University of Calgary PRISM: University of Calgary's Digital Repository
Arts Arts Research & Publications
2019-05 A Developmental Cascade from Prenatal Stress to Child Internalizing and Externalizing Problems
Hentges, Rochelle F.; Graham, Susan; Plamondon, André R.; Tough, Suzanne C.; Madigan, Sheri L.
Oxford University Press on behalf of the Society for Pediatric Psychology : Journal of Pediatric Psychology
Hentges, R. F., Graham, S. A., Plamondon, A. R., Tough, S. C., & Madigan, S. L. (2019). A Developmental Cascade from Prenatal Stress to Child Internalizing and Externalizing Problems. "Journal of Pediatric Psychology", 2019, 1-11. http://dx.doi.org/10.1093/jpepsy/jsz044 http://hdl.handle.net/1880/111910 journal article https://creativecommons.org/licenses/by/4.0 Unless otherwise indicated, this material is protected by copyright and has been made available with authorization from the copyright owner. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca Journal of Pediatric Psychology, 2019, 1–11 doi: 10.1093/jpepsy/jsz044 Original Research Article Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019 A Developmental Cascade from Prenatal Stress to Child Internalizing and Externalizing Problems
1,2 1,2 Rochelle F. Hentges, PHD, Susan A. Graham, PHD, 3,4 2,5,6 Andre Plamondon, PHD, Suzanne Tough, PHD, and 1,2 Sheri Madigan PHD
1Department of Psychology, University of Calgary, 2Owerko Centre, Alberta Children’s Hospital Research Institute, 3Department des fondements et pratiques en education, Universite Laval, 4Applied Psychology and Human Development, University of Toronto, 5Department of Community Health Sciences, University of Calgary and, 6Department of Pediatrics, University of Calgary All correspondence concerning this article should be addressed to Rochelle Hentges, PhD, Owerko Centre, Third Floor—CDC Building, c/o 2500 University Dr. N.W., Calgary, Alberta T2N 1N4, Canada. E-mail: [email protected] Received October 31, 2018; revisions received May 7, 2019; accepted May 10, 2019
Abstract Objective: This study utilized a developmental cascade approach to test alternative theories about the underlying mechanisms behind the association of maternal prenatal stress and child psychopa- thology. The fetal programming hypothesis suggests that prenatal stress affects fetal structural and physiological systems responsible for individual differences in child temperament, which fur- ther increases risk for internalizing and externalizing problems. Interpersonal models of stress transmission suggest that maternal stress influences child mental health via early parenting behav- iors. We also examined a continuation of stress hypothesis, in which prenatal stress predicts child mental health via the continuation of maternal stress in the postpartum period. Methods: Participants were 1,992 mother–child pairs drawn from a prospective pregnancy cohort. Mothers reported on their perceived stress, anxiety, and depression during pregnancy and at 4-month post- partum. Birthweight was assessed via medical records of birthweight. At 4-month postpartum, hostile-reactive parenting behaviors were assessed. Child temperamental negative affect was mea- sured at age 3. Child internalizing and externalizing problems were assessed at age 5. Results: Prenatal stress was associated with both internalizing and externalizing problems via postnatal stress and child temperament. Prenatal stress was also associated with externalizing behaviors via increased hostile-reactive parenting. After accounting for postnatal factors, prenatal stress contin- ued to have a direct effect on child internalizing, but not externalizing, symptoms. Conclusion: Results provide support for the fetal programming, interpersonal stress transmission, and continu- ation of stress models. Findings highlight the need for prenatal preventative programs that continue into the early postnatal period, targeting maternal stress and parenting behaviors.
Key words: developmental psychopathology; fetal programming; mental health; prenatal stress.
Introduction role in promoting children’s health and well-being. A robust literature within developmental science sug- However, recent advancements in research have sug- gests that the early social environment plays a critical gested that the developmental origins of health and
VC The Author(s) 2019. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: [email protected] 1 2 Hentges, Graham, Plamondon, Tough, and Madigan disease (DOHaD) can be traced back to events that oc- development through its spillover onto early parent– cur even before birth (Gillman, 2005). For example, infant interactions and relations (Hammen, Shih, & when rodents are exposed to stressors (e.g., electric Brennan, 2004). For example, Hammen et al. (2004) shocks, forced swim tests) during pregnancy, their off- found that the association between maternal depres-
spring exhibit maladjusted behaviors, including anxi- sion and child depression at age 15 years was entirely Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019 ety and depressive behaviors, heightened emotional mediated by family and interpersonal stress effects and physiological stress reactivity, and aggression (e.g., parenting). However, much of this research has (Abe et al., 2007; Eaton, Edmonds, Henry, Snellgrove, been conducted with mothers experiencing postnatal & Sloman, 2015). These findings have been replicated depression or anxiety, thus it’s unclear whether the in observational research with humans. For example, same mechanisms underlie the link between prenatal maternal perceived stress, anxiety, and depression dur- stress and child mental health. To our knowledge, ing the prenatal period have been associated with in- only two studies have examined how prenatal stress creased risk of attention-deficit hyperactivity disorder and maternal mental health impact postnatal parent- (ADHD), depression, and behavior problems ing, with mixed results. Belsky, Ruttle, Boyce, (Madigan et al., 2018; Rodriguez & Bohlin, 2005), Armstrong, & Essex (2015) found that prenatal ma- with some indications that prenatal stress might even ternal stress, defined as a composite of depression, increase the risk for more severe psychotic disorders, marital conflict, and financial stress, did not predict such as schizophrenia (Van Os & Selten, 1998). negative parenting during infancy, whereas Lereya and Wolke (2013) found that prenatal anxiety and de- A Search for Intermediary Factors: Two Proposed pression predicted higher maternal hostile and puni- Models tive parenting during preschool. Parents who are A critical next step is to move beyond cataloguing unable to control their own negative emotional reac- these associations and toward identifying intermediary tions are further proposed to engender child problems mechanisms that explain why and how prenatal stress in regulating their own behaviors (i.e., externalizing influences later child mental health. In examining the problems) and emotions (i.e., internalizing problems; underlying mechanisms, researchers often cite the fetal Grusec & Davidov, 2010). programming hypothesis, which postulates that expo- A notable limitation of the current body of research sure to events or stressors during gestation (e.g., is that most studies tend to examine the association heightened maternal anxiety) “programmes persisting between prenatal stress on either neonatal physiology changes in a range of metabolic, physiological, and and health outcomes (e.g., birth weight), or child tem- structural parameters,” which impact fetal neural de- perament, or child mental health outcomes. velopment and regulatory capabilities (Barker, 1995, Accordingly, there is a significant lack of longitudinal p. 171; O’Donnell, O’Connor, & Glover, 2009). One research that examines how these complementary pro- such “structural parameter” is fetal growth and size. cesses unfold over time. Through a developmental cas- Accordingly, prenatal stress has been associated with cade approach, the primary aim of this article is to a higher risk of prematurity and low birthweight examine the potential downstream consequences of (Bussie`res et al., 2015; Grote et al., 2010). In addition, prenatal stress on child mental health through a longi- research has suggested that prenatal stress alters the tudinal series of proposed underlying mechanisms. An physiological structure of the growing fetus, thereby advantage of a developmental cascade model is that it increases the likelihood of infants and toddlers show- allows for the examination of how early predictors cu- ing greater temperamental negative emotional reactiv- mulatively influence a developmental outcome (e.g., ity (Davis et al., 2004; Madigan et al., 2018), defined child mental health) over time through multiple pro- as early emerging, biologically based individual differ- cesses and systems (Masten & Cicchetti, 2010). Not ences in negative affect, such as anger, fear, and sad- only is this a more holistic approach to examining the ness (Rothbart, 2011). Notably, low birthweight has developmental evolution of a child exposed to prena- been tied to increased risk of both temperamental neg- tal stress, but it also allows for the identification of ative affect (Baibazarova et al., 2013) and the two multiple developmentally salient targets for interven- broadband dimensions of psychopathology, namely tion. Here, we simultaneously modeled proposed fetal internalizing and externalizing behaviors programming (i.e., birthweight, child temperament) (Aarnoudse-Moens, Weisglas-Kuperus, van and interpersonal transmission effects (i.e., hostile par- Goudoever, & Oosterlaan, 2009). enting). To our knowledge, these alternative hypothe- While much of the recent research on prenatal ses have not been examined in conjunction. stress has operated from the perspective of the fetal programming hypothesis, interpersonal models of Cascades of Stress stress transmission offer the alternative hypothesis In addition, while recent studies often control for post- that maternal stress and depression impact child partum stress (e.g., Neuenschwander et al., 2018), few Prenatal Stress Developmental Cascade 3
< 25 weeks Gestation Birth 4-months 3 years 5 years
Hostile-Reactive b Parenting b Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019
a a Temperamental a Prenatal Stress Birthweight Child Mental Health Negative Affect
c c
Postnatal Stress
Figure 1. Conceptual model of developmental cascade from prenatal stress to child mental health. Pathway (a) denotes pro- posed fetal programming processes; Pathway (b) represents the hypothesized processes from the perspective of interper- sonal models of stress transmission; and Pathway (c) denotes the effect of the continuation of stress into the postpartum period. studies assess the impact of the continuation of mater- that prenatal stress would predict poorer child mental nal stress or mental health concerns into the postpar- health through a set of biologically informed pro- tum period on child socioemotional development and cesses, including lower birth weight and toddler tem- mental health (Madigan et al., 2018). When studies in- peramental negative affect. Next, we hypothesized clude postpartum stress, they generally do so as a sta- that prenatal stress would predict poorer mental tistical covariate or by examining the relative health through its association with higher hostile- predictive roles of prenatal and postnatal stress, inde- reactive parenting during infancy. Finally, we hypoth- pendent of each other (e.g., Neuenschwander et al., esized that prenatal stress would be associated with 2018; O’Connor, Heron, Golding, Glover, & later child mental health problems through the contin- ALSPAC Study Team, 2003). Further, researchers fo- uation of maternal stress during the postpartum pe- cusing on postnatal stress are rarely able to control for riod. Given recent research suggesting that prenatal prenatal stress. Thus, it is unclear whether the effects stress might operate differently across sexes of postnatal stress can be accounted for by the contin- (Braithwaite et al., 2017; Plamondon et al., 2015), we uation of stress from the prenatal period. As a notable also examined whether the developmental cascade exception, Barker, Jaffee, Uher, & Maughan(2011) model differed as a function of child sex. examined the direct and indirect effects of prenatal de- pression and anxiety on child internalizing and exter- nalizing through its association with postpartum Methods maternal anxiety and depression, finding that both de- velopmental periods were important for subsequent Sample child mental health. However, this study did not in- Participants were drawn from an epidemiological pro- clude other potential postnatal mediating mechanisms, spective pregnancy cohort in a large city in western such as parenting and child negative affect. Canada. Pregnant women were recruited through Simultaneously modeling prenatal stress and postnatal community advertising and at local primary health- stress in a developmental cascade model will allow for care offices and the local laboratory services. Inclusion greater understanding of the timing of effects between criteria included being at least 18 years of age, ability maternal stress and child mental health, which is criti- to communicate in English, a current gestational age cal when developing targeted preventive interventions. of less than 25 weeks, and receiving prenatal care. Women with known multiple pregnancies were ex- cluded. Between August 2008 and July 2011, 3,387 The Current Study women were enrolled in the study. The original study The current study examines the intermediary processes was meant to end after 4-month postpartum, but sub- proposed by different theoretical perspectives in one sequent funding allowed for additional follow-up developmental cascade model (Figure 1), using an epi- assessments. Due to delays in funding acquisition, demiological pregnancy cohort that spans from the some participants had aged out of eligibility at these prenatal period to age five. Specifically, we hypothe- follow-up timepoints, resulting in smaller sample sizes sized that prenatal maternal stress would predict child (see Figure 2 for a flow chart of participant retention). mental health (i.e., internalizing and externalizing) Only participants with data available at age five through three unique pathways. First, we predicted (N ¼ 1992) were included in the current study. 4 Hentges, Graham, Plamondon, Tough, and Madigan
All Our Families (0.5%), or mixed/other (4.1%). Participants were gen- par cipants enrolled erally from middle- to high-socioeconomic back- n=3387 grounds, with 79.6% of participants having a university, college, or trade school degree and 80.8%
with an annual family income level of $70,000 Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019 Completed T1: Canadian dollars or more per year (Tough et al., <25 weeks pregnant Withdrawn 2017). All procedures were approved by the n=3362 (99%) par cipants a University of Calgary Research Ethics Board. n=807 (24%)
Pregnancy Completed T2: Measures loss/child death 4 months postpartum n=3057 (90%) Maternal Stress n=34 (1%) During pregnancy and at 4-month postpartum, moth- Ac ve ers completed the Perceived Stress Scale (PSS; Cohen, withdrawal Kamarck, & Mermelstein, 1983), Edinburgh n=339 (10%) Agreed to be contacted for Postnatal Depression Scale (EDPS; Cox, Holden & Passive follow-up research Sagovsky, 1987), and the Spielberger State Anxiety withdrawal n=3223 (95%) Inventory (SAI; Spielberger, Gorsuch, Lushene, Vagg, n=434 (13%) & Jacobs, 1970). Internal reliabilities were all satisfac- Completed T3: b tory (as range from .80 to .93). Consistent with previ- n=2909 eligible 3 years post birth n=1994 (69%) ous literature conceptualizing perceived stress, anxiety, and depression as indices of prenatal stress (e.g., Belsky et al., 2015; Madigan et al., 2018; Completed T4: n=2818 eligible 5 years post birth Maxwell et al., 2018), these three scales were used as n=1992 (71%) indicators of latent variables of maternal stress during the prenatal and early postnatal periods. Figure 2. Participant flowchart from recruitment through a age five of follow-up. Participants who completed at least Birthweight one prenatal questionnaire but did not complete any fol- low-up surveys, up to 5 years. bSample size at follow-up Maternal pregnancy medical records were available time points decreased due to timing of questionnaire de- for 1,760 of the 1,992 participants and provided infor- velopment and implementation, ethics approval, funding mation regarding infant birthweight (M ¼ 3353.33, constraints, and attrition. Only participants who were still SD ¼ 516.49, range: 595–5071 g) at the time of eligible, based on their child’s age, were mailed follow-up delivery. questionnaires. Adapted and reprinted with permission from Tough et al. (2017). Hostile-Reactive Parenting About 24% (n ¼ 807) of women were considered At the 4-month postpartum assessment period, moth- withdrawn from the study, defined as not completing ers responded to two items (“I have been angry with any assessments after the pregnancy period. my baby when he/she was particularly fussy” and “I Approximately 13% of participants had passively have raised my voice with or shouted at my baby withdrawn from the study, due to geographical moves, when he/she was particularly fussy”) from the lost to follow-up, or unknown reasons. Common rea- Parental Cognitions and Conduct Toward the Infant sons for active withdrawal (10%) included loss of in- Scale (PACOTIS; Boivin et al., 2005) hostile-reactive terest, lack of time, and lack of partner support. A subscale. Item responses were on a scale from 0 (not at small proportion of women (1%) withdrew due to all what I did)to10(exactly what I did). The two pregnancy loss or child death. Participants who were items (r ¼ .77) were averaged together to create a vari- retained at age five were more likely to have lower pa- able of hostile-reactive parenting. rental stress symptoms, have higher educational at- tainment and family incomes, be older, have fewer Child Negative Affect antepartum medical risks, and have infants with lower When children were three years old, mothers com- birthweights. However, differences tended to be small pleted the 12-item Negative Affect subscale of the in magnitude (see Supplementary Table S1 for mean Child Behavior Questionnaire-Very Short Form comparisons). (CBQ-VSF; Putnam & Rothbart, 2006). This scale At the first wave of data collection, the average age assesses the degree to which the child shows high tem- of mothers was 30.87 (SD ¼ 4.36) years. Mothers peramental negative affect, including anger, fear, sad- were predominantly white (81.9%), with smaller per- ness, and general upset. Item responses were on a scale centages of women identifying as Asian (10.7%), from 0 (extremely untrue)to7(extremely true). Latin American (1.6%), black (1.2%), indigenous Internal consistency was adequate (a ¼ .69). Prenatal Stress Developmental Cascade 5
Child Mental Health (Little, 1988). Results of Little’s MCAR test suggested When children were five years old, mothers completed that data were not MCAR, v2 (967) ¼ 1,181.39, p < the Behavior Assessment System for Children-Second .001. Unlike MCAR, there is no way to rigorously test Edition (BASC-2), a widely used assessment of child whether the data are missing at random (MAR) or not psychological problems with established reliability missing at random (NMAR). However, including Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019 and validity (Kamphaus & Reynolds, 2007). The cur- covariates thought to be related to missingness or val- rent study focused on the Externalizing (a ¼ .86) and ues on the outcomes (e.g., maternal education, family Internalizing (a ¼ .88) scales of the BASC. All analyses income, maternal stress/depression) can make the were conducted using the combined-sex norm-refer- MAR assumption more likely, thus limiting the poten- enced T scores. tial bias occurring if the data were NMAR (Collins, Schafer, & Kam, 2001; Graham, 2009). Therefore, Covariates missing data were estimated using full information We controlled for child sex (male¼ 1; female ¼ 2), maximum likelihood (FIML), which is robust to con- maternal education, and family income in all analyses. ditions of MAR (Enders & Bandalos, 2001). FIML Maternal education was reported on a scale of 1 (did procedures do not impute scores for missing data but not complete high school) to 6 (completed graduate rather utilize the raw data to establish parameter esti- school), while income was reported in increments of mates (Enders & Bandalos, 2001). The model was es- $10,000 (1 ¼ $10,000; 11 ¼ $100,000). We also timated in MPlus using the maximum likelihood with controlled for antepartum health risk with the inclu- robust standard errors estimator (MLR). sion of an Antepartum Risk Score (APRS; Alberta Perinatal Health Program, 2009) from the pregnancy medical records. In Alberta, the APRS is completed at Results every birth by the physician or midwife caring for the Descriptive statistics for all study variables are in- mother at the time of delivery. The APRS assigns risk cluded in Table I. Results of the developmental cas- scores to mothers based on a number of pre-existing cade model are shown in Figure 3. Only significant risk factors (e.g., obesity, maternal age <18 or 35 pathways are shown. Note that previously described years), past pregnancy history (e.g., stillbirths), current covariates were included as predictors of all longitudi- pregnancy complications (e.g., gestational diabetes or nal pathways but are not shown for clarity of presen- hypertension), and other risk factors (e.g., substance tation. Consistent with best practices, covariances use). A weighted value is assigned to each condition were also estimated between identical measurements based on the severity of the condition, and all values across time but are not shown in the figure (Marsh, are then summed to create a total risk score. Based on Morin, Parker, & Kaur, 2014). Fit indices suggested prior validation (Parboosingh, 1986), scores of 0 to 2 that the model provided a good representation of the are considered low risk, scores of 3–6 indicate moder- data (v2 [47, N ¼ 1,933] ¼ 117.23, root mean square ate risk, and scores of 7 or more reflect high risk. of error of approximation (RMSEA) ¼ .03, compara- tive fit index (CFI) ¼ 0.99, Tucker-Lewis Index (TLI) Analytic Plan ¼ 0.98). To examine the direct and indirect associations between prenatal stress and child internalizing and externalizing Direct Effects problems at age five through the set of hypothesized As shown in Figure 3, prenatal maternal stress had a mediating mechanisms, we conducted a structural equa- direct effect on child internalizing at age five, b ¼ .16, tion model (SEM) in MPlus 8.0 (Muth en & Muth en, p < .001, but not externalizing symptoms, b ¼ .05, 1998–2017). Specifically, we examined the longitudinal p ¼ .15. In addition, prenatal stress negatively pre- relations between: (a) prenatal maternal stress at dicted birthweight, b ¼ .07, p ¼ .008, and positively <25 weeks gestation; (b) birthweight; (c) maternal post- predicted both postpartum stress, b ¼ .60, p < .001, natal stress and hostile-reactive parenting at 4 months; and hostile-reactive parenting at four months, b ¼ .26, (d) child negative affect at 3 years; and (e) child inter- p < .001, and child negative affect at three years, b ¼ nalizing and externalizing symptoms at age five.All .11, p ¼ .002. Hostile-reactive parenting, in turn, pre- possible predictive pathways between exogenous and dicted higher child externalizing problems at age five, endogenous variables were estimated, as were covarian- b ¼ .12, p ¼ .001. Postnatal stress positively predicted ces between predictors assessed at the same time point child negative affect at three years, b ¼ .14, p ¼ .001, (e.g., internalizing and externalizing at 5 years). and both child internalizing b ¼ .09, p ¼ .01, and ex- ternalizing, b ¼ .11, p ¼ .001, symptoms at age five. Missing Data Birthweight did not predict subsequent parent (i.e., To evaluate whether data were missing completely at parenting, stress) or child functioning (i.e., tempera- random (MCAR), we conducted Little’s MCAR test ment, internalizing, externalizing). Finally, higher 6 Hentges, Graham, Plamondon, Tough, and Madigan
¼ temperamental negative affect at age three predicted both higher internalizing, b ¼ .32, p < .001, and ex- 3 years; b < ¼ ternalizing, ¼ .24, p .001, symptoms at age five.
Indirect Effects Downloaded from https://academic.oup.com/jpepsy/advance-article-abstract/doi/10.1093/jpepsy/jsz044/5511639 by University of Calgary user on 13 June 2019 Next, we examined the indirect effects of prenatal stress on child internalizing and externalizing symp- toms through its association with birthweight, parent- ing, postpartum stress, and child negative affect. Indirect effects were estimated using the bootstrapping 4 months postpartum; T3
¼ procedure in MPlus 8.0. Birthweight did not predict subsequent child or parental functioning and were therefore not included in the indirect effect estimates. Results for all standardized estimated indirect .04 .03 .77* .05* .74* .77* .05 .34* .32* .27* effects are presented in Table II. There were four sig- Perceived Stress Scale (possible range: 0–40); EDPS nificant indirect effects from prenatal maternal stress ¼ to child externalizing symptoms. First, there were indi- 25 weeks gestation; T2 .10*
< rect effects from prenatal stress to child externalizing
¼ through hostile-reactive parenting, b ¼ .03, p ¼ .002, postnatal maternal stress, b ¼ .07, p ¼ .001, and child .08* temperamental negative affect, b ¼ .03, p ¼ .003. In addition, the indirect pathway between prenatal stress, postnatal stress, child negative affect, and child exter- .06*