What features improve the accuracy of the clinical diagnosis of progressive supranuclear palsy-parkinsonism (PSP-P)?

Progressive supranuclear palsy-parkinsonism (PSP-P) is a primary tauopathy characterised by neurofibrillary degeneration, which is frequently mistaken for Parkinson's disease (PD), multiple system atrophy (MSA), and vascular parkinsonism (VP) at presentation. The aim of this study was to identify particular clinical features (green flags) that may be helpful in differentiating PSP-P from these other disorders. We identified 37 patients with PSP-P from 726 patients archived at the Queen Square Brain Bank. Using a retrospective case notes review the clinical features were compared between the PSP-P group and Lewy body associated parkinsonism (PD, n = 444 and dementia with Lewy bodies (DLB), n = 46), MSA (n = 90), and VP (n = 19), using the chi(2)-test for proportions for a two-by-two contingency table. The sensitivity, specificity, and positive predictive values (PPV) and negative predictive values (NPV) were calculated for individual clinical features. A specificity of >0.85 or a PPV of >0.85 were considered reliable discriminators. No clinical features were predictive of PSP-P, but late drug induced dyskinesias (specificity 0.92, PPV 0.99), late autonomic dysfunction (specificity 0.94, PPV 0.99) and any visual hallucinations (specificity 0.94, PPV 0.99) were better in distinguishing PD and PSP-P than predicted using operational diagnostic criteria for PD. PSP-P shares many clinical features with PD and DLB, MSA and VP, but visual hallucinations, drug induced dyskinesias and autonomic dysfunction are very uncommon and may be helpful exclusion criteria.