Delusional beliefs and experiences can predict the development of mental disorders within the spectrum of psychosis. The nature, content and prevalence of delusional experiences in the general population are still disputed topics. This study investigates the latent structure of delusion proneness in the non-clinical population. Eight hundred young adults (400 from Italy and 400 from the United Kingdom) completed the Peters et al. delusions inventory, a general population measure of delusional proneness. Latent class analysis was used to explore the latent structure of delusion proneness. Four classes were identified: low delusion proneness (including 28% of the sample), grandiosity (13%), paranoid thinking (41%) and positive psychotic beliefs (18%). Latent structures of sub-clinical symptoms can be observed also in non-clinical population; paranoid thinking is the most common delusional theme.
Psychotic experiences and beliefs are largely reported in the general population with prevalence between 5 to 25% (Stip and Letourneau, 2009). Similarly to other psychopathological traits (e.g. depression) it has been hypothesized that psychotic traits may be distributed in a dimensional fashion between general and clinical populations (Johns and van Os, 2001 and van Os et al., 2009). Such subjective experiences that can be ascribed to the sphere of psychoses may therefore predict the future onset of mental disorders such as schizophrenia, manic-depression, or schizoaffective disorder (Chapman et al., 1994, Krabbendam et al., 2004 and Hanssen et al., 2005).
Patients diagnosed with psychosis are extremely variable in their symptoms presentation; in particular within the dimension of positive symptoms, high levels of heterogeneity have often been observed contributing to the idea that meaningful typologies of patients could be difficultly achieved. Latent class analysis (LCA) was initially applied to the psychosis spectrum to explore the contribution of latent traits to heterogeneous symptoms presentation (Jørgensen and Jensen, 1990 and Shevlin et al., 2007). LCA posits that a heterogeneous group can be reduced to several homogeneous sub-groups by evaluating and then minimizing the associations among responses across multiple variables (Lazarsfeld and Henry, 1968 and McCutcheon, 1987). Latent class analyses therefore empirically tests for the existence of discrete groups with a similar symptom or item endorsement profile, distinguishing it from factor analysis which assumes the presence of continuous latent variables (Hudziak et al., 1998).
Studies using LCA on patients suffering from schizophrenia suggested the existence of several latent classes such as the schizoaffective class (Castle et al., 1994), the disorganized class (Murray et al., 2005), the positive and negative classes (McGrath et al., 2004) and the paranoid class (Castle et al., 1994). A number of investigations have tried to evaluate the consistency of these latent dimensions in mixed clinical samples and in the general population finding good correspondence between clinical and non-clinical classes for positive symptoms (Kendler et al., 1998 and Rocchi et al., 2008). When confined to non-clinical populations, LCA studies on positive psychotic experiences found four classes: a normative class, an intermediate class (considered as a transition state towards psychosis), a “paranoid” or global psychosis class and a positive symptoms or hallucinatory class (Shevlin et al., 2007 and Murphy et al., 2007).
This study aims to investigate the latent structure of delusion proneness in non-clinical samples recruited from different countries and evaluate the consistency of the 4-class solution in young adult samples.
3.1. General results
Internal consistency of the PDI assessed by Cronbach alpha coefficient was 0.74 (95% C.I.: 0.70 to 0.77) for the UK sample and 0.78 (95% C.I.: 0.74 to 0.81) for the Italian sample.
Sample composition and PDI-21 scores are reported in Table 1. Only the mean weighted distress, preoccupation and conviction and the total number of item endorsed were used in the analysis. PDI total scores did not differ by location (MANCOVA; Table 2). Distress and preoccupation scores were higher in the Italian sample (P < 0.001). Further, in both samples distress co-varied with gender, with higher scores in females (P = 0.004) and preoccupation co-varied with age, with younger more preoccupied (P = 0.015). No other significant differences were identified.
