Addressing Loneliness in Psychosis: A Rapid Realist Review of Multi-Level Interventions

Abstract People living with psychosis experience disproportionately high levels of loneliness, yet there is currently a lack of interventions specifically tailored to their needs. Existing approaches often focus on individual-level change, overlooking the broader social, interpersonal, and structural factors that compound disconnection. Given the complex and intersecting drivers of loneliness in this population-including stigma, marginalisation, and reduced access to meaningful social roles-a more nuanced understanding is essential to inform effective support. We conducted a rapid realist review to explore how, why, and in what contexts interventions may reduce loneliness in people with psychosis. The review synthesises findings from 65 peer-reviewed papers and is enriched by consultations with academic experts and lived experience advisors. Context-Mechanism-Outcome (CMO) configurations were developed to build a theory-driven framework of effective intervention strategies. The review identified a broad range of potential intervention targets at the individual (e.g. self-esteem, internalised stigma), interpersonal (e.g. relationship quality), community (e.g. social integration), and structural (e.g. housing) levels. For each, we suggest possible intervention strategies grounded in theory and evidence. Findings highlight the importance of intersectionality and support the development of personalised, co-produced, and multi-level approaches that are responsive to the lived realities of people with psychosis. This work will inform the design of intervention and public health strategies to reduce loneliness and promote inclusion in this underserved group. Key messages • There is currently a lack of loneliness interventions tailored for people with psychosis; we identify what works, for whom, and in what contexts. • Reducing loneliness in psychosis requires multi-level responses that address stigma, exclusion, and structural inequality.