Fong 2007.

Methods	3‐arm RCT Setting: Hong Kong
Participants	60 participants Experimental 1: n = 20; experimental 2: n = 20; control: n = 20 Number lost to follow‐up: immediate post‐treatment (day 30) assessments on 19, 20 and 15 respectively Also lost 5, 0 and 3 respectively to follow‐up (day 60) so final analysis of 14, 20 and 12 Adequacy of matching at baseline?  yes: P values are reported for all demographics and baseline data ‐ there are no significant differences Mean age (mean (SD)): experimental 1 = 69.9 (11.0), experimental 2 = 69.9 (9.8), control = 73.8 (9.9) Sex (male/female): 34/20 Side of damage: all had right brain damage Time post‐onset: experimental 1 = 12.1 (9.4), experimental 2 = 11.6 (5.1), control = 12.1 (7.1) days Inclusion criteria: first or second unilateral right lesion stroke confirmed by imaging and examination, admitted to rehabilitation hospital, < 8 weeks since stroke onset, right‐handed, left visual inattention or neglect diagnosed by < 51/54 on Star Cancelled of BIT and GCS = 15 at recruitment Exclusion criteria: severe aphasia, significantly impaired visual acuity, hemianopia Visual sensory deficit: hemianopia and visual acuity assessed (method of assessment not stated)
Interventions	Experimental 1: voluntary trunk rotation 1 hour per day, 5 days per week for 30 days = 30 hours, OT present throughout Each hour composed of 15 minutes ADLs + 45 voluntary trunk rotation with set up equipment (supine, unsupported sitting and in standing frame) reaching with ipsilateral hand into contralateral space and therefore rotating upper body/trunk by 15 to 35 degrees from midline. Used set up apparatus (peg board or shoulder arc). Voluntary or if necessary therapist provided verbal or motor prompting for 15 minutes     Experimental 2: voluntary trunk rotation and half field eye‐patching Same amount and content as experimental group 1 but wearing half field eye patches to ipsilesional (right) hemifield wearing patches on plastic goggles (over own glasses if necessary) Control: same amount of time as experimental groups 1 and 2. Conventional OT for hemiplegia (15 minutes ADLs + 45 minutes training upper extremity). No mention of any neglect‐specific treatment implying treated as if had only hemiplegia For analysis of bottom‐up and top‐down rehabilitation approaches this review coded the experimental condition as bottom‐up For analysis voluntary trunk rotation and half field eye‐patching was classed as the experimental condition and control as the control condition Profession of outcome provider: OT
Outcomes	Used 3 (some with multiple subtests) at 2 follow‐up timepoints (day 30 immediately post‐therapy + day 60) Full BIT (15 subtests with 2 category scores and an overall score) Clock drawing task "using the Watson system" FIM motor subscale (4 subtests with 1 motor subscale score)   Did not measure serious adverse events, excluded anyone rehospitalised or with deteriorating health
Notes	"Recruitment hypothesis" target both spatial representation and motor intentional deficits of personal and peripersonal space ‐ this is the voluntary rotation plus eye patches "Inexpensive and easily integrated into use in day‐to‐day rehabilitation" Lack of intention‐to‐treat analysis, no baseline data on those allocated, baseline data on those followed up suggests pre‐therapy differences described in 'Risk of bias' table below
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	High risk	Personal communication with authors: "we didn't have concealment of allocation of participants from the person who was recruiting"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"Two independent blinded assessors, who were blinded to group membership, were responsible for all repeated measures throughout the duration of the study"
Incomplete outcome data (attrition bias) All outcomes	High risk	Although reasons for post‐randomisation exclusions are stated, it would have been preferable if all participants had been included in intention‐to‐treat analysis
Free of systematic differences in baseline characteristics of groups compared?	Low risk	P values reported for all demographics and baseline data ‐ no significant differences
Did authors adjust for baseline differences in their analyses?	Low risk	No adjustment was required as there were no baseline differences