Early Training of Attention After Acquired Brain Injury

Overview

Background: To study intensive targeted cognitive rehabilitation of attention in the acute (\<4 months) and subacute rehabilitation phases (4 -12 months) after acquired brain injury and to evaluate the effects on function, activity and participation (return to work). Design: prospective, randomised, controlled study. Methods: 120 consecutive patients with stroke or traumatic brain injury are randomised to 20 hours of intensive attention training by Attention Process Training or by standard, activity based training. Progress is evaluated by Statistical Process Control and by pre and post measurement of functional and activity levels. Return to work is also evaluated in the post-acute phase. Primary endpoints are the changes in the attention measure, Paced Auditory Serial Addition Test and changes in work ability. Secondary endpoints include measurement of cognitive functions, activity and work return. There are 3, 6 and 12-month follow ups focussing on health economics. Discussion: The study will provide information on rehabilitation of attention in the early phases after acquired brain injury (ABI); effects on function, activity and return to work. Further, the application of Statistical Process Control might enable closer investigation of the cognitive changes after acquired brain injury and demonstrate the usefulness of process measures in rehabilitation.

STUDY METHODS The primary objectives of the study were: 1. In the acute stage, to evaluate the effectiveness of APT measured by the changes in performance in the attention measure, Paced Auditory Serial Addition Test, (PASAT) evaluated by Statistical Process Control 2. In the postacute stage, to evaluate and compare effects of Attention process training (APT) vs. standard rehabilitation on daily activities and return to work three months and one year after participating in a rehabilitation programme with regard to resource utilisation and health economics. STUDY DESIGN AND SETTING The study was conducted in a specialised rehabilitation clinic. Several in- and outpatient units on two sites were involved. It was a randomised, controlled (www.consort-statement.org/?o=1011) study with an open extension and a follow up at three and six months and one year. Due to the nature of the rehabilitation procedures, neither patients nor rehabilitation professionals were blinded as to the nature of the intervention, but different professionals conducted assessments and training. The study was designed to reflect normal clinical practice while allowing comparison between two rehabilitation approaches. Patients in the acute phase (\< 4 months) were recruited from both in- and outpatient units. For those, ready to be discharged from inpatient wards, the training was continued in the outpatient setting. Subacute patients (4-12 months) participated in the study as part of their outpatient rehabilitation. The study was approved by the regional ethics committee, Karolinska Institutet, Stockholm, Sweden. The participants comprised a consecutive series of patients with mild to moderate stroke or traumatic brain injury. Inclusion criteria were deficits in attention defined by the APT test, scores on the lower average and above for reasoning skills and abstract thinking, age range; 18-60 years and a good understanding of the Swedish language. Exclusion criteria were moderate to severe aphasia, ongoing psychiatric illness, a history of severe somatic disorder causing anoxic periods, ongoing substance abuse and severe pain. Patients with Hospital Anxiety and Depression Scale scores =\>10 were offered antidepressant treatment and were included in the study three weeks after pharmacological treatment had been initiated. Further, exclusion was also based on severe memory disorders, neglect, visual field defects and motor disability interfering with participation. All patients participated in a comprehensive interdisciplinary brain injury rehabilitation programme. After baseline assessment, the participants received 20 hours of attention training, at least three days a week, for a period of 5-6 weeks. They were randomly assigned to one of the two intervention programmes: intensive area-specific cognitive training with the APT or activity-based attention training provided by the occupational therapist. Evaluation of treatment effects was monitored by pre and post assessment and by repeated assessment by the primary outcome measure at baseline, after every third hour of intervention and post intervention for patients in the acute stage. Subacute patients were assessed by pre and post measurements. The open-label extension of the study comprised of client-centred standard interdisciplinary rehabilitation. Discharge was based on clinical decisions and the participants returned to the clinic for follow-up assessments. The APT programme provides a theoretically based, individualised, highly structured intervention of organised assignments at four attention levels: sustained, selective, divided and alternating attention. Progress is based on the intensity of training, on continuous feedback promoting motivation and on metacognitive training. The APT also includes education in acquired attention deficits and training for generalisation, enabling the transfer of treatment tasks and techniques to self-selected cognitive problems in everyday situations. Each session took 45-90 minutes to perform, working with material from APTs I and II. Attention training in activities included standard occupational training within an interdisciplinary rehabilitation programme. The programme consisted of a) training and the use of compensatory strategies in attention-demanding activities of daily living b) performing independent work with attention-demanding tasks at individual level c) training using computerised tasks, not specifically designed for attention remediation and d) group activities. Types of training and time devoted to a specific training procedure were individually registered. Inclusion measures were the APT-test and Matrix reasoning.Cut-off scores were 70% or less on at least two of the five subtests on the APT-test and standard scores of seven and above on Matrix Reasoning. Exclusion measures were Barthel Activity of Daily Living, (ADL)-index 50 or less; Alberts test/Line crossing with a cut off score of (\<=2); and a profile score of seven or less for The Rivermead Behavioural Memory Test. Primary outcome measures: For the acute phase: The Paced Auditory Serial Addition test, (PASAT) is presumed to measure working memory speed of information processing and sustained and divided attention. Versions A and C were administered. For the subacute phase the primary outcome measurement was the Work Ability Index (WAI). Secondary outcome measures were functional, activity and participation measures. Further measures of attention were the Digit Span task, the Block Span, the Ruff 2\&7 Selective Attention Test ,and the Letter-Number Sequencing task. For executive measures the Trail Making Test, the Color-Word Interference Test and the Tower Test were used. The latter is administered only once due to learning effects, thus allowing only group comparisons. The Rey Auditory Verbal Learning Test (RAVLT) has been used to evaluate different aspects of memory function. Depression and anxiety were assessed by the Hospital Anxiety and Depression Scale (HADS). Activity measures were the Cognitive Failure Questionnaire, the Rating Scale of Attentional Behavior and the Dysexecutive symptom questionnaire. Occupational Performance was measured by the Canadian Occupational Performance Measure (COPM). Participation and work-return measures were the Work Ability Index, used to assess self-rated work ability. The Assessment of Work Performance used to assess the skills of clients with various work-related problems during their work performance and the Work Ability Screening Questionnaire constructed by the authors including questions about education, current profession, work situation, current work ability, and approach to future work ability and return to work questions. Adverse effects were registered in field notes and standard hospital journals. Examples of adverse effects were: fatigue preventing participation in APT training, negative effects of the APT treatment, such as consequences of fatigue on following treatments, sudden emergence of exclusion criteria, etc. To assess resource utilisation and economic evaluation data for work capacity/sick leave before, and 12 months after ABI were obtained from the Registry for the Swedish Social Insurance Agency. For data regarding utilisation of health services related to ABI and medical and physical interventions, the participants were contacted by phone 12 months after injury. Resource utilisation was defined as the use of health care and social services associated with ABI. Costs of resources were calculated from a societal perspective, i.e. costs within the health-care system, transport, and caregiver's time and for sick leave. Patients' and caregivers' lost productivity was calculated based on productivity data from before ABI. The sample size calculation for the primary endpoint was based upon the estimate of 1 standard deviation (SD) improvement after 20 hours of attention training with APT or activity-based attention training. Setting an alpha at 0.05, with a power of 85%, a sample size of 19 completed data sets was needed to detect a statistically significant difference between treatment arms. Assuming a dropout rate of 25% requires the inclusion of 25 patients. Sample size calculation for secondary endpoint goals on functional level was based upon the assumption of a clinically relevant change of 1 SD in the Ruff 2\&7 test. For this effect size, a sample size of 30 complete data sets in each treatment arm, and alpha 0,0050, 2-tailed, yields a power of 0.888. Sample size calculation on activity level was based upon the assumption of a clinically relevant (2-point) change in the COPM performance measure. For this effect size, a sample size of 30 complete data sets in each treatment arm, and alpha 0,0050, 2-tailed, yields a power of 0.988. Thus the maximum number of patients to be enrolled was 120. The primary outcome measure was analysed by using statistical process control (SPC). Sigma Zone SPC, was used to explore statistical control limits and variability in improvement assuming that data plots appearing within the control limit indicate a process in stable statistical control and variations are due to chance variations, day-to-day variability in behaviour etc. Data were presented on control charts, including three additional lines; the centre line (usually based on the mean) and an upper (UCL) and lower control limit (LCL) set at ±3 standard deviations from the mean respectively). Data points outside those control limits are considered to be related to special causes of variation, such as effects of treatment. The Minimal Clinically Important Difference (MCID) in the primary outcome variable, PASAT was estimated and expressed as the minimum change of the PASAT-diff score that could be considered clinically relevant. Parametric methods, Student's t-test and Pearson correlation were used for normally distributed variables on interval level. When comparing the two treatment groups, t-tests for independent samples were used and for comparison between pre- post and follow up measures t-tests for dependent samples and analysis of variance were used. Skewed or ordinal data were analysed by non-parametric methods: Mann-Whitney U-test for comparison between the treatment groups; Wilcoxon matched pairs test, Kruskal-Wallis analysis of variance for comparison of pre- post and follow up measures and Spearman's rank correlation test. For post-hoc comparisons we used the Mann-Whitney U-test. Fisher's exact test was used for comparison between dichotomized variables. Bonferroni corrections were applied to correct for false positives due to the number of analyses. The possible effects of randomisation bias were investigated using a linear mixed-model analysis including one within-group factor treatment and one between-group factor. Cluster analysis was used for the primary outcome measure to explore and identify patterns of cognitive recovery in the acute phase. Two-tailed p-values were used with a critical significance level of 0.05. Health economics resource utilisation was calculated for each treatment arm for participants in both acute and subacute phases.