Hakim 2012.

Methods	RCT comparing risperidone and placebo for prevention of conversion of subsyndromal delirium to delirium Study took place in a single cardiosurgical ICU in Egypt
Participants	Participants included 101 elderly (aged ≥ 65) (N = 51 risperidone, mean age not provided, 33/51 male; N = 50 placebo, mean age not provided, 36/50 male) on‐pump cardiac surgery patients with diagnosis of postsurgical subsyndromal delirium (ICDSC score 1 to 3) Study enrolment between December 2007 and November 2010
Interventions	Participants with subsyndromal delirium (ICDSC score 1 to 3) received oral 0.5 mg risperidone or placebo every 12 hours until 24 hours after subsidence of subsyndromal delirium (ICDSC score of 0) or development of frank delirium (ICDSC ≥ 4). Among delirious participants, treatment allocation was revealed, and placebo‐treated patients were started on 0.5 mg oral risperidone every 12 hours. If symptoms remained uncontrolled, the dose was increased to a maximum of 4 mg/d. Among delirious risperidone‐treated patients, the dose was increased until symptoms were controlled or a maximum dose of 4 mg/d was attained. Haloperidol was used in both groups if symptoms were not controlled with maximal risperidone dose. Haloperidol was started at 0.5 mg every 8 hours and could be increased to 10 mg/d if needed. Haloperidol dose could be doubled every 24 hours until symptoms were controlled or the maximum dosage was attained Rescue medications were continued for 24 hours after a score of 0 was achieved on the ICDSC Assessment:screening for subsyndromal delirium was done using the ICDSC, began 4 hours after extubation in the ICU, and was continued for every 8‐hour nursing shift thereafter, including after discharge to the cardiosurgical ward Non‐drug strategies: not reported
Outcomes	Primary (end of study) Incidence of delirium Secondary (end of study) Duration and severity of delirium Length of ICU and hospital stay Occurrence of adverse events (e.g. extrapyramidal symptoms)
Notes	Funding Support for the study was provided solely by institutional and/or departmental sources Registration No trial registration number was included in the manuscript
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomized by a clinical pharmacist in a 1:1 ratio via a computer‐generated random numbers list created with GraphPad StatMate v.1.01i software (Graph‐Pad Software Inc., San Diego, CA, USA) using permuted blocks of size 4
Allocation concealment (selection bias)	Low risk	Allocation group codes were typed and were kept at the pharmacy in sealed envelopes. Treatment concealment was maintained until recruitment, data collection, and analysis were completed, unless an emergency warranted otherwise and was requested by an attending physician
Blinding of participants and personnel (performance bias) All outcomes	Low risk	This was a double‐blind study. Test drugs were prepared by the hospital’s pharmacy and were identical in appearance and odour. Drugs were dispensed in identical containers sealed and numbered according to the random number list
Blinding of outcome assessment (detection bias) All outcomes	Low risk	4 intensivists and 3 ward physicians, who were blinded to group allocation, were charged with screening participants for subsyndromal delirium using the ICDSC. Randomized participants were assessed by a blinded observer using the ICDSC, and those scoring > 3 were evaluated by a blinded psychiatrist to confirm delirium
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were analysed using an intention‐to‐treat principle. All randomized participants were included in the analysis
Selective reporting (reporting bias)	Unclear risk	All outcomes reported in methods were included in results. Data presented for all participants were included in the analysis. However without a published protocol or trial registration, it is unknown if all outcomes were reported as planned
Other bias	Low risk	All participants were permitted rescue haloperidol Sample size calculation was provided