Validation of the National Institute of Neurological Disorders and Stroke Criteria for Psychosis in Parkinson Disease

Abstract

Parkinson disease (PD) psychosis is a condition associated with several negative outcomes. Despite its impact, there is a lack of validated diagnostic tools for this condition. In this study, we aim to verify the validity of the proposed NINDS criteria for PD psychosis and explore its possible applications in clinical practice. We prospectively selected 104 subjects with idiopathic PD referred to a movement disorder clinic for a cross-sectional evaluation. A neurological evaluation confirmed idiopathic PD and classified PD psychosis according to the NINDS criteria. A psychiatrist then classified the subject according to DSM-IV-TR criteria for psychosis, considered the reference standard. We used Cohen's kappa (κ) to quantify reliability between methods. Finally, we designed models assigning a weighted score to each characteristic psychotic symptom from the NINDS criteria (criterion A), and plotted receiver operating curves for each model. Of the total sample, 52 (50%) met proposed criteria for NINDS PD psychosis and 16 (15.6%) met reference standard criteria. Inter-rater reliability showed only a fair agreement (κ = 0.30). By using a scoring approach for each NINDS criteria item and a cutoff total score for the diagnosis of PD psychosis, we significantly increased the agreement for diagnosis reliability (κ = 0.72), with sensitivity of 94% and specificity of 91%. Although the NINDS criteria had limited reliability for diagnosing PD psychosis, a scoring approach for symptoms showed good reliability, with sensitivity and specificity above 90%. This scoring approach may be an accurate tool for identifying patients with PD psychosis.