Studies on the prevalence of paralysis agitans by tracer methodology.

Abstract

The present study was motivated by the methodological issues revealed by a review of previous studies on Paralysis Agitans (PA) epidemiology. The study was given the following aims: To identify an easy available, inexpensive marker of PA frequency in populations; to develop, standardize and validate a model for measurement of PA prevalence by using that tracer; to study the geographical distribution of PA in Sweden, and to use the model in environmental risk analysis (ERA) for PA. Information on diagnoses and anti-parkinsonian drugs (APD) utilization from Swedish registers was analyzed, in order to select the group of APD which would best fulfill the prerequisites of specificity and preference of choice, and combined with prevalence data from direct surveys. Levodopa drugs were found to be the most suitable tracer for PA prevalence. A mathematical model was formulated to generate annual estimates for average age specific prevalences from total sales of levodopa, age specific population and figures for infant mortality rates. The following was required in order for the model to yield accurate estimates: high quality of the information to be used, good availability of levodopa drugs, a minimal size of the studied population of 100,000, an IMR lower than 28.8 per 1000 and that levodopa diffusion had reached the late adoption phase. Underdiagnosis was found to constitute an important cause of bias in classical surveys. Standardization for age and health services effectiveness by using infant mortality rates was shown to improve comparability of prevalence figures. The model was validated and showed internal consistency for age specific data on levodopa sales and on prevalence from Sweden and six Chinese towns, respectively. Estimates for prevalence for Iceland and Sardinia fitted historical data from direct surveys. The results of the ERA study showed that a high latitude and a low population density in 1900-20 were correlated with an increase in PA prevalence as measured by surveys or estimated by the use of the tracer method. Further analysis indicated that several so-called risk factors or concomitant diseases are associated with low population density. The geographical distribution of standardized PA prevalences in Swedish counties was studied for the period 1977-84. The findings were in accordance with the figures expected from the natural characteristics of the disease as determined by latitude and population density except in the county of Gävleborg which showed a significantly higher, and continuously increasing prevalence.(ABSTRACT TRUNCATED AT 400 WORDS)