Table 4. Results of univariate and multivariate hierarchical logistic regression analyses (method ENTER) for predictors of violent victimisation and victimisation of property crimes in psychiatric patients (n = 300)

Criminal victimisation prevalence was measured in 300 outpatients living in Amsterdam, The Netherlands. Face-to-face interviews were conducted with outpatients with depressive disorder (n=102), substance use disorder (SUD, n=106) and severe mental illness (SMI, n=92) using a National Crime Victimisation Survey, and compared with a matched general population sample (n=10865). Associations of socio-demographics, clinical variables and substance use were analysed for violent and non-violent crimes within the patient groups. First, a univariate analysis was conducted. Furthermore, to examine associations between psychopathology and victimisation, univariate regression analyses were performed for the different BPRS items. Additional univariate regression analysis was performed to examine the association of co-morbid substance use with victimisation; this was done only for the Depression and SMI subgroup due to multicollinearity with the SUD subgroup. Subsequently, a multivariate hierarchical logistic regression analysis was conducted (method ENTER). In this analysis two separate groups of independent variables were added subsequently. In step 1, socio-demographic characteristics were entered (age, gender, ethnicity, education, living alone and employment); in step 2 the diagnostic subgroup was entered (SMI, Depression or SUD). In all analyses the level of significance used was p<0.05.