Mizuno 2011.

Methods	RCT. Multicentre, double‐blind. Comparing training using prisms with training without prisms Setting: Japan
Participants	38 participants (444 screened) Experimental group (prisms): n = 20; control group: n = 18 Recruited from rehabilitation departments from 8 hospitals in Japan Age ‐ mean (SD): experimental: 66.0 (11.5), control 66.6 (7.7) years Time from stroke ‐ mean (SD): experimental 67.1 (18.4); control: 64.4 (20.9) days Inclusion criteria: within 3 months of first ever right hemisphere stroke, 42 to 89 years old, neglect as assessed by BIT behavioural test Exclusion criteria: unable to sit in wheelchair, aphasia or cognitive impairment resulting in inability to understand task, unable to understand Japanese, impaired vision or hearing, impaired right upper limb, previous brain injury 34 participants completed intervention and follow‐up; 4 drop‐outs (1 control, 3 prisms) ‐ 2 stroke relapse, 1 refused, 1 developed delirium 31 participants completed follow‐up BIT
Interventions	2 daily training sessions, lasting 20 minutes, 5 days per week for 2 weeks; for a total of 20 sessions Training ‐ pointing at targets, whilst sitting at a table Experimental group: prisms (shifting visual field 12° to right, Fresnel lens). Pointing task ‐ 30 times without prisms; 90 times with; 60 times without Control group: neutral plastic glasses. Pointing task as for experimental group Routine stroke rehabilitation provided as usual For analysis of bottom‐up and top‐down rehabilitation approaches this review coded the experimental condition as bottom‐up
Outcomes	BIT CBS FIM Stroke Impairment Assessment Set Outcomes were recorded at baseline, after the 2‐week intervention and immediately prior to hospital discharge
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Computerised block randomisation, with pre‐stratification according to BIT behavioural test (dichotomised to above or equal to 55 or below 55) and participating hospital
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessor was masked to treatment allocation and otherwise uninvolved in the participant's treatment
Incomplete outcome data (attrition bias) All outcomes	Low risk	A small number of participants did not complete follow‐up assessments, but there were no significant differences between those who did and did not complete the follow‐up evaluation
Free of systematic differences in baseline characteristics of groups compared?	Low risk	No significant differences were found at baseline between the prism and the control groups with regard to the mean days from onset to intervention, mean hospital stay, MMSE score, and SIAS motor score
Did authors adjust for baseline differences in their analyses?	Low risk	No differences at baseline