Rates of Down's syndrome at the upper extreme of maternal age--absence of a "leveling" effect and evidence for artifacts resulting from analyses of rates by five-year maternal age intervals.

A "leveling" in the first order exponential increase of the rate of Down's syndrome with maternal age over 40 years of age has been suggested by others in review of data reported by five-year maternal age interval. In contrast, data reviewed here analyzed by one-year interval reveal no evidence for such an effect when the observed rates (O) at each maternal age in the 40--49-year age range and in the 45--49-year age range are compared with those predicted (P) from first order exponential equations derived from data in the 33--44-year age range. The sign of O -- P is not predominantly negative as would be expected if there were a "leveling" effect. Part of the evidence for leveling in previous studies may derive from the fact that rates have been plotted at the midpoints of the intervals rather than at the true mean maternal ages of mothers having births in these five-year intervals. It is shown that in the three data sources considered here, the difference between the quin-quennial mean maternal age and the interval midpoint is greater at older ages than at younger ages. Thus, plotting rates at the midpoint of five-year intervals results in significant graphic underestimate of the rates, and may result in a slight leveling effect. It is suggested that selective underascertainment of births to those in the older maternal ages in previous studies may also have contributed to the leveling effect observed earlier.