BACKGROUND: Published studies rarely assess associations between trajectories of drinking and mortality. METHODS: We aimed to assess associations between long-term sex-specific drinking trajectories and all-cause and disease-specific mortality for 39 588 participants (23 527 women
16 061 men) enrolled in the Melbourne Collaborative Cohort Study in 1990-94 aged 40-69 years. Cox regression was used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for all-cause, cardiovascular disease- and cancer-specific mortality in relation to group-based alcohol intake trajectories. RESULTS: There were 7664 deaths (1117 cardiovascular
2251 cancer) in women over 595 456 person-years, and 7132 deaths (1283 cardiovascular
2340 cancer) in men over 377 314 person-years. We identified three distinct group-based alcohol intake trajectories for women: 'lifetime abstention', 'stable light', and 'increasing moderate'
and six for men: 'lifetime abstention', 'stable light', 'stable moderate', 'increasing heavy', 'early decreasing heavy', and 'late decreasing heavy'. We observed 9%-12% lower all-cause mortality, driven by associations with cardiovascular disease-specific deaths, for 'stable light' (women: HR 0.91
95% CI 0.87-0.96
men: HR 0.88
95% CI 0.82-0.94) and 'stable moderate' (HR 0.88
95% CI 0.81-0.96) drinking, compared with 'lifetime abstention'. In contrast, all-cause mortality was 18%-21% higher for 'early decreasing heavy' (HR 1.18
95% CI 1.05-1.32) and 'late decreasing heavy' (HR 1.21
95% CI 1.04-1.40) drinking, and cancer-specific mortality 19%-37% higher for 'increasing moderate' (HR 1.19
95% CI 1.00-1.43), 'early decreasing heavy' (HR 1.34
95% CI 1.10-1.64), and 'late decreasing heavy' (HR 1.37
95% CI 1.06-1.77) drinking. CONCLUSIONS: Our findings highlight the importance of avoiding higher levels of alcohol intake during the life course to reduce all-cause and cancer-specific mortality.