[Studies of association in complex diseases: statistical problems related to the analysis of genetic polymorphisms].

The positive results on the association between a genetic polymorphism and a specific complex disease in population-based association studies (e.g. case-control) are rarely replicated in independent studies. Furthermore, the design of these studies often lacks sufficient power to detect the association. Therefore, the "prevalence" of both type I and II errors in the scientific literature dealing with the study of complex diseases is extremely high. There are many potential problems that underlie this current, unfortunate situation such as the effect of population stratification, deficient correction for multiple tests or publication bias, among others. The present review deals with all these problems and provides guidelines that can help to minimize their effects.