Comparative and Incremental Validity of NMSQuest, NMSS, and MDS-NMS With External Validation of a Conversion Equation in Parkinson’s Disease

Abstract

Non-motor symptoms contribute to disability in Parkinson’s disease (PD), but instruments differ in format, content, and scoring. This study compared the Non-Motor Symptoms Questionnaire (NMSQuest), Non-Motor Symptoms Scale (NMSS), and Movement Disorder Society Non-Motor Rating Scale (MDS-NMS), externally evaluated the published NMSS-to-MDS-NMS conversion equation, and assessed incremental validity after reducing direct content overlap. Seventy-eight patients with idiopathic PD evaluated at a movement disorders clinic between January and March 2026 completed the NMSQuest, NMSS, MDS-NMS, Hospital Anxiety and Depression Scale, Parkinson’s Disease Sleep Scale-2, and MDS-Unified Parkinson’s Disease Rating Scale Part III. Associations were examined using Spearman correlations with bootstrap confidence intervals. External validation included calibration, predictive R², error measures, Lin’s concordance correlation coefficient, and Bland–Altman analysis. Hierarchical regression was used to assess incremental validity. The three instruments were moderately correlated (ρ = 0.506–0.528), indicating related but non-equivalent measurement of non-motor burden. The published equation underestimated MDS-NMS scores by 10.1 points (95% CI, 2.9–17.5). The calibration slope was 0.662, predictive R² was 0.121, and mean absolute error was 27.8; limits of agreement were wide, with significant proportional bias. Recalibration improved apparent performance but remained exploratory. In analyses using domain-excluded NMSS and MDS-NMS scores, the two clinician-rated scales contributed comparably to anxiety, whereas only the MDS-NMS contributed independently to sleep disturbance. No scale showed an independent association with depression. The instruments captured related but non-equivalent aspects of non-motor burden. The conversion equation showed limited transportability for patient-level score conversion and requires further multicenter external validation.