A pooled estimate can temper the impression created by a few positive amphetamine experiments. In a 2020 meta-analysis, d-amphetamine did not produce a statistically significant overall improvement in cognition among healthy adults who were not sleep deprived. That finding also failed an equivalence test: the evidence did not establish that the drug’s cognitive effect was meaningfully absent.
Which experiments were combined?
Roberts and colleagues searched through November 2019 and conducted separate analyses of modafinil, methylphenidate and d-amphetamine. The final d-amphetamine analysis contained ten published studies, 27 effect sizes and 337 participants in repeated-measures designs, after removal of one outlying effect size. Twenty-seven effect sizes are not 27 independent experiments or 27 groups of new participants.
Eligibility concerned acute administration in adults without reported psychiatric or neurological disorders. These were laboratory cognition experiments, rather than studies of ADHD treatment, chronic methamphetamine exposure, sleep-loss countermeasures or actual workplace productivity. Those populations and purposes should not be substituted for the reviewed question.
The central estimate was uncertain and heterogeneous
The overall standardized mean difference was 0.21, with a 95% confidence interval from −0.06 to 0.47; p was 0.13. Positive values favored the drug. The interval included both no average benefit and a potentially positive effect. Between-result heterogeneity was high, at I² = 91%.
There were no statistically significant benefits in the reported domain analyses either. For example, updating had an estimate of 0.03 (95% CI −0.19 to 0.24), while spatial working memory had an estimate of −0.01 (−0.50 to 0.48). These are different outcomes with different uncertainty, rather than interchangeable measures of “being smarter.”
Why a nonsignificant result is not proof of no effect
The authors additionally used equivalence testing with bounds of −0.20 and 0.20 standardized units. D-amphetamine did not establish equivalence within those bounds. Their ordinary significance test and their equivalence test therefore answered separate questions, and neither supplied a definitive universal null conclusion.
The overall result was robust to removing individual effect sizes in the sensitivity analysis. Some domain conclusions were more sensitive: the sustained-attention finding changed when one contributing study was removed. Broad pooled conclusions and narrow task conclusions deserve separate confidence.
Selection decisions shape the estimate
When experiments tested several doses, the reviewers selected the highest dose. They generally selected the most difficult task condition and a testing time intended to approximate peak plasma exposure when several times were available. These choices made the synthesis manageable, but did not identify the best dose, task difficulty or testing time for each person.
The review also excluded published papers whose necessary numerical data could not be obtained. It reported 22 such exclusions in its selection account; the methods described 23 unmet data requests. Those counts should not be silently treated as identical. Missing extractable results remain a limitation even when a publication-bias test is reassuring.
What can be said outside the laboratory?
The defensible historical conclusion is limited: this synthesis did not demonstrate an average cognitive-enhancement effect of acute d-amphetamine in rested, healthy adults. It did not prove that nobody benefits on any task, establish the effects of repeated nonmedical use, or compare d-amphetamine directly with the other drugs in one randomized population. Separate positive and negative meta-analyses do not themselves prove one drug superior to another.
A claim about academic achievement, job performance or driving would need evidence measuring that outcome. A laboratory drug effect, a subjective feeling of productivity and a real-world functional benefit are separate propositions.
Source
Roberts CA, Jones A, Sumnall H, Gage SH, Montgomery C. How effective are pharmaceuticals for cognitive enhancement in healthy adults? A series of meta-analyses of cognitive performance during acute administration of modafinil, methylphenidate and D-amphetamine. European Neuropsychopharmacology. 2020;38:40–62. doi:10.1016/j.euroneuro.2020.07.002. Historical research synthesis, not current prescribing guidance.
Leave a Reply