Is one UK death in eight really caused by drinking?
Many newspapers and websites reported last week that one in eight deaths of UK adults under the age of 65 were caused by alcohol. Here, for example, is the Daily Mail’s take: similar stories appeared in The Daily Telegraph, The Independent, and elsewhere.
The claim is a misquotation from a press release which actually claimed that one in eight deaths in Europe between the age of 15 and 64 is caused by alcohol. It originated from a programme called ALICE RAP (Addictions and Lifestyles in Contemporary Europe – Reframing Addictions Project, which published a report called Alcohol – The Neglected Addiction. (see file attachments below for report and press release.)
The report is remarkably short on evidence to substantiate this claim, or the further claims it makes that alcohol costs every man, woman and child in Europe 300 Euros a year in lost productivity and costs to the health, welfare and criminal justice system, and that it is a cause of “250 or so” different diseases and conditions.
However, the source of the mortality data quoted by ALICE RAP appears to be a 146-page report published by a Canadian group headed by Jürgen Rehm of the Centre for Addiction and Mental Health in Toronto, and backed by a grant from Lundbeck AG, a pharma company which is currently conducting trials of a drug designed to treat alcoholics by reducing the pleasure they experience.
The report covers the 27 member states of the EU, plus Norway, Switzerland and Iceland, with data relating to 2004. It concludes that in the EU, 11.9 per cent of all deaths before the age of 65 are attributable to alcohol (95 per cent CI 6.5 to 16.9 per cent). For men, the figure is 13.9 per cent (CI 8.1 to 19.2 per cent) and for women 7.7 per cent (CI 3.1 to 12.1 per cent). This is presumably the basis for the “one in eight” claim – 11.9 per cent is one in 8.40, which can be rounded to one in eight.
The UK is actually the ninth lowest of the 30 countries for male deaths, well below the EU average, with around 8 per cent of deaths under 65 attributable to alcohol. For deaths among women, the UK is eleventh lowest, with 7 per cent. Adding both sexes together, around 7.5 per cent of deaths, or one in 13, are linked to alcohol.
So rather than one in eight UK deaths before the age of 64 being caused by alcohol, one in 13 may be attributable to alcohol. The confidence intervals are very wide: it could be as few as one in 34, or as many as one in eight. The wide range indicates the degree of uncertainty in the calculations.
The estimates differ from many of those made previously, such as one published in 2008 by the North West Public Health Observatory, which concluded that 3.1 per cent of all deaths in England in 2005 were attributable to alcohol (4.4 per cent of male deaths, 2.0 per cent of female deaths). But the NWPHO was looking at all deaths, while the Toronto group only counted deaths below the age of 65.
By choosing a cut-off of 65, the calculation excludes the majority of deaths, reducing the denominator. But because many deaths from alcoholic liver disease - the commonest cause, responsible for 37.2 per cent of the deaths - occur before the age of 65, the numerator is not reduced by so much. The same applies to deaths from unintentional injury, such as road traffic accidents, and intentional injury, which are proportionately much commoner among younger people. The effect is to increase the proportion of deaths attributable to alcohol.
While it is entirely legitimate to limit the age range in this way, people should understand that the results measure the chances that alcohol will contribute to a premature death, not to a death at any age – and that the majority of us will not suffer a premature death. In the older age groups, alcohol has a marked protective effect because it reduces deaths from heart disease, so excluding deaths over 65 removes most of this benefit from the calculations.
An unusual feature of the Toronto report is the 9 per cent of deaths of alcohol-attributable deaths claimed to be the consequence of mental and neurological disorders. The figures published for England by the NHS Information Centre, however, show mental and nervous system disorders contributing relatively few deaths. Death certification procedures may differ in other countries, perhaps.
Finally, it is permissible to question the levels of consumption quoted in the Toronto report. The team used data from the World Health Organisation, based on alcohol tax payments. Such data may well be a more accurate measure of how much people actually drink, rather than how much they say they drink. But the estimates of the damage caused by alcohol come from studies based on reported levels of consumption, not actual consumption. It is therefore more appropriate, as Professor Max Parkin has argued in the British Journal of Cancer, to use the self-reported data to calculate alcohol-attributable deaths. That has the effect of reducing the risks markedly.
The Toronto authors conclude by saying: “If an effective treatment for alcohol dependence could be delivered to a sizeable portion of the affected population, it would have a profound and positive effect on overall public health through reduction of mortality”.
Lundbeck's trial drug nalmefene (Selincro) reduced alcohol consumption by 66 per cent in heavy drinkers who were also given medical advice, according to Phase III trial results, but placebo did almost as well, European Biotechnology News reported. The European Medicines Agency has yet to rule on Lundbeck’s application for a licence. Follow-up data is expected next month at the Research Society on Alcoholism Scientific Meeting in San Francisco.
Professor Peter Anderson (not verified) wrote,
Wed, 23/05/2012 - 20:24
Thank you so much for your critique and comments on the brief prepared by ALICE RAP, Alcohol - the neglected addiction [www.alicerap.eu]. The data in the brief on deaths in the European Union actually come from a WHO report Alcohol in the European Union [http://www.euro.who.int/en/what-we-do/health-topics/disease-prevention/alcohol-use/publications/2012/alcohol-in-the-european-union.-consumption,-harm-and-policy-approaches], in which 11.8% of all deaths in the age range 15-64 years are estimated as due to alcohol. In the same report, the proportion of deaths in the United Kingdom is estimated to be lower than the EU average (Figure 8, page 19), and, thus, as you point out, about 1 in 13 of UK deaths in the age range 15-64 years is due to alcohol. These are preventable deaths, and the age range is important, since this captures the final years of education and the working life, and it is tragic really that, apparently, not enough is being done to prevent these deaths. Comment posted by Peter Anderson, International Coordinator, ALICE RAP project and Professor, Substance Use, Policy and Practice, Institute of Health and Society, Newcastle University.
Sumit Rahman (not verified) wrote,
Thu, 24/05/2012 - 08:38
Another very good critical discussion of a sexed-up statistic. The point made about the choice of denominator (excluding non-premature deaths even though most of us will not suffer from premature death) is particularly illuminating, aside from the usual nonsense (EU equals UK; wide confidence intervals; "attributed to" equals "causes").
Ivan D (not verified) wrote,
Sat, 26/05/2012 - 15:25
Professor Anderson has some other difficulty with the English language based on his use of "due to" apparently as a substitute for "caused by" to describe figures that are for the most part derived from estimates and assumptions. The public health industry in general is becoming far too casual with respect to causation and correlation so I am not entirely surprised.
I would like to point out to the Professor that death is not preventable so what he and his industry aspire to is modification of how and why people die in line with what they consider acceptable. The problem is that his industry focuses far too much on top down statistics based solutions that ignore the individual nature of human beings and have historically proved to be both socially divisive and ultimately ineffective. If the Prof has some great ideas on treating addiction then I would love to hear them, but this report is just another example of the tiresome, mendacious attention seeking behaviour that typifies his industry and lines such as “In any case, where it has been studied, the health benefits of a price increase outweigh any harm that might result from increased illicit production and consumption” demonstrate a rather cavalier approach to the actual evidence that insults the readers intelligence.
I personally would not trust any analysis that had anything to do with Jürgen Rehm based on previous output including a BMJ paper in which relative risk /alcohol unit was claimed to two decimal places incredibly based on raw data derived from questionnaires. That paper was full on political rhetoric but somewhat lacking in the results section that would have allowed the reader to test its conclusions. Such are the standards of medical journals these days.