Use and misuse of statistics for epidemiological studies of multiple sclerosis

Abstract

Despite a vast literature on the epidemiology of multiple sclerosis, the cause of this disease remains elusive. The majority of the studies have focused on prevalence, so that a worldwide distribution of multiple sclerosis can now be mapped. Tantalizing apparent differences in prevalence have resulted in an abundance of hypotheses about risk factors for multiple sclerosis, none of which have been firmly established. This article reviews the difficulties inherent in the use of any type of descriptive study to investigate etiology of disease and provides an update of appropriate statistical methodology to analyze study results. The major focus is on the use of analytical study design, specifically the case-control study, to investigate a specific hypothesis about association between a risk factor and multiple sclerosis. An update on new epidemiological and statistical methodology to enhance the design of future case-control studies is presented. Future opportunities to study multiple sclerosis in new ways, such as in cohort studies or nested case-control studies, using large medical insurance and other types of health-related data bases, are discussed.