We analyzed data from the National Center for Health Statistics’ (NCHS), United States Period Linked Birth/ Infant Death Datasets for 2003–2005 [18]. The datasets include a file of all infants less than 1 year of age who died during a given calendar year (from which we drew the numerators for our mortality rate calculations), and a file of all infants born in the calendar year in which the deaths occurred (from which we drew the denominators).

The potential study population consisted of all late preterm or early term live-born infants whose mothers resided in the United States at the time of their delivery. We defined late preterm as infants born at a gestational age recorded as 34, 35, or 36 weeks, and early term as infants born at a gestational age recorded as 37 or 38 weeks.

The NCHS edited gestational age reported in the public use data files, calculated primarily from the last menstrual period (LMP), was used for gestational age. Because LMP-based estimates of infant’s gestational age at birth are subject to error because of factors such as maternal recall error, post-conception bleeding, and delayed ovulation, this estimate was then compared to the clinical/obstetric estimate also available in the public use data files. If the two estimates differed by more than 2 weeks, then the clinical/ obstetric estimate was used [19]. California did not report a clinical/obstetric estimate, so no births/deaths from California were included in this analysis [20–22]. Also excluded from this study were infants with missing birth weight or birth weight less than 500 g and infants with missing clinical or obstetric estimate of gestational age.

Among the infants who died during infancy and met inclusion criteria for the study, we assessed differences in the frequency distributions of the maternal and infant characteristics (maternal race/ethnicity, age and parity and infant sex, plurality and birth weight) between infants born late preterm and infants born early term. Using Pearson χ² tests, we considered p values <0.05 to be indicative of statistical differences.

In our primary analysis, we calculated mortality rates by race/ethnicity, and maternal age only among singleton infants. However, because a higher proportion of infants born late preterm are from multiple gestations, we performed a sub-analysis comparing mortality rates among singleton infants with rates among twins. Because infants from multiple births involving more than one other sibling are at higher risk for infant death and likely to receive additional monitoring and medical intervention, we did not include them in this sub-analysis.

We compared infant mortality rates by maternal race/ ethnicity, maternal age and plurality using confidence intervals and z-tests according to NCHS methods [5, 14, 15]. In these comparisons, non-Hispanic white infants, infants born to mothers aged 25–29 years, and singletons were the reference populations.

We also identified the three leading causes of death among infants in each of the previously described demographic categories. The cause of death categories were based on the 130 Selected Causes of Infant Death adapted for use by the Division of Vital Statistics. The 130 Selected Causes of Infant Death is a specialized list of causes of death affecting infants, created by grouping and recoding the ICD-10 codes indicated as the underlying cause of death, from which 71 rankable causes of death can be derived [23].

All statistical analyses were conducted with SAS statistical software version 9.2 (SAS Institute, Cary, NC). This study was reviewed by the Emory University Institutional Review Board, which classified it as non-human subjects research and exempt from further review.

During 2003–2005, 874,532 US infants were born late preterm and 3,164,489 infants were born early term and met the inclusion criteria for this study: 7,092 of those infants born late preterm and 10,007 of those infants born early term died during infancy. No difference was found in the distribution of race/ethnicity or gender between the late preterm infants and early term infants who died during infancy with over 50 % born to non-Hispanic white mothers and slightly over 50 % male (Table 1). Significant differences existed in the categories of plurality and birth weight. Twins constituted 11.6 % of deaths among late preterm infants and 3.9 % among early term infants. Among late preterm infant deaths, 34.0 % weighed 2,000–2,499 g at birth and 39.7 % weighed ≥ 2,500 g, whereas among early term infant deaths 15.6 % weighed 2000–2499 g at birth and 77.0 % weighed ≥ 2,500 g.

Among infants born late preterm, the infant mortality rate per 1,000 live births was highest for infants born to American Indian/Alaskan Native mothers [(14.85, (6.75 during the neonatal period and 8.10 during the post-neonatal period))] (Table 2). All of these rates were significantly higher than corresponding rates among infants born to non-Hispanic white mothers (8.20, 4.43 and 3.77, respectively). The overall infant mortality rate (9.90) and post-neonatal mortality rate (5.39) among infants born to non-Hispanic black mothers were also significantly higher than the overall and post-neonatal rates among infants born to non-Hispanic white mothers. Only the post-neonatal mortality rate among infants born to Hispanic mothers was significantly lower than the corresponding rate for infants born to non-Hispanic white mothers.

Among early term infants, mortality rates were also significantly higher for infants born to American Indian/Alaskan Native [(5.69, 1.85 during the neonatal period and 3.84 during the post-neonatal period)] and black mothers (4.49, 1.36 and 3.13, respectively) than for infants born to white mothers (2.97,1.14 and 1.83, respectively) but the rates were only a third to a half as high as the rates for infants born late preterm (Table 3). Infants born to Asian/Pacific Islander mothers had significantly lower mortality rates (2.08, 0.91 and 1.17) than infants born to white mothers while Hispanic mothers had significantly lower infant (2.61) and post-neonatal (1.55) rates but not neonatal rate.

Among infants born late preterm the overall infant mortality rate and the post-neonatal mortality rate were highest for infants born to teenage mothers (11.88 and 6.77, respectively) and generally decreased with maternal age (Table 2). The neonatal mortality rates among infants born to teenage mothers (5.12) and infants born to mothers ≥35 years of age (5.23) were significantly higher than infants born to mothers aged 25–29 years (4.25). Among early term infants, all rates were significantly higher for infants born to mothers <24 years of age [<20 years: 4.82 overall (1.42 during the neonatal period and 3.40 during the post-neonatal period); 20–24 years: 4.22, 1.30 and 2.92, respectively] than for infants born to mothers 25–29 years of age (2.81, 1.11 and 1.70, respectively)], and the neonatal mortality rate was higher for infants born to mothers ≥35 years of age (1.26) than for infants born to mothers 25–29 years of age (Table 3).

Among infants born late preterm, all mortality rates were significantly lower for twins [(5.62 overall, (2.33 during the neonatal period and 3.29 during the post-neonatal period))] than for singletons (8.59, 4.53 and 4.06, respectively) (Table 2). However, among early term infants, both the overall infant mortality rate and the neonatal mortality rate were significantly higher for twins (3.67 and 1.43, respectively vs. 3.15 and 1.17 for singletons) (Table 3).

In all categories of infants born late preterm, congenital anomalies were the most common cause of death and sudden infant death syndrome (SIDS) was the second most common (Table 4). Accidents were the third most common cause of death in all categories except infants born to Asian/Pacific Islander mothers and infants born to mothers ≥29 years of age, among whom diseases of the circulatory system were the third most common cause. The third most common causes of death among infants born to American Indian/Alaskan Native mothers were accidents and flu as they were each responsible for the same number of deaths among those infants.

Among infants born early term, congenital anomalies were the most common cause of death and SIDS was the second most common in all categories of infants except infants born to teenage mothers, for whom these two causes were reversed (Table 5). Accidents were the third most common cause of death in all categories of infants except infants born to mothers ≥35 years of age, for whom accidents were tied with diseases of the circulatory system.