Abstract
Abstract
Background
The sum of clinician‐rated MDS‐UPDRS Part III Motor Examination items is widely used in Parkinson's disease (PD), but item‐ or subset prognostic value remains unclear. Whereas reduced subsets of patient‐reported Parts IB + II show improved utility for predicting time to PD medication initiation, the validity of similar Part III subsets is unknown.
Objectives
Determine whether a subset of MDS‐UPDRS Part III items is superior in prognostic value to the full scale for time to PD medication initiation in early disease.
Methods
Harmonized longitudinal data from six early PD cohorts, with time to PD medication initiation as the primary outcome, were analyzed using a longitudinal item response theory model to rank items, combining discrimination and information. Rank‐driven cumulative subsets were formed to analyze the primary outcome using time‐varying Cox models under clinically relevant follow‐up windows. Subset performance was assessed using the C‐index and compared with the full‐item model using Wald tests. Successful item reduction required stable, significant improvement across all follow‐up windows.
Results
No item subset met the predefined criteria for stable and superior prognostic performance relative to the full Part III scale. Best reduced‐subset versus full‐model C‐indices were similar and non‐significant: 0.582 versus 0.581 under full follow‐up, 0.602 versus 0.602 at 1 year, and 0.586 versus 0.585 at 2 years.
Conclusions
Unlike Parts IB and II, no sensitivity‐driven reduced subset significantly and consistently outperformed the full MDS‐UPDRS Part III in predicting time to PD medication initiation in early PD, not supporting a reduction of the 33‐item scale in this setting.