Coffee Consumption, Additive Use, and Risk of Type 2 Diabetes- Results from Three Large Prospective US Cohort Studies.

Background: Consumption of coffee has been consistently associated with lower risk of type 2 diabetes (T2D). However, it is unknown whether the use of additives may modify the association.

Objective: To study the association between coffee consumption and risk of T2D by considering the addition of sugar, artificial sweeteners, cream or a non-dairy coffee whitener.

Methods: We used three large prospective cohorts - Nurses' Health Study (NHS 1986 - 2020), Nurses' Health Study II (NHS II 1991 - 2020) and the Health Professionals Follow-up Study (HPFS 1991 - 2020). Self-reported coffee consumption, additive use and T2D incidence were confirmed using validated questionnaires. Time-dependent Cox proportional hazards regression models were used to calculate hazard ratios with multivariable adjustment.

Results: During 3,665,408 person-years of follow-up, we documented 13,281 incident T2D cases. After multivariable-adjustment, each additional cup of coffee without any additive was associated with a 10% lower risk of T2D (HR 0.90; 95% CI 0.89 to 0.92) in the pooled analysis of the 3 cohorts. The inverse association did not change among participants who added cream. Among participants who added sugar to coffee (on average 1 teaspoon per cup), the association was significantly weakened (HR 0.95; 95 % CI 0.93 to 0.97; interaction term HR 1.17; 95 % CI 1.07 to 1.27). A similar pattern was observed among those who used artificial sweeteners (HR 0.93; 95 % CI 0.90 to 0.96; interaction term HR 1.13; 95 % CI 1.00 to 1.28). The association between coffee consumption and T2D risk among those who used coffee whitener was also attenuated, though the interaction was not significant (HR 0.95, 95 % CI 0.91 to 1.00; interaction term 1.16; 95 % CI 0.66 to 2.06).

Conclusions: Adding sugar or artificial sweetener significantly attenuated the magnitude of the inverse association between higher coffee consumption and T2D risk, whereas the use of cream did not alter the inverse association.