We sought to validate an instrument and two application procedures for screening for prevalence of parkinsonism in conjunction with a door-to-door survey in Cantalejo, Spain. Residents in Cantalejo aged 40 years or older were assigned to three groups specifically designed to optimise case-finding and cost/efficiency. A nine-item questionnaire aimed at identifying parkinsonism-related symptoms was administered and collected door-to-door by laymen and repeated by specialists at a medical facility before neurological examination. Diagnoses were then established and confirmed after a 3-year follow-up. Different indices for and concordance of answers were studied for both applications in groups of 25 individuals with parkinsonism. The scoring procedure was updated using the prevalence community sample of parkinsonism. Whereas the application administered by laymen showed, using the same cut-off scores, a higher sensitivity for parkinsonism than that administered by trained personnel, it yielded a lower positive predictive value for parkinsonism diagnosed during the study, i.e., at a 30-point cut-off (96% vs. 80% and 15% vs. 76%, respectively). Concordance was > or =64% with kappa >or=0.316 and P <or=0.031 for six questions. Instrument sensitivity, determined by different cut-off scores, had a high impact on PD detection, observed M/F ratio, severity, and mean age at onset. Choice of screening method or application can considerably affect prevalence measurements and disease natural history parameters described in PD surveys.
Copyright 2002 Movement Disorder Society