Psychometric Properties of Clinical Indicators for Identification and Management of Advanced Parkinson's Disease: Real-World Evidence From G7 Countries.
Additional Document Info
INTRODUCTION: Standardized and validated criteria to define advanced Parkinson's disease (PD) or identify patient eligibility for device-aided therapy are needed. This study assessed the psychometric properties of clinical indicators of advanced PD and eligibility for device-aided therapy in a large population. METHODS: This retrospective analysis of the Adelphi Parkinson's Disease Specific Programme collected data from device-aided therapy-nave people with PD in G7 countries. We assessed the presence of 15 clinical indicators of advancing PD and seven indicators of eligibility for device-aided therapy in patients classified with advanced PD or as eligible for device-aided therapy by the treating physician. Accuracy was assessed using area under the curve (AUC) and multivariable logistic regression models. Construct validity was examined via known-group comparisons of disease severity and burden among patients with and without each clinical indicator. RESULTS: Of 4714 PD patients, 14.9% were classified with advanced PD and 17.5% as eligible for device-aided therapy by physician judgment. The presence of each clinical indicator was 1.9- to 7.3-fold more likely in patients classified with advanced PD. Similarly, the presence of device-aided therapy eligibility indicators was 1.8- to 5.5-fold more likely in patients considered eligible for device-aided therapy. All indicators demonstrated high clinical screening accuracy for identifying advanced PD (AUC range 0.84-0.89) and patients eligible for device-aided therapy (AUC range 0.73-0.80). The Unified Parkinson's Disease Rating Scale (UPDRS) score, cognitive function, quality of life, and caregiver burden were significantly worse in indicator-positive patients. CONCLUSION: Specific clinical indicators of advanced PD and eligibility for device-aided therapy demonstrated excellent psychometric properties in a large sample, and thus may provide an objective and reliable approach for patient identification and treatment optimization.