Alcohol consumption and mortality risks in the USA.

AIMS To estimate all-cause and cause-specific mortality risks by alcohol consumption level for the US population. METHODS The survival analysis of data from National Health Interview Survey participants from 1997 to 2004 who were followed for mortality through 2006. Hazard ratios (HRs) adjusted for available demographic and health characteristics were calculated by alcohol consumption level for all-cause mortality and for mortality from circulatory and external causes. Alcohol consumption levels were generally defined as 'never drinker' (<12 drinks in life), 'former drinker' (0 drinks during last year), 'infrequent drinker' (<12 drinks in any year), 'light drinker' (1 drink per drinking day), 'moderate drinker'(2 drinks) and 'heavy drinker' (3+ drinks). RESULTS Adjusted mortality hazards for lifetime infrequent drinkers were much lower than the hazards for never drinkers among women, so lifetime infrequent drinkers were used as the reference category in survival analysis to estimate the mortality effects of alcohol consumption. Estimated all-cause mortality HRs for moderate drinkers were generally somewhat lower when compared with infrequent drinkers [HR for male moderate drinkers = 0.87, 95% confidence interval (CI) = 0.75-1.01 and HR for female occasional moderate drinkers = 0.80, 95% CI = 0.69-0.93]. Former drinkers and regular heavy drinkers had higher mortality hazards among both men and women. CONCLUSION US light to moderate drinkers may have reduced mortality risks, but some portion of their previously observed lower mortality may be due to factors other than alcohol consumption such as medical care and social integration, particularly among women. Alcohol consumption among former and heavy drinkers appears to have increased their mortality risks.