Low clinical sensitivity and unexpectedly high incidence for neuropathologically diagnosed progressive supranuclear palsy.

The objective of this study was to determine the prevalence, incidence, and clinical diagnostic accuracy for neuropathologically diagnosed progressive supranuclear palsy (PSP) with data from a longitudinal clinicopathological study using Rainwater criteria to define neuropathological PSP. Of 954 autopsy cases, 101 met Rainwater criteria for the neuropathologic diagnosis of PSP. Of these, 87 were termed clinicopathological PSP as they also had either dementia or parkinsonism or both. The prevalence of clinicopathologically defined PSP subjects in the entire autopsy dataset was 9.1%, while the incidence rate was estimated at 780 per 100 000 persons per year, roughly 50-fold greater than most previous clinically determined PSP incidence estimates. A clinical diagnosis of PSP was 99.6% specific but only 9.2% sensitive based on first examination, and 99.3% specific and 20.7% sensitive based on the final clinical exam. Of the clinicopathologically defined PSP cases, 35/87 (∼40%) had no form of parkinsonism at first assessment, while this decreased to 18/83 (21.7%) at final assessment. Our study confirms a high specificity but low sensitivity for the clinical diagnosis of PSP. The low clinical sensitivity for PSP is likely primarily responsible for previous underestimates of the PSP population incidence rate.