Yeh 2019.

Study characteristics
Methods	Design: single‐blind, multisite RCT Duration of trial: 1 year (September 2016 to September 2017) Unit of randomisation: people with ≥ 6 months after ischaemic or haemorrhagic stroke with cognitive dysfunction stratified by MMSE scores (Strata 1: MMSE score: 19 to 24; Strata 2: MMSE score: 25–30) undergoing rehabilitation and matched for baseline characteristics Recruitment and allocation: of 46 people recruited from rehabilitation units, 30 were stratified by MMSE score then randomised to 1 of 2 groups: IG (n = 15) and active CG (n = 15)
Participants	Setting: rehabilitation units Country: Taiwan Sample size: 30 adults, 70% men; IQ: 15 (53% men); CG: 15 (87% men) Inclusion criteria: ischaemic or haemorrhagic stroke occurring ≥ 6 months before enrolment; MMSE score ≥ 19; MoCA score < 26; self‐reported or informant‐reported memory or cognitive complaints or score on the Clinical Dementia Rating scale 0.5; able to follow the study instruction; adequate cardiopulmonary function to perform aerobic exercise; and able to walk with or without assistive devices Exclusion criteria: unstable medical history that might limit participation concomitant with other neurological disorders, such as Parkinson disease, amyotrophic lateral sclerosis, and multiple sclerosis; current participation in another interventional trial Age: mean: IG: 50.63 (SEM 3.99) years; CG: 60.21 (SEM 3.10) years Time since onset of stroke: mean: IG: 47.80 (SEM 11.49) months; CG: 94.43 (SEM 30.80) months Types of stroke: not reported Site of lesion: not reported
Interventions	Intervention group Brief name: SEQ (sequential) Recipients: adults > 6 months poststroke with cognitive dysfunction Why: to provide a sequential combination of aerobic exercise and CT to "prepare the brain for the compensatory recruitment process in the cognitive training sessions that follow"; CT aimed to enhance cognitive functions, to facilitate several cognitive functions, including attention, recognition, colour and shape identification, calculation, visual perception, visuospatial processing, and executive function. What (materials): stationary bicycle; BrainHQ (Posit Science) computer‐based software and personal computer What (procedures): aerobic exercise followed by CT: aerobic exercise training: using a progressive resistance stationary bicycle, 3 minutes of warm‐up, 25 minutes of aerobic resistance exercise ending with 2 minutes of cool down; target heart rate 40–70% maximal heart rate (208 − 0.7 × age); vital signs and perceived effort (Borg Perceived Exertion Scale) monitored and recorded each session computer‐based CT used to facilitate cognitive functions including attention, recognition, colour and shape identification, calculation, visual perception, visuospatial processing, and executive function Who provided: not reported in the paper but confirmed by author contact that the intervention was delivered by certified occupational therapists. How: face‐to‐face individually monitored by staff Where: not specified but within rehabilitation unit When and how much: 60 minutes per session consisting of 30 minutes of exercise and 30 minutes of CT, 2 or 3 days per week for 12–18 weeks for a total of 36 sessions (36 hours in total) Tailoring: exercise intensity was progressed as participants improved their performance throughout the training; CT programme was adjusted automatically and continuously according to each participant's level of performance. Modifications: none reported How well (planned): not reported How well (actual): not reported except that the intervention was "feasible and safe, with low dropout rates". Comparator group Brief name: active control Recipients: adults > 6 months poststroke with cognitive dysfunction Why: to provide comparable active activities What: exercise training followed by mental activities: non‐aerobic exercise training including flexibility exercise, muscle strengthening, and balance training; unstructured mental activities that did not train a specific cognitive domain; participants chose to read newspapers or magazines or watch videos of topics derived from health‐related materials. After the unstructured mental activities, participants were asked several questions about the content that they just received. Who provided: not reported How: face‐to‐face Where: not specified except within rehabilitation unit When and how much: 60 minutes of training sessions, 2 or 3 days per week for 12–18 weeks for a total of 36 sessions (36 hours in total) Tailoring: not reported Modifications: not reported How well (planned): not reported How well (actual): not reported
Outcomes	Primary: none Secondary Community Integration Questionnaire MoCA Spatial Span test from Wechsler Memory Scale – Third Edition Verbal Paired Associates subtest Wechsler Memory Scale – Third Edition Other 6‐minute walk test International Physical Activity Questionnaires EQ‐5D Methods of data collection: evaluator blinded to the group allocation assessed participants before and after the intervention programmes. Data collection time points: within 1 week before and after intervention (12–18 weeks)
Notes	Funding: yes Conflict of interest: none Published trial protocol: no Trial registration: yes; NCT03045991 Ethics approval: yes Author contact: further information requested and provided
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation scheme was generated with the web‐based Research Randomizer randomisation tool.
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding of participants or personnel for intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The evaluator for the before and after assessments was blinded to the patient's group allocation" and "… the assessments were conducted by a trained and experienced therapist who was blinded to group allocation".
Incomplete outcome data (attrition bias) All outcomes	Low risk	Flow diagram indicated there were no dropouts.
Selective reporting (reporting bias)	Low risk	All relevant outcomes from method and protocol reported.
Other bias	Low risk	No other identifiable bias.