Abstract
BACKGROUND: Estimates of the incidence and prevalence of chronic diseases can be made using established cohort studies but these estimates may have lower reliability if based purely on self-reported diagnosis. METHODS: The MRC Cognitive Function & Ageing Study (MRC CFAS) has collected longitudinal data from a population-based random sample of 13004 individuals over the age of 65 years from 5 centres within the UK. Participants were asked at baseline and after a two-year follow-up whether they had received a diagnosis of Parkinson's disease. Our aim was to make estimates of the incidence and prevalence of PD using self-reporting, and then investigate the validity of self-reported diagnosis using other data sources where available, namely death certification and neuropathological examination. RESULTS: The self-reported prevalence of Parkinson's disease (PD) amongst these individuals increases with age from 0.7% (95%CI 0.5-0.9) for 65-75, 1.4% (95%CI 1.0-1.7) for 75-85, and 1.6% (95%CI 1.0-2.3) for 85+ age groups respectively. The overall incidence of self reported PD in this cohort was 200/100,000 per year (95%CI 144-278). Only 40% of the deceased individuals reporting prevalent PD and 35% of those reporting incident PD had diagnoses of PD recorded on their death certificates. Neuropathological examination of individuals reporting PD also showed typical PD changes in only 40%, with the remainder showing basal ganglia pathologies causing parkinsonism rather than true PD pathology. CONCLUSION: Self-reporting of PD status may be used as a screening tool to identify patients for epidemiological study, but inevitably identifies a heterogeneous group of movement disorders patients. Within this group, age, male sex, a family history of PD and reduced cigarette smoking appear to act as independent risk factors for self-reported PD.