importance of the outcome but also because of the implications for the more voluminous literature on fetal growth and preterm birth. Although this study did not link GWG to those intermediate outcomes or intermediate outcomes to mortality, the strength of the patterns and their parallels with studies of fetal growth add credibility to the presumption that a causal chain from GWG to adverse birth outcomes to death is operative. Based on a limited volume of research, but one well-done study, the committee considered the evidence for a link to infant mortality to be moderate.
The relationship of GWG to fetal growth was considered in some detail in the IOM (1990) report. The association was deemed worthy of lengthy consideration because, as noted in IOM (1990) and by others, smaller size at birth is associated with increased fetal and infant mortality, cerebral palsy, hypoglycemia, hypocalcemia, polycythemia and birth asphyxia, persistent deficits in size, and persistent deficits in neurocognitive performance (Pryor et al., 1995; Goldenberg et al., 1998). Adverse health outcomes associated with small size at birth tend to follow a dose-response relationship with elevated relative risks at the lowest weights. Large size causes delivery complications, including shoulder dystocia and other forms of birth injury, as well as cesarean delivery, maternal death, and fistulae (IOM, 1990). Birth weight is a function of both duration of gestation and rate of fetal growth, so that studies using birth weight alone as a health outcome are less informative than those that distinguish between these processes. In order to isolate fetal growth rate from duration of gestation, studies often use SGA and LGA instead of birth weight as measures of fetal growth rather than birth weight. SGA and LGA are indicators that compare an infant’s weight to the distribution of birth weight of all infants born in the same week of gestation. Most commonly, infants in the lowest and highest 10th percentiles of birth weight for gestational age are classified as SGA and LGA, respectively, although some researchers use the more extreme values of the 5th percentile or two standard deviations or more below or above the mean. Some researchers use percentile cutoff points that are specific to gender, race/ethnicity, and/or parity in addition to week of gestation, although there is some controversy about the use of racial/ethnic-specific norms, in particular because their biological meaning is in doubt. Even though black infants in the United States have a markedly different weight distribution than non-black infants (of varying race/ethnicity), with deviation from group-specific norms being very informative for predicting mortality, separate group-specific norms could be interpreted as acceptance of differences in birth outcome by race/ethnicity as absolute. Such differences are not