Table 4.

Meta-analysis for the interaction effects between smoking status (ever vs. never) and variables possible or established risk factors of colorectal cancer

Risk factor of colorectal cancer	Case	Control	OR (95% C.I.) 2	I2, %	P value for multiplicative interaction b,c	RERI (95% CI) b,d	I2, %	P value for additive interaction b,d
BMI<25	2375	3137	1.14 (1.10, 1.27)	0	0.210	0.15 (−0.01, 0.31)	0	0.06
BMI≥25	4293	4472	1.24 (1.13, 1.35)	0
Male a	2341	2454	1.23 (1.09, 1.38)	0	0.496	−0.05 (−0.25, 0.16)	0	0.66
Female	2202	2164	1.18 (1.08, 1.29)	0
Fruit consumption (<median)	3114	3404	1.27 (1.14, 1.41)	0	0.112	0.16 (0.01, 0.30)	0	0.04
Fruit consumption (≥median)	3272	3951	1.13 (1.03, 1.25)	0
Vegetables consumption (<median)	2825	3156	1.18 (1.06, 1.32)	0	0.492	−0.02 (−0.19, 0.15)	0	0.81
Vegetables consumption (≥median)	3618	4235	1.21 (1.10, 1.34)	8
Any NSAID usage (Yes)	2201	3195	1.23 (1.09, 1.39)	0	0.900	0.14 (−0.04, 0.32)	0	0.14
Any NSAID usage (No)	4539	4492	1.21 (1.11, 1.32)	0

^a: In interaction analysis with sex, studies only including one sex were excluded (HPFS, NHS, and WHI);

^b: Adjusted for age, sex, BMI (<25, 25–<30), ≥30 kg/m2), education (high school graduate or less, some college or technical school, and college graduate or higher), alcohol intake (0–1 g/day, 1<−28 g/day, >28 g/day, when available), and study site (if applicable).

^c: Multiplicative interaction effects were evaluated by use of logistic regression models with interactive terms.

^d: Additive interaction effects were examined by use of linear odds ratio models with interactive terms; in the calculation of RERI, the reference groups are nerve smoker, BMI (<25 kg/m²), male, fruit consumption (≥ sex, study specific median [servings/day]), vegetables consumption (≥ sex, study specific median [servings/day]), any NSAID use (Yes), and alcohol intake (≤1 g/day).