Remington 2001.

Study characteristics
Methods	Cross‐over trial of clomipramine versus placebo
Participants	Inclusion criteria: children and adults 10‐36 years DSM‐4 diagnosis of autism recommendation, based on initial assessment, of pharmacotherapy evidence that haloperidol or clomipramine had not been used previously or, if so, that an adequate therapeutic trial was not completed Exclusion criteria: with the exception of "antiparkinsonian medication in the form of benztropine...no other psychotropic medications were permitted during the study." Location/setting: "recruited from the Autism and Pervasive Developmental Disorder Clinic at the Centre for Addiction and Mental Health, Clarke Division, a teaching hospital associated with the University of Toronto." (Canada) Sample size: 36 (phase 1) Number of withdrawals/dropouts: 5 withdrew from placebo group because of behaviour; 5 withdrew from clomipramine group, 2 because of behaviour and AEs, and 3 because of AEs alone Gender: 31/36 male Mean age: 16.3 years IQ: not reported Baseline ABC‐I or other BoC: ABC‐I 19.0 Concomitant medications: 13/36 were taking other medications such as trifluoperazine, methylphenidate, fluvoxamine or carbamazepine History of previous medications: not reported
Interventions	Intervention 1 (clomipramine) for 7 weeks: 25 mg at bedtime for 2 days, 25 mg twice a day for 2 days, 25 mg 3 times/day for 2 days, and 50 mg twice a day Intervention 2 (haloperidol): 0.25 mg at bedtime for 2 days, 0.25 mg twice a day for 2 days, 0.25 mg 3 times/day for 2 days, and 0.5 mg twice a day Comparator (placebo) for 7 weeks: placebo equivalent
Outcomes	Primary outcomes: irritability, measured using the ABC‐I subscale (Aman 1985) AEs Secondary outcomes: none reported Timing of outcome assessments: not reported
Notes	Study start date: not reported Study end date: not reported Source of funding: grant from the Ontario Mental Health Foundation Conflicts of interest: none declared Trial registry: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Participants were randomly assigned to one of three treatment groups as part of a Latin square design (clomipramine‐placebo‐haloperidol, placebo‐haloperidol‐clomipramine, and haloperidol‐clomipramine‐placebo)
Allocation concealment (selection bias)	Unclear risk	Quote: "Medications and placebo were packaged in similar capsules to maintain the double‐blind component"
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Details not provided
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Details not provided
Incomplete outcome data (attrition bias) All outcomes	Low risk	LTFU clomipramine: 20 in total withdrew: Trial one, 5 withdrew, 2 because of behaviour and AEs, and 3 because of AEs alone. Trial 2: 9 withdrew from clomipramine group; 4 due to behaviour, 1 due to AEs and behaviour, 4 due to AEs alone Trial 3: 6 withdrew from clomipramine group (3 due to behaviour and 3 due to AEs) LTFU haloperidol: 10 in total withdrew from haloperidol group. Trial 1: 3 withdrew, 2 because of AEs and 1 because of behaviour and AEs Trial 2: 4 due to AEs, and 2 due to behaviour Trial 3: 1 due to behaviour and 2 due to AEs LTFU placebo: 10 in total withdrew Trial 1: 5 because of behaviour Trial 2: 4 due to behaviour Trial 3: 1 due to behaviour
Selective reporting (reporting bias)	High risk	Outcomes including ABC subscales measured every 2 weeks, however these results were not reported.
Other bias	Low risk	No other sources identified