Analysis of multiple birth rates in Japan. 1. Secular trend, maternal age effect, and geographical variation in twinning rates.

Mean twin, triplet, and quadruplet birth rates in Japan from 1951 to 1968 were 6.41, 0.056, and 0.00094 per 1,000, respectively. In 1974 the corresponding figures were 5.83, 0.059, and 0.00329. No quintuplets were born in the former period, but a set was born in the latter year, the rate being 0.47 per million. From 1955 to 1966 the MZ twinning rate increased slightly, but decreased thereafter. This increase was limited to live-born MZ twins, particularly in the higher maternal age groups. The DZ twinning rate declined in the entire period, particularly in higher maternal age groups. This decline appeared to be essentially limited to fetal deaths. Among live births the MZ twinning rate underwent a nearly linear increase with maternal age, whereas the DZ twinning rate attained a mode in the maternal age group 35--39 years. The MZ and the DZ twinning rates among fetal deaths by maternal age had unimodal distributions with modes in maternal age groups 25--29 and 30--34 years, respectively. As to the DZ twinning rate, a geographical cline was noted, with a high rate in the northeast of Japan; the rate was positively correlated with latitude, which also positively correlated with the presence of multiple births among relatives. A negative but nonsignificant correlation was seen between the DZ twinning rate and the proportion of mothers treated with ovulation-inducing hormone. The proportion was higher in mothers of unlike-sexed twins than in those of like-sexed twins and in mothers of triplets than in those of twins. An association between DZ twinning rate and age-specific fertility per married woman is suggested to exist among higher maternal age groups in the northeast part of Japan in earlier years.