Critique 272 – New perspectives on how to formulate alcohol drinking guidelines


Shield K; Paradis C; Butt B; Naimi T; Sherk A; Asbridge M; Myran D; Stockwell T; Wells S; Poole N; Heatley J; Hobin E; Thompson K; Young M and the Low-Risk Alcohol Drinking Guidelines Scientific Expert Panel


Addiction (2023)

Author’s Abstract

Background Low-Risk Alcohol Drinking Guidelines (LRDGs) aim to reduce the harms caused by alcohol. However, considerable discrepancies exist in the ‘low-risk’ thresholds employed by different countries.

Argument/analysis Drawing upon Canada’s LRDGs update process, the current paper offers the following propositions for debate regarding the establishment of ‘low-risk’ thresholds in national guidelines: (1) as an indicator of health loss, years of life lost (YLL) has several advantages that could make it more suitable for setting guidelines than deaths, premature deaths or disability adjusted years of life (DALYs) lost. (2) Presenting age-specific guidelines may not be the most appropriate way of providing LRDGs. (3) Given past overemphasis on the so-called protective effects of alcohol on health, presenting cause-specific guidelines may not be appropriate compared with a ‘whole health’ effect derived from a weighted composite risk function comprising conditions that are causally related to alcohol consumption. (4) To help people reduce their alcohol use, presenting different risk zones associated with alcohol consumption instead of a single low risk threshold may be advantageous.

Conclusions National LRDGs should be based on years of life lost and should be neither age-specific nor cause-specific. We recommend using risk zones rather than a single drinking threshold to help people assess their own risk and encourage the adoption of behaviours with positive health impacts across the alcohol use spectrum.

Forum comments

Another paper has been published in the ‘Opinion and Debate’ section of the journal Addiction. The paper basically defends the strategy followed for the proposed updated Canadian low-risk (zero alcohol) drinking guidelines and proposes their strategy as a guideline for all national alcohol drinking guidelines (Shield et al. 2023). Following up on this proposal, this critique intends to overview and discuss the various strategies available to develop (dietary) guidelines and debate that outlined in the paper.

In general, public health recommendations on alcohol consumption are being developed to provide recommendations on safe and responsible consumption. A methodologically rigorous process that involves systematic literature reviews, evidence grading and a transparent guideline development process should provide evidence-based recommendations that promote responsible and safe alcohol consumption. The last several decades alcohol drinking guidelines recommend lower quantities of alcohol to be consumed. This may be related to new scientific evidence, in other words a new science-base, or to a changed guideline development process. At least it seems that guideline development processes are changing.

Acceptable risk. Guidelines for alcohol and other dietary components were based on the concept of acceptable risk. Knowing that all foods consumed will have inherent adverse effects (it is all in the dosage as Paracelsus remarked[1]) and benefits (we need to and we love to eat and drink) it is considered best to advice consumption with overall minimal adverse effects for the majority of the population. Such benefit may be best defined by the parameter ‘overall mortality’ since this measure includes adverse effects and benefits, both those known and unknown. Consequently, a consumption level of one to two glasses (10-20 g alcohol) daily for women and two to three glasses (20-30 g alcohol) daily for men is associated with an overall reduced mortality risk. Such an advice corresponds to what people experience: drinking in moderation (and drinking responsibly) is enjoyed and will most likely not negatively affect the health of oneself and others.

Optimal health. Some countries have been using the concept of optimal health. Leaving behind the concept of acceptable risk, it is proposed that all would have to strive for optimal health. Therefore, those specific quantities of food items are being assessed and advised which are associated with the lowest risk for the most devastating diseases in the population. This led into the recommendation of one optimal dosage minimizing health care burden. In the case of alcohol consumption, it was determined in The Netherlands that 5 g alcohol, so half a glass on average per day, was ideal for that purpose. The interpretation of that notion was that nobody would drink half a glass (also not on average) and that, therefore, the advice would have to be ‘do not drink’.

No risk. The last concept is a concept used in toxicology and the proposed updated Canadian drinking guidelines: the concept of ‘no risk’. In this concept all short-term and long-term adverse and beneficial effects of alcohol consumption are considered, weighed and modelled to generate an overall risk of alcohol consumption. The resulting approach consists of a very elaborate construction with (badly described) weighing factors and mathematical modelling ultimately showing what was intended: there is always a risk when drinking alcohol. This concept has been applied to drinking guidelines only so far, not to any other dietary component, which is logical since, in this specific context, alcohol consumption is not considered a food item but a psychoactive substance.

Hence, it seems that at least the guideline development processes are changing from an overall acceptable risk expressed as no effect or a reduced risk for overall mortality (Bryazka et al., 2022) into obtaining optimal health for the most health care burdening diseases (Food-based dietary guidelines of The Netherlands) and into no risk at all for any possible short-term and long-term alcohol consumption (Canada’s Guidance on Alcohol and Health).

Perhaps all these guideline concepts are far too complex to get a simple message across. Simple messages are proposed to be key to the acceptability of public health advice. The simple message could be ‘when drinking alcohol, drink in moderation’; moderation being one to two glasses (10-20 g alcohol) per day, not on all days of the week and preferably with meals.

Now Shield et al. (2023) propose to use a ‘whole health’ effect derived from a weighed composite risk function. The question arises what ‘whole health’ really means. Possibly in this context ‘whole health’ means all known causes of death that are or may be related to alcohol consumption without the benefits derived from pleasure and enjoyment. ‘Whole health’ as such is different from World Health Organization (WHO) definition of ‘health’, which is ‘health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity’.

Furthermore, Shield et al. (2023) wants to help people reduce their alcohol use by presenting risk zones. There are four risk zones defined: no risk at no alcohol consumption, low risk at 1-2 glasses per week, moderate risk at 3-6 glasses per week and increasingly high risk at 7+ glasses per week. Figures 1 and 2 in this paper, however, show that for men and women the lifetime alcohol-attributable deaths per 1000 remains under 0 or is equal to 0 over a range up to 5-7 gm alcohol per day, similarly for ‘years of life lost’ and ‘disability adjusted years of life lost’. The authors did not opt for this threshold since they considered that not all age effects and other adverse effects may have been underestimated. The latter underestimate what was taken into account whereas other underestimates such as alcohol consumption underreporting, which would increase a ‘no risk’ level, were not taken into account.

Also, people are prepared to take a small risk for something that will benefit them, maybe not so much in health (where a benefit does exist) as in enjoyment and pleasure. Guidelines based on optimal mainly physical health, disregard these aspects and will therefore not be taken seriously by consumers.

Furthermore, an interesting commentary was published following on from the guidelines by Greenfield (2023). He remarked that guidelines based on weekly consumption would disregard one’s drinking pattern, which is clearly related to health outcomes (his refs 4 and 5) and even with alcohol use disorder (his ref 8). Also, the call to broader implement the risk zone approach for all national drinking guidelines was considered to be premature since no evidence exists that this approach will work in stimulating people to drink less.

Specific Comments from Forum Members

Forum member Harding “disagrees with the assumption of the effectiveness this paper, as stated in the first sentence of the abstract, ‘Low-Risk Alcohol Drinking Guidelines (LRDGs) aim to reduce the harms caused by alcohol.’  The purpose of public health messages on alcohol should be to give advice to the generality of the population who choose to consume alcohol about what are sensible levels and patterns of drinking within their lifestyles, and to explain the rationale for that advice.  Experience has shown that those who indulge in harmful levels and patterns of alcohol consumption are simply not receptive to such public health advice and do not respond to it.  If they were receptive and they did respond, after decades of public health advice, then these problems would not exist.  Obviously, other approaches than public health advice are needed to address them.

Second, it is unclear why a mathematical modelling approach is an appropriate way to formulate a public health message on anything.  The usual way to formulate public health messages is to compare those with a particular lifestyle characteristic with those without that characteristic, compare the health outcomes, and then examine whether the particular characteristic causes the outcome.  Consequently, we are advised, for example, not to smoke, to take regular exercise, to consume lots of fruits and vegetables, and we required to wear seat belts in our cars, on the basis that these behaviours will make us all healthier or at least, less likely to die, based on an examination of the all the available evidence.  When the same exercise is carried with respect to alcohol consumption, a very clear picture emerges.  Those adults in western society who drink alcohol moderately and regularly tend to have longer and healthier lives than those who choose to abstain or drink more heavily.  Such an observation does not sit happily with guidelines that are based on the assumption that any alcohol consumption by its very nature carries a health risk (see Figure 3). According to the model, the highest risk category of consumption (7 or more drinks/week) equates to 13.5g/day, which is close to the bottom of the all-cause mortality curves for both men and women, i.e. the lowest overall risk from the epidemiology, for example, Di Castelnuovo et al., (2006).  Both can’t be correct.

All mathematical models need to have a central criterion to which the evidence is applied within the model.  In this case, it is that no more than one death in 1000 is due to an alcohol-related cause.  This criterion is does not emerge from the examination of the body of evidence of the effect of alcohol consumption on public health.  Rather, it is completely arbitrary, totally made up.  If a different criterion were chosen, then different public health advice would emerge from the model, and would be completely unrelated to the actual evidence.  That cannot be right.

Further, the level of alcohol consumption that emerges from the mathematical model is then, by some intellectual sleight of hand, passed off as being appropriate consumption for all individuals in the population, when it is clearly nothing of the sort.  It is the level of consumption that everyone must adhere to in order to meet the made-up criterion inherent in the model, namely, in this case, no more than one death in 1000 due to an alcohol-related cause.  This is not dissimilar to communism, in the sense that everyone is required to behave in a way to achieve some arbitrary national target, whether such behaviour is appropriate for individuals or not.

On the evidence itself, there is the general assumption throughout the paper that risks associated with alcohol consumption at various levels (see Table 2 and Figure 3) are real risks, when in fact they are based largely upon studies which can only establish association, not causation.  It has been recognised that, for example, the reduction in risk mortality from ischaemic heart disease associated with regular, moderate alcohol consumption is causal, based on the application of the Bradford-Hill criteria of causation (Roerecke and Rehm, 2014).  Where is the application of the Bradford Hill criteria for causality to the causes of death listed in Table 2?  The public expect – and have the right to expect – that all public health advice is based on causation, i.e., if they take the advice and make the effort to change their behaviour, then their health status will improve in some way, and not based on assumptions of risk that may well not be risks at all.

The paper is written, quite understandably, in the context of the harm observed from alcohol consumption within the Canadian drinking population.  One reason for this harm is mentioned in the discussion, ‘Canada is a country with a high prevalence of……heavy episodic drinking (21.2% of all adults)’.  That is one behaviour in the population, for example, that needs to be addressed, rather than trying to reduce consumption across the whole population with no clear benefit, and indeed, likely harm.”

Forum member Skovenborg considers each section of Shield et al. (2023) separately as follows.

Page 2: references

Expert opinion has a place at the bottom of the evidence hierarchy pyramid; accordingly, the level of trust in the conclusions and recommendations of the Low-Risk Alcohol Drinking Guidelines Scientific Expert Panel (SEP) rests on the quality level of the evidence base presented in the list of references. And the only way to establish an unbiased approach to the evidence is to look at the references chosen by the SEP.  Please note that Addiction uses Vancouver reference style: “Check the reference details against the actual source – you are indicating that you have read a source when you cite it.”

Let us look at the references used in this statement: “Despite hypothesized benefits of red wine on cardiovascular disease [14–18] or the disbenefits of spirits on alcohol intoxication and injury due to rapid ethanol ingestion [19] …” Four of the five references [14-17] related to “the hypothesized benefits of red wine on cardiovascular diseases” focus on resveratrol. The fifth reference [18] is a 21 years old “Fact Sheet” review of red wine and cancer prevention that firstly is not related to cardiovascular disease and secondly not any longer accessible by means of PubMed or Google Scholar. The reference regarding “the disbenefits of spirits” [19] is the 2020 American Cancer Society Guideline for diet and physical activity for cancer prevention. The Guideline section on alcohol and cancer risk have no information or references with specific relation to spirits.  

Thus, two of the six references [18, 19] are flawed and irrelevant, and the resveratrol studies in relation to “hypothesized benefits of red wine on cardiovascular disease” [14-18] are outdated as well as misleading:

•     Polyphenol composition in one-year-old red wine includes around 5-8% of catechins, 5-10% dimer procyanidins, 10-15% of anthocyanidins, 3-6% phenolic acids, <1% of flavonols, 60-80% polymeric polyphenols and only <0.3% of resveratrol (Li and Sun, 2017).

•     Resveratrol is a phytoalexin − a low molecular weight antimicrobial compound that is produced by plants as a response to biotic and abiotic stresses. In this regard, the resveratrol content of wine is usually low, highly variable and thus unpredictable. Red wine contains an average of 1.9 ± 1.7 mg trans-resveratrol, ranging from non-detectable levels to 14.3 mg/L (Stervbo et al., 2007).

•     The presence of resveratrol in the human diet is almost negligible and not enough to support healthy biological actions. Accordingly, the simple correlation of cardiovascular benefits of red wine and resveratrol is questionable, and a role for resveratrol in explaining the so-called French paradox, associated with red wine consumption, has likely been overestimated (Tomé-Carneiro et al., 2013).

•     The vast majority of studies dealing with the biological activity of resveratrol have been carried out with in vitro cell cultures and to a lesser extent in animal models with resveratrol doses from 5-100 mg/kg bodyweight. The doses used in human studies vary from 10 mg to 7,000 mg per day. With a proposed therapeutic dose of 1 g resveratrol per day, you would have to drink 505-2762 L of Pinot Noir from France or 3448 L Rosé from Serbia or 2564-17544 L Riesling from Spain every day (Weiskirchen et al., 2016).

•     Relatively few human clinical trials have been performed so far and some of the results have been contradictive. For example, in trials of obese individuals with impaired insulin sensitivity, a Dutch randomized trial (Timmers et al., 2011) demonstrated a robust decrease in both systolic and diastolic blood pressures and statistically significant improvements in HOMA-IR, suggesting a favourable effect on insulin sensitivity with a resveratrol dose of 150 mg/day, while a Danish randomized trail found no effect of 500 mg trans-resveratrol thrice daily (Poulson et al., 2012).

Why did SEP not select a recent reference such as Haseeb et al. (2017), which is a comprehensive and unbiased review published in Circulation, a medical journal with an impact factor of 37.8. In comparison, Addiction has an impact factor of 6.34. The flawed choice of references cited above might be a result of ignorance on the part of SEP, however, a more likely explanation would be “White hat Bias” − defined as bias leading to distortion of research-based information in the service of what may be perceived as “righteous ends” (Cope and Allison, 2010).

Page 3: Risk modelling

Sir David Spiegelhalter is Chair of the Winton Centre for Risk and Evidence Communication at the University of Cambridge, which aims to ensure that quantitative evidence and risk is presented to people in a fair and balanced way. In The Norm Chronicles: stories and numbers about danger, Spiegelhalter and Blastland (2013) explore the statistics for personal activities through the stories of average Norm and an ingenious measurement called the MicroLife. Imagine the duration of your adult life divided into 1 million equal parts. A MicroLife is one of these parts and lasts 30 minutes. It is based on the idea that as young adults we typically have about 1 million half-hours left to live on average. MicroLives can measure how fast you are using up your stock of life − faster or slower depending on the lifestyle habits and chronic risks to which you are exposed. One cigarette reduces life-expectancy by around 15 minutes on average, so two cigarettes cost 1 MicroLife. Twenty cigarettes mean you burn an extra 10 MicroLives a day on average, which is to say you move towards death five hours faster, every day.  Here are examples of MicroLives lost or gained with different lifestyle habits:

•     First 20 minutes of moderate exercise                      + 1

•     Subsequent 40 minutes of moderate exercise           + 1

•     2 hours watching television                                     – 1

•     Per 5 kilos above optimum weight                           – 1

•     One burger (85 g red meat)                                     – 1

•     Five servings of vegetable and fruit                          + 4

•     Smoking 15-24 cigarettes/day                                  – 10

•     2-3 cups of coffee                                                    + 1

•     First drink (10 g alcohol)                                          + 1

•     each subsequent drink (up to 6)                               – ½

Page 3: mathematical modelling and drinking pattern

“The models based on one in 1000 life-time deaths (i.e. 17.5 YLL per 1000 life-times) indicated that the thresholds for men were 6 g (g)/day when based on deaths, 5 g/day when based on premature deaths and 4 g/day when based on YLL or DALYs lost. For women, the thresholds were 7 g/day when based on deaths, 5 g/day when based on premature deaths and 4 g/day when based on YLL or DALYs lost.”

In real-life, no-one drinks 4 g of alcohol per day. The use of mathematical modelling disregards the importance of drinking pattern and the problem of missing information on drinking pattern in many studies. “One of the most serious problems in this literature, and often encountered, is the use of average amounts per day. This does not provide a realistic description of a drinking pattern (Knupfer, 1987).” Most of the results from alcohol and cancer research are expressed in terms of average daily intake, often calculated by dividing the alcohol consumption per week with seven, which discloses nothing about the pattern of consumption and is a questionable method of classifying drinkers: an average can mean occasional excess (binge drinking) or regular moderate drinking. So, what kind of drinking is actually being described when averages per day are the basis of classification: someone drinking a little every day or those who drink a lot once a week?

Page 4: age-specific advice

Age-specific guidelines are in use for the young but not recommended for the elderly by SEP. According to the recent evidence from the Global Burden of Disease study, alcohol consumption carries significant health risks and no benefits for young people, however, people aged 40+ years can safely drink small amounts of alcohol some older adults may benefit from drinking a small amount of alcohol (Bryazaka et al., 2022). Similar results were found in an analysis by White et al. (2002). A direct dose-response relation exists between alcohol consumption and risk of death in women aged 16-54 years and in men aged 16-34 years. At older ages the relationship is U-shaped. The level at which the risk is lowest increases with age, reaching 3 units a week in women aged over 65 and 8 units a week in men aged over 65. The level at which the risk is increased by 5% above this minimum is 8 units a week in women aged 16-24 and 5 units a week in men aged 16-24, increasing to 20 and 34 units a week in women and men aged over 65, respectively.

Page 5: counterproductive information

According to Shield and Rehm (2019) [SEP’s ref. 39] lifetime abstainers have a slightly elevated risk of conditions for which alcohol is protective at low-consumption levels (that is, ischaemic heart disease, ischaemic stroke and diabetes). “Additionally, the GBD study does not model the risks of ischaemic heart disease, ischaemic stroke, and injury by irregular heavy drinking status. Non-binge drinkers have a lower risk of ischaemic stroke, ischaemic heart disease, and injury. As a result of these biases, the TMREL (theoretical minimum risk exposure level) overweighs alcohol’s detrimental effects compared with its protective effects.” Shield and Rehm (2019) conclude: “Although it is an unfounded conclusion that there is no beneficial individual level of alcohol consumption, the detrimental effects at the population level far outweigh the beneficial effects.” SEP agreed that cause- or disease-specific risks should be discussed between a patient and a health professional with access to the patient’s medical history, but “shared the view that from a public health perspective aiming to educate people of the overall health risk of alcohol, it would be counterproductive to present such risks. A disease-specific approach may lead people to pick and choose a disease upon which to focus.” In other words: because the harm caused by heavy drinking, binge drinking and abuse of alcohol is detrimental, it would be counterproductive to inform the public about the beneficial effects associated with a regular, moderate intake of wine or beer with a meal.

Page 5: No safe level of alcohol consumption

“In order to identify a “safe” level of alcohol intake, valid scientific evidence would need to demonstrate that at and below a certain level, there is no risk of illness or injury associated with alcohol consumption (WHO 2023). The condition to demonstrate a certain “safe level” without risk does not conform with the current scientific method, the principle of falsifiability, a deductive standard of evaluation of scientific theories and hypotheses introduced by the philosopher of science Karl Popper. The condition of “no safe level” is pseudoscience. In a true science, the following statement can be easily made: “If x happens, it would show demonstrably that theory y is not true.” We can then design an experiment, a physical one or sometimes a simple thought experiment, to figure out if x actually does happen.  Falsification is the opposite of looking for verification; you must try to show the theory is incorrect, and if you fail to do so, thereby strengthen it. It is not possible to design an experiment that would prove the existence of a safe level of alcohol consumption.

Page 8: confounding factors

“LRDGs based on modelled life-time deaths attributable to alcohol use rely upon data from meta-analyses which aggregate results from cohort studies. These meta-analyses account for confounding, but do not account for interactions between alcohol use and other behavioural, environmental and genetic risk factors, such as body mass index, smoking, socio-economic status…”

With no randomized, controlled trials in view for the foreseeable future we are left with evidence from observational studies with: 1) information about moderate, low-risk lifestyle factors; and 2) the socio-economic status of the participants should preferably be homogeneous. Using data from the Nurses’ Health Study (1980-2014; n=78 865) and the Health Professionals Follow-up Study (1986-2014, n=44 354), Li et al. (2018) defined five low-risk lifestyle factors as never smoking, body mass index of 18.5 to 24.9 kg/m2, ≥30 min/day of moderate to vigorous physical activity, moderate alcohol intake (5-15 g/day for women, 5-30 g/day for men), and a high diet quality score (upper 40%), and estimated hazard ratios for the association of total lifestyle score (0–5 scale) with mortality (Li et al., 2018). The authors estimated that the life expectancy at age 50 years was 29.0 years (95% CI, 28.3-29.8) for women and 25.5 years (95% CI, 24.7-26.2) for men who adopted zero low-risk lifestyle factors. In contrast, for those who adopted all five low-risk factors, they projected a life expectancy at age 50 years of 43.1 years (95% CI, 41.3-44.9) for women and 37.6 years (95% CI, 35.8-39.4) for men. The projected life expectancy at age 50 years was on average 14.0 years (95% CI, 11.8-16.2) longer among female Americans with five low-risk factors compared with those with zero low-risk factors; for men, the difference was 12.2 years (95% CI, 10.1-14.2).

In a sensitivity analysis using a low-risk score without moderate alcohol consumption, the projected life expectancy at age 50 years was on average only 11.4 years (95% CI, 9.5-13.3) longer among female Americans with four low-risk factors compared with those with zero low-risk factors; for men, the difference was 10.0 years (95% CI, 9.2-10.9). In conclusion, the results of this large, state-of-the-art cohort study suggest that a moderate alcohol consumption in addition to four low-risk lifestyle factors is associated with a life expectancy at age 50 years, on average, 2.6 years longer for female nurses and 2.2 years longer for male health professionals in comparison with nurses and health professionals with four low-risk lifestyle factors and no habit of moderate alcohol intake. Supposedly it would be counterproductive to inform the general public about the results of this study and similar results from 19 other cohort studies of the same type including the latest of such studies: the Australian study of 11,340 men and women (median age 73.9 years) from the ASPirin in Reducing Events in the Elderly study, followed for a median of 6.8 years (Robb et al., 2023).

A lifestyle score was constructed based on four lifestyle factors (alcohol consumption, smoking status, physical activity and diet) ranged from 0 to 4, with higher scores indicating higher adherence to healthy lifestyle recommendations. A previous study using ASPREE data found a ‘U’ shaped association between alcohol consumption and mortality in older adults since a moderate alcohol consumption was associated with a reduced risk of CVD and all-cause mortality when compared to those reporting low/none or high levels. The lowest risk was observed for about 55 g alcohol/week = 6 standard drinks/week (Neumann et al., 2022). In the current analyses, participants were, therefore, categorised as either having moderate alcohol consumption (51–100 g of alcohol/week) or not at baseline.”

Concluding comments

Forum member Ellison states that “Other Forum reviewers have presented a very detailed and accurate criticism of the methods used by the authors of this paper in reaching their conclusions.  While I agree with their arguments, I feel that the idea of testing for the effects of ‘alcohol’ are misguided, as essentially no one consumes alcohol itself, but a variety of beverages that may contain some alcohol.  Each of these beverages, especially wine and beer, also contain many other substances, some considered harmful and others considered beneficial.  Further, the pattern of consumption is probably the most critical factor in determining the health risks or benefits of such beverages.  For example, the effects of having two drinks of a beverage each day are quite different from having 14 drinks on one day over the weekend.  Also, if a beverage containing alcohol is consumed with food it makes a large difference in its health effects.

Finally, for developing data for setting policy, it is not recommended to combine data from mature adults with data from young people who may drink to excess, that often leads to violent death at an early age and inevitably results in large number of ‘years lost to alcohol’.  The important question is what should a physician advise for a 50-to-60-year-old patient with risk factors for cardiovascular disease? This should be based on comparing data from similarly-aged adults who do, or do not, consume alcoholic beverages in a specific way.  It has clearly been shown that 50-year-olds who do not consume any beverages containing alcohol have more cardiovascular disease and earlier death than 50-year-olds who drink moderately (Streppel et al., 2009).”

As Kerr and Stockwell (2011) stated, “for consumers to follow drinking guidelines and limit their risk of negative consequences they need to track their ethanol consumption… but drinkers have difficulty defining and pouring standard drinks with over-pouring being the norm such that intake volume is typically underestimated.” What Shield et al. (2023) also fail to take into account with their ‘formula’ is that is no robust evidence that the general population changes its level of alcohol consumption in response to governments defining standard drinks and publishing low-risk drinking guidelines. Comparing the potential health impact of different nations’ guidelines would be a worthy activity for international research teams. Furthermore, guideline developers should respond to such criticisms as those levelled at the product of the Canadian Low-Risk Alcohol Drinking Guidelines Scientific Expert Panel and focus particularly on the balance between epidemiological evidence, expert judgement and pragmatic consideration (Holmes et al., 2018).


Bryazka, D., Reitsma, M. B., Griswold, M. G., Abate, K. H., Abbafati, C., Abbasi-Kangevari, M., Abbasi-Kangevari, Z., Abdoli, A., Abdollahi, M., Abdullah, A. Y. M., Abhilash, E. S., Abu-Gharbieh, E., Acuna, J. M., Addolorato, G., Adebayo, O. M., Adekanmbi, V., Adhikari, K., Adhikari, S., Adnani, Q. E. S., … Gakidou, E. (2022). Population-level risks of alcohol consumption by amount, geography, age, sex, and year: a systematic analysis for the Global Burden of Disease Study 2020. Lancet (London, England), 400(10347), 185–235.

Cope, M.B., Allison, D.B. (2010) White hat bias: examples of its presence in obesity research and a call for renewed commitment to faithfulness in research reporting. Int J Obes (Lond), 34(1):84-8; discussion 83. Epub 2009 Dec 1. PMID: 19949416;

Di Castelnuovo, A., Costanzo, S., Bagnardi, V., Donati, M.B., Iacoviello, L., de Gaetano, G. (2006). Alcohol dosing and total mortality in men and women: an updated meta-analysis of 34 prospective studies. Arch Intern Med., 166(22), 2437-45.

Greenfield, T.K. (2023). Debating Shield et al.’s perspectives on how to formulate alcohol drinking  guidelines. Addiction,

Haseeb, S., Alexander, B., Baranchuk, A. (2017). Wine and Cardiovascular Health: A Comprehensive Review. Circulation, 136(15):1434-1448.

Holmes, J., Angus, C., Meier, P.S., Buykx, P., Brennan, A. (2018) How should we set consumption thresholds for low risk drinking guidelines? Achieving objectivity and transparency using evidence, expert judgement and pragmatism. Addiction. ISSN 0965-2140

Li L., Sun B. (2017). Grape and wine polymeric polyphenols: Their importance in oenology. Crit Rev Food Sci Nutr., 59:1–17

Li, Y., Pan, A., Wang, D.D., Liu, X., Dhana, K., Franco, O.H., Kaptoge, S., Di Angelantonio, E., Stampfer, M., Willett, W.C., Hu, F.B. (2018) Impact of Healthy Lifestyle Factors on Life Expectancies in the US Population. Circulation, 138(4), 345-355. Erratum in: Circulation. 2018 Jul 24;138(4):e75. PMID: 29712712; PMCID: PMC6207481.

Knupfer, G. (1987). Drinking for health: the daily light drinker fiction. British Journal of Addiction, 82, 547-555.

Neumann, J.T., Freak-Poli, R., Orchard, S.G., Wolfe, R., Reid, C.M., Tonkin, A.M., Beilin, L.J., McNeil, J.J., Ryan, J., Woods, R.L. (2022) Alcohol consumption and risks of cardiovascular disease and all-cause mortality in healthy older adults. Eur J Prev Cardiol., 29(6):e230-e232. PMID: 34718513

Poulsen, M.M., Vestergaard, P.F., Clasen, B.F., Radko, Y., Christensen, L.P., Stødkilde-Jørgensen, H., Møller, N., Jessen, N., Pedersen, S.B., Jørgensen, J.O. (2013) High-dose resveratrol supplementation in obese men: an investigator-initiated, randomized, placebo-controlled clinical trial of substrate metabolism, insulin sensitivity, and body composition. Diabetes, 62(4):1186-95. Epub 2012 Nov 28. PMID: 2319318

Robb, C., Carr, P.R., Ball, J., Owen, A., Beilin, L.J., Newman, A.B., Nelson, M.R., Reid, C.M., Orchard, S.G., Neumann, J.T., Tonkin, A.M., Wolfe, R., McNeil, J.J. (2023) Association of a healthy lifestyle with mortality in older people. BMC Geriatr.,23(1):646. PMID: 37821846;

Roerecke, M., Rehm, J. (2014) Alcohol consumption, drinking patterns, and ischemic heart disease: a narrative review of meta-analyses and a systematic review and meta-analysis of the impact of heavy drinking occasions on risk for moderate drinkers. BMC Med., 12:182. PMID: 25567363

Shield, K., Paradis, C., Butt, P., Naimi, T., Sherk, A., Asbridge, M., Myran, D., Stockwell, T., Wells, S., Poole, N., Heatley, J., Hobin, E., Thompson, K., Young, M. (2023). New perspectives on how to formulate alcohol drinking guidelines. Addiction,

Shield, K.D., Rehm, J. (2019) Alcohol and the global burden of disease. Lancet, 393(10189):2390. PMID: 31204673.

Stervbo, U., Vang, O., Bonnesen, C. (2007) A review of the content of the putative chemopreventative phytoalexin resveratrol in red wine. Food Chemistry, 101:449–457.

Streppel, M.T., Ocké, M.C., Boshuizen, H.C., Kok, F.J., Kromhout, D. (2009) Long-term wine consumption is related to cardiovascular mortality and life expectancy independently of moderate alcohol intake: the Zutphen Study. J Epidemiol Community Health, 63(7):534-40.

Timmers, S., Konings, E., Bilet, L., Houtkooper, R.H., van de Weijer, T., Goossens, G.H., Hoeks, J., van der Krieken, S., Ryu, D., Kersten, S., Moonen-Kornips, E., Hesselink, M.K.C., Kunz, I., Schrauwen-Hinderling, V.B., Blaak, E., Auwerx, J., Schrauwen, P. (2011) Calorie restriction-like effects of 30 days of resveratrol supplementation on energy metabolism and metabolic profile in obese humans. Cell Metab., 14(5):612-22. PMID: 22055504; PMCID: PMC3880862.

Tomé-Carneiro, J., Gonzálvez, M., Larrosa, M., Yáñez-Gascón, M.J., García-Almagro, F.J., Ruiz-Ros, J.A., Tomás-Barberán, F.A., García-Conesa, M.T., Espín, J.C. (2013) Resveratrol in primary and secondary prevention of cardiovascular disease: a dietary and clinical perspective. Ann N Y Acad Sci., 1290:37-51. PMID: 23855464.

Weiskirchen, S., Weiskirchen, R. (2016) Resveratrol: How Much Wine Do You Have to Drink to Stay Healthy? Adv Nutr., 7(4):706-18. PMID: 27422505; PMCID: PMC4942868.

White, I.R., Altmann, D.R., Nanchahal, K. (2002) Alcohol consumption and mortality: modelling risks for men and women at different ages. BMJ, 325(7357):191. PMID: 12142306; PMCID: PMC117446.

The Norm Chronicles: stories and numbers about danger by Michael Blastland and David Spiegelhalter, Profile Books Ltd, London, 2013. No. of pages: 288. ISBN: 9781846686207

Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:

Henk Hendriks, PhD, Netherlands

Creina Stockley, PhD, MBA, Independent consultant and Adjunct Senior Lecturer in the School of Agriculture, Food and Wine at the University of Adelaide, Australia

Richard Harding, PhD, UK

Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark

Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany

R. Curtis Ellison, MD, Section of Preventive Medicine/Epidemiology, Boston University School of Medicine, Boston, MA, USA