Re: Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data
I am astonished to read scientific conclusions from no less than 113 researchers. Powerful opinions underpinned by a plethora of opinion leaders.
Thinking of normal human behaviour I am wondering how so many researchers came to one single and clear-cut conclusion. And this on every single detail? Reading then the BMJ’s own rules on authorship, we should assume that every of these 113 authors substantially contributed ... and drafted or critically revised it ... and gave the final approval ... and agreed ... to all aspects ... related to accuracy and so on. Superb job of such a group!
However, experience and social sciences tells me that groups larger than 7 persons tend to become ineffective and overlook errors, just because large groups tend to rely that dozens of peers might have already criticised a certain point making a repetition of criticism unnecessary.
Looking only to few words in the conclusion, I fear that this happened here:
“...associated with non-drinking and lower alcohol consumption...”
Apart from the nasty criticism concerning the logical “and”. Certainly non-drinking will finally be correlated to lower alcohol consumption, and it appears that this might be the true message. Otherwise, a non-excluding simple “or” might have fitted better.
However and more importantly, the actual data does not seems to support this conclusion.
The actual numbers I only found in Figure 2 of the article. If I interpret the data correctly, then 20 259 seemingly had that allele and 168 731 had not, yielding 10.7% with the allele. Then, the figure tells me that 12.0% of the non-drinkers had the allele and 8.6% of the “drinkers only”. Okay, those with the allele are a bit more frequent among non-drinkers, but is this difference impressive?
The split by units/week in this figure makes it even more disturbing: Light consumers with allele: 9.0%, moderate with allele: 8.7%, and heavy with allele 10.5%. These findings are in total contradiction to what the authors wrote in the introduction, namely that this allele would be correlated to “lower levels of usual alcohol consumption and blood ethanol levels”.
How can the authors explain that those with the allele were more frequently heavy drinkers?
I am astonished to read scientific conclusions from no less than 113 researchers. Powerful opinions underpinned by a plethora of opinion leaders.
Thinking of normal human behaviour I am wondering how so many researchers came to one single and clear-cut conclusion. And this on every single detail? Reading then the BMJ’s own rules on authorship, we should assume that every of these 113 authors substantially contributed ... and drafted or critically revised it ... and gave the final approval ... and agreed ... to all aspects ... related to accuracy and so on. Superb job of such a group!
However, experience and social sciences tells me that groups larger than 7 persons tend to become ineffective and overlook errors, just because large groups tend to rely that dozens of peers might have already criticised a certain point making a repetition of criticism unnecessary.
Looking only to few words in the conclusion, I fear that this happened here:
“...associated with non-drinking and lower alcohol consumption...”
Apart from the nasty criticism concerning the logical “and”. Certainly non-drinking will finally be correlated to lower alcohol consumption, and it appears that this might be the true message. Otherwise, a non-excluding simple “or” might have fitted better.
However and more importantly, the actual data does not seems to support this conclusion.
The actual numbers I only found in Figure 2 of the article. If I interpret the data correctly, then 20 259 seemingly had that allele and 168 731 had not, yielding 10.7% with the allele. Then, the figure tells me that 12.0% of the non-drinkers had the allele and 8.6% of the “drinkers only”. Okay, those with the allele are a bit more frequent among non-drinkers, but is this difference impressive?
The split by units/week in this figure makes it even more disturbing: Light consumers with allele: 9.0%, moderate with allele: 8.7%, and heavy with allele 10.5%. These findings are in total contradiction to what the authors wrote in the introduction, namely that this allele would be correlated to “lower levels of usual alcohol consumption and blood ethanol levels”.
How can the authors explain that those with the allele were more frequently heavy drinkers?
Competing interests: No competing interests