Summary: In 2008, the NHMRC commissioned the Dieticians Association of Australia to undertake systematic literature reviews to support the revision of the Dietary Guidelines for Australians. The primary aim was to undertake a series of systematic reviews of the national and international literature from the year 2002 on the food-diet-health-disease inter-relationship for different population subgroups. One of the 29 sections in the report (pp 613-678) covered the evidence for the risks and benefits of alcohol drinking. This critique is only of the alcohol section.
On balance the authors should be commended on their work. Our criticisms are primarily about the restriction in the time frame of the review, and minimal consideration of key nuances, such as quantity of alcohol consumed and its relation to risk or benefit, patterns of drinking, the balance between risks and benefits, and making statements at all when the evidence is too weak. More details (e.g. considering cardiovascular mechanisms to widen the search) would have added weight to the conclusions.
1.The search methodology used to create this review is well done and fits with best practice.
2.The authors need to be commended on adopting a balanced view toward alcohol by considering both positive and negative health effects of alcohol.
3.This lack of overt bias may have been due to the fact that all reviewers appear to have been dieticians. That may of course have introduced an inherent unknown bias from the fact that no other health disciplines had input into the process.
4.An unintended bias may have been introduced by the reliance on the 2007 WCRF/ACIR study. Although the methodology and process may have appeared the same, without the reviewers of this study's direct involvement there is always a risk that that assumption is incorrect.
5.The review only considered literature from 2002 to 2009. Although this mostly applied to all conditions and should be applauded for consistency purposes, it has ignored seminal studies from outside that time frame that alter the quality of evidence grade and evidence statement.
6.Evidence quality grades of D and possibly C suggest that the evidence for those statements may be so unreliable that it would be better to make no evidence statement at all for those conditions.
7.We note that only health benefits and risks were considered. Alcohol is a complex subject and this review misses commentary on social benefits and risks.
8.Some of the evidence statements are too simplistic. They ignore the fact that some benefits or risks of alcohol may only become relevant at high alcohol use levels, that there is a balance between risk and benefit that for any condition has to be considered in the context of any individuals non alcohol based risk factors, and that in some conditions the benefit or risk is very small in absolute terms. Indeed, throughout the text there is no clear delineation between the effects of light-to-moderate and heavy alcohol consumption, especially regarding increased risks of adverse health effects the text refers throughout to "alcohol use or alcohol consumption", where amount and pattern are not considered.
9.A dual relationship exists between alcohol consumption and diabetes mellitus. Light to moderate drinking may be beneficial while heavy drinking is detrimental. (Baliunas et al., 2009 and cited in the World Health Organisation Global Status report on Alcohol and Health 2011). The grade of D for the evidence statement about alcohol and diabetes assigned in this review appears incorrect in light of these references and may reflect the fact that predominantly the review does not cover literature reported after 2009.
10.The authors appear to have assumed any cardioprotective effect of the alcohol component was due only to the reduction in the concentration of HDL cholesterol observed following alcohol consumption. This could argue their search strategy for the cardioprotective effect of alcohol was too narrow as alcohol is also observed to produce a cardioprotective effect by a reduction in platelet stickiness and in the production of blood clotting proteins that reduces the potential for blood clots to form, as well as facilitating the breakdown of blood clots that have already formed. In addition alcohol has an endothelial anti-inflammatory effect on for example, C-reactive protein (CRP). (Rimm E et al 1999. Wannamethee S et al 2003. Booyse F et al 2007. Pai J et al 2006. Imhof A et al 2001.)
11.The evidence quality for risk of oesophageal cancer, especially in heavy alcohol drinkers appears correctly rated as B. The evidence quality for risk of oropharyngeal cancer, again especially in heavy alcohol drinkers, appears to be incorrectly rated as C, and should be rated as B. Again this review has not considered more recent literature, (World Health Organization International Agency for Research on Cancer (IARC). 2012)
12.There is a balance between the risks and benefits of alcohol for any individual. The benefits appear to outweigh the risks for light to moderate drinkers older than the mid 50s. The risks (predominantly due to accidents and binge patterns of drinking both of which were not considered in this review) outweigh the benefits in youth. (Connor et al 2005).
|Contact: R. Curtis Ellison |
Boston University Medical Center