Critique 097: Per-capita alcohol intake and all-cause mortality in Australia — 4 December 2012
Livingston M, Wilkinson C. Per-capita alcohol consumption and all-cause male mortality in Australia, 1911–2006. Alcohol and Alcoholism 2012. Pre-publication: doi: 10.1093/alcalc/ags123
Aims: Given the variety of relationships found between alcohol consumption and health using individual data (both negative and positive), the likely impact of changes in population-level alcohol consumption on health at the population level is not clear. This paper uses historical data from 1911 to 2006 to assess the relationship between changes in per-capita alcohol consumption on total male mortality in Australia.
Methods: A longitudinal aggregate study using Australian per-capita alcohol consumption and mortality data from 1911 to 2006. Analysis is undertaken using autoregressive integrated moving average time-series methods.
Results: Per-capita pure alcohol consumption has a significant association with male all-cause mortality, with an increase (decrease) of 1 liter per-capita per year associated with a 1.5% increase (decrease) in male mortality (controlling for female mortality and smoking rates). The association between per-capita consumption and mortality was significant for all age groups, with a particularly strong effect among 15–29 year olds.
Conclusion: These results place Australia in the group of countries for which a positive association between per-capita alcohol consumption and total mortality can be demonstrated. Thus, despite the beneficial effects of alcohol consumption on health found in many studies, increases in consumption at the population level in Australia are associated with declines in population health. Thus, per-capita alcohol consumption in Australia is a significant contributor to rates of male mortality, particularly among young adults, suggesting an interaction between per-capita consumption and risky episodic drinking. The policies aiming to reduce population-level alcohol consumption and episodic risky drinking have the potential to substantially improve population-health outcomes in Australia, particularly among young men.
The weaknesses of using ecologic data for judging health effects: This paper is an ecologic comparison of calculated population per-capita alcohol consumption with total mortality; it is based on aggregate data, and is not based on alcohol exposure data for individuals. Such studies are generally considered to be exploratory studies, and may be useful for generating hypotheses; however, such hypotheses must then be tested for validity in more rigorous approaches. Unfortunately, the authors of this paper seem to think that their ecologic comparison study has answered key questions about alcohol and public health.
As described in detail by Morgenstern in Modern Epidemiology,(1) there are serious problems in using aggregate historical data from a population to estimate health outcomes. He describes an inability to adjust for bias (which may increase, decrease, or even reverse the apparent risk ratio found in the analyses); unfortunately, the conditions that produce such bias cannot be checked without having individual (not group) associations. Morgenstern also describes problems from controlling for confounding, which is difficult to do in ecologic studies. Within-group misclassification (everyone is not exposed to alcohol during a period when the per-capita intake may be higher or lower) causes further problems. There are statistical approaches for dealing with these difficulties in individual-data studies, but they do not work with ecologic comparisons. While non-differential misclassification of exposure nearly always leads to bias toward the null in analyses conducted at the individual level, this principle does not hold when the exposure variable is an aggregate measure. Finally, Morgenstern points out additional problems in the analysis of ecologic comparisons due to lack of adequate data, temporal ambiguity, and collinearity. (1)
Morgenstern concludes his discussion: “Ecologic analysis poses major problems of interpretation when making ecologic inferences and especially when making biologic inferences. In contemporary epidemiology, the ‘ecologic fallacy’ reflects the failure of the investigator to recognize the need for biologic inference and thus for individual-level data. This need arises even when the primary exposure of interest is an ecologic measure and the outcome of interest is the health status of entire populations.”(1)
Modifiers of effect of population changes in per-capita alcohol consumption: An increase in per-capita consumption of alcohol can occur from heavy and/or binge-drinkers consuming more (which would be expected to increase adverse health effects) or from non-drinkers or occasional drinkers beginning to consume light-to-moderate amounts (which might be expected to improve the health of the population). Those who die may not be those who drink alcohol. Without knowing the reasons for changes over time in the calculated per-capita consumption, the implications of associations with health outcomes are difficult to interpret.
Added Forum reviewer Orgogozo, “I read this paper and was impressed by its lack of rigor, the few number of variables tested, and the apparent ignorance of the main confounders (diet, level of education, occupation during life, socio-economic status, etc.) which influence both drinking behaviors and risk of death. For those and other reasons, ecological epidemiology is often described as uninformative at best and misleading at worst. Not too difficult to decide in this case.”
The interpretation of the results of the present paper: Despite the numerous potential problems described above, the authors nevertheless have carried out very extensive analyses, with many assumptions, to support an argument that their data should be used to change national alcohol policy in Australia. This is strongly stated even though they admit that they have no data on drinking patterns of individuals or on most social and economic factors operating in each time period (e.g., consumption of home-brewed beverages, educational levels of the population during the period, disposable income during the period, the number of automobiles being driven at the time, etc.). They find (as expected) that effects of mortality rates for the young are higher when the young are consuming more alcohol, but do not emphasize the minimal effects seen among the elderly, when most deaths occur.
Other comments of Forum reviewers: One Forum reviewer comments: “It seems that their analysis of mortality over the last 90 years takes into account only four factors: alcohol, smoking, the 1911 flu epidemic and the ‘war years.’ The authors infer that by ignoring all improvements in public health or medical care, changes in diet or living standards, etc., it is possible to correlate alcohol consumption with higher mortality. I am sure that with a similar approach, carefully avoiding the inclusion of major factors that drive mortality rates, it would be possible to show that shorter hair styles or tie width in men’s fashion is responsible for increased mortality in that population, and the authors would call for higher taxes on haircuts or tax ties by the millimeter.”
Reviewer Skovenborg makes several observations that do not appear to have been taken into consideration by the authors of this paper:
1. Binge-drinking by young men is associated with a high prevalence of other health risk behaviors.(2)
2) In recent years per capita alcohol consumption among young people in Sweden has deceased while at the same time alcohol-related hospitalizations has increased sharply. Changes in per capita consumption can hide significant shifts in the drinking habits of heavy drinkers, and alcohol policies appear to influence the drinking behaviour of some individuals (like moderate drinkers) more than others (like heavy drinkers).(3)
3) A recent analysis of the potential role of alcohol control and policy measures and the current pattern of drinking and drunkenness among adolescents in 40 European and North American countries found that stronger alcohol controls were associated with a lower proportion of weekly drinking but a higher proportion of drunkenness. Strict policies might in fact lead to rebellion and increased prevalence of binge drinking.(4)
- Morgenstern H. Ecologic Studies. In Rothman KJ, Greenland S (eds), Modern Epidemiology. 2nd edition. 1998. Philadelphia: Lippincott, Williams & Wilkins.
- Hallgren M, Leifman H, Andréasson S. Drinking less but greater harm: Could polarized drinking habits explain the divergence between alcohol consumption and harms among youth? Alcohol and Alcoholism 2012;47:581-590.
- Miller JW, Naimi TS, Brewer RD, Jones SE. Binge drinking and associated health risk behaviors among high school students. Pediatrics 2007;119:76-85.)
- Gilligan C, Kuntsche E, Gmel G. Adolescent drinking patterns across countries: Associations with alcohol policies. Alcohol and Alcoholism 2012;47:732-737.
An ecologic analysis from a group in Australia has related trends in the per-capita consumption of alcohol in Australia with rates of all-cause mortality. Unfortunately, the authors do not have data on individual consumption of alcohol (and no way of knowing if the people who drink or do not drink are those who die) and do not have data on many key factors that relate to both alcohol consumption and mortality (e.g., education, income, diet, occupation, etc.). Especially, they have no data on the patterns of consumption of members of the population (regular moderate intake or binge drinking). They “adjust” for death rates in women but do not report sex-specific mortality rates.
This paper illustrates many of the dangers of the so-called “ecologic fallacy,” in which comparisons are made between aggregate population data and biologic outcomes. It has been shown that such comparisons may be of interest in generating hypotheses, but not for answering questions of causation, for which different types of studies are required. As stated in a leading textbook of epidemiology: “Ecologic analysis poses major problems of interpretation when making ecologic inferences and especially when making biologic inferences. In contemporary epidemiology, the ‘ecologic fallacy’ reflects the failure of the investigator to recognize the need for biologic inference and thus for individual-level data. This need arises even when the primary exposure of interest is an ecologic measure and the outcome of interest is the health status of entire populations.”(1)
With such serious limitations from the use of ecologic comparisons, this paper does not provide data upon which alcohol policy can be based. The authors of this paper appear to have ignored previous research providing very different interpretations of an inverse association between per-capita alcohol intake and mortality. A plethora of studies based on alcohol intake among individuals, and especially patterns of consumption, provide data needed for such policy decisions.
* * *
Comments on this paper have been provided by the following members of the International Scientific Forum on Alcohol Research:
Yuqing Zhang, MD, DSc, Epidemiology, Boston University School of Medicine, Boston, MA, USA
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark
Andrew L. Waterhouse, PhD, Marvin Sands Professor, Department of Viticulture and Enology, University of California, Davis; Davis, CA, USA
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Jean-Marc Orgogozo, MD, Professor of Neurology and Head of the Neurology Divisions, the University Hospital of Bordeaux, Pessac, France
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Fulvio Ursini, MD, Dept. of Biological Chemistry, University of Padova, Padova, Italy;
Pierre-Louis Teissedre, PhD, Faculty of Oenology – ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA