Critique 139: Does adjustment for stress levels explain the protective effect of moderate drinking on the risk of mortality? — 18 May 2014
Ruf E, Baumert J, Meisinger C, Döring A, Ladwig K-H, and for the MONICA/KORA Investigators. Are psychosocial stressors associated with the relationship of alcohol consumption and all-cause mortality? BMC Public Health 2014, 14:312. http://www.biomedcentral.com/1471-2458/14/312.
Background: Several studies have shown a protective association of moderate alcohol intake with mortality. However, it remains unclear whether this relationship could be due to misclassification confounding. As psychosocial stressors are among those factors that have not been sufficiently controlled for, we assessed whether they may confound the relationship between alcohol consumption and all-cause mortality.
Methods: Three cross-sectional MONICA surveys (conducted 1984–1995) including 11,282 subjects aged 25–74 years were followed up within the framework of KORA (Cooperative Health Research in the Region of Augsburg), a population-based cohort, until 2002. The prevalences of diseases as well as of lifestyle, clinical and psychosocial variables were compared in different alcohol consumption categories. To assess all-cause mortality risks, hazard ratios (HRs) were estimated by Cox proportional hazards models which included lifestyle, clinical and psychosocial variables.
Results: Diseases were more prevalent among non-drinkers than among drinkers: Moreover, non-drinkers showed a higher percentage of an unfavourable lifestyle and were more affected with psychosocial stressors at baseline. Multivariable-adjusted HRs for moderate alcohol consumption versus no consumption were 0.74 (95% confidence interval (CI): 0.58-0.94) in men and 0.87 (95% CI: 0.66-1.16) in women. In men, moderate drinkers had a significantly lower all-cause mortality risk than non-drinkers or heavy drinkers (p = 0.002) even after multivariable adjustment. In women, moderate alcohol consumption was not associated with lowered risk of death from all causes.
Conclusions: The present study confirmed the impact of sick quitters on mortality risk, but failed to show that the association between alcohol consumption and mortality is confounded by psychosocial stressors.
Among the most consistent of findings from prospective epidemiologic studies is an inverse association between moderate drinking and total mortality, even after adjustments for all known potential confounders. The present analysis was carried out specifically to focus on the effects of psychosocial stressors on the association, using a large population-based German cohort from the WHO MONICA study. Standard risk factors assessed included smoking, physical activity, body size, blood pressure, hypertension, serum cholesterol, and HDL cholesterol; further, a history of diabetes, heart failure, myocardial infarction, cancer and liver diseases were assessed to help judge “sick quitters.” A total of 15.3% of males and 41.8% of females reported no alcohol consumption; “moderate drinking” was defined as an average intake of 0.1-39.9 g/day for men (making up 51.1% of the cohort) and 0.1 – 19.9 g/day for women (38.8% of the cohort). Although data are not presented, there were apparently few women in the higher drinking categories.
The authors related the effects of including, or not including, in their equations a large number of psychosocial stressors, including educational level, occupational status, several indices of social support, job strain symptoms, depressive symptoms, somatic symptoms, and self-perceived health status, in the estimation of the effects of alcohol consumption on risk of total mortality over an average follow-up period of 12 years. The majority of the factors evaluated in the present paper have been included in previous studies, but the authors combined an extensive list of such factors for these analyses, which showed little effect on risk estimates when added to the multi-variable analysis. The authors conclude: “The observed protective effect of moderate drinking could not be attributed to misclassification or confounding by psychosocial stressors.”
The authors have demonstrated among men a “U-shaped “ curve, as the risk of mortality was the same for men reporting ≥ 40 grams/day of alcohol as that of non-drinkers; the risk for moderate drinkers was reduced by 25-30%. For women, there was a lower estimated risk ratio for all drinkers than for non-drinkers, although the confidence intervals included 1.0 in all categories (perhaps, as the authors state, there were few women in their higher categories of alcohol intake).
Overall, this analysis of a large population-based population in Germany provides additional evidence that the observed reduction in total mortality seen among moderate drinkers is not due to confounding by other lifestyle factors, including psychosocial stressors.
Specific comments by Forum reviewers: Several Forum reviewers thought it interesting that the apparent underlying reason for this analysis was to show how confounding factors, not alcohol, was the reason that moderate drinkers have lower risk of mortality. As stated by the authors, their results indicate that psychosocial stressful factors do not explain the protective effect of moderate drinking. Reviewer Keil stated: “This paper is straight forward and easy to digest. This paper is very good, because the people who wrote it are good scientists and the cohort is of good quality. When you try everything to find an excuse and are unsuccessful at the end, this supports a protective effect of moderate drinking.”
Reviewer Van Velden commented: “There is no doubt that psychological stressors may cause alcohol abuse, and it is also true that people with health problems may abstain from alcohol consumption. This does not negate the positive effects on health by the majority of ‘normal’ people consuming alcohol moderately for social reasons, and not to ‘escape’ from reality. The authors should have stated this more clearly.”
Forum member Mattivi added: “This paper is a good example of useful dissemination of the negative results of a study (which is, unfortunately, not so common). Here the working hypothesis was that the protective association of moderate alcohol intake with mortality could be explained by misclassification confounding by psychosocial stressors. After evaluating a most extensive combination of psychosocial stressors, it was substantially rejected. This is a typical example where a null result is as interesting as a ‘positive’ one. We can safely conclude that the hypothesis of the protective effects of a moderate alcohol intake on mortality risk has passed a ‘stress test’!”
Reviewer Waterhouse commented on the results of this paper: “At least among men, the happiness associated with moderate drinking still cannot explain the reduced mortality in that population. Thus, the study suggests that happiness helps reduce mortality, but light drinking appears to help even the morose.”
Reviewer Lanzmann-Petithory stated: “The type of alcoholic beverage and the pattern of drinking are not taken into account. Further, the age of the population is especially heterogeneous, as the investigators have mixed people born during the 1st World War, some having gone through the 2nd World War, and young adults; there may well have been cohort effects for stress among this German population.”
She continues: “Further, the size of the population is somewhat low for a mortality study, and they were apparently not enough deaths to distinguish as many groups in the multi-variable adjustment analysis (3 groups) as in the crude rate (6 groups). Although the authors assert they could not evaluate drinking pattern, with their separation between week-end and a work day in their alcohol intake assessment, they could maybe have tried to identify a week-end binge drinking profile.”
Finally, Forum member De Gaetano wrote that he agreed with the positive comments of other reviewers, and added: “In our own meta-analysis quoted in this paper (Di Castelnuovo et al, Arch Int Med, 2006) we observed that, when adjusted and unadjusted data were compared, the maximum protection against mortality was only reduced from 19% to 16%. We commented that even if the confounders were twice as important of that considered in our analysis, a protection of 13% would have still been highly significant. The present paper confirms our hypothesis. The inverse association was found both in men and in women, but in women it disappeared at doses lower than in men.” The latter could reflect true gender differences in alcohol effect, but could just also relate to small number of women with high alcohol intake in this study.
An inverse association between moderate alcohol consumption and total mortality has been reported in most prospective epidemiologic studies, even after adjustments for all known potential confounders. The present analysis was carried out specifically to focus on the effects of psychosocial stressors on the association, using a large population-based German cohort from the WHO MONICA study. No alcohol intake was reported by 15.3% of males and 41.8% of females; “moderate drinking” was defined as an average intake of 0.1-39.9 g/day for men (making up 51.1% of the cohort) and 0.1 – 19.9 g/day for women (38.8% of the cohort). Although data are not presented, there were apparently few women in the higher drinking categories.
The authors related the effects of including, or not including, in their equations a large number of psychosocial stressors (including educational level, occupational status, several indices of social support, job strain symptoms, depressive symptoms, somatic symptoms, and self-perceived health status) in the estimation of the effects of alcohol consumption on risk of total mortality over an average follow-up period of 12 years. In their analyses, there was little effect on risk estimates for mortality when these factors were added to the multi-variable analysis. The authors conclude: “The observed protective effect of moderate drinking could not be attributed to misclassification or confounding by psychosocial stressors.”
The authors have demonstrated among men a “U-shaped” curve, with the risk for moderate drinkers being 25-30% lower than that of both non-drinkers and heavier drinkers. For women, there was a lower estimated mortality risk ratio for all drinkers than for non-drinkers, although the confidence intervals included 1.0 in all categories (perhaps, as the authors state, there were few women in their higher categories of alcohol intake).
Forum reviewers thought that this was a well-done analysis of a large population-based population. It did not support the hypothesis that social support, job strain, depressive symptoms, and other such psychosocial factors have a strong influence on the demonstrated inverse relation between moderate alcohol consumption and total mortality. Thus, this study provides additional evidence that the observed reduction in total mortality seen among moderate drinkers is not due to confounding by other lifestyle factors, including psychosocial stressors.
* * *
Contributions to this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Ulrich Keil, MD, PhD, Institute of Epidemiology and Social Medicine, University of Münster, Münster, Germany
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
Fulvio Mattivi, PhD, Head of the Department Good Quality and Nutrition, Research and Innovation Centre, Foundazione Edmund Mach, in San Michele all’Adige, Italy
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Tedd Goldfinger, DO, FACC, Desert Cardiology of
Dag S. Thelle, MD, PhD, Senior Professor of Cardiovascular Epidemiology and Prevention, University of Gothenburg, Sweden; Senior Professor of Quantitative Medicine at the University of Oslo, Norway
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner,
Dominique Lanzmann-Petithory,MD, PhD, Nutrition/Cardiology, Praticien Hospitalier Hôpital Emile Roux, Paris, France
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA