Covariate-dependent genetic counseling in Huntington's disease.

From detailed study of a population-based sample of active cases of Huntington's disease it is known that the age of onset of Huntington's disease is strongly affected by whether Huntington's disease was inherited from mother or father. Standard methods for genetic counseling risk computations use age of unaffected consultant as a covariate affecting the quoted risk, but no other covariates are used in these risk computations. We demonstrate the feasibility of using a proportional hazards approach to computation of the penetrance function which in turn enables a risk to be computed incorporating other covariates besides age of consultant. When there is only one significant covariate other than age, direct estimation of age-of-onset distributions corresponding to various values of the covariate also gives useful results. Our analyses show that genetic counseling risks based on covariates are feasible to compute and offer additional accuracy to the recipient of the information.