Goal Management Training (GMT) for Improvement of Cognitive Control Function After Acquired Brain Injury

Overview

The goal of this clinical trial is to investigate predictors of treatment outcome, and the effect of individual treatment components of Goal Management Training (GMT) for improvement of cognitive control function in people with acquired brain injury (ABI). Primary aim: To identify demographic, clinical and cognitive predictors of treatment response in Goal Management Training after acquired brain injury (ABI)? Secondary aims: To investigate the effects of a) extended cuing (via a smartphone) and b) a booster module? * All included participants will receive Goal Management Training in groups of 4-6 patients, as implemented in a rehabilitation hospital setting (St. Olavs Hospital, Trondheim University Hospital). The standard treatment ("GMT Usual") consists of 10 sessions, delivered as 2 sessions a day, one day per week, over 5 weeks. * All participants will be asked to complete self-report measures and performance-based cognitive testing at baseline (T1), immediately after the main treatment period (T2), at 6 months (T3), and 1 year (T4) after treatment. * After baseline assessment, 50% of participants will be randomized to receive extended cuing through a smartphone application ("GMT Cuing") - intended to facilitate the effect of between-session tasks (homework) completed by the participants. These participants will, in addition to the standard treatment, on a daily basis receive a message that says "STOP" as a reminder to do their home assignments. * After completion of the 10 GMT sessions and the first post-treatment assessment immediately after the main treatment period, 50% of the participants will be randomized to receive an additional booster module ("GMT Boost") 3 months after the last ordinary GMT module - intended to facilitate a prolonged treatment response. The remaining 50% will receive no booster module ("GMT No Boost") * Randomization will be carried out on treatment group-level (all patients in the same group receive the same treatment). The total anticipated sample size is N = 116 patients. * The Global Executive Composite (GEC) score derived from BRIEF-A will be used as the primary outcome measure. A selection of other included measures will be used as secondary outcome measures.

The Global Executive Composite (GEC) score derived from BRIEF-A will be used as the primary outcome measure. A selection of other included measures will be used as secondary outcome measures. Data will be analyzed based on an intention-to-treat approach. Penalized linear regression by the elastic net approach (a combination of the Lasso and Ridge regression approaches) will be used to identify demographic, clinical and cognitive predictors of outcome at 6 months after treatment (T3), which is the primary aim of the study. For the secondary aim of investigating the differences in outcome for primary and secondary outcomes between "GMT Cuing" and "GMT Usual", and between "GMT Boost" and "GMT No Boost", linear mixed models (LMMs) will be used. Data for all time points will be included, but of primary interest are differences at T2 (immediately after treatment) for assessing the effect of cuing, and at T3 (6 months after treatment) for the effect of boosting. The LMMs can account for within-subject correlations due to repeated measurements. In addition, the investigators will perform exploratory moderation and mediation analyses across both treatment groups. For the penalized regression models, complete case analyses will be performed as long as the number of missing observations is small. Otherwise, imputation will be considered, but imputation is not straightforward for variable selection models. Linear mixed models can handle missing data for the outcome variable. Considering multiple testing linked to several secondary outcomes, p-values will be interpreted with care rather than using a formal p-value adjustment. Results will be interpreted according to the magnitude of the group difference (effect size) as well as the p-values. Data will be analyzed using IBM SPSS, STATA and R.