Fig. 4

HRs (95% CI) of separate tea and coffee consumption for all-cause mortality in subgroups. Models were adjusted for age, sex, ethnicity, education, socioeconomic status, smoking status, alcohol consumption frequency, physical activity frequency, BMI, diet, cholesterol-lowering medication, BP medication, insulin, depression, and tea or coffee intake in coffee or tea analysis. Non-tea drinkers or non-coffee drinkers were considered the reference group. p_{− interaction}: the p for interaction was estimated by including an interaction term beta of the tea/coffee-baseline characteristics in fully adjusted Cox models