Maximising information on smell, quantitative motor impairment and probable REM-sleep behaviour disorder in the prediction of Parkinson’s disease

ABSTRACT Background Hyposmia, motor impairment and probable REM-sleep behaviour disorder (RBD) are markers for Parkinson’s disease (PD). Proposed PD risk prediction models have dichotomised test results and applied likelihood ratios (LRs) to scores above and below cut-offs. We investigate whether LRs for specific test values could enhance prediction models. Methods Smell and probable RBD data for PD patients were taken from the Tracking Parkinson’s study (n=1046). For motor impairment previously published data were supplemented (n=87). PREDICT-PD pilot study participants were the controls. Smell, motor impairment and RBD were assessed using the University of Pennsylvania Smell Identification Test (UPSIT), the Bradykinesia-Akinesia Incoordination (BRAIN) test and the REM sleep behaviour disorder Screening Questionnaire (RBDSQ). UPSIT and RBDSQ data were analysed using logistic regression to determine which items were predictive of PD, or using total scores. Gaussian distributions were fitted to BRAIN test scores. LRs were calculated from logistic regression models or from score distributions. False-positive rates (FPRs) for specified detection rates (DRs) were calculated. Results Logistic regression modelling yielded a greater range of LRs. 16 odours were associated with PD; LRs ranged from 0.005-5511. 6 RBDSQ questions were associated with PD; LRs ranged from from 0.34-69. BRAIN test LRs ranged from 0.16-1311. For a 70% DR the FPR for the 16 odours was 2.4%, for 50% DRs, the BRAIN test FPR was 6.6% and 12.2% for the RBDSQ. Conclusions Maximising information on PD markers can potentially improve the ability of algorithms to detect PD by generating LRs with a larger range of values than using dichotomised results.