Estimation of incidence and prevalence of Parkinson's disease in the elderly using pharmacy records.

Abstract

We determined the prevalence and incidence of Parkinson's disease among persons aged 55 years and older in pharmacy records. Data came from the PHARMO database which includes information on drug dispensing for all residents of six Dutch cities. We selected all persons aged 55 years and older who had used antiparkinsonian drugs, and calculated a chance for having Parkinson's disease with use of a previously validated logistic regression model. We used a cut-off of 0.5 (sensitivity 62%, positive predictive value 92%). Prevalence was estimated on the first Wednesday of October 1997, incidence on the first Wednesday of October 1993 until 1997. A patient was incident, if the first prescription for an anti-parkinsonian drug was dispensed at least 180 days after entry into PHARMO. Prevalence and incidence were standardized to the Netherlands population of 1 January 1998. The prevalence was adjusted for the sensitivity and positive predictive value of the model. The unadjusted prevalence (per 100,000) for those aged 55-64 years was 111, 65-74 years 598, 75-84 years 1551 and for persons aged 85 years and older 1847. The adjusted and standardized prevalence was 970 per 100,000 (95% confidence interval 869 to 1071). The incidence (per 100,000 person years) for persons aged 55-64 years was 12, 65-74 years 108, 75-84 years 257 and for persons aged 85 years and older 247. The standardized incidence was 109 per 100,000 person years (96 to 121). Prevalence and incidence were in range with the literature. Pharmacy records therefore seem to be a useful tool for continuous monitoring of incidence and prevalence of Parkinson's disease.