Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 1.
Published in final edited form as: Epidemiology. 2021 Sep 1;32(5):664–671. doi: 10.1097/EDE.0000000000001387

Fetal growth trajectories among small for gestational age babies and child neurodevelopment

Kelly K Ferguson 1,*, Sara Sammallahti 2,3,†, Emma Rosen 1,†, Michiel van den Dries 2, Anjoeka Pronk 4, Suzanne Spaan 4, Mònica Guxens 2,5,6,7, Henning Tiemeier 2,3, Romy Gaillard 8, Vincent WV Jaddoe 8,9
PMCID: PMC8338787  NIHMSID: NIHMS1708212  PMID: 34086648

Abstract

Background:

Being born small for gestational age (SGA, <10th percentile) is a risk factor for worse neurodevelopmental outcomes. However, this group is a heterogeneous mix of healthy and growth-restricted babies, and not all will experience poor outcomes. We sought to determine whether fetal growth trajectories can distinguish who will have the worst neurodevelopmental outcomes in childhood among babies born SGA.

Methods:

The present analysis was conducted in Generation R, a population-based cohort in Rotterdam, the Netherlands (N=5487). Using group-based trajectory modeling, we identified fetal growth trajectories for weight among babies born SGA. These were based on standard deviation scores of ultrasound measures from mid- and late pregnancy in combination with birth weight. We compared child non-verbal IQ and attention deficit hyperactivity disorder (ADHD) symptoms at age 6 between SGA babies within each growth trajectory to babies born non-SGA.

Results:

Among SGA individuals (n=656), we identified three distinct fetal growth trajectories for weight. Children who were consistently small from mid-pregnancy (n=64) had the lowest IQ (7 points lower compared to non-SGA babies, 95% CI=−11.0, −3.5) and slightly more ADHD symptoms. Children from the trajectory that started off larger but were smaller at birth showed no differences in outcomes compared to children born non-SGA.

Conclusions:

Among SGA children, those who were smaller beginning in mid-pregnancy exhibited the worst neurodevelopmental outcomes at age 6. Fetal growth trajectories may help identify SGA babies who go on to have poor neurodevelopmental outcomes.

Keywords: ultrasound, SGA, fetal growth, neurodevelopment, birthweight, trajectories

Introduction

It is well established that babies who are born small for gestational age (SGA, <10th percentile birth weight for gestational age) face an increased risk of various adverse childhood health outcomes, including cognition and behavior problems.1–7 However, this group is a heterogeneous mix of healthy and growth-restricted babies,8 and not all are at risk for poor outcomes. Distinguishing these two groups is essential for understanding the causes and consequences of growth restriction, yet doing so has proven extraordinarily difficult.

Information on growth trajectories across gestation that incorporate data from fetal ultrasounds could inform the distinction between these groups. Previous research demonstrates that estimated fetal weight9–15 and head circumference10, 12, 16–19 measured by ultrasound in utero, as well as rates of change in growth during gestation,10, 12–14 are associated with worse birth outcomes and childhood health consequences. However, few studies consider both size of the fetus at individual time points as well as rates of change in fetal growth simultaneously.14, 20, 21

In the present study we applied group-based trajectory modeling to identify distinct fetal growth trajectories among babies born SGA in order to examine associations between trajectories and neurodevelopmental outcomes assessed at age 6. We hypothesized that among babies born SGA, fetal growth trajectories would identify those who had worse neurodevelopmental outcomes compared to non-SGA children. In our primary analysis, we investigated trajectories of weight measured at mid- and late pregnancy by ultrasound and at birth. Additionally, although measurements were available on fewer participants, we examined trajectories of head circumference as a secondary analysis.

Methods

Study population

The Generation R Study, a prospective population-based birth cohort, has previously been described.22 All mothers with an expected delivery date between 4/2002 and 1/2006 living in Rotterdam, the Netherlands, were invited to participate; a total of 8,879 pregnant women (61% of eligible mothers) were enrolled.22 Mothers provided written informed consent for themselves and their child at the time of enrollment, and the study protocol underwent human subject review at Erasmus Medical Center, Rotterdam, Netherlands.

We included only those participants who: 1) had singleton pregnancies; 2) had at least one of the two growth measures (weight or head circumference) measured at all three time points (mid- and late pregnancy and birth); and 3) had either non-verbal IQ or attention deficit hyperactivity disorder (ADHD) symptoms data available at age 6. Finally, we excluded one sibling per sibling pair; we prioritized inclusion of SGA children with complete exposure and outcome data, and otherwise selected the included sibling at random. This yielded a total analytical sample of 5487 mother-child dyads. Among these, 660 infants were born SGA (birth weight <10th percentile for gestational age based on an internal population-based standard).23

Fetal growth measures

Women underwent fetal ultrasound examinations during early, mid-, and late pregnancy.23 Early-pregnancy ultrasound or medical record data was used for dating purposes.24 At mid- (~20 weeks) and late pregnancy (~30 weeks), head circumference was measured to the nearest millimeter using standardized techniques, and estimated fetal weight was calculated using Hadlock’s formula.25 We standardized these fetal measurements to standard deviation scores (SDS) for gestational age based on internal population-based growth curves, which are described in detail elsewhere.24 Briefly, we derived reference curves from ultrasound measures of fetal growth on approximately 8,000 live-born singletons who were non-anomalous and whose mothers enrolled in the study before 25 weeks of gestation. Birth measurements of head circumference and weight were derived from midwife and hospital registries and standardized for gestational age and sex.26

Childhood non-verbal IQ and ADHD symptoms

When the child was 6 years old (standard deviation=0.5 years), we invited families to participate in an in-person follow-up. Non-verbal IQ at this visit was assessed using the Mosaics and Categories subtests from the Snijders-Oomen Nonverbal Intelligence Test-Revised (SON-R), a test developed and validated in the Netherlands; similarly to other widely used IQ tests, it yields IQ estimates with mean=100 and SD=15 points.27 Raw scores were standardized into estimates of non-verbal IQ using age-specific, Dutch references from the SON-R 2.5–7 manual.

Child emotional and behavioral problems were assessed by self-report from the primary caregiver with the Child Behavior Checklist (CBCL) 1.5–5,28 an internationally validated and reliable measure of emotional and behavioral problems.28, 29 The CBCL measures these problems on continuous scales of severity, and has been shown to predict adult psychiatric disorders.30, 31 Each of 99 items within seven different scales is scored as 0 = ’not true’, 1 = ’somewhat or sometimes true’, or 2 = ’very true or often true’, based on behaviors in the preceding 2 months. From the CBCL, we used the DSM-oriented subscale score of ADHD symptoms; scores in this scale range from 0 to 12, with an average of 3.1 (SD=2.6) within our population-based sample.

Covariates

We included potential confounders identified a priori from a directed acyclic graph. We assessed information on maternal height, pre-pregnancy body mass index (BMI, kg/m2), education (secondary or lower vs. tertiary), ethnicity (Dutch vs. non-Dutch), and smoking during pregnancy (no vs. yes) using the prenatal questionnaire. We assessed maternal psychopathology using the Brief Symptom Inventory,32 reported as the Global Severity Index (GSI) score.33 At the 6-year follow-up, child age was recorded and maternal non-verbal IQ was assessed using a computerized Ravens Advanced Progressive Matrices Test.34, 35

Statistical analysis

For calculating fetal growth trajectories for weight, we used data from SGA participants with complete information on weight at mid- and late pregnancy and at delivery, which represented almost all SGA participants from the overall study population (n=656 out of 660). For our secondary analysis of head circumference trajectories, we used data from SGA participants with complete information on head circumference at mid- and late pregnancy and at delivery, which was a much smaller subset (n=309 out of 660).

Within SGA babies, we created latent growth trajectory classes for weight and head circumference standard deviation scores using group-based trajectory modeling, a procedure that allows for the classification of individuals based on shared features of longitudinal data points.36 For this approach we utilized the Traj procedure in SAS ®.37 We began with a two-group solution and then proceeded to test the three-, four-, and five-group solutions, aiming to arrive at the simplest and most suitable model based on model fit statistics, posterior probabilities, and group size.36 Independent variables included gestational age at the time of the mid- and late pregnancy ultrasounds and gestational age at birth; other covariates were not included in determination of the trajectories. Difference in Bayesian Information Criterion (BIC) value between the simpler model and the consecutive more complicated model (i.e., with more groups) was assessed. An average posterior probability of at least 0.7 (for those assigned to the group in question) was considered adequate.36 We considered group size of <10% in the smallest group, or fewer than 30 cases per group for subsequent analyses, to be suboptimal.

Subsequently, we established trajectory membership by assigning each individual to the group with highest membership probability, and included all participants with the data in question (weight or head circumference) in one of the three latent growth trajectory class groups. We outputted trajectory membership for each individual for subsequent analyses.

Next, we examined demographic characteristics within each trajectory and compared those to non-SGA babies (>10th percentile, including large for gestational age). Non-SGA was chosen as the primary reference group because this contrast is made most commonly in the literature (i.e., SGA compared to non-SGA). Our goal was to demonstrate whether SGA sub-groups had worse outcomes than others when drawing this typical comparison. We created multivariable linear regression models for the association between trajectory membership and non-verbal IQ and ADHD symptoms with non-SGA individuals as the reference group. Models were created separately for weight and head circumference groups. We imputed 50 datasets (with 10 iterations) using multiple chained imputation equation to handle missing covariate data, pooling estimates by Rubin’s rules.38 Separate imputations were run for each anthropometric dataset; the demographic predictors were consistent. The percentage of missing values was <18% and distributions of covariates in imputed datasets were very similar to those in the observed dataset. Sample sizes for each model varied based on availability of outcome measures (eFigure 1).

We performed several sensitivity analyses. First, we compared associations between trajectories, selecting the most common trajectory as the reference group. We also examined these models with adjustment for posterior probabilities of group membership, as a way to account for the uncertainty from the first model.39 Second, we examined associations among a more homogenous sample of Dutch participants only. Third, we examined associations excluding preterm children, to ensure our results were not solely driven by preterm birth. Fourth, we similarly examined associations excluding mothers who were diagnosed with hypertensive disorders in pregnancy. Fifth, we examined associations restricting to pregnancies that had verification of gestational age based on 1st trimester ultrasound, to reduce the risk of misclassifying growth because of less reliable gestational age estimates. Sixth, we restricted the comparison group to participants born appropriate for gestational age, excluding babies born large for gestational age (LGA, >90th percentile). Finally, for comparison, we examined how SGA status (vs. non-SGA status) itself, and continuous standard deviation scores of weight and head circumference at each time point among SGA children, were associated with non-verbal IQ and ADHD symptoms.

Results

Mothers of SGA children (n=656), compared to those of non-SGA children, were more often non-Dutch, primiparous, or smokers, had lower education levels, and were slightly shorter (Table 1). Being born SGA was associated with lower non-verbal IQ in adjusted models (mean difference [MD]= −2.5, 95% CI= −3.8, −1.3), but not with ADHD symptoms (MD=−0.04, 95% CI=, −0.26, 0.19).

Table 1.

Characteristics of the Generation R Study population by weight trajectory group (shown in Figure 1).

Characteristic Total (N=5487) Non-SGAa (n=4827) SGAb (n=656) Group 1: Consistent low weight SDS SDS (n=64) Group 2: Moderate decrease in weight SDS (n=418) Group 3: Greatest decrease in weight SDS (n=174)
Maternal age, years (mean [SD]) 30.3 (5.0) 30.4 (4.9) 29.6 (5.6) 29.4 (5.4) 29.6 (5.6) 29.6 (5.6)
Maternal ethnicity, Dutch (n [%]) 2949 (55) 2662 (56) 285 (44) 34 (55) 180 (44) 71 (42)
 non-Dutch (n [%]) 2458 (45) 2100 (44) 356 (56) 28 (45) 229 (56) 99 (58)
Maternal education, tertiary (n [%]) 2431 (47) 2196 (48) 233 (38) 24 (40) 151 (39) 58 (35)
 secondary or lower (n [%]) 2768 (53) 2387 (52) 379 (62) 36 (60) 236 (61) 107 (65)
Maternal BMI, kg/m2 (mean [SD]) 23.6 (4.2) 23.7 (4.2) 22.9 (4.1) 22.3 (3.9) 22.5 (3.7) 24.0 (4.8)
Maternal height, cm (mean [SD]) 168 (7.3) 168 (7.3) 164 (6.8) 166 (6.9) 164 (6.9) 164 (6.3)
Maternal IQ, points (mean [SD]) 96 (15) 96 (15) 93 (16) 95 (14) 93 (16) 90 (16)
Maternal GSI score, points (mean [SD]) 0.3 (0.4) 0.3 (0.4) 0.3 (0.4) 0.2 (0.3) 0.3 (0.4) 0.4 (0.5)
Mother primiparous (n [%]) 3211 (59) 2742 (57) 465 (71) 46 (72) 269 (71) 123 (72)
 Multiparous (n [%]) 2244 (41) 2058 (43) 186 (29) 18 (28) 120 (29) 48 (29)
Mother did not smoke during pregnancy (n [%]) 3638 (74) 3241 (74) 395 (66) 38 (63) 249 (65) 108 (71)
 smoked during pregnancy (n [%]) 1309 (26) 1107 (25) 201 (34) 22 (37) 134 (35) 45 (29)
Child sex, female (n [%]) 2746 (50) 2420 (50) 324 (50) 34 (53) 213 (51) 77 (44)
 Male (n [%]) 2741 (50) 2407 (50) 332 (50) 30 (47) 205 (49) 97 (56)
Child IQ score, points (mean [SD]) 101 (15) 101 (15) 97 (15) 94 (18) 97 (15) 99 (16)
Child ADHD symptom score, points (mean [SD]) 3.1 (2.6) 3.1 (2.6) 3.3 (2.5) 3.8 (2.6) 3.3 (2.5) 3.1 (2.6)
Open in a new tab
a

Defined as 10th–90th percentile birth weight for gestational age and sex.

b

Defined as <10th percentile birth weight for gestational age and sex.

ADHD: Attention deficit/hyperactivity disorder; BMI: body-mass-index; GSI: Global Severity Index; IQ: intelligence quotient; kg: kilogram; m: meter; n: number of cases; N: number of participants with available data; SD: standard deviation; SDS, standard deviation score; SGA, small for gestational age; %: proportion of cases among participants with data available. Note: n may not add up to total sample size due to missingness. Total n for each variable is as follows: Maternal age (5487); maternal ethnicity (5407); maternal education (5199); maternal BMI (4584); maternal height (5472); maternal IQ (4817); maternal GSI score (4551); mother parity (5455); maternal smoking (4947); child sex (5487); child IQ (4674); child age at IQ (4674); child ADHD score (4780); child age at behavioral assessment (4780).

Fetal growth trajectories for weight

Based on the criteria defined above, we identified three distinct fetal growth trajectories by weight for SGA babies (Figure 1). The 3-group solution was deemed to be the best fit based on the BIC and because it had sufficient average posterior probability (0.86) (eTable 1). Average posterior probability (with range of posterior probabilities across participants and group size given in brackets) was 0.8 (0.5–1.0, n=64), 0.9 (0.5–1.0, n=418), and 0.9 (0.5–1.0, n=174) in weight trajectory groups 1, 2, and 3, respectively.

Figure 1. Fetal weight trajectories among babies born small for gestational age in the Generation R Study.

Figure 1.

Open in a new tab

Abbreviations: IQR, interquartile range; SDS, standard deviation score

The greatest separation in weight standard deviation scores was at the mid-pregnancy ultrasound, with weight becoming more similar as pregnancy progressed. Group 1 for weight (those with consistent low weight SDS, n=64, 10% of SGA babies) had the lowest weight in mid- and late pregnancy and at delivery and had slightly earlier deliveries (eTable 2). Compared to other SGA mothers, those from Group 1 were younger, taller, more often Dutch, highly educated, and primiparous, and had higher IQ and lower GSI scores (Table 1). Group 2 (moderate decrease in weight SDS, n=418, 64% of SGA babies) was intermediate in weight across pregnancy and Group 3 (greatest decrease in weight SDS, n=174, 27% of SGA babies) had the highest weight in mid- and late pregnancy.

In adjusted models, children from the consistent low weight SDS group, on average, 7 lower IQ points (95% CI=−11.0, −3.5) in tests performed at age 6 compared to non-SGA children (Figure 2), which translates into a difference of approximately half a standard deviation (SD). They also had slightly more ADHD symptoms in the unadjusted model (0.75-point increase, meaning approximately 0.3 SD); however, this association was attenuated in the fully adjusted model. Children from the moderate decrease in weight SDS group also had lower IQ scores (mean difference [MD]= −4.2, 95% CI=−5.9, −2.6) in adjusted models, but had no difference in ADHD symptoms (MD=0.01, 95% CI=−0.27, 0.28). Children from the greatest decrease in weight SDS group had IQ and ADHD symptoms that were not different from non-SGA children.

Figure 2. Adjusteda differences in IQ or ADHD symptoms for children from each growth trajectory compared to children who were not born small for gestational age (SGA).

Figure 2.

Open in a new tab

Note: Child IQ and ADHD symptoms are in the units of points. aAdjusted for child sex (girl vs boy), child age at assessment (continuous), maternal age (continuous), maternal pre-pregnancy body-mass-index (continuous), maternal height (continuous), maternal education (less than tertiary vs tertiary), maternal ethnicity (Dutch vs non-Dutch), parity (primiparous vs multiparous), maternal intelligence quotient (continuous), maternal global severity score describing psychopathology (continuous), and maternal smoking during pregnancy (no vs yes). Abbreviations: IQ, intelligence quotient; ADHD, attention deficit hyperactivity disorder; CI, confidence interval.

When we examined associations among SGA groups, we also observed that those born in the consistent low weight SDS group had lower IQ compared to the moderate decrease in weight SDS group (Table 2). This association was similar in models additionally adjusted for posterior probabilities of inclusion. We also observed higher IQ among children born in the greatest decrease in weight SDS group compared to the moderate group. However, we observed no differences between SGA groups for ADHD symptoms.

Table 2.

Mean difference in child IQ and ADHD symptoms among small for gestational age participants only, with Group 2 weight trajectory (moderate decrease in z-score, n=418) as the reference group.

IQ ADHD symptoms
n Mean difference (95% CI) p n Mean difference (95% CI) p
Group 1: Consistent low weight SDS Unadjusted 54 −3.4 (−7.8, 1.0) 0.13 55 0.49 (−0.23, 1.2) 0.19
Adjusteda 54 −4.4 (−8.5, −0.3) 0.03 55 0.55 (−0.16, 1.3) 0.13
Group 2: Moderate decrease in weight SDS Unadjusted 418 ref 418 ref
Adjusteda 418 ref 418 ref
Group 3: Greatest decrease in weight SDS Unadjusted 148 1.9 (−1.0, 4.9) 0.20 147 −0.20 (−0.69, 0.29) 0.42
Adjusteda 148 3.3 (0.6, 6.1) 0.02 147 −0.19 (−0.67, 0.30) 0.45
Open in a new tab

Note: Child IQ and ADHD symptoms are in the units of points.

a

Adjusted for child sex (girl vs boy), child age at assessment (continuous), maternal age (continuous), maternal pre-pregnancy body-mass-index (continuous), maternal height (continuous), maternal education (less than tertiary vs tertiary), maternal ethnicity (Dutch vs non-Dutch), parity (primiparous vs multiparous), maternal intelligence quotient (continuous), maternal global severity score describing psychopathology (continuous), and maternal smoking during pregnancy (no vs yes).

Abbreviations: IQ: intelligence quotient; ADHD: Attention deficit/hyperactivity disorder; CI: Confidence Interval; n: number of participants in group of interest; SDS, standard deviation score; p: p-value for difference between the weight gain group in question and SGA children in weight group 2.

Results were very similar in models where we examined Dutch participants only (n=232 SGA, 35%), or where we excluded preterm children (n=126 of SGA, 19%), mothers who experienced hypertensive disorders in pregnancy (n=177 of SGA, 27%), participants whose gestational age was not based on 1st trimester ultrasound (n=372 of SGA, 57%), or when the comparison group was restricted to appropriate-for-gestational age births, excluding LGA children (eTable 3).

As a comparison, we estimated the association between non-verbal IQ or ADHD symptoms and a one standard deviation change in estimated fetal weight for mid- and late pregnancy among all SGA children (eTable 4). In adjusted models, a 1-unit increase in weight standard deviation score at mid-pregnancy was associated with 1.9 points higher IQ (95% CI=0.6, 3.2), at late pregnancy with 2.4 points higher IQ (95% CI=1.0, 3.8), and at delivery with 3.0 points higher IQ (95% CI=0.3, 5.8). Only birth weight standard deviation scores were associated with ADHD symptoms (MD=−0.52, 95% CI=−0.96, −0.08).

Fetal growth trajectories for head circumference

For our secondary analysis, a smaller number of study participants had complete data for identifying head circumference growth trajectories (n=309), mostly due to missingness in delivery measurements. We also identified three latent classes among SGA babies (eFigure 2; eTable 1). Average posterior probability (and range and group size) was 0.8 (0.4–1.0, n=206) for head trajectory group 1, 0.7 (0.5–1.0, n=59) for group 2, and 0.8 (0.5–1.0, n=44) for group 3. Group 1 for head circumference (n=206, 67%) was the most typical group, with slightly smaller than average head size consistently across gestation. Group 2 (n=59, 19%) had close to average in head size in mid-pregnancy but then declined, with delivery measurements much lower (~3 standard deviations) compared to the other SGA births (~1 standard deviation). For Group 3 (n=44, 14%), head size in mid-pregnancy was slightly larger compared to the overall SGA group and declined across gestation. Demographic characteristics by group are shown in eTable 5.

In adjusted models, children from Group 2, but not others, had lower IQ (MD=5.0, 95% CI=−8.7, −1.2) and slightly higher ADHD symptoms (MD 0.46, 95%CI= −0.25, 1.18) compared with non-SGA children (eFigure 3). However, there were no differences in IQ or ADHD symptoms between head circumference groups (eTable 6). Results were similar in other sensitivity analyses (eTable 7). Among SGA children, continuous measures of head circumference standard deviation scores during gestation had stronger associations with child IQ than measures at delivery (eTable 8).

Discussion

Within the Generation R prospective population- based birth cohort, we examined fetal growth trajectories among babies born SGA, and tested the hypothesis that childhood neurodevelopmental outcomes would differ based on trajectory. Among SGA babies, those with consistently low weight across gestation had the lowest non-verbal IQ at age 6. The mean IQ of children from this group was 7 points lower than that of non-SGA babies (i.e., >10th percentile birth weight for gestational age at delivery), and 3 points lower than that of other SGA babies. These findings show that fetal growth trajectories aid in identifying newborns who go on to have poor neurodevelopmental outcomes. If these findings are replicated, this innovation could improve detection of associations in studies investigating risk factors, and may have potential for prognostic purposes in the clinical setting.

Defining fetal growth restriction is a necessity and a challenge. Clinical definitions are typically based on ultrasound measurements (e.g., estimated fetal weight <10th percentile)41, an approach that is not without controversy40. However, researchers have multiple options for using the same data. These include the following: 1) single ultrasound measurements; 2) difference in two ultrasound measurements (i.e., change in weight over time); 3) conditional growth measurements (i.e., difference between expected and actual weight over time); and 4) growth mixture models.11 The latter, employed in the present study, have been applied minimally but are advantageous because they incorporate information on the size of the fetus compared to other fetuses at each time point, as well as changes in that relative size over time.

Our findings are consistent with the literature, although no previous study has utilized the exact same analytic approach. Among other studies using growth mixture models, two identified a single group of growth-restricted babies who were at an increased risk of neonatal complications.20, 21 A third observed stable as well as changing fetal growth trajectories in a population where all participants had at least one estimated fetal weight measurement below the 10th percentile, and babies who were consistently small across pregnancy had the worst neurodevelopmental outcomes at ages 1 and 5.14 Several studies have also examined rate of change in growth over pregnancy as a potential predictor of poor neurodevelopmental outcomes. Findings indicate that fetal weight gain in mid-pregnancy may be most important for various neurodevelopmental outcomes.10, 12 Last, a study where SGA was associated with lower IQ at ages 5 and 8 found that this association was greatest in magnitude among SGA newborns who had an estimated fetal weight 2 SD below the mean at 25–37 week ultrasound.15 Overall these findings and ours indicate that small babies who were consistently small across gestation have the worst outcomes.

Furthermore, our results are in line with the literature on neurodevelopment in pregnancy. In early- to mid-gestation, where we see the greatest differences in our groups, the basic structures of the brain are forming.42 This development is crucial for cognitive ability, as measured by IQ. Later in pregnancy, however, myelinization and frontal lobe development occur, which are highly relevant to executive function. The fact that our groups were more similar in size in later pregnancy may thus explain the lesser associations observed with ADHD symptoms.

We also know that not all smallness is detrimental. Instead, it is likely that there are distinct etiologies for SGA, only some of which are pathologic and place the child at risk for adverse long-term neurodevelopmental outcomes.43 As examples, genetic anomalies and infections in early pregnancy, typically resulting in early growth restriction, may also influence brain development. Placental dysfunction, on the other hand, with growth manifestations that occur later in gestation, may not have the same magnitude of impact.

We expected that babies who showed decline in relative size across pregnancy would have poor neurodevelopmental outcomes, and that those who were consistently small across pregnancy would represent a “constitutionally small” group and fare better. However, based on our findings and the literature described above, it appears that for babies born SGA, growth from early to mid-pregnancy is key. For these fetuses, variation in growth toward the end of pregnancy may be less important for neurodevelopmental outcomes later in life.

Customized growth curves or percent optimal birthweight, accounting for features such as maternal height and parity,44–46 have also been useful for distinguishing which SGA babies will go on to have the worst health outcomes.45 Our results do not seem to be driven by these factors. In fact, it is quite intriguing that mothers of SGA babies who had poor neurodevelopmental outcomes (consistent low weight SDS, Group 1) had demographic profiles that are generally considered more favorable for fetal growth, including a higher proportion Dutch and higher education levels compared to mothers of other SGA babies. Thus, the information gained from fetal growth trajectories may be independent of what could be gained from customized curves.

Future directions for this work would should include replication to determine whether patterns hold in populations with differing demographic, anthropometric, and geographic characteristics. Additionally, examination of trajectories in populations with more than two ultrasound measures is warranted. A study with a small number of ultrasounds would logically result in fewer, less-varied groups than a study with more time points. Finally, an important next step is the exploration of risk factors associated with a poor growth trajectory, including diet, lifestyle characteristics, and environmental exposures.

Strengths and limitations

Our study has three main strengths. First, it is a large, prospective study with multiple ultrasound measures during pregnancy. Second, we had rich childhood follow-up information available at age 6 and the ability to assess associations between fetal growth trajectories and child non-verbal IQ as well as ADHD symptoms, the latter of which has not been examined in relation to in utero fetal growth measurements previously. Third, we applied a method that incorporates information both on relative fetal size and rate of change in the identification of distinct fetal growth trajectories among SGA babies. This may be more informative than examining size at any individual time point or velocities of growth independently.

Some limitations of our approach are that the fetal growth trajectories are data driven, are based on a relatively small number of SGA births, and may not be generalizable. The Generation R Study is a multi-ethnic cohort and does not oversample for at-risk pregnancies, as in other studies with similar goals,14–16 so these patterns may be observed in other populations. Findings from our secondary analysis of head circumference must also be interpreted with caution; measurements at delivery were available only among a subset of participants, who could have different outcomes compared to those without measurements. Finally, clinical utility our findings may be limited since fetal growth trajectories included measures taken at birth. However, there are many instances where defining cases of fetal growth restriction at delivery could be valuable if: 1) recurrence exists among siblings; 2) modifiable risk factors occurring during childhood modulate associations with poor child health outcomes, and interventions can be targeted there; and 3) definitions based on fetal growth trajectories could be used for improving the ability to identify mechanisms, causes, and risk factors that could be used for prediction or prevention in the future.

Conclusion

Researchers are thwarted in efforts to identify true fetal growth restriction by insufficient criteria. Within a population-based cohort from Rotterdam, the Netherlands, we demonstrate that among babies born SGA, there is a distinct group with consistent smallness across gestation that have lower neurodevelopmental outcomes in childhood. IQ scores in this group were 6 points lower compared to those among of babies born non-SGA, which could be hugely consequential given that a reduction in just 1 IQ point are estimated to lead to a lifetime loss of earnings of 2%.47 This effect size is also similar to what has been observed for the change in child IQ in families where a mother has less than a high school education compared to a college education or greater.48 Identification of fetal growth restricted babies using this approach could refine outcome definitions in research studies investigating the developmental origins of health and disease.

Supplementary Material

SDC

Acknowledgments:

We would like to thank Whitney Cowell for her statistical guidance in the early stages of this project. The authors thank the participating parents and their children, general practitioners, hospitals, midwives, and pharmacies for their contribution. The Generation R Study is conducted by the Erasmus Medical Center, Rotterdam, in close collaboration with the Faculty of Social Sciences of the Erasmus University Rotterdam, the Municipal Health Service, Rotterdam Homecare Foundation, and Stichting Trombosedienst en Artsenlaboratorium Rijnmond.

Sources of financial support: This work was supported by the Intramural Research Program of the National Institute of Environmental Health Sciences, National Institutes of Health [ZIA ES101575]; the European Research Council Consolidator Grant [ERC-2014-CoG-648916 to VWVJ]; and a grant of the Netherlands Organization for Scientific Research [NWO grant 016.VICI.170.200 to HT].

Footnotes

Conflicts of interest: None

Process by which data may be obtained: Because of restrictions based on privacy regulations and informed consent of participants, data cannot be made freely available in a public repository. Data can be obtained upon request. Requests should be directed towards the management team of the Generation R Study (secretariaat.genr@erasmusmc.nl), which follows a protocol of approving data requests.

References

  • 1.Lundgren EM, Tuvemo T. Effects of being born small for gestational age on long-term intellectual performance. Best practice & research Clinical endocrinology & metabolism 2008; 22: 477–88. [DOI] [PubMed] [Google Scholar]
  • 2.Leonard H, Nassar N, Bourke J, et al. Relation between intrauterine growth and subsequent intellectual disability in a ten-year population cohort of children in Western Australia. American journal of epidemiology 2007; 167: 103–11. [DOI] [PubMed] [Google Scholar]
  • 3.Huang C, Martorell R, Ren A, Li Z. Cognition and behavioural development in early childhood: the role of birth weight and postnatal growth. International journal of epidemiology 2012; 42: 160–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Gu H, Wang L, Liu L, et al. A gradient relationship between low birth weight and IQ: A meta-analysis. Scientific reports 2017; 7: 18035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Tanis JC, van der Ree MH, Roze E, et al. Functional outcome of very preterm–born and small-for-gestational-age children at school age. Pediatric research 2012; 72: 641. [DOI] [PubMed] [Google Scholar]
  • 6.Goldenberg RL, DuBard MB, Cliver SP, et al. Pregnancy outcome and intelligence at age five years. American journal of obstetrics and gynecology 1996; 175: 1511–5. [DOI] [PubMed] [Google Scholar]
  • 7.Heinonen K, Räikkönen K, Pesonen A-K, et al. Behavioural symptoms of attention deficit/hyperactivity disorder in preterm and term children born small and appropriate for gestational age: a longitudinal study. BMC pediatrics 2010; 10: 91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Peleg D, Kennedy CM, Hunter SK. Intrauterine growth restriction: identification and management. American family physician 1998; 58: 453–60, 66–7. [PubMed] [Google Scholar]
  • 9.Alkandari F, Ellahi A, Aucott L, Devereux G, Turner S. Fetal ultrasound measurements and associations with postnatal outcomes in infancy and childhood: a systematic review of an emerging literature. J Epidemiol Community Health 2015; 69: 41–8. [DOI] [PubMed] [Google Scholar]
  • 10.Henrichs J, Schenk JJ, Barendregt CS, et al. Fetal growth from mid-to late pregnancy is associated with infant development: the Generation R Study. Developmental Medicine & Child Neurology 2010; 52: 644–51. [DOI] [PubMed] [Google Scholar]
  • 11.Larose TL, Turner SW, Hutcheon JA, et al. Longitudinal Ultrasound Measures of Fetal Growth and Offspring Outcomes. Current Epidemiology Reports 2017; 4: 98–105. [Google Scholar]
  • 12.Norris T, Johnson W, Petherick E, et al. Investigating the relationship between fetal growth and academic attainment: secondary analysis of the Born in Bradford (BiB) cohort. International journal of epidemiology 2018; 47: 1475–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.van Batenburg-Eddes T, de Groot L, Steegers EA, et al. Fetal programming of infant neuromotor development: the generation R study. Pediatric research 2010; 67: 132. [DOI] [PubMed] [Google Scholar]
  • 14.Von Ehrenstein OS, Mikolajczyk RT, Zhang J. Timing and trajectories of fetal growth related to cognitive development in childhood. American journal of epidemiology 2009; 170: 1388–95. [DOI] [PubMed] [Google Scholar]
  • 15.Rogne T, Engstrøm AA, Jacobsen GW, Skranes J, Østgård HF, Martinussen M. Fetal growth, cognitive function, and brain volumes in childhood and adolescence. Obstetrics & Gynecology 2015; 125: 673–82. [DOI] [PubMed] [Google Scholar]
  • 16.Harvey D, Prince J, Bunton J, Parkinson C, Campbell S. Abilities of children who were small-for-gestational-age babies. Pediatrics 1982; 69: 296–300. [PubMed] [Google Scholar]
  • 17.Henrichs J, Schenk JJ, Schmidt HG, et al. Fetal size in mid-and late pregnancy is related to infant alertness: The generation R study. Developmental Psychobiology: The Journal of the International Society for Developmental Psychobiology 2009; 51: 119–30. [DOI] [PubMed] [Google Scholar]
  • 18.Whitehouse AJ, Zubrick SR, Blair E, Newnham JP, Hickey M. Fetal head circumference growth in children with specific language impairment. Archives of disease in childhood 2012; 97: 49–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Walker SP, Thame MM, Chang SM, Bennett F, Forrester TE. Association of growth in utero with cognitive function at age 6–8 years. Early human development 2007; 83: 355–60. [DOI] [PubMed] [Google Scholar]
  • 20.Barker ED, McAuliffe FM, Alderdice F, et al. The role of growth trajectories in classifying fetal growth restriction. Obstetrics & Gynecology 2013; 122: 248–54. [DOI] [PubMed] [Google Scholar]
  • 21.Slaughter JC, Herring AH, Thorp JM. A Bayesian latent variable mixture model for longitudinal fetal growth. Biometrics 2009; 65: 1233–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Kooijman MN, Kruithof CJ, van Duijn CM, et al. The Generation R Study: design and cohort update 2017. European journal of epidemiology 2016; 31: 1243–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Gaillard R, Steegers EA, de Jongste JC, Hofman A, Jaddoe VW. Tracking of fetal growth characteristics during different trimesters and the risks of adverse birth outcomes. Int J Epidemiol 2014; 43: 1140–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Verburg B, Steegers E, De Ridder M, et al. New charts for ultrasound dating of pregnancy and assessment of fetal growth: longitudinal data from a population-based cohort study. Ultrasound in Obstetrics and Gynecology: The Official Journal of the International Society of Ultrasound in Obstetrics and Gynecology 2008; 31: 388–96. [DOI] [PubMed] [Google Scholar]
  • 25.Hadlock FP, Harrist RB, Carpenter RJ, Deter RL, Park SK. Sonographic estimation of fetal weight. The value of femur length in addition to head and abdomen measurements. Radiology 1984; 150: 535–40. [DOI] [PubMed] [Google Scholar]
  • 26.Niklasson A, Ericson A, Fryer J, Karlberg J, Lawrence C, Karlberg P. An update of the Swedish reference standards for weight, length and head circumference at birth for given gestational age (1977-1981). Acta Paediatrica 1991; 80: 756–62. [DOI] [PubMed] [Google Scholar]
  • 27.Tellegen PJ, Winkel M, Wijnberg-Williams BJ, Laros JA. Snijders-oomen nonverbal intelligence test. Son-r 21/2–7 manual and research report: Lisse: Swets & Zeitlinger B.V.; 1998. [Google Scholar]
  • 28.Achenbach TM, Rescorla LA. Manual for the ASEBA preschool forms and profiles. Burlington: University of Vermont. Research Center for Children, Youth, and Families; 2000. [Google Scholar]
  • 29.Ivanova MY, Achenbach TM, Rescorla LA, et al. Preschool psychopathology reported by parents in 23 societies: testing the seven-syndrome model of the child behavior checklist for ages 1.5–5. Journal of the American Academy of Child & Adolescent Psychiatry 2010; 49: 1215–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Hofstra MB, Van Der Ende J, Verhulst FC. Child and adolescent problems predict DSM-IV disorders in adulthood: a 14-year follow-up of a Dutch epidemiological sample. Journal of the American Academy of Child & Adolescent Psychiatry 2002; 41: 182–9. [DOI] [PubMed] [Google Scholar]
  • 31.Roza SJ, Hofstra MB, Van Der Ende J, Verhulst FC. Stable prediction of mood and anxiety disorders based on behavioral and emotional problems in childhood: A 14-year follow-up during childhood, adolescence, and young adulthood. American Journal of Psychiatry 2003; 160: 2116–21. [DOI] [PubMed] [Google Scholar]
  • 32.De Beurs E Brief symptom inventory, handleiding [Dutch Manual]. Leiden, The Netherlands: 2004. [Google Scholar]
  • 33.Arindell W, Ettema J. Handleiding bij een multidimensionele psychopathologie-indicator. Symptom Checklist SCL-90.(Herziene uitgave). 2003. [Google Scholar]
  • 34.Prieler J Raven’s advanced progressive matrices. Austria: Schufried; 2003. [Google Scholar]
  • 35.Chiesi F, Ciancaleoni M, Galli S, Primi C. Using the Advanced Progressive Matrices (Set I) to assess fluid ability in a short time frame: An item response theory–based analysis. Psychological assessment 2012; 24: 892. [DOI] [PubMed] [Google Scholar]
  • 36.Nagin DS. Group-Based Modeling of Development: Harvard University Press; 2005. [Google Scholar]
  • 37.Jones BL, Nagin DS, Roeder K. A SAS procedure based on mixture models for estimating developmental trajectories. Sociological methods & research 2001; 29: 374–93. [Google Scholar]
  • 38.Buuren Sv, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. Journal of statistical software 2010: 1–68. [Google Scholar]
  • 39.Zou R, Tiemeier H, Van Der Ende J, et al. Exposure to maternal depressive symptoms in fetal life or childhood and offspring brain development: a population-based imaging study. American Journal of Psychiatry 2019; 176: 702–10. [DOI] [PubMed] [Google Scholar]
  • 40.Unterscheider J, O’Donoghue K, Malone FD. Guidelines on fetal growth restriction: a comparison of recent national publications. American journal of perinatology 2015; 32: 307–16. [DOI] [PubMed] [Google Scholar]
  • 41.McCowan LM, Figueras F, Anderson NH. Evidence-based national guidelines for the management of suspected fetal growth restriction: comparison, consensus, and controversy. American journal of obstetrics and gynecology 2018; 218: S855–S68. [DOI] [PubMed] [Google Scholar]
  • 42.Schoenwolf GC, Bleyl SB, Brauer PR, Francis-West PH. Larsen’s human embryology E-book: Elsevier; 2021. [Google Scholar]
  • 43.Nardozza LMM, Caetano ACR, Zamarian ACP, et al. Fetal growth restriction: current knowledge. Archives of gynecology and obstetrics 2017; 295: 1061–77. [DOI] [PubMed] [Google Scholar]
  • 44.Gardosi J, Francis A, Turner S, Williams M. Customized growth charts: rationale, validation and clinical benefits. American journal of obstetrics and gynecology 2018; 218: S609–S18. [DOI] [PubMed] [Google Scholar]
  • 45.McCowan LM, Harding JE, Stewart AW. Customised birthweight centiles predict SGA pregnancies with perinatal morbidity. BJOG: An International Journal of Obstetrics & Gynaecology 2005; 112: 1026–33. [DOI] [PubMed] [Google Scholar]
  • 46.Blair EM, Liu Y, de Klerk NH, Lawrence DM. Optimal fetal growth for the Caucasian singleton and assessment of appropriateness of fetal growth: an analysis of a total population perinatal database. BMC pediatrics 2005; 5: 13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Salkever DS. Updated estimates of earnings benefits from reduced exposure of children to environmental lead. Environmental Research 1995; 70: 1–6. [DOI] [PubMed] [Google Scholar]
  • 48.Breslau N, Paneth N, Lucia VC, Paneth-Pollak R. Maternal smoking during pregnancy and offspring IQ. International journal of epidemiology 2005; 34: 1047–53. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SDC
NIHMS1708212-supplement-SDC.pdf (418.3KB, pdf)

RESOURCES