Brodaty 2003 RIS‐AUS‐05.

Study characteristics
Methods	Study design: randomised controlled trial Study grouping: parallel group Study duration: 12 weeks Rescue medication: anticholinergic medication was allowed to treat EPS only if a reduction in trial medication dose was not effective. Treatment of urinary incontinence with low‐dose tricyclic antidepressants or anticholinergic medication was allowed to continue. Low‐dose oxazepam was permitted to treat agitation, provided that usage did not exceed 4 days in a 7‐day period. Short‐acting sedative/hypnotic agents prescribed chronically for insomnia at baseline were permitted if the clinician judged that they could not be discontinued. Under exceptional circumstances, initiation of night sedation for insomnia was allowed using a short‐acting benzodiazepine (preferably oxazepam at the lowest effective dose). Narcotic analgesics were permitted, provided that the dosage had been stable for at least 3 months and that they were not prescribed to control agitation or aggression.
Participants	Number randomised: 345 (173 risperidone 172 placebo) Mean age: 83.5 years Sex (female): 84.9% Type of dementia: Alzheimer's disease, vascular dementia, mixed type dementia Severity of dementia: severe Indication: aggression (is subtype of agitation); subgroup analysis in patients with psychosis (additionally) was reported (Baseline CMAI total aggression: 33.5) Setting: nursing home Country: Australia and New Zealand
Interventions	Intervention characteristics Risperidone dose: 0.25mg to 1 mg twice daily Placebo Mean risperidone dosage was 1.03 ± 0.61 mg/day.
Outcomes	Agitation: Cohen‐Mansfield Agitation Inventory (CMAI) total aggression subscale Number of responders for agitation Extrapyramidal symptoms Somnolence Death Any adverse event Any serious adverse event Discontinuation (any reason) Discontinuation due to adverse events
Identification
Notes	Sponsorship source: Janssen‐Cilag Australia and Johnson & Johnson, L.L.C.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Comparability of groups (selection bias)	Unclear risk	No baseline characteristics for all randomized per group are shown (n=156 instead of 172 for placebo; n=153 instead of 173 for risperidone). Small differences in the reported baseline characteristics, e.g. more aggression and NPS in risperidone group. Most are not adjusted for in the analyses. Unclear how the differences could have affected the estimated effects.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	No information
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	High risk	Agitation: In the article, 152 for placebo and 149 for risperidone were reported instead of 170 and 167 who received at least once the study medication.
Selective reporting (reporting bias)	High risk	Results on all reported measures are reported. However, effect on MMSE and FAST not reported in numbers. Post‐hoc subgroup analysis with respect to effect on psychosis.One site excluded. Only adjusted least square means are reported. Crude means are only mentioned in a figure without providing exact numbers. No CI or SD or SE are reported for the change from baseline. Only difference of LS between placebo and risperidone.
Other bias	High risk	The double‐blind treatment period was preceded by a maximum 7‐day, single‐blind washout period, during which patients took 0.5 mL of placebo oral solution each evening while existing psychotropic medication was discontinued.