The Norwegian Women and Cancer Study started in 1991 (Lund et al. 2003).The main purpose was to create a large prospective cohort designed to study the relationship between internal and external hormones and female cancer with focus on breast cancer. The women were mailed an invitation with a photo-booklet of all oral contraceptives (Kumle et al. 2002) or hormonal replacement therapy brands (Bakken et al. 2001) and a questionnaire, 4 or 8 pages. Questions on diet, especially fish consumption (Lund 2002) , and sun habits (Veierod et al. 2003) were added in the later mailings of the study. During the years 1991-1997 altogether 179 387 women were invited of whom 102 540 women (57%) aged 30-70 years returned a questionnaire to the Institute of Community Medicine, University of Tromsø, Norway. All women filled in an informed consent for later linkages to the Norwegian Cancer Registry, the register of the National Mammographic Screening Program, and the register of death certificates in Statistics Norway. All women have been sampled randomly from the Norwegian Central Person Register which contains information on all Norwegian inhabitants including a unique birth number consisting of the date of birth and five additional numbers giving a unique combination (Lunde et al. 1980). This person number is used in all linkage in order to give a complete follow-up through the national registers. At the last linkage to the Norwegian Cancer Registry out 2004, 5.569 incident cases of cancer of which 1.983 was breast cancer, had been registered (funded by The Norwegian Cancer Society, (NCS), the Norwegian Research Council, (NRC), National Cancer Institute, US (NCI).
Expansion of the cohort
During 2003-2004 another 60 000 women born 1943-57 or 45-60 years old was invited of whom about 28 000 returned a filled in 8 pages questionnaire. In 2005-2006,in collaboration with Statistics Norway, we invited 60 000 new women born in the same years (funded by NRC).
In 1998-2002 all women received an invitation to fill in a second questionnaire and 80 693 women replied (response rate 81% corrected for death and emigration). Of these, 37 226 women constitute the Norwegian part of the European Prospective Investigation into Cancer and Nutrition – the EPIC study (Riboli & Kaaks 1997). All women completed an 8-paged questionnaire, with a semi-quantitative food frequency part as part of the EPIC database (Bingham & Riboli 2004). Of the participating EPIC women 1.3% had answered less than 50% of the questions or had an energy intake lower than 2500 kJ or higher than 15 000 kJ (funded by NRC, NCS).
Third mailing Spring 2004 we started mailing the third questionnaire to those participating in 1991-1995. From 5 077 invited in March 2004 3 827 returned an 8 pages questionnaire before the first remainder. In autumn 2004 we did invite another 25 000, and in autumn 2005 the last 17 000. The third questionnaire was mailed about 10-12 years after the first (funded by NRC).
In 2003 a test-retest study was undertaken among 2 000 invited women. The kappa estimator of agreement varied from 0.95 for oral contraceptive use to typical 0.50-0.70 for dietary questions. This study is part of a PhD work done in collaboration with Department of Statistics, Institute of Basic Medical Sciences, University of Oslo (funded by NCS).
During 2004 we completed our validation study of the food frequency questionnaire that has been used in NOWAC and which is also part of the EPIC study by adding another 24-hours recall study; 250 women called four times or once each season. The estimated intake of foods will be compared to the 1820 24-hours recall done as part of the EPIC calibration study (Slimani et al. 2002). The estimated food intake from the questionnaire, which is computed according to the rules laid down by EPIC and by use of the 2001 version of the Norwegian food composition table. Previously a validation of questionnaire information about omega fatty acids showed a good correlation with measured serum levels.
The external validity of the study has been verified and found acceptable (Lund et al. 2003;Lund & Gram 1998) . The figure below compares observed and expected number of incident cases of all cancers in NOWAC, and the differences were small.
Cumulative age-specific incidence rates per 100,000 women per year for all sites of cancer 1999, a comparison between NOWAC rates and national figures from the Cancer Registry of Norway (Cancer Registry of Norway 2002)