A paper entitled The relationship between alcohol use and long-term cognitive decline in middle and late life: a longitudinal analysis using UK Biobank was published last week in the reputable Journal of Public Health. The authors used data on 13,342 people who had been asked about their alcohol consumption and then given reaction time tests both at baseline and after five years, and came to the strong conclusion that “Consuming more than one UK standard unit of alcohol per day is detrimental to cognitive performance and is more pronounced in older populations”. Current guidelines are to keep to two units a day, but the authors claim that “Our findings suggest that to preserve cognitive performance 10 g/day is a more appropriate upper limit. This would translate into not more than one UK standard unit of alcohol each day”. The Times covered the story uncritically, and the Sun had this wonderful headline.
But did these boffins’ data justify their claims? Just a quick look at the academic paper’s Figure 1 (reproduced below) suggests the authors’ conclusions are bizarre, to say the least.
The response times improve rather dramatically with increasing alcohol consumption, down to a minimum when daily alcohol consumption is around 16g, which is exactly the two units a day under current guidelines. Then there is a (rather gentle) slowing in reaction times for increasing consumption up to very heavy drinking. The confidence intervals are very wide, but the estimated response time at two units a day is clearly estimated to be lower than at one unit a day, so if anything the study supports the current guidelines.
So the conclusions that one unit a day is preferable to two units do not seem in the least justified by the fitted model, and may have arisen from a misinterpretation of the curve-fitting technique**. Whatever the cause, it would seem inappropriate that this paper, as it stands, is part of the scientific literature.
And in fact a more suitable headline might be: drinking up to current guidelines linked to improved cognitive performance.
- * The threshold of 10g did not come from the data on response times, but was chosen as a rather arbitrary position of a ‘knot’ in the fitted cubic spline.