Barker‐Collo 2009.

Study characteristics
Methods	Design: parallel RCT Duration of trial: 18 months (dates not specified) Unit of randomisation: inpatient adult survivors of incident stroke with a confirmed attention deficit Recruitment and allocation: 334 survivors of stroke (all pathological subtypes) were approached within 2 weeks after stroke. 107 (32%) consented and were checked for eligibility. Of these 95 (88.8%) were eligible for attention screening, 84 of these (88%) were assessed as having deficits and 78 of these (5 withdrew and 1 moved) were randomised by stratified minimisation (by age, gender, ethnicity (European, non‐European) and Barthel Index (≤ 18, > 18) to 2 groups, APT (n = 38) and SC (n = 40).
Participants	Setting: 2 metropolitan hospitals Country: New Zealand Sample size: 78 adults, 60% men; APT: 38; SC: 40 Inclusion criteria: 1 SD below the normative mean on any of the following tests: Bells Test, IVA‐CPT, TMT‐A and TMT‐B, and 2 slowest Paced Auditory Serial Addition Test trials; newly diagnosed stroke (first‐ever‐in‐a‐lifetime stroke) Exclusion criteria: inability to give informed consent, severe cognitive deficit precluding participation (MMSE < 20), medical instability or condition that could impact results (e.g. dementia), not fluent in English as required for standardised assessment Age: mean: APT: 70.2 (SD 15.6) years; SC: 67.7 (SD 15.6) years Time since onset of stroke: mean: APT: 18.48 (SD 11.95) days; SC: 18.58 (SD 7.62) days Types of stroke: 61.5% had had an ischaemic stroke Site of lesion: APT: left hemisphere 14 (43.8%), right 15 (46.9%), other 3 (9.1%); SC: left hemisphere 25 (58.1%), right 17 (39.5%), other 1 (2.3%)
Interventions	Intervention group Brief name: APT Recipients: inpatient adult survivors of incident stroke with a confirmed attention deficit Why: to provide a theoretically based, hierarchical, and multilevel treatment, involving cognitive exercises for remediation and improvement of aspects of attention including sustained, selective, alternating, and divided attention (Sohlberg 1987; Sohlberg 2001). Sohlberg 2001 outlined 6 treatment principles for attention process training, including theoretically grounded, hierarchically organised, providing sufficient repetition, based on client performance data, with active facilitation of generalisation throughout treatment, and providing a flexible and adaptable format. What (materials): APT package including paper and pencil tasks, set of CDs including auditory CDs that produced auditory stimuli as well as a distraction (like 'white noise') to overlay some tasks where selective attention was needed. For the visual tasks, this distraction was produced by using acetate overlays with patterns on them (Barker‐Collo 2009). The latest version of APT program can be purchased at: lapublishing.com/apt-attention-process-training/, which provides details of latest APT software and training programme; materials include a manual and tracking sheets for exercises. See also Sohlberg 1987 for an appendix of materials used for training each aspect of attention and Sohlberg 2001 for an outline of an APT programme that included computer activities, auditory tapes, and pen‐and‐paper tasks. What (procedures): provider used a hierarchy of treatment tasks targeting different aspects of attention starting at sustained attention then progressing to selective, alternating, and divided attention. Each task was considered "mastered" once the client was able to complete the task with no more than 1 error (Barker‐Collo 2009). Current APT programme involves "a set of drill based, hierarchically organized exercises that tap different domains of attention that are matched to the client's impairment profile and administered repetitively. They are paired with generalization real world, individualized exercises that are selected to promote generalization" (Barker‐Collo 2009). Sohlberg 2001 has examples of tasks in the APT computer program for addressing each of these aspects and appendices with example recording protocols and a case study. Examples of tasks include listening for target words or sequences on auditory attention tapes, mental math activities, putting words in alphabetical order, placing a visual distractor (e.g. a plastic overhead sheet with distractor lines) over the top of a paper‐and‐pencil activity. Who provided: a registered clinical neuropsychologist provided APT training in this study; APT could be administered by neuropsychologists, occupational therapists, speech language therapists, and other rehabilitation specialists. There were no training requirements in addition to the manual. How: face‐to‐face individual sessions. Sohlberg 2001 reported that delivery should be flexible and adaptable including delivery to individuals or groups. Where: in hospital prior to discharge and then in clients' primary residences after discharge; Sohlberg 2001 reported that delivery should be flexible and adaptable including delivery in clinics or at home. When and how much: up to 30 hours of individual APT conducted for 1 hour on weekdays for 4 weeks (mean 13.5 (SD 9.44) hours) Tailoring: because of issues such as fatigue, a 30‐hour maximum was set in this study. Sohlberg 2001 described the hierarchical nature of the programme tasks and how clinicians use client performance data to tailor the intervention. The hierarchy of programme tasks place increasing demands on attentional control and WM. Client performance data were used to make treatment decisions, such as when to start, stop, or modify a programme. For example, clinicians examine the participants' error profiles to assess where errors were occurring, such as at the beginning or end of a task, reflecting a different attentional demand and adjust the training tasks accordingly. Graphs of performance were shown to the client to provide objective and powerful feedback. Modification: none reported How well (planned): none reported in the paper and not formally assessed; the author advised that "there was tracking of the order in which tasks were administered to ensure that the protocol was adhered to". How well (actual): none reported Comparator group Brief name: SC Recipients: inpatient adult survivors of incident stroke with a confirmed attention deficit Why: to provide standard rehabilitation care poststroke. What: not reported Who: not reported How: not reported except standard inpatient care Where: metropolitan hospital When and how much: not reported Tailoring: none reported Modifications: not reported How well (planned): not reported How well (actual): not reported
Outcomes	Primary mRS Secondary IVA‐CPT FSAQ, which combines auditory and visual attention scores TMT‐A; TMT‐B PASAT CFQ Other Bells Test SF‐36 MCS SF‐36 PCS GHQ‐28 Methods of data collection: a trained assessor blind to randomisation administered and scored the assessments using standard procedures. Assessments lasted 2.5 hours, occurring over 2 sessions if required. Data collection time points: baseline, 5 weeks (1‐week postintervention), and 6 months
Notes	Funding: yes (supported by the New Zealand Health Research Council (HRC Refs 06/063C and 07/070C) and a National Heart Foundation (New Zealand) Fellowship Conflict of interest: none Published trial protocol: not reported Trial registration: yes (ACTRN12607000045415) Ethics approval: yes Author contact: further details about intervention implementation were provided and unpublished data (means and SDs) supplied.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization was concealed using an online internet randomization service whose procedures ensure enrollment and check eligibility before allowing randomization. Stratified minimization randomization was used to ensure the balance for possible prognostic factors … across the groups".
Allocation concealment (selection bias)	Low risk	Quote: "Implementation of the randomization sequence was via secured online contacting of the treating clinician, who had no access to assessment data. Randomization information was not accessible to any other study staff during the study".
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and therapists not blinded. Quote: "All APT sessions were administered by a registered clinical neuropsychologist, who was the only member of the study team (e.g. named investigators, statisticians, data management, assessors) who did not remain blind to randomization status throughout the study".
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Assessments were repeated at 5 weeks and 6 months by a trained assessor blind to randomization … Randomization information was not accessible to any other study staff during the study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Attrition overall provided, not numbers lost for each group. Reasons for losses not provided separately for both groups. Used last observation carried forward for missing data.
Selective reporting (reporting bias)	Low risk	All outcomes included in this review were reported as per the protocol.
Other bias	Low risk	No other identifiable bias.