Table 4.

Data adjustments for included studies.

Study	Adjusted for
Joh et al. (20)	3 multivariable models with adjustments for potential confounders. Model 1 adjusted for age, time period of endoscopy, time since most recent endoscopy, number of endoscopies, reason for endoscopy. Model 2 additionally adjusted for family history of CRC, menopausal status/menopausal hormone use, current aspirin use greater than or equal to 2x/week, history of type 2 diabetes, adult height, body mass index (BMI) (at age 18 and current), smoking (adolescent and current), alcohol consumption (18–22 years and current), physical activity (adolescent and current). Model 3 additionally adjusted for adolescent and adult intake of total calories, total calcium, vitamin D, total folate, fiber, fruits, vegetables and dairy, currently total red meat intake, western dietary pattern score during adolescence, corresponding adult variables to adolescent exposure variables.
Molmenti et al. (23)	Calcium, energy, protein, total fat, saturated fat and supplemental folate intake were rescaled by a factor of 100 to provide large enough coefficient estimation to reasonable capture the change in each variable and its effect of metachronous adenomas. Odds rations and 95% CI adjusted for waist circumference, gender, energy and trial arm.
Zheng et al. (22)	Age-adjusted models controlled for age, total caloric intake, time period of endoscopy, number of reported endoscopies, time in years since most recent endoscopy, reason for current endoscopy. Multivariable models additionally adjusted for height, BMI, history of CRC in first-degree relative, menopausal status, menopausal hormone use, history of type 2 diabetes, pack-years of smoking, physical activity, current use of multivitamin, regular use of aspirin or non-steroidal anti-inflammatory drugs DASH diet further adjusted for alcohol intake.
Nguyen et al. (21)	Covariates adjusted for included age (5-year intervals), time period (2-year intervals), first degree family history of CRC, height, BMI, menopausal status, menopausal hormone use, personal history of type 2 diabetes, pack-years smoking, physical activity, current use of multivitamin, regular use of aspirin or non-steroidal anti-inflammatory drugs, number of reported endoscopies, time in years since most recent endoscopy, reason for most recent endoscopy, total caloric intake (quartiles). For analyses considering high school diet, covariates most proximate to the exposure used. High school sulfur microbial diet scores calculated without alcohol (primary), or assuming alcohol consumption was all beer, all liquor, or spilt between both equally.
Yue et al. (24)	Models stratified by age and follow-up cycle Models adjusted for energy intake (kcal/day) Multivariable models additionally adjusted for total alcohol consumption, height, race, family history of CRC, history of diabetes, smoking pack-years, regular use of aspirin or non-steroidal anti-inflammatory drugs, multivitamin use, menopausal status and hormone use, history of lower endoscopy within the past 10 years. Five dietary indices further adjusted for BMI and physical activity.
Peters et al. (29)	Adjusted for age and education.
Khan et al. (28)	Univariable odds ratios were unadjusted Multivariable odds ratios were adjusted for all independent variables.
Chang et al. (27)	Multivariable models adjusted for covariates that included age, sex, family history of CRC, aspirin/non-steroidal anti-inflammatory drug use, smoking, physical activity, BMI, alcohol consumption, red/processed meat intake, fruit and vegetable intake, high-fiber food intake, calcium supplement use.
Rosato et al. (26)	Unconditional multiple logistics regression models included terms for age, sex, center, study, year of interview, education, family history of CRC, alcohol drinking. The analysis of dietary items models further included terms for total energy intake using the residual method.
Archambault et al. (25)	Adjusted for age, sex, family history, study, total energy consumption.