Safety Denominators in the Autism Crossover Trial Exceeded the Initial Fifty-Person Arms

The autism trial’s safety summaries included exposure from both treatment periods. The number exposed to each preparation was therefore different from the initial randomized group size.

Participants could contribute to more than one treatment category

Fifty participants began in each of the three first-period groups. Across both periods, ninety-five received whole-plant extract, ninety-three purified cannabinoids and ninety-four placebo. These exposure counts reflect the crossover structure and are not three independent groups of new participants.

Efficacy and safety used different scopes

Because of a treatment-order effect, the principal efficacy comparisons were restricted to the first period. Safety and tolerability data from both periods were retained. An adverse-event percentage should therefore use the appropriate exposed denominator rather than automatically dividing by fifty.

Event counts also differ from people with an event

Repeated reports can arise in one participant, while a participant can experience several types of symptoms. The study reported both events and affected participants. Those quantities should not be interchanged or added to imply an independent-patient total. The trial found more somnolence with cannabinoids, but its safety summaries do not establish an equivalent long-term rate in an unselected population. Accurate denominators preserve both the crossover design and the distinction between reports, exposures and individuals.

Related analysis: the study’s broader findings and limitations.

Research cited

Aran and colleagues (2021). Cannabinoid treatment for autism: a proof-of-concept randomized trial.