in vegetarians and vegans could be attributed to their lower BMI when compared with regular meat-eaters.
In a paper published in November 2020, fracture risk was compared across the four diet groups, controlling for a wide variety of factors including exercise and BMI. Compared with meat-eaters, risks of hip fracture were 25% higher in vegetarians, and more than twice as high in vegans, equivalent to 29 (95% CI: 9 to 52) and 149 (95% CI: 79 to 245) more cases per 10,000 people over 10 years, respectively.
The vegans also had a 43% higher risk (95% CI: 20% to 70% higher) of fractures overall, including twice the risk of leg fracture (RR 2.05; 95% CI: 1.23 to 3.41) and a 59% higher risk of fracture at a combination of other main sites (collarbone, rib or vertebra; RR 1.59; 95% CI: 1.02-2.50), compared with meat-eaters.
The associations described above remained statistically significant after additional adjustment for dietary intake of calcium and protein, two nutrients known to be important for bone health, so the findings were hard to explain.
Interestingly, no significant differences in risk by diet group were observed for wrist or ankle fractures, nor for fractures of the arm after adjusting for BMI.
A paper published in 2011 looked at the association between diet and diverticular disease, a common condition in Western countries that has been linked to a low-fibre diet. In this study of 47,033 participants, including 812 diverticular disease cases, vegetarians (including vegans) had a 31% (95% CI: 14% to 45%) lower risk of developing diverticular disease compared with
meat-eaters, a reduction that remained significant after controlling for dietary fibre intake.
A surprising finding from another paper published in 2011 showed that vegetarians and vegans combined had a 26% (95% CI: 14% to 37%) lower risk of eye cataract compared with meat-eaters.
A 2015 paper looked at mortality (death rate) before age 90 by diet group based on a total of 5,294 deaths. Regular meat-eaters (participants eating at least 50 grams of meat per day) were used as the reference group against which we compared mortality in the combined vegetarian and vegan diet group.
There was no difference in overall (allcause) mortality between the vegetarian and vegan groups when compared with
regular meat-eaters (RR 1.02; 95% CI: 0.94 to 1.10). However, when we repeated the analysis and excluded the data for participants known to have changed diet group during followup, all-cause mortality was 8% lower in vegetarians and vegans combined compared with regular meat-eaters (95% CI: 1% to 16% lower), This gave us a clearer picture of long-term diet based on data for 4,270 deaths before age 90, as shown in the graph below.
In addition, vegetarians and vegans had lower death rates for all malignant cancers combined, pancreatic cancer and
haematological (blood) cancer, compared with regular meat-eaters. It was also possible to investigate mortality in vegans alone compared with regular meat-eaters for six major causes of death. However, there were no significant differences in mortality between the two diet groups, perhaps because of the relatively small number of vegans in the study (2,228, including 166 deaths).
Conclusions and acknowledgements
Based on findings from the EPICOxford study, vegetarians and vegans may enjoy a number of health benefits compared with meat-eaters even when other variables in lifestyle, such as exercise and not smoking, are adjusted for. These include, having a lower BMI (by about 1 kg/m2), slightly lower blood pressure, lower serum cholesterol (by about 0.5 mmol/L), a lower incidence of ischaemic heart disease, lower overall cancer incidence and lower rates of type-2 diabetes, diverticular disease and eye cataract. They also have low mortality compared with the general population (but similar to that of comparable nonvegetarians).
However, it’s not all good news. Vegetarians and vegans appear to have a higher risk of stroke and, especially in vegans, a higher risk of bone fracture, particularly fractures of the hip. Therefore, a vegetarian or vegan diet is not a panacea. Clearly, further research is warranted to establish the reasons for these increased risks and whether they can be avoided without compromising dietary principles.
Nevertheless, in a position paper on vegetarian diets published in 2016, the US Academy of Nutrition and Dietetics declared that “appropriately planned vegetarian, including vegan, diets are healthful, nutritionally adequate, and may provide health benefits for the prevention and treatment of certain diseases.”
Thanks are owing to all participants in EPIC-Oxford, and to my former colleagues at the Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford.
A full list of papers arising from the EPIC-Oxford study can be found on the study website: www.epic-oxford.org.
Published papers mentioned in the above article are:
Sobiecki JG, Appleby PN, Bradbury KE, Key TJ, High compliance with dietary recommendations in a cohort of meat eaters, fish eaters, vegetarians, and vegans: results from the European Prospective Investigation into Cancer and Nutrition–Oxford Study, Nutrition Research 2016;36:464-477.
Bradbury KE, Crowe FL, Appleby PN, Schmidt JA, Travis RC, Key TJ, Serum concentrations of cholesterol, apolipoprotein A-I, and apolipoprotein B in a total of 1694 meat-eaters, fish-eaters, vegetarians, and vegans. European Journal of Clinical Nutrition 2014;68:178-83
Appleby PN, Davey GK, Key TJ, Hypertension and blood pressure among meat eaters, fish eaters, vegetarians and vegans in EPIC-Oxford. Public Health Nutrition 2002; 5: 645-654.
Tong TY, Appleby PN, Bradbury KE, Perez-Cornago A, Travis RC, Clarke R, et al. Risks of ischaemic heart disease and stroke in meat eaters, fish eaters, and vegetarians over 18 years: results from the prospective EPIC-Oxford study. BMJ 2019;366: l4897. doi: 10.1136/bmj.l4897
Key TJ, Appleby PN, Crowe FL, Bradbury KE, Schmidt JA, Travis RC, Cancer in British vegetarians: updated analyses of 4998 incident cancers in a cohort of 32,491 meat-eaters, 8612 fisheaters, 18,298 vegetarians and 2246 vegans. American Journal of Clinical Nutrition 2014; doi:10.3945/ajcn.113.071266.
Papier K, Appleby PN, Fensom GK, Knuppel A, Perez-Cornago A, Schmidt JA, et al. Vegetarian diets and risk of hospitalisation or death with diabetes in British adults: results from the EPIC-Oxford study. Nutrition and Diabetics 2019; 9(7). doi: 10.1038/s41387-019-0074-0
Tong TYN, Appleby PN, Armstrong MEG, Fensom GK, Knuppel A, Papier K, et al. Vegetarian and vegan diets and risks of total and site-specific fractures: results from the prospective EPIC-Oxford study. BMC Medicine 2020;18:353-367. doi:10.1186//s12916-020-01815-3.
Crowe FL, Appleby PN, Allen NE, Key TJ. Diet and risk of diverticular disease in Oxford cohort of European Prospective Investigation into Cancer and Nutrition (EPIC): prospective study of British vegetarians and non-vegetarians. British Medical Journal 2011; 343: d4131 doi: 10.1136/bmj.d4131.