Diagnostic Accuracy of Noun- and Verb-Naming Tasks in Detecting Cognitive Impairment in Parkinson's Disease

Background: In Parkinson's disease (PD), verb-naming tasks (VNTs) have been proposed as superior to noun-naming ones in detecting language deficits, although such a hypothesis is not supported at a statistical level. Objectives: The main aim of this study was to provide diagnostic accuracy evidence for a VNT and noun-naming task (NNT) in detecting cognitive impairment (CI) in PD patients. Method: Thirty-three consecutive PD patients were subdivided into participants with (PD-CI; N = 12) or without CI (cognitively unimpaired, PD-CU; N = 21), based on a raw score <= 25 or >25 on the Mini-Mental State Examination, respectively. The NNT and VNT by Neuropsychologia [2006 Jan;44(1):73-89] were administered. Diagnostic accuracy of the NNT and VNT was assessed through receiver-operating characteristics analyses by comparing PD-CU to PD-CI patients. At the optimal cut-off, sensitivity, specificity, positive and negative predictive values (PPV, NPV), and likelihood ratios (LR+, LR-) were separately tested for the NNT and VNT against PD-CU versus PD-CI classification. Results: Diagnostic accuracy was higher for the NNT (AUC = 0.85; p = 0.001) versus the VNT (AUC = 0.68; p = 0.092). Consistently, the NNT yielded higher sensitivity, specificity, and post-test features than the VNT (NNT: sensitivity = 0.75, specificity = 0.81, PPV = 0.69, NPV = 0.85, LR+ = 3.94, LR- = 0.31; VNT: sensitivity = 0.67, specificity = 0.67, PPV = 0.53, NPV = 0.78, LR+ = 2, LR- = 0.5). Conclusions: In accordance with the Movement Disorders Society guidelines, NNTs are diagnostically sound psychometric instruments to discriminate PD patients with versus without CI. However, these findings need replication by (1) employing a gold standard different from the Mini-Mental State Examination, which does not capture the full range of CI in this population and (2) subdividing PD patients into those with mild CI and dementia.