[Ordinal logistic regression in epidemiological studies].