What consumers eat 19 Sweden and the United Kingdom) and is co-ordinated by the International Agency for Research on Cancer(IARC)in Lyon, france EPIC was initiated in 1990 with pilot and methodological studies to test the alidity and feasibility of drawing, determining representative portions and storing biological samples, of taking anthropometric measurements and of col- lecting data through different types of questionnaires and from variable European populations(Riboli and Kaaks, 1997). Upon the finalisation and the standardi sation of the study protocol, the fieldwork was undertaken from 1993 to 1998 and more than 480000 subjects were included in the cohort Eligible subjects were generally drawn from the general population, residin in a given geographical area. In some countries different sampling frames were used in order to ensure a good participation rate and complete follow-up; this did not seriously violate the sampling scheme of a cohort study. According to the study protocol, men over 40 years and women over 35 years of age were recruited. The lower age limit for women was selected to ensure a sufficient number of sub- jects for investigating risk factors for premenopausal cancers. The upper age limit of the cohort is less precisely defined and varies between 60 and 74 years, depend g on the study centre Standardised protocols have been developed to collect data on the subjects medical history, current medication, several lifestyle factors, anthropometry, diet and collection of blood samples. Sections of optional questions were added in some cases to address country-specific objectives. The questionnaire on physical activity was the same in the majority of the countries. A standard common pro- tocol was used for the anthropometric measurements, allowing for tests of within- and between-observer variability. For the dietary assessment, study subjects were requested to complete a centre- specific dietary questionnaire on their habitual food habits. In most countries a semi-quantitative, interviewer-or self-administered food frequency questionnaire was used. Preliminary analyses of the data colleted through the baseline ques tionnaire were undertaken in the coordinating center and included the estimation of the energy and selected nutrient intake and the daily consumption of major food groups. Additionally, a random sample from each cohort selected on the basis of the number and age-gender distribution of expected cancer cases was interviewed by trained interviewers using a computerised software programme(the EPIC-SOFT specially designed to collect standardised 24-hour recalls of foods consumed during the preceding day(Slimani et al, 1999) In the field of collecting dietary data at a European level, the EPIC pro contributed by developing methods for collecting comparable individual dietary intake data in culturally diverse populations. Although EPIC was not primarily aiming at documenting dietary patterns in Europe, the central database includes information on the habitual(using the food frequency method) and the sporadic (using one 24-hour recall) diet of more than 480000 Europeans with heteroge neous dietary habits, covering the diet of Mediterranean regions, the central European food patterns and the dietary habits of the Nordic populatiSweden and the United Kingdom) and is co-ordinated by the International Agency for Research on Cancer (IARC) in Lyon, France. EPIC was initiated in 1990 with pilot and methodological studies to test the validity and feasibility of drawing, determining representative portions and storing biological samples, of taking anthropometric measurements and of col￾lecting data through different types of questionnaires and from variable European populations (Riboli and Kaaks, 1997). Upon the finalisation and the standardi￾sation of the study protocol, the fieldwork was undertaken from 1993 to 1998 and more than 480 000 subjects were included in the cohort. Eligible subjects were generally drawn from the general population, residing in a given geographical area. In some countries different sampling frames were used in order to ensure a good participation rate and complete follow-up; this did not seriously violate the sampling scheme of a cohort study. According to the study protocol, men over 40 years and women over 35 years of age were recruited. The lower age limit for women was selected to ensure a sufficient number of sub￾jects for investigating risk factors for premenopausal cancers. The upper age limit of the cohort is less precisely defined and varies between 60 and 74 years, depend￾ing on the study centre. Standardised protocols have been developed to collect data on the subjects’ medical history, current medication, several lifestyle factors, anthropometry, diet and collection of blood samples. Sections of optional questions were added in some cases to address country-specific objectives. The questionnaire on physical activity was the same in the majority of the countries. A standard common pro￾tocol was used for the anthropometric measurements, allowing for tests of within￾and between-observer variability. For the dietary assessment, study subjects were requested to complete a centre￾specific dietary questionnaire on their habitual food habits. In most countries a semi-quantitative, interviewer- or self-administered food frequency questionnaire was used. Preliminary analyses of the data colleted through the baseline ques￾tionnaire were undertaken in the coordinating center and included the estimation of the energy and selected nutrient intake and the daily consumption of major food groups. Additionally, a random sample from each cohort selected on the basis of the number and age-gender distribution of expected cancer cases was interviewed by trained interviewers using a computerised software programme (the EPIC-SOFT) specially designed to collect standardised 24-hour recalls of foods consumed during the preceding day (Slimani et al, 1999). In the field of collecting dietary data at a European level, the EPIC project contributed by developing methods for collecting comparable individual dietary intake data in culturally diverse populations. Although EPIC was not primarily aiming at documenting dietary patterns in Europe, the central database includes information on the habitual (using the food frequency method) and the sporadic (using one 24-hour recall) diet of more than 480 000 Europeans with heteroge￾neous dietary habits, covering the diet of Mediterranean regions, the central European food patterns and the dietary habits of the Nordic populations. What consumers eat 19