Abstract
BACKGROUND: Non-linear associations have been reported between baseline measures of alcohol consumption and type 2 diabetes (T2DM). However, given that drinking varies over the adult life course, we investigated whether differences existed in the longitudinal trajectory of alcohol consumption according to T2DM status. METHODS: For a case-cohort (916 incident cases; 7376 controls) of British civil servants nested within the Whitehall II cohort, the self-reported weekly volume of alcohol consumption was traced backwards from the date of diagnosis or censoring to the beginning of the study, covering a period of up to 28 years. Mean trajectories of alcohol intake were estimated separately by diagnosis status using random-effects models. RESULTS: Drinking increased linearly among male cases before diagnosis, but declined among male non-cases prior to censoring. At the time of diagnosis or censoring, consumption among those who developed T2DM was 33.4 g/week greater on average. These patterns were not apparent among women. Here, alcohol intake among female cases was consistently below that of non-cases, with the difference in consumption most pronounced around 15 years prior to diagnosis or censoring, at ∼28.0 g/week. Disparities by diagnosis status were attenuated following adjustment for potential confounders, including the frequency of consumption and metabolic factors. Drinking among male and female cases declined following diagnosis. CONCLUSIONS: Differences in the weekly volume of alcohol consumption are reported in the years leading up to diagnosis or censoring. Although male and female cases predominantly consumed alcohol at volumes lower than or equal to those who were not diagnosed, these disparities appear to be largely explained by a range of socio-demographic and lifestyle factors. Where disparities are observed between cases and non-cases, adjusted absolute differences are small in magnitude. The decision to drink alcohol should not be motivated by a perceived benefit to T2DM risk.