Cottam 1987.

Study characteristics
Methods	2‐group parallel RCT Setting: rehabilitation centre, USA
Participants	12 stroke rehabilitation inpatients with left hemi‐spatial neglect Experimental: n = 6, control = 6 Mean age, years: experimental = 66.2, control = 71.3 Sex (men/women): 7/5 Side of damage: all had right middle cerebral artery lesions Time post onset, mean, weeks: experimental = 6, control = 16.3 Inclusion: right‐handedness; visual acuity > 20/100 corrected on Snellen's; oriented in person, place, and time; evidence of left hemi‐spatial neglect on ≥ 3 of the tests used; either WAIS‐R VIQ > 80 or minimum scaled score = 8 on 4/6 verbal subtests; arm and leg able to propel wheelchair
Interventions	3‐phase intervention, each phase consisting of 5 half‐hour sessions per day Visually scanning a light board when stationary, taught to verbally self‐prompt to start on left and scan from left to right Same activity but while self‐propelling Did not use the light board, but participants named objects presented on both sides while self‐propelling vs no information other than participants were inpatients at a rehabilitation facility and were assessed after same periods as experimental group
Outcomes	Study collected 3 types of outcomes: Data scanning and attention skills: single target cancellation (3 minutes letter H) and double target cancellation (3 minutes letters C and E); scores are for average number of far left‐sided omissions Light board: point at light and say the colour, allowing 10 seconds; scores are for average number of left‐sided omissions ADL: avoidance of obstacles on wheelchair course, as rated by 2 observers Assessed pre‐intervention, after each phase (5 days), and at follow‐up 6 weeks post discharge from hospital This review used only cancellation data, immediate and persisting effects
Notes	Single‐letter cancellation outcome data are entered as left‐sided omissions (i.e. low score is better outcome)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	No detail given; exactly 6 per group
Allocation concealment (selection bias)	High risk	No detail given; exactly 6 per group
Blinding of participants	High risk	Blinding not possible
Blinding of personnel	High risk	Blinding not possible
Blinding of outcome assessment (detection bias) All outcomes	High risk	Not stated. Not mentioned so unlikely to be blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	1 control lost to follow‐up
Selective reporting (reporting bias)	High risk	Comprehensive dissertation report with massive multiplicity. Apparently post‐hoc combinations of quintiles (far left, left, centre, right, far right) of areas on visual tasks
Other bias	Low risk	Nothing obvious