Kim 2018.

Study characteristics
Methods	RCT, 3‐group parallel Setting: Korea, 'patients receiving rehabilitation'
Participants	Patients with left hemi‐spatial neglect due to first stroke, right hemisphere stroke, no cognitive decline, muscle strength > 2, no visual defect, sitting balance to perform intervention Age, years: intervention 1 = 70.3 ± 9.6, intervention 2 = 66.6 ± 12.2, intervention 3 = 62.5 ± 16.5 Men/women: intervention 1 = 5/5, intervention 2 = 5/5, intervention 3 = 5/5 Time post stroke, days: intervention 1 = 19.2 ± 13.4, intervention 2 = 24.5 ± 22.4, intervention 3 = 15.3 ± 9.8
Interventions	Intervention 1: rTMS therapy with a coil stimulator shaped like a figure 8 at diameter of 70 mm using MagPro (MagVenture Inc., Farum, Denmark). Session included 900 stimuli applied over contralesional posterior parietal cortex at an intensity of 95% motor thresholds and a frequency of 0.9 Hz Intervention 2: patients in the robot group received additional treatment for hemi‐spatial neglect using a rehabilitation robot (Neuro‐X; Apsun Inc., Seoul, Korea) for upper limbs. During robot therapy, patients sat on the right side of the robot with the robot’s monitor on their left side. Robot therapy programme was conducted through games that induced passive and active assistive range of motion of the wrist, elbow, and shoulder joints. These games consisted of 2 isometric exercises and 2 range of motion exercises. The 2 isometric exercises used wrist extension and wrist flexion, in which the default muscle strength for wrist extension and wrist flexion were measured quantitatively before the start of the game, so the game was continued only when a force exceeding a certain level of strength was applied Intervention 3: rTMS + upper limb robot
Outcomes	Motor‐Free Visual Perception Test (MVPT‐3) Line bisection test Star cancellation test Albert’s test Catherine Bergego Scale (CBS) Mini‐Mental State Examination (MMSE) Korean version of the Modified Barthel Index (K‐MBI) Measured before and after intervention
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation table
Allocation concealment (selection bias)	High risk	No mention of concealment, but blocks of size 3 highly predictable
Blinding of participants	High risk	No sham for either intervention
Blinding of personnel	High risk	No sham for either intervention
Blinding of outcome assessment (detection bias) All outcomes	High risk	“Occupational therapist who did not directly participate in the treatment”, but patient clearly aware
Incomplete outcome data (attrition bias) All outcomes	High risk	Early discharges excluded from analyses
Selective reporting (reporting bias)	Low risk	Outcomes reported in equal detail
Other bias	Low risk	Nothing obvious