Smoking and Colorectal Cancer: Different Effects by Type of Cigarettes?

↵* Number of controls/number of cases. The analysis of smoking intensity and duration gave rise to missing values for this variable in three male and one female subjects. These case-control pairs were excluded from the analysis.

↵† OR and 95% CI are estimated by conditional logistic regression, matched on age and ethnicity, compared with never smokers, and adjusted for family history of colorectal cancer, body mass index 5 y ago (males only), lifetime hours in recreational activities (males only) and egg, total calcium, non-starch polysaccharides from vegetables, methionine (females only), and alcohol (males only) intake. All levels of the smoking characteristics are compared against never smokers. Separate models were run for intensity, duration, and time since smoking commenced.

↵‡ Further adjusted for smoking status.

↵§ P_trend is estimated by fitting the medians of each category as a continuous variable in the multivariate model.

↵∥ Further adjusted for smoking status and duration.

NOTE: OR and 95% CI are estimated by conditional logistic regression, matched on age and ethnicity, and adjusted for family history of colorectal cancer, body mass index 5 y ago (males only), lifetime hours in recreational activities (males only) and egg, total calcium, non-starch polysaccharides from vegetables, methionine (females only), and alcohol (males only) intake. Two separate models were run in each sex, both with never smokers as the reference category: one model included indicator variables for each quartile of pack-years of all tobacco products, and the other model included indicator variables for each quartile of pack-years for each type of cigarettes (filtered and non-filtered cigarettes), and in men, levels (<median; >median) of pack-years of cigars and pipes. In women, the latter model was further adjusted for pack-years of cigars and pipes (continuous).

↵* Q1, 1st quartile; Q2, 2nd quartile; Q3, 3rd quartile; Q4, 4th quartile. Based on male controls who smoked, the quartile cut points (25th-75th percentile) for total pack-years and pack-years of filtered and non-filtered cigarettes are as follows: >0-12, >12-28.3, >28.3-48, and >48 for all tobacco; >0-10.4, >10.4-23, >23-40, and >40 for filtered cigarettes; and >0-4.8, >4.8-11, >11-24 and >24 for non-filtered cigarettes. Median cut points for pipes and cigars are 1.8 pack-years of cigars and 1.0 pack-years of pipes. Based on female controls who smoked, the quartile cut points (25th-75th percentile) for total pack-years and pack-years of filtered and non-filtered cigarettes are as follows: >0-5.3, >5.3-15, >15-34.7, and >34.7 for all tobacco; >0-7, >7-15.6, >15.6-30, and >30 for filtered cigarettes; and >0-1.4, >1.4-5, >5-11, and >11 for non-filtered cigarettes.

↵† P_trend is estimated by fitting the medians of each quartile as a continuous variable in the multivariate model.

↵‡ Number of controls/number of cases (never smokers: 363/282 and 528/489 for men and women, respectively).

NOTE: OR and 95% CI are estimated by conditional logistic regression, matched on age and sex, and adjusted for family history of colorectal cancer, body mass index 5 y ago, lifetime hours in recreational activities as well as egg, total calcium, non-starch polysaccharides from vegetables, methionine, and alcohol intake. Two separate models were run in each ethnicity, both with never smokers as the reference category: one model included indicator variables for each quartile of pack-years of all tobacco products, and the other model included indicator variables for each quartile of pack-years for each type of cigarettes (filtered and non-filtered) and was further adjusted for pack-years of cigars and pipes (continuous).

↵* Q1, 1st quartile; Q2, 2nd quartile; Q3, 3rd quartile; Q4, 4th quartile. Based on controls who smoked, the quartile cut points (25th-75th percentile) for total pack-years are as follows: >0-9.8, >9.8-24.6, >24.6-44, and >44; for pack-years of filtered cigarettes: >0-9.5, >9.5-21, >21-38, and >38; for pack-years of non-filtered cigarettes: >0-4, >4-10, >10-22, and >22.

↵† P_trend is estimated by fitting the medians of each quartile as a continuous variable in the multivariate model.

↵‡ Number of controls/number of cases (never smokers: 506/452, 196/150 and 72/68 for Japanese, Caucasians, and Native Hawaiians, respectively).

NOTE: OR and 95% CI are estimated by conditional logistic regression, matched on age, sex and ethnicity, and adjusted for family history of colorectal cancer, body mass index 5 y ago, lifetime hours in recreational activities as well as egg, total calcium, non-starch polysaccharides from vegetables, methionine, and alcohol intake. Two separate models were run for each subsite, both with never smokers as the reference category: one model included indicator variables for each quartile of for pack-years of all tobacco products, and the other model included indicator variables for each quartile of pack-years for each type of cigarettes (filtered and non-filtered) and was further adjusted for pack-years of cigars and pipes (continuous).

↵* Q1, 1st quartile; Q2, 2nd quartile; Q3, 3rd quartile; Q4, 4th quartile. Based on controls who smoked, the quartile cut points (25th-75th percentile) for total pack-years are as follows: >0-9.8, >9.8-24.6, >24.6-44, and >44; for pack-years of filtered cigarettes: >0-9.5, >9.5-21, >21-38, and >38; for pack-years of non-filtered cigarettes: >0-4, >4-10, >10-22, and >22.

↵† P_trend is estimated by fitting the medians of each quartile as a continuous variable in the multivariate model.

↵‡ Number of controls/number of cases (never smokers: 607/547 and 276/211 for males and females, respectively).