Sequential aspects of spontaneous abortion: maternal age, parity, and pregnancy compensation artifact

At least 3 hypotheses predict that spontaneous abortion risk differs during reporductive history: genuine maternal age effects change individual risks; genuine birth order effects change individual risks; and variant individual risks, which are really independent of both age and parity, produce an artifactual association of risk with age in populations of women. The availability of large numbers of reproductive histories recorded on magnetic tape by the Collaborative Study on Cerebral Palsy provides an opportunity to weigh these hypotheses. Information was gathered between mid-1959 and mid-1966 by 13 hospitals, mostly east of the Mississippi. Random samples of essentially all women registering in the obstetric clinics of the collaborating institutions entered the study. Generally, these women came from poorer urban areas. Data are taken from the interviews at 1st registrations only. At this time, women had prior reproductive histories of varying lengths. The data are analyzed to yield broad comparative evaluations of the maternal age, parity, and artifact hypotheses. When the logit transforms of abortion risks were regressed on maternal age, the linear component was positive and significant at the 1% level in every ethnic group. In all categories except blacks, the fit to such a simple model was quite adequate. Fit in the case of the blacks was disturbed by the high rate among 13-year olds and the low rate among 37-year olds. The 37-year old black sample was the only one to depart markedly from the trend of increased risk at high age. Primary analysis of birth order defects used Slater's (1962) rank order statistic on a group of histories. In every ethnic category the observed mean value of Slater's statistic exceeds its expected value of 0.5; every standardized deviation has a negligibly small probability when tested against the normal distribution. The conclusin is that spontaneous abortions tend to come late in a reproductive history. The white data showed a definite trend contrary to expectation under the pregnancy compensation hypothesis. Although not significant in the "o" (liveborn) versus "x" (abortion) contrast, the lowering trent in maternal age with prior abortion experience was signifixant for the longer histories. Equally surprising was the apparent positive finding in the black data. In sum, the data clearly showed that among women with histories mixing spontaneous abortions and live births, risk of abortion was greater at higher parity. Although the women sampled tended to be young, and increase of risk with age was demonstrated in the white sample. These effects were not because of sample biases. Black age effects were possibly confounded with pregnancy compensation artifact which can mimic aging influence in unselected samples.