Registration Artifacts in International Comparisons of Infant Mortality

Abstract

Although nearly all countries have noted a marked drop in infant mortality in recent decades, there are still large disparities between countries, particularly between developed and developing nations but also among industrialized countries. Nations employ different gestational age and/or birth-weight cutoff points for reporting fetal deaths and live births. Interpretations of signs of life, which classify deaths as fetal or infant, may vary widely despite attempts at standardization by the World Health Organization. There also may be differences in registering tiny preterm infants as live births or fetal deaths. The authors analyzed data on six national groups from the Internal Collaborative Effort on Perinatal and Infant Mortality (whites and blacks in the United States, Israeli Jews and non-Jews, Norwegians, and Swedes) for the years 1987 and 1988. In particular, two major potential artifacts that might help explain intercountry differences in infant mortality were examined: classification as fetal or infant death (especially at very low birth weights) and underregistration of borderline-viable infants as either fetal deaths or live births. Crude infant mortality rates (infant deaths per 1000 live births) were lowest in Sweden, where they were 25% below those for Norway, US whites, and Israeli Jews. Rates were twice as high for US blacks and Israeli non-Jews. Mortality risk was reduced in both US groups and Israeli non-Jews after adjusting for maternal age, parity, and multiple births. Excluding Sweden (because of insufficient data), stillbirths of 20 weeks' or more gestation as a proportion of all perinatal deaths were highest in Norway and lowest in US blacks. Differences were most marked when birth weight was less than 500 g. Live births of less than 500 g varied more than 50-fold between Sweden and both Israeli groups at the low end and US blacks at the high end. US whites had a 15-fold higher proportion of live births less than 500 g than did Swedes. Rates for live births at 500 to 749 g varied 7-fold among the various groups, but again were lowest in Sweden and highest in US blacks. Live births less than 750 g were much likelier to be registered in the United States than in the other countries, and this was not simply a reflection of differential classification of live births and stillbirths. Excluding births less than 750 g, the relative risk of infant mortality was somewhat reduced for Swedes, Israeli Jews and non-Jews, and US whites, and it was markedly reduced for US blacks. Relative risk figures for both Israeli groups and both US groups (especially US blacks) declined after adjusting for maternal age, parity, and multiple gestation. These findings confirm the existence of marked differences in registering infants near borderline viability as well as differences in classifying fetal versus infant deaths. Information updating these figures, now more than a decade old, is needed. In addition, more in-depth studies are required to ascertain the reasons for these differences, whether cultural, religious, or economic.