Citation

Werner M. Clin. Chim. Acta 1996; 246(1-2): 5-20.

Affiliation

Department of Pathology, George Washington University, Washington, DC 20037, USA.

Copyright

(Copyright © 1996, Elsevier Publishing)

DOI

unavailable

PMID

8814969

Abstract

Alcohol consumption has been associated with favorable as well as adverse health effects. Risk factors represent formal predictors that such outcomes will occur more frequently than due to chance alone, but these epidemiological indicators are only average statistical associations, and by themselves do not prove causality. However, the concept of risk factors assumes that the total health hazard for any defined age-sex-race cohort can be segmented into one component accounted for by the effects of specified risks and one component accounted for by the "basic average health hazard". Typically, the risk-outcome relationships of drinking are not linear. A so-called U-shaped association between drinking and total mortality has been proposed, whereby moderate drinkers have lower mortality than either abstainers or heavy drinkers. When the relative risk of drinking is formulated as a quotient (odds ratio) with the basic health hazard used as the denominator, appropriate definition and selection of the reference population used to derive the latter becomes pivotal. Equally crucial is the choice of the yardstick for outcome. To date such objective measures as morbidity or mortality have been used almost exclusively, but the complex impact of drinking would demand assessment of physical, mental and social functioning. Finally, the threshold value above which drinking should be considered epidemiologically significant is a judgmental issue, where ethical bias can express itself. By their very nature epidemiological risk estimates are inadequate if not inapplicable to advise everyone that moderate drinking is a personal medical benefit or risk. To render any health factor identified by statistical relationships applicable in individual medical counselling demands validation by clinical trial of its sensitivity, specificity, predictive values, and associated benefits as well as costs. This information has not been compiled, but nevertheless, epidemiological correlations can be used to compare risks and benefits among different health factors to arrive at informed personal choices.

Language: en