Meta-analysis of alcohol and all-cause mortality: a validation of NHMRC recommendations

Abstract

Objective: To compare the National Health and Medical Research Council (NHMRC) recommendations on responsible, hazardous and harmful alcohol intake with their effects on all-cause mortality in men and women and on the occurrence of 10 specific neoplastic, cardiovascular and alimentary diseases.

Design: Meta-analyses of relative risks of mortality in relation to usual level of alcohol intake pooled from 16 cohort studies (mostly of adults over 35 years), and alcohol and selected conditions from a further 132 epidemiological studies. Results reported by authors were assigned to sex-specific exposure categories defined by the NHMRC based on median alcohol intakes. Pooled estimates of relative risk were calculated using precision-based weighting.

Setting: The assessment was part of comprehensive meta-analysis of epidemiological research undertaken for the National Drug Strategy.

Results: Relative risk of all-cause mortality in male drinkers compared with abstainers fell to 0.84 at 1.0-1.9 standard drinks per day, returned to 1.01 by 3.0-3.9 drinks, and increased to 1.37 at six or more drinks. In female drinkers the lowest relative risk (of 0.88) was at 0-0.9 drinks per day, and by 2.0-2.9 drinks the risk exceeded that in abstainers by 1.13; at six drinks the relative risk was 1.58. Based on NHMRC categories, the relative risks of mortality were 0.93 (0.93-0.94) in responsible drinkers, 1.24 (1.22-1.27) in hazardous drinkers and 1.37 (1.35-1.49) in harmful drinkers. Risk of cancers of the oropharynx, oesophagus, liver, larynx and female breast and of cirrhosis of the liver increased with increasing alcohol intake level.

Conclusions: A pattern of usual alcohol intake consistent with the NHMRC recommendations will confer a mortality risk similar to or less than that observed in abstainers. The biologically effective dose of alcohol on mortality in women is approximately two standard drinks per day less than in men. Our validation is most reliable for drinkers aged 35 years or older.