Childhood adversity (CA) is a well-established risk factor for psychotic disorders. The reviewed preprint aimed to synthesise and quantify mediating pathways linking CA and psychosis, focusing on treatable psychological and biological mechanisms. The review followed PRISMA guidance and was registered with PROSPERO (CRD42024542972). The search was performed in January 2026 on Ovid platforms covering PsycINFO, Medline, and Embase for terms related to psychosis, CA, and mediation analyses.
This systematic review identified studies that tested mediation between CA and psychosis-related outcomes. The authors extracted effect estimates and, where available, calculated the percentage of the total effect mediated for each reported mediator. Mediators were grouped into clinically meaningful families to summarise evidence across studies. When feasible, quantitative synthesis used two-stage meta-analytic structural equation modelling (METASEM) to pool indirect effects across studies.
A total of 117 studies met inclusion criteria. Sample composition included 54 studies in clinical samples, 59 in non-clinical samples, and 4 that included both clinical and non-clinical participants. Of the included studies, 107 examined psychological mediators and 12 examined biological mediators. The authors report that fewer biological mediator studies precluded aggregation into a pooled quantitative estimate for biological pathways.
Across the pooled descriptive analyses, the authors report median percentages of the total CA–psychosis effect mediated by different mediator families. Reported medians were: dissociation 49% (k = 24), psychosocial stressors 45% (k = 6), negative schemas 37.9% (k = 23), post-traumatic symptoms 35.2% (k = 10), depressive symptoms 31.5% (k = 14), anxiety 27.8% (k = 11), attachment styles 27.1% (k = 12), and mentalization domains 8.7% (k = 5). These medians summarise the proportion of the CA–psychosis association attributed to each mediator family across included studies.
When quantitative meta-analysis was possible, the authors report pooled indirect effects and the approximate percentage of total effect mediated for selected mediators:
Dissociation: Meta-analytic synthesis using METASEM included k = 7 studies (N = 2,143) for overall psychosis with a pooled indirect effect of 0.42 (95% CI 0.17 to 0.66), representing approximately 50.49% of the total effect. Dissociation was also pooled specifically for delusions (k = 7; N = 1,053; I.E. = 0.36; 46.44%) and hallucinations (k = 10; N = 5,705; I.E. = 0.28; 57.59%).
Depressive symptoms: A pooled mediation effect was reported (k = 5; N = 5,028) with an indirect effect of 0.33 (95% CI 0.31 to 0.35), corresponding to about 31.05% of the total CA–psychosis effect in the included studies.
Negative schemas: Meta-analysis across k = 7 studies (N = 10,791) yielded an indirect effect of 0.26 (95% CI 0.17 to 0.35), or approximately 26.36% mediated between trauma and psychosis broadly defined.
These pooled estimates indicate meaningful mediation via dissociation, depressive symptoms, and negative schemas in the sampled literature.
The authors report high heterogeneity across all quantitative syntheses. High between-study variability may reflect differences in study populations (clinical vs non-clinical), outcome definitions (psychosis broadly, delusions, hallucinations), measurement of CA and mediators, study design, and analytical approaches used in the primary studies. The report notes that only a subset of mediator types and studies could be combined using METASEM, and that observed heterogeneity limits the certainty with which pooled indirect effects can be generalized.
Additional limitations inherent to the review include the preprint status of the article (the work has not been peer reviewed) and the descriptive nature of some aggregated median percentages when full quantitative pooling was not possible. The authors declare no competing interests. The preprint was posted July 21, 2026.
Twelve included studies examined biological mediators, but the authors state that the number and consistency of these studies were insufficient to permit quantitative synthesis. As a result, the review cannot provide pooled estimates for biological pathways linking CA and psychosis. The authors emphasise that further research is needed to evaluate biological mechanisms and their interactions with psychological mediators.
Across 117 studies, the review concludes that CA contributes to psychosis mainly through psychosocial and psychological mediators, with dissociation emerging as a particularly strong mediator. Depressive symptoms and negative schemas also showed robust mediating effects in pooled analyses. The paucity of comparable biological mediator studies prevented meta-analytic conclusions about biological pathways. The authors recommend further work to clarify biological contributions and to investigate how biological and psychological mechanisms interact.
Clinicians and researchers should note that this synthesis is based on a preprint and awaits peer review. The findings underscore potential targets for intervention (for example addressing dissociation and trauma-related cognitive and affective processes) but do not constitute clinical guidance until supported by peer-reviewed evidence.