Pharmacological and dietary-supplement treatments for autism spectrum disorder: a systematic review and network meta-analysis.

Most trials were small (median 40 participants) and short (median 12 weeks), so the effects are imprecise and may not hold up in larger or longer studies. Evidence quality was low or very low for nearly every comparison, and the authors downgraded several results for reporting bias and indirectness. There was little data in adults, no usable data for methylphenidate, and anxiety and depressive symptoms were poorly studied, so the adult and mood findings rest on a handful of tiny trials. The networks were mostly star-shaped around placebo with only one or two trials per drug, so the treatment rankings and league tables should be read with great caution, and masked heterogeneity or incoherence cannot be ruled out. About half of the trials said they randomised participants without describing how, so the reader cannot check that the groups were comparable. Trials of antipsychotics were mainly conducted in children selected for irritability, so the improvements in core symptoms may be a by-product of reduced challenging behaviour rather than a direct effect. Two large phase 3 trials of bumetanide (422 children in total) were negative and stopped early but reported no usable data, so the positive bumetanide result in this review may be biased. About 30% of the included studies were funded by industry or had investigators who applied for a patent. The results for folinic acid, carnosine and vitamin D rested on one or two small trials and did not survive sensitivity analyses.

The work was funded by the Innovative Medicines Initiative, the Spanish Ministry of Science, Innovation and Universities, Instituto de Salud Carlos III, the Horizon 2020 Framework Programme and several Spanish foundations (Fundación Mutua Madrileña, Fundación Alicia Koplowitz, Fundación Familia Alonso). Within the trials reviewed, about 30% were funded by industry or had investigators who applied for a patent. No individual author declarations were reported in the supplied text.

Do not read the improvements in social-communication difficulties and repetitive behaviours as evidence that antipsychotics treat the core features of autism. Most antipsychotic trials enrolled children chosen because they had irritability, and the authors state that the core-symptom gains could be a by-product of reduced challenging behaviour rather than a direct drug effect. The evidence was also graded low or very low.