aje oxford bse

American Journal of Epidemiology

ª The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health.

All rights reserved. For permissions, please e-mail: [email protected].

Vol. 168, No. 10

DOI: 10.1093/aje/kwn233

Advance Access publication September 30, 2008

Original Contribution

What Is Spared by Fetal Brain-Sparing? Fetal Circulatory Redistribution andBehavioral Problems in the General Population

Sabine J. Roza, Eric A. P. Steegers, Bero O. Verburg, Vincent W. V. Jaddoe, Henriette A. Moll,Albert Hofman, Frank C. Verhulst, and Henning Tiemeier

Received for publication January 22, 2008; accepted for publication July 11, 2008.

Intrauterine growth restriction has been linked to infant behavioral problems. While typically only birth weight isexamined, here the authors assessed fetal circulatory redistribution, also called the ‘‘brain-sparing effect,’’ which isa fetal adaptive reaction to placental insufficiency. They aimed to investigate whether fetal circulatory redistributionprotects against behavioral problems. Within the Generation R Study (Rotterdam, the Netherlands, 2003–2007),fetal circulation variables for the umbilical artery and the middle and anterior cerebral arteries were assessed withDoppler ultrasound in late pregnancy. Ratios between placental resistance and cerebral resistance were related tobehavioral problems, as measured by the Child Behavior Checklist, in 935 toddlers aged 18 months. The umbilical/anterior cerebral ratio was associated with the Total Problems summary score from the Child Behavior Checklist(per standard-deviation increase, odds ratio ¼ 1.2, 95% confidence interval: 1.0, 1.5). Children with higher umbilical/anterior cerebral ratios had higher risks of internalizing problems, emotional reactivity, somatic complaints, andattention problems. A high umbilical/middle cerebral ratio was related to higher scores on the Internalizing andSomatic Complaints scales. The authors conclude that infants with circulatory redistribution in gestation are morelikely to have behavioral problems. This suggests that ‘‘brain-sparing’’ does not completely spare the brain andindicates underlying pathology with consequences for later behavior.

cerebrovascular circulation; fetal blood; fetal hypoxia; infant behavior

Abbreviations: CI, confidence interval; OR, odds ratio; SD, standard deviation; U/C, umbilical/cerebral.

An adverse intrauterine environment may have lifelongconsequences for the developing fetus. A nonoptimal environ-ment in utero is related to somatic (1) and psychiatric (2–5)disorders. Maternal undernutrition during pregnancy has beenassociated with schizophrenia (2, 3), antisocial personalitydisorder (4), and depression (5). Moreover, researchers inseveral population-based studies have described relationsbetween low birth weight and behavior and cognition inchildhood and adulthood (6–9). Animal and clinical studieshave shown that intrauterine growth restriction is associatedwith alterations in brain tissue volume (10, 11) and withdisabilities in motor skills, cognitive function, concentra-tion, attention, and mood (12–14).

Intrauterine growth restriction is most commonly causedby placental insufficiency (15). Animal studies have shown

that in the presence of intrauterine growth restriction due toplacental insufficiency, the central nervous system is pref-erentially perfused, which is intended to maintain oxygensupply to the brain as much as possible (16, 17). In humans,this fetal adaptive mechanism, the ‘‘brain-sparing effect,’’ canbe demonstrated with Doppler ultrasound. A raised pulsatilityindex in the umbilical artery, reflecting a reduced number ofarterioles, infarction, and thrombosis (18), identifies fetusesthat are growth-restricted due to placental dysfunction. Thefetus adapts to placental insufficiency by vasodilatation of thecerebral circulation, which can be detected as a decreasedpulsatility index in the cerebral arteries (19). Thus, the in-dicator of the ‘‘brain-sparing effect’’ is a raised ratio betweenthe umbilical artery pulsatility index and the middle cerebralartery pulsatility index (hereafter called the U/C ratio) (20).

Correspondence to Dr. Henning Tiemeier, Department of Child and Adolescent Psychiatry, Erasmus University Medical Center, P.O. Box 2060,

3000 CB Rotterdam, the Netherlands (e-mail: [email protected]).

1145 Am J Epidemiol 2008;168:1145–1152

by guest on April 18, 2014

http://aje.oxfordjournals.org/D

ownloaded from

http://aje.oxfordjournals.org/


The term ‘‘brain-sparing’’ refers to relative protection of thebrain as compared with other organs during fetal develop-ment, but this does not guarantee normal development afterbirth. Scherjon et al. (21) showed in a clinical sample ofmainly growth-restricted preterm infants that brain-sparingpredicts cognitive deficits at the age of 5 years. Experimen-tally induced placental insufficiency in animals altered brainstructure in the offspring, causing reduced brain weight,ventriculomegaly, and volumetric reductions in the basalganglia and the hippocampus (22).

We examined the hypothesis that changes in blood flow inthe middle and anterior cerebral arteries are associated withbehavioral and emotional problems in a population-basedcohort. Doppler ultrasound was used to measure U/C ratiosfor both the anterior and middle cerebral arteries. We in-vestigated whether raised U/C ratios were related to an in-creased risk of behavioral problems.

MATERIALS AND METHODS

Setting

The study was conducted within a subgroup of theGeneration R Study, a population-based prospective studyof a cohort from fetal life to young adulthood in Rotterdam,the Netherlands, which has been previously described (23).Briefly, this cohort included 9,778 mothers of different eth-nicities and their children. Midwives and gynecologists inthe study area informed eligible pregnant women about thestudy at their first antenatal visit. Detailed prenatal ultra-sound assessments were conducted in a subgroup of 1,232Dutch pregnant women, the Focus cohort, which is ethnicallyhomogeneous, to exclude confounding or effect modificationby ethnicity. No other inclusion or exclusion criteria for par-ticipation in the Focus cohort were used. All children wereborn between February 2003 and August 2005 and formeda prenatally enrolled birth cohort.

The study was approved by the Medical Ethics Committeeof Erasmus Medical Center, Rotterdam. Written informedconsent was obtained from all adult participants.

Study population

A total of 1,232 women and their children were enrolledin the Generation R Focus Study during pregnancy. Weexcluded twin pregnancies (n ¼ 15) and pregnancies lead-ing to perinatal death (n ¼ 2). Moreover, we randomly ex-cluded 1 child of each sibling pair (n ¼ 19) to avoid biasdue to paired data. Of the remaining 1,196 fetuses, interpret-able Doppler measurement of the umbilical artery and ce-rebral arteries was performed in 1,135. Information onbehavioral and emotional problems occurring between theages of 17.5 months and 22 months was available for 935(82%) of these children.

Fetal circulation

Three trained sonographers carried out fetal ultrasoundexaminations in late pregnancy (median gestational age,

30 weeks). Online measurements included Doppler meas-urements of the umbilical artery and the anterior and middlecerebral arteries (24). The umbilical artery pulsatility indexwas measured in a free-floating loop of the umbilical cord.Measurements of the middle and anterior cerebral arterypulsatility indices were performed with color Doppler visu-alization of the circle of Willis in the fetal brain. Forreliability analyses, intraclass correlation coefficients be-tween and among observers were calculated in 12 subjects.Intraobserver intraclass correlation coefficients varied from0.93 to 0.98, and interobserver intraclass correlation coef-ficients varied from 0.82 to 0.91 for the Doppler measure-ments (24). During the ultrasound examination, several fetalgrowth measurements, including head circumference andabdominal circumference, were assessed using standardizedprocedures. In line with other reports on fetal brain sparing(20, 21, 25–28), we calculated the U/C ratio by dividing thepulsatility index of the umbilical artery by the pulsatilityindex of one of the cerebral arteries. Although mothersand their caregivers were informed about clinically relevantfindings such as small-for-gestational-age birth, pregnantwomen were not informed of the results of the Dopplerexaminations.

Child behavioral and emotional problems

The Child Behavior Checklist (29) for toddlers (ages 18months to 5 years) was used to obtain standardized parentalreports of children’s problem behaviors. This questionnairecontains 99 problem items, which are scored with regard to7 empirically based syndromes that were derived by factoranalyses: Emotionally Reactive, Anxious/Depressed, SomaticComplaints, Withdrawn, Sleep Problems, Attention Problems,and Aggressive Behavior. The summary Internalizing scale isa summary score for items on the first 4 syndrome scales, andthe Externalizing scale is a summary score for Attention Prob-lems and Aggressive Behavior. The sum for all 99 problemitems is the Total Problems score. Each item is scored 0, 1, or2 (0 ¼ not true, 1 ¼ somewhat or sometimes true, and 2 ¼very true or often true) on the basis of the child’s behaviorduring the preceding 2 months. Good reliability and validityhave been reported for the Child Behavior Checklist (29). Forhomogeneity, we included only children in the age rangebetween 17.5 months and 22 months. To investigate thepossibility of a ‘‘floor effect,’’ we used data on a Dutch normsample of 669 children aged 12–60 months (30). Comparingchildren younger (n ¼ 79) and older than 24 months, wefound that younger children on average scored lower on theInternalizing, Emotionally Reactive, Somatic Complaints,and Attention Problems scales. However, the percentageof children scoring at or near the minimum possible scorewas not statistically significantly smaller in the younger agegroup than in the older age group. Within our sample, thepercentages of children scoring at or near the minimumpossible score were 18.7% for Total Problems, 11.0% forInternalizing, 9.8% for Externalizing, 38.1% for Emotion-ally Reactive, 48.4% for Anxious/Depressed, 32.3% forSomatic Complaints, 60.3% for Withdrawn, 43.3% forSleep Problems, 22.9% for Attention Problems, and 7.4%for Aggressive Behavior.

1146 Roza et al.

Am J Epidemiol 2008;168:1145–1152



ownloaded from



Covariates

Data on gestational age, birth weight, gender, Apgarscores 1 and 5 minutes after birth, infant’s umbilical arteryblood pH, and hypertensive complications of pregnancywere obtained from midwife and hospital registries at birth.We established gestational age by using the fetal ultrasoundexaminations within the Generation R Study. Maternal age,maternal height, and maternal educational level were deter-mined at enrollment. To assess maternal psychopathology inmidpregnancy, we used the Brief Symptom Inventory (31, 32).Maternal smoking and alcohol use during pregnancy wereassessed using 3 prenatal questionnaires. Postnatal question-naires administered at ages 6, 12, and 24 months gatheredinformation on breastfeeding, hospitalization of the child,and visits to physicians.

Statistical analyses

We calculated standard deviations (SDs) for the U/C ra-tios for the middle and anterior cerebral arteries. The ChildBehavior Checklist summary and syndrome scales were di-chotomized, since the resulting scores were right-skewedand could not be transformed to satisfy the assumption ofnormality. Primarily, we defined a nonoptimal score as thehighest 15% of problem item scores. As a test of consis-tency, we additionally examined results for Total Problemsusing the highest 25% and highest 10% of scores as cutoffpoints for behavioral problems.

Differences in baseline characteristics between childrenin the top 15% of Total Problems and children without be-havioral problems were compared with independent t testsfor normally distributed continuous variables, Mann-WhitneyU tests for non-normally distributed continuous variables,and chi-squared statistics for categorical variables.

We used linear regression models and logistic regressionmodels to test the associations between SD scores of mea-sures of fetal circulatory redistribution, on the one hand, andgrowth measures and behavioral problems on the other hand.

All associations were adjusted for gender and age. Theother covariates were selected as a result of exploratoryanalyses and were included in the models if the effect esti-mate changed meaningfully (defined as more than 5%).Measures of association are presented with 95% confidenceintervals. Statistical analyses were carried out using theStatistical Package for the Social Sciences, version 11.0 forWindows (SPSS, Inc., Chicago, Illinois).

Response analysis

Analyses of missing behavioral outcome data showed thatmothers of children without information on behavior were2.1 years younger (95% confidence interval (CI): 1.3, 2.8;t ¼ 5.2, P < 0.001), were less educated (percent with auniversity-level education, 25.1 vs. 39.1; v2 ¼ 13 (1 df),P < 0.001), more often continued to smoke during preg-nancy (prevalence of 26.4% vs. 11.2%; v2 ¼ 32 (1 df),P < 0.001), and had higher total psychopathology symptomscores (median score of 0.1 (95% range, 0–1.4) vs. 0.1 (95%range, 0–0.7); Mann-Whitney U test: P ¼ 0.003) comparedwith mothers of children with behavioral information. Chil-

dren with and without behavioral information did not differin terms of blood-flow parameters for the umbilical andcerebral arteries, birth weight, gestational age at birth, orhospitalization.

RESULTS

Table 1 compares demographic characteristics and poten-tially confounding variables for the 135 children whoshowed problem behavior at the age of 18 months and the800 children who had no behavioral problems. Childrenwith a high score on Total Problems were more often first-born than children with a normal Total Problems score.Maternal psychopathology was also significantly higher inchildren with behavioral problems than in children withoutbehavioral problems.

The unadjusted and adjusted associations between mea-sures of fetal circulatory redistribution and growth measuresare presented in Table 2. For each SD increase in the U/Cratio for the middle cerebral artery, birth weight was 78 glower and, adjusted for gestational age, 0.14 SDs lower.Similarly, a higher U/C ratio for the anterior cerebral arterywas related to lower birth weight. Furthermore, a higher U/Cratio was related to a smaller ratio of abdominal circumfer-ence to head circumference.

Table 3 shows the relation between fetal circulatory re-distribution and the summary scales of the Child BehaviorChecklist. The U/C ratio of the anterior cerebral artery wasrelated to Total Problems and to internalizing problems. Ahigher U/C ratio for the middle cerebral artery was related toa higher score only on the summary Internalizing scale.Fetal circulatory redistribution did not predict externalizingproblems. Odds ratios for the Internalizing scale were notsignificantly different from odds ratios for the Externalizingscale (for the middle cerebral artery, the mean difference inlog odds ratios was 0.12, 95% CI: �0.14, 0.37 (P ¼ 0.3);for the anterior cerebral artery, the mean difference in logodds ratios was 0.10, 95% CI: �0.18, 0.37 (P ¼ 0.5)). Theodds ratios for statistically significant associations were all1.2 per SD increase in U/C ratio. When analyses were re-peated after dichotomization of the determinant on the up-per 15% of U/C ratio, the effect size amounted to an oddsratio of 2.11 (95% CI: 1.27, 3.49; P ¼ 0.004).

Confounding variables that changed the effect estimatesmore than 5% were gestational age at birth, birth weight,birth order, maternal educational level, maternal psychopa-thology, and physician consultations. Although we found noindication that birth weight was directly related to behav-ioral problems (for Total Problems, per 1-kg increase inbirth weight, unadjusted odds ratio (OR) ¼ 1.32, 95% CI:0.93, 1.88; P ¼ 0.13), inclusion of birth weight in the modelchanged the effect estimates for fetal circulatory redistribu-tion meaningfully. Maternal smoking was not a confoundingvariable in the association between U/C ratio and behavioraloutcome. When mothers smoked during pregnancy, the U/Cratio for the middle cerebral artery was 0.02 SDs lower(95% CI: �0.19, 0.15; P ¼ 0.8), whereas the U/C ratiofor the anterior cerebral artery was 0.12 SDs higher (95%CI: �0.06, 0.28; P ¼ 0.2). Since maternal smoking also was

Fetal Brain-Sparing and Behavioral Problems 1147

Am J Epidemiol 2008;168:1145–1152



ownloaded from



not related to behavioral outcome (Table 1) and did notchange the effect estimates meaningfully, it was excludedfrom the final models in Table 3. Similarly, pregnancy com-plications such as maternal hypertension or maternal pre-eclampsia did not confound the association between U/Cratio and problem behavior (data not shown).

Additionally, we examined whether a measure of asym-metrical growth—that is, the ratio between abdominal cir-cumference and head circumference at 30 weeks ofgestation—was related to behavioral outcome. The oddsratio for Total Problems per SD increase in the ratio ofabdominal circumference to head circumference was 1.11(95% CI: 0.90, 1.37; P ¼ 0.3). Similarly, we found no in-dication that the ratio of abdominal circumference to headcircumference was related to internalizing (OR ¼ 1.12,95% CI: 0.91, 1.38; P ¼ 0.3) or externalizing (OR ¼ 1.08,95% CI: 0.88, 1.32; P ¼ 0.5) of problems. When TotalProblems score was dichotomized at the 75th or 90th per-centile, the odds ratios per SD increase in anterior cerebralartery U/C ratio were 1.20 (95% CI: 1.01, 1.41; P ¼ 0.03)and 1.22 (95% CI: 0.97, 1.54; P ¼ 0.09), respectively.Ten fetuses were small for gestational age (<�2 SDs) at

30 weeks of gestation; exclusion of these fetuses did notmaterially change the effect estimates.

Next, we analyzed the association between fetal circula-tory redistribution and specific syndrome scales of the ChildBehavior Checklist (Table 4). Fetal redistribution to the an-terior cerebral artery was significantly related to 3 subscalesof the Child Behavior Checklist. Per SD increase in U/Cratio, the odds for having high scores on the EmotionallyReactive scale were 26% higher. Secondly, a higher anteriorcerebral artery U/C ratio also made it more likely that infantswould have attention problems or somatic complaints; theincreases in odds were 26% and 23%, respectively. Fetal re-distribution to the middle cerebral artery was related only toSomatic Complaints (per SD increase in U/C ratio, OR ¼ 1.23,95% CI: 1.03, 1.47). Physician consultation only marginallyaccounted for this association.

DISCUSSION

The main finding of this study was that signs of fetal brainsparing were associated with infant problem behavior.

Table 1. Characteristics of Subjects in the Generation R Focus Study, Rotterdam, the Netherlands, 2003–2007a

No Behavioral Problems(n 5 800)

Behavioral Problems(Top 15% of Total Problems Score)

(n 5 135) P Valueb

Mean(SD)

Median(95% Range)

%Mean(SD)

Median(95% Range)

%

Child

Gestational age, weeks 40.3 (36.0–42.4) 40.4 (37.0–42.6) 0.4

Birth weight, g 3,512 (529) 3,589 (532) 0.1

Small-for-gestational-age birth (��1 SD) 13.0 12.6 1.0

Male gender 50.0 57.0 0.1

Gestational age at time of Dopplermeasurements, weeks

30.4 (28.5–32.6) 30.3 (28.4–32.3) 0.6

Age, months 18.1 (0.6) 18.2 (0.6) 0.2

Firstborn child 62.6 71.1 0.07

Hospitalization during first 18 months of life 18.7 24.4 0.1

Regular visits to a general practitioner(�3 per 6 months)

28.4 35.8 0.1

Outpatient clinic visits (�2 per 6 months) 24.7 29.6 0.2

Mother

Maternal age, years 32.0 (3.8) 31.6 (3.6) 0.2

Maternal height, cm 171.1 (6.2) 170.6 (6.8) 0.4

Maternal educational level, university level 39.7 35.8 0.4

Maternal smoking in pregnancy, yes 19.9 18.5 0.8

Maternal alcohol use in pregnancy, yes 56.4 56.7 1.0

Maternal psychopathology score 0.10 (0.00–0.64) 0.17 (0.00–1.00) <0.001

Maternal hypertension, yes 3.7 5.3 0.3

Maternal preeclampsia, yes 1.3 3.0 0.1

Abbreviation: SD, standard deviation.a Values presented are means (standard deviations) for normally distributed continuous variables, medians (95% range) for non-normally

distributed continuous variables, and percentages for categorical variables.b P values were derived from independent t tests for normally distributed continuous variables, Mann-Whitney U tests for non-normally distrib-

uted continuous variables, and chi-squared tests for categorical variables.

1148 Roza et al.

Am J Epidemiol 2008;168:1145–1152



ownloaded from



Children with fetal circulatory redistribution to the anteriorcerebral artery had higher risks of total problems and inter-nalizing problems at the age of 18 months. This preferentialperfusion of the frontal lobes during pregnancy was alsorelated to higher scores on the syndrome scales EmotionalReactivity, Somatic Complaints, and Attention Problems.The association between preferential perfusion of the mid-dle cerebral artery and problem behavior was less evident.We found no indication that the effects were specific for

internalizing, since the effect estimates for the Internalizingand Externalizing scales were not significantly different.

The present study shows, in line with an earlier report(25), that measures of fetal circulatory redistribution arestrong predictors of lower birth weight. Furthermore, Dopplersigns of circulatory redistribution were associated witha smaller ratio of abdominal circumference to head circum-ference, indicating a relative sparing of the growth of thefetal head in a suboptimal environment. However, we found

Table 3. Associations Between Fetal Circulatory Redistribution and Summary Scales of Child Behavioral

Problems Within the Generation R Focus Study, Rotterdam, the Netherlands, 2003–2007a

Measure of FetalCirculatory Redistribution

Total Problems Internalizing Externalizing

OR 95% CI OR 95% CI OR 95% CI

Models A

U/C ratiob of middle cerebralartery (per SD)

1.10 0.92, 1.31 1.18 1.00, 1.40 1.05 0.88, 1.24

U/C ratio of anterior cerebralartery (per SD)

1.17 0.97, 1.40 1.18 0.98, 1.41 1.09 0.91, 1.30

Models B

U/C ratio of middle cerebralartery (per SD)

1.14 0.95, 1.37 1.23* 1.03, 1.47 1.09 0.91, 1.30


1.22* 1.00, 1.49 1.25* 1.02, 1.52 1.14 0.94, 1.37


* P < 0.05.a Values presented are odds ratios from logistic regression models. Odds ratios from models A were unadjusted,

and odds ratios from models B were adjusted for gender, age, gestational age at birth, birth weight, birth order,

maternal educational level, maternal psychopathology, and child hospitalization/physician consultations.b Ratio between the umbilical artery pulsatility index and the middle cerebral artery pulsatility index.

Table 2. Associations Between Fetal Circulatory Redistribution and Measures of Fetal and Neonatal Growth

Within the Generation R Focus Study, Rotterdam, the Netherlands, 2003–2007a


Birth Weight, gBirth Weight (g)Adjusted for

Gestational Ageb

Fetal AC/HCRatio at 30 Weeks

of Gestationb

b 95% CI b 95% CI b 95% CI

Models A

U/C ratioc of middle cerebralartery (per SD)

�86.6* �119.2, �52.4 �0.15* �0.21, �0.08 �0.07* �0.12, �0.02


�120.6* �156.2, �85.1 �0.20* �0.27, �0.13 �0.08* �0.14, �0.03

Models B

U/C ratio of middle cerebralartery (per SD)

�78.5* �110.1, �46.9 �0.14* �0.20, �0.08 �0.07* �0.12, �0.02


�105.9* �140.3, �71.6 �0.18* �0.25, �0.12 �0.08* �0.14, �0.02

Abbreviations: AC/HC ratio, abdominal circumference/head circumference ratio; CI, confidence interval; SD, stan-

dard deviation; U/C, umbilical/cerebral.

* P < 0.01.a Values presented are regression coefficients from linear regressionmodels. Regression coefficients frommodels

A were unadjusted. Regression coefficients from Models B were adjusted for gender, birth order, maternal height,

maternal smoking during pregnancy, maternal hypertension, and maternal preeclampsia.b Standard deviation scores, adjusted for gestational age.c Ratio between the umbilical artery pulsatility index and the middle cerebral artery pulsatility index.


Am J Epidemiol 2008;168:1145–1152



ownloaded from



little indication in this cohort that fetal or neonatal growthmeasures were associated with behavioral outcome. Studieson fetal programming have frequently used low birth weightas a proxy for an adverse intrauterine environment (7, 8, 33).However, having a low birth weight or being born small forgestational age can also be a result of constitutional factors(34). A small fetus resulting from reduced placental perfu-sion and chronic hypoxia during pregnancy can be betteridentified using blood flow parameters of the umbilical ar-tery and the cerebral arteries (35). Several other studiesshowed that Doppler signs of fetal circulatory redistributionwere associated with perinatal complications such as emer-gent cesarean sections, prematurity, and stays in intensivecare units (29, 36, 37). Animal studies provided informationon abnormalities in brain structure and function after uni-lateral ligation of the maternal uterine artery (10, 22, 38).This work suggests that the goal, to prevent cerebral hyp-oxia, cannot always be achieved by brain sparing. In otherwords, the brain is protected at the cost of other organs, butfetal circulatory redistribution does not necessarily spare thebrain in absolute terms.

Earlier studies directly related blood flow parameters ofbrain sparing to visual, neuromotor, and cognitive outcomesin clinical populations. Scherjon et al. (26) found that chil-dren with fetal brain-sparing may have accelerated myeli-nation of visual pathways during the first year of life, butthey could not demonstrate a difference in visual function-ing at age 11 years (28). Moreover, although gross neuro-motor outcome at age 3 years was not affected by fetalcirculatory redistribution (27), brain-sparing had a detrimen-tal effect on cognitive outcome at 5 years of age (21). Thesefindings seem contradictory and were not replicated in othersamples. Since the authors used different endpoints in timeand different outcome measures, it is difficult to interpret theresults. Chronic fetal hypoxia may adversely affect specificbrain regions, whereas other regions are spared. Dubiel et al.(35) suggested that the redistribution of fetal cerebral floodflow in the presence of chronic hypoxia preferentially pro-tects the frontal region of the brain, which is involved inemotional, cognitive, and motivational processes (39, 40).Our findings show that signs of brain sparing in the anterior

cerebral artery were related to a higher risk of attentionproblems and emotional reactivity, which may indicate lessmature frontal-subcortical circuitry (39). Cautiously inter-preting our results and the results of the clinical studies ofScherjon et al. (26), those parts of the brain that are involvedin higher-order brain functions may be more vulnerable tohypoxemia. Although the fetal adaptive mechanism servesto protect the frontal regions at the expense of temporal andoccipital regions in case of chronic fetal hypoxia, cognitiveand behavioral functioning may be affected earlier than vi-sual and motor function. Further research into the effects offetal circulatory redistribution on several neurobehavioraloutcomes at similar endpoints in time is needed to strengthenthis hypothesis.

Some methodological considerations need to be dis-cussed. First, our sample represented predominantly chil-dren with low vulnerability to emotional and behavioralproblems, probably because of favorable environmentalcharacteristics. However, this selection also yields advan-tages, since effects of confounding are smaller in a morehomogeneous group. Next to selective nonresponse, selec-tive loss to follow-up of healthier mothers could lead to bias.We found no indication that children with intrauterinegrowth restriction or raised U/C ratios were more often lostto follow-up. Multiple imputation methods could not beused to complete data on the outcome, since for childrenwith missing data on the outcome, highly related informa-tion (e.g., information on temperament at younger ages) wasnot available. Therefore, the possibility of biased results dueto selective attrition cannot be ruled out. Second, we usedthe Child Behavior Checklist to assess toddler behavior.Although the Child Behavior Checklist has been shown tohave good reliability and validity from age 18 months on-wards, we used this questionnaire at the lower boundary ofits age frame. Although younger children in a Dutch normsample on average score lower on several subscales, wefound no indication that a higher number of children scoreat or near the minimum possible score. Both in the Dutchnorm group and in our sample, there is a considerable flooreffect; that is, many children score at or near the minimumpossible score. Therefore, the Child Behavior Checklist

Table 4. Associations Between Fetal Circulatory Redistribution and Syndrome Scales of Child Behavioral Problems Within the Generation R

Focus Study, Rotterdam, the Netherlands, 2003–2007a


Internalizing Syndrome Scale Externalizing Syndrome Scale

EmotionallyReactive

Anxious/Depressed

SomaticComplaints

WithdrawnAttentionProblems

AggressiveBehavior

OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI

U/C ratiob of middle cerebralartery (per SD)

1.06 0.90, 1.25 1.07 0.91, 1.25 1.23* 1.03, 1.47 1.07 0.93, 1.24 1.15 0.97, 1.35 1.07 0.89, 1.29


1.26* 1.05, 1.49 1.07 0.91, 1.27 1.23* 1.01, 1.49 1.13 0.97, 1.31 1.26* 1.06, 1.51 1.11 0.91, 1.35


* P < 0.05.a Values presented are odds ratios from logistic regression models. Odds ratios were adjusted for gender, age, gestational age at birth, birth

weight, birth order, maternal educational level, maternal psychopathology, and child hospitalization/physician consultations.b Ratio between the umbilical artery pulsatility index and the middle cerebral artery pulsatility index.

1150 Roza et al.

Am J Epidemiol 2008;168:1145–1152



ownloaded from



scores were dichotomized at the level of the highest 15%. Itis unlikely that the lack of variance in the lower boundariesof the scales influenced our findings. Using a parent-reportmeasure might introduce reporter bias. Differential misclas-sification of the outcome is less likely, since mothers wereblinded to the results of the Doppler velocimetry. Exclusionof small-for-gestational-age fetuses at 30 weeks of gestationdid not materially change the effect estimates. Furthermore,we adjusted our associations for maternal symptoms of psy-chopathology, which are known to color the reporting ofactual child behavior (41). Third, residual confounding isa potential limitation. We were able to rule out the possibil-ity that our associations were explained by prematurity,sociodemographic variables, maternal smoking during preg-nancy, or medical complications after birth. Although med-ical complications after birth and maternal psychologicalsymptoms did not fully explain our results, maternal healthconcerns could be a possible residual confounder. Finally, asindicated by the small-to-moderate effect sizes, the possi-bility of chance findings cannot be ruled out, although asso-ciations were consistent when using different cutoff points.

In conclusion, the brain-sparing mechanism does not pro-tect the child from all neurodevelopmental problems. In-fants who experience circulatory redistribution withpreferential perfusion of the brain in utero are more likelyto have behavioral and emotional problems in young child-hood. Our findings suggest that the U/C ratio is a sensitivemarker of fetal growth restriction in which the fetal head isrelatively spared and is a risk indicator for behavioral out-come. Further research is needed to evaluate the consequen-ces of fetal circulatory redistribution for cognitive andbehavioral development at preschool age and school age.

ACKNOWLEDGMENTS

Author affiliations: Generation R Study Group, ErasmusUniversity Medical Center, Rotterdam, the Netherlands(Sabine J. Roza, Bero O. Verburg, Vincent W. V. Jaddoe);Department of Child and Adolescent Psychiatry, ErasmusUniversity Medical Center, Rotterdam, the Netherlands(Sabine J. Roza, Frank C. Verhulst, Henning Tiemeier);Department of Obstetrics and Gynecology, Erasmus Univer-sity Medical Center, Rotterdam, the Netherlands (Eric A. P.Steegers, Bero O. Verburg); Department of Pediatrics,Erasmus University Medical Center, Rotterdam, theNetherlands (Vincent W. V. Jaddoe, Henriette A. Moll);and Department of Epidemiology, Erasmus UniversityMedical Center, Rotterdam, the Netherlands (VincentW. V. Jaddoe, Albert Hofman, Henning Tiemeier).

The Generation R Study is conducted by ErasmusUniversity Medical Center (Rotterdam, the Netherlands) inclose collaboration with the Faculty of Social Sciences ofErasmus University, the Municipal Health Service (Rotterdamarea), the Rotterdam Homecare Foundation, and theStichting Trombosedienst and Artsenlaboratorium Rijnmond.The first phase of the Generation R Study was made possibleby financial support from Erasmus University MedicalCenter, Erasmus University, and the Netherlands Organization

for Health Research and Development. The present studywas supported by an additional grant from the NetherlandsOrganization for Health Research and Development(‘‘Geestkracht’’ program grant 10.000.1003).

The authors gratefully acknowledge the contributions ofthe general practitioners, hospitals, midwives, and pharma-cies in Rotterdam.

Conflict of interest: none declared.

REFERENCES

1. Barker DJ. Fetal origins of coronary heart disease. BMJ. 1995;311(6998):171–174.

2. Susser E, Neugebauer R, Hoek HW, et al. Schizophrenia afterprenatal famine. Further evidence. Arch Gen Psychiatry. 1996;53(1):25–31.

3. St Clair D, Xu M, Wang P, et al. Rates of adult schizophreniafollowing prenatal exposure to the Chinese famine of 1959–1961. JAMA. 2005;294(5):557–562.

4. Neugebauer R, Hoek HW, Susser E. Prenatal exposureto wartime famine and development of antisocial person-ality disorder in early adulthood. JAMA. 1999;282(5):455–462.

5. Brown AS, van Os J, Driessens C, et al. Further evidence ofrelation between prenatal famine and major affective disorder.Am J Psychiatry. 2000;157(2):190–195.

6. Gunnell D, Harrison G, Whitley E, et al. The association offetal and childhood growth with risk of schizophrenia. Cohortstudy of 720,000 Swedish men and women. Schizophr Res.2005;79(2–3):315–322.

7. Wiles NJ, Peters TJ, Leon DA, et al. Birth weight and psy-chological distress at age 45–51 years: results from theAberdeen Children of the 1950s cohort study. Br J Psychiatry.2005;187:21–28.

8. Costello EJ, Worthman C, Erkanli A, et al. Prediction fromlow birth weight to female adolescent depression: a test ofcompeting hypotheses. Arch Gen Psychiatry. 2007;64(3):338–344.

9. Nigg JT, Breslau N. Prenatal smoking exposure, low birthweight, and disruptive behavior disorders. J Am Acad ChildAdolesc Psychiatry. 2007;46(3):362–369.

10. Mallard C, Loeliger M, Copolov D, et al. Reduced number ofneurons in the hippocampus and the cerebellum in the post-natal guinea-pig following intrauterine growth-restriction.Neuroscience. 2000;100(2):327–333.

11. Rees S, Inder T. Fetal and neonatal origins of altered braindevelopment. Early Hum Dev. 2005;81(9):753–761.

12. Geva R, Eshel R, Leitner Y, et al. Memory functions of chil-dren born with asymmetric intrauterine growth restriction.Brain Res. 2006;1117(1):186–194.

13. Tolsa CB, Zimine S, Warfield SK, et al. Early alteration ofstructural and functional brain development in premature in-fants born with intrauterine growth restriction. Pediatr Res.2004;56(1):132–138.

14. Leitner Y, Fattal-Valevski A, Geva R, et al. Neurodevelop-mental outcome of children with intrauterine growth retarda-tion a longitudinal, 10-year prospective study. J Child Neurol.2007;22(5):580–587.

15. Kingdom J, Huppertz B, Seaward G, et al. Development of theplacental villous tree and its consequences for fetal growth.Eur J Obstet Gynecol Reprod Biol. 2000;92(1):35–43.

16. Baschat AA. Fetal responses to placental insufficiency: anupdate. BJOG. 2004;111(10):1031–1041.


Am J Epidemiol 2008;168:1145–1152



ownloaded from



17. Cohn HE, Sacks EJ, Heymann MA, et al. Cardiovascular re-sponses to hypoxemia and acidemia in fetal lambs. Am JObstet Gynecol. 1974;120(6):817–824.

18. Sheldon RE, Peeters LL, Jones MD Jr, et al. Redistribution ofcardiac output and oxygen delivery in the hypoxemic fetallamb. Am J Obstet Gynecol. 1979;135(8):1071–1078.

19. Chauhan SP, Magann EF. Screening for fetal growth restric-tion. Clin Obstet Gynecol. 2006;49(2):284–294.

20. Scherjon SA, Kok JH, Oosting H, et al. Fetal and neonatalcerebral circulation: a pulsed Doppler study. J Perinat Med.1992;20(1):79–82.

21. Scherjon S, Briet J, Oosting H, et al. The discrepancy betweenmaturation of visual-evoked potentials and cognitive outcomeat five years in very preterm infants with and without hemo-dynamic signs of fetal brain-sparing. Pediatrics. 2000;105(2):385–391.

22. Rehn AE, Van Den Buuse M, Copolov D, et al. An animalmodel of chronic placental insufficiency: relevance to neuro-developmental disorders including schizophrenia. Neurosci-ence. 2004;129(2):381–391.

23. Jaddoe VW, Mackenbach JP, Moll HA, et al. The Generation RStudy: design and cohort profile. Eur J Epidemiol. 2006;21(6):475–484.

24. Verburg BO, Jaddoe VW, Wladimiroff JW, et al. Fetal hemo-dynamic adaptive changes related to intrauterine growth: theGeneration R Study. Circulation. 2008;117(5):649–659.

25. Scherjon SA, Oosting H, de Visser BW, et al. Fetal brainsparing is associated with accelerated shortening of visualevoked potential latencies during early infancy. Am J ObstetGynecol. 1996;175(6):1569–1575.

26. Scherjon SA, Oosting H, Smolders-DeHaas H, et al. Neuro-developmental outcome at three years of age after fetal ‘brain-sparing’. Early Hum Dev. 1998;52(1):67–79.

27. Kok JH, Prick L, Merckel E, et al. Visual function at 11 yearsof age in preterm-born children with and without fetal brainsparing. Pediatrics. 2007;119(6):e1342–e1350.

28. Makhseed M, Jirous J, Ahmed MA, et al. Middle cerebralartery to umbilical artery resistance index ratio in the predic-tion of neonatal outcome. Int J Gynaecol Obstet. 2000;71(2):119–125.

29. Achenbach TM, Rescorla LA. Manual for the ASEBA Pre-school Forms and Profiles. Burlington, VT: Research Centerfor Children, Youth, and Families, University of Vermont;2000.

30. Tick NT, van der Ende J, Koot HM, et al. 14-Year changes inemotional and behavioral problems of very young Dutchchildren. J Am Acad Child Adolesc Psychiatry. 2007;46(10):1333–1340.

31. de Beurs E. Brief Symptom Inventory. Leiden, the Netherlands:Handleiding; 2004.

32. Derogatis LR, Melisaratos N. The Brief Symptom Inventory:an introductory report. Psychol Med. 1983;13(3):595–605.

33. Kelly YJ, Nazroo JY, McMunn A, et al. Birthweight and be-havioural problems in children: a modifiable effect? Int JEpidemiol. 2001;30(1):88–94.

34. Bryan SM, Hindmarsh PC. Normal and abnormal fetal growth.Horm Res. 2006;65(suppl 3):19–27.

35. Dubiel M, Gunnarsson GO, Gudmundsson S. Blood redistri-bution in the fetal brain during chronic hypoxia. UltrasoundObstet Gynecol. 2002;20(2):117–121.

36. Vergani P, Roncaglia N, Locatelli A, et al. Antenatal predictorsof neonatal outcome in fetal growth restriction with absentend-diastolic flow in the umbilical artery. Am J Obstet Gyne-col. 2005;193(3 pt 2):1213–1218.

37. Sterne G, Shields LE, Dubinsky TJ. Abnormal fetal cerebraland umbilical Doppler measurements in fetuses with intra-uterine growth restriction predicts the severity of perinatalmorbidity. J Clin Ultrasound. 2001;29(3):146–151.

38. Tashima L, Nakata M, Anno K, et al. Prenatal influence ofischemia-hypoxia-induced intrauterine growth retardation onbrain development and behavioral activity in rats. Biol Neo-nate. 2001;80(1):81–87.

39. Alvarez JA, Emory E. Executive function and the frontallobes: a meta-analytic review. Neuropsychol Rev. 2006;16(1):17–42.

40. Krain AL, Castellanos FX. Brain development and ADHD.Clin Psychol Rev. 2006;26(4):433–444.

41. Youngstrom E, Izard C, Ackerman B. Dysphoria-related biasin maternal ratings of children. J Consult Clin Psychol. 1999;67(6):905–916.

1152 Roza et al.

Am J Epidemiol 2008;168:1145–1152



ownloaded from



aje oxford bse

Documents