Science and Technology Committee HC 1536 Alcohol GuidelinesSupplementary written evidence submitted by Dr Richard Harding (AG 13a)
1. At the oral evidence session on Wednesday 12 October, a number of issues arose on which I would like to comment further.
The Cardio-Protective Effect of Moderate Alcohol Consumption
2. There are two separate bodies of evidence for the cardio-protective effect of moderate alcohol consumption. These are observation studies, based on observation of subjects over time, and the effect of alcohol consumption on biological markers for coronary heart disease in humans and in animals.
The observational studies
3. Professor Heather has drawn my and the Committee’s attention to the work of Professor Kaye Fillmore, and a further paper by Tanya Chikrtzhs, a co-author of Professor Fillmore’s paper.
4. When evidence for the cardio-protective effect emerged in the 1970s and 1980s, Shaper1 suggested that the considerable apparent benefit conferred by moderate alcohol consumption compared with abstainers could be explained by the abstainer category containing persons terminating or decreasing their alcohol consumption to very occasional drinking—in other words, previous heavy drinkers. Further, as people age and become ill or frail or increase use of medications, their alcohol consumption decreases, some abstaining altogether. If these people are included in the abstainer category, then it is not the absence of alcohol that is elevating their risk of coronary heart disease (CHD), but, rather, their compromised health. This is the “sick quitter” hypothesis.
5. This hypothesis has been taken seriously by the epidemiological community. Some analysts strove to correct for it by excluding ex-drinkers from the abstainers category, and by including only life-long abstainers. A meta-analysis2 of seven studies concluded that the “sick-quitter” hypothesis had been eliminated, and by the mid-1990s the hypothesis was relegated, by most, to history.
6. About 10 years later Professor Fillmore revisited this hypothesis, and re-examined the classification of “abstainers” in 54 epidemiological studies for two characteristics, which she called “errors”. These were:
inclusion of former drinkers (no drinks in the past year) in the abstainer category, and
inclusion of occasional drinkers (one drink a month or less) in the abstainer category.
Professor Fillmore’s concerns were that former drinkers might include those whose health had been damaged by alcohol and had given up, and occasional drinkers may include those who have decreased their consumption because of age or infirmity. Only two studies relating to CHD of were found free of these characteristics, and these remaining two studies found no significant cardio-protective effect. It is impossible to draw any general conclusions from two studies.
7. I was present at a scientific meeting in 2006 when Professor Fillmore presented her work. In the report of the following discussion (which has been published),3 other workers had no concerns about including occasional drinkers in the abstainer group, since they had found that these drinkers had the essentially the same risk as lifetime abstainers. There was agreement that it was inappropriate to use “sick quitters” as controls.
8. In her paper4 that was published following this meeting, Professor Fillmore stated that her own group’s position on the health benefits of alcohol resulted from extensive discussion among themselves and what seemed to be a herculean effort to eliminate—rather than prove—the “sick quitter” hypothesis. On the basis of fine contributions of laboratory science demonstrating plausible and real mechanisms for cardiac protection, their conclusion is that alcohol (among other substances, lifestyles and behaviours) conveys benefit to the heart. But, she said, the lot falls to epidemiology to demonstrate whether the population actually does benefit.
9. Many epidemiological papers published since 2006 had data permitting them to include only lifetime abstainers in the reference category, and have found a marked and significant cardio-protective effect for those drinkers in the light to moderate category. Fuller5 re-examined the data from a major US study in a way that avoided the potential pitfalls described by Fillmore, and found a protective effect for moderate consumption compared to abstainers. A recent comprehensive meta-analysis by Ronksley et al6 in the British Medical Journal describes in detail how earlier objections have been dealt with, and concluded that consumption levels up to 15g/day (nearly two UK units) were associated with about 25% reduction in risk of cardiovascular disease mortality. It seems to be clear that epidemiology has demonstrated that the population actually does benefit.
Effect of alcohol consumption on biological markers for coronary heart disease
10. As I mentioned when giving oral evidence, at the time of the 1995 Report on Sensible Drinking, the biological markers accepted at the time were essentially the increase in high density lipoprotein cholesterol (HDL-C) and the beneficial effect on clotting factors. There have since been many hundreds of experimental studies showing numerous other mechanisms by which moderate alcohol consumption lowers the risk of atheroscerosis and coronary heart disease. A recent review in the British Medical Journal by Brien et al7 described 21 of these mechanisms. The conclusion of this study is there is compelling, indirect evidence in support of a causal protective effect of alcohol for coronary heart disease. Another recent review by Collins et al.8 on the effect of alcohol consumption on biomarkers concluded that there is convincing evidence that light-moderate, non-binge alcohol intake reduces the risk of CHD.
11. One of the most important of these mechanisms remains the elevation of HDL-C. It is important because the ratio of low density lipoprotein cholesterol (LDL-C) to HDL-C a much better indicator of CHD risk than total cholesterol concentration. HDL-C removes LDL-C from the bloodstream and transports it to the liver, so a high level of HDL-C is desirable. People with high level of total cholesterol but favourable ratio have a lower risk of disease than those with a low level of total cholesterol and poor ratio. Apart from choosing your own grandparents, there are essentially two lifestyle changes anyone can make which would increase HDL-C—taking exercise and drinking alcohol. A physical activity level equivalent to 60 minutes/day of leisure time activity (walking, climbing stairs, jogging) raises HDL-C by about 7%, whereas up to 7–14 drinks/week raises it by about 20% 9 in men. With a rise in HDL-C of that magnitude, it would be astonishing if moderate alcohol consumption did not markedly affect CHD risk.
The Chikritzhs Paper
12. This paper raises four concerns:
misclassification error—already commented on above;
confounding;
self-report, recall bias and drinker “drift”; and
drinking patterns.
13. Confounding. The argument here is that those who drink alcohol in moderation also are likely to be those who also live healthy lifestyles, and therefore it is to expected that they will have lower levels of cardiovascular disease compared to abstainers. What is more, it is impossible to correct for all possible confounding factors.
14. This point is essentially about causation, ie epidemiological studies only identify associations, and can never establish causation. It is possible that other factors, or combination of factors, can explain the apparent protection observed. Bradford-Hill identified10 a number of factors which should be considered when assessing whether an association is causative. I quote directly from the recent paper by Ronksley et al,6 already cited above, on this point.
15. “The lingering question is whether this association is causal. Clearly, observational studies cannot establish causation. However, when the present results are coupled with those from our companion review paper7 summarising intervention mechanistic studies focusing on biomarkers associated with cardiovascular disease, the argument for causation becomes more compelling. Indeed, the mechanistic biomarker review shows biological plausibility for a causal association by showing favourable changes in pathophysiologically relevant molecules.
Therefore, we can now examine the argument for causation based on Hill’s criteria.10 Beyond the biological plausibility argument discussed above, there is an appropriate temporal relation with alcohol use preventing cardiovascular disease. Secondly, we have observed a greater protective association with increasing dose, except that it seems to be offset somewhat by negative associations with the risk of haemorrhagic stroke. Thirdly, the protective association of alcohol has been consistently observed in diverse patient populations in both men and women. Fourthly, the association is specific: moderate drinking (up to one drink or 12.5g alcohol per day for women and two drinks or 25g alcohol per day for men) is associated with lower rates of cardiovascular disease but is not uniformly protective for other conditions such as cancer. Lastly, the reduction in risk is notable even when controlling for known confounders (such as smoking, diet and exercise). Any potential unmeasured confounder would need to be very strong to explain away the apparent protective association.”
16. Finally, some observational studies can isolate those in their study population those with the healthiest combination of lifestyles. In my first submission (paragraph 30, final bullet), I pointed out that moderate alcohol consumption is now identified as one element of a healthy lifestyle (along with lifetime non-smoking, a healthy diet and physical activity) by researchers who are exploring the concept of “successful ageing”. The mortality risk for those with all four lifestyle characteristics was significantly lower than who had three, but were abstainers. Others have reported that for the most healthy members of society, moderate alcohol consumption can make them healthier. Mukamal11 explored exactly this issue, and found that for those who were already at a low risk of cardiovascular disease (on the basis of body mass index, physical activity, smoking and diet), moderate alcohol consumption lowered the risk much further. Joosten et al12 reported, “In subjects already at lower risk of type 2 diabetes on the basis of multiple low-risk lifestyle behaviors, moderate alcohol consumption was associated with an approximately 40% lower risk compared with abstention.” So among the healthiest people, those who consume alcohol moderately have much better health outcomes, and the major confounding factors are therefore not playing a role.
17. Self-report, recall bias and drinker “drift”. Finding out how much alcohol people drink is really quite difficult. It is true that people generally under-estimate their consumption, and the best that researchers can do is rely on questionnaires. As Professor Fillmore says,4 “It should not be forgotten that epidemiology deals with crude approximations, beset by confounding—often unmeasured—especially when single estimates or limited measurements are used.” Yet despite these inbuilt sources of inaccuracy, which would tend to weaken any protective effect, studies around the world consistently find that the cardio-protective effect of moderate consumption is still apparent.
18. Drinking patterns. Yes, drinking patterns are very important to health outcomes, and there will be great variation among the drinkers in any study. In studies that have had data on the pattern of drinking, subjects consuming alcohol on a regular basis consistently have better health outcomes than subjects who consume alcohol only on one or two days per week, usually in a binge-drinking pattern. But again, despite lack of information on the pattern of drinking in many other studies, the protective effect of alcohol is still apparent from them.
Other Points that arose in the Oral Evidence Session
The Australian approach
19. I have seen the Australian Alcohol Policy Coalition Position Statement on Cancer, Cardiovascular Disease and Alcohol Consumption, September 2011.13 I realise that the Coalition is only concerned with reducing the level of alcohol misuse in Australia, but I am afraid that in my view it is not the result of a sound, balanced and up to date appraisal of the science, but is misleading and a misrepresentation of the actual position. As such is it not a proper basis for policy.
Q11
20. I gave an unsatisfactory response to Pamela Nash’s question, “Would you advocate the guidelines being reinforced by the evidence base?” I said I would not change them, which is true, but I did not explain why. It is not really a question of the guidelines being reinforced by the evidence base, because in the 1995 Sensible Drinking report, the guidelines were transparently argued from the evidence base. The evidence base has grown over time, and so the question is whether any new evidence changes the rationale for the guidelines. My view is that the current evidence base has reinforced the existing guidelines rather than undermined them, hence my answer.
21. The current guideline range of consumption, 1–4 units/day for men and 1–3 units/day for women, coincided broadly with the bottom of the all-cause mortality curve. For men over 40 and post-menopausal women, risk increases therefore for consumption on either side of this range, both higher and lower. Hence also the advice for these people to consider drinking 1–2 drinks/day if they wished to take advantage of the beneficial effects. For younger people, provided the anti-intoxication messages were respected and alcohol was consumed responsibly, there did not appear to be any significant health risk which would accrue at this level of consumption.
Q13
22. In an answer on how messages should be conveyed to the public, Professor Heather advised that there should be both daily and weekly limits that should never be exceeded, and advocated a return to the old weekly limits. The weekly “limit” of 21 units for men and 14 units for men were the UK Guidelines before the 1995 Report. The basis for these limits was never clear to me. It was never clear to me. The Joint Royal Colleges Report published in June 1995, which I cited in my first submission, was entitled, “Alcohol and the Heart in Perspective, Sensible Limits Confirmed”, and yet, despite the title, nowhere in that report was I able to find any justification for limits of 21 and 14 units/week for individuals. But it is clear from the Joint Royal Colleges Report that the concern was the need to reduce the mean level of consumption, and, although acknowledging the cardio-protective effect, advised against raising the “sensible limits” because there would be an adverse effect on public health.
23. As the summary of this Report explains on page 1, if the mean increases, the proportion of people drinking in the higher risk categories is likely to increase, with a consequent increased risk of alcohol associated harm. This is the so-called “whole population theory”, and was addressed in the 1995 Report on Sensible Drinking Report in paragraphs 9.2–9.4. Essentially it requires the whole population to conform to a mean level of consumption in order that there are fewer drinkers at the higher end of consumption, which is where most of the misuse is.
24. The problem with this is that public health messages on alcohol are addressed to populations, but received by individuals, and those individuals are encouraged to believe that following this advice will benefit their health. But this is not necessarily the case if the purpose of the advice is to peg individual consumption to a particular population mean so that other people, towards the end of the graph, are less likely to misuse alcohol. It seems to me that this is manipulative, and in my view any advisory level of intake justified on this basis is politically indefensible. That is why, in my response in oral evidence to Question 2 on the purpose of alcohol guidelines, I said that the guidelines should be meaningful to adults as individuals.
Alcohol and cancer
25. In her response to Question 11, Dr Morgan cited the International Agency for Research on Cancer Report on Alcohol Consumption, and said that it included a statement that there is clear evidence of about 10% increase in the risk of cancer for every 10g of alcohol/day. I believe that she meant to say breast cancer, rather than all or any other cancer. I could not find that statement in the IARC Monograph, but the data presented there seems to suggest the figure is in the 5%–10% range for breast cancer. On page 1271, it says, “for regular consumption of 18g/day, the relative risk is still significantly increased at 1.13”, which equates to 7% per 10g/day.
26. I would like to draw the Committee’s attention to the transcript14 of an interview given on the BBC Radio 4 programme “More or Less” in May 2009 with Professor David Spiegelhalter, Winton Professor of the Public Understanding of Risk at Cambridge University. He was talking about alcohol and cancer in the context of the Million Women Study,15 and I drew in part from his comments in my oral evidence. It demonstrates how important (and difficult) it is to gain a proper perspective on the overall effect of alcohol on public health, rather on specific health outcomes, and not be swayed by how the science is reported, even sometimes by the researchers themselves.
19 October 2011
References
1. Shaper A G, Wannamethee G, Walker M. Alcohol and mortality in British men: Explaining the U-shaped curve. Lancet 1988; 2:1267-1273
2. Maclure M. Demonstration of deductive meta-analysis: Ethanol intake and risk of myocardial infarction. Epidemiol. Rev. 1993; 15:328-335
3. Panel Discussion I: Does alcohol consumption prevent cardiovascular disease? Proceedings of an international conference. Annals of Epidemiology 2007;17: S37-S39.
4. Fillmore, K M, Stockwell, T, Chikritzhs T, Bostrom, A and Kerr, W. Moderate Alcohol use and reduced mortality risk: Systematic error in prospective studies and new hypotheses. Annals of Epidemiology 2007; 17: S16-S23
5. Fuller TD. Moderate alcohol consumption and the risk of mortality. Demography 2011. DOI 10.1007/s13524-011-0035-2
6. Ronksley P E, Brien S E, Turner B J, and Mukamal, K J. Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis. British Medical Journal 2011; 342:d671; doi:10.1136/bmj.d671
7. Brien S E, Ronksley P E, Turner B J, Mukamal K J, and Ghali A G. Effect of alcohol consumption on biological markers associated with risk of coronary heart disease: systematic review and meta-analysis of intervention studies. British Medical Journal 2011; 342:d636; doi: 10.1136/bmj.d636
8. Collins M A, Neafsey E J, Mukamal K J, Gray M O, Parks D A, Das K D, and Kortuis R J. Alcohol in moderation, cardioprotection, and neuroprotection: epidemiological considerations and mechanistic studies. Alcoholism: Clinical and Experimental Research 2009; 33(2):206-219. DOI: 10.1111/j.1530-0277.2008.00828x
9. Ellison R C, Zhang Y, Qureshi M M, Knox S, Arnett D K, and Province M A. Lifestyle determinants of high-density lipoprotein cholesterol: The National Heart, Lung and Blood Institute Family Heart Study. American Heart Journal 2003; 147: 529-535, DOI: 10.1016/j.ahj.2003.10.033
10. Bradford-Hill A. The environment of disease : association or causation? Proceedings of the Royal Society of Medicine. 1965; 58:295-300
11. Mukamal K J, Chiuve S E, Rimm E B. Alcohol consumption and risk for coronary heart disease in men with healthy lifestyles. Archives of Internal Medicine 2006; 166: 2145-2150
12. Joosten M M, Grobbee D E, van der A D L, Verschuren W W M, Hendriks H F J, Beulens J W J. Combined effect of alcohol consumption and lifestyle behaviors on risk of type 2 diabetes. Am J Clin Nutrition, published on-line 21 April 2010. doi:10.3945/ajcn.2010.29170
13. Alcohol Policy Coalition Position Statement. Cancer, Cardiovascular Disease and Alcohol Consumption. September 2011
14. Transcript of an interview with Professor David Spiegelhalter, “More or Less”, BBC Radio 4, 9 May 2009
15. Allen N E, Beral V, Casabonne D, Kann S W, Reeves G K, Brown A, Green J. Moderate alcohol intake and cancer incidence in women. Journal of the National Cancer Institute 2009; 101(5): 296-30