Friday, 8 April 2011

More fun with statistics

You can’t have missed the fanfare over the new study showing an increased cancer risk associated with alcohol consumption. As usual the statistics are presented in such a way as to scare the living bejesus out of even moderate drinkers. The statement “research has shown a link between alcohol consumption and cancers of the oesophagus, liver, bowel and female breast.” is not one I would dispute. However lets not forget that correlation does not equal causation, as we are all to eager to point out when dealing with quackery.

The three cancers linked to alcohol consumption in men are mouth/throat oesophagus, liver and bowel. So lets look at these in perspective.

Bowel.  This is the commonest of the three with an incidence of about 45 per 100,000, or roughly one in two thousand. According to this articleIn Great Britain male bowel cancer incidence rates rose slowly by an average of 1% each year between 1979 and 1999, since when there has been a slight decrease. Over the same period the female rates have changed very little” What increase there has been has been attributed to screening and improved diagnosis. These figures hardly speak of an increasing epidemic. 

Liver. This is a much less common cancer than bowel with an incidence of about 4 per 100,000, although this has increased significantly over the last few years. However the majority of new cases present in the over 70s. Lets face it you have to go somehow.
Mouth/throat/oesophagus. this is the least common with an incidence of 2 per 100,000. Although there has been an increase in the incidence lately this has been attributed to human papilloma virus, transmitted during oral sex. This is a cancer where the correlation/causation factor is most dubious. Is it oral sex, booze, or both? Is there a correlation between drinking and oral sex? Are you more likely to indulge in muff diving while pissed? I don’t suppose that the researchers have asked themselves these questions.

As usual I think this research has been blown up out of all proportion. I don’t believe anyone will take the slightest notice of it. I certainly won’t. As a last word I think this is a far more sensible view. And so is this.

1 comment:

  1. Couldn't agree more. When a scientist starts his study discussion with the words "If we assume causality", then my hackles rise. These observational studies, churned out over time from EPIC, seem to be politically skewed to fit with the preconcieved notions of the researchers and their masters.

    Even this study acknowledges the reduced incidence of CHD/CVD that moderate consumption of alcohol affords. So then are we to choose between a heart attack or the much lesser risk of cancer, which according to the ASR's for 2006 are 154.9 per 100,000 for all cancers?

    There currently is little to no evidence to connect cancer to alcohol, with the possible exception of liver cancer which has some correlation with fatty liver disease. But as you say observation (and correlation) does not prove causation. Wibble!