Confounding in epidemiological studies on assessment of the impact of genetic factors on disease risk: the problem of redundant adjustment.