The psychometric discriminative properties of the Peters et al Delusions Inventory: a receiver operating characteristic curve analysis

Objective: To establish sensitivity, specificity, and discriminative validity of the Peters et al Delusions Inventory (PDI), a recently released self-compiled questionnaire aimed at measuring psychosis proneness in the general population. Method: Eighty-one patients diagnosed with a mental disorder with psychotic features (schizophrenia, schizoaffective disorder, bipolar disorder, major depression) and 210 putatively healthy individuals were invited to complete the PDI and the brief version of the General Health Questionnaire (GHQ-12). The receiver operating characteristic analysis has been used to extract the best threshold to discriminate between cases, those with a psychotic disorder, and noncases, those without a psychotic disorder. Results: The best PDI threshold in discriminating between cases and noncases was 8 (sensitivity, 74%; specificity, 79%). The corresponding figures of sensitivity and specificity for a GHQ-12 cutoff of 6 were 63% and 83%, respectively. The value of the area under the receiver operating characteristic curve was found to be higher for the PDI (0.815) than for the GHQ-12 (0.770), although at a barely detectable, statistically significant difference (Z = 1.45, P = .073). Conclusion: The expected high negative predictive value of the PDI (96%) in the general population suggests it will be a valuable tool in future research on psychosis proneness. © 2007 Elsevier Inc. All rights reserved.