Can the prediction model using regression with optimal scale improve the power to predict the Parkinson's dementia?

BACKGROUND: Efficiently detecting Parkinson's disease (PD) with dementia (PDD) as soon as possible is an important issue in geriatric medicine. AIM: To develop a model for predicting PDD based on various neuropsychological tests using data from a nationwide survey conducted by the Korean Centers for Disease Control and Prevention and to present baseline data for the early detection of PDD. METHODS: This study comprised 289 patients who were 60 years or older with PD [110 with PDD and 179 Parkinson's Disease-Mild Cognitive Impairment (PD-MCI)]. Regre-ssion with optimal scaling (ROS) was used to identify independent relationships between the neuropsychological test results and PDD. RESULTS: In the ROS analysis, Korean version of mini mental state ex-amination (MMSE) (KOREAN version of MMSE) (b = -0.52, SE = 0.16) and Hoehn and Yahr staging (b = 0.44, SE = 0.19) were significantly effective models for distinguishing PDD from PD-MCI (P < 0.05), even after adjusting for all of the Parkinson's motor symptom and neuropsychological test results. The optimal number of categories (scaling factors) for KOREAN version of MMSE and Hoehn and Yahr Scale was 10 and 7, respectively. CONCLUSION: The results of this study suggest that among the various neuropsychological tests conducted, the optimal classification scores for KOREAN version of MMSE and Hoehn and Yahr Scale could be utilized as an effective screening test for the early discrimination of PDD from PD-MCI.