Glass bottle with alcohol.

    The problem with the public health message on alcohol

    1 December 2017

    Christopher Snowdon recently wrote about an exchange of e-mails between Public Health England (PHE) and the Sheffield University group that was tasked to provide modelling on the revised alcohol guidelines. What he uncovered was surprising to many. You can read his report, and PHE’s response here.

    It is important to remind ourselves of the purpose of a public health message. It is to provide an opportunity for members of the public to improve their health. For example, by taking regular exercise, eating more fruits and vegetables, eating oily fish and not smoking. Whatever it may be, people are being advised to share a lifestyle habit with those in the population who are healthier. Before the last review, that used to be the case with public health message on alcohol consumption.

    I was a member of the Working Party set up by the Department of Health that produced the 1995 Report ‘Sensible Drinking’, which led to the previous public health message on alcohol. We undertook a comprehensive review of the relationship between alcohol consumption and chronic disease, and we concluded that excessive alcohol consumption leads to accidents and addiction, and increases the risk of certain chronic diseases, including cirrhosis of the liver, some cancers, psychiatric disorders, and high blood pressure, with consequential social and economic harm.

    However, we also found that at lower, moderate levels of consumption, it is a different story. There was – and still is a great deal of evidence that shows moderate consumption of alcohol is associated with significant protection for important diseases compared with those who choose to abstain, namely for coronary heart disease, ischaemic stroke, and diabetes. Now, dementia and cognitive decline can be added to this list.

    These important factors defined the 1995 Working Party’s approach to shaping our public health message, which was to separate alcohol misuse from sensible drinking. So we advised:

    • to avoid intoxication, and

    • not to drink in situations where it is important that judgement is not impaired, e.g when driving, taking part in physical sport, before using machinery, ladders, etc, and in the workplace.

    In other words, don’t misuse alcohol.

    Sensible drinking in 1995 was regarded as no more than 3 to 4 units per day for men and 2 to 3 units per day for women, levels of consumption we said were unlikely to accrue significant harm. This is also the level of intake that coincides approximately with the bottom of the all-cause mortality curve. This is because the beneficial effect this level of consumption has on important diseases. We also suggested men over 40 and post-menopausal women could if they wished take advantage of the health benefits by consuming 1 to 2 units per day.

    Of course, some have tried to explain away these beneficial effects. The ‘sick quitter’ hypothesis suggests that the reason for the higher disease burden among abstainers is due to former drinkers in the abstainer category whose health was ruined by alcohol consumption. But the same elevated risk is seen among lifetime abstainers. It has also been proposed that the decrease in risk among moderate drinkers is due their greater prevalence of healthy lifestyle factors, that is, they tend to be non-smokers, take regular exercise, eat a healthy diet, etc. But when researchers only included the healthiest people in an epidemiological study, they too benefited from moderate alcohol consumption.

    So although the 1995 advice on sensible drinking represented a level of alcohol consumption where risk was already at rock bottom, all this was swept aside in 2016 with new ‘low risk’ guidelines, which advised:

    • that there was no safe level of alcohol consumption,

    • that both men and women should consume no more than 14 units per week over 3 days or more,

    • that everyone should have several alcohol-free days each week, and

    • that alcohol is linked to the risks of heart disease, cancer, liver disease and epilepsy.

    Further, the Chief Medical Officer said that the protective effect of moderate consumption was ‘an old wives tale’.

    But how did such a dramatic volte-face come about? To understand that, it is necessary to explore how the 2016 review by Public Health England was carried out.

    First, they asked the Committee on Carcinogenicity for advice on the relationship between alcohol consumption and cancer, as the 1995 Working Party had done. The available body of evidence was greater than in 1995, particularly on breast cancer. The Committee concluded that even at low levels of alcohol intake, below 1.5 units per day, there was an increased risk of cancers of the oral cavity and pharynx, oesophagus and breast in women. This is the basis of the statement, ‘there is no safe level of alcohol consumption’. Also, the Committee concluded that the risk of getting cancer increases the more alcohol a person drinks.

    Rather than consider similar evidence on the relationship between alcohol consumption and other diseases, the team at Sheffield University used the Sheffield Alcohol Policy Model, a mathematical model, to identify a level of consumption whereby no more than 1 per cent of deaths in men and women would be attributable to alcohol, whether that be from chronic causes (i.e. diseases) or acute causes (i.e. accidents). For causes for which alcohol was regarded as partially responsible, the Sheffield group developed a ‘relative risk function’ to feed into their model.

    While the leading e-mail exchange has rightfully questioned where the Sheffield group provided valid recommendations, there were already a number of problems with this approach.

    1. The 1 per cent target is completely arbitrary, and could equally be 2 per cent or 0.1 per cent, or anything. Also the relative risk factors applied to diseases which are partially attributable to alcohol are a matter of subjective assessment based on particular meta-analyses of the data, so they could easily be changed. If different numbers are chosen, then the public health advice changes as well, so it was possible to justify any advisory level.

    2. The levels of intake generated by the Sheffield model are designed to achieve this 1 per cent population target. The idea is that everyone has to change their behaviour in order to achieve the target, whether or not that level of intake is appropriate to individuals, a policy more at home in Soviet Russia than in a western democracy. It has also led to some absurd conclusions, for example, that the maximum intake for men and women should be the same at 14 units per week, when in fact there are very good reasons for having different levels of intake for men and women. Women tend to be of smaller stature, they have a higher proportion of fat in their bodies (alcohol resides in the water phase), and they break down alcohol more slowly. But according to the model, having different levels would not achieve the target, so everyone is pinned down to 14.

    3. The chronic and acute effects of alcohol consumption are mixed up in the model, and not treated separately. So the risk of someone, for example, of driving a car when intoxicated and killing themselves is built into advice given to everyone, whether or not they drink and drive, or indeed, drive at all. I suspect that the reason why the advice not to get intoxicated and not to drink when driving, etc has now disappeared from the public health message, is that the risks run by those indulging in these activities are now built into the advice given to all, whether or not they are relevant.

    It is hard to avoid the conclusion that the 2016 review was undertaken with the principal objective of lowering the advisory level of alcohol intake, rather than updating the message to take account of the latest science.

    The big loser in all this is of course the British public. The previous message sought to eliminate alcohol misuse while advising a level of intake at which no significant harm accrued, and also taking advantage of the protection moderate alcohol consumption offers to the principal diseases of ageing, with enormous potential benefits to individuals and the country as a whole. Now, the protective effect has been mathematically-modelled out of existence (except, apparently, for women over 55 drinking 5 units per week), and there is no safe level of consumption.

    The bottom line is that men and women who drink alcohol moderately and regularly tend to have longer and healthier lives than those who do not. An important opportunity to tell the country the full truth and communicate an appropriate public health message has been thrown away.