Akerlund 2013.

Study characteristics
Methods	Design: parallel, single‐site RCT Duration of trial: 31 months (March 2008 to December 2010) Unit of randomisation: rehabilitation outpatients of working age (18–65 years) in the postacute phase after a brain injury with identified WM deficits. Recruitment and allocation: of 331 rehabilitation outpatients of working age in the postacute phase after a brain injury, 203 people were screened for WM deficits. 121 met the WM inclusion criteria; 24 of these were excluded due to communication problems or being unfit for testing. Of the remaining number who accepted participation, 45 participants were randomised by lot to either to IG (n = 25) or CG (n = 20). 34 of these participants had stroke, 19 in IG and 15 in CG.
Participants	Setting: metropolitan rehabilitation outpatient clinic Country: Sweden Sample size: 34 adults allocated (IG: 19; CG: 15), 29 analysed (gender of subsample with stroke not specified; original sample had 24 men (51%) and 23 women (49%)) Inclusion criteria: ≤ 5 digits/blocks forwards or ≤ 4 digits/blocks reversed in the WAIS‐III Digit Span and WAIS‐III‐NI Span board Exclusion criteria: inability to communicate, medical reasons according to the physician's health assessment (i.e. pronounced fatigue, pain, or depression) Age: mean: IG: 47.38 years; CG: 52.86 years Time since onset of stroke: mean: IG: 30.31 (SD 13.41) weeks; CG: 27.79 (SD 13.94) weeks Types of stroke: 66% of participants with stroke had a right hemisphere stroke, 41% had a left hemisphere stroke, and the rest bilateral.
Interventions	Intervention group Brief name: Cogmed QM computerised training program Recipients: rehabilitation outpatients of working age in the postacute phase after a brain injury with identified WM deficits Why: to provide an evidence‐based remedial approach using a computerised training program with intensive repetition of tasks for visuospatial and verbal WM that are adapted for the participant at a challenging performance level. What (materials): Cogmed WM training. Stockholm: Pearson Assessment AB. 2006. Available online at: www.cogmed.com/. Online version requires a stable broadband Internet connection, preferably 0.5 Mbit/second or better. What (procedures): the Cogmed training program in addition to usual rehabilitation routines. The training programme included a battery of visuospatial and verbal auditory WM tasks: visuospatial WM tasks require recall of the position of stimuli in a 4‐by‐4 grid and then reproduction of the stimuli in the same order, in the reverse order, or in a rotated grid; verbal WM tasks require recall of sequences of letters and digits forwards or backwards or both. All parts of the battery must be trained at each session, 90 trials each day. The tasks are introduced with a voice‐over transmitted by the computer's speaker. The person responds by localising and remembering multiple stimuli at the same time. The tasks have a unique sequencing order in each trial and short delays that require the representation of stimuli to be held in the person's WM. Who provided: an occupational therapist certified as coach for the Cogmed QM training, confirmed in personal communication with the authors. According to the Cogmed website, Cogmed is "currently used by psychologists, speech pathologists, occupational therapists and other clinical specialists working with individuals with attention and learning difficulties" and requires providers to undertake Cogmed Coach Training and Accreditation courses, either by self‐paced online coursework that is free with the purchase of a Cogmed Coach Starter Pack or attending a 1‐day face‐to‐face course offered around the world. How: the person can work individually and independently using the online software on a personal computer. Participants were able to ask staff for assistance and were provided with personal and individual feedback once per week, by a specially trained coach. Where: in a quiet room in the occupational therapy department. When and how much: 30–45 minutes per session, 5 days per week for 5 weeks (25 sessions) (12.5–18.75 hours in total) Tailoring: the difficulty level of the tasks adapts according to the individual's performance. The software included direct reinforcement via scores and positive verbal feedback. Participants were able to ask staff for assistance. Once per week the coach gave personal and individual feedback about results. Modification: none reported How well (planned): there appeared to be no formal assessment of fidelity. The program was reportedly used according to the guidelines and that the coaches provided input to help participants "adhere to the training" but no evaluation of this was reported. How well (actual): not reported. The researchers commented that the training was tiring and that the schedule of 5 days per week for 5 weeks and the need to attend the outpatient clinic for the training was challenging for many of the participants and was the reason for dropouts. Comparator group Brief name: usual rehabilitation Recipients: rehabilitation outpatients of working age in the postacute phase after a brain injury with identified WM deficits Why: provide usual rehabilitation services What: outpatient rehabilitation in accordance with the usual routines at the clinic, based on their rehabilitation needs Who: several different professionals in the team How: face‐to‐face but not further specified Where: rehabilitation clinic When and how much: 5 weeks Tailoring: rehabilitation services based on individual needs Modifications: not reported How well (planned): not reported How well (actual): not reported
Outcomes	Primary AMPS Secondary WAIS‐III Digit Span reversed RBMT‐II BNIS DEX WM Questionnaire Other HADS FIS Methods of data collection: demographic data were collected from the medical chart; the AMPS was administered by a clinical occupational therapist trained, calibrated, and certificated in the administration of the AMPS; other information not provided. Data collection time points: baseline, after the 5‐week intervention, and at 3 months' follow‐up
Notes	Funding: no Conflict of interest: none Published trial protocol: no Trial registration: no Ethics approval: no Author contact: further information and stroke‐specific data from authors requested and provided.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were "randomized by lot".
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	The IG received Cogmed training from trained occupational therapist coaches and the CG received regular occupational therapy and rehabilitation services but with computerised CT excluded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Not reported and there was no ethical process of informed consent; however, the participants completed the outcome measures and knowledge of receiving the CT or not could have impacted outcome measurement.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	For the whole sample, there was 15% attrition after training and 19% after 18 weeks with missing outcome data balanced in numbers across groups. However, reasons for attrition were not provided separately for both groups. About half of the reasons appeared unrelated to outcome, and the other half were unclear. We used data provided by authors for the subsample of stroke participants for which attrition was unclear.
Selective reporting (reporting bias)	Low risk	No protocol located but all outcomes reported across 2 papers.
Other bias	Low risk	No other identifiable bias.