Bayesian mapping of multiple sclerosis prevalence in the province of Pavia, northern Italy

• Published in: Journal of the neurological sciences Vol. 244; no. 1; pp. 127 - 131
• Main Authors: Bergamaschi, R., Montomoli, C., Candeloro, E., Monti, M.C., Cioccale, R., Bernardinelli, L., Fratino, P., Cosi, V.
• Format: Journal Article
• Language: English
• Published: Shannon Elsevier B.V 15-05-2006; Elsevier Science
• Subjects: Adolescent; Adult; Aged; Bayes Theorem; Bayesian mapping; Biological and medical sciences; Cluster Analysis; Cross-Sectional Studies; Epidemiologic Methods; Female; Humans; Immunomodulators; Italy; Italy - epidemiology; Male; Medical sciences; Middle Aged; Multiple sclerosis; Multiple Sclerosis - epidemiology; Multiple sclerosis and variants. Guillain barré syndrome and other inflammatory polyneuropathies. Leukoencephalitis; Neurology; Pharmacology. Drug treatments; Prevalence; Italy; Europe; Bayesian mapping; Italy; Multiple sclerosis; Prevalence; Cartography; Nervous system diseases; Central nervous system disease; Inflammatory disease
• ISSN: 0022-510X; 1878-5883
• DOI: 10.1016/j.jns.2006.01.013

The geographical analysis of a disease risk is particularly difficult when the disease is non-frequent and the area units are small. The practical use of the Bayesian modelling, instead of the classical frequentist one, is applied to study the geographical variation of multiple sclerosis (MS) across the province of Pavia, Northern Italy.
464 MS-affected individuals resident in the province of Pavia were identified on December 31st 2000. The overall prevalence was 94 per 100,000 inhabitants. This estimate indicates an increasing MS prevalence in the province, in accordance with the vast majority of the Italian areas where prevalence studies have been repeated.
We mapped the geographical variation of MS prevalence across the 190 communes of the province both with a classical approach and a Bayesian approach. The frequentist approach produced an extremely dishomogeneous map, while the Bayesian map was much smoother and more interpretable.
Our study underlines the usefulness of Bayesian methods to obtain reliable maps of disease prevalence and to identify possible clusters of disease where to carry out further epidemiological investigations.