reaserch

Stress during pregnancy is associated with developmental outcome in infancy

Anja C. Huizink,1,3 Pascale G. Robles de Medina,2 Eduard J.H. Mulder,2

Gerard H.A. Visser,2 and Jan K. Buitelaar1 1University Medical Center Utrecht, Department of Child and Adolescent Psychiatry and Rudolf Magnus Institute for Neurosciences, Netherlands; 2University Medical Center Utrecht, Department of Obstetrics, Neonatology and

Gynecology and Rudolf Magnus Institute for Neurosciences, Netherlands; 3Department of Social Medicine, Institute for Research in Extramural Medicine, VU University Medical Center, Netherlands

Background: Animal studies show that prenatal maternal stress may be related to cognitive impair- ments in offspring. Therefore, we examined whether psychological and endocrinologic measures of stress during human pregnancy predicted developmental outcome of the infant at 3 and 8 months. Method: Self-report data about daily hassles and pregnancy-specific anxiety and salivary cortisol levels were collected in 170 nulliparous women in early, mid- and late pregnancy in a prospective design, in which healthy infants born at term were followed up after birth. Results: High levels of pregnancy- specific anxiety in mid-pregnancy predicted lower mental and motor developmental scores at 8 months (p < .05). High amounts of daily hassles in early pregnancy were associated with lower mental devel- opmental scores at 8 months (p < .05). Early morning values of cortisol in late pregnancy were negat- ively related to both mental and motor development at 3 months (p < .05 and p < .005, respectively) and motor development at 8 months (p < .01). On average a decline of 8 points on the mental and motor development scale was found. All results were adjusted for a large number of covariates. Conclu- sion: Stress during pregnancy appears to be one of the determinants of delay in motor and mental development in infants of 8 months of age and may be a risk factor for later developmental problems. Further systematic follow-up of the present sample is needed to determine whether these delays are transient, persistent or even progressive. Keywords: Prenatal stress, development, infancy. Abbrevi- ations: ACTH: adrenocorticotropic hormone; HPA-axis: hypothalamic-pituitary-adrenal axis; MDI: Mental Developmental Index; PDI: Psychomotor Developmental Index.

In pregnant animals induced stress has been shown to adversely affect behavioral adaptation and motor and mental development of the offspring (Weinstock, 1997). In a series of studies, Schneider and co-workers have shown that prenatal stressors adversely affect the motor and mental development of rhesus monkeys (Schneider, 1992; Schneider, Coe, & Lubach, 1992; Schneider, Roughton, Koeh- ler, & Lubach, 1999). Exposure to mild stress during mid-pregnancy, operationalized as three noise bursts over a 10-minute period five times a week, resulted in decreased motor maturity (a delay in learning to self-feed, low muscle tone, inferior bal- ance reactions, a slowed response speed, poorer coordination), and a declined attention in the first months of life in comparison with control infants (Schneider, 1992). Recently, Schneider et al. (1999) showed that these effects were most profound after exposure to stress in early gestation, but could still be found after mid- to late gestational stress. The same mild prenatal stressor appeared to have a negative effect on cognition as well. A delay in object permanence was found on a sequence of Piagetian tasks after prenatal stress (Schneider et al., 1992). Although the mechanisms underlying the transfer of maternal stress to the fetus are only partly under- stood, the hypothalamic–pituitary–adrenal (HPA) axis presumably mediates the influence of maternal

stress on the developing fetus in animals (Weinstock, 1997; Barbazanges, Piazza, Le Moal, & Maccari, 1996; McCormick, Smythe, Sharma, & Meaney, 1995). In primates, the effects of a mild stressor during pregnancy could be mimicked by prenatal exposure to adrenocorticotropic hormone (ACTH) during a 2-week period (Schneider, 1992).

In humans, the effects of naturally occurring stressors on birth outcome have been studied in prospective designs. There is substantial evidence that maternal stress is associated with premature delivery and lower birth weight adjusted for gesta- tional age (Glynn, Wadhwa, Dunkel-Schetter, Chicz- Demet, & Sandman, 2001; Dunkel-Schetter, 1998; Copper et al., 1996; Lou et al., 1994; Wadhwa, Sandman, Aporto, Dunkel-Schetter, & Garite, 1993) and with a smaller head circumference (Lou et al., 1994). The latter finding may reflect suboptimal brain development and may be a predictor of im- paired cognitive development (Hack et al., 1991; Ounsted, Moar, & Stott, 1988). One previous study investigated the effect of stress during pregnancy on the postnatal development of human infants (Van den Bergh, 1990). Measures of general anxiety in the third trimester of pregnancy were positively correla- ted with a difficult temperament of the infant at 10 weeks and 7 months after birth, but were unrelated to infant mental or motor development (Van den

Journal of Child Psychology and Psychiatry 44:6 (2003), pp 810–818

� Association for Child Psychology and Psychiatry, 2003. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Bergh, 1990). A recent large prospective epidemio- logical study found that prenatal maternal anxiety predicted behavioral and emotional problems in children at 4 years (O’Connor, Heron, Golding, Beveridge, & Glover, 2002).

The present prospective longitudinal study was designed to examine the effects of stress in human pregnancy on both motor and mental development early in life. We took account of several potential confounders. To avoid confounding influence of prematurity and previous pregnancies, only infants were included born near term of nulliparous women. We used two aspects of prenatal maternal stress in early, mid-, and late pregnancy as predictors of in- fant outcome: daily hassles and pregnancy anxiety. Daily hassles, rather than major life events, may be seen as proximal measures of stress and have been found to affect health more rapidly than major life events (DeLongis, Coyne, Dakof, Folkman, & Lazarus, 1982; McEwen & Seeman, 1999). Preg- nancy anxiety reflects a unique aspect of human pregnancy. In a previous study, we showed the ex- istence of pregnancy anxieties which were only partly related to measures of general anxiety and depres- sion (Huizink, 2000). Therefore, we were interested in the effect of daily hassles and pregnancy anxiety in pregnancy on later infant development.

The hypothalamic–pituitary–adrenal (HPA) axis has been found to be one of the mediators of the effects of prenatal maternal stress on the developing fetus in animal studies (e.g., Weinstock, 1997; Weinstock, Matlina, Maor, Rosen, & McEwen, 1992; Fride, Dan, Feldon, Halevy, & Weinstock, 1986; McCormick et al., 1995). Therefore, physiological parameters reflecting the activity of the maternal HPA axis during pregnancy were also included as predictors of postnatal infant development. For that purpose, cortisol day profiles were assessed in early, mid-, and late pregnancy.

Methods

Participants

Nulliparous women with a singleton pregnancy were recruited from the outpatient clinic of the Department of Obstetrics of the University Medical Center Utrecht (UMCU). The clinic is a first-line referral center where the majority of pregnant women seen carry a low to normal risk status and have their pregnancy supervised by midwives. The UMCU is located outside the city of Utrecht and attracts a mixed rural and urban popula- tion of patients. Eligibility criteria were: first-time sin- gleton pregnancy resulting in birth after 37 weeks of gestation, no major pregnancy complications, no major birth complications, Apgar scores >7, and good health of the baby. Women were excluded from the study if they were not fluent in Dutch, used drugs, or used medication with risks for the fetus. The institutional review board at the UMCU approved the study. All participants gave written informed consent.

Study design and measures

The participants were included in a prospective longi- tudinal study with assessments at 15–17 weeks (early pregnancy), 27–28 weeks (mid-pregnancy), and 37–38 weeks of gestation (late pregnancy), and at 10 days and at 3 and 8 months after birth. At study entry, data were collected on demographic characteristics, and on ob- stetric, medical and psychiatric histories. The assess- ments at early, mid- and late pregnancy included ultrasound recording of fetal behavior to be reported elsewhere, and a set of self-report questionnaires on various aspects of stress during pregnancy.

Predictors

To predict infant development, we used two different aspects of maternal stress, which were only moderately intercorrelated in each period of pregnancy (r’s ranging from –.03 to .26). The first psychological predictor was the rate of occurrence of daily hassles in the past months, as measured by the Everyday Problem List (Vingerhoets, Jeninga, & Menges, 1989). Examples of items are ‘You could not find important belongings’ and ‘You were trapped in a traffic jam’. Cronbach’s alphas were >.85. Daily hassles have been found to be associ- ated with increased cortisol secretion in adult males (Van Eck, Berkhof, Nicolson, & Sulon, 1996).

The second predictor, pregnancy anxiety, was as- sessed by means of the Pregnancy Related Anxieties Questionnaire-Revised (PRAQ-R; Huizink, 2000). This questionnaire was developed by confirmatory factor analysis from the PRAQ of Van den Bergh (Van den Bergh, 1990) and consisted of ten items that fitted to a three-factor model: fear of giving birth, fear of bearing a physically or mentally handicapped child, and concern about one’s own appearance. We used two subscales in the present study: fear of giving birth (3 items; scores ranging from 3 to 15) and fear of bearing a physically or mentally handicapped child (4 items; scores ranging from 4 to 20). Examples of items are: ‘I am worried about the pain of contractions and the pain during de- livery’ (fear of giving birth) and ‘I am afraid the baby will be mentally handicapped or will suffer from brain damage’ (fear of bearing a physically or mentally han- dicapped child). The items were answered on a 5-point scale, ranging from ‘never’ to ‘very often’. Cronbach’s alpha’s of the subscales were all >.76 throughout pregnancy. In another study, we found that only 17% of the variance in pregnancy anxiety could be explained by general trait anxiety and depression measures (Huiz- ink, 2000). It has been shown that pregnancy anxiety rather than general anxiety was related to birth out- come and activation of the neuroendocrine axis in pregnancy (Killingsworth Rini, Dunkel-Schetter, Wad- wha, & Sandman, 1999; Wadhwa et al., 1993).

The concentration of cortisol in maternal saliva was taken as an endocrinologic measure of maternal stress. Salivary cortisol levels reliably reflect levels of the un- bound hormone in blood (Meulenberg & Hofman, 1990; Kirschbaum & Hellhammer, 1989). Seven saliva sam- ples were collected every two hours between 8:00 a.m. and 8:00 p.m., to obtain cortisol daytime curves in each of the three periods of pregnancy. All samples were stored at –70 �C until assayed. Salivary cortisol was

Prenatal stress and infant development 811

measured without extraction using an in-house com- petitive radioimmunoassay employing a polyclonal anticortisol-antibody (K7348). [1,2]-3 H(N)-Hydrocorti- sone (NET 185, NEN-DUPONT, Dreiech, Germany) was used as a tracer following chromatic verification of its purity. The lower limit of detection was 0.5 nmol/L and interassay variation was 11.0%, 8.2%, and 7.6% at 4.7, 9.7 and 14.0 nmol/L, respectively (n ¼ 20). Reference values for adults are 4–28 nmol/L at 8:00 a.m. The early morning (8 a.m.) cortisol level is a dynamic measure and was thought to reflect the early morning peak as an anticipatory stress response. It was entered as a predictor in the present study. Also, the mean cortisol secretion during the day (from 8:00 a.m. to 8:00 p.m.) was entered as a predictor in the present study. The psychological and endocrinologic predictors proved to be uncorrelated, except for the 8 a.m. cortisol value and daily hassles (r ¼ .27, p < .05) in late pregnancy.

Outcome variables

The main dependent measures were the developmental indices of the infant at the age of 3 and 8 months after birth as assessed by the Bayley Scales of Infant Devel- opment (BSID; Bayley, 1969) in a standard test situ- ation. The examinations were performed by a psychologist who was blind to the data on stress during pregnancy. The BSID has been translated and validated in a Dutch population of infants (Van der Meulen & Smrkovky, 1983) and results in a Mental Develop- mental Index (MDI) and a Psychomotor Developmental Index (PDI). The MDI is designed to assess sensory- perceptual acuities, discriminations, and the ability to respond to these; the early acquisition of ‘object con- stancy’ and memory, learning, and problem-solving ability; vocalizations and the beginnings of verbal communication; and early evidence of the ability to form generalizations and classifications. The PDI is a measure of the degree of body control, coordination of the large muscles, and finer manipulatory skills of the hands and fingers.

Confounders

Data on other factors from the prenatal, perinatal and postnatal periods that may influence infant develop- ment served as covariates in our analyses. The educa- tional level and professional level of the pregnant woman and her partner (Van Westerlaak, Kropman, & Collaris, 1976) defined socio-economic status (SES). Smoking (number of cigarettes smoked per day) and alcohol-intake (number of beverages per week) were assessed by self-report in each period of pregnancy. Biomedical risk factors during pregnancy were added up in a cumulative score (1 point per factor). Perinatal covariates included birth weight (in grams) and gesta- tional age at birth (in weeks). Also, complications dur- ing delivery were taken into account, by calculating a cumulative score (1 point per factor). Postnatal covari- ates were breastfeeding, psychological well-being and perceived stress of the mother at 3 and 8 months fol- lowing childbirth. Psychological well-being was de- termined by the Dutch version of the General Health Questionnaire (GHQ-30; Koeter & Ormel, 1991). This

questionnaire contains 30 questions to be answered on a 4-point scale. Cronbach’s alpha was .92. Perceived stress was assessed with a Dutch translation of the 14- item Perceived Stress Scale (Cohen & Williamson, 1987). The scale measures perceived stress over the last month on a 4-point scale, ranging from ‘never’ to ‘al- ways’. Cronbach’s alpha was .92. Postnatal depression scores were determined by means of the 10-item Edin- burgh Postnatal Depression Scale (EPDS; Cox, Holden, & Sagovsky, 1987). Cronbach’s alpha was .86.

Statistical analysis

Categorical or interval-scaled covariates (SES, maternal age, gestational age, birth weight, postnatal stress and depression of the mother) were tested for their re- lationships with the dependent and the independent variables by means of correlations (Pearson product- moment or Spearman rank-order correlations where appropriate) and regression analysis. Only covariates which were significantly related to the predictors and dependent variables were included in further analyses. For each predictor one MANCOVA was performed, with MDI and PDI scores at 3 months and 8 months as dependent variables, resulting in 4 tests in each pregnancy period. A high–low contrast on the between- subjects factor represented the upper and lower quar- tile scores on the predictors. Dichotomous covariates (smoking and alcohol use, infants’ sex, breastfeeding) were entered as a between-subjects factor in the MAN- COVA. Only in case of a significant multivariate Ho- telling’s T2 test were univariate analyses performed subsequently to locate the source of the difference. The clinical relevance of prenatal predictors was explored in logistic regression analyses that attempted to differen- tiate mental and motor scores below the lower quartile from scores above the upper quartile. The associations between continuous predictor variables and the dicho- tomized dependent variables in the logistic regression models are reported as standardized odds ratios (SOR) and 95% confidence intervals (CI). The SOR represents the change in risk due to one standard deviation change in the independent variable. With all tests, statistical significance was assumed at the level of p < .05. Given the exploratory set-up of the study and the reasonable though not large sample size, the alpha level was not corrected for multiple comparisons. In this way, an adequate balance was established between the risk for type I and type II errors.

Results

Descriptive analyses

Two hundred and thirty women satisfied the inclu- sion and exclusion criteria and agreed to participate in the study. To check for selection bias, we ap- proached 70 non-participants of whom 69% (n ¼ 48) returned the questionnaire that collected data on the mental health status (trait-anxiety), appraisal of pregnancy, maternal age, educational level, em- ployment status, and health behavior during preg- nancy (smoking and drinking habits). Comparisons

812 Anja C. Huizink et al.

were made between a random sample of participants (n ¼ 52) and non-participants by means of t-tests and Chi-square tests. Participants smoked more than did non-participants (23% versus 7%, p < .01). On all other aspects, non-participants did not differ from participants. Of the 230 women who completed the questionnaires on the first occasion, 217 and 172 did so on the second and third occasion, re- spectively. The main reasons for the drop in the number of participants towards late pregnancy were delivery before 37 weeks of gestation, or delivery before the assessment in late pregnancy had taken place; other reasons were lack of interest, lack of time, stillbirth, pregnancy complications that re- quired intensive follow-up, or relocation to another city. Only healthy infants born near term (more than 37 completed weeks of gestation) were included in the follow-up study after birth, to remain free from confounding factors involved with prematurity or health problems of the infant. The total number of mother–infant dyads with complete datasets, inclu- ding an assessment of infant development at 3 and 8 months of age, was 170. Cortisol data of the third prenatal measurement was not complete, due to the fact that the last sample of cortisol was collected in the week after the last prenatal visit, which was planned near term. Other reasons for incomplete cortisol data throughout pregnancy were insufficient amounts of collected saliva. The sample of partici- pants consisted largely of Caucasian middle class women, although both lower and higher social clas- ses were represented (Table 1). The majority of

women (93.7%) lived together with their partner, ei- ther in wedlock or unmarried. We noted the following biomedical risks during pregnancy: pregnancy complications (n ¼ 30: 17.6%), use of medication during pregnancy (n ¼ 25: 14.7%), medication with risks for the fetus (n ¼ 4: 2.4%), fertility problems (n ¼ 48: 28.2%), in vitro fertilization (n ¼ 13: 7.6%), high blood pressure (n ¼ 15: 8.8%), gestational dia- betus mellitus (n ¼ 3: 1.8%), gynecological risk (n ¼ 12: 7.1%), and pre-existent disease (n ¼ 12: 7.1%). Since only 7 participants reported smoking more than 10 cigarettes per day, smoking was coded as a dichotomous variable: non-smokers, and smokers of 1 or more cigarettes per day. Further, since only 11 subjects consumed more than 2 alco- hol-containing beverages per week, a dichotomous variable was created: non-drinkers, and drinkers of 1 or more beverages/week. We observed the follow- ing perinatal complications: intrapartum complica- tions (n ¼ 25: 14.7%), use of medication during delivery (n ¼ 88: 51.8%), elective caesarean section (n ¼ 24: 14.1%), and artifical delivery due to fetal distress (n ¼ 20: 11.8%). Descriptives of the poten- tial covariates are summarized in Table 1 as well.

Descriptives of the scores on the predictor and dependent variables are presented in Table 2. The MDI scores at 3 and 8 months of age were signifi- cantly correlated (r ¼ .26, p < .0005), and so were the PDI scores at these ages (r ¼ .23, p < .0005). The MDI and PDI scores were highly correlated at both 3 months (r ¼ .52, p < .0005) and 8 months (r ¼ .38, p < .0005).

Table 1 Descriptives of potential covariates of the prenatal, perinatal and postnatal periods (N ¼ 170). Values are presented as proportions or as mean (SD) and range

Prenatal Maternal age, years 31.3 (4.9) Socioeconomic status: Educational level mother Low 13.6% Middle 67.5% High 18.9% Educational level partner Low 23.4% Middle 59.8% High 16.8% Professional level mother Low 8.0% Middle 54.6% High 37.4% Professional level partner Low 18.0% Middle 29.2% High 52.8% Paid job 87.4% (44.7% full-time, 55.3% part-time) Ethnic background Caucasian (96%) Smoking (cigarettes/day) Smokers (1 or more cig/day) n ¼ 29; Non-smokers: n ¼ 141 Alcohol-intake (beverages/week) Drinkers (1 or more beverage/week) n ¼ 26; Non-drinkers: n ¼ 144 Biomedical risks: Cumulative score 1.0 (1.2), range 0–5 Number of risks No risk: n ¼ 85, 1 or more risks, n ¼ 85 Perinatal Birth weight, grams 3386 (487) Gestational age at birth, weeks 39.6 (1.9) Sex 84 boys, 86 girls Perinatal complications: Cumulative score 1.0 (1.3), range 0–5 Postnatal Psychological well-being (GHQ-30) 4.7 (5.1), range 0–25 Perceived stress 25.9 (5.8), range 14–45 Postnatal depression (EPDS) 15.8 (4.6), range 10–32

*low level: primary school, high-school education; middle level: secondary school education; high level: college or academic education.

Prenatal stress and infant development 813

Psychological stress measures during pregnancy and infant mental and motor development

MANCOVAs were performed with a high–low contrast set on the predictor variable as between-subject factor, the MDI and PDI scores at 3 and 8 months as dependent variables, and gestational age at birth, birth weight and the postnatal stress and depression levels of the mother as covariates. An overall effect of daily hassles in early pregnancy on 8-month scores was found to show a trend towards significance, after correction for these covariates (F ¼ 2.31, p ¼ .10). No effect of daily hassles in mid- or late pregnancy was found on infant development. A significant overall effect was found of fear of giving birth in

mid-pregnancy on 8-month scores of infant devel- opment (F ¼ 5.04, p < .005). Subsequent univariate analyses (Table 3 and Figure 1) revealed that a high amount of daily hassles in early pregnancy was as- sociated with lower MDI scores at 8 months (F ¼ 3.9, p ¼ .05). Strong fear of giving birth in mid-pregnancy was associated with lower MDI and PDI scores at 8 months (F ¼ 5.58, p < .05 and F ¼ 7.67, p < .01, respectively). Strong fear of giving birth in late pregnancy was associated with lower MDI scores at 8 months (F ¼ 5.34, p < .05). No effect was found of fear of bearing a handicapped child on infant development.

Logistic regression showed that of the two different aspects of maternal stress (daily hassles, pregnancy anxieties), daily hassles in early pregnancy were an independent risk factor for low (i.e., £ P25) MDI scores of infants at 8 months of age (SOR ¼ 1.1, 95% CI 1.02–1.18). Logistic regression furthermore showed that high levels of fear of giving birth in mid- pregnancy increased the risk of having an infant with a low (i.e., £ P25) PDI score at 8 months of age (SOR ¼ 1.3, CI 1.12–1.56).

Endocrinologic stress measures during pregnancy and infant mental and motor development

A MANCOVA, with a high–low contrast on the 8 a.m. salivary cortisol level in late pregnancy as between- subject factor, the MDI and PDI scores at 3 and 8 months as dependent variables, and gestational age at birth, birth weight and the postnatal stress and depression level of the mother as covariates, showed a significant overall effect (F ¼ 4.61, p < .05). In univariate follow-up tests high cortisol was related to lower MDI scores at 3 months of age (F ¼ 6.38, p < .05) and lower PDI scores at both 3 and 8 months of age (F ¼ 9.15, p < .005; and F ¼ 9.38, p < .005) (see Table 3). Cortisol in early and mid-pregnancy did not show overall significant effects on infant development.

Table 2 Descriptives of the psychological and endocrinologic predictors

Predictors Mean SD Range N

Psychological Daily hassles T1 10.0 6.3 0–45 170 Daily hassles T2 7.8 5.5 0–26 170 Daily hassles T3 6.4 4.3 0–23 170 Fear of giving birth T1 6.2 2.9 3–15 170 Fear of giving birth T2 5.9 2.7 3–15 170 Fear of giving birth T3 6.0 2.7 3–15 170 Fear of handicapped child T1 9.3 3.5 4–20 170 Fear of handicapped child T2 8.6 3.1 4–19 170 Fear of handicapped child T3 8.5 3.2 4–20 170 Endocrinologic Cortisol 8 AM T1 19.8 7.4 6.0–44.2 142 Cortisol 8 AM T2 23.3 6.8 9.3–41.0 130 Cortisol 8 AM T3 23.6 6.3 2.5–43.0 85 Cortisol mean value T1 10.6 2.3 5.2–19.8 142 Cortisol mean value T2 14.4 3.1 6.3–22.3 130 Cortisol mean value T3 17.4 3.8 2.8–30.8 85 Dependent variables MDI 3 months 114.9 15.0 71–150 170 MDI 8 months 117.7 15.5 76–150 170 PDI 3 months 101.3 13.7 61–150 170 PDI 8 months 109.4 13.5 77–150 170

T1 ¼ 15–17 weeks of gestational age; T2 ¼ 27–28 weeks of gestational age; T3 ¼ 37–38 weeks of gestational age.

Table 3 Stress during pregnancy and mental and motor developmental scores at 3 and 8 months. Means and standard deviations (in brackets) are presented

MDI PDI

3 months 8 months 3 months 8 months

Low High Low High Low High Low High

Psychological measure of stress Early pregnancy Daily hassles 114 (15) 112 (13) 120 (15)* 113 (15) 103 (16) 98 (10) 110 (12) 109 (11) Mid pregnancy Fear of giving birth 115 (17) 112 (12) 122 (15)* 114 (11) 102 (17) 99 (8) 114 (12)** 106 (13) Late pregnancy Fear of giving birth 117 (17) 113 (13) 123 (17)* 116 (15) 101 (17) 101 (12) 114 (14) 108 (12) Endocrinologic measure of stress Late pregnancy Cortisol 8 AM 117 (13)* 107 (14) 117 (14) 112 (13) 103 (7)** 95 (9) 114 (16)* 102 (10)

*post-hoc univariate analyses p < .05; ** post-hoc univariate analyses p < .01.

814 Anja C. Huizink et al.

Discussion

Stress during pregnancy, as reflected by a high amount of daily hassles in early pregnancy or strong fear of giving birth in mid-pregnancy, was associated with an average decline of 8 points in mental and psychomotor developmental scores of the infant 8 months after birth (Table 3). The effects of preg- nancy anxiety remained significant after adjusting for possible confounders, such as SES, maternal age, birth weight, gestational age, biomedical risks during pregnancy, perinatal complications, and the mothers’ postnatal stress and depression levels. The effect on mental development at 8 months appeared to be non-linear since it was only found when high– low contrast groups were formed. In contrast, the negative effect of fear of giving birth on psychomotor development at 8 months proved to be linear. Corti- sol levels in saliva at 8 a.m. in late pregnancy as an endocrinologic index of stress was also linearly re- lated to psychomotor development at 3 and 8 months and mental development at 3 months. Overall, the negative effects of prenatal stress on developmental outcome were more clearcut at 8 than at 3 months. Relevant issues are greater measurement error of developmental outcome at 3 months due to less reliability and less variation in the scores.

At first sight, the effects of prenatal stress on in- fant development seem to be rather mild. The pre- sent study, however, probably provides an underestimate of the influence of prenatal stress on infant development, because the follow-up was lim- ited to healthy infants born near term, and stress effects that are mediated by an adverse birth out- come (Dunkel-Schetter, 1998; Copper et al., 1996; Wadhwa et al., 1993) were not taken into account in our study. Another consideration is that the adverse effects on infant development were found when studying the influence of commonly occurring and relatively minor stressors rather than that of

experimentally induced stress such as applied in animal designs or of circumscribed major life-events. Nonetheless, our results concur with evidence from studies in monkeys indicating that experimental prenatal stress induced neuromotor deficits in off- spring (Schneider et al., 1999; Schneider, 1992; Schneider et al., 1992). We note that although the MDI is classified as a ‘mental developmental index’, over half of the items contributing to the MDI be- tween the age of 3 and 8 months are motor or sen- sorimotor tasks. A recent large study showed effects of prenatal maternal anxiety on child behavior at 4 years (O’Connor et al., 2002). The effects found in the present study and in the few other prospective human studies (O’Connor et al., 2002; van den Bergh, 1990) are small, but appear to be consistent. The effects of prenatal stress and anxiety on the in- fant appear to be on various aspects of behavior and development, suggesting a diffuse effect. The clinical relevance of our findings is apparent from the in- creased risk of obtaining developmental scores below the lowest quartile, given high amounts of daily hassles and distress in early pregnancy and a strong fear of giving birth in mid-pregnancy. It is further important to note that Bayley test scores of mental development in infancy were found to correlate significantly with intelligence test scores in later childhood (Laucht, Esser, & Schmidt, 1994; Siegel, 1989).

We examined two relatively independent aspects of the emotional state of pregnant women as potential predictors of infant development. Pregnancy anxiety represented as fear of giving birth in mid-pregnancy emerged as the strongest predictor. Pregnancy anxiety reflects a unique element of human preg- nancy and was previously found to predict adverse pregnancy outcome (Killingsworth Rini et al., 1999). Since pregnancy anxiety is not quite comparable with the stressors applied in studies in rodents and non-human primates, replication should be sought

130

120

110

100

130

120

110

100

90

80

9 5 %

C l m

e n ta

l d e ve

lo p m

e n t

9 5 %

C l m

o to

r d e ve

lo p m

e n t

Fear of giving birth in mid pregnancy and mental development at 3 and 8 months

Cortisol at 8 am in late pregnancy and motor development at 3 and 8 months

N= 58 58 75 75 37 37 low mediate high

N= 24 24 38 38 22 22 low mediate high

= 3 months scores = 8 months scores

Figure 1 Effect of high fear of giving birth in mid-pregnancy on mental development (left) and of high cortisol at 8 a.m. in late pregnancy on motor development (right). The error bars reflect 95% confidence intervals

Prenatal stress and infant development 815

in future human studies. Van den Bergh (1990) showed that measures of general anxiety in the third trimester of pregnancy were unrelated to infant mental or motor development. Our measures of pregnancy anxiety, on the contrary, were related to infant mental and motor development. In another study, we found that pregnancy anxiety was only partly related to measures of general anxiety, and we suggested that pregnancy-related anxiety may be a rather distinctive syndrome (Huizink, 2000). There- fore, it might be of interest to focus on pregnancy anxiety as predictor of infant development in future human studies.

Our findings underline the importance of the study of relatively minor but stressful daily hassles in addition to pregnancy anxiety in the identification of pregnant women with psychological high-risk status. Thus far, the results of stress reduction programs in pregnancy are inconclusive (Villar et al., 1992). The present study suggests that a sharper focus on pregnancy anxieties and daily hassles may increase the effectiveness of intervention studies.

It remains difficult to establish during which period of pregnancy exposure to stress matters most in affecting the postnatal development of the infant. Our stress measures throughout pregnancy are correlated over time and thus are not independent. Measurement of effects of stress during a particular period in pregnancy on fetal behavior and physiology would provide more short-cut information about the timing issue of stress exposure. Studies in rhesus macaques reported sensitivity to prenatal stress to peak during early gestation and to taper off during later gestation (Schneider et al., 1999). Further, the California earthquake in 1994 was perceived as most stressful when it occurred early in pregnancy com- pared with late pregnancy (Glynn et al., 2001). In addition, stress experienced early in pregnancy was associated with shorter gestational length in this earthquake study (Glynn et al., 2001). We found the strongest effects on infant development of psycholo- gical measures of stress and cortisol in mid- to late pregnancy. O’Connor et al. (2002) found that anxiety at 32 weeks of gestation had the strongest effect on later child behavior, which is comparable to our results.

The relationship between psychological and endo- crinologic measures of stress in pregnancy may be complicated by the physiologic changes in neuro- endocrine function during pregnancy. The fetal- placental-decidual unit produces steroids and peptides hormones, among them corticotropin- releasing hormone (CRH). Maternal cortisol creates a positive feedback loop, by stimulating the synthesis and release of placental CRH which in turn further activates the maternal HPA-axis. As a result, over the course of pregnancy there is a progressive increase in maternal plasma levels of stress hormones, in- cluding CRH, ACTH and cortisol. This may have implications for the responsivity of the HPA-axis to

stress. Indeed, only in late pregnancy was the early morning value of cortisol correlated with psycholo- gical measures of stress.

The mechanisms that underlie the association between prenatal stress and infant development are unknown. There are a number of plausible hypo- theses. First, maternal stress may reduce uteropla- cental blood flow since cortisol and catecholamines are known to affect vessel tone (Teixeira, Fisk, & Glover, 1999). Reduced supply of oxygen and nu- trients to the fetus in turn mobilizes a response of the fetal HPA-axis that is operative from mid-pregnancy on. Second, maternal stress may lead to increased production of placental CRH that further activates the fetal HPA-axis (Majzoub & Karalis, 1999). Third, maternal cortisol may be directly transported across the placenta and enter the fetal circulation. The fetus is relatively protected from raised levels of maternal cortisol by the 11 ß-hydroxysteroid dehydrogenase (11 ß-HSD) enzyme in the placenta that metabolizes cortisol to inactive cortisone. In spite of this, how- ever, maternal cortisol has been found to account for about 40% of the variance in fetal concentrations of cortisol in high stress conditions (Gitau, Cameron, Fisk, & Glover, 1998).

A common element of these hypotheses is that the fetus is exposed to excess levels of cortisol. Studies in the rat indicate that the fetal brain is protected from glucocorticoids (in rat: corticosterone) by 11 ß-HSD which is highly expressed in all areas of the brain in mid-pregnancy (Diaz, Brown, & Seckl, 1998). The expression of 11-HSD, however, is dra- matically reduced in the last period of pregnancy, which allows glucocorticoids to interact with their receptor systems and influence brain development (Diaz et al., 1998). This would fit with our finding of the association between cortisol in late pregnancy and infant developmental outcome. Glucocorticoids are critical in promoting neuronal and glial matura- tional events under normal circumstances but are neurotoxic in high concentration (Uno et al., 1994). In particular, the hippocampus is highly vulnerable to excess levels of glucocorticoids that may lead to dose-dependent degeneration and depletion of hip- pocampal pyramidal neurons (Uno et al., 1994) and persistent reductions in hippocampal glucocorticoid receptor systems (Barbazanges et al., 1996; Maccari et al., 1995).

Animal studies provide extensive documentation that prenatal stress results in persisting alterations of the regulation of the HPA-axis and concomitant lifelong problems in behavioral adaptation (Wein- stock, 1997) and neurocognitive deficits (Schneider et al., 1999; Schneider & Coe, 1993). Early neuro- motor dysfunction in children has been associated with academic, cognitive, and behavioral problems at later ages (Gillberg & Gillberg, 1989). Decades ago, Bayley (1969) stated: ‘Motor abilities play im- portant roles in the development of the child’s ori- entation toward its environment, and they influence

816 Anja C. Huizink et al.

the quality of its interaction with the environment. Locomotion and control of the body serve to enlarge the potential sphere for new and varied experiences and for individual choices in seeking or avoiding different kinds of experience.’ Thus, the effects of prenatal stress on early motor and mental develop- ment may hamper the subsequent development of the child in various ways. We are currently involved in a further follow-up of our sample to examine whether the negative effects of prenatal stress on developmental outcome at 3 and 8 months are transient, persistent or even progressive, and to gain more insight into the neurobiological basis of these effects.

Acknowledgement

This research was supported by the Van der Gaag Stichting and the Praeventiefonds (28-2685).

Correspondence to

A.C. Huizink, Department of Social Medicine, Insti- tute for Research in Extramural Medicine, VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands; Email: [email protected]

References Barbazanges, A., Piazza, P.V., Le Moal, M., & Maccari, S. (1996). Maternal glucocorticoid secretion mediates longterm effects of prenatal stress. Journal of Neuroscience, 16, 3943–3949.

Bayley, N. (1969). Bayley Scales of Infant Development. New York: Psychological Corp.

Cohen, S., & Williamson, G.M. (1987). Perceived stress in a probability sample of the United States. In S. Spacapan & S. Oskamp (Eds.), The social psycho- logy of health (pp. 31–47). Newbury Park, California: SAGE Publications.

Copper, R.L., Goldenberg, R.L., Das, A., Elder, N., Swain, M., Norman, G., et al. (1996). The preterm prediction study: Maternal stress is associated with spontaneous preterm birth at less than thirtyfive weeks’ gestation. National Institute of Child Health and Human Development Maternal Fetal Medicine Units Network. American Journal of Obstetrics and Gynecology, 175, 1286–1292.

Cox, J.L., Holden, J.M., & Sagovsky, R. (1987). Detec- tion of postnatal depression: Development of the Edinburgh Postnatal Depression Scale. British Jour- nal of Psychiatry, 150, 782–786.

DeLongis, A., Coyne, J.C., Dakof, G., Folkman, S., & Lazarus, R.S. (1982). Relationship of daily hassles, uplifts, and major life events to health status. Health Psychology, 1, 119–136.

Diaz, R., Brown, R.W., & Seckl, J.R. (1998). Distinct ontogeny of glucocorticoid and mineralocorticoid receptor and 11beta-hydroxysteroid dehydrogenase

types I and II mRNAs in the fetal rat brain suggest a complex control of glucocorticoid actions. Journal of Neuroscience, 18, 2570–2580.

Dunkel-Schetter, C. (1998). Maternal stress and pre- term delivery. Prenatal and Neonatal Medicine, 3, 39–42.

Fride, E., Dan, Y., Feldon, J., Halevy, G., & Weinstock, M. (1986). Effects of prenatal stress on vulnerability to stress in prepubertal and adult rats. Physiological Behavior, 37, 681–687.

Gillberg, I.C., & Gillberg, C. (1998). Children with preschool minor neurological disorders IV: Behavior and school achievement at age 13. Developmental Medicine and Child Neurology, 31, 3–13.

Gitau, R., Cameron, A., Fisk, N.M., & Glover, V. (1998). Fetal exposure to maternal cortisol. The Lancet, 353, 707–708.

Glynn, L.M., Wadhwa, P.D., Dunkel-Schetter, C., Chicz-Demet, A., & Sandman, C.A. (2001). When stress happens matters: Effects of earthquake timing on stress responsivity in pregnancy. American Jour- nal of Obstetrics and Gynecology, 184, 637–642.

Hack, M., Breslau, N., Weissman, B., Aram, D., Klein, N., & Borawski, E. (1991). Effect of very low birth weight and subnormal head size on cognitive abilities at school age. New England Journal of Medicine, 325, 231–237.

Huizink, A.C. (2000). Prenatal stress and its effect on infant development. Unpublished PhD Thesis, Utrecht University, The Netherlands.

Killingsworth Rini C., Dunkel-Schetter, C., Wadhwa, P.D., & Sandman, C.A. (1999). Psychological adapta- tion and birth outcomes: The role of personal resources, stress, and sociocultural context in preg- nancy. Health Psychology, 18, 333–345.

Kirschbaum, C., & Hellhammer, D.H. (1989). Salivary cortisol in psychobiological research: An overview. Neuropsychobiology, 22, 150–169.

Koeter, M.W.J., & Ormel, J. (1991). General Health Questionnaire. Nederlandse bewerking [Dutch adap- tation]. Swets test services, The Netherlands.

Laucht, M., Esser, G., & Schmidt, M.H. (1994). Contrasting infant predictors of later cognitive func- tioning. Journal of Child Psychology and Psychiatry, 35, 649–662.

Lou, H.C., Hansen, D., Nordentoft, M., Pryds, O., Jensen, F., Nim, J., & Hemmingsen, R. (1994). Prenatal stressors of human life affect fetal brain development. Developmental Medicine and Child Neurology, 36, 826–832.

Maccari, S., Piazza, P.V., Kabbaj, M., Barbazanges, A., Simon, H., & Le Moal, M. (1995). Adoption reverses the longterm impairment in glucocorticoid feedback induced by prenatal stress. Journal of Neuroscience, 15, 110–116.

Majzoub, J.A., & Karalis, K.P. (1999). Placental cortico- tropin-releasing hormone: Function and regulation. American Journal of Obstetrics and Gynecology, 180, S242–246.

McCormick, C.M., Smythe, J.W., Sharma, S., & Meaney, M.J. (1995). Sex-specific effects of prenatal stress on hypothalamic-pituitary-adrenal responses to stress and brain glucocorticoid receptor density in adult rats. Brain Research and Developmental Brain Research, 84, 55–61.

Prenatal stress and infant development 817

McEwen, B.S., & Seeman, T. (1999). Protective and damaging effects of mediators of stress. Elaborating and testing the concepts of allostatis and allostatic load. Annals of the New York Academy of Science, 896, 30–47.

Meulenberg, P.M.M., & Hofman, J.A. (1990). The effect of oral contraceptive use and pregnancy on the daily rhythm of cortisol and cortisone. Clinica Chimica Acta, 190, 211–222.

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

Ounsted, O.H., Moar, V.A., & Stott, A. (1988). Head circumference and developmental ability at the age of seven years. Acta Paediatrica Scandinavia, 77, 374– 379.

Schneider, M.L., Roughton, E.C., Koehler, A.J., & Lubach, G.R. (1999). Growth and development fol- lowing prenatal stress exposure in primates: An examination of ontogenetic vulnerability. Child Devel- opment, 70, 263–274.

Schneider, M.L., & Coe, C.L. (1993). Repeated social stress during pregnancy impairs neuromotor devel- opment of the primate infant. Journal of Development and Behavorial Pediatrics, 14, 81–87.

Schneider, M.L. (1992). The effect of mild stress during pregnancy on birthweight and neuromotor matur- ation in rhesus monkey infants (Macaca mulatta). Infant Behavior and Development, 15, 389–403.

Schneider, M.L., Coe, C.L., & Lubach, G.R. (1992). Endocrine activation mimics the adverse effects of prenatal stress on the neuromotor development of the infant primate. Developmental Psychobiology, 25, 427–439.

Siegel, L.S. (1989). A reconceptualisation of prediction form infant test scores. In M.H. Bornstein & N.A. Krashnegor (Eds.), Stability and continuity in mental development: Behavioral and biological perspectives (pp. 89–103). Hillsdale, NJ: Erlbaum.

Teixeira, J.M.A., Fisk, N.M., & Glover, V. (1999). Association between maternal anxiety in pregnancy and increased uterine artery resistance index: Cohort based study. British Medical Journal, 318, 153–157.

Uno, H., Eisele, S., Sakai, A., Shelton, S., Baker, E., DeJesus, O., & Holden, J. (1994). Neurotoxicity of glucocorticoids in the primate brain. Hormones and Behavior, 28, 336–348.

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

Van Eck, M.M.M., Berkhof, H., Nicolson, N.A., & Sulon, J. (1996). The effects of perceived stress, traits, mood states, and stressful daily events on salivary cortisol. Psychosomatic Medicine, 158, 447–458.

Van der Meulen, B.F., & Smrkovky, M. (1983). BOS 2-30. Bayley ontwikkelingsschalen: Handleiding [Bayley Developmental series in Dutch: A manual]. Lisse, The Netherlands: Swets and Zeitlinger B.V.

Van Westerlaak, J.M., Kropman, J.A., & Collaris, J.W.N. (1976). Beroepenklapper [List of occupations: in Dutch]. Nijmegen, the Netherlands: Instituut voor Toegepaste Sociologie.

Villar, J., Farnot, U., Barros, F., Victoria, C., Langer, A., & Belizan, J.M. (1992). A randomized trial of psychosocial support during high-risk preg- nancies. The New England Journal of Medicine, 327, 1266–1271.

Vingerhoets, A.J.J.M., Jeninga, A.J., & Menges, L.J. (1989). Het meten van chronische en alledaagse stressoren: Eerste onderzoekservaringen met de Alle- daagse Problemen Lijst (APL) II [Measuring chronic and everyday stressors: First research findings with the Everyday Problem List II: in Dutch]. Gedrag en Gezondheid, 17, 10–17.

Wadhwa, P.D., Sandman, C.A., Porto, M., DunkelS- chetter, C., & Garite, T.J. (1993). The association between prenatal stress and infant birth weight and gestational age at birth: A prospective investigation. American Journal of Obstetrics and Gynecology, 169, 858–865.

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

Weinstock, M., Matlina, E., Maor, G.I., Rosen, H., & McEwen, B.S. (1992). Prenatal stress selectively alters the reactivity of the hypothalmaic-pituitary adrenal system in the female rat. Brain Research, 595, 195–200.

Manuscript accepted 14 August 2002

Appendix A: Pregnancy anxiety

Fear of giving birth:

I am worried about the pain of contractions and the pain during delivery. I am anxious about the delivery because I have never experienced one before. I am worried about not being able to control myself during labor and fear that I will scream.

Fear of bearing a physically or mentally handicapped child:

I am afraid the baby will be mentally handicapped or will suffer from brain damage. I am afraid our baby will be stillborn, or will die during or immediately after delivery. I am afraid that our baby will suffer from a physical defect or worry that something will be physically wrong with the baby. I sometimes think that our child will be in poor health or will be prone to illnesses.

818 Anja C. Huizink et al.

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.

Users should refer to the original published version of the material.