Association between stressful life events and psychosis relapse: a 2-year prospective study in first-episode psychosis

Stressful life events occurring after the onset of psychosis have been associated with poorer long-term outcomes1. However, methodological issues with existing evidence, such as inadequate consideration of the confounding effect of illness stage and of socio-demographic and clinical variables, limit a clear understanding of the implications of this finding. Further, as most available evidence is based on retrospective studies, which are susceptible to recall bias, prospective evidence that life events preceded and were in reasonable temporal proximity to the psychosis relapse is needed to support the notion that these events might have a precipitating role. About one in two patients will present with a relapse severe enough to require hospital readmission within the first two years of their first episode of psychosis2. Relapses not only cause considerable suffering to the individuals and their families, but also have implications for utilization of health care resources. Here, we employed a prospective cohort approach to investigate the effect of stressful life events on the risk of relapse, as indexed by hospital admission, over the first two years following psychosis onset. First-episode non-organic psychosis patients (ICD-10: F20-F29 and F30-F33) aged 18-65 years, admitted to psychiatric services in the catchment area of South London, were prospectively recruited and followed up for at least two years. Stressful life events that occurred over the follow-up period were assessed using the Brief Life Events Questionnaire (BLEQ), a tool that allows the assessment of the time of occurrence of each event and its emotional impact, with high validity and reliability3. A full treatment history was recorded by the World Health Organization (WHO) Life Chart Schedule4. Relapse was defined as admission to a psychiatric inpatient unit because of exacerbation of psychotic symptoms within two years of first presentation to psychiatric services and receiving a diagnosis of psychosis. Separate survival analyses were carried out to investigate the effect of any life events and of total number of life events (occurring within the two-year period following onset of psychosis) on time to first relapse, using Cox proportional hazards regression in a multivariable model controlling for the effect of potential confounders (gender, ethnicity, relationship status, age of psychosis onset, care intensity at onset, diagnosis at onset, medication adherence, alcohol use, cigarette use, other illicit drug use). As the proportional hazards assumption was violated at different levels of cannabis use, the model was stratified by that variable. Kaplan-Meier plots (created using the ‘survminer’ package in R) were used to depict unadjusted survival data. Two hundred fifty-six patients with first-episode psychosis were recruited into the study. Most of them were men (61%), of non-White ethnic origin (66%), and not in a relationship (74%). The prevalence of cigarette use was 57%, that of problematic alcohol use 14%, that of cannabis use 39%, and that of other illicit drug use 18%. The mean age at psychosis onset was 28.06±8.03 years. Most patients presented with non-affective psychosis (82%), were admitted to hospital close to the psychosis onset (78%) and in the context of a compulsory admission (60%). Within two years from the onset of the disorder, 36% of recruited patients experienced at least one relapse of psychosis requiring hospital admission. The highest number of relapses recorded in the study period was three, with the longest hospital stay lasting 14.8 months. Patients who had experienced any stressful life event following their psychosis onset (42%) had a significantly higher risk of relapse (as indexed by hospitalization) compared to those who did not experience any stressful life event (hazard ratio, HR=1.71, 95% CI: 1.11-2.64, p=0.016), after controlling for the above-mentioned socio-demographic and clinical factors (see also supplementary information). Including medication adherence into the model, while still controlling for the socio-demographic and clinical factors, did not substantially change the results (HR=1.77, 95% CI: 1.13-2.79, p=0.013). A higher risk of relapse was observed as a function of the number of experienced stressful life events, but this was statistically significant only after adjusting for medication adherence as well as for the above-mentioned socio-demographic and clinical factors (HR=1.23, 95% CI: 1.01-1.50, p=0.037). Among the socio-demographic and clinical factors controlled for, African ethnic origin, not being in a relationship, being a cigarette user, receiving a higher care intensity at onset (i.e., being hospitalized) and having poor medication adherence were all significantly associated with increased risk of relapse in survival analyses (see supplementary information). In this study, we attempted to address most limitations of previous research. In particular, we used a prospective longitudinal design to avoid the recall bias that is inherent to retrospective studies5. Our results, therefore, provide evidence to support a temporal relationship between exposure to stressful life events and subsequent psychosis relapse, in line with the “triggering” hypothesis of psychosis6. Further, by restricting recruitment to first-episode patients, we were able to mitigate the potentially confounding effect of a highly variable clinical course of psychosis, that is especially relevant to patients suffering from psychotic disorders of longer duration. Higher clinical severity at onset7 and poor medication adherence2 have been found to be robust indicators of subsequent admissions and poor outcome in patients with psychosis. Also, ­converging evidence supports higher odds of poor outcome a­mong psychosis patients of non-White ethnic origin8, and in cigarette users9. By including these predictors in our model, we found that our results were consistent with previous work, but we were able to add stressful life events to the list of risk factors for psychosis relapse that are supported by robust evidence. By lending support to the notion that stressful life events can have a significant role in psychotic relapse, the present results may have clinical and public health implications for the prevention and treatment of psychosis. In particular, they call for approaches allowing for real-time measurement of life events in clinical settings, so that timely interventions can be implemented to pre-empt potential adverse consequences.