Towards zero

For the last couple of years I have been predicting that the time will come when the recommended drinking guidelines (colloquially known as "limits") will fall to zero. One day we will be told that the daily/weekly units system is too confusing for the plebs and that there is no safe level of alcohol consumption. The best advice, the experts will say, is to not drink at all.

Gradually, the evidence is being gathered for this zero tolerance line. Most recent was this, reported in the Independent:

Even light drinking increases cancer risk

New study indicates that it's not just heavy drinkers who need to worry about the health implications of alcohol

Just one alcoholic drink a day may increase the risk of cancer, according to a new study, which estimates that light drinking is responsible for 34,000 deaths a year worldwide.

This is based on a study—or, more likely, a meta-analysis—the results of which have been extrapolated. The relative risks that the 34,000 figure have been extrapolated from are things like this...

One drink a day increased the risk of cancer of the oesophagus by almost a third, according to the study being reported in the Annals of Oncology, which analysed data from more than 200 research projects. Low alcohol intake increased the risk of oral cavity and pharynx cancer by 17 per cent, and breast cancer in women by 5 per cent.

Let's leave aside the unlikelihood of questionnaire-based epidemiology being able to identify one risk to two decimal points (until relatively recently, if you'd have told an epidemiologist that x increases risk by 5% (ie. a relative risk of 1.05), he would have laughed in your face).

Instead, and for the sake of argument, let's assume that these risks are accurate. What do they mean for the individual? I have not been able to find the mortality rate for oesophageal cancer, but there were 1,975 deaths from cancer of the oral cavity & pharynx in England & Wales in 2010. Since there were 493,242 deaths altogether, we can say as a rough but useable estimate that the risk of dying from those diseases is about 0.4%.

In the same year, 10,290 women died from breast cancer, out of a total of 255,326 female deaths. The chance of a British woman dying from breast cancer is therefore roughly 4%.

The absolute risk of mortality from the former, and possibly the latter, cancer will be somewhat lower for teetotallers, but we shall be conservative and stick with the figures above as being the baseline risks.

If having a drink a day raises the individual's risk of dying from oral cavity & pharynx by 17%, then the absolute lifetime risk rises from 0.40% to 0.47%. If having a drink a day increases the risk of breast cancer by 5%, then the absolute lifetime risk rises from 4% to 4.2%.

I wonder if this information—if given in relevant, rather than relative terms, as I have above—would make anyone think it was worth living a life of self-denial in order to reduce their risk by a fraction of one percent. Considering that the lifetime risk of dying from all cancers combined is somewhere between a quarter and a third, I doubt it would.

Risks of this order simply have no practical relevance to human beings. The relative risks are so tiny—even if we assume them to be real—that it is irrational to worry about them. The only way they can be made to sound scary is by doing what the authors of this study have done, ie...

(a) only give relative risk and hope the public mistakes it for absolute risk (a common error)

(b) extrapolate your numbers across the largest population you can think of. In this instance, they've gone for the nuclear option of using the population of the entire world—this should always ring alarm bells

If you do both of these things, you can then use the old line about the risks not being very high, but that "in terms of public health" they are of great significance. This get-out clause is often to be found in the text of epidemiological studies when the findings are of negligible significance for a person's health, and we see it again here:

"Alcohol increases the risk of cancer even at low doses," say the researchers. "Given the high proportion of light drinkers in the population, and the high prevalence of these tumours, especially of breast cancer, even small increases in cancer risk are of great public health relevance."

No matter how many times this line gets used, it never gains any meaning. A risk that has no practical significance to an individual should be of equally little significance to public health. The public, after all, is no more than a collection of individuals. Spouting variations of "many a mickle makes a muckle" does not alter that fact. If the public health industry considers risks which are of no significance to people to be of great importance to them, it only shows how irrelevant the public health industry is to the lives of normal men and women.

Professor Peter Johnson, Cancer Research UK's chief clinician, said: "This study adds to the evidence linking alcohol consumption to several types of cancer, and confirms that even light drinkers have a small but definite increase in the risk, particularly for those parts of the body, such as the throat and oesophagus, that come into direct contact with alcohol.

"People who wish to minimise their risk of cancer can help by cutting down on their drinking."

This thinly-veiled plea for abstinence is not justified by the scientific evidence. Only this week we have seen the latest in a long line of epidemiological research concluding that moderate drinking is very good for health, and not just for rare diseases. As Crampton mentions, a study in the European Heart Journal found a statistically significant reduction in the risk of total mortality of 22 per cent and a 42 per cent reduction in mortality from heart disease. It also looked at the "sick quitter" hypothesis so beloved of Ben Goldacre and David Nutt and found that it did not explain the results. There is now masses of evidence that there is a U-shaped curve for drinking and mortality, and especially heart disease (which is the single biggest cause of death in the UK).

That study did not get reported in the Independent and we continue to wait for a study to be published showing how many people are being "killed" worldwide through teetotalism. Don't hold your breath.