Science and Technology Committee HC 1536 Alcohol GuidelinesWritten evidence submitted by Alcohol Research UK (AG 06)

1. What evidence are Government’s guidelines on alcohol intake based on and how regularly is the evidence base reviewed?

We have decided to focus our document on the current evidence base.

2. Could the evidence base and sources of scientific advice to Government on alcohol be improved?

The first point to make is that alcohol causes nearly 10% of all ill-health and premature deaths in Europe. The World Health Organization’s Global Burden of Disease Study finds that alcohol is the third most important risk factor, after smoking and raised blood pressure, for European ill-health and premature death.

The number of alcohol-related deaths in the United Kingdom has consistently increased since the early 1990s, rising from the lowest figure of 4,023 (6.7 per 100,000) in 1992 to the highest of 9,031 (13.6 per 100,000) in 2008. (1)

The following summary of current evidence and key concepts relating to guidelines relies heavily on recent work carried out in Australia and Canada by expert committees producing low risk guidelines: labelled low risk because even regular small amounts of alcohol have been linked to disease. A Special Issue of the Drug & Alcohol Review devoted to the topic of drinking guidelines will be published next March, with early online availability before that. The Special Issue will include papers by world experts on a range of issues to do with drinking guidelines and will be directly relevant to the Committee’s deliberations.

2.1 A Relative or Absolute Risk Approach

The Canadian approach was to base guidelines on comprehensive reviews of the Relative Risk of health or social outcomes for different levels of consumption compared with the risk experienced by abstainers. Sixteen systematic reviews and meta-analyses conducted by Rehm, Taylor and colleagues at the Centre for Addiction and Mental Health, Ontario were examined (2). The aim was to identify a level of average daily consumption where overall net risk of premature death is the same as that of a lifetime abstainer (zero net risk). At this level of consumption potential health risks and benefits from drinking exactly cancel each other out.

The Australian approach [3] focused upon the absolute risk of harmful outcomes. Daily drinking levels were estimated which would increase the lifetime risk of premature death, injury or illness to more than 1%: this risk of one in 100 is often adopted when assessing the harmful effects of exposure to other hazardous events (eg toxins in the environment).

It should be noted that both the Australian and the Canadian methodology differ from the approach adopted in the Department of Health Sensible Drinking recommendation in 1995.Their recommendations were derived from an analysis of the point on the J-shaped curve relating consumption to the risk of all-cause mortality at which the curve shows a significant increase relative to its lowest point (see Department of Health, 1995, p.21, para 7.8), rather than relative to abstainers as in the Canadian analysis.

2.2 Summary of Canadian findings

2.2.1 Lifetime risk of alcohol- related disease

The Canadian review considered findings separately for males and females from four mostly well-designed meta-analyses linking level of alcohol consumption and risk of death from all causes. The point at which there was a zero net risk in these studies compared with the abstainer reference group was between 1.5 and 2.5 Canadian standard drinks for women and between two and three standard drinks for men. One Canadian standard drink is 13.45g of absolute alcohol—the UK standard drink (known as a “unit”) is 8g. However, there is an important potential confounding factor when using abstainers as a reference group. As well as lifetime abstainers these groups may include both ex-drinkers and occasional drinkers who are likely to comprise many people who have cut down their drinking for health reasons. These people have a higher risk of premature death than lifetime abstainers. Di Castelnuovo et al [4] claimed to have taken account of this issue. They provide the best available estimate of a net zero risk point. This was found to be at an average of two Canadian standard drinks per day for women and three Canadian standard drinks per day for a man: the equivalent of 3.4 UK units for women and five UK units for men.

2.2.2 Lifetime risk of death from alcohol-related injury

A systematic search identified 17 published studies from 1995 onwards which had quantified the relative risk of injury following drinking. Six of the 12 case-control studies reported stronger effects for women while none of those with case-crossover designs found significant gender effects.

The Canadian expert group concluded that recommended upper levels for occasional consumption on a single day should be three Canadian standard drinks for a woman and four Canadian standard drinks for a man: equivalent to five UK units for women and seven for men. These guidelines should always be accompanied by “the following caveats and stipulations for minimising risk of acute problems: (i) avoiding high-risk situations and activities; (ii) minimising intoxication by drinking slowly, selecting lower alcohol content beverages, drinking in association with food, alternating alcoholic drinks with caffeine-free non-alcoholic drinks and not combining use with other psychoactive substances; (iii) minimising frequency of consuming at upper levels; and (iv) advising persons with low tolerance due to young or old age and/or low bodyweight to never exceed two Canadian standard drinks for women, three for men (3.5 and 5 UK units).

It was clearly acknowledged that each drink from one onwards increases the risk of acute problems in many situations. It was reasoned though that if certain high-risk situations are avoided and if other precautions are taken these levels can still be considered low risk.

2.3 Summary of the Australian evidence

2.3.1 Lifetime risk of alcohol-related disease

When considering the absolute risk of dying from alcohol-attributable disease it is clear that, as the average volume of alcohol consumption increases, the lifetime risk of death from alcohol-related disease increases. For both sexes, the lifetime risk of death from alcohol-related disease more than triples when consumption increases from two to three Australian standard drinks a day (2.5 to 3.75 UK units). At higher levels of drinking, large differences by gender are seen, with the risk for women being significantly higher than that for men. At one standard drink, the lifetime risk for women is lower than for men. At 10 standard drinks a day, the lifetime risk for women is almost one in 10, 25 times the risk at two standard drinks. For men, there is an almost 12-fold increase and the risk is six in 100.

At levels of alcohol consumption recommended in the Australian report (two Australian standard drinks or less on any day), there is little difference in the risk of alcohol-related harm for men and women over a lifetime. This level is equivalent to 2.5 UK units.

2.3.2 Lifetime risk of death from alcohol-related injury

A modelled analysis was used to determine the accumulated lifetime risk of death from the type of injuries for which alcohol has an accepted causal effect, based on established epidemiological data.

Lifetime risk was calculated for an increasing number of drinks per occasion and for various numbers of drinking occasions over a lifetime.

The risk of death from injury remains below one in 100 for both men and women if they always drink two Australian standard drinks or less on an occasion, even if the occasions are every day. (Equivalent to 2.5 UK units: about one pint of 4.5% ABV beer).

2.4 Comparison of Canadian and Australian reports

A comparison of these two reports indicates that different methods of establishing low risk guidelines will lead to differing low risk limits. These upper limits for daily low risk consumption, in UK units, are summarised in the following table:

Table 1


Low risk limits for disease

Low risk limits for injury















Both the Australian and the Canadian reports consider a similar evidence base relating to young people and to pregnancy.

2.5 Children and young people under 18 years of age

The results of a comprehensive review of the available evidence completed by the Australian team (2) indicate that:

Young people under the age of 15 years are much more likely than older drinkers to experience risky or antisocial behaviour connected with their drinking.

The rates of risky behaviour are also elevated among drinkers aged 15−17 years.

These conclusions are based upon an assessment of the potential harms of alcohol for these age groups, as well as a range of epidemiological research that indicates that alcohol may adversely affect brain development children and be linked to alcohol-related problems later in life. However, this evidence is not conclusive enough to allow definitive statements to be made about the risks of drinking for young people. As a result it has not been possible to set a “safe” or “no-risk” drinking level for children and young people. The safest option for those under fifteen is not to drink at all and the safest option for 15–17-year-olds is to delay the initiation of drinking for as long as possible.

2.6 Pregnancy and breastfeeding

2.6.1 Whilst here has been a great deal of research on the effects of alcohol on the foetus, the complexity of the issue makes development of policy and provision of definitive advice difficult. Although the risk of birth defects is greatest with high, frequent maternal alcohol intake during the first trimester, alcohol exposure throughout pregnancy (including before pregnancy is confirmed) can have consequences for development of the foetal brain.

However, it is not clear at what dose harmful effects become detectable. This uncertainty is reflected in policy and guidance regarding alcohol use in pregnancy (5). Several authoritative guidelines emphasise that a safe level has not been established and conclude that not drinking at any stage in pregnancy is the safest option.

2.6.2 The Australian guideline document provides a comprehensive review of the evidence (p71). For example, the National Perinatal Epidemiology Unit in the UK has published two high quality systematic reviews (6, 7). The reviews were based on a thorough search of the literature and were critically assessed using established criteria.

The first review, which included 46 studies, addressed the effects of low to moderate prenatal alcohol exposure (less than 12g alcohol) on pregnancy outcomes. In five of the eight studies reporting on miscarriage, an increased rate was observed in the exposed group; however two of these studies had methodological limitations, including failure to adjust for confounding factors. Overall, there was no convincing evidence that low-moderate maternal alcohol intake conferred an increased risk of miscarriage, still-birth, prematurity, intrauterine growth restriction (or small for gestational age at birth) and birth defects, including Foetal Alcohol Syndrome.

There were several limitations in the studies included in the review and the authors concluded that it “is difficult to determine whether there was any adverse effect on pregnancy outcome associated with low-moderate levels of prenatal alcohol consumption” but also that the paucity and inconsistency of available evidence “preclude the conclusion that drinking at these (low-moderate) levels during pregnancy is safe”.

2.6.3 In the second systematic review the foetal effects of prenatal “binge-drinking” (defined in most studies as six standard drinks on a single occasion) in women who were pregnant or trying to become pregnant were examined.

In the 14 included papers, there was no consistent evidence that “binge-drinking” influenced rates of these outcomes, with the exception of neurodevelopmental outcomes. Individual studies suggest that disinhibited and delinquent behaviour, reduction in verbal IQ, and learning difficulties and poor educational performance are more common in children whose mothers “binge”. Furthermore, the risk increased with higher alcohol intake and greater frequency of “binges”. Many of the included studies had methodological weaknesses and different definitions for “binge-drinking” were used (eg some studies included women who had a “binge” on a single occasion and others included only women who “binged” throughout pregnancy).

A further limitation of both this and the earlier review is the short duration of follow-up of exposed children in the included papers. Alcohol-related neurodevelopmental disorders, for example, may not be diagnosed until school age. The study authors conclude that, despite the lack of good evidence of harm from heavy drinking in human studies, the animal literature suggests that the appropriate public health message may be “recommending pregnant women to avoid binge drinking”.

The reviewers note that, when pregnant women who do not drink regularly report having had isolated episodes of heavy drinking, they should be reassured that the evidence for risk of harm is minimal.

2.6.4 Breastfeeding

There is a lack of good quality evidence from human studies regarding the effects of maternal alcohol consumption on lactation, infant behaviour and development. As a result, as for pregnancy, it was not possible to set a “safe” or “no-risk” drinking level for breastfeeding women.

2.7 Alcohol consumption and risk of social harm

2.7.1 Alcohol consumption is not only linked to acute and chronic diseases, it is also connected with social harms such as physical and sexual violence, vandalism, public disorder, family and interpersonal problems, financial problems, unwanted sex, work and school-related problems, with levels of risk rising with increased consumption. When socio-demographic variables and volume of overall alcohol use are controlled, the likelihood of social harm increases with the frequency of heavy drinking. A large US national alcohol survey (8) and found that frequency of drinking five or more US standard drinks per day was a strong predictor of violent behaviour, driving license revocation, spousal abuse, divorce/separation as well as work and school problems. (Five US standard drinks is equivalent to about 8.5 UK units.)

2.7.2 A study of patterns and levels of drinking by US college students (9) used a mathematical modelling approach to identify changing risk of harmful outcomes as a function of quantity of alcohol consumed on one occasion. Clear threshold effects were identified for this relatively young group at two drinks for females and three drinks for males. (2.5 and four UK units). These findings were considered to support the need for lower recommended levels for younger drinkers and for mitigating risk by limiting frequency of consuming at upper levels of consumption. As a consequence the Australian report advised substantially lower levels of consumption (one or two Australian standard drinks no more than one or two times per week) were recommended for older teenagers and for young people under 25 years.

2.8 Limitations of the research evidence

The following are the main limitations of the evidence base used to set low-risk limits.

2.8.1 Under-reporting of personal alcohol consumption

It is well established that self-reported consumption of alcohol can often underestimate actual consumption. Some of the larger cohort studies from which risks of specific or all-cause mortality have been estimated use the most reliable recent recall methods or validate self-report using diary methods: but many do not. Under-reporting of consumption in studies results in overestimation of the risks of alcohol consumption.

2.8.2 Failure to take account of heavy drinking episodes

The frequency of heavy consumption occasions involving five or more drinks in one day is associated with increased risk of alcohol-related morbidity and mortality. Recent studies relating level of alcohol use to risk of adverse health outcomes are more likely to control for this variable, but the meta-analyses presented in the Australian and Canadian reports include many studies that have not. This will also result in overestimation of risks from a given level of alcohol use.

2.8.3 Misclassification of former and occasional drinkers as lifelong-abstainers

Abstainers who have been drinking in the past are more susceptible than lifelong abstainers to alcohol-related harms. Recent studies and meta-analyses have been more careful to compare drinking risk against that for a strictly defined group of lifelong abstainers, making separate estimates for former drinkers.

2.8.4 Failure to control for confounding effects of personality and lifestyle

There are also a number of reasons why the beneficial effects of moderate alcohol use may be overestimated. These include evidence of publication bias (whereby researchers are more likely to publish papers finding evidence of cardiac protection than not) and a failure to control for lifestyle factors. Light and moderate alcohol use is significantly associated with a multitude of positive health behaviours which are especially likely to reduce risk of cardiac and vascular illness, eg healthy diet, regular exercise, lower bodyweight and high socio-economic status. In other words, moderate alcohol use can be a sign of healthy living rather than a cause of extra longevity.

3. How well does the Government communicate its guidelines and the risks of alcohol intake to the public?

We have no robust evidence that enables us to answer this question but we would like to follow suggestions made in the Australian guidelines and propose the use of the following methods in the future:

3.1 Materials to support health professionals providing assessment and brief advice to early-stage problem drinkers or persons whose health is compromised in some way by their drinking.

3.2 Web-based interactive materials that enable individuals to assess their own level of alcohol consumption and risk profile. Examples of such websites are and and NHS Choices

3.3 Brief and informative leaflets to be available in various healthcare settings and for the general public summarising the low-risk guidelines.

3.4 Materials that enable drinkers to better understand the number of standard drinks they consume (eg, using illustrations of popular drinks showing numbers of units they contain or the labeling of alcohol containers with the number of units they contain).

3.5 Widespread social marketing to both youth and adults to increase their knowledge of guidelines, and inform their individual choices. This includes the use of Smart phone Apps.

3.6 The Canadian expert group suggest that a “useful and engaging dissemination strategy would be to develop an Internet-based resource which would allow individuals to estimate their absolute risk of an alcohol caused death at different levels of consumption based on information about gender, age, personal health and perhaps family histories of illness. This would complement the more simplistic population-wide Guideline advice based on relative risk”.

3.7 The message to be disseminated must be simple but it also must be accurate. Heather (10) has pointed out that it must give a guideline on a limit for regular consumption as well as a different one for one-off consumption, although these two different kinds of limit are often confused in communications to the general public. He argues that: “Conventional wisdom has it that, to be effective, communications to the general public should consist of no more than one piece of information or two at the most. However, if the accurate information it is desired to communicate is unavoidably complicated, what alternative is there? At the very least, there should be a discussion in public health circles about how the conflicting imperatives of avoiding overly complex messages and telling the public what the available evidence indicates can be reconciled.

4. How do the UK Government’s guidelines compare to those provided in other countries?

4.1 The Australian document provides a useful chart that summarises recommendations relating to low risk limits in OECD countries

From this chart it can be seen that for women 12 countries are recommending limits that are lower than the UK. They are: Australia, Austria, Czech Republic, Finland, France, Ireland, Netherlands, New Zealand, Poland, Slovenia, Sweden & United States. Only five countries recommend limits that are greater than the UK. They are: Canada, Portugal, Romania, Singapore & Spain. For men 15 countries recommend limits that are lower than the UK. They are: Australia, Austria, Canada, Czech Republic, Finland, France, Ireland, New Zealand, Poland, Singapore, Slovenia, Spain, Sweden, Switzerland & United States. Six countries recommend higher limits: Denmark, Italy, Japan, Netherlands, Portugal & South Africa.

5. References

1. Alcohol Deaths—UK rates increase in 2008; Office for National Statistics, General Register Office for Scotland, Northern Ireland Statistics and Research Agency, 28 January 2010.

2. Australian Guidelines to Reduce Health Risks from Drinking Alcohol

3. The basis for Canada’s new low risk drinking guidelines: a relative risk approach to estimating hazardous levels and patterns of alcohol use. Tim Stockwell, Peter Butt, Doug Beirness, Louis Gliksman and Catherine Paradis. Based on a paper presented at the Annual Symposium of the Kettil Bruun Society for Social and Epidemiological research on Alcohol, 30 May–4 June 2010.

4. Di Castelnuovo, A Costanzo, S, Bagnardi, V, Donati, M B, Iacoviello, L. and de Gaetano, G. (2006) Alcohol dosing and total mortality in men and women: an updated meta-analysis of 34 prospective studies. Archives of Internal Medicine,166:2437-2445.

5. O’Leary, C, Nassar, N, Zubrick, S, Kurinczuk, J, Stanley, F & Bower, C (2010). Evidence of a complex association between dose, pattern and timing of prenatal alcohol exposure and child behaviour problems. Addiction, 105, (1), 74 – 86.

6. Henderson J, Gray R, Brocklehurst P (2007a). Systematic review of effects of low-moderate prenatal alcohol exposure on pregnancy outcome. BJOG 114: 243–52.

7. Henderson J, Kesmodel U, Gray R (2007b). Systematic review of the fetal effects of prenatal binge-drinking. J Epidemiol Community Health 61: 1069–73.

8. Dawson, D A, Li, T K & Grant, B F (2008). A prospective study of risk drinking: At risk for what? Drug and Alcohol Dependence, 95, 62–72

9. Gruenewald, P, Johnson, F, Ponicki, W R, & LaScala, E A (2010). A dose-response perspective on college drinking and related problems. Addiction, 105(2), 257-269.

10. Heather N, (2009). The importance of keeping regular: accurate guidance to the public on low-risk drinking levels. Alcohol & Alcoholism Vol. 44, No. 3, pp. 226–228.

September 2011

Prepared 5th January 2012