• Alcohol;
  • drinking patterns;
  • coronary heart disease;
  • mortality;
  • prospective cohort study


Aims  To investigate the relationship between three measures of alcohol consumption obtained simultaneously in a large cohort and the validated risk of coronary heart disease and all-cause mortality during follow-up.

Design  Prospective cohort study with median follow-up of 11 years.

Setting  The Whitehall II Cohort Study: London-based civil service.

Participants  A total of 10 308 (33% female) civil servants aged 35–55 years at baseline (1985–88).

Measurements  Self-reported volume of alcohol consumed during past week, frequency of drinking over past year, usual amount consumed per drinking session.

Main outcome measures  Coronary heart disease and all-cause mortality until 1999.

Findings  A U-shaped relationship was found between volume of alcohol consumed per week and outcome. Compared to those who drank moderately (10–80 g alcohol per week), non-drinkers and those drinking more than 248 g per week had approximately a twofold increased risk of mortality. The optimal frequency of drinking was between once or twice a week and daily, after adjustment for average volume consumed per week. Those drinking twice a day or more had an increased risk of mortality (male hazard ratio 2.44 95% CI 1.31–4.52) compared to those drinking once or twice a week. Drinking only once a month or only on special occasions had a 50% increased risk of mortality. The usual amount consumed per drinking session was not indicative of increased health risk in this cohort.

Conclusions  Epidemiological studies should collect information on frequency of drinking in addition to average volume consumed in order to inform sensible drinking advice.