The most common figure I see is 30%, and the AAN conservatively says 20%. This breakdown referenced by Medscape is probably more accurate but still unacceptable, imo.

The diagnosis of PD can be challenging when based solely on clinical history and examination. A review of the literature has revealed misdiagnoses in 47% of cases diagnosed by general physicians in the community, in 25% of cases diagnosed by a general neurologist, and in 6%-8% of cases diagnosed by a neurologist specializing in movement disorders.[1]

The most current data is from 2006. No wonder our trials typically cannot meet endpoints.