Food and diet

Alcohol limit 'should be cut for public health'

If people drank just half a unit of alcohol a day it could cut deaths from chronic conditions such as cancer, The Guardian has reported today.

The claim is based on new research investigating both the harmful and the protective effects of alcohol, and how changing average drinking habits could cut fatal chronic disease. Researchers built a mathematical model to estimate the effect of changing habits, and found that if drinkers in England cut down to an average daily consumption of 5g of alcohol (about half a unit), it could prevent or delay nearly 4,600 deaths each year. Current advice is that men should consume no more than three to four units of alcohol a day and women no more than two to three units.
This is a complex study that created a detailed model from existing evidence on alcohol consumption and the incidence of chronic disease. However, this theoretical model has several limitations. As the authors note, it relies on the quality and reliability of the existing studies used to devise the model. Also, the study is based on average levels of alcohol consumption so is unable to take into account different patterns of drinking (for example binge drinking), which are thought to play an important part in risk of disease.

Where did the story come from?

The study was carried out by researchers from Oxford University and Deakin University Australia. They were supported by grants from the British Heart Foundation and the Australian Government’s Department of Health and Ageing.

The study was published in the peer-reviewed medical journal BMJ Open.

The research was covered fairly, if uncritically, in the press. The Guardian included comments from independent experts and from industry funded sources.

What kind of research was this?

A host of previous research has linked alcohol consumption to a range of chronic diseases, including cancer, obesity, high blood pressure and epilepsy. There is also some research that suggests that moderate drinking can have a protective effect against conditions such as heart disease, although the matter is open to some debate and critics point out that the potential benefits may not outweigh the potential harms.

The authors say that, as previous research has suggested that alcohol consumption is a risk factor for many chronic diseases while providing ‘modest protection’ from others, this gives rise to contradictory advice about the level of alcohol consumption that is optimum for health. They also say that the impact of current guidelines on incidence of chronic disease is unclear.

In this study, researchers built a detailed model called a ‘macro-simulation model’ to estimate the average level of alcohol consumption that would theoretically be needed to minimise deaths from a range of chronic diseases. They also calculated whether increasing the number of non-drinkers (those who drink either no alcohol or very low amounts) could have a similar effect.

What did the research involve?

The researchers built a macro-simulation model that assessed the impact that levels of alcohol consumption had on death rates from various chronic diseases. The model estimated this impact in the English population.

The researchers identified an initial list of 11 chronic diseases, including five cancers, linked to alcohol consumption using data from the World Health Organization and the World Cancer Research Fund Report. The non-cancerous diseases were coronary heart disease, stroke, hypertension (high blood pressure), diabetes, liver cirrhosis and epilepsy. The five cancers were of the liver, mouth and throat, oesophagus, breast and bowel.

The researchers searched two large databases for meta-analyses of prospective cohort or case-control studies that quantified the risk of chronic disease associated with different levels of alcohol consumption. A meta-analysis is a type of study that combines the statistical results of several studies into a single set of results. The associations they found in these meta-analyses included protective effects (for coronary heart disease), linear increases in risk and ‘U’ or ‘J’ shaped relationships indicating protection only at low or moderate consumption (for example, for stroke). These types of relationships are named in this way because their results broadly resemble the shape of a ‘U’ or ‘J’ when plotted on a graph. 

The researchers identified the average weekly alcohol consumption among people aged 16 or over in England, using a General Household Survey from 2006. Non-drinkers and very-low alcohol consumers were analysed as a separate category (referred to as non-drinkers).

The official statistics on deaths from the list of 11 chronic diseases according to age and sex were used as the basis for estimating the number of deaths due to alcohol. The degree to which reducing alcohol reduced risk of each disease was quantified in a variety of meta-analyses. Together these were used to estimate the number of chronic disease deaths prevented and delayed at current levels of consumption.

They then modelled the number of deaths that would result from these chronic diseases using two theoretical scenarios.

  • In the first scenario they changed the average level of alcohol consumption among drinkers, (keeping the proportion of non-drinkers the same). They varied the amount of alcohol consumed by drinkers between 1 and 48g (or six units) daily, while keeping the existing age and sex distribution for alcohol consumption.
  • In the second scenario they varied the proportion of non-drinkers in the population while keeping the average consumption among drinkers the same. They varied the percentage of non-drinkers in the population between 0% and 100% (again maintaining existing age and sex distribution).

They then analysed the data in order to find the level of average alcohol consumption that would be likely to result in the lowest number of deaths from overall chronic disease.

What were the basic results?

In the first scenario they found that about 5g of alcohol a day (just over half a unit) was the optimum level of alcohol consumption, resulting in the prevention or delay of 4,579 deaths (95% credibility interval 2,544 to 6,590). This represents a 3% reduction in all deaths from alcohol-related chronic disease on the 2006 figures.

  • They predict that this level of consumption would result in 2,668 fewer deaths from cancer (a reduction of 8%), 2,828 fewer deaths from liver disease (a reduction of 49%) but an additional 843 deaths a year from cardiovascular disease (a rise of 0.7%).
  • Their model of the second scenario, in which the proportion of non-drinkers was increased, showed no benefit in terms of reducing deaths from chronic disease.

How did the researchers interpret the results?

They say their model shows that the optimum average daily alcohol consumption appears to be substantially lower than the currently recommended levels for safe drinking in the UK. The level of consumption they recommend would equate to about one-quarter of a glass of wine or one-fifth of a pint of beer a day on average, much lower than the recommended daily maximum of two to three units for women and three to four units for men. Public health targets, they argue, should be to reduce average alcohol consumption in England to half a unit a day for both men and women.


There has long been debate over the level of both harm and protection offered by alcohol, particularly over whether any protective effect alcohol may have on the heart actually outweighs its role in various chronic diseases. This complex study created a detailed model to simulate, using the best available evidence, the relationship between alcohol levels and the incidence of chronic disease. The researchers’ aim was to calculate an optimum level of alcohol consumption for a minimum risk of a range of chronic diseases.

However, it has a number of limitations, as the authors themselves have noted. Its calculations rely on the accuracy of meta-analyses of previous cohort and case-control studies on the risk to health of regular alcohol consumption. It is unclear how reliable the original studies were in terms of their design or methods, but these types of studies are often unable to take into account ‘confounders’, which are factors that may also affect disease risk. They also often rely on participants estimating their own levels of alcohol consumption. Given the difficulty of accurately estimating or recalling alcohol consumption, it may have led to an underestimation or overestimation by the researchers of the benefits of reducing alcohol.

One key limitation is that the study is based on average levels of alcohol consumption and was unable to take into account different patterns of drinking (such as binge drinking or consumption of particular dinks, such as red wine), which are thought to play an important part in differing risks of disease.

As the authors concede, the results are based on data about current alcohol consumption and disease levels in England. The level of alcohol consumption associated with least risk of disease may vary substantially between different populations and social groups.

As well as the limitations of this study, it is also open to debate whether the public would find the estimated ideal levels of consumption realistic or acceptable.

NHS Attribution