On the inadequacy of quinquennial data for analyzing the paternal age effect on Down's dyndrome rates

Abstract

Investigations of the influence of paternal age on the rate of Down's syndrome are complicated by the high correlation between parental ages and the strong dependence of the incidence rate upon maternal age. Two possible approaches to isolating an independent paternal age effect are shown to lead to erroncous results if based on data by quinquennial age intervals rather than by single-year intervals. For a multiple regression method the discrepancy can be removed by using the mean maternal and mean paternal age within each quinquennial cell. Failure to do so results in an artifactual paternal age effect.