Objective To examine the relationship between binge drinking at baseline and the onset of new depression during 24 years of follow-up after adjustment for age, socio-economic status, education and marital status.

Design and setting Data from phases 1 (1985–1988) to 9 (2007–2009) of the Whitehall II prospective cohort were used.

Participants 5985 (male=4161, female=1824) British civil servants aged 35–55 years who were free from depression at baseline.

Comparison groups Cohort members were grouped, at phase 1, according to their self-reported usual and maximum number of alcoholic drinks consumed in a single sitting—abstainers, non-bingers (reference category) and bingers. Alcohol consumption was split into two categories and number of drinks consumed was converted to units for analysis: wine and spirits (1 unit per drink), and beer (2 units per drink). For usual drinking sessions those who reported consuming 5+ units of wine/spirits and 10+ units of beer were categorised as bingers, those consuming 1–4 units of wine/spirits or 1–9 units of beer were classified as non-bingers. For maximum drinking sessions, participants were defined as bingers following the Department of Health guidelines as those consuming 8+ or 6+ units of alcohol for males and females respectively for both categories of consumption. Those who reported consuming no drinks were classified as abstainers in all analyses.

Main outcome measures The 30-item General Health Questionnaire (GHQ-30) was administered at all phases of data collection. The depression subscale of the GHQ-30 was used to identify new cases of depression (scores of 4 or more) across all phases.

Results Adjusted HRs and 95% CIs were estimated using Cox proportional hazard models fitted in the total cohort and stratified by gender. Usual drinking session spirit/wine bingers had an elevated risk of depression (HR 1.28, CI 1.02 to to 1.60) compared to non-bingers in the total sample. Maximum drinking session spirit/wine bingers had a greater risk of depression in the total (HR 1.23, CI 1.04 to to 1.44) and male (HR 1.27, CI 1.03 to to 1.56) samples. There were no statistically significant effects when using beer measures as exposures or for abstainers in any alcohol measures after adjustment for confounders.

Conclusion Binge drinking on wine and spirits, but not beer, in midlife increases the risk of having a depressive episode over the course of the following 22–24 year period. Future work will examine other covariates and explore bidirectional issues in this relationship.