Re: Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis
Based on meta-analysis Ronksley and colleagues  asserted that the association between moderate alcohol consumption and reduced mortality risk was "beyond question". A companion paper  concluded that strong evidence for underlying biological mechanisms warranted refocusing research towards investigating how best to communicate health benefits to the general population.
We reviewed all of the 67 studies which generated the 84 articles in the Ronksley meta-analysis. All but two suffered from at least one of six serious methodological problems (see Figure 1 and also online supplementary material at http://www.carbc.ca including bibliography].
Figure 1 here.
Problem 1: A failure to control for confounding lifestyle factors
Naimi et al  reported that moderate drinkers were significantly healthier than abstainers on 27 risk factors for heart disease unrelated to drinking. We lowered the bar and allowed all but four studies which failed to assess a minimum set of age, smoking status and health status.
Problem 2: Baseline drinking assessed over too short a time period
To assess an individual's pattern and level of drinking in order to predict future health outcomes, it is necessary to enquire about alcohol consumption over a reasonably long period. We excluded 4 further studies which failed to assess drinking at baseline over a period of at least 30 days.
Problem 3: Failure to assess both quantity and frequency of consumption
All studies included in the Ronksley et al  meta-analysis were required to provide estimates of average daily intake of ethanol, either in terms of grams or other "units" of ethanol. However, 10 of the remaining studies failed to adequately assess either frequency or quantity of typical drinking (or only assessed quantity for frequent drinkers).
Problem 4: Inclusion of former drinkers in abstainer reference group
The methodological problem of including former drinkers in the abstainer reference group is widely recognised  (the "sick quitter" hypothesis). Former drinkers often abstain because of ill-health and therefore make the health status of moderate drinkers look good by comparison . While Ronksley et al adjusted the estimates in some studies which failed to separate lifetime abstainers from former drinkers, there were many (n=32) such studies and other serious errors remained. Only 22 studies met this criterion in the entire sample, of which 17 also met the previous criteria. This bias is not restricted to heavy or problem drinkers since moderate drinkers also sometimes quit drinking when their health deteriorates .
Problem 5: Inclusion of occasional drinkers in the abstainer reference group
The former drinker hypothesis described above is generally attributed to Shaper and colleagues . Shaper et al also noted that people do not only abstain with age and increasing frailty, they may also just reduce consumption. Many studies have definitions of abstention which permit the inclusion of occasional drinkers, a practice which may also bias towards finding apparent health benefits from light to moderate drinking . Defining occasional drinking as drinking less than one 12g drink per week, a further 7 studies failed to meet this criterion.
Problem 6: Failure to report results for occasional drinkers separately from moderate drinkers
It has also been suggested that occasional drinkers might actually have enhanced health status, not as a result of any effects of alcohol but through a constellation of other health protective factors . Only two of the remaining studies minimised potential bias by reporting results for occasional drinkers separately both from abstainers and light or moderate drinkers [6,7]. One found protective effects from moderate drinking in relation to CVD but not CHD . The other found evidence for reduced CHD and CVD mortality but only for women with high alcohol consumption .
The problems identified in this literature are many and serious. Most studies had two or more of these problems plus other issues were identified. For example, only two of the 67 unique studies made efforts to control for the effects of episodic heavy drinking which is an oversight since benefits do not accrue to individuals with an occasional heavy drinking pattern even if their average daily consumption is moderate .
While Ronksley et al  identified the 84 best published articles relevant to the issue of moderate drinking and health, because of the poor quality of the studies it is premature to draw firm conclusions. We suggest there are still strong competing hypotheses to explain the association of health benefits with moderate drinking which have yet to be discounted - in particular, the possibility of uncontrolled confounding from other lifestyle factors . This possibility is also supported by meta-analyses finding biologically implausible benefits from moderate alcohol consumption such as protection against liver cirrhosis  and birth defects . We hope that by drawing attention to the shortcomings in this literature future studies may avoid these errors and provide a clearer answer to the issue of moderate drinking and health benefits.
1. Ronksley R, Brien S, Turner B, Mukamal K, Ghali W. (2011) Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis. British Medical Journal, 342:doi:10.1136/bmj.d671.
2. Brien S, Ronksley P, Turner T, Mukamal K, Ghali W. (2011) Effect of alcohol consumption on biological markers associated with risk of coronary heart disease: systematic review and meta-analysis of interventional studies. British Medical Journal 342:d636.
3. Naimi T, Brown D, Brewer R, et al. (2005). Cardiovascular risk factors and confounders among nondrinking and moderate drinking U.S. adults. American Journal of Preventive Medicine 28:369–373.
4. Fillmore K, Stockwell T, Chikritzhs T, Bostrom A. (2006). Moderate Alcohol Use and Reduced Mortality risk: Systematic Error in Prospective Studies and New Hypotheses. Annals of Epidemiology 95:135-146.
5. Shaper AG, Wannamethee SF. (1998) J-shaped curve and changes in drinking habit. In Chadwick, DJ & Goode JA, editors. Alcohol and Cardiovascular Diseases: Novartis Foundation Symposium 216. New York: John Wiley and Sons. pp 173–192.
6. Murray R, Connett J, Tyas S, Bond R, Ekuma O, Silversides C, Barnes G. (2002). Alcohol volume, drinking pattern, and cardiovascular disease morbidity and mortality: Is there a U-shaped function? American Journal of Epidemiology, 155, 3:242-248.
7. Harriss LR, English DR, Hopper JL, Powles J, Simpson JA, O’Dea K, et al. (2007). Alcohol consumption and cardiovascular mortality accounting for possible misclassification of intake: 11-year follow-up of the Melbourne Collaborative Cohort Study. Addiction 102:1574-85.
8. Roerecke M, Rehm J. (2010) Irregular Heavy Drinking Occasions and Risk of Ischemic Heart Disease: A Systematic Review and Meta-Analysis. American Journal of Epidemiology (2010) doi: 10.1093/aje/kwp451. First published online: February 8, 2010•
9. Rehm J, Taylor B, Mohapatra S, Irving H, Baliunas D, Patra J, Roerecke M, Rehm J. (2010) Alcohol as a risk factor for liver cirrhosis – a systematic review and meta-analysis. Drug and Alcohol Review 29, 4:437-445.
10. Henderson J, Gray R, Brocklehurst P. (2007). Systematic review of effects of low-moderate prenatal alcohol exposure on pregnancy outcome. British Journal of Obstetrics and Gynaecology 114,3:243–252.
Competing interests: No competing interests