Does Socioeconomic Status Modify the Association Between Preterm Birth and Children’s Early Cognitive Ability and Kindergarten Academic Achievement in the United States?

Does Socioeconomic Status Modify the Association Between Preterm Birth and Children’s Early... Abstract Being born preterm and being raised in poverty are each linked with adverse cognitive outcomes. Using data from 5,250 singletons born in the United States in 2001 and enrolled in the Early Childhood Longitudinal Study, Birth Cohort, we examined whether household socioeconomic status (SES) modified the association between preterm birth (PTB) and children’s scores on cognitive assessments at age 2 years and reading and mathematics assessments at kindergarten age. Gestational age was measured from birth certificates and categorized as early preterm, moderate preterm, late preterm, early term, and term. SES was measured at age 9 months using a composite of parental education, occupation, and income. PTB was associated with 0.1- to 0.6-standard deviation-deficits in 2-year cognitive ability and kindergarten mathematics scores and with 0.1- to 0.4-standard-deviation deficits in kindergarten reading scores. Children living in the lowest (versus highest) SES quintile scored 0.6 standard deviations lower on 2-year cognitive ability, 1.1 standard deviations lower on kindergarten reading, and 0.9 standard deviations lower on kindergarten mathematics. The association between PTB and cognitive outcomes did not differ by postnatal SES. However, children who were both born preterm and lived in lower-SES households had the poorest performance on all 3 outcomes and therefore may represent a uniquely high-risk group. cognitive development, interaction, poverty, preterm birth Being born preterm and being raised in poverty are each linked with adverse cognitive outcomes (1, 2). Children’s early cognitive ability and being ready to enter school are important predictors of later social and academic achievement as well as socioeconomic and health trajectories over the life course (3, 4). In the United States, both of these factors are important public health concerns, with 10% of children born preterm and half residing in low-income households (5, 6). Preterm children may be born with developmentally immature brains, and they face higher risk of central nervous system and brain injuries (7). In the first months and years of children’s development, sensory enrichment and deprivation have important influences on brain growth and development as children are experiencing, responding to, and learning from stimuli in their environments (8). Children raised in families of lower socioeconomic status (SES) may be more likely to experience stressors such as material deprivation and to have fewer cognitively stimulating experiences (9). Preterm children raised in more advantaged environments have been observed to have better developmental outcomes compared with preterm children in more disadvantaged environments, suggesting that growing up in a more advantaged environment may attenuate the adverse impacts of preterm birth (10, 11). Studies of cognitive outcomes of preterm birth often treat SES as a confounding factor because it is associated with risk of being born preterm (12) and with children’s developmental outcomes (2, 9). It is plausible that SES also acts as a modifier of the effect of preterm birth on children’s cognitive development, but only a few epidemiologic studies have tested this hypothesis. In studies assessing additive interaction between preterm birth and SES, investigators have found either that lower childhood SES exacerbated the association between preterm birth and worse cognitive or school outcomes or that there was no evidence of interaction between the 2 factors. Results from 3 studies in Taiwan (13), Sweden (14), and the United States (public school children in Georgia) (15) showed that lower childhood SES exacerbated associations between preterm birth and lower school test scores in adolescence (13, 14) as well as lower first-grade statewide test scores (15). SES was measured at the family level in the Taiwanese and Swedish studies and at the neighborhood level in the US study. Another study in Canada found no evidence of additive interaction between preterm birth and parenting characteristics (used as downstream markers of SES) in their associations with children’s developmental scores in early childhood (16). No national, population-based studies have assessed this question in the United States, where both preterm birth and childhood poverty are major public health concerns. Due to differences across studies to date in terms of study populations, ages and tests used for assessing cognitive and academic outcomes, and methods of measuring SES, it is unknown whether findings are generalizable to the general US population. In this study, we assessed whether household SES in early childhood modified associations between preterm birth and children’s early cognitive development as well as kindergarten reading and mathematics scores in a population-based cohort of US children born in 2001. METHODS This study analyzed data from the Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), a population-based longitudinal study of about 10,700 children born in 2001 and followed through kindergarten (17). ECLS-B was designed to provide nationally representative estimates with the application of sampling weights. Children were selected through stratified, clustered sampling of 2001 birth certificates with oversampling of twins, children with low or very low birth weight, and certain racial/ethnic groups. More details on sampling procedures are provided in Web Appendix 1 (available at https://academic.oup.com/aje). Data were collected from birth certificates and prospective assessment at 9 months (2001–2002), 2 years (2003–2004), preschool age (2005–2006), and kindergarten age (2007–2008). Kindergarten data were collected over 2 successive years in order to assess children when they entered kindergarten, which differed based on birth date. Singletons born at 24–40 weeks’ gestation were eligible for our analyses (n = 6,150). Children were excluded if they were part of a multiple birth (n = 1,800), were born outside 24–40 weeks (n = 700), were missing gestational age (GA) (n = 1,700) or birth weight (n = 200), or had a congenital anomaly or chromosomal abnormality (list in Web Table 1; n = 550). They could also be excluded if the primary survey respondent was not the biological mother (n = 150) or if the birth weight reported on the birth certificate was implausible for GA based on cutoffs established by Alexander et al. (18) (n = 200). Some children were excluded for multiple reasons. Figure 1 displays hypothesized relationships among GA at delivery, prenatal and postnatal household SES, and cognitive outcomes based on existing guidance for representing effect modification in directed acyclic graphs (19). SES is represented both as a potential confounder and effect modifier of the relationship between GA at delivery and cognitive outcomes. Prenatal SES represents a family’s SES prior to a child’s birth, which may influence both the child’s GA at delivery and postnatal household SES during the child’s early life. In the present study, only postnatal measurements of household SES were available, and so our testable hypothesis was limited to assessing whether postnatal household SES modified the associations between preterm birth and cognitive outcomes. Hypothesized potential confounders of both the exposure-outcome and modifier-outcome relationships were child’s race/ethnicity, child’s sex, maternal parity, maternal age at delivery, and household structure. Child’s age at assessment was also included as a covariate due to the rapid pace of development and associated increases in cognitive scores over the course of childhood. Figure 1. View largeDownload slide Directed acyclic graph representing potential modification of the relationship between preterm birth and children’s cognitive outcomes by household socioeconomic status, among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. Prenatal socioeconomic status reflects the family’s socioeconomic status prior to a child’s birth, which may influence both the child’s gestational age at delivery and the family’s socioeconomic status during the child’s early life. Postnatal socioeconomic status reflects the family’s socioeconomic status during the child’s early life, which may be influenced by the family’s socioeconomic history and may influence the child’s cognitive outcomes. Figure 1. View largeDownload slide Directed acyclic graph representing potential modification of the relationship between preterm birth and children’s cognitive outcomes by household socioeconomic status, among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. Prenatal socioeconomic status reflects the family’s socioeconomic status prior to a child’s birth, which may influence both the child’s gestational age at delivery and the family’s socioeconomic status during the child’s early life. Postnatal socioeconomic status reflects the family’s socioeconomic status during the child’s early life, which may be influenced by the family’s socioeconomic history and may influence the child’s cognitive outcomes. GA at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). Nearly all (97%) of the 1,700 children missing GA were born in California, where the state birth certificate did not report the clinical estimate in 2001. Household SES was measured at wave 1 using a composite index created by ECLS-B reflecting mother’s and father’s education, mother’s and father’s occupation, and household income. Briefly, the composite index was derived from the average of standardized values for each component variable; see Web Appendix 1 for more detail on index construction (17). Because impacts of childhood poverty may differ based on timing and duration of exposure (20), we considered using multiple measurements of SES to define each child’s exposure to poverty. However, SES measurements at age 9 months, age 2 years, and kindergarten were strongly consistent (Cronbach’s α = 0.96) and therefore we used wave-1 measurements as our primary measure. ECLS-B cognitive assessments were administered by trained survey staff in subjects’ homes using standard procedures (17). Cognitive ability at 2 years of age was measured at wave 2 with the Bayley Short Form–Research Edition (BSF-R) mental scale. The BSF-R mental scale was an abbreviated version of the Bayley Scales of Infant Development–Second Edition that was developed specifically for the ECLS-B (21). Children’s “scale scores” on the BSF-R mental scale estimated what their raw scores would have been on the full Bayley Scales of Infant Development–Second Edition; scale scores in the full ELCS-B sample ranged from 92.35 to 174.14 (22). Bayley Scales of Infant Development–Second Edition scores are typically normed with mean of 100 and standard deviation of 10. Because the ECLS-B did not report normed scores, our primary analyses used children’s scale scores. We also standardized scale scores to a normal distribution (mean = 0, standard deviation (SD), 1) in order to interpret results in terms of standard-deviation differences between groups. We adjusted all 2-year outcome analyses for child’s age at assessment. When determining a child’s “testing age” for development assessments, it is standard clinical practice to correct for prematurity until 2 or 3 years of age (23). Scale scores available from ECLS-B were not adjusted for prematurity, so we corrected children’s testing ages for the estimated number of weeks before 40 weeks that a child was born. For example, a child born at 36 weeks and tested at 24 months chronological age was assigned a corrected testing age of 23 months. Reading and mathematics achievement at kindergarten age were measured at waves 4 and 5 using ECLS-B–designed assessments of knowledge and skills in reading and mathematics (17). We used each child’s kindergarten scores from their first entry into kindergarten. We analyzed scale scores, which estimated the number of items that a child would have answered correctly if they had been asked all of the scored questions (24). Scores in the full sample ranged from 12.39 to 82.48 on the reading assessment and 10.85 to 69.69 on the mathematics assessment. All kindergarten outcome analyses were adjusted for age at assessment, and testing ages were not corrected for prematurity. Univariate frequencies and percentages of GA, household SES, and covariates were computed for the eligible target population, the 2-year analytical cohort, and the kindergarten analytical cohort. To describe the observed joint distribution of gestational and household SES, we computed mean scores and standard deviations for the 25 subgroups of 5 GA categories and 5 household SES quintiles. Multivariable analyses with generalized estimating equations (GEE) based on the normal distribution were used to estimate differences in scores for each outcome accounting for clustering of subjects by ECLS-B–designed sample clusters using an exchangeable correlation structure. Sampling weights were not applied in our main analyses because our primary interest was in obtaining internally valid estimates of relationships between GA, household SES, and outcome scores among study participants rather than in ensuring that our findings were representative of 2001 US births. We fitted 2 sets of models: 1) estimating the mutually adjusted independent associations of GA (early preterm, moderate preterm, late preterm, and early term compared with full term (referent)) and household SES (quintiles, with the highest SES quintile as the referent) with each outcome; and 2) adding terms for interaction between GA and household SES (e.g., early preterm × SES quintile 1). Presence of additive interaction was assessed by testing statistical significance of the set of interaction terms using the generalized score test statistic, by plotting model-predicted scores by GA and socioeconomic quintile at mean covariate values, and by examining associations between preterm birth and each outcome after stratifying by socioeconomic quintile. Crude models controlled only for children’s age at assessment. Adjusting models also controlled for child’s race/ethnicity, child’s sex, parity, maternal marital status, and maternal age at delivery. Finally, we assessed whether associations between GA, household SES, and cognitive/academic outcomes differed according to race/ethnicity by adding 3-way interaction terms as well as stratifying by race/ethnicity. Sensitivity analyses were used to assess the robustness of our findings. First, we repeated all analyses using GA based on reported date of last menstrual period; these analyses included California births. Second, wave-2 measurements of SES were used for the kindergarten analyses. Third, we assessed the robustness of our results to our choice of SES measurement variables (25) by defining SES using maternal education (less than high school, completed high school, some college, bachelor’s degree or higher), household income at wave 1 (<$25,000, $25,000–$75,000, >$75,000), household poverty at wave 1 (<130% of federal poverty level, 130%–185% of federal poverty level, >185% of federal poverty level), and using the SES composite index as a continuous variable. Fourth, we repeated the 2-year analyses without a prematurity correction. Fifth, we repeated kindergarten analyses with a prematurity correction. Finally, we repeated all analyses using complex sample weights. Analyses were conducted using SAS, version 9.3 (SAS Institute, Inc., Cary, North Carolina). The study was approved by the Emory Institutional Review Board. Unweighted counts are rounded to the nearest 50 per National Center for Education Statistics guidelines. RESULTS Of the 6,150 children who were eligible for the analyses, 5,700 completed 2-year data collection and 3,900 completed kindergarten data collection (Web Figure 1). After further excluding children with missing cognitive scores, there were 5,250 in the 2-year analytical sample and 3,800 in the kindergarten analytical sample. Due to oversampling of low-birth-weight children, 17%–19% of the 2-year and kindergarten analytical samples were born preterm, and another 26%–27% were born at early term (Table 1). Less than 0.5% of children were missing any covariate data at 2 years or kindergarten. The distributions of GA, SES, and covariates were similar in the 2-year and kindergarten analytical cohorts compared with the eligible target population. Table 1. Characteristics of Singleton Children Born at 24–40 Weeks’ Gestation and Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Abbreviation: N/A, not applicable. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Percentages may not sum to 100 due to rounding. c Gestational age at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). d Counts round to zero. e Values are expressed as mean (standard deviation). Table 1. Characteristics of Singleton Children Born at 24–40 Weeks’ Gestation and Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Abbreviation: N/A, not applicable. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Percentages may not sum to 100 due to rounding. c Gestational age at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). d Counts round to zero. e Values are expressed as mean (standard deviation). The unweighted mean score on the 2-year BSF-R mental scale assessment among children in our analytical sample was 126.2 (SD, 11.0); scores ranged from 92.4 to 174.1. Unweighted mean scores on kindergarten reading and mathematics achievement tests were 44.6 (SD, 14.9; range, 12.4–82.5) and 44.0 (SD, 10.5; range, 11.2–69.7), respectively (Table 2). Earlier delivery and lower household SES were both associated with lower mean scores on all assessments. Table 2. Mean Cognitive and Academic Achievement Scores, Stratified by Gestational Age and Household Socioeconomic Status, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Abbreviations: BSF-R, Bayley Short Form–Research Edition; SD, standard deviation. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Counts round to zero. Table 2. Mean Cognitive and Academic Achievement Scores, Stratified by Gestational Age and Household Socioeconomic Status, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Abbreviations: BSF-R, Bayley Short Form–Research Edition; SD, standard deviation. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Counts round to zero. When mutually adjusted, there were significant deficits in BSF-R mental scale score for both preterm birth and lower household SES (Table 3). In fully adjusting models, 0.1- to 0.6-standard-deviation deficits were observed for children born at early preterm (−7.0 points, 95% confidence interval (CI): −8.6, −5.4), moderate preterm (−2.9 points, 95% CI: −4.1, −1.7), and late preterm (−1.3 points, 95% CI: −2.3, −0.3). Children in lower SES quintiles performed 0.2–0.6 standard deviations lower than those in the highest quintile. Estimated deficits for lower household SES compared with the highest SES quintile ranged from −2.5 points (95% CI: −3.5, −1.5) for the second-highest quintile to −6.7 points (95% CI: −7.8, −5.6) for the lowest quintile. There was no statistical evidence of additive interaction between GA and household SES (P = 0.47; Figure 2A). Stratifying by socioeconomic quintile, estimated risk differences for the preterm and early term groups compared with term children were similar across socioeconomic quintiles (Web Table 2). Table 3. Associations Between Gestational Age at Delivery, Socioeconomic Status, and Cognitive Outcome at Age 2 Years and Reading and Mathematics Achievement Outcomes at Kindergarten Age, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Abbreviations: BSF-R, Bayley Short Form–Research Edition; CI, confidence interval. a Adjusted only for child’s age at assessment. b Adjusted for child’s age at assessment, child’s race/ethnicity (non-Hispanic black, Hispanic, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, more than 1 race, non-Hispanic white), child’s sex (female, male), maternal age at delivery (15–17 years, 18–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40 years or older), parity (nulliparous, 1–2 previous live births, 3 or more previous live births), and maternal marital status (married; never married; divorced, widowed, or separated). c Estimated β reflects estimated difference in mean outcome score. Table 3. Associations Between Gestational Age at Delivery, Socioeconomic Status, and Cognitive Outcome at Age 2 Years and Reading and Mathematics Achievement Outcomes at Kindergarten Age, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Abbreviations: BSF-R, Bayley Short Form–Research Edition; CI, confidence interval. a Adjusted only for child’s age at assessment. b Adjusted for child’s age at assessment, child’s race/ethnicity (non-Hispanic black, Hispanic, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, more than 1 race, non-Hispanic white), child’s sex (female, male), maternal age at delivery (15–17 years, 18–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40 years or older), parity (nulliparous, 1–2 previous live births, 3 or more previous live births), and maternal marital status (married; never married; divorced, widowed, or separated). c Estimated β reflects estimated difference in mean outcome score. Figure 2. View largeDownload slide Model-predicted cognitive and academic achievement scores and 95% confidence intervals derived from models estimating interaction between gestational age and household socioeconomic status for Bayley Short Form–Research Edition (BSF-R) mental scale scores at age 2 years (A), reading scale scores at kindergarten age (B), and mathematics scale scores at kindergarten age (C), among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. All models adjusted for child’s age at assessment (corrected for prematurity for the BSF-R mental scale), child’s race/ethnicity, child’s sex, maternal age at delivery, parity, and maternal marital status. Predicted scores are at mean or reference covariate values. Shaded circle: early preterm; shaded square: moderate preterm; shaded diamond: late preterm; open circle: early term; shaded triangle: term. Figure 2. View largeDownload slide Model-predicted cognitive and academic achievement scores and 95% confidence intervals derived from models estimating interaction between gestational age and household socioeconomic status for Bayley Short Form–Research Edition (BSF-R) mental scale scores at age 2 years (A), reading scale scores at kindergarten age (B), and mathematics scale scores at kindergarten age (C), among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. All models adjusted for child’s age at assessment (corrected for prematurity for the BSF-R mental scale), child’s race/ethnicity, child’s sex, maternal age at delivery, parity, and maternal marital status. Predicted scores are at mean or reference covariate values. Shaded circle: early preterm; shaded square: moderate preterm; shaded diamond: late preterm; open circle: early term; shaded triangle: term. Preterm birth and lower household SES were each independently associated with lower kindergarten reading scale scores (Table 3). In fully adjusting models, 0.1- to 0.4-standard-deviation deficits were observed for early preterm (−5.3 points, 95% CI: −7.7, −3.0), moderate preterm (−2.6 points, 95% CI: −3.9, −1.3), and late preterm (−1.5 points, 95% CI: −3.0, 0.0) groups. Children in lower SES quintiles performed 0.3–0.9 standard deviations lower than did those in the highest quintile. Estimated deficits for lower household SES compared with the highest SES quintile ranged from −5.1 points (95% CI: −6.5, −3.7) for the second-highest quintile to −12.6 (95% CI: −14.4, −10.9) for the lowest quintile. The addition of terms for interaction between GA group and household SES was not statistically significant (P = 0.82; Figure 2B). Qualitative examination of stratified results suggested that deficits for early preterm children worsened with lower SES quintile, and deficits for moderate preterm and late preterm were attenuated with lower SES quintile, although there was variation across quintiles (Web Table 2). Preterm birth and lower household SES were also each independently associated with lower mathematics scale scores (Table 3). After adjusting for household SES and other covariates, 0.1- to 0.6-standard-deviation deficits were observed for children born at early preterm (−6.6 points, 95% CI: −8.5, −4.7), moderate preterm (−3.2 points, 95% CI: −4.3, −2.1), and late preterm (−1.4 points, 95% CI: −2.5, −0.4). Children in lower SES quintiles performed 0.3–0.9 standard deviations lower than those in the highest quintile. The estimated deficits for lower household SES compared with the highest SES quintile ranged from −3.6 points (95% CI: −4.4, −2.7) for the second-highest quintile to −9.1 points (95% CI: −10.2, −7.9) for the lowest quintile. There was no statistical evidence of additive interaction between GA and household SES (P = 0.68; Figure 2C). Socioeconomic stratum–specific analyses indicated that deficits for early preterm worsened slightly with lower SES quintile, and deficits for moderate preterm and late preterm were attenuated (Web Table 2). For all 3 outcomes, 3-way interaction terms with race/ethnicity were not significant, and there was no significant interaction between GA and household SES in any racial/ethnic group in analyses stratified by race/ethnicity. Findings were similar to our main analyses when using measurements of GA with last menstrual period, using alternative SES measures (including using the wave-2 measurement of SES quintile), and applying complex sample weights. In the 2-year analyses, not correcting for prematurity resulted in larger estimated deficits for preterm groups, while in the kindergarten analyses, correcting for prematurity resulted in attenuated estimated deficits for preterm groups. DISCUSSION Preterm birth and lower household SES were each associated with substantial deficits in children’s cognitive scores at 2 years of age and in reading and mathematics achievement at kindergarten age. While household SES did not modify the associations between preterm birth and cognitive outcomes, in the absence of additive interaction the 2 factors separately and adversely affected children’s scores. Score deficits associated with being exposed to both preterm birth and lower household SES were approximately equal to the sum of the estimated deficits associated with the 2 separate exposures. Children exposed to both preterm birth and lower household SES performed at the lowest levels on all outcomes. Prior studies on additive interactions between preterm birth and SES in their associations with cognitive and academic outcomes have found either that lower SES exacerbated the adverse impacts of preterm birth (13–15) or that there was no evidence of interaction between the 2 factors (16). Two of these studies included assessments of children at similar ages to those investigated in our study—early childhood and around the time of starting elementary school (15, 16). Richards et al. (15) found that both preterm birth and living in a more-deprived neighborhood were associated with higher risk of failing first-grade statewide tests among 330,000 public school children in the state of Georgia. There was also positive additive interaction between the 2 factors, such that living in a more-deprived neighborhood exacerbated the adverse impact of preterm birth on test failure (15). It is possible that neighborhood-level deprivation captured not only families’ socioeconomic status but also community resources and school quality, which may contribute to exacerbating impacts of preterm birth on children’s school-related outcomes. Further, it is possible that performance on statewide competency tests was affected not only by students’ cognitive skills but also by school quality and climate as well as noncognitive skills such as behavior and socioemotional well-being. Brown et al. (16) found no evidence of modification by parenting factors—quality of interactions, effectiveness, and consistency—of the association between late preterm or early term birth and cognitive outcomes among 15,000 Canadian children enrolled in a national longitudinal study. The results of their study may be driven by its inclusion of only children born at 34 weeks or later; our study similarly found small cognitive deficits for late-preterm children and no significant deficit for early-term children. While our analysis found that household SES did not exacerbate the association between preterm birth and worsened cognitive outcomes, it must be emphasized that lower household SES was still found to independently and adversely influence children’s cognitive scores at both age 2 years and kindergarten age. These findings echo previous research linking childhood poverty with adverse impacts on cognitive development and academic performance (2, 26). It may be possible that impacts of poverty on children’s cognitive and academic outcomes are so substantial that the additional influence of preterm birth may reach a bottom “threshold” in terms of affecting scores. In our study, estimated disparities comparing the highest and lowest socioeconomic groups were larger than those comparing early preterm with term children and term-born children in the lowest SES quintile had predicted scores lower than most preterm groups in the highest SES quintile. A major strength of the present study is the utilization of a large, population-based sample of US children with rigorous, prospective measurements of cognitive development at multiple time points. Using data from a population-based cohort for studying outcomes of preterm birth, however, comes with the caveat that it may underrepresent children who are more ill and unable or unwilling to participate in such a study. We were able to analyze the outcomes of the same group of children as they grew and developed through early childhood to kindergarten, although we were not able to analyze trajectories of cognitive growth due to noncongruity of the 2-year and kindergarten outcomes. It is important to acknowledge the limitations of this study in assessing potential complexities of pathways through which preterm birth and SES may separately and jointly influence children’s cognitive outcomes. The present analysis was not able to assess pathways involving prenatal SES. It is possible that the impacts of SES may be mediated through preterm birth and also, postnatally, may modify the associations between preterm birth and cognitive outcomes. Studies with both prenatal and postnatal SES measurements are needed to disentangle these pathways. Experiencing a preterm birth may affect family’s SES through incurring medical costs and forcing changes in income and employment due to increased needs for caregiving (27). While we were not able to assess whether such changes in SES occurred, the composite SES index—reflecting income, educational attainment, and occupation—was plausibly more stable against this concern compared with using only household income to measure SES. Although there was a large overall sample size, some GA-SES exposure groups had sparse data, especially for the early and moderate preterm groups and in analyses stratified by race/ethnicity. Because of the known socioeconomic disparities across racial/ethnic groups in the United States, future studies should specifically examine whether relationships among preterm birth, SES, and cognitive development vary across racial/ethnic groups. About 85% and 61% of eligible children were included in the 2-year and kindergarten analyses, respectively. However, the analytical samples were similar to the eligible target population in terms of gestational age, SES, and demographic factors. Dropout rates were exaggerated because about half of those lost between the 2 waves were part of a planned sample-size reduction carried out by systematically reducing sampling rates (17). In our study, preterm birth and lower SES were associated with substantial deficits in cognitive scores at 2 years and in academic achievement scores at kindergarten age, although there was no evidence of additive interaction between the 2 exposures. However, even in the absence of interaction, both factors had important impacts on cognitive scores. Given that preterm birth and childhood poverty are prevalent in the United States and often overlap, more research is needed to better understand the independent and joint associations of preterm birth and childhood poverty with children’s development. ACKNOWLEDGMENTS Author affiliations: Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia (Jennifer L. Beauregard, Carolyn Drews-Botsch, W. Dana Flanders, Michael R. Kramer); and Department of Behavioral Sciences and Health Education, Rollins School of Public Health, Emory University, Atlanta, Georgia (Jessica M. Sales). J.L.B. received support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, T32 Predoctoral Training Program in Reproductive, Perinatal, and Pediatric Epidemiology under (award T32HD052460), and from the Health Resources and Services Administration of the US Department of Health and Human Services (grant T03MC07651). M.R.K. received support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (award K01-HD074726). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Health Resources and Services Administration, Department of Health and Human Services, or the US Government. Conflict of interest: none declared. Abbreviations BSF-R Bayley Short Form–Research Edition CI confidence interval ECLS-B Early Childhood Longitudinal Study, Birth Cohort GA gestational age SD standard deviation SES socioeconomic status REFERENCES 1 Behrman RE , Butler AS , eds. Preterm Birth: Causes, Consequences, and Prevention . Washington, DC : National Academies Press ; 2007 . http://www.ncbi.nlm.nih.gov/books/NBK11362/. Accessed August 18, 2014. 2 Brooks-Gunn J , Duncan GJ . The effects of poverty on children . Future Child . 1997 ; 7 ( 2 ): 55 – 71 . Google Scholar CrossRef Search ADS PubMed 3 Conti G , Heckman JJ . Understanding the early origins of the education-health gradient: a framework that can also be applied to analyze gene-environment interactions . Perspect Psychol Sci . 2010 ; 5 ( 5 ): 585 – 605 . Google Scholar CrossRef Search ADS PubMed 4 Heckman JJ . Skill formation and the economics of investing in disadvantaged children . Science . 2006 ; 312 ( 5782 ): 1900 – 1902 . Google Scholar CrossRef Search ADS PubMed 5 Hamilton BE , Martin JA , Osterman MJ , et al. . Births: final data for 2014 . Natl Vital Stat Rep . 2015 ; 64 ( 12 ): 1 – 64 . 6 Addy S , Engelhardt W , Skinner C . Basic Facts About Low-Income Children. New York, NY : National Center for Children in Poverty ; 2013 . http://www.nccp.org/publications/pub_1076.html. Accessed November 18, 2014. 7 Burd I , Balakrishnan B , Kannan S . Models of fetal brain injury, intrauterine inflammation, and preterm birth . Am J Reprod Immunol . 2012 ; 67 ( 4 ): 287 – 294 . Google Scholar CrossRef Search ADS PubMed 8 Stiles J , Jernigan TL . The basics of brain development . Neuropsychol Rev . 2010 ; 20 ( 4 ): 327 – 348 . Google Scholar CrossRef Search ADS PubMed 9 Johnson SB , Riis JL , Noble KG . State of the art review: poverty and the developing brain . Pediatrics . 2016 ; 137 ( 4 ): e20153075 . Google Scholar CrossRef Search ADS PubMed 10 Bendersky M , Lewis M . Environmental risk, biological risk, and developmental outcome . Dev Psychol . 1994 ; 30 ( 4 ): 484 – 494 . Google Scholar CrossRef Search ADS 11 Sameroff AJ . Environmental context of child development . J Pediatr . 1986 ; 109 ( 1 ): 192 – 200 . Google Scholar CrossRef Search ADS PubMed 12 Messer LC , Vinikoor LC , Laraia BA , et al. . Socioeconomic domains and associations with preterm birth . Soc Sci Med . 2008 ; 67 ( 8 ): 1247 – 1257 . Google Scholar CrossRef Search ADS PubMed 13 Wang WL , Sung YT , Sung FC , et al. . Low birth weight, prematurity, and paternal social status: impact on the basic competence test in Taiwanese adolescents . J Pediatr . 2008 ; 153 ( 3 ): 333 – 338 . Google Scholar CrossRef Search ADS PubMed 14 Ekeus C , Lindström K , Lindblad F , et al. . Preterm birth, social disadvantage, and cognitive competence in Swedish 18- to 19-year-old men . Pediatrics . 2010 ; 125 ( 1 ): e67 – e73 . Google Scholar CrossRef Search ADS PubMed 15 Richards JL , Chapple-McGruder T , Williams BL , et al. . Does neighborhood deprivation modify the effect of preterm birth on children’s first grade academic performance? Soc Sci Med . 2015 ; 132 : 122 – 131 . Google Scholar CrossRef Search ADS PubMed 16 Brown HK , Speechley KN , Macnab J , et al. . Mild prematurity, proximal social processes, and development . Pediatrics . 2014 ; 134 ( 3 ): e814 – e824 . Google Scholar CrossRef Search ADS PubMed 17 Snow , K , Derecho , A , Wheeless , S , et al. . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Kindergarten 2006 and 2007 Data File User’s Manual (2010-010). Washington, DC: National Center for Education Statistics . 2009 . 18 Alexander GR , Himes JH , Kaufman RB , et al. . A United States national reference for fetal growth . Obstet Gynecol . 1996 ; 87 ( 2 ): 163 – 168 . Google Scholar CrossRef Search ADS PubMed 19 VanderWeele TJ , Robins JM . Four types of effect modification: a classification based on directed acyclic graphs . Epidemiology . 2007 ; 18 ( 5 ): 561 – 568 . Google Scholar CrossRef Search ADS PubMed 20 Guo G . The timing of the influences of cumulative poverty on children’s cognitive ability and achievement . Soc Forces . 1998 ; 77 ( 1 ): 257 – 287 . Google Scholar CrossRef Search ADS 21 National Center for Education Statistics . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), User’s Manual for the ECLS-B Longitudinal 9-month-2-year Data File and Electronic Codebook. Washington, DC : National Center for Education Statistics ; 2006 . 22 Andreassen C , Fletcher P . Early Childhood Longitudinal Study, Birth Cohort (ECLS–B) Psychometric Report for the 2-Year Data Collection (NCES 2007-084). Washington, DC : National Center for Education Statistics ; 2007 . http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2007084. Accessed March 11, 2015. 23 Engle WA ; American Academy of Pediatrics Committee on Fetus and Newborn . Age terminology during the perinatal period . Pediatrics . 2004 ; 114 ( 5 ): 1362 – 1364 . Google Scholar CrossRef Search ADS PubMed 24 Najarian M , Snow K , Lennon J , et al. . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Preschool—Kindergarten 2007 Psychometric Report. Washington, DC : National Center for Education Statistics ; 2010 . https://nces.ed.gov/pubs2010/2010009.pdf. 25 Rehkopf DH , Glymour MM , Osypuk TL . The consistency assumption for causal inference in social epidemiology: when a rose is not a rose . Curr Epidemiol Rep . 2016 ; 3 ( 1 ): 63 – 71 . Google Scholar CrossRef Search ADS PubMed 26 Duncan GJ , Yeung WJ , Brooks-Gunn J , et al. . How much does childhood poverty affect the life chances of children? Am Sociol Rev . 1998 ; 63 ( 3 ): 406 . Google Scholar CrossRef Search ADS 27 Hodek JM , von der Schulenburg JM , Mittendorf T . Measuring economic consequences of preterm birth—methodological recommendations for the evaluation of personal burden on children and their caregivers . Health Econ Rev . 2011 ; 1 : 6 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Epidemiology Oxford University Press

Does Socioeconomic Status Modify the Association Between Preterm Birth and Children’s Early Cognitive Ability and Kindergarten Academic Achievement in the United States?

Loading next page...
 
/lp/ou_press/does-socioeconomic-status-modify-the-association-between-preterm-birth-VNw5goiVrF
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0002-9262
eISSN
1476-6256
D.O.I.
10.1093/aje/kwy068
Publisher site
See Article on Publisher Site

Abstract

Abstract Being born preterm and being raised in poverty are each linked with adverse cognitive outcomes. Using data from 5,250 singletons born in the United States in 2001 and enrolled in the Early Childhood Longitudinal Study, Birth Cohort, we examined whether household socioeconomic status (SES) modified the association between preterm birth (PTB) and children’s scores on cognitive assessments at age 2 years and reading and mathematics assessments at kindergarten age. Gestational age was measured from birth certificates and categorized as early preterm, moderate preterm, late preterm, early term, and term. SES was measured at age 9 months using a composite of parental education, occupation, and income. PTB was associated with 0.1- to 0.6-standard deviation-deficits in 2-year cognitive ability and kindergarten mathematics scores and with 0.1- to 0.4-standard-deviation deficits in kindergarten reading scores. Children living in the lowest (versus highest) SES quintile scored 0.6 standard deviations lower on 2-year cognitive ability, 1.1 standard deviations lower on kindergarten reading, and 0.9 standard deviations lower on kindergarten mathematics. The association between PTB and cognitive outcomes did not differ by postnatal SES. However, children who were both born preterm and lived in lower-SES households had the poorest performance on all 3 outcomes and therefore may represent a uniquely high-risk group. cognitive development, interaction, poverty, preterm birth Being born preterm and being raised in poverty are each linked with adverse cognitive outcomes (1, 2). Children’s early cognitive ability and being ready to enter school are important predictors of later social and academic achievement as well as socioeconomic and health trajectories over the life course (3, 4). In the United States, both of these factors are important public health concerns, with 10% of children born preterm and half residing in low-income households (5, 6). Preterm children may be born with developmentally immature brains, and they face higher risk of central nervous system and brain injuries (7). In the first months and years of children’s development, sensory enrichment and deprivation have important influences on brain growth and development as children are experiencing, responding to, and learning from stimuli in their environments (8). Children raised in families of lower socioeconomic status (SES) may be more likely to experience stressors such as material deprivation and to have fewer cognitively stimulating experiences (9). Preterm children raised in more advantaged environments have been observed to have better developmental outcomes compared with preterm children in more disadvantaged environments, suggesting that growing up in a more advantaged environment may attenuate the adverse impacts of preterm birth (10, 11). Studies of cognitive outcomes of preterm birth often treat SES as a confounding factor because it is associated with risk of being born preterm (12) and with children’s developmental outcomes (2, 9). It is plausible that SES also acts as a modifier of the effect of preterm birth on children’s cognitive development, but only a few epidemiologic studies have tested this hypothesis. In studies assessing additive interaction between preterm birth and SES, investigators have found either that lower childhood SES exacerbated the association between preterm birth and worse cognitive or school outcomes or that there was no evidence of interaction between the 2 factors. Results from 3 studies in Taiwan (13), Sweden (14), and the United States (public school children in Georgia) (15) showed that lower childhood SES exacerbated associations between preterm birth and lower school test scores in adolescence (13, 14) as well as lower first-grade statewide test scores (15). SES was measured at the family level in the Taiwanese and Swedish studies and at the neighborhood level in the US study. Another study in Canada found no evidence of additive interaction between preterm birth and parenting characteristics (used as downstream markers of SES) in their associations with children’s developmental scores in early childhood (16). No national, population-based studies have assessed this question in the United States, where both preterm birth and childhood poverty are major public health concerns. Due to differences across studies to date in terms of study populations, ages and tests used for assessing cognitive and academic outcomes, and methods of measuring SES, it is unknown whether findings are generalizable to the general US population. In this study, we assessed whether household SES in early childhood modified associations between preterm birth and children’s early cognitive development as well as kindergarten reading and mathematics scores in a population-based cohort of US children born in 2001. METHODS This study analyzed data from the Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), a population-based longitudinal study of about 10,700 children born in 2001 and followed through kindergarten (17). ECLS-B was designed to provide nationally representative estimates with the application of sampling weights. Children were selected through stratified, clustered sampling of 2001 birth certificates with oversampling of twins, children with low or very low birth weight, and certain racial/ethnic groups. More details on sampling procedures are provided in Web Appendix 1 (available at https://academic.oup.com/aje). Data were collected from birth certificates and prospective assessment at 9 months (2001–2002), 2 years (2003–2004), preschool age (2005–2006), and kindergarten age (2007–2008). Kindergarten data were collected over 2 successive years in order to assess children when they entered kindergarten, which differed based on birth date. Singletons born at 24–40 weeks’ gestation were eligible for our analyses (n = 6,150). Children were excluded if they were part of a multiple birth (n = 1,800), were born outside 24–40 weeks (n = 700), were missing gestational age (GA) (n = 1,700) or birth weight (n = 200), or had a congenital anomaly or chromosomal abnormality (list in Web Table 1; n = 550). They could also be excluded if the primary survey respondent was not the biological mother (n = 150) or if the birth weight reported on the birth certificate was implausible for GA based on cutoffs established by Alexander et al. (18) (n = 200). Some children were excluded for multiple reasons. Figure 1 displays hypothesized relationships among GA at delivery, prenatal and postnatal household SES, and cognitive outcomes based on existing guidance for representing effect modification in directed acyclic graphs (19). SES is represented both as a potential confounder and effect modifier of the relationship between GA at delivery and cognitive outcomes. Prenatal SES represents a family’s SES prior to a child’s birth, which may influence both the child’s GA at delivery and postnatal household SES during the child’s early life. In the present study, only postnatal measurements of household SES were available, and so our testable hypothesis was limited to assessing whether postnatal household SES modified the associations between preterm birth and cognitive outcomes. Hypothesized potential confounders of both the exposure-outcome and modifier-outcome relationships were child’s race/ethnicity, child’s sex, maternal parity, maternal age at delivery, and household structure. Child’s age at assessment was also included as a covariate due to the rapid pace of development and associated increases in cognitive scores over the course of childhood. Figure 1. View largeDownload slide Directed acyclic graph representing potential modification of the relationship between preterm birth and children’s cognitive outcomes by household socioeconomic status, among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. Prenatal socioeconomic status reflects the family’s socioeconomic status prior to a child’s birth, which may influence both the child’s gestational age at delivery and the family’s socioeconomic status during the child’s early life. Postnatal socioeconomic status reflects the family’s socioeconomic status during the child’s early life, which may be influenced by the family’s socioeconomic history and may influence the child’s cognitive outcomes. Figure 1. View largeDownload slide Directed acyclic graph representing potential modification of the relationship between preterm birth and children’s cognitive outcomes by household socioeconomic status, among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. Prenatal socioeconomic status reflects the family’s socioeconomic status prior to a child’s birth, which may influence both the child’s gestational age at delivery and the family’s socioeconomic status during the child’s early life. Postnatal socioeconomic status reflects the family’s socioeconomic status during the child’s early life, which may be influenced by the family’s socioeconomic history and may influence the child’s cognitive outcomes. GA at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). Nearly all (97%) of the 1,700 children missing GA were born in California, where the state birth certificate did not report the clinical estimate in 2001. Household SES was measured at wave 1 using a composite index created by ECLS-B reflecting mother’s and father’s education, mother’s and father’s occupation, and household income. Briefly, the composite index was derived from the average of standardized values for each component variable; see Web Appendix 1 for more detail on index construction (17). Because impacts of childhood poverty may differ based on timing and duration of exposure (20), we considered using multiple measurements of SES to define each child’s exposure to poverty. However, SES measurements at age 9 months, age 2 years, and kindergarten were strongly consistent (Cronbach’s α = 0.96) and therefore we used wave-1 measurements as our primary measure. ECLS-B cognitive assessments were administered by trained survey staff in subjects’ homes using standard procedures (17). Cognitive ability at 2 years of age was measured at wave 2 with the Bayley Short Form–Research Edition (BSF-R) mental scale. The BSF-R mental scale was an abbreviated version of the Bayley Scales of Infant Development–Second Edition that was developed specifically for the ECLS-B (21). Children’s “scale scores” on the BSF-R mental scale estimated what their raw scores would have been on the full Bayley Scales of Infant Development–Second Edition; scale scores in the full ELCS-B sample ranged from 92.35 to 174.14 (22). Bayley Scales of Infant Development–Second Edition scores are typically normed with mean of 100 and standard deviation of 10. Because the ECLS-B did not report normed scores, our primary analyses used children’s scale scores. We also standardized scale scores to a normal distribution (mean = 0, standard deviation (SD), 1) in order to interpret results in terms of standard-deviation differences between groups. We adjusted all 2-year outcome analyses for child’s age at assessment. When determining a child’s “testing age” for development assessments, it is standard clinical practice to correct for prematurity until 2 or 3 years of age (23). Scale scores available from ECLS-B were not adjusted for prematurity, so we corrected children’s testing ages for the estimated number of weeks before 40 weeks that a child was born. For example, a child born at 36 weeks and tested at 24 months chronological age was assigned a corrected testing age of 23 months. Reading and mathematics achievement at kindergarten age were measured at waves 4 and 5 using ECLS-B–designed assessments of knowledge and skills in reading and mathematics (17). We used each child’s kindergarten scores from their first entry into kindergarten. We analyzed scale scores, which estimated the number of items that a child would have answered correctly if they had been asked all of the scored questions (24). Scores in the full sample ranged from 12.39 to 82.48 on the reading assessment and 10.85 to 69.69 on the mathematics assessment. All kindergarten outcome analyses were adjusted for age at assessment, and testing ages were not corrected for prematurity. Univariate frequencies and percentages of GA, household SES, and covariates were computed for the eligible target population, the 2-year analytical cohort, and the kindergarten analytical cohort. To describe the observed joint distribution of gestational and household SES, we computed mean scores and standard deviations for the 25 subgroups of 5 GA categories and 5 household SES quintiles. Multivariable analyses with generalized estimating equations (GEE) based on the normal distribution were used to estimate differences in scores for each outcome accounting for clustering of subjects by ECLS-B–designed sample clusters using an exchangeable correlation structure. Sampling weights were not applied in our main analyses because our primary interest was in obtaining internally valid estimates of relationships between GA, household SES, and outcome scores among study participants rather than in ensuring that our findings were representative of 2001 US births. We fitted 2 sets of models: 1) estimating the mutually adjusted independent associations of GA (early preterm, moderate preterm, late preterm, and early term compared with full term (referent)) and household SES (quintiles, with the highest SES quintile as the referent) with each outcome; and 2) adding terms for interaction between GA and household SES (e.g., early preterm × SES quintile 1). Presence of additive interaction was assessed by testing statistical significance of the set of interaction terms using the generalized score test statistic, by plotting model-predicted scores by GA and socioeconomic quintile at mean covariate values, and by examining associations between preterm birth and each outcome after stratifying by socioeconomic quintile. Crude models controlled only for children’s age at assessment. Adjusting models also controlled for child’s race/ethnicity, child’s sex, parity, maternal marital status, and maternal age at delivery. Finally, we assessed whether associations between GA, household SES, and cognitive/academic outcomes differed according to race/ethnicity by adding 3-way interaction terms as well as stratifying by race/ethnicity. Sensitivity analyses were used to assess the robustness of our findings. First, we repeated all analyses using GA based on reported date of last menstrual period; these analyses included California births. Second, wave-2 measurements of SES were used for the kindergarten analyses. Third, we assessed the robustness of our results to our choice of SES measurement variables (25) by defining SES using maternal education (less than high school, completed high school, some college, bachelor’s degree or higher), household income at wave 1 (<$25,000, $25,000–$75,000, >$75,000), household poverty at wave 1 (<130% of federal poverty level, 130%–185% of federal poverty level, >185% of federal poverty level), and using the SES composite index as a continuous variable. Fourth, we repeated the 2-year analyses without a prematurity correction. Fifth, we repeated kindergarten analyses with a prematurity correction. Finally, we repeated all analyses using complex sample weights. Analyses were conducted using SAS, version 9.3 (SAS Institute, Inc., Cary, North Carolina). The study was approved by the Emory Institutional Review Board. Unweighted counts are rounded to the nearest 50 per National Center for Education Statistics guidelines. RESULTS Of the 6,150 children who were eligible for the analyses, 5,700 completed 2-year data collection and 3,900 completed kindergarten data collection (Web Figure 1). After further excluding children with missing cognitive scores, there were 5,250 in the 2-year analytical sample and 3,800 in the kindergarten analytical sample. Due to oversampling of low-birth-weight children, 17%–19% of the 2-year and kindergarten analytical samples were born preterm, and another 26%–27% were born at early term (Table 1). Less than 0.5% of children were missing any covariate data at 2 years or kindergarten. The distributions of GA, SES, and covariates were similar in the 2-year and kindergarten analytical cohorts compared with the eligible target population. Table 1. Characteristics of Singleton Children Born at 24–40 Weeks’ Gestation and Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Abbreviation: N/A, not applicable. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Percentages may not sum to 100 due to rounding. c Gestational age at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). d Counts round to zero. e Values are expressed as mean (standard deviation). Table 1. Characteristics of Singleton Children Born at 24–40 Weeks’ Gestation and Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Characteristic Target Population 2-Year Cohort Kindergarten Cohort No.a %b No.a %b No.a %b Total 6,150 100 5,250 100.0 3,800 100.0 Gestational agec  Early preterm 200 3.3 150 3.1 100 2.9  Moderate preterm 450 7.2 350 7.0 250 7.1  Late preterm 550 8.6 450 8.6 300 7.7  Early term 1,600 26.4 1,400 26.3 1,000 26.8  Full term 3,350 54.5 2,900 55.0 2,100 55.5 Socioeconomic index  Quintile 1 (lowest) 1,200 19.5 1,000 18.6 700 17.9  Quintile 2 1,250 20.6 1,050 19.9 750 20.0  Quintile 3 1,250 20.4 1,100 20.7 750 20.1  Quintile 4 1,150 18.8 1,050 19.6 750 19.3  Quintile 5 (highest) 1,250 20.6 1,100 21.1 850 22.6 Child’s race/ethnicity  Black, non-Hispanic 1,150 18.4 950 18.1 700 18.8  Hispanic 1,100 17.7 900 17.5 650 17.7  Asian 650 10.3 500 9.3 400 10.9  Native Hawaiian or other Pacific Islander 50 0.4 –d 0.4 –d 0.3  American Indian or Alaska Native 150 2.7 150 2.7 100 3.2  More than 1 race, non-Hispanic 500 8.2 450 8.7 350 9.0  White, non-Hispanic 2,600 42.2 2,250 43.2 1,500 39.8  Missing –d 0.2 –d 0.2 –d 0.3 Child’s age at assessment, months  Uncorrectede N/A 24.4 (1.2) 68.3 (4.2)  Corrected for prematuritye N/A 23.9 (1.4) N/A Child’s sex  Female 3,000 48.8 2,550 48.7 1,850 49.2  Male 3,150 51.2 2,700 51.3 1,900 50.8 Maternal age group at birth, years  15–17 250 4.1 200 4.1 150 3.8  18–19 500 8.4 450 8.3 300 7.8  20–24 1,650 26.9 1,350 26.1 950 25.5  25–29 1,500 24.4 1,300 24.5 950 24.7  30–34 1,400 22.8 1,200 23.2 900 24.2  35–39 650 10.8 600 11.2 450 11.4  ≥40 150 2.5 150 2.6 100 2.7 Parity  0 2,500 40.8 2,100 40.3 1,550 40.9  1–2 3,050 49.2 2,600 49.5 1,850 49.4  ≥3 600 9.6 500 9.7 350 9.3  Missing –d 0.4 –d 0.5 –d 0.4 Marital status  Married 3,950 64.0 3,400 65.0 2,500 65.7  Never married 1,800 29.3 1,500 28.6 1,050 28.0  Divorced, widowed, or separated 400 6.6 350 6.4 250 6.3 Abbreviation: N/A, not applicable. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Percentages may not sum to 100 due to rounding. c Gestational age at delivery was measured using the clinical estimate of gestation reported on the birth certificate and categorized as early preterm (24–27 weeks), moderate preterm (28–33 weeks), late preterm (34–36 weeks), early term (37–38 weeks), or full term (39–40 weeks). d Counts round to zero. e Values are expressed as mean (standard deviation). The unweighted mean score on the 2-year BSF-R mental scale assessment among children in our analytical sample was 126.2 (SD, 11.0); scores ranged from 92.4 to 174.1. Unweighted mean scores on kindergarten reading and mathematics achievement tests were 44.6 (SD, 14.9; range, 12.4–82.5) and 44.0 (SD, 10.5; range, 11.2–69.7), respectively (Table 2). Earlier delivery and lower household SES were both associated with lower mean scores on all assessments. Table 2. Mean Cognitive and Academic Achievement Scores, Stratified by Gestational Age and Household Socioeconomic Status, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Abbreviations: BSF-R, Bayley Short Form–Research Edition; SD, standard deviation. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Counts round to zero. Table 2. Mean Cognitive and Academic Achievement Scores, Stratified by Gestational Age and Household Socioeconomic Status, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Assessment and Gestational Age Total Socioeconomic Index Quintile 1 (Lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Highest) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) No.a Mean (SD) Outcome at 2 Years BSF-R mental scale score  Total 5,250 126.2 (11.0) 1,000 122.2 (9.8) 1,050 123.5 (9.9) 1,100 126.2 (10.8) 1,050 128.1 (11.4) 1,100 130.3 (10.7)  Full term 2,900 127.9 (10.5) 500 123.9 (9.4) 550 124.8 (9.4) 600 128.0 (10.3) 600 130.2 (10.5) 700 131.2 (10.6)  Early term 1,400 126.4 (10.8) 250 122.8 (9.5) 250 124.8 (9.7) 300 125.4 (11.2) 300 127.6 (11.2) 300 130.7 (10.5)  Late preterm 450 124.0 (10.5) 100 120.6 (9.0) 100 122.1 (9.4) 100 124.7 (10.4) 100 126.7 (12.3) 50 128.4 (9.6)  Moderate preterm 350 120.0 (10.2) 100 117.7 (9.8) 100 118.8 (9.1) 50 122.7 (9.4) 50 119.1 (11.4) 50 123.6 (11.1)  Early preterm 150 113.9 (9.7) 50 111.0 (9.3) 50 110.3 (9.1) 50 115.7 (8.8) 50 115.1 (11.4) –b 118.2 (8.0) Outcome at Kindergarten Reading scale score  Total 3,800 44.6 (14.9) 700 36.4 (13.7) 750 39.7 (13.1) 750 44.1 (13.5) 750 47.1 (13.3) 850 53.8 (14.1)  Full term 2,100 45.6 (14.8) 350 36.8 (13.6) 400 40.0 (13.0) 400 44.9 (13.7) 400 47.9 (12.7) 550 54.5 (14.0)  Early term 1,000 44.4 (14.9) 200 36.9 (13.0) 200 39.7 (13.2) 200 42.7 (13.1) 200 47.2 (14.2) 250 53.4 (14.7)  Late preterm 300 42.4 (14.7) 100 36.7 (15.0) 50 38.7 (13.7) 50 44.3 (12.7) 50 44.6 (14.5) 50 52.6 (12.5)  Moderate preterm 250 42.1 (13.9) 50 34.4 (14.6) 50 40.5 (12.4) 50 45.1 (12.6) 50 45.0 (12.1) 50 49.4 (13.1)  Early preterm 100 39.9 (15.3) –b 30.5 (13.7) –b 32.7 (11.7) 50 41.7 (15.1) 50 42.7 (13.7) –b 52.3 (13.7) Mathematics scale score  Total 3,800 44.0 (10.5) 700 37.7 (9.9) 750 40.4 (9.8) 750 43.8 (9.2) 750 46.0 (9.2) 850 50.8 (9.0)  Full term 2,100 45.0 (10.2) 350 38.2 (9.5) 400 41.0 (9.5) 400 44.3 (9.3) 400 47.0 (8.5) 550 51.4 (8.6)  Early term 1,000 44.1 (10.6) 200 38.6 (10.2) 200 40.6 (9.7) 200 43.1 (9.2) 200 46.1 (9.6) 250 51.0 (9.6)  Late preterm 300 42.2 (10.5) 100 37.6 (9.7) 50 38.9 (12.0) 50 44.1 (8.1) 50 45.1 (9.8) 50 48.9 (8.1)  Moderate preterm 250 40.9 (10.1) 50 35.1 (10.4) 50 39.7 (9.5) 50 43.5 (8.4) 50 42.3 (8.7) 50 46.7 (10.0)  Early preterm 100 38.1 (11.1) –b 31.3 (9.7) –b 33.3 (10.1) 50 41.1 (11.0) 50 38.2 (10.6) –b 45.8 (7.9) Abbreviations: BSF-R, Bayley Short Form–Research Edition; SD, standard deviation. a Unweighted sample sizes are rounded to the nearest 50 per National Center for Education Statistics guidelines. b Counts round to zero. When mutually adjusted, there were significant deficits in BSF-R mental scale score for both preterm birth and lower household SES (Table 3). In fully adjusting models, 0.1- to 0.6-standard-deviation deficits were observed for children born at early preterm (−7.0 points, 95% confidence interval (CI): −8.6, −5.4), moderate preterm (−2.9 points, 95% CI: −4.1, −1.7), and late preterm (−1.3 points, 95% CI: −2.3, −0.3). Children in lower SES quintiles performed 0.2–0.6 standard deviations lower than those in the highest quintile. Estimated deficits for lower household SES compared with the highest SES quintile ranged from −2.5 points (95% CI: −3.5, −1.5) for the second-highest quintile to −6.7 points (95% CI: −7.8, −5.6) for the lowest quintile. There was no statistical evidence of additive interaction between GA and household SES (P = 0.47; Figure 2A). Stratifying by socioeconomic quintile, estimated risk differences for the preterm and early term groups compared with term children were similar across socioeconomic quintiles (Web Table 2). Table 3. Associations Between Gestational Age at Delivery, Socioeconomic Status, and Cognitive Outcome at Age 2 Years and Reading and Mathematics Achievement Outcomes at Kindergarten Age, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Abbreviations: BSF-R, Bayley Short Form–Research Edition; CI, confidence interval. a Adjusted only for child’s age at assessment. b Adjusted for child’s age at assessment, child’s race/ethnicity (non-Hispanic black, Hispanic, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, more than 1 race, non-Hispanic white), child’s sex (female, male), maternal age at delivery (15–17 years, 18–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40 years or older), parity (nulliparous, 1–2 previous live births, 3 or more previous live births), and maternal marital status (married; never married; divorced, widowed, or separated). c Estimated β reflects estimated difference in mean outcome score. Table 3. Associations Between Gestational Age at Delivery, Socioeconomic Status, and Cognitive Outcome at Age 2 Years and Reading and Mathematics Achievement Outcomes at Kindergarten Age, Among Singleton Children Enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008 Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Exposure BSF-R Mental Scale Score (Age 2 Years) Reading Scale Score (Kindergarten Age) Mathematics Scale Score (Kindergarten Age) Model 1a Model 2b Model 1a Model 2b Model 1a Model 2b βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI βc 95% CI Gestational age  Early preterm −7.3 −8.8, −5.7 −7.0 −8.6, −5.4 −5.6 −7.9, −3.2 −5.3 −7.7, −3.0 −6.8 −8.6, −4.9 −6.6 −8.5, −4.7  Moderate preterm −2.7 −3.8, −1.5 −2.9 −4.1, −1.7 −2.4 −3.8, −1.0 −2.6 −3.9, −1.3 −3.2 −4.3, −2.1 −3.2 −4.3, −2.1  Late preterm −1.2 −2.2, −0.2 −1.3 −2.3, −0.3 −1.7 −3.3, −0.1 −1.5 −3.0, 0.0 −1.6 −2.7, −0.5 −1.4 −2.5, −0.4  Early term −0.4 −1.0, 0.2 −0.2 −0.8, 0.4 −0.5 −1.5, 0.5 −0.5 −1.4, 0.5 −0.2 −0.8, 0.4 −0.2 −0.8, 0.5  Full term 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Socioeconomic status  Quintile 1 (lowest) −8.3 −9.2, − 7.3 −6.7 −7.8, −5.6 −16.8 −18.3, −15.2 −12.6 −14.4, −10.9 −12.3 −13.3, −11.3 −9.1 −10.2, −7.9  Quintile 2 −6.9 −7.8, −6.1 −6.0 −7.0, −5.1 −13.9 −15.3, −12.4 −10.5 −12.0, −8.9 −10.0 −10.9, −9.1 −7.5 −8.4, −6.5  Quintile 3 −4.3 −5.2, −3.4 −3.8 −4.8, −2.9 −9.2 −10.7, −7.8 −6.6 −8.0, −5.2 −6.6 −7.6, −5.6 −4.5 −5.5, −3.6  Quintile 4 −2.4 −3.4, −1.4 −2.5 −3.5, −1.5 −6.8 −8.2, −5.4 −5.1 −6.5, −3.7 −4.8 −5.7, −3.9 −3.6 −4.4, −2.7  Quintile 5 (highest) 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent 0.0 Referent Abbreviations: BSF-R, Bayley Short Form–Research Edition; CI, confidence interval. a Adjusted only for child’s age at assessment. b Adjusted for child’s age at assessment, child’s race/ethnicity (non-Hispanic black, Hispanic, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, more than 1 race, non-Hispanic white), child’s sex (female, male), maternal age at delivery (15–17 years, 18–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40 years or older), parity (nulliparous, 1–2 previous live births, 3 or more previous live births), and maternal marital status (married; never married; divorced, widowed, or separated). c Estimated β reflects estimated difference in mean outcome score. Figure 2. View largeDownload slide Model-predicted cognitive and academic achievement scores and 95% confidence intervals derived from models estimating interaction between gestational age and household socioeconomic status for Bayley Short Form–Research Edition (BSF-R) mental scale scores at age 2 years (A), reading scale scores at kindergarten age (B), and mathematics scale scores at kindergarten age (C), among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. All models adjusted for child’s age at assessment (corrected for prematurity for the BSF-R mental scale), child’s race/ethnicity, child’s sex, maternal age at delivery, parity, and maternal marital status. Predicted scores are at mean or reference covariate values. Shaded circle: early preterm; shaded square: moderate preterm; shaded diamond: late preterm; open circle: early term; shaded triangle: term. Figure 2. View largeDownload slide Model-predicted cognitive and academic achievement scores and 95% confidence intervals derived from models estimating interaction between gestational age and household socioeconomic status for Bayley Short Form–Research Edition (BSF-R) mental scale scores at age 2 years (A), reading scale scores at kindergarten age (B), and mathematics scale scores at kindergarten age (C), among singleton children enrolled in the Early Childhood Longitudinal Study, Birth Cohort, United States, 2001–2008. All models adjusted for child’s age at assessment (corrected for prematurity for the BSF-R mental scale), child’s race/ethnicity, child’s sex, maternal age at delivery, parity, and maternal marital status. Predicted scores are at mean or reference covariate values. Shaded circle: early preterm; shaded square: moderate preterm; shaded diamond: late preterm; open circle: early term; shaded triangle: term. Preterm birth and lower household SES were each independently associated with lower kindergarten reading scale scores (Table 3). In fully adjusting models, 0.1- to 0.4-standard-deviation deficits were observed for early preterm (−5.3 points, 95% CI: −7.7, −3.0), moderate preterm (−2.6 points, 95% CI: −3.9, −1.3), and late preterm (−1.5 points, 95% CI: −3.0, 0.0) groups. Children in lower SES quintiles performed 0.3–0.9 standard deviations lower than did those in the highest quintile. Estimated deficits for lower household SES compared with the highest SES quintile ranged from −5.1 points (95% CI: −6.5, −3.7) for the second-highest quintile to −12.6 (95% CI: −14.4, −10.9) for the lowest quintile. The addition of terms for interaction between GA group and household SES was not statistically significant (P = 0.82; Figure 2B). Qualitative examination of stratified results suggested that deficits for early preterm children worsened with lower SES quintile, and deficits for moderate preterm and late preterm were attenuated with lower SES quintile, although there was variation across quintiles (Web Table 2). Preterm birth and lower household SES were also each independently associated with lower mathematics scale scores (Table 3). After adjusting for household SES and other covariates, 0.1- to 0.6-standard-deviation deficits were observed for children born at early preterm (−6.6 points, 95% CI: −8.5, −4.7), moderate preterm (−3.2 points, 95% CI: −4.3, −2.1), and late preterm (−1.4 points, 95% CI: −2.5, −0.4). Children in lower SES quintiles performed 0.3–0.9 standard deviations lower than those in the highest quintile. The estimated deficits for lower household SES compared with the highest SES quintile ranged from −3.6 points (95% CI: −4.4, −2.7) for the second-highest quintile to −9.1 points (95% CI: −10.2, −7.9) for the lowest quintile. There was no statistical evidence of additive interaction between GA and household SES (P = 0.68; Figure 2C). Socioeconomic stratum–specific analyses indicated that deficits for early preterm worsened slightly with lower SES quintile, and deficits for moderate preterm and late preterm were attenuated (Web Table 2). For all 3 outcomes, 3-way interaction terms with race/ethnicity were not significant, and there was no significant interaction between GA and household SES in any racial/ethnic group in analyses stratified by race/ethnicity. Findings were similar to our main analyses when using measurements of GA with last menstrual period, using alternative SES measures (including using the wave-2 measurement of SES quintile), and applying complex sample weights. In the 2-year analyses, not correcting for prematurity resulted in larger estimated deficits for preterm groups, while in the kindergarten analyses, correcting for prematurity resulted in attenuated estimated deficits for preterm groups. DISCUSSION Preterm birth and lower household SES were each associated with substantial deficits in children’s cognitive scores at 2 years of age and in reading and mathematics achievement at kindergarten age. While household SES did not modify the associations between preterm birth and cognitive outcomes, in the absence of additive interaction the 2 factors separately and adversely affected children’s scores. Score deficits associated with being exposed to both preterm birth and lower household SES were approximately equal to the sum of the estimated deficits associated with the 2 separate exposures. Children exposed to both preterm birth and lower household SES performed at the lowest levels on all outcomes. Prior studies on additive interactions between preterm birth and SES in their associations with cognitive and academic outcomes have found either that lower SES exacerbated the adverse impacts of preterm birth (13–15) or that there was no evidence of interaction between the 2 factors (16). Two of these studies included assessments of children at similar ages to those investigated in our study—early childhood and around the time of starting elementary school (15, 16). Richards et al. (15) found that both preterm birth and living in a more-deprived neighborhood were associated with higher risk of failing first-grade statewide tests among 330,000 public school children in the state of Georgia. There was also positive additive interaction between the 2 factors, such that living in a more-deprived neighborhood exacerbated the adverse impact of preterm birth on test failure (15). It is possible that neighborhood-level deprivation captured not only families’ socioeconomic status but also community resources and school quality, which may contribute to exacerbating impacts of preterm birth on children’s school-related outcomes. Further, it is possible that performance on statewide competency tests was affected not only by students’ cognitive skills but also by school quality and climate as well as noncognitive skills such as behavior and socioemotional well-being. Brown et al. (16) found no evidence of modification by parenting factors—quality of interactions, effectiveness, and consistency—of the association between late preterm or early term birth and cognitive outcomes among 15,000 Canadian children enrolled in a national longitudinal study. The results of their study may be driven by its inclusion of only children born at 34 weeks or later; our study similarly found small cognitive deficits for late-preterm children and no significant deficit for early-term children. While our analysis found that household SES did not exacerbate the association between preterm birth and worsened cognitive outcomes, it must be emphasized that lower household SES was still found to independently and adversely influence children’s cognitive scores at both age 2 years and kindergarten age. These findings echo previous research linking childhood poverty with adverse impacts on cognitive development and academic performance (2, 26). It may be possible that impacts of poverty on children’s cognitive and academic outcomes are so substantial that the additional influence of preterm birth may reach a bottom “threshold” in terms of affecting scores. In our study, estimated disparities comparing the highest and lowest socioeconomic groups were larger than those comparing early preterm with term children and term-born children in the lowest SES quintile had predicted scores lower than most preterm groups in the highest SES quintile. A major strength of the present study is the utilization of a large, population-based sample of US children with rigorous, prospective measurements of cognitive development at multiple time points. Using data from a population-based cohort for studying outcomes of preterm birth, however, comes with the caveat that it may underrepresent children who are more ill and unable or unwilling to participate in such a study. We were able to analyze the outcomes of the same group of children as they grew and developed through early childhood to kindergarten, although we were not able to analyze trajectories of cognitive growth due to noncongruity of the 2-year and kindergarten outcomes. It is important to acknowledge the limitations of this study in assessing potential complexities of pathways through which preterm birth and SES may separately and jointly influence children’s cognitive outcomes. The present analysis was not able to assess pathways involving prenatal SES. It is possible that the impacts of SES may be mediated through preterm birth and also, postnatally, may modify the associations between preterm birth and cognitive outcomes. Studies with both prenatal and postnatal SES measurements are needed to disentangle these pathways. Experiencing a preterm birth may affect family’s SES through incurring medical costs and forcing changes in income and employment due to increased needs for caregiving (27). While we were not able to assess whether such changes in SES occurred, the composite SES index—reflecting income, educational attainment, and occupation—was plausibly more stable against this concern compared with using only household income to measure SES. Although there was a large overall sample size, some GA-SES exposure groups had sparse data, especially for the early and moderate preterm groups and in analyses stratified by race/ethnicity. Because of the known socioeconomic disparities across racial/ethnic groups in the United States, future studies should specifically examine whether relationships among preterm birth, SES, and cognitive development vary across racial/ethnic groups. About 85% and 61% of eligible children were included in the 2-year and kindergarten analyses, respectively. However, the analytical samples were similar to the eligible target population in terms of gestational age, SES, and demographic factors. Dropout rates were exaggerated because about half of those lost between the 2 waves were part of a planned sample-size reduction carried out by systematically reducing sampling rates (17). In our study, preterm birth and lower SES were associated with substantial deficits in cognitive scores at 2 years and in academic achievement scores at kindergarten age, although there was no evidence of additive interaction between the 2 exposures. However, even in the absence of interaction, both factors had important impacts on cognitive scores. Given that preterm birth and childhood poverty are prevalent in the United States and often overlap, more research is needed to better understand the independent and joint associations of preterm birth and childhood poverty with children’s development. ACKNOWLEDGMENTS Author affiliations: Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia (Jennifer L. Beauregard, Carolyn Drews-Botsch, W. Dana Flanders, Michael R. Kramer); and Department of Behavioral Sciences and Health Education, Rollins School of Public Health, Emory University, Atlanta, Georgia (Jessica M. Sales). J.L.B. received support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, T32 Predoctoral Training Program in Reproductive, Perinatal, and Pediatric Epidemiology under (award T32HD052460), and from the Health Resources and Services Administration of the US Department of Health and Human Services (grant T03MC07651). M.R.K. received support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (award K01-HD074726). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Health Resources and Services Administration, Department of Health and Human Services, or the US Government. Conflict of interest: none declared. Abbreviations BSF-R Bayley Short Form–Research Edition CI confidence interval ECLS-B Early Childhood Longitudinal Study, Birth Cohort GA gestational age SD standard deviation SES socioeconomic status REFERENCES 1 Behrman RE , Butler AS , eds. Preterm Birth: Causes, Consequences, and Prevention . Washington, DC : National Academies Press ; 2007 . http://www.ncbi.nlm.nih.gov/books/NBK11362/. Accessed August 18, 2014. 2 Brooks-Gunn J , Duncan GJ . The effects of poverty on children . Future Child . 1997 ; 7 ( 2 ): 55 – 71 . Google Scholar CrossRef Search ADS PubMed 3 Conti G , Heckman JJ . Understanding the early origins of the education-health gradient: a framework that can also be applied to analyze gene-environment interactions . Perspect Psychol Sci . 2010 ; 5 ( 5 ): 585 – 605 . Google Scholar CrossRef Search ADS PubMed 4 Heckman JJ . Skill formation and the economics of investing in disadvantaged children . Science . 2006 ; 312 ( 5782 ): 1900 – 1902 . Google Scholar CrossRef Search ADS PubMed 5 Hamilton BE , Martin JA , Osterman MJ , et al. . Births: final data for 2014 . Natl Vital Stat Rep . 2015 ; 64 ( 12 ): 1 – 64 . 6 Addy S , Engelhardt W , Skinner C . Basic Facts About Low-Income Children. New York, NY : National Center for Children in Poverty ; 2013 . http://www.nccp.org/publications/pub_1076.html. Accessed November 18, 2014. 7 Burd I , Balakrishnan B , Kannan S . Models of fetal brain injury, intrauterine inflammation, and preterm birth . Am J Reprod Immunol . 2012 ; 67 ( 4 ): 287 – 294 . Google Scholar CrossRef Search ADS PubMed 8 Stiles J , Jernigan TL . The basics of brain development . Neuropsychol Rev . 2010 ; 20 ( 4 ): 327 – 348 . Google Scholar CrossRef Search ADS PubMed 9 Johnson SB , Riis JL , Noble KG . State of the art review: poverty and the developing brain . Pediatrics . 2016 ; 137 ( 4 ): e20153075 . Google Scholar CrossRef Search ADS PubMed 10 Bendersky M , Lewis M . Environmental risk, biological risk, and developmental outcome . Dev Psychol . 1994 ; 30 ( 4 ): 484 – 494 . Google Scholar CrossRef Search ADS 11 Sameroff AJ . Environmental context of child development . J Pediatr . 1986 ; 109 ( 1 ): 192 – 200 . Google Scholar CrossRef Search ADS PubMed 12 Messer LC , Vinikoor LC , Laraia BA , et al. . Socioeconomic domains and associations with preterm birth . Soc Sci Med . 2008 ; 67 ( 8 ): 1247 – 1257 . Google Scholar CrossRef Search ADS PubMed 13 Wang WL , Sung YT , Sung FC , et al. . Low birth weight, prematurity, and paternal social status: impact on the basic competence test in Taiwanese adolescents . J Pediatr . 2008 ; 153 ( 3 ): 333 – 338 . Google Scholar CrossRef Search ADS PubMed 14 Ekeus C , Lindström K , Lindblad F , et al. . Preterm birth, social disadvantage, and cognitive competence in Swedish 18- to 19-year-old men . Pediatrics . 2010 ; 125 ( 1 ): e67 – e73 . Google Scholar CrossRef Search ADS PubMed 15 Richards JL , Chapple-McGruder T , Williams BL , et al. . Does neighborhood deprivation modify the effect of preterm birth on children’s first grade academic performance? Soc Sci Med . 2015 ; 132 : 122 – 131 . Google Scholar CrossRef Search ADS PubMed 16 Brown HK , Speechley KN , Macnab J , et al. . Mild prematurity, proximal social processes, and development . Pediatrics . 2014 ; 134 ( 3 ): e814 – e824 . Google Scholar CrossRef Search ADS PubMed 17 Snow , K , Derecho , A , Wheeless , S , et al. . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Kindergarten 2006 and 2007 Data File User’s Manual (2010-010). Washington, DC: National Center for Education Statistics . 2009 . 18 Alexander GR , Himes JH , Kaufman RB , et al. . A United States national reference for fetal growth . Obstet Gynecol . 1996 ; 87 ( 2 ): 163 – 168 . Google Scholar CrossRef Search ADS PubMed 19 VanderWeele TJ , Robins JM . Four types of effect modification: a classification based on directed acyclic graphs . Epidemiology . 2007 ; 18 ( 5 ): 561 – 568 . Google Scholar CrossRef Search ADS PubMed 20 Guo G . The timing of the influences of cumulative poverty on children’s cognitive ability and achievement . Soc Forces . 1998 ; 77 ( 1 ): 257 – 287 . Google Scholar CrossRef Search ADS 21 National Center for Education Statistics . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), User’s Manual for the ECLS-B Longitudinal 9-month-2-year Data File and Electronic Codebook. Washington, DC : National Center for Education Statistics ; 2006 . 22 Andreassen C , Fletcher P . Early Childhood Longitudinal Study, Birth Cohort (ECLS–B) Psychometric Report for the 2-Year Data Collection (NCES 2007-084). Washington, DC : National Center for Education Statistics ; 2007 . http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2007084. Accessed March 11, 2015. 23 Engle WA ; American Academy of Pediatrics Committee on Fetus and Newborn . Age terminology during the perinatal period . Pediatrics . 2004 ; 114 ( 5 ): 1362 – 1364 . Google Scholar CrossRef Search ADS PubMed 24 Najarian M , Snow K , Lennon J , et al. . Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Preschool—Kindergarten 2007 Psychometric Report. Washington, DC : National Center for Education Statistics ; 2010 . https://nces.ed.gov/pubs2010/2010009.pdf. 25 Rehkopf DH , Glymour MM , Osypuk TL . The consistency assumption for causal inference in social epidemiology: when a rose is not a rose . Curr Epidemiol Rep . 2016 ; 3 ( 1 ): 63 – 71 . Google Scholar CrossRef Search ADS PubMed 26 Duncan GJ , Yeung WJ , Brooks-Gunn J , et al. . How much does childhood poverty affect the life chances of children? Am Sociol Rev . 1998 ; 63 ( 3 ): 406 . Google Scholar CrossRef Search ADS 27 Hodek JM , von der Schulenburg JM , Mittendorf T . Measuring economic consequences of preterm birth—methodological recommendations for the evaluation of personal burden on children and their caregivers . Health Econ Rev . 2011 ; 1 : 6 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

American Journal of EpidemiologyOxford University Press

Published: Aug 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off