Buitelaar 2001.

Methods	Double‐blind, randomised controlled trial of risperidone and placebo
Participants	Setting: inpatients (2 centres) Sample size: 38; 19 active treatment, 19 placebo Sex: 33 males, 5 females Age range: 12 to 18 years Mean age: active treatment 14 (SD = 1.5) years; placebo 13.7 (SD = 2.0) years IQ range: 60 to 90 Inclusion criteria: persistent overt aggressive behaviour as evidenced by ≥ 1 on Overt Agression Scale ‒ Modified; failure of behavioural treatment. Participants were included if "their aggressive behavior failed to respond to behavioral treatment approaches (typically, these behavioral treatments involve contingency management and social skills training delivered on an individual basis for at least 2 months)." (p 240) Diagnosis: DSM‐IV criteria conduct disorder, oppositional defiant disorder Comorbidity: ADHD (14 active treatment, 12 placebo) Withdrawn/dropouts: 2 withdrawals (2 placebo)
Interventions	Intended dose: from 0.5 mg twice daily increased by 1 mg up to 5 mg. As fixed as possible, could be adjusted down if adverse event present. Mean end dose: 63% on 3 mg a day, mean 2.9 mg (1.5 to 4 mg)
Outcomes	Primary outcomes: Clinical Global Impression ‒ Severity scale Secondary outcomes: Overt Aggression Scale ‒ Modified, Aberrant Behaviour Checklist (all scales nurse and teacher‐rated) Follow‐up interval: 6 weeks
Notes	Imputation method for incomplete data: last observation carry forward (LOCF) (p 242) Funding/support: "Supported by Janssen‐Cilag, BV, Tilburg, the Netherlands" (p 239)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization code had been generated by computer in block of four numbers" (p 241).
Allocation concealment (selection bias)	Unclear risk	Not stated.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	"Dosage was adjusted by the responsible psychiatrist who was blind to the treatment" (p 241).
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not stated.
Incomplete outcome data (attrition bias) All outcomes	Low risk	"Two participants (11%) in the placebo group stopped treatment during the double‐blind period because of lack of therapeutic effects and uncontrollable aggressive behavior" (p 242).
Selective reporting (reporting bias)	Unclear risk	Protocol unavailable.
Other bias	High risk	No reason given why 145 approached and 49 found to be eligible. "Greater severity of psychosocial stressors in the risperidone group" (p 242).