van den Brandt 1993a

Methods	Cohort/sub‐cohort‐controlled cohort study Country: the Netherlands
Participants	Name of parent cohort: Netherlands Cohort Study (NLCS) Recruitment: 1986 van den Brandt 1993b: Participants: 120,852: 58,279 men and 62,573 women; aged 55 to 69 years; returned baseline questionnaire; no history of cancer at baseline Outcome assessment: n.r. Number of cases: Stomach cancer: 104 (male/female: 84/20) Colon cancer: 234 (male/female: 121/113) Rectal cancer: 113 (male/female: 77/36) van den Brandt 1993a: Participants: 120,852: 58,279 men and 62,573 women; age 55 to 69 years; returned baseline questionnaire; no history of cancer at baseline Outcome assessment: n.r. Number of cases: Lung cancer: 370 (male/female: 335/35) van den Brandt 1994: Participants: 62,573 post‐menopausal women Outcome assessment: 1989 Number of cases: Breast cancer (post‐menopausal): 355 (male/female: 0/355) Breast cancer (post‐menopausal), multivariate analysis: 270 (male/female: 0/270) Zeegers 2002: Participants: 120,852: 58,279 men and 62,573 women Outcome assessment: December 1992 Number of cases: Bladder cancer: 431 (male/female: 372/59) van den Brandt 2003: Participants: 58,279 men Outcome assessment: n.r. (probably December 1992) Number of cases: Prostate cancer: 540 (male/female: 540/0) Case definition: incidence Years of follow‐up: 3.3 years (Brandt 1993a; Brandt 1993b; Brandt 1994), 6.3 years (Zeegers 2002; Brandt 2003) Type of selenium marker: toenail
Interventions	d.n.a.
Outcomes	van den Brandt 1993b: Analysed cases: 234 of 351 colon cancer cases / 104 of 176 stomach cancer cases / 113 of 185 rectal cancer cases analysed (reasons for non‐inclusion: history of cancer at baseline not available, no pathological confirmation or CIS, no toenail clipping available) Statistical methods: Mantel‐Haenszel Variables controlled in analysis: age, gender van den Brandt 1993a: Analysed cases: 370 of 617 cases analysed (reasons for non‐inclusion: history of cancer at baseline not available, no toenail clipping, no pathological confirmation, problems with selenium measurement) Statistical methods: Statistical methods: Mantel‐Haenszel Variables controlled in analysis: age, gender van den Brandt 1994: Analysed cases: 355 of 553 cases analysed (reasons for non‐inclusion: history of cancer at baseline not available, CIS, no toenail sample or problems with selenium detection) Statistical methods: multivariate case‐cohort analysis Variables controlled in analysis: age, history of benign breast disease, maternal breast cancer, breast cancer in sister(s), age at menarche, age at menopause, oral contraceptive use, parity, age at first birth, body mass index, education, current cigarette smoking, alcohol intake, energy intake Zeegers 2002: Analysed cases: 431 of 619 cases analysed (reason for non‐inclusion: no toenails available) Statistical methods: exponentially distributed failure time regression models Variables controlled in analysis: age, gender, number of cigarettes/day, years of cigarette smoking van den Brandt 2003: Analysed cases: 540 of 704 cases analysed (reason for non‐inclusion: no toenail samples or selenium detection not possible) Statistical methods: exponentially distributed failure time regression models Variables controlled in analysis: age, family history of prostate cancer, number of cigarettes/day, years of cigarette smoking, level of education
Risk estimates [95% CI]	Reference category: lowest quartile/quintile Results: van den Brandt 1993b: Stomach cancer both genders: highest quintile: RR 0.61 (95% CI 0.33 to 1.11); highest quintile: RR 0.64 (95% CI 0.33 to 1.27) (max. adj.) men: highest quintile: RR 0.40 (95% CI 0.17 to 0.96) (max. adj.) women: highest quartile: RR 1.68 (95% CI 0.43 to 6.54) (max. adj.) Colon cancer   both genders: highest quintile: RR 0.77 (95% CI 0.49 to 1.19); highest quintile: RR 0.80 (95% CI 0.50 to 1.29) (max. adj.) men: highest quintile: RR 0.82 (95% CI 0.43 to 1.58) (max. adj.) women: highest quintile: RR 0.77 (95% CI 0.41 to 1.45) (max. adj.) Rectal cancer  both genders: highest quintile: RR 1.01 (95% CI 0.55 to 1.84); highest quintile: RR 1.05 (95% CI 0.54 to 2.03) (max. adj.) men: highest quintile: RR 0.91 (95% CI 0.41 to 2.00) (max. adj.) women: highest quartile: RR 1.58 (95% CI 0.59 to 4.22) (max. adj.) van den Brandt 1993a: Lung cancer  both genders: highest quintile: RR 0.40 (95% CI 0.27 to 0.59) men: highest quintile: RR 0.50 (95% CI 0.30 to 0.82) women: highest quartile: RR 0.40 (95% CI 0.13 to 1.24) van den Brandt 1994: Breast cancer          multivariate analysis: highest quintile: RR 0.84 (95% CI 0.55 to 1.27) age‐stratified analysis: highest quintile: RR 0.93 (95% CI 0.65 to 1.33) Zeegers 2002: Bladder cancer   both genders: highest quintile: RR 0.67 (95% CI 0.46 to 0.97) van den Brandt 2003: Prostate cancer     highest quintile: RR 0.69 (95% CI 0.48 to 0.99)
Selenium levels in exposure categories	van den Brandt 1993b: lowest quintile: ≤ 0.483 µg/g highest quintile: ≥ 0.631 µg/g  lowest quartile: ≤ 0.497 µg/g highest quartile: ≥ 0.613 µg/g  van den Brandt 1993a: both genders and men: lowest quintile: ≤ 0.483 µg/g highest quintile: ≥ 0.631 µg/g women: lowest quartile ≤ 0.497 µg/g highest quartile ≥ 0.613 µg/g van den Brandt 1994: women: lowest quintile: ≤ 0.499 µg/g highest quintile: ≥ 0.646 µg/g Zeegers 2002 : lowest quintile: ≤ 0.483 µg/g highest quintile: ≥ 0.631 µg/g van den Brandt 2003: men: lowest quintile: ≤ 0.467 µg/g highest quintile: ≥ 0.617 µg/g
Notes	Primary publication: van den Brandt 1993b Other publications: Zeegers 2002, van den Brandt 1993a, van den Brandt 1994, van den Brandt 2003