Public Health Nutrition: 7(7), 871–878 DOI: 10.

1079/PHN2004620

The UK Women’s Cohort Study: comparison of vegetarians,

fish-eaters and meat-eaters
JE Cade*, VJ Burley, DC Greenwood and the UK Women’s Cohort Study Steering Group†
Nutritional Epidemiology Group, University of Leeds, 71 – 75 Clarendon Road, Leeds LS2 9PL, UK

Submitted 31 July 2003: Accepted 11 March 2004

Abstract
Background: This paper describes the development of the UK Women’s Cohort Study
and presents cohort baseline characteristics.
Methods: In total, 35 372 women, aged 35– 69 years at recruitment, were selected to
ensure a wide range of dietary intakes. Diet was assessed by a 217-item food-
frequency questionnaire (FFQ). Detailed lifestyle information was collected by postal
questionnaire. Vegetarians, fish-eaters and meat-eaters were compared.
Results: The cohort women are mainly white, well-educated, middle-class and
married with children. They are health-conscious with only 11% current smokers and
58% taking dietary supplements. Twenty-eight per cent of subjects self-report as
being vegetarian and 1% as vegan. However, only 18% are defined as ‘vegetarian’
from the FFQ. Fat provides 32% of energy; vitamin and mineral intakes are high, with
a broad range of intakes. Meat-eaters are older, with a higher body mass index (BMI)
and the lowest intakes of carbohydrate, fibre, vitamin C, folate, iron and calcium.
Other fish-eaters are similar to vegetarians. Vegetarians have the lowest intakes of
protein, fat and saturated fat. Oily fish-eaters have the lowest BMI; are the least likely
to smoke or use full-fat milk; and are the most likely to use dietary supplements and
consume the most fruit and vegetables. Oily fish-eaters have the highest total energy
intake and vegetarians the lowest. Semi-skimmed milk, bread, potatoes, wine, Keywords
Cohort
bananas and muesli are important contributors to energy for all groups.
Nutrition
Conclusion: A large cohort of middle-aged women has been created encompassing a Food intake
wide range of different eating patterns, including diets currently of interest to research Vegetarian
into protection against cancer and coronary heart disease. Participants will be Fish intake
followed up to study the effects of different food and nutrient intakes on long-term Women
health outcomes. Diet

Over recent years, a number of cohort studies have been due in part to the fact that diet is a complex exposure with
carried out to explore diet and disease relationships1 – 8. measurement being subject to a range of errors and bias.
The UK Women’s Cohort Study (UKWCS) is one of these, There are many different methods of assessing diet, each
and was started in 1993 to investigate diet and cancer with its own shortcomings12. Also, within any single
relationships in a group of women in the United Kingdom. population group, diet often appears too homogeneous to
Through the selection procedure used, the study was be able to detect subtle effects of dietary differences.
designed to optimise exposure to dietary factors of One of the key questions that needs to be addressed is
interest. While our current knowledge on diet and cancer ‘Which dietary pattern makes us live the longest?’13.
is sufficient to make some broad recommendations9, many Cohort studies have often not addressed or have been
important questions remain unanswered on the links unable to assess the risks associated with a particular
between diet and cancer. dietary pattern. One recent collaborative analysis of five
Nutritional epidemiology has often produced conflicting large prospective studies with a high proportion of people
results when exploring diet and chronic disease10,11. This is who defined themselves as ‘vegetarian’ showed no
protective effect on cancer mortality of this type of dietary
pattern14. However, there are many different definitions of
†The UK Women’s Cohort Study Steering Group: Professor Rhys the term ‘vegetarian’15,16, making interpretation of these
Williams, University of Wales; Professor David Forman, University of
Leeds; Professor Margaret Thorogood, University of Warwick;
results problematic.
Dr Jennifer Barrett, University of Leeds; Dr Barrie Margetts, The design of the UKWCS is described in this paper with
University of Southampton. an emphasis on characterisation of the subjects in the

*Corresponding author: Email [email protected] q The Authors 2004

872 JE Cade et al.
cohort. Vegetarians, fish-eaters and meat-eaters are able to detect fairly small differences in total cancer
compared. registrations. For site-specific cancers, the difference in
risk would obviously have to be more marked to be
Methods detectable with near certainty. Current projections,
allowing for the substantial healthy participant effect,
The UKWCS has been established to study the relation- staggered entry and time lag, estimate that, by the end of
ships between diet and cancer incidence and mortality 2006, we will have 1600 incident cancer cases identified of
(from selected causes) in a group of middle-aged women which 550 will be breast cancer and 1500 deaths from all
in the UK. Specifically: to establish a cohort with a wide causes within the cohort. This sample size should have
range of exposure to nutrients of interest, including total approximately 80% power to detect a protective effect of a
energy, types of fat, fibre and antioxidant nutrients; to test vegetarian diet with an odds ratio of 0.8 for either
the hypothesis that eating a vegetarian diet is protective endpoint, dependent on definition of vegetarian.
against ill health; and to study specific diet–disease
research questions in the cohort. Baseline data collection
One hundred and seventy-four local research ethics
Cohort sample committees were contacted and permission to carry out
Subjects in the cohort were selected to ensure that there the baseline study was obtained19. Baseline data were
was a wide range of dietary patterns represented. The collected between 1995 and 1998 via a postal ques-
cohort was constructed to have similar, large numbers of tionnaire to each subject. The self-administered ques-
subjects in three main groups: vegetarian, eating fish (not tionnaires consisted of a detailed assessment of diet using
meat) and meat-eaters. This ensured higher power for a food-frequency questionnaire (FFQ) developed from the
important comparisons involving fruit, vegetables or fish one used in the Oxford arm of the European Prospective
intake as well as associated nutrients that may be the focus Investigation into Cancer and Nutrition (EPIC) study20. To
of research into potential relationships between diet and ensure that foods frequently consumed by vegetarians
cancer. were included, a pilot study was undertaken in June 1994
The cohort was taken from responders to the World on the sample of 71 vegetarian women in the cohort living
Cancer Research Fund’s (WCRF) direct mail survey. This in Southampton. An FFQ and a 7-day weighed food diary
included people living in England, Wales and Scotland and were completed by these women. As a result, more
used direct mail lists, targeted towards females, with an vegetable-based composite dishes were added to the food
overall response rate of 17%. This identified about 16 000 list21. The resulting FFQ consists of a list of 217 foods with
self-reported vegetarians and a similar number of other 10 pre-coded classifications of frequency of consumption
non red-meat-eaters aged 35 –69 years, out of a total of ranging from never to 6 or more times per day22. The FFQ
500 000 responders. Eighty-five per cent of the responders has recently been validated on a sub-sample of 303 cohort
were women. Seventy-five per cent of the responders to subjects. Nutrient values from the FFQ were compared
the original survey indicated that they would be willing to with values from a 4-day food diary and also fasting blood
participate in a more detailed survey. These women measures of specific nutrients23. Results showed highly
formed the population to be contacted to become part of significant correlations between all diary and FFQ
the UK Women’s Cohort. nutrients. In addition, there were significant correlations
Women aged 35 –69 years at completion of the original between plasma and FFQ measures for vitamin B12,
mail survey were eligible for inclusion. All of the vitamin C and calcium.
vegetarians and the non red-meat-eaters were invited to Although 28% of subjects defined themselves as
take part in the study. A comparison group was selected vegetarian, in this analysis we used a more accurate
from the remaining eligible women by selecting, for each definition of ‘vegetarian’ from the FFQ data. Women who
vegetarian, the next non-vegetarian in the list aged within said that they ate meat or fish less than once a week were
10 years of the vegetarian. Further women were recruited classified as being vegetarians. This pragmatic definition
from responders to the baseline data collection, who were reflects a realistic vegetarian diet with very low meat or fish
asked to identify friends and relatives of a similar age consumption. Other groups defined were: meat-eaters,
group who were vegetarians and meat-eaters. who ate meat once a week or more; oily fish-eaters, who
ate oily fish 2–4 times per week and meat less than once a
Sample size week; and other fish-eaters, who ate fish once a week or
A sample from the WCRF initial mail survey was used to more and meat less than once a week (oily fish less than
estimate numbers available for the main cohort. Original 2–4 times per week).
sample size estimates were based on data from 1989, the Analysis of the FFQ generated nutrient intakes. This was
most recent cancer registration statistics available at the done by applying standard values from McCance &
time17,18. Expected figures over a 1-year follow-up period Widdowson’s The Composition of Foods (5th edition)24 and
showed that a total of around 30 000 subjects would be associated supplements for nutrients for each food item on
UKWCS: vegetarians, fish- and meat-eaters 873
the FFQ. This was then multiplied by the frequency of subject and her family. A large database of ingredients of
consumption of the item, expressed per day. The sum of dietary supplements that were used by the cohort has
these items produced the total nutrient intake. Type and been created. The need for such as database was noted in
amount of milk and amount of sugar consumed per day a recent paper on potential harm caused by sup-
were generated from separate, more detailed questions, plements25. Since we are still in the process of contacting
since these common items are not well characterised non-responders to this phase of the study, all results
through an FFQ. Vitamin and mineral intakes reported are presented here relate solely to the baseline data collection
from food only from the FFQ and do not include phase. Results are presented for the sample as a whole and
supplement intakes. To explore which foods were also split according to whether subjects were classified as
important sources of energy, the energy contribution meat-eaters, oily fish-eaters, other fish-eaters or
from each of the 217 items was ranked. (The full list of vegetarians.
foods included in the FFQ can be obtained from the
corresponding author.) Results
A lifestyle questionnaire, to identify other possible
confounders and provide information to allow registration Response
with the National Health Service (NHS) Central Register, A total of 35 372 women returned the baseline ques-
was also included. All questionnaire information was tionnaire (a response rate of 58% from the 61 000 subjects
double-entered by hand to reduce error. All subjects who who were mailed). Each subject has provided information
returned completed questionnaires were included in the on approximately 600 variables.
analysis. As of January 2002, we had 34 998 (99%) subjects traced
on the NHS central register. In addition to this, we are
Mortality and morbidity data following up a number of cases that require further
All subjects were submitted to the Office of National information in order to be traced successfully. So far we
Statistics to be flagged on the NHS central register using have had a total of 876 deaths (March 2004) and 1361
their NHS number, full name and date of birth where incident cancer registrations, including 65 cases of
possible. Deaths and cancer registrations for the cohort are colorectal cancer and 440 cases of breast cancer.
being recorded. Cause of death is coded according to the
International Classification of Diseases 9 and 10. Cancers Cohort baseline characteristics
registered after a subject returned their questionnaire were Baseline characteristics of the cohort are presented in
taken as newly incident cancers. Prevalent cases were also Table 1. The mean (standard deviation (SD)) age of the
notified. cohort subjects was 52 (9) years at baseline. The majority
of the women were white (99%) and middle-class (63%
Phase 2 data collection NS-SEC class 1 – Professional and Managerial26,27).
Contacting all subjects for a second time began in April Seventy-five per cent of the women were married or
1999 and finished in 2002. This phase of data collection living as married, and 86% had children. The cohort was
consisted of a 4-day food and drink weighed diary, a 1-day well-educated in general, with 27% having degree-level
physical activity diary and a questionnaire covering many qualifications and over half were currently in employment.
aspects of both nutrition and the medical history of the The cohort women live in all areas of England, Scotland

Table 1 Characteristics of the total sample, meat-eaters, oily fish-eaters, other fish-eaters and vegetarians

Total sample Meat-eaters Oily fish-eaters Other fish-eaters Vegetarians

(n ¼ 35 372) (n ¼ 24 738) (n ¼ 870) (n ¼ 3286) (n ¼ 6478)

% of total sample 70 3 9 18
Age (years), mean (SD) 52 (9) 54 (9) 51 (9) 50 (9) 49 (8)
BMI (kg m22), mean (SD) 24.5 (4.3) 25.0 (4.5) 23.2 (3.6) 23.3 (3.5) 23.3 (3.9)
Degree-level education (%) 27 23 32 37 37
Lives in Greater London (%) 12 11 16 16 14
Smokes daily (%) 8 9 6 7 7
Alcohol .1/week (%) 52 54 52 55 45
Full-fat milk (%) 12 13 6 10 10
Supplements (%) 58 55 73 67 62
Self-reported illness (%)
Heart attack 1.5 1.8 1.7 0.8 0.8
Angina 2.2 2.6 1.6 1.4 1.1
High blood pressure 17.3 19.6 14.0 13.0 11.4
High cholesterol/lipids 7.6 8.9 6.7 5.7 4.0
Cancer 7.5 8.2 6.7 7.1 5.2

SD – standard deviation; BMI – body mass index.

874 JE Cade et al.
and Wales, with one-third living in the South and East of the least and meat-eaters to have the most self-reported
England. The mean (SD) body mass index (BMI) of the illness. The other fish-eaters were rather similar to the
women was 24.5 (4.3) kg m22. Only 11% of the cohort vegetarians. The oily fish-eaters differed in that they had
were current smokers. In terms of self-reported illness, 8% the lowest BMI, were least likely to smoke or use full-fat
said that they had or had ever had cancer, 2% diabetes, milk, and were most likely to use dietary supplements.
17% high blood pressure and 8% high cholesterol or The meat-eaters reported the lowest number of servings of
hyperlipidaemia. fruit/dishes (10) and vegetable/dishes (11) per week and
the oily fish group reported the highest number (12 and
Cohort food and nutrient characteristics 13, respectively) from additional cross-check questions on
Eighteen per cent (6478) of the women were vegetarian the FFQ.
based on the FFQ. Fifty-eight per cent of the women An exploration of nutrient intake according to these
reported taking dietary supplements and 27% reported groupings also revealed differences (Table 2). The oily
changing their diet in the past 12 months. In general, meat fish-eaters had the highest total energy intakes and the
consumption was rather low in the cohort at a mean of 5 vegetarians had the lowest. This tended to result in higher
portions per week and fish at 2 portions per week. nutrient intakes for the oily fish consumers than for the
Nutrient intakes of the cohort are presented in Table 2. other groups. Percentage energy from protein and fat were
The mean energy intake was 2361 kcal (median 2261 highest in the meat-eaters and percentage energy from
kcal). Fat, carbohydrate and protein provided 32%, 53% carbohydrate was highest in the vegetarians. The meat-
and 15% of energy, respectively. Vitamin and mineral eaters had the lowest intakes of carbohydrate and
intakes were high as illustrated by the mean vitamin C percentage energy from carbohydrate, fibre, vitamin C,
intake (172 mg, median 156 mg). folate, iron and calcium, but the highest intakes of vitamin
A and zinc. The vegetarians had the lowest absolute
Differences by dietary pattern protein, fat and saturated fat intakes, although percentage
To explore variation in the cohort, subjects were classified of energy from fat was lowest among the oily fish
according to whether they were meat-eaters, oily fish- consumers. When the analysis was repeated using nutrient
eaters, other fish-eaters or vegetarians, which resulted in density (i.e. amount of nutrient per 1000 kcal) to adjust for
differences emerging in a range of characteristics. Results the higher energy intakes of the oily fish consumers,
of significance tests of the differences are not shown differences between the groups still remained. For
because the sample size is so big that even very small example, oily fish consumers still had higher intakes of
differences are ‘statistically significant’. Table 1 shows that dietary fibre, vitamin C and folate.
the meat-eaters were slightly older, had a higher BMI, were Table 3 summarises the top 10 foods, from the 217 items
less likely to have a degree, live in Greater London or use on the FFQ, that contributed to energy for each group, to
dietary supplements, and were more likely to smoke daily indicate differences between the groups in relative
and use full-fat milk than the other groups. The proportions of a number of important food items. For all
vegetarians were younger and less likely to have alcohol groups, the top 10 items contributed at least one-quarter of
more than once a week. Self-reported illness also differed their daily energy intake. Semi-skimmed milk, bread,
by group. There was a tendency for vegetarians to have potatoes, wine, bananas and muesli were important

Table 2 Nutrient intakes for the total sample and by meat-, fish-eating and vegetarian status. Data are expressed as
mean (standard deviation)

Total sample Meat-eaters Oily fish-eaters Other fish-eaters Vegetarians

Calories including alcohol 2361 (801) 2370 (810) 2552 (866) 2350 (748) 2303 (773)
Protein (g) 90 (32) 95 (33) 96 (31) 81 (25) 75 (25)
% energy from protein 15.1 (2.5) 15.7 (2.4) 15.2 (2.2) 13.8 (2.0) 13.1 (2.0)
Carbohydrate (g) 315 (113) 310 (112) 348 (127) 324 (111) 323 (115)
% energy from carbohydrate 52.6 (7.0) 51.5 (6.7) 54.2 (6.7) 54.7 (6.7) 55.7 (7.0)
Fat (g) 85 (36) 86 (37) 90 (38) 84 (34) 83 (35)
% energy from fat 32.4 (5.8) 32.6 (5.6) 31.4 (6.2) 31.9 (5.9) 32.0 (6.3)
Saturated fat (g) 29.5 (14.4) 31 (15) 28 (14) 28 (14) 27 (14)
Polyunsaturated fat (g) 16.4 (7.7) 16 (7) 20 (9) 18 (8) 18 (9)
Monounsaturated fat (g) 27.8 (12.4) 28 (12) 29 (13) 27 (12) 26 (12)
Fibre (Englyst) (g) 26 (11) 24 (11) 32 (13) 28 (11) 29 (12)
Vitamin C (mg) 172 (92) 168 (89) 212 (113) 178 (89) 179 (99)
Folate (mg) 404 (146) 397 (142) 465 (176) 416 (143) 416 (154)
Vitamin A (mg) 1249 (633) 1318 (673) 1216 (587) 1077 (457) 1075 (496)
Iron (mg) 18.9 (8.1) 18.8 (8.1) 21.8 (9.1) 19.2 (8.1) 18.9 (8.1)
Calcium (mg) 1141 (411) 1133 (403) 1280 (491) 1182 (409) 1134 (423)
Zinc (mg) 11.5 (4.3) 12.0 (4.4) 11.8 (4.4) 10.4 (3.5) 10.2 (3.7)
UKWCS: vegetarians, fish- and meat-eaters 875
Table 3 Top 10 contributors to energy by meat-, fish-eating and vegetarian status (% contri-
bution to total energy)

Food Meat-eaters Oily fish-eaters Other fish-eaters Vegetarians

Semi-skimmed milk 3.7 2.5 3.1 2.7

Wholemeal bread 3.4 5.4 5.0 5.4
Boiled potato 3.2 2.2 2.4 2.4
White bread 2.7 – – 1.8
Jacket potato 2.2 2.7 2.5 2.7
Wine 2.1 2.1 2.3 1.9
Banana 2.2 2.6 2.5 2.6
Brown bread 1.9 – 1.8 –
Muesli 1.7 2.1 2.3 2.2
Mini chocolate bars 1.7 – – –
Oily fish – 3.4 – –
Apples – 2.2 2.0 2.1
Porridge – 1.8 – –
Cheddar cheese – – 2.3 2.4

Where cells are empty does not imply that the food was not eaten, rather that the food was not featured in
the top 10 energy-contributing foods for that group.

contributors to energy for each group. However, there difference or a selection effect – where vegetarians who
were some differences between the groups. Oily fish was are unhealthy are less willing to volunteer for a study such
the second most important contributor to energy in the as this, or it could be due to different uses of health
oily fish group. Mini chocolate bars featured in the top 10 services leading to different likelihood of diagnosis
foods for meat-eaters but not in any of the other groups. between the groups35.
Apples appeared in the top 10 foods for all groups except As anticipated, about one-third of the cohort self-
the meat-eaters. Cheese was in the top 10 for other fish- defined themselves to be vegetarians. However, only 18%
eaters and vegetarians only. Porridge was in the top 10 were defined as vegetarians from the FFQ and ate meat or
foods for the oily-fish-eating group alone. fish less than once a week. The recent National Diet and
Nutrition Survey of British Adults (NDNS) has shown that
Discussion 4% of women aged 50 –64 years report being vegetarian or
vegan36. The definition of terms is important, to ensure
A large cohort of middle-aged British women has been that similar groups and dietary patterns are being
created, with detailed dietary and lifestyle information. compared37. In this current analysis, vegans are included
This is one of the largest cohorts looking at diet and cancer within the definitions of vegetarian used. Future work will
in women in the UK. The EPIC study has much larger differentiate these particular groups. Aspects of the
numbers across the whole of Europe28. The Oxford– EPIC protective elements of the vegetarian diet in Western
cohort also has a large number of non meat-eaters with populations have been widely studied. There is still a need
rather similar characteristics to the UK Women’s Cohort29. to elucidate why a vegetarian diet tends to be healthier
Other cohorts with an emphasis on vegetarian diets are than a non-vegetarian diet. Three main routes have been
generally smaller30,31. suggested: vegetarians have healthier lifestyles in other
This cohort is not representative of the British female ways (such as lower smoking); vegetarians eat less of
population, nor indeed was it ever intended to be. Rather, potentially harmful components in the diet; and they eat
it was designed to optimise power for relevant compari- more potentially beneficial items38. The UK Women’s
sons relating to intake of fruit and vegetables and Cohort will be able to explore these issues in detail and
associated nutrients and their impact on the development relate them to cancer incidence rates and mortality from all
of cancer and coronary heart disease. Other cohorts have and specific causes.
been designed with a wide range of nutrient intakes where Subjects in the UK Women’s Cohort generally follow a
the unrepresentative nature of the cohort is seen as a healthier lifestyle than average. They have lower rates of
strength32,33. smoking and higher intakes of fruit and vegetables
In comparison with women of a similar age from the recorded on the FFQ than the average British woman.
most recent British census (2001), the women in the Despite this, it has been possible to show interesting and
UKWCS are more likely to be married, have a high social potentially important differences in food and nutrient
class, have a degree and to live in the South and East of intake within the cohort that may have implications for
England34. Differences between the groups were apparent long-term health outcomes. Women in this cohort are
at baseline in terms of self-reported illness, with meat- more likely to be taking dietary supplements than are
eaters appearing to have most and vegetarians the least women of a similar age from the recent NDNS. In the
amount of pre-existing illness. This may be due to a real NDNS, 55% took supplements36; this is the same as for the
876 JE Cade et al.
meat-eaters in our cohort. In addition, nutrient values information. There is potential to link the data from this
varied between the previous NDNS adults39 and the UK cohort with others within the UK to provide even larger
Women’s Cohort. In particular, energy intakes overall and numbers of cases. On the other hand, the cohort does
hence micronutrient levels in general are higher in this not represent a random sample of the British middle-
cohort. Percentage energy from carbohydrate is higher aged female population in that they tend to be healthier
and percentage energy from fat is lower in this cohort. In than usual and there are known weaknesses in the FFQ
part, this will be due to the different methodology used to methodology51.
assess diet. The NDNS used a 7-day weighed intake with In conclusion, a large cohort of middle-aged British
the potential for underreporting compared with our use of women has been created which includes a wide range of
an FFQ. Also, survey fieldwork for the first NDNS adults different dietary patterns. Two distinct groups of fish-
was undertaken in 1986 and so dietary patterns may have eaters, oily fish-eaters and other fish-eaters, have been
changed. described. In addition, vegetarians and meat-eaters have
Splitting the cohort by dietary patterns, according to the also been characterised. Participants will be followed up
amount of meat, oily fish and other fish consumed, to explore the effects of different food and nutrient intakes
generated some interesting differences40. Fish-eaters were and food patterns on long-term health outcomes.
split into two groups since oily fish contains high levels of
n– 3 fatty acids which may be beneficial. In our cohort, we Acknowledgements
found that the meat-eaters were more likely to be older
and less well-educated than the others. These character- The UK Women’s Cohort Study is funded by the World
istics were also seen in subjects who were most likely to Cancer Research Fund.
eat meat in the East Anglian EPIC cohort41. At an Special thanks go to James Thomas for his work on the
international level, these characteristics do not necessarily database. Thanks to Clare Calvert and Alyson Greenhalgh
hold out; for example, a large survey of Norwegian for previous cohort management and to all the nutrition
women found that it was the older women who ate more students and others who have contributed to the data
fish42. It is important therefore to have nationally relevant collection process: Amanda Woodhouse, Rosie Dennison,
data for this and other dietary patterns that can be Karen Lawson, Cheryl Golding, Jennie Pollard, Clare
identified13. By careful definition of dietary patterns the Jaggers, Sinead Boylan, Aine McConnon, Alison Long,
high consumers of oily fish have been shown to be Michelle Spence, Emma Elliott, Gavin McArt, Ula Nur,
different from the other fish-eaters and the vegetarians, Caroline Owen, Briege McNulty, Barbara Bailey, Carole
who were more similar to each other. For example, the Burton and Cathy Britten. We are indebted to all the
oily fish-eaters not only had the highest fat intakes but also women in the cohort for their time and trouble in
the highest fibre and vitamin C intakes. completing the forms.
In terms of future health risk, a fish-eating dietary
pattern may be important. More good-quality epidemio- References
logical evidence is needed to further characterise the links
between fish intake and risk of heart disease or cancer. 1 Ascherio A, Rimm EB, Giovannucci EL, Spiegelman D,
The epidemiological literature surrounding potential Stampfer M, Willett WC. Dietary fat and risk of coronary
beneficial effects of fish consumption on coronary heart heart disease in men: cohort follow up study in the United
States. British Medical Journal 1996; 313: 84– 90.
disease is confused43 – 45. A systematic review of 11 cohort
2 Bandera EV, Freudenheim JL, Marshall JR, Zielezny M, Priore
studies concluded that fish consumption was not RL, Brasure J, et al. Diet and alcohol consumption and lung
associated with reduced coronary heart disease mortality cancer risk in the New York State Cohort (United States).
in the cohorts from low-risk populations. However, fish Cancer Causes & Control 1997; 8: 828– 40.
3 Elmstahl S, Holmqvist O, Gullberg B, Johansson U, Berglund
consumption at 40 – 60 g daily was associated with G. Dietary patterns in high and low consumers of meat in a
markedly reduced coronary heart disease mortality in the Swedish cohort study. Appetite 1999; 32: 191– 206.
rather smaller cohort studies in high-risk populations. The 4 Goldbohm RA, van den Brandt PA, van’t Veer P, Brants HA,
underlying biochemical mechanism is not fully known Dorant E, Sturmans F, et al. A prospective cohort study on
the relation between meat consumption and the risk of
and causal inference is premature46. The evidence colon cancer. Cancer Research 1994; 54: 718–23.
surrounding fish consumption and cancer risk is, if 5 Key TJ, Thorogood M, Appleby PN, Burr ML. Dietary habits
anything, even less clear. Ecological and case– control and mortality in 11,000 vegetarians and health conscious
studies have shown a possible protective effect of higher people: results of a 17 year follow up. British Medical
Journal 1996; 313: 775–9.
fish consumption against breast cancer47 – 49, whereas 6 Mills PK, Beeson WL, Phillips RL, Fraser GE. Bladder cancer
others have not50. in a low risk population: results from the Adventist Health
The UK Women’s Cohort has particular strengths Study. American Journal of Epidemiology 1991; 133: 230– 9.
related to the broad spectrum of dietary patterns 7 Potter JD, Sellers TA, Folsom AR, McGovern PG. Alcohol,
beer, and lung cancer in postmenopausal women. The Iowa
represented. It has detailed FFQ and subsequent Women’s Health Study. Annals of Epidemiology 1992; 2:
food diary information along with extensive lifestyle 587– 95.
UKWCS: vegetarians, fish- and meat-eaters 877
8 Riboli E. Nutrition and cancer: background and rationale Social Class. Colchester: Institute for Social and Economic
of the European Prospective Investigation into Cancer Research, University of Essex, 2001.
and Nutrition (EPIC). Annals of Oncology 1992; 3: 27 Bravo Y, Greenwood DC, Cade JE. The impact of social class
783– 91. on a healthy diet: analysis from the UK Women’s
9 Department of Health. Nutritional Aspects of the Develop- Cohort Study. Proceedings of the Nutrition Society 2002;
ment of Cancer. Report of the Working Group on Diet and 61: 142A.
Cancer of the Committee on Medical Aspects of Food and 28 Riboli E, Hunt KJ, Slimani N, Ferrari P, Norat T, Fahey M, et al.
Nutrition Policy. Report on Health and Social Subjects No. European Prospective Investigation into Cancer and Nutri-
48. London: The Stationery Office, 1998. tion (EPIC): study populations and data collection. Public
10 Kaaks R, Riboli E. The role of multi-centre cohort studies in Health Nutrition 2002; 5: 1113 – 24.
studying the relation between diet and cancer. Cancer Letters 29 Davey GK, Spencer EA, Appleby PN, Allen NE, Knox KH,
1997; 114: 263 –70. Key TJ. EPIC– Oxford: lifestyle characteristics and nutrient
11 Hankin JH. Role of nutrition in women’s health: diet and intakes in a cohort of 33 883 meat-eaters and 31 546 non
breast cancer. Journal of the American Dietetic Association meat-eaters in the UK. Public Health Nutrition 2003; 6:
1993; 93: 994– 9. 259–68.
12 Margetts BM, Nelson M. Design Concepts in Nutritional 30 Frentzel-Beyme R, Chang-Claude J. Vegetarian diets and
Epidemiology. Oxford: Oxford University Press, 1997. colon cancer: the German experience. American Journal of
13 Greenwood DC, Cade JE, Draper A, Barrett JH, Calvert C, Clinical Nutrition 1994; 59: 1143S– 52S.
Greenhalgh A. Seven unique food consumption 31 Fraser GE. Associations between diet and cancer, ischemic
patterns identified among women in the UK Women’s heart disease, and all-cause mortality in non-Hispanic white
Cohort Study. European Journal of Clinical Nutrition 2000; California Seventh-day Adventists. American Journal of
54: 314– 20. Clinical Nutrition 1999; 70: 532S –8S.
14 Key TJ, Fraser GE, Thorogood M, Appleby PN, Beral V, 32 Kaaks R, Riboli E. Validation and calibration of dietary intake
Reeves G, et al. Mortality in vegetarians and nonvegetarians: measurements in the EPIC project: methodological con-
detailed findings from a collaborative analysis of 5 siderations. International Journal of Epidemiology 1997; 26:
prospective studies. American Journal of Clinical Nutrition S15– 25.
1999; 70: 516S – 24S. 33 Schatzkin A, Subar A, Thompson FE, Harlan LC, Tangrea J,
15 Draper A, Lewis J, Malhotra N, Wheeler E. The energy and Hollenbeck AR, et al. Design and serendipity in establishing
nutrient intakes of different types of vegetarian: a case for a large cohort with wide dietary intake distributions – the
supplements? [published erratum appears in British Journal National Institutes of Health – American Association of
of Nutrition 1993; 70(3): 812] British Journal of Nutrition Retired Persons Diet and Health Study. American Journal
1993; 69: 3 – 19. of Epidemiology 2001; 154: 1119– 25.
16 Sanders TA, Reddy S. Vegetarian diets and children. 34 Office for National Statistics. 2001 Census [online]. Available
American Journal of Clinical Nutrition 1994; 59: at http://www.statistics.gov.uk/census2001/default.asp.
1176S–81S. Accessed 13 April 2003.
17 Office of Population Censuses and Surveys. Mortality 35 Knutsen SF. Lifestyle and the use of health services.
Statistics for 1991. Series DH2 No. 18. London: The American Journal of Clinical Nutrition 1994; 59: 1171S– 5S.
Stationery Office, 1993. 36 Henderson L, Gregory J, Swan G. The National Diet and
18 Office of Population Censuses and Surveys. 1989 Cancer Nutrition Survey: Adults aged 19 to 64 years. Vol. 1. Norwich:
Registrations, England and Wales. Series MBI No. 22. HMSO, 2002.
London: The Stationery Office, 1994. 37 Jaggers C, Cade J, Greenwood D, Greenhalgh A. Who are the
19 Woodhouse A, Calvert C, Cade J. The UK Women’s Cohort vegetarians? Proceedings of the Nutrition Society 2001; 60:
Study: background and obtaining local ethical approval. 5A.
Proceedings of the Nutrition Society 1997; 56: 64A. 38 Palmer ME, Haller C, McKinney PE, Klein-Schwartz W,
20 Riboli E, Kaaks R. The EPIC Project: rationale and study Tschirgi A, Smolinske SC, et al. Adverse events associated
design. European Prospective Investigation into Cancer and with dietary supplements: an observational study. Lancet
Nutrition. International Journal of Epidemiology 1997; 2003; 361: 101.
26(Suppl. 1): S6– 14. 39 Gregory J, Foster K, Tyler H, Wiseman M. The Dietary and
21 Cade J, McClelland I, Woodhouse A. What are vegetarian Nutritional Survey of British Adults. London: HMSO, 1990.
women eating? Appetite 1995; 24: 267. 40 Cade JE, Burley VJ, Greenwood DC. A comparison of fish
22 Calvert C, Cade J, Barrett JH, Woodhouse A. Using cross- eaters with other meat eaters and vegetarians taking part in
check questions to address the problem of mis-reporting of the UK Women’s Cohort Study. Proceedings of the Nutrition
specific food groups on food frequency questionnaires. Society 2001; 60: 228A.
UKWCS Steering Group. United Kingdom Women’s Cohort 41 Fraser GE, Welch A, Luben R, Bingham SA, Day NE. The
Study Steering Group. European Journal of Clinical effect of age, sex, and education on food consumption of a
Nutrition 1997; 51: 708 – 12. middle-aged English cohort – EPIC in East Anglia.
23 Spence M, Cade JE, Burley VJ, Greenwood DC. Ability of the Preventive Medicine 2000; 30: 26 –34.
UK Women’s Cohort food frequency questionnaire to rank 42 Hjartaker A, Lund E. Relationship between dietary habits,
dietary intakes: a preliminary validation study. Proceedings age, lifestyle, and socio-economic status among adult
of the Nutrition Society 2002; 61: 117A. Norwegian women. The Norwegian Women and Cancer
24 Holland B, Welch AA, Unwin ID, Buss DH, Paul AA, Study. European Journal of Clinical Nutrition 1998; 52:
Southgate DAT. McCance & Widdowson’s The Composition 565–72.
of Foods. London: Royal Society of Chemistry and Ministry of 43 Kromhout D, Bloemberg BP, Feskens EJ, Hertog MG,
Agriculture, Fisheries and Food, 1991. Menotti A, Blackburn H. Alcohol, fish, fibre and antioxidant
25 Willett WC. Convergence of philosophy and science: the vitamins intake do not explain population differences in
third international congress on vegetarian nutrition. Amer- coronary heart disease mortality. International Journal of
ican Journal of Clinical Nutrition 1999; 70: 434S – 8S. Epidemiology 1996; 25: 753– 9.
26 Rose D, Pevalin D. The National Statistics Socio-Economic 44 Morris MC, Manson JE, Rosner B, Buring JE, Willett WC,
Classification: Unifying Official and Sociological Hennekens CH. Fish consumption and cardiovascular
Approaches to the Conceptualisation and Measurement of disease in the Physicians’ Health Study: a prospective
878 JE Cade et al.
study. American Journal of Epidemiology 1995; 142: 48 Lund E, Bonaa KH. Reduced breast cancer mortality among
166– 75. fishermen’s wives in Norway. Cancer Causes & Control 1993;
45 Albert CM, Hennekens CH, O’Donnell CJ, Ajani UA, Carey 4: 283– 7.
VJ, Willett WC, et al. Fish consumption and risk of sudden 49 Braga C, La Vecchia C, Negri E, Franceschi S, Parpinel M.
cardiac death. Journal of the American Medical Association Intake of selected foods and nutrients and breast cancer risk:
1998; 279: 23 –8. an age- and menopause-specific analysis. Nutrition &
46 Marckmann P, Gronbaek M. Fish consumption and coronary Cancer 1997; 28: 258– 63.
heart disease mortality. A systematic review of prospective 50 Caygill CPJ, Hill MJ. Fish n2 3 fatty acids and human
cohort studies. European Journal of Clinical Nutrition 1999; colorectal and breast cancer mortality. European Journal of
53: 585–90. Cancer Prevention 1995; 4: 329– 32.
47 Tretli S, Gaard M. Lifestyle changes during adolescence 51 Cade J, Thompson R, Burley V, Warm D, Margetts B.
and risk of breast cancer: an ecologic study of the effect of Consensus document on the development, validation and
World War II in Norway. Cancer Causes & Control 1996; 7: utilisation of food frequency questionnaires. Public Health
507– 12. Nutrition 2002; 5: 567– 87.