The place of child-spacing as a factor in infant mortality: a recursive model.

This study investigates the role of a preceding birth interval in infant mortality by considering a causal ordering of seven variables. Data are from the 1974 World Fertility Survey for Korea and cover the survival of 6,161 index children. A two-stage logit model was used. The analysis suggests that infant mortality is directly influenced by the preceding birth interval which, in turn, is influenced by five other explanatory variables: maternal age, birth order, immediately preceding infant's death, education of mother, and place of residence. Maternal age and prior infant death also exert direct effects on mortality. In terms of relative risk, prior infant death has the strongest effect of all the explanatory variables. A longer birth interval increases the odds of an infant's survival by 25 per cent, whereas the death of a preceding child decreases the odds by 45 per cent. However, infant deaths in Korea occur infrequently in comparison with short birth intervals. Thus, the two factors present comparable attributable risks in unadjusted measurements. The first-stage causal structure affecting a birth interval is more complicated than the second-stage structure affecting infant mortality.