Evolving methods in genetic epidemiology. I. Analysis of genetic and environmental factors in family studies.

Using family data (either sibships, nuclear families, or extended pedigrees) to test models of inheritance is the essence of human genetics. The rigor and precision of this testing process can range from a strictly observational process (eliminating simple single-locus models one at a time on individual pedigrees) through formal tests of hypotheses about the expected proportion of affected sibs up to maximum likelihood estimation of parameters in a generalized model of inheritance. The goal of this process is to use family data to identify Mendelian mechanisms. If a phenotype (either a quantitative or a qualitative trait) is directly controlled by gene(s), the phenotypic distribution in families (and ultimately in populations) should follow the predictions of Mendelian genetics. Unfortunately, the information content of family data from observational studies of humans is limited, and the statistical tools available have important limitations. As the observed phenotype becomes further and further removed from the direct action of genes, however, its distribution in families will be more influenced by environmental factors and is less likely to follow predictable Mendelian patterns. Such phenotypes are traditionally called "multifactorial" because both genetic and environmental factors control the trait (n.b., the term "complex" disease is often used as a synonym). Many important diseases fall under this multifactorial label, including most of the major chronic diseases (cardiovascular disease, cancer, diabetes, some birth defects, etc.). Epidemiologic studies of most, if not all, of these multifactorial diseases have identified one or more observable environmental fac-