Appendix: A general regression model for analysis of independent maternal and paternal age effects for 47,+21 and other disorders that may arise from mutant gametes from either parent

Abstract

Biological considerations suggest that regression equations used to model the rate of mutational outcomes as a possible function of maternal age and paternal age (or other parental factors) are most appropriately additive models of the type: r=(h(x)+j(y) or r=(h(x)+(k(x)·j(y)), where r is the rate of the outcome event, x is maternal age, y is paternal age, and h, j and k are functions to be specified. The first, simpler model assumes that there is no independent maternal age effect upon formation of a gamete or zygote with a paternally derived mutation or upon survival of the consequent conceptus. The second more general model relaxes this assumption. These models appear preferable to those used previously, such as log r=(h(x)+j(y)) or equivalently r=exp (h(x)·j(y)), which posit complex relationships closer to a multiplicative interaction for which it is difficult to suggest obvious biological interpretations.