Correspondence
Alcohol Consumption and Mortality in U.S. Adults
N Engl J Med 1998; 338:1385-1386May 7, 1998
- Article
To the Editor:
Thun and coworkers (Dec. 11 issue) 1 report some benefits of moderate alcohol consumption in nearly 500,000 subjects who were middle-aged or older. This is a large group, but an even larger group was excluded because alcohol consumption was not quantified or data were missing. This creates a problem, because the listed death rates for the excluded subjects with unclassified drinking habits are higher than those in the analytic cohort. We can only guess at the consumption habits of the excluded subjects, but a person unwilling to answer questions about his or her alcohol consumption is more likely to consume alcohol than not. As Dr. Potter notes in his accompanying editorial, alcohol consumption in the analytic cohort was lower than that reported for the United States as a whole, a finding consistent with underreporting of alcohol consumption.2 If we assume that a substantial proportion of the subjects in the excluded group consumed alcohol and that their mortality rate was elevated, the result could be an apparent reduction in the mortality rates in the groups consuming alcohol that were included in the final analytic cohort.
The simplistic interpretation of the report, despite implicit warnings in the discussion and the accompanying editorial, is that abstaining from alcohol increases mortality. We should emphasize that studies of subjects who do not consume alcohol as compared with the general population reveal a lower mortality rate: one such study reported a reduction of about 25 to 30 percent in the risk of all cancers and a decrease of about 20 percent in total mortality.3
3 ReferencesAlbert B. Lowenfels, M.D.
New York Medical College, Valhalla, NY 105951
Thun MJ ,Peto R ,Lopez AD , et al. Alcohol consumption and mortality among middle-aged and elderly U.S. adults. N Engl J Med 1997;337:1705-1714
Full Text | Web of Science | Medline2
Potter JD . Hazards and benefits of alcohol. N Engl J Med 1997;337:1763-1764
Full Text | Web of Science | Medline3
Kjaerheim K ,Andersen A ,Helseth A . Alcohol abstainers: a low-risk group for cancer -- a cohort study of Norwegian teetotalers. Cancer Epidemiol Biomarkers Prev 1993;2:93-97
Web of Science | Medline
To the Editor:
Interpretation of the observed J-shaped curve between mortality from all causes and alcohol consumption has intrigued many people for decades.1 As with all observational studies, however, subtle differences may contribute to the apparent protective effect of moderate drinking. Specifically, some subjects, embarrassed by their high levels of alcohol consumption, may provide socially desirable responses to questions about their consumption and claim to drink nothing. Consequently, any contamination in the group classified as having no consumption in the article by Thun et al. could have distorted the results.
Adverse lifestyles are both a major determinant of health and a major problem to document on survey research. The observed protective effect of moderate alcohol consumption might be exaggerated by bias from people who have particularly bad habits, including, perhaps, withholding information about their drinking from the volunteers who collected the data for this study.
1 ReferencesDavid R. Urbach, M.D.
Chaim M. Bell, M.D.
University of Toronto, Toronto, ON M5G 1N8, Canada1
Marmot MG ,Rose G ,Shipley MJ ,Thomas BJ . Alcohol and mortality: a U-shaped curve. Lancet 1981;1:580-583
CrossRef | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: The large American Cancer Society prospective study of middle-aged and elderly Americans we conducted found that those who initially reported having one or two drinks a day had lower death rates than those who reported no alcohol use, thereby reinforcing the similar findings of a prospective study of British doctors.1,2 Those who chose to complete the alcohol section of the mailed questionnaire were self-selected, as both we and Dr. Lowenfels noted, but it is still valid to compare those who answered the questions and said they did drink with those who said they did not and to conclude from this comparison that there is a protective effect of moderate alcohol consumption that slightly reduces the overall death rate in middle-aged or elderly Americans. We make, however, no particular claims about the overall effects of heavy drinking nor any claims about the net effects of alcohol on U.S. mortality rates. The half-million persons who completed the alcohol section of our questionnaire included disproportionately few who described themselves as consuming six or more drinks per day; moreover, of those who did so and who answered a second questionnaire eight years later, only 18 percent persisted in reporting such a habit. Finally, because the questionnaire was to be completed privately and mailed to a distant office, there was no question of the alcohol questions' being unanswered or consumption being understated because of embarrassment due to the presence of a local volunteer. (As we reported,3 a later survey suggested that most who left the alcohol questions blank were, in fact, nondrinkers.)
3 ReferencesMichael J. Thun, M.D.
American Cancer Society, Atlanta, GA 30329-4251Richard Peto, F.R.S.
University of Oxford, Oxford OX2 6HE, United KingdomClark W. Heath, Jr., M.D.
American Cancer Society, Atlanta, GA 30329-42511
Doll R ,Peto R ,Hall E ,Wheatley K ,Gray R . Mortality in relation to consumption of alcohol: 13 years' observations on male British doctors. BMJ 1994;309:911-918
CrossRef | Web of Science | Medline2
Doll R . One for the heart. BMJ 1997;315:1664-1668
CrossRef | Web of Science | Medline3
Thun MJ ,Peto R ,Lopez AD , et al. Alcohol consumption and mortality among middle-aged and elderly U.S. adults. N Engl J Med 1997;337:1705-1714
Full Text | Web of Science | Medline
Author/Editor Response
In my editorial,1 I noted that the study of Thun et al. showed us that more than one to two drinks per day was harmful to health and that even our ready and comfortable insistence that one drink per day is beneficial may be misplaced. This latter point was derived from the fact that the American Cancer Society cohort consumed only about half the amount of alcohol that is known to be consumed by the U.S. population as a whole. I noted two possible explanations — either a considerable proportion of heavy drinkers were not included in the cohort or consumption was underreported, perhaps differentially underreported, within the cohort. Both phenomena make it difficult to be sure about the level at which any possible benefit may begin or end.
Lowenfels supports the argument that either data or drinkers are missing and notes the lower mortality rate of a cohort of Norwegian nondrinkers. Urbach and Bell argue for differential underreporting and draw attention to the less favorable mortality among the persons in the cohort who did not provide alcohol-consumption data.
Perhaps it is time to reexamine our conclusions about any benefits that accrue from alcohol; it is certainly time to renew attention to its deleterious health consequences. Furthermore, it may be that if we had embraced the call to increase consumption of vegetables and fruit 2 with as much enthusiasm as we have embraced the increasingly dubious benefits of alcohol, our rates of cancer and heart disease would be dropping considerably faster.
2 ReferencesJohn D. Potter, M.D., Ph.D.
Fred Hutchinson Cancer Research Center, Seattle, WA 981091
Potter JD . Hazards and benefits of alcohol. N Engl J Med 1997;337:1763-1764
Full Text | Web of Science | Medline2
World Cancer Research Fund Panel. Food, nutrition and the prevention of cancer: a global perspective. Washington, D.C.: American Institute for Cancer Research, 1997.
- Citing Articles (1)
Citing Articles
1
Michael Thomas French, Didra BrownTaylor, Ricky Neville Bluthenthal. (2006) Price elasticity of demand for malt liquor beer: Findings from a US pilot study. Social Science & Medicine 62:9, 2101-2111
CrossRef
