Programme Grants for Applied Research (Jul 2024)
Cognitive remediation therapy to enhance cognition and improve recovery in early psychosis: the ECLIPSE research programme including an RCT
Abstract
Background Despite the effectiveness of cognitive remediation, it is not widely implemented because we do not know whether teams will accept it, how much therapist time is needed, whether there are factors which predict lower benefits, whether it is cost-effective and what is required for large-scale roll-out. Objective To understand the factors that will enhance implementation and benefits of cognitive remediation in Early Intervention Services. Design Four work packages: (1) focus groups and interviews exploring the development of satisfaction and preference measures for staff and service users; participant team interviews to collect data, before and after introducing cognitive remediation, to understand team dynamics; (2) an observational study of a newly developed therapist e-training programme; (3) a multiarm multistage four-arm randomised controlled trial comparing different amounts of therapist input with Treatment as Usual; and (4) an analysis of trial data to understand potential mediating and moderating factors that affect treatment benefits. Setting Early Intervention Services in the United Kingdom National Health Service. Participants Staff and service users in touch with Early Intervention Services. Interventions For the e-training, we piloted and then provided an e-learning system for training cognitive remediation therapists. For the randomised trial, we provided a cognitive remediation software programme (CIRCuiTS™,King’s College London, London) that was delivered in three conditions, all offering up to 42 sessions of cognitive remediation. The conditions were: Intensive (one to one with a therapist), Group treatment with a therapist, Independent with drop-in sessions. Main outcome measures Work package 1: We developed two satisfaction measures and tested a team dynamic model. Work package 2: Feasibility and acceptability questionnaire, time to complete e-training modules. Work package 3: The personal recovery measure – Goal Attainment Scale. Results Work package 1: The service user satisfaction with cognitive remediation was reliable and valid. Although it did not show statistically significant differences between the arms of the trial, the most preferred methods (Group and Intensive) had higher associated satisfaction. Team leadership and especially a flattened hierarchy, resources and time were identified as vital for implementation. Our team dynamic model supported the importance of leadership in influencing organisational climate, which then affected staff attitudes. However, this was only significant before staff had any experience of their patients receiving cognitive remediation. Although the sample was much smaller after therapy, this may indicate that experience of the beneficial therapy changes team dynamics. Work package 2: The e-training modules were completed by 43% of the recruited participants. They judged the training to be feasible and acceptable, but it did take longer to complete than expected. COVID-19 with the increased workload may have had some effects, but our data exploration shows that it was individuals who had most recently qualified who had the best outcomes. This may be because of a lighter workload or that they were more used to online training. Adaptations suggested are now being implemented. Work package 3: Following the interim analysis we closed two arms – Independent therapy and Treatment as Usual. Four hundred and forty-eight patients consented and 377 were eligible and completed baseline assessment. They were randomised: Group 134, Independent 65, Intensive 112 and Treatment as Usual 66. At post therapy, there were no differences between Group and Intensive or between Independent and Treatment as Usual, but the combined Group and Intensive versus Treatment as Usual was significant (mean difference: 5.734; standard error = 1.958; p = 0.003; lower confidence interval = 1.898 to upper confidence interval = 9.571). Our economic analysis showed that Group and Intensive cognitive remediation were not different with respect to quality-adjusted life-years (difference £150, 95% confidence interval –£1132 to £1905). Both conferred significant benefit compared with standard care (Group and Treatment as Usual: difference £257, 95% confidence interval –£1694 to £2615; Intensive vs. Treatment as Usual: difference £260, 95% CI –£1654 to £2239). Their cost–benefit for quality-adjusted life-year improvement was well below the National Institute for Health and Care Excellence threshold for adopting the intervention to National Health Service services. Work package 4: Cognition had a small mediation effect, and negative symptoms moderated the transfer of cognitive benefits to goal attainment. Limitations The trial suffered from recruitment difficulties which were overcome when we switched from block to individual randomisation. The final target was large enough to test our main outcomes and moderating and mediating variables. Conclusions Cognitive remediation should be provided in the National Health Service, involving a trained therapist on a Group or Intensive format with team and training support. Future work We have a large database and will continue to investigate factors that affect cognitive remediation benefits. Study registration This study is registered as ISRCTN14678860 https://doi.org/10.1186/ISRCTN14678860. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Programme Grants for Applied Research Programme (NIHR award ref: RP-PG-0612-20002) and is published in full in Programme Grants for Applied Research; Vol. 12, No. 4. See the NIHR funding and Awards website for further award information. Plain language summary Cognitive problems in people with schizophrenia predict functional recovery even with the best-possible rehabilitation opportunities. One psychological treatment, cognitive remediation can improve both cognitive and functional recovery, such as in social relationships and occupational achievements. We investigated the optimal way of delivering computerised cognitive remediation in Early Intervention Services by incorporating service user and staff views. Our work involved exploring what treatment factors are important to service users, developing two bespoke cognitive remediation satisfaction measures for service users and service providers. Our investigation of team dynamics suggested that a more flexible leadership style together with increasing resources might be most suited to successful delivery. One challenge for large-scale roll-out of cognitive remediation is therapist training opportunities, so we developed an online programme and then tested it in the United Kingdom and, with a few adaptations, it can be provided to the National Health Service. Although cognitive remediation is effective, we do not know how much therapist time produces therapy benefits, so we conducted a trial comparing three options for cognitive remediation treatment that differed in the amount of the therapist time (Independent, Group and Intensive). We compared these options with a no-treatment group. We investigated benefits of each option in 377 participants from Early Intervention Services. The main benefit assessed was whether participants could achieve their personal goals that were set before therapy began. Although we found no change in a social or occupational functioning scale rated by an observer, the Group and Intensive cognitive remediation produced significant achievements in the participants’ own personal goals compared with those who did not receive therapy or who carried out therapy on their own. We also found that any cognitive remediation therapy option improved thinking skills. Group and Intensive therapy were also cost-effective. Overall, cognitive remediation was acceptable and well liked by service users, so all our data suggest that it should be more easily accessible in the National Health Service. Scientific summary Cognitive remediation (CR) therapy is a psychological intervention targeting cognitive dysfunction associated with psychosis with the goal of improving functional outcomes. Cognitive remediation can improve cognitive and functional recovery, as well as increasing confidence and motivation. Evidence from the largest meta-analysis demonstrated that these effects are durable. National Health Service Early Intervention for psychosis services provide care soon after a diagnosis of psychosis to optimise recovery. Embedding CR in those services may confer long-lasting benefits such as improved social relationships and the opportunity to work or take part in further education. The overarching aim of the Enhancing Cognition and quality of Life in early PSychosiEs (ECLIPSE) research programme was to determine the optimal method of introducing computerised CR therapy (CIRCuiTS™) into NHS Early Intervention Services (EIS) considering improvement in cognitive and social function and cost-effectiveness. The programme was centred on work package 3 (WP3), a randomised, blinded multiarm multisite controlled trial comparing three CR implementation methods with each other and with routine clinical care [Treatment as Usual (TAU)]. To aid the interpretation of WP3 and bolster our recommendations to the NHS, WP1 included a series of studies investigating: patient and staff satisfaction with CR; preference for the methods of CR delivery; and the influence of team organisational climate. Although the CR was computerised and can be completed independently, its effects are boosted via learning, practice and feedback supervised by a therapist. WP2 was a series of studies to design and test an online CR therapist training package to use in their individual teams. Finally, WP4 examined mediators and moderators of the trial results to inform teams about how the clinical features of their individual patients may influence the effect of CR. In WP3, the CR modalities were: Intensive 1 : 1 therapist supervision; Group supervision with one therapist; and Independent (self-determined with infrequent access to therapist supervision). Participants were recruited from EIS in 12 mental health trusts, linked to 6 universities and serving diverse urban/rural populations. They were clinically stable and presented with a non-affective psychotic episode not <3 months earlier. All participants were offered 42 hours of CR. Therapy was delivered by a supervised assistant psychologist using the CIRCuiTS software. The primary outcome was achievement of self-determined social goals measured by the Goal Attainment Scale (GAS). Secondary outcomes were changes in cognitive function, general social function, self-esteem and symptoms. An interim analysis was planned so that any treatment arm inferior to the others would be dropped. Assessments were conducted at three time points: baseline, post therapy and 6 months later. For the primary and secondary outcomes, intention-to-treat analyses estimated treatment effect contrasts for the following: Group versus Intensive; Independent versus TAU; Group + Intensive versus TAU using a linear mixed model with baseline scores, site, period (before and after interim analysis) and baseline missingness as covariates. The average treatment effect per hour of CR on the primary outcome and sensitivity analyses were performed. The interim analysis used end-point data from 100 participants: TAU (n = 20), Group (n = 33), Independent (n = 22), Intensive (n = 25). The Data Monitoring Committee recommended closing recruitment to the TAU and Independent arms. At the end of the trial, 448 patients consented and 377 were eligible and completed baseline assessment. They were randomised as follows: 134 to Group, 65 to Independent, 112 to Intensive and 66 to TAU. Post therapy, there was no statistical difference in GAS scores between the Group and Intensive arms [mean difference 0.737; standard error (SE) = 1.652; p = 0.655; lower confidence interval (CI) = –2.50 to upper CI = 3.975], or in the Independent arm compared with TAU (mean difference 0.695; SE = 2.448; p = 0.777; lower CI = –4.104 to upper CI = 5.493). There was a significant difference between the combined Group and Intensive sample versus TAU post therapy (mean difference 5.734; SE = 1.958; p = 0.003; lower CI = 1.898 to upper CI = 9.571). Post-therapy analyses of global cognition (composite score from several broad-ranging neuropsychological tests) demonstrated no difference between the Group and Intensive arms (mean difference 0.192; SE = 0.499; p = 0.699; lower CI = −0.785 to upper CI = 1.170), but there was a significant difference between the combined Group and Intensive conditions versus TAU at post therapy (mean difference 1.479; SE = 0.553; p = 0.008; lower CI = 0.395 to upper CI = 2.564). There was also a trend for significance between Independent and TAU (mean difference 1.348; SE = 0.701; p = 0.054; lower CI = −0.026 to upper CI = 2.722). There were no other secondary outcome differences for measures of social function (Social and Occupational Functioning Assessment Scale and time use survey), negative symptoms (Clinical Assessment Interview for Negative Symptoms) and self-esteem (Rosenberg scale) at post therapy and no differences at 6 months for any measure. Ninety-six adverse events (AEs) were reported for 59 individuals with 60 classified as serious adverse events (SAEs) for 44 participants. The rates of AEs and SAEs were similar across arms. Two participants died, both in the Intensive CR arm, but neither were judged as related to the study intervention. Work package 3 also included a cost-effective analysis using quality-adjusted life-years (QALYs) as the outcome. QALYs were derived from the EuroQol-5 Dimensions (baseline to 6 months). Service use was measured by the Client Service Receipt Inventory and combined with unit cost information (Personal Social Services Research Unit) to calculate costs. Intervention costs were calculated from therapist training, supervision and access to CIRCuiTS. Differences in costs were not significant (Group vs. Intensive, £26,383 per QALY; Group vs. TAU, £4306 per QALY; Intensive vs. TAU, £3170 per QALY). Group versus TAU and Intensive versus TAU had significantly higher QALYs than TAU, suggesting that Group and Intensive CR provided more benefit compared with TAU. There were no significant differences in costs and outcomes for GAS. Work package 3 summary. CR provided in Intensive and Group modalities significantly enabled participants to achieve their self-determined goals to a greater extent than standard care alone. Global cognition was also significantly improved by Intensive and Group modalities compared with standard care. Group and Intensive modalities were indistinguishable. There was an indication that CR used independently may lead to improved global cognition compared with standard care. Group and Intensive CR were not different with respect to QALYs, and both conferred significant benefit compared with standard care and their cost–benefit for QALY improvement was well below the National Institute for Health and Care Excellence (NICE) threshold for adopting the intervention to NHS services. To assess CR acceptability, a new satisfaction scale was developed with service users, representative of trial participants and EIS staff in focus groups using thematic analysis (WP1, Study 1). These were validated (WP1, Study 2) with a survey sent to participants and site teams at the end of the trial. Test–retest analysis showed good reliability; concurrent reliability using the Working Alliance Inventory and validity for the service user scale. The investigation of differences in the different CR approaches (WP1, Study 2) demonstrated that mean differences in satisfaction score were in favour of Group CR over Intensive CR and Independent CR and in favour of Intensive CR over Independent CR, but these differences were not statistically different. Focus groups conducted with staff and service users assessed service user and staff views of implementation using thematic analysis (WP1, Study 3). There was strong preference for regular open-ended 1 : 1 sessions with options for drop-in sessions. Organisational climate was measured in two ways. A qualitative study involving semistructured interviews with staff with thematic analysis (WP1, Study 4) indicated that consideration of local conditions and organisational microclimates mediates the successful implementation of new interventions and is needed in addition to generic, context-free variables such as resources before new interventions can be introduced. A quantitative survey (WP1, Study 5) with staff in the participating sites used a set of validated scales assessing team climate, leadership and team attitude to evidence-based practice (before and after CR implementation). We tested whether leaders affect the climate for use of CR and whether that then influences staff attitudes to CR. Before the introduction of CR, that model was supported. After staff had experience of CR, it was not a significant model. This may be because staff experienced the patient benefit and so the effect of leaders was reduced. If we take the two sets of results together, then improving the organisational climate through flexible leadership and increasing the resources for the intervention will enhance its adoption. Experience of beneficial interventions can reduce the effects of leadership and bodes well for continued adoption. In WP2, an e-learning CR programme based on previous training materials was further developed through focus groups with mental health clinicians from a range of specialities. It includes multimedia presentations (film, video and short exercises) (Study 6). In Study 7, the acceptability and feasibility of this package were tested with volunteers from eight mental health trusts. Completion of training took longer than expected, partly explained by constraints during the COVID-19 pandemic. Feedback from the participants was positive and their suggestions for what would have helped them complete the training more quickly will be used to refine the training programme. Following completion of the trial, mediators and moderators of the effects were investigated in WP4. The assumption was made that variation in the intention-to-treat estimates across arms was derived principally from the variation in the time-on-task achieved in each arm. This was tested using analysis of covariance, and a significant effect was found for time-on-task increasing the GAS by 0.187 (95% CI 0.062 to 0.312; p = 0.003) for each hour of therapy. There were no significant difference in the effects of time-on-task between the Group and Intensive arms (Wald: 1df p = 0.753) and no difference between these pooled therapist-assisted and independently undertaken treatment effects (Wald: 1df p = 0.630) when time-on-task was considered. We tested a model that cognitive improvement (using the post-therapy global cognitive score) mediated the effect of CR time-on-task on Goal Attainment. The model, including baseline variables, fitted well [root mean square error of approximation = 0.00; Tucker Lewis index = 1.0122; (27df) p = 0.668] and showed that the mediated path from time-on-task to cognition and cognition to functional outcome was small and non-significant (Coeff = 0.014, p = 0.248, 95% CI –0.009 to 0.037). Baseline global symptoms did not moderate the effect of time-on-task on cognition (p = 0.098). However, negative symptoms significantly moderated the effect of improved cognition on goal attainment (p = 0.016). Participants with high negative symptoms spent on average an hour more on CR tasks than those with low negative symptoms. The mediation effect of change in cognitive score on the change in Goal Attainment was estimated for high and low negative symptom groups. Larger changes in cognition were associated with larger changes in the GAS score overall, but this was not evident in the high negative symptom group, suggesting that negative symptoms can interfere with mobilising improvements in general cognition to achieve a desired functional outcome. The ECLIPSE programme overall provides valuable evidence of the sort of factors to consider specifically with the introduction of CIRCuiTS into EIS. It also provides insights for introducing cognitive interventions in general, as well as novel interventions in all services. We need to understand the team climate, influence the leaders and provide evidence of benefits. We were not able to differentiate between the two types of therapy (Group and Intensive); both improved personal goals and were cost-effective. The cost of improving by one QALY was lower than the threshold for the NICE, and so there is now ample evidence that it should be more generally available. As the therapy needs trained therapists, the inclusion of an e-learning programme will boost large-scale roll-out. We do not know if everyone improves, but there is an indication that those with higher negative symptoms benefit less from the cognitive improvements. This will be investigated further, but the result suggests that therapy adaptations may be required, including increasing the number of sessions and a more integrated transfer of cognitive skills into everyday life. Study registration This study is registered as ISRCTN14678860 https://doi.org/10.1186/ISRCTN14678860. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Programme Grants for Applied Research Programme (NIHR award ref: RP-PG-0612-20002) and is published in full in Programme Grants for Applied Research; Vol. 12, No. 4. See the NIHR funding and Awards website for further award information.
Keywords
