Strasser 2008.

Study characteristics
Methods	Aim of study: to determine the effectiveness of a team training intervention in improving patient outcomes Study design: cluster randomised trial Unit of randomisation: rehabilitation unit Mean cluster size: information not available Unit of analysis: patient Sample size calculation: study was designed with 80% power to detect a difference of 6 points in primary outcome measure (motor FIM) between intervention and control groups. ICC in motor FIM score gain was estimated to be 4.6%, (SD 12.8%) using data from 45 rehabilitation sites participating in a previous observational study by author. Assuming 20 patients per site, it was anticipated 28 sites would be required.
Participants	Healthcare professionals: 31 rehabilitation teams, comprising 464 staff of 6 disciplines; medicine, nursing, occupational therapy, speech‐language pathology, physiotherapy and social work/case management. Demographics of staff within all clusters not provided. Patients: stroke survivors (n = 1368) < 90 days' poststroke Preintervention Experimental group: n = 350, 98.3% men, mean age 65.9 (SD 11.4) years Control group: n = 439, 96.8% men, mean age 66.6 (SD 12.0) years Postintervention Experimental group: n = 233, 96.6% men, mean age 67.6 (SD 11.1) years Control group: n = 346, 97.4% men, mean age 66.9 (SD 11.9) years Ethnicity: Preintervention Experimental group: white 57.7%, black 30.9%, Hispanic 8.9%, Asian 0.6%, Native American 0.6%, other/unknown 1.5% Control group: white 55.6%; black 24.6%; Hispanic 17.1%; Asian 1.1%; Native American 1.1%, other/unknown 0.4% Postintervention Experimental group: white 62.7%; black 22.7%; Hispanic 9.4%; Asian 0.9%; Native American 0.9%; other/unknown 3.4% Control group: white 53.8%; black 24.6%; Hispanic 19.9%; Asian 0.6%; Native American 0.3%; other/unknown 0.9% Setting: inpatient (acute or subacute) Country: USA
Interventions	Targeted behaviour change: for multidisciplinary staff to function more effectively as a team Theory used: Lichstein's treatment implementation model Category of implementation intervention: delivery arrangement intervention/implementation strategies Multifacetedintervention: yes Tailoring to barriers: yes Implementation action plans were modified according to perceived barriers by team leaders Description of intervention: interprofessional education, local opinion leaders, audit and feedback (see Appendix 2 for full details of intervention) Delivery: off‐site, face‐to‐face workshops conducted by research staff Frequency: 1 × 2.5‐day workshop (16 hours), written feedback 3–5 weeks after workshop, consultation 2–3 months after written feedback Duration of intervention: 6 months Control: received summary of performance on process measures taken pre‐ and postintervention
Outcomes	Main outcome: Patient outcomes: functional improvement measured by change in the motor items of Functional Independence Measure (FIM) Quality of care outcomes: not reported Secondary outcome: patient length of stay and percentage discharged home from inpatient rehabilitation Follow‐up: 12‐months after completion of the intervention Loss of clusters and individuals: 4/31 clusters lost due to incomplete ethics process (1) and unreported data (3) Adjusted for clustering for each outcome: yes Method of cluster adjustment for each outcome: cluster‐adjusted Chi2 t‐tests used in analysis ICC reported for each outcome: no
Notes	Outcomes used in this review: Functional improvement measured by change in the motor items of FIM Percentage of stroke patients discharged home from inpatient rehabilitation Patient length of stay Unit of analysis error: no Ethical approval and informed consent obtained (yes/no): site ethics obtained from local Veterans Affairs research committees and institutional review boards. Informed consent not reported. Funding source: supported by Veterans Administration Rehabilitation Research and Development Service Declarations of interest: no conflict of interest reported. Some evidence cited to support implementation research was author's own work. Contact with author? yes Additional outcome data provided from author? no Trial registration: NCT00237757
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'Computer' used to randomise sites to intervention or control groups (process not described) Clusters were stratified into 4 strata according to volume of patients and FIM scores. Each stratum 'force randomised' to have 4 sites in 1 arm.
Allocation concealment (selection bias)	Low risk	Concealment of allocation prior to assignment not described, computer used for randomisation.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Information on blinding not available. Staff likely unblinded due to participating or not participating in training. Patients likely blinded to group allocation as data taken from an outcomes database (Veterans Affairs Functional Status Outcomes Database) and was information collected in usual practice.
Blinding of outcome assessment (detection bias) All outcomes	High risk	FIM scores assessed by staff who were likely unblinded to group allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	4 clusters lost overall following randomisation (3 in control, 1 in intervention). 2/16 control sites did not report postintervention data, only patients with complete data (FIM) were eligible for inclusion so no patient loss to follow‐up. The proportion of missing data was less than the effect size and therefore unlikely to overturn the study result.
Selective reporting (reporting bias)	Unclear risk	No published protocol, study registered retrospectively on ClinicalTrials.gov. Secondary outcomes not mentioned in registered project details.
Other bias	Unclear risk	Recruitment bias: low risk Incorrect analysis: unclear whether clustering accounted for in analysis. Similar baseline characteristics: unclear risk, comparable characteristics of sites from available information. Details of sites and composition of teams not provided. Similar baseline outcome measures: unclear risk, control sites had patients with a lower initial motor FIM score (mean: 43.5 control group vs 48.2 intervention group) at baseline. Reliability of primary outcome measures: low risk. FIM has demonstrated reliability and validity, though could be questioned as an adequate measure of team effectiveness. Adequate protection against contamination: low risk, allocation was by rehabilitation unit at separate hospitals.