Fig 4. Associations between selected emulsifier intakes and cancer risks among study participants from the NutriNet-Santé cohort, 2009–2021 (N = 92,000).^a,b DAG, diglyceride of fatty acid; FAs, fatty acids; HR, hazard ratio; MAG, monoglyceride of fatty acid.

^aEmulsifiers with at least 1 statistically significant association with cancer risk are represented here. The detail of all investigated associations between emulsifier intakes and cancer risk with corresponding HRs and 95% CIs is provided in eTable D, as well as cut-offs for categories of emulsifier intakes, and number of cancer cases per category of emulsifier intakes. Mean values for emulsifier intake within each category is presented in eTable B. Groups of emulsifiers were defined as follows (European codes): total carrageenans (E407, E407a). The following emulsifiers were coded as sex-specific tertiles: total carrageenans, E407, E412, E415, E440, E450, E471, and E500. Due to a higher proportion of non-consumers among the included participants, the following emulsifiers were coded as non-consumers (first category), low consumers (second category), and high consumers (third category), with low- and high-consumptions defined according to sex-specific median intakes among consumers: E340, E410, E414, E475, and E901. ^bMultivariable Cox proportional hazard models were adjusted for age (time-scale), sex, BMI (continuous, kg/m²), height (continuous, cm), physical activity (categorical IPAQ variable: high, moderate, low), smoking status (never smoked, former smoker, current smokers), number of smoked cigarettes in pack-years (continuous), educational level (less than high school degree, <2 y after high school degree, ≥2 y after high school degree), number of dietary records (continuous), family history of cancer (yes/no), energy intake without alcohol (continuous, kcal/d), daily intakes of alcohol (continuous, g/d), lipids (continuous, g/d), sugars (continuous, g/d), sodium (continuous, g/d), fibre (continuous, g/d), consumption levels of fruits and vegetables (continuous, g/d), red and processed meats (continuous, g/d), and dairy products (continuous, g/d). Breast cancer models were additionally adjusted for oral contraception (yes/no, in total and premenopausal models only), age at menarche (never, <12 y, ≥12 y), number of biological children (continuous), age at first biological child (no child, <30 y, ≥30 y), menopausal status at baseline (premenopausal, postmenopausal, in total models only), hormonal treatment for menopause (yes/no, in total and postmenopausal models only).