Cherry-picking the evidence on alcohol and health

You can prove anything if you discard 99.9% of the evidence

Tim Stockwell has been up to his old tricks. In a study that was widely publicised this week despite being published in January, he claims - yet again - that moderate drinking does not confer health benefits. The study is largely a rehash of his meta-analysis from last year (which I wrote about here) so there isn’t much more to say except to note the extraordinary amount of cherry-picking that is required to come to such a conclusion. 

He and his team started with 3,248 relevant studies of which 3,125 were immediately discarded. This left 123 cohort studies to which they added 87 relatively recent cohort studies. They then discarded 103 of these because they didn’t meet Stockwell’s increasingly stringent and somewhat arbitrary criteria. This left 107 studies, but there was still work to do. 

In the previous meta-analysis, Stockwell decided that only 21 of these studies “strictly defined lifetime abstainers” and were therefore worthy of his attention. As I mentioned at the time (but he didn’t), three quarters of these found clear evidence of a J-Curve, i.e. lower risk of all-cause mortality among moderate drinkers than among teetotallers. 

In the new study, he introduces yet another filter for “quality” and reduces the number of studies down from 21 to 18 and then 15, but these still show lower risk for moderate drinkers, so he introduces some more criteria until a vast literature built up over 50 years is whittled down to just six studies. This gets rid of the apparent benefits of moderate drinking. He then removes one more study and, voila!, moderate drinkers are now at greater risk than teetotallers.

If this isn’t blatant cherry-picking, I don’t know what is. The five studies below are the ones he is left with...

I’d be interested to read these studies, but he doesn’t reference them and I can’t find relevant studies that match the name and date. All these authors were in his list of “bias-free” studies last year, but the dates don’t match so perhaps he simply got the dates wrong this time? If so, he must be referring to these...

Rehm et al. (2001) - Supports the J-Curve. It found ‘a significant influence of drinking alcohol on mortality with a J-shaped association for males and an insignificant relation of the same shape for females’.

Sempos et al. (2003) - Doesn’t support the J-Curve. Study of African Americans found no J-Curve. The authors say this ‘may be the result of the more detrimental drinking patterns in this ethnicity and consequently the lack of protective effects of alcohol on coronary heart disease’.

Makela et al. (2005) - Supports the J-Curve. ‘Moderate drinking is associated with a lower risk of IHD [ischemic heart disease], whereas drinking in a heavy episodic manner (often referred to as "binge drinking") is not.’ Women who drank moderately had lower rates of all cause mortality but this was not statistically significant among men.

Nakaya et al. (2004) - Doesn’t support the J-Curve. This study from Japan found a linear increase in all cause mortality risk as people drank more. The associations were not statistically significant.

Zaridze et al. (2014) - Insufficient evidence either way. This study from Russia found all cause mortality higher among vodka drinkers although ‘there were too few deaths to determine reliably whether a little vodka consumption provided a slight protective effect or a slight hazard.’ 

Considering he had thrown out literally 99.9% of the evidence to get this far, the remaining studies hardly make a compelling case for Stockwell’s contrarian claim. Nor is it easy to see what is so great about any of them compared to many of the studies he binned. 

It’s just the usual Stockwell garbage designed to attract headlines and clog up the search engines with content to sow doubt in the minds of casual readers. In his comments to the Guardian he more or less admits that he is only doing this because the benefits of alcohol consumption are inconvenient to people like him who want to regulate booze like tobacco...

“It’s been a propaganda coup for the alcohol industry to propose that moderate use of their product lengthens people’s lives,” said Dr Tim Stockwell, first author on the study and a scientist at the Canadian Institute for Substance Use Research at the University of Victoria.

“The idea has impacted national drinking guidelines, estimates of alcohol’s burden of disease worldwide and has been an impediment to effective policymaking on alcohol and public health,” he added.  

Speaking of national drinking guidelines, it’s probably no coincidence that this six month old study was press released days before a consultation on alcohol and health is due to close in the USA. Stockwell's long time collaborator Tim “two drinks a week” Naimi is already on the panel. As usual, it’s not really about science. It’s about politics.

If you want a more accurate analysis of the evidence, read this article. It’s from 2000, but the evidence of a protective effect from moderate drinking has only grown stronger in the years since.

The evidence amassed to date on the link between moderate alcohol intake and reduced risk of dying of cardiovascular disease might be thought already sufficient to bracket sceptics of alcohol’s protective effect with doubters of manned lunar missions and members of the Flat Earth Society. Published studies demonstrating this link can now be counted in the hundreds, and no fewer than six plausible underlying biological mechanisms have been identified.

… Alternative explanations for the protective effect of moderate alcohol intake, relying on ever more tenuous confounding effects, have been discredited one by one. The apparently protective effect of moderate alcohol consumption has so far survived the use of controls for sociodemographic status; for the “sick quitter hypothesis”! (ie, the suggestion that many abstainers have stopped drinking because of serious illness), for the amount of cholesterol in the diet, and even for the degree of social isolation. As evidenced in the study by Simons et al. the protective effect is fairly specific to cardiovascular disease and does not operate for other major causes of death in older people such as cancer. While a handful of recent studies have failed to find a protective effect for moderate drinking, these are still heavily outnumbered by those with positive findings. In fact, the range of different countries and cultures in which the phenomenon has been documented is also testimony to its robustness, even if different levels of consumption appear to provide the benefit in different drinking cultures.

That article was written by Tim Stockwell back in the days before he became president of the neo-temperance Kettil Bruun Society and before he started taking money from state-owned Scandinavian alcohol monopolies that are keen to emphasise the benefits of heavy regulation.