Differential predictors of expressed emotion toward individuals with schizophrenia between families and halfway houses

Background This study investigated patient- and caregiver-related predictors of expressed emotion (EE) toward individuals with schizophrenia in families and halfway houses and yet understudied differential effects across settings.Methods We included 40 individuals with schizophrenia living with their families ("outpatients") and 40 "inpatients" in halfway houses and recorded the EE of 56 parents or 22 psychiatric nurses, respectively, through Five Minutes Speech Sample. Each outpatient was rated by one to two parents; each inpatient was rated by two to five nurses. As EE ratings had a multilevel structure, EE predictors were investigated in backward stepwise generalized linear mixed models using the "buildmer" R package. We first fitted models including either caregiver- or patient-related predictors in each setting and finally included both types of predictors. Setting-specific patient-related effects were investigated in interaction analyses. Adjustment for multiple tests identified the most robust associations.Results In multivariate models including either caregiver- or patient-related predictors, nurses' higher age, shorter work experience and lower inpatients' negative symptoms robustly predicted higher emotional overinvolvement (EOI). In the final models including both types of predictors, nurses robustly displayed lower EOI (i.e., reduced concern and disengagement) toward inpatients with higher negative symptoms. Several other features were nominally associated with criticism and EOI in each setting. However, no feature robustly predicted criticism in inpatients and criticism/EOI in outpatients after adjustment for multiple tests. In interaction analyses, higher negative symptoms differentially predicted lower EOI in nurses only.Conclusion Our findings suggest setting-specific pathogenetic pathways of EOI and might help customize psychoeducational interventions to staff in halfway houses.