Prospective Predictive Accuracy Versus Previous-Fall Discriminative Validity of Clinical Assessment Tools for Fall Risk in Parkinson’s Disease: A Systematic Review and Meta-Analysis

Background/Objectives: Many clinical tools are used to assess fall risk in Parkinson’s disease (PD). We examined whether tools that discriminate people with a history of falls also predict future falls. Methods: Six electronic databases were searched through January 2025, and the search was later updated. Hedges’ g and the area under the receiver operating characteristic curve (AUC) were pooled separately for previous-fall and prospective designs using three-level models with cluster-robust variance estimation and one estimate per cohort per clinical domain or instrument. Results: Forty-five reports were included; 42 reports from 36 independent cohorts contributed to the meta-analyses. Fallers performed worse than non-fallers in previous-fall (g = −0.82; 95% confidence interval [CI] −1.03 to −0.62; 20 cohorts) and prospective designs (g = −0.61; 95% CI −0.75 to −0.48; 14 cohorts), most markedly on balance and fall-confidence measures. The pooled AUC was 0.80 (95% CI 0.74–0.85) for previous falls and 0.70 (95% CI 0.63–0.76) for future falls, and prospective accuracy remained lower within the balance domain. Prospective AUCs of individual domains and instruments ranged from 0.60 to 0.78, and the certainty of evidence was low to moderate. Conclusions: Discrimination of fall history overstates prospective predictive accuracy. Clinical tools characterize fall-related impairments but should not be used as stand-alone predictors of future falls in PD.