Reacción a "Reaction to study showing ADHD drugs may work against Alzheimer's disease"

This systematic review has several quality indicators: (1) The study protocol was previously registered and therefore the study appears free of selective reporting bias (reporting what is convenient for us). (2) It is reported according to the PRISMA guidelines, so the report aims to be transparent and reproducible. (3) It is based on clinical trials and therefore will not be affected by confounding of effects. (4) It combines effects using a more robust random effects model. (5) Its wording is cautious, tentative, warning of the steps needed before proposing a particular intervention ("there is a case for further clinical trials").

I have not found mention of the clinical value for patients of the main result: a standardised mean difference of 0.14, with CI [confidence interval] from 0.03 to 0.25. This means that the effect is about one seventh (0.14) of the usual differences between individuals. A very very small effect, therefore. As this is opinionated, my disappointment is that it does not dispute this.

That said, the main result (and most secondary analyses) has a curious value of the I2 statistic, which they round to 0%, implying an implausible homogeneity of effect. This invites a careful reanalysis of these results.