The association between infant birthweight, head circumference and neurodevelopmental outcomes

Abstract

Objective:

To evaluate whether being small (SGA) or large for gestational age (LGA) or having a small or large head circumference (HC) at birth is associated with adverse neurodevelopmental outcomes.

Study Design:

Secondary analysis of a multicenter negative randomized trial of thyroxine therapy for subclinical hypothyroid disorders in pregnancy. The primary outcome was child intelligence quotient (IQ) at 5 years of age. Secondary outcomes included several neurodevelopmental measures. Associations between the outcomes in children with SGA (<10^th percentile) or LGA (>90^th percentile) birthweights, using ethnicity- & sex-specific population nomogram as well as nomograms from the National Fetal Growth (NFG) study, were compared with the referent of those with AGA birthweight. Similar analyses were performed for HC.

Results:

Using the population nomogram, 90 (8.2%) were SGA and 112 (10.2%) were LGA. SGA neonates were more likely to be born preterm to mothers who were younger, smoked, and less likely to have less than a high school education, whereas LGA neonates were more likely to be born to mothers who were older and have higher body mass index, compared with AGA neonates. SGA at birth was associated with a decrease in the child IQ at 5 years of age by 3.34 (95% CI 0.54 – 6.14) points, and an increase in odds of child with an IQ <85 (aOR 1.9, 95% CI 1.1–3.2). There was no association between SGA and other secondary outcomes, or between LGA and the primary or secondary outcomes. Using the NFG standards, SGA at birth remained associated with a decrease in the child IQ at 5 years of age by 3.14 (95% CI 0.22 – 6.05) points and higher odds of an IQ <85 (aOR 2.3, 95% CI 1.3 – 4.1), but none of the other secondary outcomes. HC was not associated with the primary outcome and there were no consistent associations of these standards with the secondary outcomes.

Conclusions:

In this cohort of pregnant individuals with hypothyroid disorders, SGA birthweight was associated with a decrease in child IQ and greater odds of child IQ < 85 at 5 years of age. Using a fetal growth standard did not appear to improve the detection of newborns at risk of adverse neurodevelopment.

Précis:

In a cohort of pregnant individuals with subclinical hypothyroid disorders, small for gestational age birthweight was associated with a decrease in child IQ at 5 years of age.

Introduction

Disturbances in fetal growth are associated with increased neonatal and infant morbidities as well as adverse long-term health consequences.^1–4 Neonates born small for gestational age (SGA), assessed by birthweight or aberrant head circumference (HC) measurement after birth, are at increased risk of adverse perinatal outcomes including low pH and Apgar scores, need for emergency cesarean delivery, neonatal intensive care admission, cerebral palsy, and neonatal death.^5–8 Those born large for gestational age (LGA) are also at increased risk of cesarean delivery, birth trauma, and neonatal intensive care admissions.^9,10 Many of these adverse perinatal outcomes are associated with long-term neurodevelopmental and cognitive adverse outcomes.^3–5,8,11

Traditionally, evaluation of fetal growth has been accomplished by comparing fetal weight or birthweight and infant HC to population-based norms.^12–15 These population norms may not reflect the current demographics of the US, and may misclassify fetal growth, as they are typically based on birthweight and measurement of the HC, which requires a delivery. A neonate that needed delivery is not reflective of the normal status of a fetus in an ongoing pregnancy. In addition, they do not account for the fact that preterm born neonates are more likely to be growth restricted.^12,14,15 Recently, the NICHD reported the results of new contemporary racial and ethnic standards for fetal growth. These national fetal growth standards were based on prospectively followed 1,737 fetuses born to healthy low-risk pregnant individuals of different racial and ethnic backgrounds, and the longitudinal evaluations of fetal growth was performed using ultrasound. Racial and ethnic-specific fetal growth curves were created.¹⁶ Whether these nomograms, which may reflect fetal growth better than birthweight nomograms, improve the detection of abnormal fetal growth that is associated with long-term adverse outcomes, remains to be investigated.

Therefore, our objectives in this study were 1) to determine whether SGA or LGA newborns and those with aberrant HC are at risk of lower child intelligence quotient (IQ) at 5 years of age and other adverse neurodevelopmental outcomes compared with those of normal size, and 2) to determine whether assessing birthweight and HC using a population nomogram^12,13 or nomograms from the National Fetal Growth study¹⁶, improve the detection of newborns at risk of adverse neurodevelopmental outcomes compared with those of normal size. We hypothesize that SGA and LGA newborns and those with aberrant HC are at risk of adverse neurodevelopmental outcomes compared with those with normal growth and that there are no differences in the association using population or National Fetal Growth study nomograms.

Materials & Methods

Study Design

We performed a secondary analysis of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network study of thyroxine therapy for subclinical hypothyroidism or hypothyroxinemia diagnosed during pregnancy (www.clinicaltrials.gov NCT00388297). ¹⁷ The details of the study, which was conducted at 15 centers in the U.S. between October 2006 and October 2009, are described elsewhere. ¹⁷ Briefly, it consisted of two multi-center double-blinded placebo – controlled trials conducted in parallel, in which a total of 1023 pregnant individuals, with singleton gestation, diagnosed with subclinical hypothyroidism or hypothyroxinemia during pregnancy were randomized to daily treatment with either thyroxine supplement in capsule form or matching placebo. The primary outcome of the trial was the child’s IQ at 5 years of age, and therapy did not influence neurodevelopment; therefore, treatment groups were combined for this secondary analysis. Overall, the five year follow-up rate was ~90%. We excluded patients whose pregnancy was complicated by congenital fetal anomaly, antepartum fetal demise, neonatal death, and those for which child neurodevelopmental outcomes, birthweight or HC measurements were not available or not realistic (attributed to documentation errors). The original protocol was approved by the Institutional Review Board at each center, and this secondary analysis was deemed exempt by the institutional review board.

Study outcomes

Our primary outcome was the same as the original trial primary paper - child intelligence quotient (IQ) at 5 years of age assessed using the WPPSI–III Full Scale test. Secondary outcomes included: cognitive, motor, and language scale scores from the Bayley Certified Scales of Infant Development III at 12 and 24 months corrected age; cognitive and achievement levels from the Differential Ability Scales (DAS II) at 36 months; behavioral and social competencies from the Child Behavior Checklist (CBCL) at 36 and 60 months of age; cognitive and achievement levels from two DAS II subtests, recall of digits forward and recognition of pictures, and the ADHD index score from the Connors’ Rating Scale(S) – Revised at 48 months of age; and selected cognitive abilities from the subscales of the Wechsler Preschool and Primary Scale of Intelligence (WPPSI-III) at 60 months of age.

Statistical Analysis

Birthweight and HC were obtained immediately after birth. Aberrant birthweight (SGA, birthweight <10% and LGA, birthweight >90%) was assessed using an ethnicity- & sex-specific population birthweight nomogram (Alexander)¹³ and aberrant HC (<10% and >90%) using a population nomogram (Lubchenco) ¹² as well as those from the National Fetal Growth study.¹⁶ Baseline maternal and neonatal data were compared between groups using the Chi-square or Fisher exact test for categorical data and the Wilcoxon Rank Sum test for continuous data as appropriate. The odds of developing the primary and secondary outcomes were calculated for SGA and LGA using children born appropriate for gestational age (AGA birthweight 10–90% and HC 10–90%, respectively) as a reference group. Odds ratio (OR) with 95% CI for the various primary and secondary outcomes were calculated. The association between aberrant growth and neurodevelopmental outcomes were then assessed using logistic and linear regressions as appropriate adjusting for the TSH stratum and other factors considered to be potential confounders, including mother’s education, insurance type, smoking status, maternal age, parity, infant sex, gestational age at birth, treatment group assignment and race and ethnicity. Analyses were performed for birth weight and head circumference. A two-sided p-value of less than 0.05 was considered to indicate statistical significance, and no adjustment for multiple testing was performed as these analyses were considered exploratory. Statistical analyses were performed using SAS statistical software (SAS Institute, Inc, Cary, NC).

Results

1096 participants were included in this analysis and were allocated to levothyroxine (n=555) or placebo (n=541). Using the population nomogram, 90 (8.2%) were SGA and 112 (10.2%) were LGA. Based on this classification, SGA neonates were more likely to be born preterm to mothers who were younger mothers, smoked, and less likely to have less than a high school education compared with AGA neonates. Overall, there was no difference in baseline maternal thyroid status nor infant sex between the groups for SGA. (Table 1)

Table 1.

Maternal and neonatal characteristics of patients enrolled in this analysis according to birthweight status using population nomogram.

	SGA (n=90)	AGA (n=894)	LGA (n=112)	P value Overall	P value SGA vs AGA	P value LGA vs AGA
Maternal age (years)	26.4±6.1	27.6±5.7	29.3±5.8	<0.001	0.04	0.003
Ethnicity/Race				0.06	0.14	0.13
Hispanic	48.9	53.6	59.8
Non-Hispanic white	41.1	31.5	22.3
Non-Hispanic black	10.0	14.9	17.9
Pre-pregnancy BMI (kg/m2)	27.7±5.8	29.0±6.6	31.4±7.2	<0.001	0.08	<0.001
Source of medical insurance				0.35	0.75	0.15
Government assisted	58.9	57.2	66.1
Private	30.0	28.9	20.5
Self-pay	11.1	14.0	13.4
Smoking during pregnancy	20.0	6.5	6.3	<0.001	<0.001	0.92
Education <high school	37.8	46.0	47.3	0.02	0.02	0.13
Treatment group				0.65	0.42	0.59
Levothyroxine	54.4	50.0	52.7
Placebo	45.6	50.0	47.3
Baseline thyroid status:				0.07	0.88	0.02
Subclinical hypothyroidism	57.8	56.9	45.5
Hypothyroxinemia	42.2	43.1	54.5
Gestational age at delivery, weeks	38 [37–39]	39 [38–40]	39 [39–40]	<0.001	<0.001	<0.001
Gestational age at delivery < 37 weeks	15.6	7.7	1.8	0.001	0.01	0.02
Mode of delivery				0.002	<0.001	0.07
Vaginal	56.7	73.3	65.2
Cesarean	43.3	26.7	34.8
Nulliparous	34.4	32.6	23.2	0.11	0.72	0.05
Male sex	47.8	51.2	53.6	0.71	0.53	0.64
Apgar score 5 minutes < 7	2.2	0.56	0	0.17	0.13	1.00

BMI = Body mass index; SGA = small for gestational age; AGA = appropriate for gestational age; LGA = large for gestational age.

Data are reported as mean± standard deviation, median [IQR], or %.

SGA at birth was associated with decrease in the child IQ at 5 years of age (primary outcome) by 3.34 (95% CI 0.54 – 6.14) points and increase in the odds of child with IQ <85 (aOR 1.9, 95% CI 1.1–3.2). (Table 2) There was no association between SGA and other secondary outcomes, (Table 3) nor a significant interaction between SGA and thyroid status for any of the outcomes. Additionally, there was no association between LGA and the primary (−0.21 points, 95% CI −2.74 – 2.31) or any secondary outcomes. (Tables 2, 3)

Table 2:

Primary outcome and rates of primary outcome (child IQ) and IQ <85 for aberrant birthweight by different nomograms

	Birtdweight (Alexander)
	SGA (n=88)	AGA (n=875)	LGA (n=110)	P	SGA vs. AGA $ #	P	LGA vs. AGA $ #	P
Primary Outcome	91.9 ± 15.2	95.1 ± 14.9	93.6 ± 13.7	0.56	β −3.34 (95% CI −6.14, −0.54)	0.02	β −0.21 (−2.74, 2.31)	0.87
Primary Outcome < 85	31 (35.2)	213 (24.3)	27 (24.6)	0.08	aOR 1.91 (1.13, 3.22)	0.02	aOR 0.93 (0.57, 1.53)	0.79
	Birthweight (National Fetal Growth)
	SGA (n=82)	AGA (n=777)	LGA (n=82)	P	SGA vs. AGA $ #	P	LGA vs. AGA $ #	P
Primary Outcome	93.5 ± 15.6	95.2 ± 14.7	91.8 ± 13.0	0.46	β −3.14 (95% CI −6.05, −0.22)	0.04	β −0.03 (−2.91, 2.84)	0.98
Primary Outcome < 85	29 (35.4)	179 (23.0)	24 (29.3)	0.03	aOR 2.33 (1.33, 4.11)	0.003	aOR 1.07 (0.63, 1.84)	0.80

Data are presented as mean ± standard deviation or n (%).

^*The primary outcome was child intelligence quotient (IQ) at 5 years of age assessed using the WPPSI–III Full Scale test. Data for the primary outcome were missing on 23 children using the Alexander nomogram and 18 children for the National Fetal Grwoth standards.

^$For binary outcomes, the adjusted OR (95% CI) from logistic models is presented. For continuous outcomes, the regression coefficient (β, 95% CI) from linear regression models is presented and reflects the adjusted difference in score (i.e. score for aberrant birthweight – score for AGA).

^# Regression models adjusted for gestational age at delivery, race-ethnicity, maternal age, parity, education, type of insurance, smoking status, baseline thyroid status (subclinical hypothyroidism or hypothyroxinemia), treatment group, and child sex.

Table 3:

Secondary outcomes for aberrant birthweight by different nomograms

	Birthweight (Alexander)					Birthweight (National Fetal Growth)
	SGA	AGA	LGA	SGA vs. AGA	LGA vs. AGA	SGA	AGA	LGA	SGA vs. AGA	LGA vs. AGA
One Year Exam
Bayley-III
Cognitive score	100.2±12.7	99.5±13.3	99.4±12.3	β −0.14 (95% CI −2.88, 2.61)	β 0.52 (95% CI −2.01, 3.05)	100.0±12.2	99.7±13.0	97.5±11.5	−1.34 (−4.15, 1.47)	−0.37 (−3.24, 2.50 )
Cognitive score < 85	10.2	8.6	7.7	1.56 (0.72, 3.36)	0.79 (0.36, 1.74)	10.8	7.6	10.4	2.20 (0.98, 4.95)	1.26 (0.56, 2.86)
Motor score	96.6±11.2	97.0±11.6	96.2±10.8	β −0.54 (95% CI −3.08, 1.99)	β −1.22 (95% CI −3.55, 1.11)	97.0±10.4	97.1±11.3	95.6±11.0	−0.69 (−3.31, 1.94)	−1.17 (−3.85, 1.51)
Motor score < 85	12.6	8.8	13.5	1.45 (0.72, 2.95)	1.71 (0.90, 3.24)	9.8	8.7	13.0	1.19 (0.52, 2.75)	1.36 (0.64, 2.89)
Language score	96.1 ±14.1	94.9±12.5	93.5±12.2	β 1.03 (95% CI −1.74, 3.80)	β −0.88 (95% CI −3.43, 1.67)	95.8±14.5	94.6±12.4	92.8±12.3	−0.36 (−3.25, 2.54)	−0.48 (−3.44, 2.49)
Language score < 85	20.5	17.6	19.2	1.18 (0.67, 2.09)	1.05 (0.62, 1.80)	19.3	17.9	20.8	1.17 (0.63, 2.16)	0.98 (0.54, 1.79)
Two Year Exam
Bayley-III
Cognitive score	91.8 ±12.5	90.2 ±12.4	89.8 ±12.4	β 0.48 (95% CI −2.02, 2.98)	β 1.03 (95% CI –1.19, 3.24)	92.6 ±13.0	90.6 ±12.5	87.6 ±10.5	−0.30 (−2.88, 2.28)	−0.04 (−2.58, 2.51)
Cognitive score < 85	13.8	24.3	24.0	0.58 (0.29, 1.17)	0.84 (0.50, 1.41)	14.5 p.142	23.7	29.0	0.81 (0.39, 1.67)	0.95 (0.54, 1.68)
Motor score	98.0 ±10.5	97.3 ±12.2	96.8 ±11.2	β 0.06 (95% CI −2.54, 2.67)	β 0.65 (95% CI −1.65, 2.96)	98.7 ±10.2	97.6 ±12.0	95.5 ±12.4	−0.87 (−3.56, 1.82)	0.05 (−2.57, 2.68)
Motor score < 85	6.3	10.1	12.6	0.61 (0.23, 1.61)	1.11 (0.57, 2.16)	5.3	9.2	16.9	0.66 (0.22, 2.04)	1.73 (0.85, 3.54)
Language score	91.8 ±17.2	90.8 ±15.1	90.3 ±12.5	β 0.02 (95% CI −2.86, 2.89)	β 1.54 (95% CI −1.03, 4.11)	94.0 ±15.8	90.8 ±14.9	87.9 ±12.4	−0.20 (−3.17, 2.77)	1.65 (−1.28, 4.58)
Language score < 85	33.3	33.0	29.3	1.10 (0.63, 1.91)	0.69 (0.42, 1.13)	27.0	33.4	33.8	1.05 (0.57, 1.92)	0.63 (0.37, 1.10)
Three Year Exam
DAS-II score	88.7 ±15.2	90.6 ±15.7	88.8 ±14.1	β −1.94 (95% CI −4.84, 0.95)	β −0.87 (95% CI −3.43, 1.70)	91.5±14.2	90.7 ±15.7	85.6 ±13.3	−1.56 (−4.53, 1.42)	−1.45 (−4.38, 1.48)
DAS-II score < 85	38.8	35.5	39.1	1.27 (0.75, 2.15)	1.07 (0.68, 1.69)	31.3	35.0	46.9	1.11 (0.63, 1.98)	1.13 (0.67, 1.90)
Child Behavioral Checklist > 60	16.5	9.0	11.8	1.73 (0.90, 3.33)	1.29 (0.67, 2.46)	16.3	9.2	12.4	1.83 (0.91, 3.70)	1.09 (0.52, 2.27)
Four Year Exam
Conners ADHD score > 60	18.1	15.6	20.6	1.07 (0.58, 1.98)	1.57 (0.93, 2.67)	23.4	15.2	19.2	1.30 (0.71, 2.37)	1.63 (0.87, 3.06)
Recall of Digits Forward – low score	21.7	25.0	26.9	0.99 (0.53, 1.83)	0.92 (0.54, 1.54)	16.9	23.6	32.5	1.09 (0.53, 2.23)	1.03 (0.58, 1.84)
Recognition of Pictures – low score	22.0	19.4	24.3	1.06 (0.59, 1.90)	1.34 (0.81, 2.22)	23.4	19.6	24.4	1.18 (0.63, 2.18)	1.14 (0.64, 2.05)
Five Year Exam
Child Behavioral Checklist > 60	10.2	10.2	11.7	1.0 (0.47, 2.14)	1.16 (0.61, 2.21)	12.2	10.5	11.0	1.33 (0.62, 2.84)	0.79 (0.37, 1.69)

For binary outcomes, the adjusted OR (95% CI) from logistic models is presented. For continuous outcomes, the regression coefficient (β, 95% CI) from linear regression models is presented and reflects the adjusted difference in score (i.e. score for aberrant birthweight – score for AGA).

None of the comparisons were statistically significant.

Regression models adjusted for gestational age at delivery, race-ethnicity, maternal age, parity, education, type of insurance, smoking status, baseline thyroid status (subclinical hypothyroidism or hypothyroxinemia), treatment group, and child sex.

Using the national fetal growth standards, 84 (8.8%) and 82 (8.6%) children were born SGA and LGA, respectively. SGA at birth remained associated with a decrease in the child IQ at 5 years of age (primary outcome) by 3.14 (95% CI 6.05 – 0.22) points and higher odds of an IQ <85 (aOR 2.33, 95% CI 1.3 – 4.1). The strength of association between SGA and primary outcome was similar to the one seen using population norms. There were no associations between SGA status using the national fetal growth standards and any of the other secondary outcomes. Likewise, there was no association between LGA and the primary (−0.03 points, 95% CI −2.91 – 2.84) or any secondary outcomes. (Tables 2, 3) The strength of associations between SGA and primary outcome was similar using either standard.

A total of 40 (3.7%) children were born with a HC <10% using population-based nomogram and similarly 35 (3.7%) using National Fetal growth curves. Whether using a population-based nomogram for HC at birth or the national fetal growth standards, aberrant HC was not associated with the primary outcome. (Table 4) HC <10% by population-based nomogram was associated with a decrease in the motor score compared with normal size HC, −4.17 (95% CI −7.91 – −0.43) points at 12 months of age and −4.54 (95% CI −8.29 – −0.79) points at 24 months of age. HC >90% by the national fetal growth standard was associated with an increase in language score, 1.92 (0.03 – 3.81) and lower odds of language score <85 (odds ratio 0.65, 95% CI 0.45 – 0.95) compared with normal size HC at 24 months of age. (Table 5) There were no other significant associations between aberrant HC with either standard and any other secondary outcomes.

Table 4:

Primary outcome (child IQ) and rates of primary outcome <85 for aberrant HC by different nomograms

	HC (Luchenco)
	<10% (n=38)	10–90% (n=865)	>90% (n=141)	P	<10% vs. 10–90%	P	>90% vs. 10–90%	P
Primary Outcome	94.3 ± 17.5	94.6 ± 14.7	95.1 ± 15.3	0.70	β −0.63 (95% CI −4.74, 3.48)	0.76	β −0.70 (−2.99, 1.58)	0.55
Primary Outcome < 85	13 (34.2)	219 (25.3)	33 (23.4)	0.40	aOR 1.62 (0.77, 3.45)	0.21	aOR 0.98 (0.62, 1.57)	0.95
	HC (National Fetal Growth)
	<10% (n=34)	10–90% (n=627)	>90% (n=256)	P	<10% vs. 10–90%	P	>90% vs. 10–90%	P
Primary Outcome	96.1 ± 18.4	94.5 ± 14.4	95.2 ± 14.6	0.76	β −1.00 (95% CI −5.34, 3.34)	0.65	β 1.26 (−0.61, 3.13)	0.19
Primary Outcome < 85	10 (29.4)	159 (25.4)	57 (22.3)	0.50	aOR 1.52 (0.64, 3.62)	0.34	aOR 0.78 (0.53, 1.15)	0.21

Data are presented as mean ± standard deviation or n (%).

^*The primary outcome was child intelligence quotient (IQ) at 5 years of age assessed using the WPPSI–III Full Scale test. Data for the primary outcome were missing on 23 children using the Alexander nomogram and 18 children for the National Fetal Grwoth standards.

^$For binary outcomes, the adjusted OR (95% CI) from logistic models is presented. For continuous outcomes, the regression coefficient (β, 95% CI) from linear regression models is presented and reflects the adjusted difference in score (i.e. score for aberrant birthweight – score for AGA).

^#Regression models adjusted for gestational age at delivery, race-ethnicity, maternal age, parity, education, type of insurance, smoking status, baseline thyroid status (subclinical hypothyroidism or hypothyroxinemia), treatment group, and child sex.

Table 5:

Secondary outcomes for aberrant head circumference by different nomograms

	Head circumference (Luchenco)					Head circumference (National Fetal Growth)
	<10%	10–90%	>90%	<10% vs. 10–90%	>90% vs. 10–90%	<10%	10–90%	>90%	<10% vs. 10–90%	>90% vs. 10–90%
One Year Exam
Bayley-III
Cognitive score	98.7±14.3	99.2±13.2	101.8±10.9	−2.25 (−6.27, 1.78)	0.78 (−1.50, 3.07)	101.3±14.3	99.0±12.9	100.9±11.8	−0.99 (−5.21, 3.24)	1.33 (−0.50, 3.17 )
Cognitive score < 85	10.5	9.1	5.3	1.50 (0.49, 4.61)	0.71 (0.31, 1.63)	5.9	9.2	5.7	1.08 (0.23, 4.99)	0.63 (0.33, 1.20)
Motor score	93.9 ±12.6	97.0 ±11.4	96.3 ±11.2	−4.17 (−7.91, −0.43) #	−0.80 (−2.90, 1.31)	95.0±11.9	97.0±11.2	97.0±11.0	−2.91 (−6.88, 1.05)	0.25 (−1.46, 1.96)
Motor score < 85	16.2	9.2	10.5	2.20 (0.85, 5.70)	1.16 (0.62, 2.18)	15.2	9.0	8.6	2.17 (0.73, 6.48)	0.89 (0.51, 1.54)
Language score	96.2 ±12.7	94.5 ±12.6	95.7 ±11.9	0.66 (−3.42, 4.74)	1.05 (−1.24, 3.35)	95.3±13.4	94.5±12.7	94.2±11.7	−1.21 (−5.58, 3.16)	0.29 (−1.60, 2.18)
Language score < 85	18.9	18.4	16.5	1.06 (0.45, 2.52)	0.91 (0.55, 1.52)	18.2	18.3	18.9	1.06 (0.41, 2.74)	0.99 (0.66, 1.48)
Two Year Exam
Bayley-III
Cognitive score	90.4 ±12.8	90.1 ±12.3	91.8 ±12.5	−1.87 (−5.48, 1.74)	0.25 (−1.77, 2.29)	94.7±10.8	90.4±12.4	90.7±12.2	0.43 (−3.42, 4.27)	0.47 (−1.16, 2.11)
Cognitive score < 85	13.9	23.9	19.4	0.74 (0.27, 2.05)	0.91 (0.54, 1.51)	9.4	23.2	23.1	0.69 (0.19, 2.49)	0.98 (0.66, 1.45)
Motor score	94.1 ±12.9	97.3 ±11.9	98.1 ±12.1	−4.54 (−8.29, −0.79) #	0.03 (−2.08, 2.15)	97.1±10.3	97.5±11.9	97.9±12.0	−2.83 (−6.81, 1.15)	0.56 (−1.14, 2.26)
Motor score < 85	11.1	9.9	10.2	1.38 (0.44, 4.35)	1.21 (0.63, 2.36)	6.3	9.4	10.0	0.75 (0.15, 3.71)	1.14 (0.66, 1.97)
Language score	92.4 ±16.2	90.5 ±15.1	92.4 ±14.2	−1.04 (−5.11, 3.04)	0.88 (−1.48, 3.23)	96.8±19.1	90.4±14.9	91.3±13.8	1.63 (−2.70, 5.95)	1.92 (0.03, 3.81) #
Language score < 85	35.1	33.1	26.6	1.67 (0.78, 3.57)	0.70 (0.44, 1.13)	30.3	33.8	29.1	1.61 (0.68, 3.83)	0.65 (0.45, 0.95) #
Three Year Exam
DAS-II score	91.7 ±17.1	90.1 ±15.3	91.1 ±15.6	−0.39 (−4.54, 3.76)	−0.68 (−3.01, 1.66)	95.1±16.5	89.9±15.0	91.0±15.7	1.27 (−3.08, 5.61)	1.44 (−0.45, 3.33)
DAS-II score < 85	36.8	36.2	30.9	1.38 (0.65, 2.94)	0.90 (0.57, 1.40)	32.4	36.3	32.0	1.34 (0.57, 3.15)	0.76 (0.53, 1.09)
Child Behavioral Checklist > 60	0	10.4	8.03	*	0.79 (0.40, 1.57)	0	10.2	10.0	*	0.97 (0.58, 1.63)
Four Year Exam
Conners ADHD score > 60	18.9	15.9	17.6	0.91 (0.38, 2.19)	1.36 (0.81, 2.27)	18.8	16.8	14.0	0.82 (0.31, 2.14)	0.89 (0.57, 1.39)
Recall of Digits Forward – low score	18.9	25.4	21.9	1.09 (0.43, 2.79)	0.96 (0.58, 1.61)	9.4	25.6	19.8	0.67 (0.18, 2.53)	0.68 (0.45, 1.04)
Recognition of Pictures – low score	21.6	19.6	22.5	1.16 (0.50, 2.70)	1.19 (0.73, 1.91)	18.8	19.7	21.9	0.87 (0.32, 2.35)	1.10 (0.74, 1.64)
Five Year Exam
Child Behavioral Checklist > 60	7.9	10.9	7.1	0.72 (0.21, 2.46)	0.62 (0.30, 1.25)	5.9	11.8	8.6	0.56 (0.13, 2.50)	0.62 (0.36, 1.05)

For binary outcomes, the adjusted OR (95% CI) from logistic models is presented. For continuous outcomes, the regression coefficient (β, 95% CI) from linear regression models is presented and reflects the adjusted difference in score (i.e. score for aberrant birthweight – score for AGA).

None of the comparisons were statistically significant except for these ones marked with (^#) which had a p<0.05

^*Not computed due to 0 cases SGA.

Regression models adjusted for gestational age at delivery, race-ethnicity, maternal age, parity, education, type of insurance, smoking status, baseline thyroid status (subclinical hypothyroidism or hypothyroxinemia), treatment group, and child sex.

Discussion

In this analysis of neurodevelopmental outcomes of children born to pregnant individuals with either subclinical hypothyroidism or hypothyroxinemia disorders, we found that SGA birthweight was associated with a decrease in average child IQ and greater odds of child IQ < 85 at 5 years of age. The strength of association did not appear to be stronger using national fetal growth compared with population nomograms. We did not find any consistent association between LGA birthweight or aberrant head circumference measurements after birth and neurodevelopmental outcomes.

The association between SGA and adverse pregnancy and neonatal outcomes has been previously evaluated. However these studies were limited by shorter follow up than our study, only relying on birthweight, and had inconsistent results.^6,7,18,19 Moreover, a limitation to prior studies is the different methods used to assess neurodevelopmental and cognitive outcomes and the various definitions for SGA. Using data from the Early Childhood Longitudinal Study–Birth Cohort, the cognitive function of 585 children who were severely SGA (<3% for GA at birth) was 12 percentile points lower than their AGA counterparts when evaluated at 9 months of age. However no difference was seen when the assessment was repeated at 2 years of age, nor were there any differences in academic performance for reading or math at preschool age (around 3.5 years).¹⁹ Other studies showed that fetal growth <10^th percentile was more prevalent among children who developed neonatal encephalopathy, whether it was secondary to an ischemic event or not, suggesting that antepartum injury, that may manifest as reduced growth potential may be a causative character of neonatal encephalopathy.⁸ Last, a systematic review of studies of neurodevelopment in term SGA newborns (n=7,861) compared with AGA (n= 91,619) ones showed that standardized neurodevelopmental scores of SGA newborns were 0.32 SD (95% CI, 0.25–0.38) below those for normal controls.¹¹

On the other hand, the association between HC and neurocognitive outcomes in children is not consistent. In a cohort of 4,046 infants born < 34 weeks, a small HC at birth (z score below 2 SD) at birth was associated with higher risk of adverse neurodevelopmental outcomes.²⁰ In another study, children with consistently small HC (2 SDs below the mean) were 7 times more likely to have an neurocognitive disorders, but 85% of children with small HC had normal neurocognition, and 93% of children with neurocognitive disorders had HC within the normal range.²¹ In our study, we found that HC <10% by population-based nomogram was associated with a decrease in the motor score at 12 months of age and 24 months of age. Whereas, HC >90% by the national fetal growth standard was associated with a decrease in language score <85 at 24 months of age. However, these inconsistent associations could be related to type I error, as we did not correct for multiple comparisons.

The mechanism of the association between abnormal growth and child neurodevelopment is not known, but studies detected the presence of fetal brain lactate peak among growth restricted fetuses,²² which may represent a marker of cerebral injury, predisposing these children to long term adverse neurodevelopmental outcomes. Other potential mechanisms may include alterations in the hypothalamic-pituitary-adrenal and hypothalamic-pituitary-thyroid axes, leading to reduced dendritic growth and synaptic formation, especially in the hippocampus.^23–26

Traditionally, evaluation of fetal growth has been accomplished by comparing fetal or birthweight and infant head circumference to population-based norms. These population norms are relatively old, may not reflect the current demographics of the US, were usually derived from either heterogeneous or highly selected patient cohorts that included abnormally grown fetuses (whether large or small), thus they may misclassify fetal growth. In addition, these population norms are typically based on birthweight and measurement of the HC, which requires a delivery. A neonate that needed delivery is not reflective of the normal status of a fetus in an ongoing pregnancy. The NICHD - National Fetal Growth standards represent a contemporary racial and ethnic standard for fetal growth, and were based on data from 1,737 prospectively followed fetuses born to healthy low risk pregnant individuals of different racial and ethnic backgrounds.¹⁶ Findings from our study do not support that these nomograms, which theoretically may reflect fetal growth better than birthweight nomograms, are superior at detecting abnormal fetal growth that is associated with long-term adverse outcomes.

Our study has several strengths. More than 90% of the original trial participants were followed through 5 years. In addition, trained and certified blinded examiners, blinded to maternal randomization status, assessed the child annually from age 1 to 5 to provide a comprehensive assessment of neurocognitive and neurodevelopmental outcomes. In addition, we had clearly pre-specified primary analyses and were able to adjust in our analysis for a number of key covariates such as maternal age and socio-demographic indicators of disparity including maternal education, insurance type, and race and ethnicity, all of which have been associated with the primary outcome of interest. Last, the original trial enrolled racially, ethnically, and geographically diverse patient population drawn from 33 geographically varied hospitals across the United States, making the findings more generalizable. However, generalizability is limited given the background of baseline thyroid function in the study population and study demographics. Moreover, similar to prior studies that evaluated child IQ, we are aware that the spectrum of child neurodevelopment is affected by many biological and social factors, which may not have been collected for this study. For example, parental IQ and the home and educational environment were not available to include in these analyses. We also did not evaluate postnatal growth as studies suggest it may affect long term neurocognitive outcomes in school age children.^27,28 Due to sample size limitations, we did not conduct separate analyses for term and preterm born children and for SGA children born to mothers with or without a hypertensive disorder of pregnancy, or those who smoked during pregnancy, as prior studies suggested that there are variations in neurodevelopmental outcomes within SGA children.²⁹ However we corrected for GA at delivery in our multivariable analysis. Although the association between SGA and lower odds of low IQ score in children persisted despite adjusting for multiple covariates, we acknowledge the possibility of unmeasured residual confounding. Similar to other observational studies, our study cannot address causality, or is able to elucidate the potential direct mechanisms underlying the association between SGA status and child neurodevelopment. It is also plausible that a birthweight < 10^th percentile may reflect a constitutionally small neonate with normal outcomes, and that a small HC < 10^th percentile may reflect a constitutionally small fetus who is also near or below the 10^th percentile for birthweight. Due to the limitations in our data, we are not able to accurately differentiate between those who are constitutionally small vs. those who are growth restricted. Also, due to sample size limitations, we did not conduct separate analysis of the outcomes of children had small HC but appropriate birthweight or those who had a birthweight or HC <3^rd or 5^th percentile. Finally, in addition to the limitations in detecting small differences in outcome measures due to the fixed sample size, no statistical adjustments were made for multiple comparisons.

Conclusion

In conclusion, in a cohort of pregnant individuals with hypothyroid disorders, SGA birthweight, but not aberrant HC measurements or LGA birthweight, was associated with a decrease in child IQ and greater odds of child IQ < 85 at 5 years of age. Using a fetal growth standard did not appear to improve the detection of SGA born children at risk of adverse neurodevelopment.

Figure 1:

Study flow chart

Figure 2:

Distribution of SGA (birthweight <10%) and LGA (birthweight >90%) neonates by population nomogram and national fetal growth (NFG) standards for those with both assessments.

Figure 3:

Distribution of aberrant HC neonates by population (Lubchenko) nomogram and NFG standards (<41 weeks) for those with both assessments.

Appendix

In addition to the authors, other members of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network are as follows:

University of Texas Medical Branch, Galveston, TX – G. Saade, A. Salazar, A. Acosta, K. Smith, G. Hankins, S. Jain, M. Munn, L. Pacheco

The University of Utah Health Sciences Center, Salt Lake City, UT – K. Hill, A. Sowles, S. Timothy. P. Reed (deceased) and S. Esplin (Intermountain Healthcare)

University of Texas Southwestern Medical Center, Dallas, TX – L. Moseley, J. Price, C. Melton, M. Garcia, J. Gerald, M. Santillan

University of Pittsburgh, Pittsburgh, PA – M. Cotroneo, D. DeAngelis, M. Luce, R. Kennedy, D. Nowinski

University of Alabama at Birmingham, Birmingham, AL – S. Harris, F. Biasini, M. Parks, J. Grant, C. Lee, A. Todd, K. Domnanovich, W. Andrews

Wayne State University, Detroit, MI – N. Hauff, Y. Sorokin, L. Goldston, D. Driscoll

The Ohio State University, Columbus, OH – F. Johnson, J. Iams, S. Wylie, R. Devlin, B. Selegue, C. Latimer, J. Bauer

Brown University, Providence, Rhode Island – D. Allard, T. Leach, V. Watson, B. Hughes

Columbia University, New York, NY – S. Bousleiman, V. Carmona, A. Zygmunt, Y. Williams (Drexel University), M. Grant (Drexel University), C. Kitto (Christiana Care Health Systems), B. Higley (Christiana Care Health Systems), M. Falk (St. Peter’s University Hospital); L. Padovano (St. Peter’s University Hospital)

MetroHealth Medical Center-Case Western Reserve University, Cleveland, OH – C. Milluzzi, B. Nielsen, W. Dalton, H. Cozart, E. Chien

The University of Texas Health Science Center at Houston, McGovern Medical School-Children’s Memorial Hermann Hospital, Houston, TX – F. Ortiz, S. Blackwell, B. Rech, M. Hutchinson, P. Givens

University of North Carolina at Chapel Hill, Chapel Hill, NC –K. Clark, S. Timlin, K. Dorman, E. Campos, H. Byers, S. Brody (WakeMed Health & Hospitals)

Northwestern University, Chicago, IL – G. Mallett, M. Ramos-Brinson, M. Weissbourd (Lurie Children’s Hospital), M. Dinsmoor (NorthShore University HealthSystem), K. Paychek (NorthShore University HealthSystem), P. Campbell

Oregon Health & Science University, Portland, OR – M. Rincon, J. Tolosa, L. Pereira, P. Blasco, S. Saxton, K. Beach, J. Snyder

George Washington University Biostatistics Center, Washington, DC – E.A. Thom, B. Jones-Binns, L. Mele

National Institute of Neurological Disorders and Stroke, Bethesda, MD – D.G. Hirtz

Eunice Kennedy Shriver National Institute of Child Health and Human Development, Bethesda, MD – C. Spong, S. Tolivaisa

MFMU Network Steering Committee Chair (Medical University of South Carolina, Charleston, SC) – J. P. VanDorsten, M.D.