Stillbirths in multiple births: test of independence

The stillbirth rate in twins is a more sensitive indicator of environmental hazards than the stillbirth rate in singletons. Medical care or other socioeconomic factors may be more influential for perinatal survival in twin than in single deliveries. Studies have indicated that stillbirths among children in a set of multiple maternities are not independent. Models were considered assuming independent outcomes within a set of multiple maternities. Analyses of the stillbirth rates confirm that the risk of stillbirth among males is almost constantly higher than among females. Any model introduced should assume different stillbirth rates for males and females. The models were tested with both maximum likelihood and minimum chi2 methods. Data was analyzed from Sweden, the Aland Islands, Saxony, England and Wales, and significant discrepancies obtained from the independence models. The same-sexed twin data contain both monozygotic and dizygotic twin sets with apparently different stillbirth rates. Consequently, for same-sexed twins the proposed model could be considered too simple. After improvement by splitting the same-sexed data into monozygotic and dizygotic twin sets, the dependence still remains. The proportion of both same-sexed and opposite-sexed twin pairs that contain two stillborn is greater than what the stillbirth rates and the independence should indicate. Consequently, stillbirth rate estimates based on the relative frequency of twin sets with two stillborn children have a positive bias. When the stillbirth rate decreases, the number of sets with two stillborn children decreases more slowly than would be indicated by independence.