Drug tracer estimates of Parkinson's disease prevalence in British Columbia (Canada): Validation of specificity

Introduction: Drug tracer methodology employs prescription data to estimate disease prevalence. Estimates must be corrected for the proportion of patients without the disease who are on the drugs, and for the proportion of patients with the disease who are untreated. We present estimates of Parkinson's disease (PD) prevalence in British Columbia (Canada) and provide an empirical estimate of the specificity of the drug tracer method used. Methods: The British Columbia PharmaCare database was used to identify patients receiving anti-PD prescriptions from 1996 and 2002. A sample of these patients was screened for self-reported PD, later confirmed with clinical assessment. Patients reporting no-PD were asked about other chronic conditions. We compiled the reported chronic conditions, cross-referencing with drug use, to estimate the specificity of the gross anti-PD prescription categorization. These estimates were combined with published data to correct the initial estimates of PD prevalence. Results: Preliminary findings estimated prevalence to be between 100 and 150 per 100 000, after correction for specificity derived from a previous study in the 1980s. In the current validation, of a sample of approximately 1000 patients receiving anti-PD medication, approximately 25% of these patients reported no PD. Other conditions commonly reported with anti-PD drug use included restless leg syndrome (12% of anti-PD prescriptions), and pituitary conditions (4% of anti-PD prescriptions). Conclusions: A prescription database can be used to estimate disease prevalence with corrections applied as anti-PD treatments are also used for the treatment of conditions such as restless leg syndrome and pituitary conditions.