reaserch

J Abnorm Child Psychol (2006) 34:789–798 DOI 10.1007/s10802-006-9054-7

O R I G I N A L PA P E R

Does Maternal Prenatal Stress Adversely Affect the Child’s Learning and Memory at Age Six? Barbara M. Gutteling · Carolina de Weerth · Noortje Zandbelt · Eduard J. H. Mulder · Gerard H. A. Visser · Jan K. Buitelaar

Published online: 25 October 2006 C© Springer Science+Business Media, LLC 2006

Abstract Prenatal maternal stress has been shown to affect postnatal development in animals and humans. In animals, the morphology and function of the offspring’s hippocampus is negatively affected by prenatal maternal stress. The present study prospectively investigated the influence of prenatal maternal stress on learning and memory of 112 children (50 boys, 62 girls, Age: M = 6.7 years, SD = 8.4 months), with the Test of Memory and Learning (TOMAL). Maternal stress levels were determined three times during pregnancy by self- report questionnaires. Furthermore, maternal saliva cortisol samples were used as a measure of hypothalamus-pituitary- adrenal axis functioning. Results of hierarchical multivariate

B. M. Gutteling Department of Child and Adolescent Psychiatry, Utrecht University Medical Center and the Rudolf Magnus Institute of Neuroscience, Utrecht, The Netherlands

B. M. Gutteling (�) · J. K. Buitelaar Department of Psychiatry (961), Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB, Nijmegen, The Netherlands e-mail: [email protected]

C. de Weerth Department of Developmental Psychology, Radboud University Nijmegen, Nijmegen, The Netherlands

N. Zandbelt Department of Psychiatry, Radboud University Nijmegen University Medical Centre, Nijmegen and Department of Developmental Psychology, Radboud University Nijmegen, Nijmegen, The Netherlands

E. J. H. Mulder · G. H. A. Visser Department of Perinatology and Gynaecology, Utrecht University Medical Center, Utrecht, The Netherlands

regression analyses showed that maternal life events mea- sured during the first part of pregnancy were negatively as- sociated with the child’s attention/concentration index, while controlling for overall IQ, gender, and postnatal stress. No associations were found between prenatal maternal cortisol and the offspring’s learning and memory.

Keywords Prenatal maternal stress . Learning . Memory .

Child

Barker’s (1995) hypothesis of fetal programming opened the field for extensive research into the fetal origins of diseases at adult age. The hypothesis states that the environment in utero can alter the development of the fetus during particular sensitive periods, with a permanent effect on the set point of physiological systems and the phenotype in later years. The prenatal environment can be affected by external prenatal factors, such as maternal smoking (Cnattingius, Granath, Petersson, & Harlow, 1999), maternal alcohol intake (Henriksen et al., 2004), and maternal use of drugs dur- ing pregnancy (Thadani et al., 2004), all of which can result in worse birth outcomes. Furthermore, the in utero environ- ment can be influenced by internal prenatal factors, which are likewise related to a less optimal postnatal outcome. One of these factors, prenatal maternal stress, can have long-lasting consequences on the development of the offspring (Gitau, Fisk, & Glover, 2001; Huizink, Mulder, & Buitelaar, 2004a; Weinstock, 2001). In animal studies, prenatal maternal stress has been shown to affect postnatal physical outcome, and the development, behavior, and stress responses of the offspring (Huizink et al., 2004a; Mulder et al., 2002; Weinstock, 1997; Weinstock, 2001). Furthermore, in humans, prenatal mater- nal stress is related to a poor birth outcome (low birth weight, premature delivery, and a small head circumference; Gitau et al., 2001; Mulder et al., 2002), to a difficult temperament

Springer

790 J Abnorm Child Psychol (2006) 34:789–798

(Gutteling et al., 2005b; Huizink, Robles de Medina, Mulder, Visser, & Buitelaar, 2002; van den Bergh, 1990), and to behavioral and emotional problems in toddlers and in children (Gutteling et al., 2005b; O’Connor, Heron, Golding, Beveridge, & Glover, 2002).

The mechanisms involved in transduction of prenatal ma- ternal stress to the fetus are only partly understood, but three possible mechanisms have been proposed: 1) transplacen- tal transport of cortisol to the fetus, 2) release of placental hormones, induced by prenatal maternal stress, 3) maternal stress-induced decreases of the blood flow to the placenta (Huizink et al., 2004a). These mechanisms could act inde- pendently or in concert.

Particularly the hippocampus has been shown to be sensi- tive to the neurotoxic effects of high levels of endogeneous or exogeneous steroid hormones or of fetal hypoxia (Sapolsky, Uno, Rebert, & Finch, 1990). For example; treatment of mid- aged rats with corticosterone was associated with impaired hippocampus dependent spatial learning (Bodnoff et al., 1995). Further, in humans, acute glucocorticoid treatment with dexamethasone or prednisone was associated with de- creased declarative memory performance (Newcomer, Craft, Hershey, Askins, & Bardgett, 1994; Wolkowitz et al., 1990).

Recently, research in humans has also been carried out on the influence of prenatal maternal stress on the child’s cogni- tive functioning. The influence of prenatal stress (State-Trait Anxiety Inventory: STAI; Spielberger, Gorsuch, & Leshene, 1970) on attention and working memory was studied. More cognitive problems were found in prenatally stressed chil- dren at 8- to 9-years of age (van den Bergh & Marcoen, 2004), and in adolescence (van den Bergh et al., 2005a). These problems could be due to negative effects of prena- tal anxiety on the fetus’ hippocampus. On the other hand, O’Connor, Heron, Golding, and Glover (2003) did not find associations between prenatal maternal stress (Crown-Crisp index, Birtchnell, Evans, & Kennard, 1988) and attention problems in 6-year old children. However, van den Bergh and colleagues (2004, 2005a) measured prenatal stress at 12–22 weeks of gestation, while O’Connor et al. (2003) measured stress at 18 and 32 weeks prenatally. These differ- ences in pregnancy assessment moments could at least partly explain the lack of consensus in results. From the above, it is clear that further research on the relations between pre- natal maternal stress and its timing, and offspring cognitive development, is necessary before solid conclusions on these issues can be drawn.

The goal of the present study is to test the hypothesis that prenatal maternal stress is associated with a poor per- formance on memory and learning tasks in the 6-year-old offspring. Further, if so, to obtain more insight into which period of pregnancy is most sensitive for these effects, and to determine which stress factors play the most important role. A recent review on the differential effects of the tim-

ing of stress during pregnancy concluded that the results of 16 studies were inconsistent in this respect (van den Bergh et al., 2005b). Therefore, we chose not to hypothesize about which period in pregnancy would be most vulnerable for the prenatal stress effects.

Method

Participants

This study is part of an ongoing prospective longitudinal project which investigated the influence of prenatal mater- nal stress factors and endocrine factors on fetal behavior and postnatal development. A total of 230 mothers were ini- tially recruited from a population of women who visited the Outpatient Clinic of the Department of Obstetrics of the Uni- versity Medical Center Utrecht, The Netherlands, between January 1996 and July 1998. Eligibility criteria were: Dutch fluency, no drug use, no use of medication with risks for the fetus, first-time singleton pregnancy resulting in birth after 37 weeks of gestation, no major pregnancy or birth com- plications, Apgar scores > 7, and good health of the baby. All women were 15–17 weeks pregnant of their first single- ton child and participated on a voluntary basis. The local ethics committee approved the study and the participants gave written informed consent. The maternal written con- sent and child assent (verbally by the investigator) were (re)- established at the time of follow-up testing. Participants were asked to fill out questionnaires on stress and possible con- founding variables, such as educational level and smoking, at 15–17 weeks, 27–28 weeks, and 37–38 weeks of gestation. The first and last measurement moments were chosen to take place as early and late in pregnancy as possible, and the second measurement moment was taken as an extra mea- surement in between. In the Netherlands women have their first medical check at approximately 12–16 weeks of ges- tation, and this was therefore the first opportunity to invite mothers to participate in the study.

Of the 230 women who completed the questionnaires on the first occasion, 217 completed the questionnaires on the second occasion and 172 on the third occasion. Of these 172 mother-child pairs, 32 were excluded because of medical rea- sons, such as preterm delivery, stillbirth, child’s illness, and serious pregnancy complications, or because they delivered twins.

The present follow-up study took place between March 2004 and August 2004 when the children were between 5 and 8 years old. A total of 145 participants who were still taking part of the longitudinal study were invited to partic- ipate in this follow-up study (five of the mother-child pairs participated in this postnatal part of the study but not in one of the earlier phases). Of these 145 children, 113 actually

Springer

J Abnorm Child Psychol (2006) 34:789–798 791

Table 1 Descriptives of maternal and child characteristics

Maternal characteristics

Maternal age (years) 32.2 (SD = 5.0) Maternal education:

Low educational level (primary school) 4% Middle educational level (high school) 54% High educational level (college or academic education).

43%

Smoking during first trimester of pregnancy 16% Alcohol during first trimester of pregnancy 20% Child characteristics Mean (SD) Birth weight (grams) 3404 (496) Gestational age (days) 280 (8.6)

Note: SD = Standard Deviation.

participated in the home testing procedure. Loss of partic- ipants was due to lack of time (n = 4), not being able to reach the participants due to the family having moved away (n = 18), lack of interest (n = 7), school problems (n = 2) and illness of mother’s partner (n = 1). One child was unwilling to cooperate during the test-session (n = 1).

The children who participated in the present study (n = 112) and those who had participated at approxi- mately 16 weeks of gestation (n = 217) were compared by independent t-tests to check whether there was at- trition bias based upon maternal differences. No signifi- cant differences between the groups were found in gen- der, t(1,112) = − .46, p = .65; maternal educational levels, t (1,112) = − 1.06, p = .32; early period maternal stress ques- tionnaires, Daily hassles, t(1,107) = − .63, p = .53, Fear of giving birth, t(1,102) = 1.44, p = .14, Fear of having a handi- capped child, t(1,102), p = .85, Life events, t(1,115) = − .48 p = .64; or cortisol data, Morning cortisol, t(1,99) = .34, p = .74, Evening cortisol, t(1,100) = 1.02, p = .31, Slope, t(1,99) = − .36, p = .72.

Furthermore, the children who participated in the present study (n = 113) and those whose temperament and problem behavior was measured at the age of 27-months (n = 110)

were compared. No significant differences were found for temperament, t(1,110) = − .36, p = .72 or problem behavior, t(1,108) = − .07, p = .94.

Of the 112 children who participated, 50 were boys and 62 were girls. The mean age of the children was 6.7 years (SD = 8.4 months). A t-test for independent samples showed no significant difference in age between boys and girls (Mean age boys = 6.6 years and the mean age for girls = 6.9 years, t(1,110) = 1.87, p = .07).

Missing data

About 10% of the prenatal stress data, which were measured three times during pregnancy, were missing. For each partic- ular variable at each assessment moment, the missing values were replaced with the mean score of the group as a whole. Although the imputation with group means smoothes the data by reducing differences among the participants, it was pre- ferred above other alternatives, such as imputation with the subject’s score on the missing variable at another assessment moment, because the prenatal maternal stress variables were not stable throughout pregnancy (i.e. increased or decreased over time, see Table 2).

Measures

Prenatal maternal questionnaire measures. Several mater- nal questionnaires were collected in the three pregnancy periods. First, the frequency of daily hassles, in the past 2 months, was measured with the Everyday Problem List (Alledaagse Problemen Lijst, APL; Vingerhoets, Jeninga, & Menges, 1989). This questionnaire is based on different ques- tionnaires: the Daily Life Experience Questionnaire (Stone & Neale, 1982), the Daily Hassles Scale (Kanner, Coyne, Schaefer, & Lazarus, 1981), and the Everyday Problem Scale (Burks & Martin, 1985). This self-report questionnaire was used to assess the number of daily hassles during the past 2 months. Mothers completed the list by marking ‘yes’ or

Table 2 Descriptives of the prenatal and postnatal maternal stress factors

Stress factor First period of pregnancy N; Mean (SD)

Second period of pregnancy N; Mean (SD)

Third period of pregnancy N; Mean (SD)

Postnatal mean at 3 and 8 months postpartum N; Mean (SD)

Daily hassles 105; 9.9 (5.8) 111; 8.1 (6.1) 104; 6.9 (4.5) Life events 112; 283.2 (132.0) 112; 159.2 (115.2) 112; 141.1 (123.2) Fear of having a handicapped child

86; 9.1 (3.4) 85; 8.3 (2.6) 85; 8.3 (2.8)

Fear of giving birth 86; 5.5 (2.6) 85; 5.4 (2.6) 86; 5.5 (2.5) Perceived stress 111; 25.6 (4.7) Morning cortisol (nmol/l) 79; 19.7 (5.9) 77; 24.0(6.7) 56; 23.6(6.8) Evening cortisol (nmol/l) 80; 5.2 (2.0) 77; 8.3 (2.7) 56; 12.4 (3.7) Slope 81; − 2.39 (3.09) 76; − 1.57 (2.97) 56; − 4.41 (4.25) Note. N = number of participants; SD = Standard Deviation.

Springer

792 J Abnorm Child Psychol (2006) 34:789–798

‘no’ on 49 items, which were very diverse in nature (e.g. ‘You had to wait long for an appointment’, ‘You couldn’t be yourself’, ‘You had problems with friends’). The APL has been proven reliable and valid in a non-problem sample (Vingerhoets et al., 1989). Re-test (1 week later) reliability of a normative sample was .85. Norm values are available for the non-pregnant population. Norm scores of a commu- nity sample showed that < 9 is normal, 10–16 is high, and > 17 is very high. Internal consistency was adequate for the present study sample (Cronbach’s α = .84 in the first period of pregnancy, α = .81 in the second period, and α = .96 in the third period).

Second, the impact score on the Life Events Questionnaire (Vragenlijst Recent Meegemaakte Gebeurtenissen (VRMG); Van de Willige, Schreurs, Tellegen, & Zwart, 1985) was in- cluded. The Life Events Questionnaire was used to measure the negative emotionality or ‘impact’ of the ‘participants’ life events of the past year (at 15–17 weeks of gestation) or past 2 months (at 27–28 weeks of gestation and at 38–39 weeks of gestation). The impact score of this questionnaire is based on the Social Readjustment Rating Questionnaire (Holmes & Rahe, 1967). The participants rate the negative impact of the events they experienced within a list of 63 possible events on 5-point Likert scales (ranging from ‘very negative’ to ‘not negative’). The higher the impact score, the more life event-related negative emotionality had been expe- rienced by the subject. Each of these scores is then multiplied by a standard score indicating the impact of that specific life event. For example, dying of a partner or family member is multiplied by 100, moving is multiplied by 20, and being a victim of a criminal offence is multiplied by 53. The result- ing values are then summed up into a total score that is used for analysis. Thus, the total score is partly determined by how negative the participant rated a life event, and partly by an objective ‘stressor’ weight given to each particular event. More information on this questionnaire can be obtained from the first author. No norm values, reliability or validity scales of norm groups are available for this questionnaire. The in- ternal consistency for this study sample was adequate for the three pregnancy periods (Cronbach’s α = .70 for the first period, α = .76 for the second period, and α = .77 for the last period).

Third, pregnancy related anxiety was measured with the Pregnancy Related Anxiety Questionnaire-Revised, PRAQ- R (van den Bergh, 1990). This self-report questionnaire con- sists of 34 items, and has three subscales. Questions concern specific fears and worries related to pregnancy. The items are answered nominally, on a 5-point Likert scale (ranging from ‘never’ to ‘very often’). Scores can range from 3 to 15 for the scale “fear of delivery” and from 4 to 20 for the scale “fear of a handicapped child”. No norm values nor reliability or validity evidence for community samples are available. For this study, two scales were used: “fear of bear-

ing a physically or mentally handicapped child” (4 items) and “fear of giving birth” (3 items) (Huizink, Robles de Medina, Mulder, Visser, Buitelaar, 2004b). This choice was based on earlier findings between these scales and postnatal outcome in the same study sample (Gutteling, de Weerth, & Buite- laar, 2004, 2005a; Gutteling et al., 2005b). For the present study sample, construct validity was supported, since factor analysis showed that pregnancy-related anxiety in general and both subscales in particular were distinctively different from general anxiety (Huizink et al., 2004b). Also, the in- ternal consistency was satisfactory for both subscales during the total pregnancy (Cronbach’s α = .79–.83 for fear of giv- ing birth, and α = .87–.88 for fear of a handicapped child; Huizink et al., 2004b).

Prenatal maternal cortisol measures. Cortisol saliva sam- ples were collected in each of the three periods of preg- nancy (15–17 weeks, 27–28 weeks, and 37–38 weeks of gestation) on a pre-selected day every 2 h between 8 a.m. and 8 p.m. Mothers were asked to keep strictly to the sam- pling times. However, because electronic sampling time de- vices were not used, compliance can only be assumed. The mothers collected saliva by means of special cotton swabs (Salivettes, Sarstedt Inc., Germany) and stored the samples in their refrigerator until mailing them to the hospital. Fur- thermore, mothers were instructed to perform their normal daily activities and asked not to perform vigorous physical activities.

Samples were stored at − 20◦C until analysis. The cortisol concentrations were measured without extraction, using an in house competitive radioimmunoassay with polyclonal anticortisol-antibody (K7348). [1,2 − 3H(N)]- Hydrocortisone (NET 185, NEN-DUPONT, Dreiech, Ger- many) was used as a tracer after chromatographic verifica- tion of its purity. The lower limit of detection was 0.5 nmol/L and inter-assay variation was 11.0; 8.2; and 7.6% at 4.7; 9.7 and 14.0 nmol/L respectively (N = 20).

Because earlier results in this longitudinal project showed associations between prenatal morning cortisol and postnatal child characteristics (Gutteling, et al., 2004, 2005a; Huizink, Robles de Medina, Mulder, Visser, & Buitelaar, 2003), the 8 a.m. values of each pregnancy period were used as indepen- dent variables. Additionally, the evening (8 p.m.) values and slopes of the cortisol curves for each pregnancy period were included as independent variables in an exploratory fashion. SPSS version 12.1 was used to calculate the slope that con- sisted of the 7 diurnal measurements. It was calculated as follows: We performed general linear modeling using the mothers as random factor, the slope as the dependent fac- tor and the 7 cortisol variables (8 a.m. till 8 p.m.) as the covariates. The Beta’s of the main effects were considered the slope. This slope in cortisol levels throughout the day is related to the shape of the circadian curve and to recovery

Springer

J Abnorm Child Psychol (2006) 34:789–798 793

from daily stress, and as such is often used to study relations between HPA axis activity and behavior (see e.g. Li et al., 2005; White, Gunnar, Larson, Donzella, & Barr, 2000). Be- cause the cortisol data were not normally distributed, we transformed with logarithm and used the transformed data in the regression analyses.

Child measures. The children’s learning and memory func- tion was assessed with the neuropsychological Test of Mem- ory and Learning (TOMAL, Reynolds & Bigler, 1994). The TOMAL is intended to measure memory and learning in children and adolescents between the ages of 5 and 20 years. It consists of 14 subtests, 10 of which form the core subtests.

Several indexes can be calculated to provide information on specific components of memory. The use of compos- ite indexes is recommended, because their results are more reliable than those of the individual subtests (Reynolds & Bigler, 1994). Five composite indexes can be calculated. These indexes are the Attention/Concentration Index (ACI), the Associative Recall Index (ARI), the Free Recall Index (FRI), the Learning Index (LI), and the Sequential Recall Index (SRI). All indexes have a mean of 100 and a standard deviation of 15 (Reynolds & Bigler, 1994).

The TOMAL has shown good reliability (both internal consistency and stability; e.g. all used subtests show an in- ternal consistency of Cronbach’s alphas of above .74, with 63% of the subtests revealing coefficients of above .90) as well as a good validity (content, construct and criterion va- lidity; Reynolds & Bigler, 1994). However, the memory and learning test is only validated in American children.

The first author translated the TOMAL into Dutch. In order to help avoid possibly biased results due to anxiety, nervousness and other interfering symptoms (Reynolds & Bigler, 1994), the TOMAL was administered in the home environment of each child. Four trained master students and the first author administered the TOMAL, which took approximately 1.5 h to complete. There were no significant differences in children’s scores among the different test leaders.

In our sample, the reliability for the total TOMAL was adequate (Cronbach’s α = .77 for internal consistency of the 10 core subtests). This is considered in accordance with the findings of the sample of the original authors. Internal con- sistency was moderate to good for the ACI, the LI, and the SRI (all α’s > .65) but low for the ARI and the FRI (both lower than α = .60). Both the ACI and the SRI contain the subtests digits forward, letters forward, and manual imita- tion, all indicators of concentration and working memory. Because of the overlap between these two indexes and the earlier findings of associations between prenatal maternal stress and attention and concentration (van den Bergh & Marcoen, 2004), we decided to use the ACI in the analyses. Since the ACI and the LI showed good reliability and are

completely non-overlapping indices from the TOMAL, they were used as the dependent variables in the analyses.

Confounding variables. The following possible confounding variables were assessed: child’s gender, gestational age, and birth weight, maternal age, maternal educational level, prena- tal smoking (yes/no) and alcohol use (yes/no), and postnatal maternal stress. Postnatal maternal stress levels were deter- mined when the children were 3 and 8 months old with the State-Trait Anxiety Index (STAI; Spielberger et al., 1970) and the Perceived Stress Scale (PSS, Cohen & Williamson, 1987). The STAI measured anxiety in two ways. Namely state anxiety that is conceptualized as a transitory emotional anxiety and trait anxiety that refers to relatively stable prone- ness to anxiety. The STAI-state contains 20 statements that ask the respondent how she feels at the moment (e.g. ‘I feel upset’, ‘I feel at ease’). Individuals respond to each item on a four-point Likert scale, indicating the frequency with which they feel like the statement (‘not at all’—‘very much so’). The STAI-trait contains 20 statements that ask the respon- dent how she generally feels (e.g. ‘I am a steady person’, ‘I lack self-confidence’), and which are also answered on 4-point Likert scales (‘almost never’- ‘almost always’). The PSS measured perceived stress during the last month, on a 4- point Likert scale varying from ‘never’ to ‘always’ (e.g. ‘how often have you felt nervous and “stressed”?’, ‘how often have you felt that things were going your way?’). The question- naire is reported valid and reliable (Cohen & Williamson, 1987).

Pearson’s correlations showed significant relations be- tween the 3 and 8 months’ measurements for both ques- tionnaires. Therefore, mean scores over these ages were cal- culated and used as measures of early postnatal stress and possible confounding variables.

Finally, because IQ is known to be closely related to mem- ory and learning abilities (Ackerman, Beier, & Boyle, 2005), the results of an intelligence test were included as a con- founding variable in the analyses. IQ was measured by using four subtests of the Dutch version of the Wechsler Intelli- gence Scale for Children III (WISC III, Wechsler, 2002). The four subtests were similarities, vocabulary (both verbal subtests), picture completion, and block design (both per- formance subtests). All subtests have a mean of 10 and a SD of 3 (Wechsler, 2002). A sum score of the subtests was calculated (IQ-score) and used as a possible confounding variable.

Statistical analyses

Hierarchical multiple regression analyses were performed to investigate the links between prenatal stress and memory and learning. Confounding variables that were significantly cor- related (at the .05 level or below) to the dependent variables

Springer

794 J Abnorm Child Psychol (2006) 34:789–798

Table 3 Correlations between the prenatal stress variables and memory and learning

DH1 DH2 DH3 Birt1 Birt2 Birt3 Child1 Child2 Child3 LE1 LE2 ACI LI

DH1 .04 .12 DH2 .56∗∗ .03 .11 DH3 .41∗∗ .72∗∗ .14 .07 Birth1 −.03 .08 .15 −.03 .03 Birth2 −.02 .13 .15 .76∗∗ −.08 .02 Birth3 .05 .22∗ .23∗ .63∗∗ .74∗∗ .08 .07 Child1 −.12 .02 .10 .28∗∗ .17 .16 −.00 −.14 Child2 .03 .10 .22∗ .32∗∗ .21∗ .33∗∗ .61∗∗ .04 −.19 Child3 .07 .16 .27∗∗ .18 .17 .31∗∗ .56∗∗ .74∗∗ .01 −.10 LE1 .29∗∗ .16 .15 .16 .19 .21∗ −.04 .01 −.01 −.16 .02 LE2 .16 .19∗ .14 .09 .14 .19∗ .10 .13 .14 .66∗∗ −.13 −.01 LE3 .08 .19∗ .19∗ .09 .13 .20∗ .03 .04 .05 .60∗∗ .71∗∗ −.10 .11 Note.∗∗Correlation is significant at the 0.01 level. ∗Correlation is significant at the 0.05 level. Abbreviations: DH = Daily hassles, LE = Life events, Child = fear of having a handicapped child, Birth = Fear of giving birth, ACI = Attention/concentration index, LI = Learning index, 1 = pregnancy period 1, 2 = pregnancy period 2, 3 = pregnancy period 3.

were included as predictors in the regression analyses together with the prenatal stress levels. Regression analyses were carried out separately for the prenatal maternal stress factors, and for the maternal cortisol data. With all tests, p-values < 0.05 were considered statistically significant.

Outliers and influential data points were determined using Cook’s distance and leverage values. When Cook’s distance was greater than 1 or the leverage had a greater value than 2p/N (p is the number of parameters) the data point was considered an outlier. Eliminating the influential data points (n = 1) did not change the results.

Results

Preliminary analyses

Maternal measures. The total group results for prenatal ma- ternal stress variables are presented in Table 2. The partici- pants showed normal levels of stress scores.

Prenatal maternal stress variables. Correlations among the mean prenatal maternal stress variables were in gen- eral low to modest. Table 3 shows the correlations among the prenatal maternal stress variables, and between these variables and the memory and learning outcome variables.

Cortisol data. The three cortisol measures within each as- sessment period were mostly modestly to strongly correlated, and measures between adjacent periods were also often mod- estly correlated: those of period 1 with period 2, and those of period 2 with period 3 (see Table 4).

Relations between prenatal maternal stress questionnaire data and cortisol data. The correlations between prenatal maternal stress and prenatal cortisol of each pregnancy pe- riod were non-significant.

Child measures. The children showed normal memory and learning scores on the TOMAL indexes (ACI, M = 99.5, SD = 11.1, range = 71–120; LI, M = 101.1, SD = 15.3,

Table 4 Correlations between the prenatal cortisol variables and memory and learning

Morning 1 Evening 1 Slope 1 Morning 2 Evening 2 Slope 2 Morning 3 Evening 3 ACI LI

Morning 1 −.03 .12 Evening 1 .24∗ −.08 −.06 Slope 1 .22 .36∗∗ .00 .06 Morning 2 .15 −.11 −.08 −.03 −.14 Evening 2 .18 .37∗∗ .31∗ −.03 −.16 −.04 Slope 2 .31∗ .37∗∗ .27∗ .37∗∗ .74∗∗ −.12 −.15 Morning 3 −.03 −.11 −.05 .34∗ .15 .33∗ .13 .10 Evening 3 .06 .09 .22 .14 .33∗ .40∗∗ .45∗∗ .11 .15 Slope 3 .01 .06 .09 .23 .31∗ .49∗∗ .65∗∗ .85∗∗ .18 .22

Note. ∗Correlation is significant at the 0.05 level.∗∗Correlation is significant at the 0.01 level. Abbreviations: ACI = Attention/concentration index, LI = Learning index, 1 = pregnancy period 1, 2 = pregnancy period 2, 3 = pregnancy period 3.

Springer

J Abnorm Child Psychol (2006) 34:789–798 795

Table 5 Hierarchical regression analyses of the prenatal maternal stress predictors measured during the first period of pregnancy and postnatal memory and learning

Attention/concentration index Learning index

R-square df p-value R-square df p-value Model 1 .25 4;110 .00∗∗ .18 4;110 .00∗∗

Beta t p-value Beta t p-value (Constant) 7.81 .00∗∗ 5.67 .00∗∗

WISC .47 5.52 .00∗∗ .41 4.63 .00∗∗

Gender .21 2.44 .02∗ .13 1.51 .14 Postnatal perceived stress .02 .18 .86 −.06 −.72 .47 Smoking −.03 −.38 .71 −.08 −.94 .35

R-square change df p-value R-square change df p-value Model 2 .05 8;110 .00∗∗ .03 8;110 .00∗∗

Beta t p-value Beta t p-value (Constant) 8.22 .00∗∗ 5.73 .00∗∗

Daily hassles −.01 −.07 .94 .08 .82 .41 Life events −.22 −2.42 .02∗ −.01 −.13 .90 Fear of giving birth −.02 −.18 .86 .09 .94 .35 Fear of having a handicapped child −.06 −.66 .51 −.16 −1.71 .09 WISC .49 5.79 .00∗∗ .40 4.38 .00∗∗

Gender .18 2.16 .03∗ .16 1.75 .08 Postnatal perceived stress .03 .37 .71 −.08 −.83 .41 Smoking .03 .31 .76 −.10 −1.03 .31

range 53–125) and the WISC (M = 43.3, SD = 7.6, range = 29–63). Relations between confounding variables and dependent variables. An analysis of the correlations (Pearson or Spear- man, where appropriate) between the continuous confound- ing variables; gestational age, birth weight, maternal age, and maternal educational level, and the TOMAL indexes indicated significant positive correlations between the IQ- scores and the learning and memory variables (ACI and IQ r(1,110) = .46, p < .01, LI and IQ r(1,110) = .39, p < .01). Since smoking, alcohol use, and gender were dichotomous variables, independent t-tests were performed to investigate differences in memory and learning performance behavior between the groups. No significant results were found for smoking and alcohol use. Results showed that girls per- formed better than boys on the ACI, t(1,110) = 2.0, p = .05. Based on these results, IQ and gender were included in the regression analyses as confounding variables. Furthermore, smoking was also entered in the analyses, because earlier results within the same study sample have shown signifi- cant effects of maternal smoking (Gutteling et al., 2005b). Postnatal perceived stress and postnatal state anxiety cor- related significantly (r(1,108) = .68, p < .01). Therefore, we decided to exclude the STAI, as earlier in this longitudinal study postnatal maternal perceived stress had been found to influence the effect of prenatal stress on postnatal outcome (Gutteling et al., 2005b).

Main analyses

Hierarchical regressions were conducted using prenatal stress and cortisol levels to predict the ACI index and the learning index. At step one of each regression, confounding variables (IQ, gender, smoking and postnatal stress) were entered, and at step two the predictor (either prenatal stress or cortisol levels) was added to see if it contributed unique incremental variance in the outcome variable. Separate anal- yses were performed for the prenatal maternal stress scores and the cortisol levels (entering both types of variables to- gether does not alter the results). The analyses showed that the impact score of life events measured during the first pe- riod of pregnancy was negatively associated with the atten- tion/concentration index (ACI). The models for the prenatal maternal stress factors of the first pregnancy period are pre- sented in Table 5.

No significant associations were found for prenatal ma- ternal stress measures assessed later in pregnancy, nor for prenatal maternal cortisol levels.

Discussion

Our main finding was in the expected direction, i.e. that a higher negative impact of maternal life events measured during the first period of pregnancy (15–17 weeks) pre- dicted lower scores on the 6-year-old child’s attention/

Springer

796 J Abnorm Child Psychol (2006) 34:789–798

concentration index (ACI). This result is similar to that of van den Bergh and colleagues (van den Bergh & Marcoen, 2004; van den Bergh et al., 2005a) who studied the influence of prenatal stress on attention and working memory at childhood and adolescence. The authors did not find significant results for stress measured later in pregnancy, and concluded that prenatal stress in early pregnancy was associated with more cognitive problems in both childhood and adolescence. These findings point in the direction of possible programming effects of early maternal pregnancy stress on the fetal brain, resulting in attention and concentra- tion problems in childhood and adolescence. Future studies using, for example, functional Magnetic Resonance Imaging should give more insight in which area of the human brain is particularly affected by prenatal maternal stress.

Our results are not in line with those of O’Connor et al. (2003) who showed no associations between prenatal ma- ternal anxiety and attention problems, measured by parental reports, in 6-year old children. Furthermore, we did not find an association between prenatal stress and attention prob- lems, measured with the Child Behavior Checklist at 5-years of age in this sample (Gutteling, de Weerth, & Buitelaar, submitted). However, it is important to note that the associ- ations between prenatal maternal stress and attention were consistently found when the information on the offspring’s attention was obtained by neuropsychological or develop- mental tests and/or trained observers’ information, such as in the present and earlier studies (van den Berg et al., 2005a, 2005b; Huizink et al., 2002), while no associations were found when using questionnaires filled in by the mother (Gutteling et al., submitted, O’Connor et al., 2002). Because the effects found in the different studies are significant but of small effect size, it is possible that the mothers of young children are not aware of memory and/or attention deficits, as these might only be detectable through specialized tests and observations.

In the present study, the offspring’s’ learning and mem- ory scores were associated only with the negative impact of life events and not with other prenatal maternal stress vari- ables. Earlier reports of the same longitudinal study showed the offspring’s’ temperament and cortisol levels to be related to other indicators of prenatal maternal stress, namely daily hassles, perceived stress and fear of having a handicapped child (Gutteling et al., 2004, 2005a, 2005b). However, the measures of prenatal maternal stress are all different in na- ture and assess different aspects of the stress and anxiety experienced by the pregnant women. This is reflected in the weak to modest correlations among the women’s scores on the different questionnaires. These variations in experienced stress measured by the different questionnaires could also be reflected in different effects on maternal physiology that in turn could produce different effects on postnatal outcome. Nonetheless, the findings with life events of the present study

are in line with earlier published reports of other researchers. Lee, Chang and Lung (2006) also determined negative asso- ciations between prenatal maternal life events and offspring attention and concentration problems in a retrospective study. Also Obel (2003) reported life events in the first and second trimester of pregnancy to be related to attention problems in the offspring.

No relationships were found between prenatal mater- nal basal cortisol measures and the child’s memory and learning. Apparently, prenatal maternal basal cortisol as as- sessed in this study is not related to memory and learning in 6-year-old children. Earlier results of this longitudinal study showed positive relations between prenatal maternal corti- sol and offspring cortisol reactivity levels (Gutteling et al., 2004, 2005a), but no relation between maternal cortisol and offspring behavior (Gutteling et al., 2005b). However, it is possible that series of maternal cortisol samples taken on 1 day per pregnancy period, as was the case in our study, do not adequately reflect a pregnant woman’s physiological stress status. Future studies should measure basal cortisol on series of days, cortisol reactivity to stressors (de Weerth & Buite- laar, 2005), and also sample other HPA axis hormones, such as CRH and ACTH, in order to more rigorously investigate possible links between physiological measures of maternal pregnancy HPA axis functioning and offspring memory and learning.

There are several possible explanations for finding only one association between our prenatal maternal stress mea- sures, and the offspring’s learning and memory scores. First, the mothers in our study group had low to moderate stress levels. Perhaps higher stress levels are necessary for obtain- ing more general and severe learning and memory deficits. Second, it is known that negative effects of cortisol levels on memory performance appear in aged humans (Lupien et al., 2005). Also, memory impairment as a result of early mater- nal deprivation has been found to become more visible in aged rats as compared to younger rats (Oitzl, Workel, Flut- tert, Frosch, & de Kloet, 2000). It is therefore possible that greater effects of prenatal stress on memory and learning are not yet very visible in the children of the present study. Longitudinal studies and follow-ups are therefore necessary to obtain more insight into the long-term effects of prenatal maternal stress on memory and learning.

Limitations of the present study are the relatively small sample size, and the lack of a genetic informed design, which precludes to take account of the influence of genetic factors. Maternal genetic vulnerability factors, which can be passed on to the child, are most probably at least partly responsi- ble for both prenatal maternal stress levels and the child’s memory and learning. Finally, clinical cut-off scores of the TOMAL are unavailable for the Dutch population; therefore the clinical relevance of the differences in memory/attention performance is not clear.

Springer

J Abnorm Child Psychol (2006) 34:789–798 797

The overall conclusion from this study is that in a women with low to moderate levels of early pregnancy mater- nal stress, higher levels of stress appear to be associated with slight decreases in attention/concentration scores in the school-aged offspring. More research is needed to determine whether these effects can be replicated in other populations, and whether they become larger with age and in the off- spring of pregnant women with abnormally high amounts of prenatal stress.

Acknowledgments The authors wish to express their gratitude to the parents and children who participated in this study, to Anja C. Huizink and Pascale G. Robles de Medina for data collection in the prenatal phase, to Reinier K. J. Hoogendorp for his helpful comments, and to the Van der Gaag Stichting (KNAW) and the Praeventie fonds for financial support (Zorg Onderzoek Nederland, 28-2685).

References

Ackerman, P. L., Beier, M. E., & Boyle, M. O. (2005). Working memory and intelligence: The same or different constructs. Psychological Bulletin, 31, 30–60.

Barker, D. J. P. (1995). The fetal origins of adult disease. Proceedings of the Royal Society of London. Series B. Biological Sciences, 262, 37–43.

Birtchnell, J., Evans, C., & Kennard, J. (1988). The total score of the Crown-Crisp experiential index: a useful and valid measure of psychoneurotic pathology. British Journal of Medical Psychology, 61, 255–266.

Bodnoff, S. R., Humphreys, A. G., Lehman, J. C., Diamond, D. M., Rose, G. M., & Meaney, M. J. (1995). Enduring effects of chronic corticosterone treatment on spatial learning, synaptic plasticity, and hippocampal neuropathology in young and mid-aged rats. Journal of Neuroscience, 15, 61–69.

Burks, N., & Martin, B. (1985). Everyday problems and life change events: ongoing versus acute sources of stress. Stress, spring, 27– 35.

Cnattingius, S., Granath, F., Petersson, G., & Harlow, B. L. (1999). The influence of gestational age and smoking habits on the risk of subsequent preterm deliveries. New England Journal of Medicine, 341, 943–948.

Cohen, S., & Williamson, G. M. (1987). Perceived stress in a probability sample of the United States. In S. Spacapan & S. Oskamp (Eds.), The social psychology of health (pp. 31–49). Newbury Park, CA: Sage Publications.

de Weerth, C., & Buitelaar, J. K. (2005). Physiological stress reactivity in human pregnancy—a review. Neuroscience & Biobehavioral Reviews, 29, 293–310.

Gitau, R., Fisk, N. M., & Glover, V. (2001). Maternal stress in pregnancy and its effect on the human foetus: An overview of research finding. Stress, 4, 195–203.

Gutteling, B. M., de Weerth, C., & Buitelaar, J. K. (2004). Prenatal maternal stress and 4-6 year old children’s salivary cortisol con- centrations pre-and post-vaccination. Stress, 7, 257–260.

Gutteling, B. M., de Weerth, C., & Buitelaar, J. K. (2005a). Prenatal stress and children’s cortisol reaction to the first day of school. Psychoneuroendocrinology, 30, 541–549.

Gutteling, B. M., de Weerth, C., Willemsen-Swinkels, S. H. N., Huizink, A. C., Mulder, E. J. H., Visser, G. H. A., & Buitelaar, J. K. (2005b) The effects of prenatal stress on temperament and problem be- havior of 27-month-old toddlers. European Child & Adolescent Psychiatry, 14, 41–51.

Henriksen, T. B., Hjollund, N. H., Jensen, T. K., Bonde, J. P., Andersson, A. M., Kolstad, H., Ernst, E., Giwercman, A., Skakkebaek, N. E., & Olsen, J. (2004). Alcohol consumption at the time of conception and spontaneous abortion. American Journal of Epidemiology, 160, 661–667.

Holmes, T. H., & Rahe, R. H. (1967). The social readjustment rating scale. Journal of Psychosomatic Research, 11, 213–218.

Huizink, A. C., Mulder, E. J. H., & Buitelaar, J. K. (2004a). Prenatal stress and risk for psychopathology. Specific effects or induc- tion of general susceptibility. Psychological Bulletin, 130, 115– 142.

Huizink, A. C., Mulder, E. J. H., Robles de Medina, P. G., Visser, G. H. A., & Buitelaar, J. K. (2004b). Is pregnancy anxiety a distinctive syndrome? Early Human Development, 79, 81–91.

Huizink, A. C., Robles de Medina, P. G., Mulder, E. J. H., Visser, G. H. A., & Buitelaar, J. K. (2002). Psychological measures of prenatal stress as predictors of infant temperament. Journal of American Academy of Child and Adolescent Psychiatry, 41, 1078– 1085.

Huizink, A. C., Robles de Medina, P. G., Mulder, E. J. H., Visser, G. H. A., & Buitelaar, J. K. (2003). Stress during pregnancy is associated with developmental outcome in infancy. Journal of Child Psychology and Psychiatry, 44, 810–818.

Kanner, A. D., Coyne, J. C., Schaefer, C., & Lazarus, R. S. (1981). Com- parison of two models of stress measurement: daily hassles and uplifts versus major life events. Journal of Behavioral Medicine, 4, 1–39.

Lee, C. Y., Chang, Y. Y., & Lung, F. W. (2006). The marriage-related risk factors during maternal pregnancy in children with attention- deficit hyperactivity disorder. Child: Care, Health & Development, 32, 205–212.

Li, G., Cherrier, M. M., Tsuang, D. W., Petrie, E. C., Colasurdo, E. A., Craft, S., Schellenberg, G. D., Peskind, E. R., Raskind, M. A., & Wilkinson, C. W. (2005). Salivary cortisol and memory function in human aging. Neurobiology of Aging, in press.

Lupien, S. J., Fiocco, A., Wan, N., Maheu, F., Lord, C., Schramek, T., & Tu, M. T. (2005). Stress hormones and human memory function across the lifespan. Psychoneuroendocrinology, 30, 225– 242.

Mulder, E. J. H., Robles de Medina, P. G., Huizink, A. C., van den Bergh, B. R. H., Buitelaar, J. K., & Visser, G. H. A. (2002). Prenatal maternal stress: effects on pregnancy and the (unborn) child. Early Human Development, 70, 3–14.

Newcomer, J. W., Craft, S., Hershey, T., Askins, K., & Bardgett, M. E. (1994). Glucocorticoid-induced impairment in declarative mem- ory performance in adult humans. Journal of Neuroscience, 14, 2047–2053.

Obel, C. (2003). Stress during pregnancy and fetal brain develop- ment. Unpublished doctoral dissertation, University of Aarhus, Denmark.

O’Connor, T. G., Heron, J., Golding, J., Beveridge, M., & Glover, V. (2002). Maternal antenatal anxiety and children’s behav- ioral/emotional problems at 4 years. British Journal of Psychiatry, 180, 502–508.

O’Connor, T. G., Heron, J., Golding, J., & Glover, V. (2003). Ma- ternal antenatal anxiety and behavioural/emotional problems in children: a test of a programming hypothesis. Journal of Child Psychology and Psychiatry and Allied Disciplines, 44, 1025– 1036.

Oitzl, M. S., Workel, J. O., Fluttert, M., Frösch, F., & de Kloet, E. R. (2000). Maternal deprivation affects behaviour from youth to senescence: amplification of individual differences in spatial learning and memory in senescent Brown Norway rats. European Journal of Neuroscience, 12, 3771–3780.

Reynolds, C. R., & Bigler, E. D. (1994). Test of Memory and Learning. TX: Pro-ED, Inc.

Springer

798 J Abnorm Child Psychol (2006) 34:789–798

Sapolsky, R. M., Uno, H., Rebert, C. S., & Finch, C. E. (1990). Hip- pocampal damage associated with prolonged glucocorticoid ex- posure in primates. Journal of Neuroscience, 10, 2897–2902.

Spielberger, C. D., Gorsuch, I., & Leshene, R. E. (1970). Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psy- chologists Press.

Stone, A. A., & Neale, I. M. (1982). Development of a methodology for assessing daily experiences. In A. Blaum & E. F. Singer (Eds.), Advances in Environmental Psychology. Environment and health. Vol. 4 pp. 49–83. Hillsdale, NJ: Erlbaum.

Thadani, P. V., Strauss, J. F., III, Dey, S. K., Anderson, V. M., Audus, K. L., Coats, K. S., Cross, J. C., Erlebacher, A., Ganapathy, V., Linzer, D. I., Miller, R. K., Novak, D. A., Rapaka, R. S., Sadovsky, Y., Salafia, C. M., Soares, M., & Unadkat, J. (2004). National Insti- tute on Drug Abuse Conference report on placental proteins, drug transport, and fetal: development. American Journal of Obstetrics and Gynecology, 191, 1858–1862.

van de Willige, G., Schreurs, P., Tellegen, B., & Zwart, F. (1985). Het meten van ‘life events’: de Vragenlijst Recent Meegemaakte Gebeurtenissen (VRMG). Nederlands Tijdschrift voor Psycholo- gie, 40, 1–19.

van den Bergh, B. R. H. (1990). The influence of maternal emotions during pregnancy on fetal and neonatal behavior. Pre and Peri Natal Psychology Journal, 5, 119–130.

van den Bergh, B. R. H., & Marcoen, A. (2004). High antenatal maternal anxiety is related to ADHD symptoms, externalizing problems, and anxiety in 8- and 9-year-olds. Child Development, 75, 1085– 1097.

van den Bergh, B. R. H., Mennes, M., Oosterlaan, J., Stevens, V., Stiers, P., Marcoen, A., & Lagae, L. (2005a). High antenatal maternal anx-

iety is related to impulsivity during performance on cognitive tasks in 14-and 15-year-olds. Neuroscience & Biobehavioral Reviews, 29, 259–269.

van den Bergh, B. R. H., Mulder, E. J. H., Mennes, M., & Glover, V. (2005b). Antenatal maternal anxiety and stress and the neurobe- havioural development of the fetus and child: links and possible mechanisms. A review. Neuroscience & Biobehavioral Reviews, 29, 237–358.

Vingerhoets, A. J. J. M., Jeninga, A. J., & Menges, L. J. (1989). Het meten van chronische en alledaagse stressoren: Eerste onder- zoekservaringen met de Alledaagse Problemen Lijst (APL)II [The measurement of daily hassles and chronic stressors: the develop- ment of the Everyday Problem Checklist (EPCL, Dutch: APL)]. GedragGezond, 17, 10–17.

Wechsler, D. (2002). Wechsler Intelligence Scale for Children Derde Editie [Third Edition] NL: handleiding [manual]. Amsterdam: Nederland Instituut voor Psychologen Dienstencentrum.

Weinstock, M. (1997). Does prenatal stress impair coping and regu- lation of hypothalamic-pituitary-adrenal axis? Neuroscience and Biobehavioral Reviews, 21, 1–10.

Weinstock, M. (2001). Alterations induced by gestational stress in brain morphology and behavior of the offspring. Progress in Neurobi- ology, 65, 427–451.

White, B. P., Gunnar, M. R., Larson, M. C., Donzella, B., & Barr, R. G. (2000). Behavioral and physiological responsivity, sleep, and patterns of daily cortisol production in infants with and without colic. Child Development, 71, 862–877.

Wolkowitz, O. M., Reus, V. I., Weingartner, H., Thompson, K., Breier, A., Rubinow, & Pickar, D. (1990). Cognitive effects of corticos- teroids. American Journal of Psychiatry, 10, 1297–1303.

Springer