Severe small size for gestational age and cognitive function: catch-up phenomenon possible




Objective


We sought to compare cognitive functioning in children born with birthweight <3% vs ≥3% for gestational age (GA) between 9 months and kindergarten.


Study Design


Nonanomalous singletons from the Early Childhood Longitudinal Study–Birth Cohort were included. Associations between weight for GA and cognitive functioning were examined using a series of confounder-adjusted general linear models.


Results


Of 3633 cohorts, 585 (16%) were <3% for GA. At 9 months, cognitive performance of newborns <3% was about 12 percentile points lower than their normal counterparts ( P < .001). By 2 years, however, no significant cognitive differences between these groups were observed ( P = .668). Academic performance at preschool age (around 3.5 years) was not different for reading ( P = .245) or math ( P = .880), nor different at kindergarten age.


Conclusion


Newborns <3% for GA exhibit catch-up cognitive functioning by 2 years, with relatively no decrements in academic functioning observed by kindergarten.


Severe small size for gestational age (SGA), defined as birthweight <3% for gestational age (GA), is linked with neonatal morbidity such as neonatal seizures, proven sepsis, need for intubation, retinopathy, and mortality within 28 days of birth. Additionally, the Practice Bulletin Intrauterine Growth Restriction from the American Congress of Obstetricians and Gynecologists (ACOG) acknowledges that while normal infants with SGA will exhibit “catch-up” growth by 2 years, they are at 2-fold increased risk for major neurologic sequelae. The link between abnormal growth and cognitive delay is based on a comparison of 714 neonates of appropriate size for age with 347 SGA neonates, derived from several studies. The potential problem with linking neurologic sequelae with suboptimal growth is that the neurologic sequelae vary across studies, as do the definitions of growth restriction and the assessments used for neurologic or cognitive functioning. Thus, there is a need to reassess the association between growth restriction and cognitive development using consistent standardized measurements and methods.




For Editors’ Commentary, see Contents



The purpose of this study was to utilize the nationally representative Early Childhood Longitudinal Study–Birth Cohort (ECLS-B) data to determine if newborns at <3% for GA, when compared to those ≥3%, are at risk for sustained cognitive deficits and subsequent problems with academic achievement. For this analysis we sought to assess the difference between <3% vs ≥3% because the likelihood of morbidity is significantly greater when birthweight is <3% rather than 4-10%.


Materials and Methods


The ECLS-B (US Department of Education) is a national multisource, multimethod study, conducted by the National Center for Education Statistics, designed to evaluate the influences on children’s early development, school readiness, and early academic performance. The sample was drawn from children born in 2001, with an oversampling of specific minority groups (eg, American Indian and Asian and Pacific Islander infants), low birthweight infants, and twins, which allowed for more statistical power in examining outcomes related to these subpopulations. Births were sampled using a 2-stage cluster-based system developed by the National Center for Health Statistics. It uses a nationwide cluster sampling strategy of birth certificate data to select a representative sample of infants born in 2001. The ECLS-B utilized an oversampling strategy that included low and very low birthweight infants, creating an overall sample that allowed for more statistical power in examining outcomes related to these subpopulations. Participants were followed through their kindergarten year, with data collection waves occurring at 9 months, 2 years, 3.5 years, and 5 years.


During the 9-month and 2-year visits, participants were administered the research edition of the Bayley Scales of Infant Development (BSID). This test of early child cognitive and motor development is age-normed and designed for children from early infancy to 3 years of age. It produces 2 subscale scores: the Motor score taps fine and gross motor skills; and the Mental (MNT) score taps expressive and receptive language, reasoning and problem-solving, and emotion regulation. Since the focus of our study was cognitive function, we used the MNT score in analyses.


Generally, tests of abilities or characteristics must show a consistency in measurement within the test and between administrations of the tests, with coefficients measuring these criteria being considered good when above a threshold of about 0.70. Within the ECLS-B, the research edition of the BSID demonstrated good internal reliability (MNT score, coefficient alpha = 0.78-0.93) as well as good test-retest reliability (MNT score, r = 0.83). Correlations between the BSID and the McCarthy Scales of Children’s Abilities, another well-regarded early childhood cognitive assessment, ranged from 0.57–0.79.


At the preschool (3.5 years) and kindergarten (5 years) waves of the ECLS-B, participants were administered direct cognitive tests that tapped early reading and math abilities. These cognitive assessments were developed for the ECLS-B using item-response theory and adaptive testing to allow for accurate measurement of children with widely varying abilities. The reading portion of this assessment taps basic skills, vocabulary, global understanding of context, interpretation of stories, and overall comprehension. The math portion of the assessment taps sense of numbers, measurement, spatial sense, understanding of probability, and identification of patterns. As the ECLS-B team developed these tests, several rounds of testing and refinement were used to generate tests that ultimately showed good internal consistency (reading reliability coefficient = 0.84-0.93; math = 0.89-0.92).


From the birth certificate, GA and birthweight were obtained. Using the normative birthweight conversion tables published by Oken et al, birthweight for GA was calculated as a percentile for each case. We did not use the more frequently cited national reference published by Alexander et al because it categorized newborns as <5% or <10% for GA, and not <3%, the threshold for significantly higher neonatal morbidity.


A series of multivariate general linear models (GLM) that adjusted for the complex sampling design of the ECLS-B were fitted to examine the overall differences in developmental trajectories of cognitive functioning between newborns with weight <3% vs ≥3% for GA. Additional univariate GLM at each time point were used to explore this further. All models were adjusted to control for child race (categorized as Caucasian, African American, Hispanic, and other), child sex, maternal education (years of school), maternal socioeconomic status (a composite index of occupational status, income, and education), and prematurity. All analyses were conducted using the standard and complex samples modules of SPSS software (version 19; SPSS Inc, Chicago, IL). Because the ECLS-B dataset used in this study contains no information that can be used to identify subjects, the Institutional Review Board at Eastern Virginia Medical School reviewed this study and determined that it does not fall under human subjects research regulatory requirements.




Results


The ECLS-B sampled 13,921 live births in 2001 and 10,688 newborns completed the first major data collection point, when the child was 9 months old. A subsampling scheme determined by the ECLS-B study team yielded a total of 6420 children who completed the kindergarten wave, which represents a 5-year retention rate of 74% for the selected subsample. We excluded 2787 newborns for the following reasons: 1643 twins, 599 with incomplete birth certificate data, 516 missing cognitive and academic data, and 29 anomalous. There were no significant differences in SGA rates between cases retained and lost to follow-up (16% vs 15.8%; P = .72). Our final sample size for analysis was 3633 children ( Figure 1 ).




FIGURE 1


Case selection

F/U, follow-up.

*1643 twins, 599 with incomplete birth certificate data, 516 who were missing data, and 29 anomalous newborns.

Paulson. Catch-up phenomenon in SGA. Am J Obstet Gynecol 2012.


Among the 3633 children, 51% were male with a racial distribution of: 42% (1516) white, 15% (567) Asian, 15% (554) Hispanic, 14% (497) African American, and 14% (499) others. Preterm births comprised 9.2% of our sample and 16% (581) were SGA, consistent with the ECLS-B oversampling strategy. Among infants who were SGA, 8.4% were born <37 weeks’ gestation, vs 1.7% among newborns with weight of ≥3% ( P < .001). Moreover, severe SGA children were more likely to have 5-minute Apgar scores <7 (1.3%) than those with weight of 3% or more (0.6%; P = .038). No differences were observed between severe SGA and children who weighed >3% for GA on race/ethnicity, maternal age, or route of delivery ( Table 1 ). Comparisons between severe SGA vs newborns >3% indicate significantly poorer cognitive functioning (by 12 percentile points) at age 9 months, but no significant differences in cognitive or academic performance at later time points ( Table 2 ).



TABLE 1

Sample characteristics























































































































Characteristic BW <3% for GA, n = 581 BW ≥3% for GA, n = 3025 P value
Maternal age, y 28.7 (28.2–29.3) 29.3 (29.0–29.5)
<19 2.0% (22) a 1.8% (80) .767
19-34 75.8% (436) 78.0% (2343)
≥35 22.2% (127) 20.2% (633)
Ethnicity
African American 15.7% (118) 12.1% (379) .155
Caucasian 54.4% (270) 61.3% (1246)
Hispanic 19.1% (84) 17.8% (470)
Asian 6.5% (66) 3.9% (501)
Other 4.2% (47) 5.0% (452)
GA, wk 37.6 (37.4–37.9) 39.4 (39.37–39.52)
24-28 0 0.1% (30) < .001
29-32 2.3% (58) 0.4% (73)
33-34 2.9% (72) 0.4% (21)
34-36 3.1% (45) 0.7% (28)
≥37 91.6% (410) 98.3% (2835)
Route of delivery
Vaginal 74.1% (347) 76.1% (2284) .536
Cesarean 25.9% (237) 23.9% (729)
Apgar score at 5 min
<7 1.3% (27) 0.6% (41) .038
≥7 98.7% (558) 99.4% (2987)

Data presented as % (n) or means and 95% confidence intervals (for maternal age and GA).

BW, birthweight; GA, gestational age.

Paulson. Catch-up phenomenon in SGA. Am J Obstet Gynecol 2012.

a Actual cell counts reported. Percentages and χ 2 reflect weighted rates.


Tags:
May 15, 2017 | Posted by in GYNECOLOGY | Comments Off on Severe small size for gestational age and cognitive function: catch-up phenomenon possible

Full access? Get Clinical Tree
Get Clinical Tree app for offline access