Constantinidou 2008.

Methods	Design: prospective randomized controlled trial. Randomization: 2‐arm, parallel group, multi‐centre trial. Duration of study: 2004‐2008.
Participants	Number randomized: 49 people undergoing rehabilitation following TBI. 29 assigned to intervention group, 20 to control group. Inclusion criteria: aged 18‐55 years; moderate‐to‐severe closed head injury; Ranchos Los Amigos scale score ≥ 6; no aphasia; resolved post‐traumatic amnesia; enrolment in a residential postacute rehabilitation programme; participants within 4 years of brain injury. Exclusion criteria: penetrating head injuries; diagnosis of stroke; premorbid central nervous system disorder or learning disability; premorbid psychiatric disorder; active alcohol abuse; deficits in auditory comprehension; English as second language; colour blind; diagnosis of depression.
Interventions	Intervention: categorization programme Intervention consisted of 2 types of tasks: object categorization tasks consisted of 5 different levels. Tasks began with teaching perceptual features to describe objects or living things and move to higher levels of cognition including analyses, synthesis, linguistic flexibility and abstract reasoning; new category learning tasks consisted of 3 levels. Under each level, there were 5 steps that increasingly demanded a higher level of rule‐governed responses. Errorless learning principles and cueing hierarchies were applied under each step. Duration: mean of 13 weeks to complete categorization programme. Participants received approximately 57 hours of individual cognitive treatment, averaging 2‐3 hours per week on the categorization programme‐related tasks, for a total of 27 hours of categorization programme treatment and about 4.5 hours of total individual therapy per week. Control: standard rehabilitation programme at each rehabilitation centre. retraining therapy programmes to improve attention, memory and problem solving and also integrated functional skills such as time and money management and psychosocial training as part of their treatment regimens. Duration: mean 80 hours of individual cognitive treatment over an 18‐week period, averaging 4.5 hours of individual therapy per week.
Outcomes	Community Integration Questionnaire (CIQ) along with the following cognitive assessment tools: Wechsler Abbreviated Scale of Intelligence, Scales of Cognitive Ability for Traumatic Brain Injury, Rey Complex Figure Test, Trail Making Tests, Wechsler Memory Scale III, California Verbal Learning Test II, Wisconsin Card Sorting Test, The Booklet Category Test, Symbol Digits Modalities Test, Control Oral Word Association, subsets from Woodcock‐Johnson III, Mayo‐Portland Adaptability Inventory III (MPAI‐3).
Notes	Setting: 5 residential brain injury rehabilitation centres. Country: US. Duration of follow‐up: none. Dropouts: intervention group: 2 discontinued rehabilitation, 2 developed complications, 5 discharged due to insurance‐related issues. Control group: 6 discharged due to insurance‐related issues. Funding: grants from the National Institute of Child Health and Human Development, National Institutes of Health, and the Center for NeuroSkills, Bakersfield, CA.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomly assigned by project investigators who were off location and did not have direct contact with participants." Comment: method of random sequence generation not reported. Author could not provide specific details to clarify this.
Allocation concealment (selection bias)	Low risk	Quote: "Randomly assigned by project investigators who were off location and did not have direct contact with participants." Comment: allocation concealment was adequate since it was performed off‐location.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: not blinded, self‐reported outcomes are likely to be influenced by the knowledge of allocation.
Blinding of outcome assessment (detection bias) Objective outcomes	Low risk	Quote: "The functional outcome measures in most cases were conducted by the case management staff who was not involved in patient training and, therefore, was not informed of the participant's group assignment."
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "Data from patients unable to complete the assigned treatment regimen were included in the analyses to the fullest extent possible. If partial data were useful for certain analyses, then those data were analysed. Therefore, the intention‐to‐treat principle was followed." Comment: we rated this as high risk of bias because there were 15 dropouts (31%).
Selective reporting (reporting bias)	Low risk	Comment: published report contains all expected outcomes.
Other bias	Low risk	Comment: no additional biases detected.