Article Text

OP05 The longitudinal association between alcohol consumption and adiponectin: a prospective cohort study using Whitehall II data
  1. S Bell,
  2. A Britton
  1. Research Department of Epidemiology and Public Health, University College London, London, UK


Background Studies have shown that moderate alcohol consumption is associated with enhanced insulin sensitivity and a reduced incidence of type 2 diabetes. One suggested pathway is through alcohol-induced increases in the hormone adiponectin. Lower levels of adiponectin have also been linked to coronary heart disease, cancer, hypertension and metabolic syndrome. The majority of studies linking alcohol intake to adiponectin rely on only one measure of alcohol consumption at baseline and adiponectin level ascertained at a single follow-up occasion. It is important to consider the longitudinal development of both processes to determine how, if at all, the two are related. However, studies with repeat measures of alcohol consumption and adiponectin are scarce, so few studies have been able to examine the relationship simultaneously. The purpose of this study was to investigate how prospectively measured alcohol consumption is related to changes in adiponectin levels over time.

Methods This study uses data from the Whitehall II cohort. Prospective data on alcohol consumption (UK units per week; 1 unit = 8 g ethanol) and serum adiponectin (ng/ml) were collected at phase 3 (1991–94), phase 5 (1997–99) and phase 7 (2003–04). Adiponectin data were log-transformed. Those with known type 2 diabetes at phase 3 were excluded from the analytic sample, as were those who abstained from alcohol throughout follow-up. A dynamic structural equation model was estimated whereby lagged alcohol consumption predicted upcoming change in (log) adiponectin (the model included longitudinal trajectories for both processes as well as covariances between the intercepts and slopes of both trajectories). Age, gender and ethnicity were entered as time-invariant predictors whilst body mass index, waist circumference, fasting serum insulin and smoking status were entered as time-varying covariates. Models were fit in Mplus 7.11. The final sample size was 2098.

Results Increased alcohol consumption was associated with increases in (log) adiponectin (Beta=0.005, CI 0.004, 0.006; p < 0.001) after adjustment for confounding factors. Both the alcohol intercept (Rho=-0.543) and slope (Rho=-1.051) were significantly associated with the adiponectin slope (p < 0.01 in both instances) indicating that both the initial level as well as change in alcohol intake impact the rate of change in adiponectin levels.

Conclusion Our findings support the hypothesis that the reduced risk of type 2 diabetes (and coronary heart disease) among moderate alcohol consumers may be partly mediated through increases in adiponectin levels. Understanding mechanisms is important in establishing whether associations are causally linked.

  • alcohol
  • diabetes
  • CHD

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.