for those comparison groups. This may lead to the inaccurate determination of cases of preterm birth and a biased establishment of risk characteristics and high-risk areas. As indicated earlier, this represents a major concern for international comparisons of preterm birth rates.
Epidemiological studies of preterm birth in large populations, which often use vital records, typically rely on LMP or, more recently, LMP and the clinical estimate as reported on the birth certificate to define gestational age. It is these studies that have typically established current national trends and international comparisons in preterm birth rates. Meanwhile, clinical studies may more typically have access to early ultrasound data, although the study populations selected may be less representative of the larger population at risk of preterm birth. These measurement issues hinder comparisons among study findings, limit the interpretation and generalizability of the results, and persist as an ever lurking potential bias to research on preterm birth.
During the last 2 decades of the 20th century, preterm birth rates in the United States exhibited a steady increase. As depicted in Figure B-2, an approximately 30 percent increase was observed for both the preterm birth and the very preterm birth rates between 1980 and 2000.
Furthermore, examination of the gestational age distribution during the latter part of this period reveals a slight decrease in mean gestational age from 39.2 weeks for 1985 to 1988 to 38.8 weeks for 1995 to 2000. Additionally, there is an overall shift in the distribution, resulting in a great proportion of preterm births and a decrease in postterm births (42-plus weeks of gestation). These patterns are illustrated in Figure B-3, which displays the gestational age distributions of live births to U.S. resident mothers, using data from the NCHS linked live birth infant death cohort files.
For the two time periods portrayed in Figure B-3, Table B-4 provides the proportion of births for the various preterm birth categories. Between 1985–1988 and 1995–2000, the proportion of preterm birth rose approximately 15 percent.
These increasing trends in preterm delivery have not been consistent among racial groups in the United States. Figure B-4 provides trends in the percentages of preterm and very preterm births for whites and African Americans on the basis of the reported race of the mother. Although a steady increase in these rates is evident for whites, this temporal pattern is not evident for African Americans. This divergence in trends in the rates of preterm delivery has been the subject of investigation (8, 51). Differential