Predicting Fall Risk in Parkinson’s Disease with Clinical Measures

PREDICTING falls in Parkinson’s disease (PD) with routine clinical measures has shown moderate promise, according to a study that developed and externally validated models of next-year risk for first and recurrent falls. 

Predicting Falls in Parkinson’s Disease 

Falls remain a major concern in PD, making early identification of at-risk patients a priority. In this study, models to predict fall risk were developed in the Parkinson’s Progression Markers Initiative (PPMI) and validated externally in University of Pennsylvania (UPenn) cohorts. Incident falls were evaluated among visits without current or prior falls, and recurrent falls among visits with current falls. 

The incident model comprised the Movement Disorder Society–Unified Parkinson’s Disease Rating Scale Part III, Montreal Cognitive Assessment and Geriatric Depression Scale, and was developed in 773 visit-pairs and validated in 589. 

The recurrent model comprised Hoehn and Yahr stage, the Geriatric Depression Scale and Symbol Digit Modalities Test, and was developed in 302 visit-pairs and validated in 137.  

Discrimination was assessed using the area under the receiver operating characteristic curve (AUROC), the area under the precision-recall curve and development-derived risk strata; exploratory analyses added imaging and fluid biomarkers. 

How Accurate Were the Predictions? 

The incident model achieved moderate discrimination in PPMI (AUROC: 0.704) and UPenn (AUROC: 0.630), and in the UPenn cohorts, incident fall risk rose from 7.1% to 18.5% across low- to high-risk groups. 

Similarly, the recurrent model achieved moderate discrimination in PPMI (AUROC: 0.683) and UPenn (AUROC: 0.641), with recurrent fall rates rising from 22.2% to 65.0% across risk groups. 

Additionally, biomarker augmentation modestly improved discrimination in PPMI but not in external validation, where estimates were imprecise. 

Next Steps for Risk Stratification 

The findings show that predictive models using clinical measures can distinguish yearly fall risk groups in PD. Motor severity, disease stage, cognition, and depressive symptoms have emerged as the most informative predictors. These models offer a framework for evaluating risk stratification, including in trial design, while biomarker findings warrant larger cohorts. 

Reference 

Vizcarra JA et al. Prediction of first and recurrent falls in Parkinson’s disease. Annals of Neurology. 2026;DOI:10.1002/ana.78400.