RESEARCH LETTERS Studies of alcohol and mortality show a U-shaped relation. 1 Heavy drinkers and abstainers have higher mortality rates than moderate drinkers. Explanations for the relation include a protective effect of moderate drinking, the presence of people who abstain because of illness, and other unidentified risk factors for mortality in abstainers. Studies of alcohol and mortality typically start in mid-life or later, and, therefore, there is little information about the lifetime characteristics of abstainers, including factors that may influence mortality. We investigated, in a large nationally representative sample of young adults in the UK, the association between alcohol consumption and factors predictive of mortality, specifically, psychological distress, self-rated health and limiting illness. 2,3 The 1958 birth cohort includes all individuals born from March 3 to March 9 in England, Scotland, and Wales. 4 At a 23-year and 33-year follow-up, 9605 individuals reported their consumption of beers, spirits, wine, and other alcoholic drinks in the previous week. Units of alcohol were estimated, with one unit of alcohol equivalent to half a pint of beer, one measure of spirits, or one glass of wine. We classified light drinkers as consuming 0–5 units (women) or 0–10 units (men) per week; moderate drinkers as 6–20 units (women) or 11–35 units (men); and heavy drinkers as more than 20 units (women) or more than 35 units (men). Psychological distress was indicated by scores of 7 or more on the malaise inventory, a self-completion 24-item symptom list with acceptable internal consistency (Cronbach’s  0·78–0·80). All individuals also rated their health as poor, fair, good, or excellent, and reported any limiting longstanding illnesses. Alcohol consumption was significantly associated with all three health measures (p<0·001) among men and women at age 33. There were higher rates of ill health among non- drinkers and heavy drinkers than among moderate drinkers (table). To investigate the hypothesis of higher rates among abstainers because of abstinence due to ill health, we repeated analyses excluding heavy or problem drinkers (as indicated by the CAGE scale of problem drinking 5 ) at age 23 years, and the U-shaped relation persisted. U-shaped relations with alcohol consumption were found for psychological distress, poor health, and limiting illness in early adulthood. The U-shaped relations between alcohol and mortality documented elsewhere, 1 are, therefore, preceded by similar relations between alcohol and ill health, evident by age 33 years, which implies that underlying causal mechanisms commence earlier in life. The poorer health of non-drinkers was not attributable to past heavy drinkers giving up alcohol. At least four other general explanatory models for the U- shaped relation can be identified: moderate drinking may have a causal protective effect on health; the causal direction may be reversed, which requires that poor health leads to abstinence or heavy drinking dependent on other factors; abstainers and heavy drinkers may share common risk factors for a particular health outcome; and non-drinkers and heavy drinkers are at increased health risk, but risk factors may differ between the two groups. These explanations are not mutually exclusive and the magnitude of their contribution may vary for the specific health outcomes assessed. For each of these models, the underlying mechanisms may also be implicated in the emergence of the U-shaped relation between alcohol and mortality. For example, unemployment and associated financial hardship might be common risk factors for abstinence and heavy drinking (model three) that, in turn, influence morbidity and mortality. Another example would be if abstainers were socially isolated and lacked support, whereas heavy drinkers experienced more adverse life events (model four). Abstainers and heavy drinkers are, however, similar in several respects long before the emergence of higher mortality in both groups. Attempts to explain the relation between mortality and alcohol consumption should, therefore, include the possible contribution of earlier life influences. We thank the Department of Health (England and Wales) and the National Health and Medical Research Council (Australia) for financial support. Chris Power is supported by the Canadian Institute for Advanced Research as a Weston Fellow. 1 Thun MJ, Peto R, Lopez AD, et al. Alcohol consumption and mortality among middle-aged and elderly US adults. N Engl J Med 1997; 337: 1705–14. 2 Huppert FA, Whittington JE. Symptoms of psychological distress predict 7-year mortality. Psychol Med 1995; 25: 1073–86. 3 Kaplan GA, Kamacho T. Perceived health and mortality: a nine year follow-up of the human population laboratory cohort. Am J Epidemiol 1983; 117: 292–304. 4 Ferri E, ed. Life at 33: the fifth follow-up of the National Child Development Study. London: National Children’s Bureau and City University, 1993. 5 Mayfield D, McLeod G, Hall P. The CAGE questionnaire: validation of a new alcoholism screening instrument. Am J Psychiatr 1974; 131: 1121–23. Epidemiology and Public Health, Institute of Child Health, London WC1N 1EH, UK (C Power; e-mail C.Power@ich.ucl.ac.uk); and NH&MRC NHMRC Psychiatric Epidemiology Research Centre, The Australian National University, Canberra, Australia THE LANCET • Vol 352 • September 12, 1998 877 Research letters U-shaped relation for alcohol consumption and health in early adulthood and implications for mortality Chris Power, Bryan Rodgers, Steven Hope Alcohol consumption p† None Light Moderate Heavy All All men Total 602 1689 1753 608 4652 Psychological distress 10·0% 4·6% 5·3% 9·1% 6·2% <0·001 Poor/fair rated health 20·2% 10·9% 10·0% 15·7% 12·4% <0·001 Limiting illness 12·0% 5·4% 3·8% 6·7% 5·8% <0·001 All women Total 1376 1956 1446 175 4953 Psychological distress 14·9% 10·1% 9·5% 20·0% 11·6% <0·001 Poor/fair rated health 16·7% 12·0% 11·5% 16·2% 13·3% <0·001 Limiting illness 9·2% 4·9% 4·7% 7·4% 6·1% <0·001 Restricted sample, men* Total 519 1424 1224 292 3459 Psychological distress 9·8% 4·0% 5·3% 11·0% 5·9% <0·001 Poor/fair rated health 20·1% 10·1% 9·4% 14·8% 11·8% <0·001 Limiting illness 12·1% 5·0% 3·7% 5·1% 5·6% <0·001 Restricted sample, women* Total 1320 1826 1239 123 4508 Psychological distress 14·8% 9·8% 9·3% 22·8% 11·5% <0·001 Poor/fair rated health 16·7% 11·5% 11·0% 17·2% 13·0% <0·001 Limiting illness 8·9% 4·9% 4·7% 7·3% 6·1% <0·001 *Excluding heavy or problem drinkers at age 23 years. † 2 test (3 degrees of freedom). Alcohol consumption and health status at age 33 years