Overview of research This thesis consists of five studies investigating the effects of psychological interventions for psychosis patients. Each study utilises a different methodology allowing a variety conclusions to be drawn. Chapter 1 provides a general introduction to the topic, including the background and history of psychological interventions for psychosis and their development in context to the dominant medical model, alongside an overview of the current evidence for psychological interventions. The critical questions that this thesis aims to address are also introduced. Four of the five studies included in this thesis use meta-analytical methods. Meta-analysis is a statistical procedure that allows the comparison of multiple existing published studies to provide an overall estimate. When conducted carefully, meta-analyses can provide more information than individual trials alone and forms the basis of most guidelines on healthcare interventions. Meta-analyses typically include randomised controlled trials (RCTs), which are scientific studies testing interventions against control conditions. These RCTs are conducted under strict conditions to improve the validity of results. The final study in this thesis uses RCT methodology to test a brief intervention. A general discussion is then provided which considers the implications of the findings alongside strengths, weaknesses and suggestions for future research. Summary of research Chapter 2 provides a comparative meta-analysis of psychological interventions, which includes any major form of psychological interventions for psychosis for which there was sufficient available research comparing them against other interventions to qualify. The meta-analysis therefore included randomised controlled trials on cognitive-behavioural therapy for psychosis (CBTp), social skills training, cognitive remediation, psycho-education, supportive therapy and befriending. A systematic search was conducted of four key databases resulting in the selection of 48 RCTs including 3,295 participants with psychosis. The data from all RCTs was analysed to pool the effect size from each study, which provided an aggregated statistic for each comparison. The quality of the RCTs was also assessed and the results of this assessment were used in further analyses to ensure the validity of results. Results showed that CBTp was more beneficial than other interventions for positive symptoms (which include key psychosis symptoms such as hallucinations and delusions) while social skills training was more beneficial for negative symptoms (which include apathy and lack of motivation). Cognitive remediation also showed a beneficial effect for overall symptoms, as did CBTp, although these findings did not survive the extra ‘sensitivity analyses’ we conducted based on methodological quality of the included RCTs. Based on these results, it was concluded that there are small but reliable differences between psychological interventions for psychosis. Chapter 3 provides a network meta-analysis focused on the impact of psychological interventions on psychotic symptoms. Network meta-analysis is an alternative methodology which allows researchers to draw statistical conclusions not only from direct comparisons between interventions but also indirect evidence using the network model. A systematic search was conducted resulting in the inclusion of 90 RCTs and 8,440 randomised participants with psychosis. Network meta-analysis was used to examine direct and indirect evidence for ‘total symptoms’ of psychosis, which is an overall measure including all relevant psychotic symptoms. Study quality was again assessed to help ensure validity of results. Results demonstrated that psychological interventions were of significant benefit compared to control groups. Mindfulness-based psycho-education was shown as the intervention most likely to reduce total symptoms. However, all included RCTs for this intervention were from China, meaning that future research investigating the efficacy of mindfulness-based psycho-education in a variety of cultural contexts may help determine whether these findings generalise to other international settings. While the previous chapters apply conventional meta-analytic techniques, chapter 4 provides a cumulative meta-analysis investigating the impact of individualised, case-formulation based CBTp on hallucinations and delusions, which are the key features of positive symptoms. Case formulation refers to an essential technique in CBT that helps to individualise a patient’s treatment and allows a close conceptual link between research and clinical practice. Cumulative meta-analysis is a novel technique that, alongside providing information on the effectiveness of a treatment, can also help us determine whether the evidence base for that treatment is sufficient and stable. A systematic search resulted in the inclusion of 35 RCTs and 2407 participants with psychosis. Meta-analyses were conducted and study quality was again assessed to help determine the validity of results. Results demonstrated that the evidence base for CBTp has been sufficient and stable since 2016 for hallucinations and 2015 for delusions. CBTp was demonstrated as beneficial for hallucinations compared to any control, treatment as usual and active controls. For delusions, CBTp was beneficial when compared to any control and treatment as usual, but did not demonstrate significant benefit against active controls although there were a limited number of RCTs included in this comparison, which may limit validity. The effects of CBTp were also shown as stronger when case-formulation was used and also when the primary focus of the study was the reduction of hallucinations or delusions instead of other outcomes. The fact that the evidence for CBTp has been shown as sufficient and stable means that there may be limited worth in continuing to spend vital resources on similar RCTs testing ‘generic’ CBTp and resources may better be directed into developing new or improved variants. Chapter 5 utilises another novel meta-analytic technique allowing the application of individual-participant data (IPD). In this approach, the original databases from published RCTs are requested from authors meaning that the individual data for each participant can be used in analyses rather than relying on the summary effect size data available in published manuscripts. The IPD approach allows a more precise estimation of effects and allows the investigation of ‘moderator’ variables, which refer to demographic or clinical variables at the individual level that may impact who benefits most from treatment. This study was initially developed as follow-on from chapter 1 and attempted to source databases all the included CBTp RCTs alongside conducting a new systematic search to determine whether any new RCTs were eligible. After contacting relevant authors, databases for 14 of 23 eligible RCTs were included resulting in the data for 898 participants with psychosis being included. CBTp was demonstrated as beneficial for total psychotic symptoms and general symptoms, although not for positive symptoms. This finding contrasts results from the previous chapters, although may be explained by the exclusion of a proportion of eligible RCTs due to failure to obtain these databases from original authors. The moderator analysis did not show any demographic or clinical variables as influencing treatment outcome although the number of therapy sessions a patient received had impact on outcome. The results of this IPD meta-analysis suggest that patient characteristics, including severity of psychotic symptoms, do not significantly influence treatment outcome while sufficient ‘dosage’ of CBTp is important. Finally, chapter 6 reports on a secondary analysis of an RCT conducted in a clinical setting in the UK National Health Service (NHS). This study examined the effects of a brief psychological intervention aiming to address overconfidence in perceptual decision-making among patients diagnosed with psychosis. 31 patients aged 16-65 were randomly assigned to one of two groups; 1) a brief intervention based on ‘metacognitive training’ which aimed to address a common thinking bias called the “jumping-to-conclusions” bias, or 2) an attention-control condition designed to account for therapist time and attention. Participants completed outcome measures assessing overconfidence and the “jumping-to-conclusions” bias. Results demonstrated that those receiving meta-cognitive training experienced a significant reduction in overconfident reasoning when compared to those receiving the control condition. This RCT provides preliminary evidence that meta-cognitive training is a worthwhile method by which to address overconfident reasoning in psychosis. There were however methodological limitations of this RCT due to limited resources. A larger RCT with stronger methodology is therefore warranted. Conclusions When considered collectively, the findings from the body of research included in this thesis provide strong evidence for the validity of psychological interventions for psychosis. The evidence base for CBTp was demonstrated as sufficient and stable, while social skills training was demonstrated as an effective intervention for negative symptoms. The results for CBTp are important in the on-going debate about effectiveness and whether or not it has been “oversold.” In light of the accumulated evidence, future research on psychological interventions for psychosis may best focus on the development of new or improved approaches and move on from the debate on whether psychological intervention “works” or not.