Replication analysis in the ESTHER study.

aModels adjusted for age (5-year groups), alcohol drinking (none, 1-39, 40-99, 100-199, ≥200 g/w), coffee consumption (none, <1, 1, >1 cup/d), sex, and BMI category. bThe assay used in the ESTHER study measured γ-GT at 37 °C, and the logistic models predicted elevated γ-GT defined according to the male cutoff (>50 U/L at 37 °C) provided by the manufacturer. Geometric means in the table refer to values at 25 °C, calculated by multiplying the 37 °C values with a conversion factor of 0.57 (Thomas L, et al. 2005; J Lab Med 29:301). cP-values refer to a 2 degrees of freedom Wald test for the significance of an interaction term of the smoking trend variable with the BMI category. Given are results of trend modelsa (median smoking intensity of 0, 8, 20, 30 and 40 cigarettes/day assigned to individuals belonging to the different smoking intensity classes; former smokers excluded) estimating the effects of smoking on serum γ-GT levels nested in BMI strata.