3

A recent study on the cancer risks of drinking alcohol has been widely reported in news headlines. The Times headline is:

Half-bottle of wine a night ‘as bad as three cigarettes’

The BBC summarise slightly differently:

Drinking one 750ml bottle of wine a week increases the risk of developing cancer over a lifetime by the equivalent of 10 cigarettes a week for women and five for men, a study says

They're both reporting the same study.

Not every expert agrees. David Spiegelhalter says (my emphasis):

If cigarette-equivalents were to be used to communicate the cancer risk of alcohol consumption, it is vital that their impact is properly evaluated to check they do not produce unreasonable concern – particularly as the overall health effect of moderate alcohol consumption is still contested.

Other experts' views can be found from the same source.

I've used one specific way of expressing the key issue as the headline for the question. But don't get bogged down with that specific formulation: other reports were all paraphrases of the results from the same study.

The key question is: given all we know about the harms of alcohol (which are very contested for moderate consumption) is the comparison with smoking (where the harms are clear and mostly uncontested) fair and a good summary of the totality of evidence?

matt_black
  • 56,186
  • 16
  • 175
  • 373
  • 1
    Both sources are basically saying the same thing in summarizing the same study. This is a report on the results of this one study with a particular methodology for data collection. I don't see what there is to be skeptical about; one can have an opinion about whether the way this study's findings were reported but that's an opinion, likely about how to describe relative risks and the value of using comparisons/benchmarks. From a stats perspective, it's pretty irresponsible to report estimated means without some measure of variability but that is in no way specific to this particular example. – Bryan Krause Mar 28 '19 at 18:19
  • 2
    @BryanKrause Hence why I asked whether the reports from this single study are a reasonable summary of what else we know. There is enough bad science that any *single* study should always be treated with skepticism. So when a surprising result appears it must be evaluated in the light of other studies. Especially when it gets such widespread, fear-inducing headlines. – matt_black Mar 28 '19 at 18:23
  • 1
    Even the original article contains some cautions about interpreting their results. Yes, of course, any single study should be treated with caution. But there's no reason to be skeptical of the actual claim made by the articles, which is that in this study, the relative cancer risks between smoking and drinking were as reported. – Bryan Krause Mar 28 '19 at 18:26
  • 2
    I could certainly see half a bottle of wine every night causing more problems than half a bottle of wine per week. – JAB Mar 28 '19 at 19:42
  • 2
    I don't think *is the comparison with smoking fair (and a good summary of the totality of evidence)* is a question for this site. –  Mar 29 '19 at 08:20
  • 1
    @JanDoggen Why not? The claim has been expressed this way in numerous headlines. So there is a claim using this comparison. Debunking claims like that is totally on topic here either when the headlines misrepresent the science or when that particular pice of science is at variance with other scientific knowledge. Perhaps I could phrase it better but I'm trying to avoid a descent into narrowly checking the headline and ignoring the broader science. – matt_black Mar 29 '19 at 10:20
  • 1
    Probably worth pointing out that the study is only comparing cancer risks, and ignoring other health risks. – DJClayworth Mar 29 '19 at 17:55

0 Answers0