SEARCH

SEARCH BY CITATION

Keywords:

  • pregnancy;
  • childhood;
  • psychosocial stress;
  • life course;
  • prospective cohort study

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

Objective

To evaluate the impact of childhood stressors, recalled childhood stress, and stressors in adulthood on perceived stress in pregnancy.

Design

Prospective cohort study.

Setting

Pregnant women were recruited from pre-birth clinics in two communities.

Participants

Four-hundred and twenty-one pregnant women.

Results

Perceived prenatal maternal stress was the main outcome measure. Recalled childhood stress was positively associated with prenatal stress (β = .54) after adjusting for other child and adult factors. Low family cohesion during childhood was indirectly associated with prenatal stress through its effects on recalled childhood stress, current family cohesion, and current subjective socioeconomic position (SEP). Low levels of adult subjective SEP (β = −.44) and family cohesion (β = −.25) were directly associated with prenatal stress. Perceived social support during childhood was indirectly associated with prenatal stress through its effect on recalled childhood stress and perceived social support in adulthood. Childhood subjective SEP indirectly influenced prenatal stress through its effect on adult subjective SEP and recalled childhood stress.

Conclusions

Prenatal stress is a result of the interplay between factors from childhood and adulthood. The study findings can be used to inform psychosocial risk assessment and interventions across the lifespan to decrease prenatal stress and its adverse outcomes.

Psychological distress is one of the most common morbidities experienced by women during pregnancy (Roy-Matton, Moutquin, Brown, Carrier, & Bell, 2011), with prenatal maternal stress representing a prevalent form of distress (Matthey et al., 2004). Rates of prenatal stress range from 13% to 36% (Kingston, Heaman, Fell, Dzakpasu, & Chalmers, 2010; Whitehead, Brogan, Blackmore-Prince, & Hill, 2003; Woods, Melville, Guo, Fan, & Gavin, 2010) and are comparable to those of other pregnancy-related morbidities (e.g., maternal obesity, gestational diabetes, and gestational hypertension) (Roy-Matton et al.). Prenatal stress has detrimental short- and long-term consequences that extend far beyond the immediate health of the mother and infant. It has been associated with maternal anxiety and depression in both the prenatal and postpartum periods (Beck, 2001; Brummelte & Galea, 2009), risky health behaviors (Ahluwalia, Mack, & Mokdad, 2004), pregnancy complications (Woods et al., 2010), and inadequate prenatal care (Heaman, Gupton, & Moffatt, 2005).

Prenatal stress has been implicated in the delivery of preterm and small for gestational age infants (Beydoun & Saftlas, 2008), and a substantial body of evidence indicates that prenatal stress has long-term negative effects of on infant, child, and adolescent behavioral, cognitive, psychomotor, and socio-emotional development and school performance (Beydoun & Saftlas; Kingston, 2011). Other studies exploring child outcomes have concluded that independent of postnatal effects, prenatal stress has an impact on chronic disease in adulthood, which is supportive of the fetal programming hypothesis (Kajantie, 2006).

Compared to the volume of research exploring the impact of prenatal stress on adverse maternal and child outcomes, relatively few studies have assessed factors that contribute to prenatal stress, limiting our ability to identify women at risk and to provide effective and timely interventions. This research focused almost exclusively on risk factors during pregnancy (Tiedje, 2003). Researchers generally have not considered women's psychosocial health as a continuum from pre-conception to post-partum, despite knowledge that “some of the most powerful influences on pregnancy outcomes are related to influences on women's health that occur long before pregnancy begins” (Misra, Guyer, & Allston, 2003, p. 65).

Few studies have been conducted on predictors of prenatal maternal stress. Researchers generally have not considered women's psychosocial health as a continuum from pre-conception to postpartum.

Additionally, the prioritization of healthy child development has refocused attention on the lifelong impact of childhood adversity on adult health outcomes, including mental health. Indeed, recent studies have demonstrated the long-term effects of common forms of adversity such as family dysfunction and socioeconomic disadvantage on adult health. Some of the most striking findings of this body of research are that common childhood stressors have enduring effects on adult mental health (Hazel, Hammen, Brennan, & Najman, 2008; Seedat et al., 2009), are linked to neuroendocrine changes (Carpenter et al., 2004; Elzinga et al., 2008; Gonzalez, Jenkins, Steiner, & Fleming, 2009; Shea et al., 2007), and their influences do not appear to attenuate with time (Clark, Caldwell, Power, & Stansfeld, 2010). Based on a systematic review of markers and mechanisms leading to child and adolescent psychopathology, Grant et al. (2003) concluded that, “Stressors remain central to current etiological theories of child and adolescent psychopathology” (p. 451).

A report by the World Health Organization Commission of Social Determinants described the necessity of promoting healthy foundations in early life by addressing maternal health, stating that “Implementing a more comprehensive approach to early life includes…comprehensive support to and care of mother before, during, and after pregnancy – including interventions that help to address prenatal and postnatal maternal mental health problems” (Commission on Social Determinants of Health, 2008, p. 53). This recommendation is consistent with the conclusions of a review of evidence examining the role of prenatal influences on child neurodevelopment and the effect and timing of early childhood intervention (Doyle, Harmon, Heckman, & Tremblay, 2009). The authors found that the earlier the childhood developmental intervention was introduced (e.g., 0 to 3 years), the larger and more enduring the effects. Doyle et al. concluded with a compelling argument for the antenatal investment hypothesis, which suggests that the greatest impact on child development results from prevention efforts that begin during the first trimester of pregnancy. Together, these findings imply that the most favorable time to address prenatal stress is before conception or during early pregnancy.

It is important to understand the pathways across a woman's life that contribute to prenatal stress as this could inform the components of a comprehensive psychosocial assessment and guide the provision of effective care. The purpose of this study was to test and refine the Life Course Prenatal Maternal Stress Model in order to understand how common stressors during childhood and adulthood influence prenatal stress.

The Life Course Prenatal Maternal Stress Model

Overview of Model Structure. The Life Course Prenatal Maternal Stress Model (Fig. 1) was developed from a systematic review of 55 studies that examined factors influencing stress during childhood or the prenatal period, or that evaluated the relationship between stress in childhood and adulthood (unpublished). Articles that described potential mechanisms by which childhood stress influenced stress in adulthood were used to guide the development of model pathways. From this review, the most common stressors were determined to be family-related and socioeconomic. The model describes relationships between stressors related to socioeconomic position (SEP) and family environment in childhood and pregnancy, and perceived prenatal stress, mediated by perceived stress in childhood and social support in childhood and adulthood. The model integrates psychological, social, and demographic factors across the life course through a number of direct and indirect pathways.

image

Figure 1. Proposed Life Course Prenatal Maternal Stress Model. In diagrams of structural equation models, ovals represent latent variables (e.g., stress in pregnancy) and rectangles depict indicator variables (e.g., adult social support). Latent variables represent complex constructs that are measured using a variety of instruments (Table 1). Indicator variables are typically measured by one instrument, item, or index (Table 1). SEP = socioeconomic position.

Download figure to PowerPoint

Theoretical Framework. The model is based on Lazarus and Folkman's (1984) transactional theory, which defines perceived stress as a process involving “a relationship between the person and the environment that is appraised by the person as taxing or exceeding his or her resources and endangering his or her well-being” (p. 21). The foundational tenet of this theory is that the impact of a stressor is dependent upon how it is perceived by an individual. Based on evidence that objective SEP (e.g., income, education) does not fully explain the relationship between SEP and ill health and that subjective SEP (e.g., an individual's perception of his/her relative position and social status in the socioeconomic structure) plays an important role (Adler, Epel, Castellazzo, & Ickovics, 2000; Ostrove, Adler, Kuppermann, & Washington, 2000; Singh-Manoux, Marmot, & Adler, 2005), our conceptualization of SEP as a stressor involves both objective and subjective components.

The specific hypotheses that were tested included (a) pregnant women who recalled experiencing high levels of stress during childhood or reported high stress in adulthood would be more likely to have high levels of prenatal stress; (b) after controlling for adult stressors, childhood stressors and recalled stress in childhood would be significantly and positively associated with prenatal stress; (c) family cohesion, subjective SEP, and social support reported in both childhood and adulthood would mediate the relationships between early stressors in childhood and prenatal stress; (d) subjective SEP in childhood and adulthood would play a more important role in childhood and prenatal stress than objective SEP; and (e) recalled stress in childhood would mediate the relationship between childhood stressors and prenatal stress.

Methods

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

This study was part of a prospective cohort study exploring the influence of stressors in childhood, recalled childhood stress, and stressors in adulthood on pregnancy outcomes. The primary outcome was perceived stress in pregnancy and the secondary outcomes were related to neonatal health. Ethics approval for the study was obtained from the Hamilton Health Sciences/McMaster University Faculty of Health Sciences Research and Ethics Board and the ethics committee that serves both hospitals involved in the study.

Participants and Setting

Women were recruited consecutively from pre-birth clinics in two community hospitals in two south-western Ontario cities between September 2007 and March 2008. Both hospitals serve a largely middle-class population and are the only local hospitals with obstetric services in their communities, reporting annual deliveries of 1500 (site one) and 4500 (site two). Over 90% of women who plan to deliver at these centers attend the pre-birth clinics during their second trimester. Women were eligible for the study if they were ≥ 16 years of age; competent in English; pregnant with singleton infants; and delivering their infants at one of the recruitment sites. As with other studies that have explored predictors of prenatal stress (Woods et al., 2010), pregnant adolescent women were not invited to participate in the study because we anticipated that predictors of perceived stress may differ in pregnant adolescents as compared to non-adolescent women. Similarly, the study was limited to women with singleton pregnancies because those with multiple pregnancies may have unique risk factors for prenatal stress (Roca, Gutierrez, & Gris Martinez, 2009). Trained registered nurses in each pre-birth clinic approached eligible women to determine their interest in participating in the study and obtained their signed informed consent.

Data Collection Procedures

Consenting women were given a questionnaire to complete at home and mail back to the principal investigator in a self-addressed, pre-stamped envelope within 1 week (Dillman, 2007). As recommended by Dillman, a follow-up thank-you/reminder letter was mailed to each participant at 1 week and reminder letters to women who had not returned the questionnaire were sent 3 and 6 weeks after the clinic visit, with a second questionnaire included in the 6-week mailing. One research assistant coordinated questionnaire follow-up mailings and a second research assistant entered data into the statistical database.

Model Variables and Measures

The prenatal questionnaire included measurement instruments as well as a sociodemographic questionnaire. The prenatal questionnaire was reviewed by survey methodology and content experts for face and content validity and was pre-tested in a sample of 10 women who were 20 to 40 years of age. Minor revisions to the questionnaire were made based on pre-testing. A description of the instruments used to measure the model constructs and the composition of latent variables is found in Table 1. The main outcome of prenatal stress was measured by the Perceived Stress Scale (PSS), which has been used in other studies of pregnant women (Heaman, Blanchard, Gupton, Moffatt, & Currie, 2005; Kramer et al., 2009).

Table 1. Model Constructs and Data Collection Instruments
ConstructInstrument and Description
Perceived stress in childhood (*components of latent variable)Global Perceived Early Life Stress scale (GPELS).* item self-report questionnaire to assess perceived stress during pre-teen (age 6–12) and teen (age 13-16) years (Carpenter et al., 2004). Psychometric data for this tool are not available
 Screen for Child Anxiety Related Emotional Disorders (SCARED) (Generalized Anxiety Disorder [GAD] subscale).* GAD subscale 7-item self-report questionnaire to screen for anxiety disorders in children (Birmaher et al., 1997, 1999). This subscale has good internal consistency of .70 to .90 (Birmaher et al., 1999). Psychometric data for retrospective use are not available.
 Perceived impact of socioeconomic and family stressors* Investigator-developed questions that measured perceived stress due to specific socioeconomic and family stressors, (e.g., How stressful was ________ for you as a child?) Rated on a 5-point Likert scale ranging from very stressful to not stressful at all. Correlations between the study-developed childhood stress items and Carpenter et al.'s global perceived childhood stress items ranged from .52 to .66.
Perceived stress in pregnancy (*components of latent variable)Global perceived stress: Investigator-developed, single-item asking about amount of stress experienced in past month, past year, and past 3 years, e.g., Overall, how much stress have you experienced during the past month?* Rated on a 5-point Likert scale ranging from none to a great deal.
 Perceived Stress Scale (PSS).* 10-item version used (Cohen, Kamarck, & Mermelstein, 1983). Cronbach's alpha ranges from .78 to .93 (Cohen & Williamson, 1988). Psychometric data for retrospective use are not available.
 Perceived impact of socioeconomic and family stressors* Investigator-developed questions that measured perceived stress due to specific socioeconomic and family stressors, (e.g., How stressful is ________ for you?). Rated on a 5-point Likert scale ranging from very stressful to not stressful at all.
Family environment in childhood (*components of latent variable)Parental Bonding Inventory (PBI). 50-item self-report questionnaire intended for retrospective assessment of the quality of parental relationships during childhood (Parker, 1989, 1990). Reliability of the care scales is .91–.93 and for the protection scales it is .87–.88 (Parker, 1990).
 Family Adaptability and Cohesion Evaluation Scale III (FACES).*
 10-item cohesion subscale used retrospectively to assess cohesion/togetherness (Olson, 1986). The cohesion subscale has an internal consistency of .77 (Neabel, Fothergill-Bourbonnais, & Dunning, 2000). Psychometric data for retrospective use are not available.
 McMaster Family Assessment Device (FAD). 12-item General Functioning subscale used retrospectively to assess family functioning. This subscale has been found to have good internal consistency (Cronbach's α = .86) and construct validity (Byles, Byrne, Boyle, & Offord, 1988). Psychometric data for retrospective use are not available.
Family environment in adulthood (*components of latent variable)Family Adaptability and Cohesion Evaluation Scale III (FACES).* 10-item cohesion subscale used to assess cohesion/togetherness of current family (Olson, 1986). The 10-item cohesion subscale has an internal consistency of .77 (Neabel et al., 2000). Psychometric data for retrospective use are not available.
 McMaster Family Assessment Device (FAD). 12-item General Functioning subscale used to assess current family functioning (Neabel et al., 2000). This subscale has been found to have good internal consistency (Cronbach's α = .86) and construct validity (Byles et al., 1988). Psychometric data for retrospective use are not available.
Socioeconomic position (SEP) in childhoodObjective SEP: Index comprising average household social class based on National Statistics Socioeconomic Classification (NS-SEC) of mother and father; average household education of father and mother; and wealth (ownership of house)
 Subjective SEP: MacArthur Scales of Subjective Social Status (http://www.macSEP.ucsf.edu/research/psychosocial/usladder.php) “x” placed on rung of image of ladders (each ladder has 10 rungs): (1) community ladder (perception related to social standing/personal importance of childhood family in community) (Singh-Manoux, Adler, & Marmot, 2003) and (2) SEP ladder (perception of childhood family's standing based on occupation, education, income compared to others in country) (Singh-Manoux et al.). In a study of adolescents, reliability of the SES ladder was .73 and .79 for the community ladder (Goodman, McEwen, Dolan, Schafer-Kalkhoff, & Adler, 2005). No specific reliability estimates have been reported for adults. One study has used this tool retrospectively (Gianaros et al., 2008). The participant instructions for this instrument were modified slightly to instruct respondents to mark the social status of their family when they were children, and their current status as adults.
Socioeconomic position (SEP) in adulthoodObjective SEP Index comprising average household social class based on National Statistics Socioeconomic Classification of woman and partner (NS-SEC); average household education of woman and partner; wealth (ownership of house); and household income.
 Subjective SEP: MacArthur Scales of Subjective Social Status. “x” placed on rung of image of ladders: (1) community ladder (perception of social standing/personal importance of current family in community); (2) SEP ladder (perception of current family's standing based on occupation, education, income compared to others in country)
Social support in childhoodChild and Adolescent Social Support Scale (CASSS). Measures childhood perceived social support from teachers, classmates, and close friends (Malecki & Demaray, 2002). Coefficient alpha of the total scale is high in both elementary (.96 –. 97) and high school (.97) students (Demaray & Malecki, 2002).
Social support in adulthoodInterpersonal Support Evaluation List (ISEL-12). 12-item scale assesses tangible assistance/material aid, support or availability of a confidant (Cohen, Mermelstein, Kamarck, & Hoberman, 1985). Cronbach's alpha is .81 in adults (Cohen et al., 1985).
Number of weeks gestation at completion of surveySelf-reported
ParitySelf-reported
Maternal age at completion of surveySelf-reported
Pregnancy complicationsA composite variable (none vs 1 or more) of the following conditions: asthma, thyroid disease, heart disease, non-gestational diabetes, chronic hypertension. Data were retrieved from the Niday Perinatal Database. This database is a repository of maternal and infant data that are routinely collected on all births in Ontario, Canada (http://www.bornontario.ca/about-born/founding-members/niday-perinatal-and-nicu-scn-database)

Because no psychometrically evaluated measure of childhood perceived stress was found, seven items assessing the appraised impact of specific childhood socioeconomic and family stressors were developed to provide understanding of their contributions to recalled childhood stress. Item generation was guided by the same systematic review that informed model development. Face and content validity of these items also were assessed by content and survey methodology experts; in addition, these questions were evaluated as part of the questionnaire pre-testing process. No changes to these investigator-developed questions were made as a result of pre-testing.

We also included a global measure of perceived childhood stress (Carpenter et al., 2004) that used a single-item measure to assess overall perceived stress for ages 6–12 years and 13–16 years. Studies that have psychometrically evaluated brief, global measures of psychosocial stress have demonstrated reliability and validity of this form of measure (Elo, Leppanen, & Jahkola, 2003; Littman, White, Satia, Bowen, & Kristal, 2006). Given that family functioning, cohesion, and parental bonding have been associated with psychological distress in childhood (Buschgens et al., 2010; Repetti, Taylor, & Seeman, 2002; Van Oort, Verhulst, Ormel, & Huizink, 2010) and adulthood (Luecken, Rodriguez, & Appelhans, 2005), we included measures of each of these concepts.

Statistical Analysis

Structural equation modeling (SEM) is an analytic approach that allows the simultaneous estimation of both direct and indirect a priori hypothesized

relationships using regression coefficients while adjusting for other factors in the model (Kline, 2005). It particularly useful for exploring the complex etiology of prenatal stress because it allows for the testing of multiple theoretical pathways and permits the inclusion of both early life and adult variables in the same model for analysis.

After testing the assumption of multivariate normality, we used a two-step modeling procedure as recommended by Kline (2005). First, we analyzed and refined the fit of the measurement components of the model. Once these were well-fitting, we analyzed the fit of the model as a whole. Multiple model fit indices were used to assess model fit, including the model chi-square, the Steiger-Lind root mean square error of approximation (RMSEA), and the Bentler comparative fit index (CFI) (Kline). We used maximum likelihood estimation for estimation of means, variances, and covariances of the variables in order to retain records with missing data in our analysis.

Modifications to the model were guided by the theoretical plausibility of the pathway and model fit statistics (Kline, 2005). Statistically non-significant pathways (p < .05) were removed from the final model to promote parsimony if the model fit was not adversely affected by their deletion. We used Baron's four-step process to assess variables for mediation (Baron & Kenny, 1986).

Results

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

The Sample

The sample comprised 441 women, 263 from site one and 178 from site two. The participation rate was 68.2% at site one and 44.0% at site two for an overall rate of 56.4% (639/1133). The questionnaire return rate was 74% (472/639). Of these 472 women, 12 were excluded because their questionnaires were returned post-delivery and one delivered twins. The loss to follow-up rate was 4.1%. Sample characteristics are displayed in Tables 2 and 3.

Table 2. Sociodemographic and Family Characteristics of the Sample (n = 441)
CharacteristicM (SD)n (%)
Note
  1. = mean. SD = standard deviation. n = number of respondents. SEP = socioeconomic position. FACES = Family Adaptability and Cohesion Evaluation Scale III.

Maternal age (years)29.8 (5.8)  
Primiparity 190(43.1)
Marital status
 Married/Common-law 420(95.2)
 Single/Separated 21(4.8)
Born in Canada 365(83.0)
Maternal education
 High school or less 80(18.1)
 Some or completed community college or bachelor's degree 303(68.9)
 Graduate degree 57(13.0)
Income
 < $20,000 22(5.2)
 $20000–$39,999 61(14.3)
 $40000–$59,999 63(14.7)
 $60000–$79,999 70(16.4)
 $80000 212(49.5)
Pregnancy complications (1 or more) 34(7.7)
Child family community ladder6.6 (1.7)  
Child family SEP ladder6.2 (1.7)  
Adult community ladder6.5 (1.5)  
Adult SEP ladder6.4 (1.6)  
Child FACES III36.2 (7.9)  
Adult FACES III42.3 (5.8)  
Table 3. Psychosocial Characteristics of the Sample (n = 441)
CharacteristicM (SD)n (%)
Note
  1. M = sample mean; SD = standard deviation; n = number of respondents by characteristic. ISEL = Interpersonal Support Evaluation List.

  2. aQuestion was asked of respondents who reported that their family experienced financial stress.

Child stress in elementary school
 Essentially stress free 116(26.4)
 Less stressful than most kids my age 106(24.1)
 About average 147(33.5)
 More stressful than most kids my age 54(12.3)
 Extremely stressful 16(3.6)
Child stress in teen years
 Essentially stress free 47(10.7)
 Less stressful than most kids my age 108(24.7)
 About average 159(36.2)
 More stressful than most kids my age 100(22.8)
 Extremely stressful 25(5.7)
Overall stress in childhood
 Extremely or very stressful 46(10.5)
 Moderately stressful 108(24.6)
 Mildly stressful 200(45.6)
 Not at all stressful 85(19.4)
Child family stress
 Extremely or very stressful 54(12.4)
 Moderately stressful 86(19.8)
 Mildly stressful 158(36.3)
 Not at all stressful 137(31.5)
Child financial stressa
 Extremely or very stressful 27(11.7)
 Moderately stressful 56(24.2)
 Mildly stressful 110(47.6)
 Not at all stressful 38(8.6)
Adult Perceived Stress Scale17.0 (3.9)  
Adult overall how stressful past month
 Extremely or very stressful 53(12.1)
 Moderately stressful 115(26.1)
 Mildly stressful 209(47.5)
 Not at all stressful 63(14.3)
Adult overall how stressful past year
 Extremely or very stressful 54(12.3)
 Moderately stressful 140(31.8)
 Mildly stressful 204(46.4)
 Not at all stressful 42(9.5)
Adult overall how stressful past 3 years
 Extremely or very stressful 55(12.5)
 Moderately stressful 154(35.0)
 Mildly stressful 196(44.5)
 Not at all stressful 35(8.0)
Adult family stress
 Extremely or very stressful 14(3.3)
 Moderately stressful 40(9.2)
 Mildly stressful 178(40.9)
 Not at all stressful 203(46.0)
Adult financial stressa
 Extremely or very stressful 44(17.3)
 Moderately stressful 78(30.6)
 Mildly stressful 114(44.7)
 Not at all stressful 19(7.5)
Adult ISEL42.0 (5.3)  

Women's Characteristics. Most women were married or living common-law, had some post-secondary education, and were born in Canada. Almost half of the women had household incomes greater than CAN $80000 and the mean MacArthur ladder ratings of socioeconomic position were slightly higher than those reported in another study of pregnant women (Ostrove et al., 2000). Almost one-half of the women indicated that they experienced moderate to extreme levels of stress during the past year and the past three years.

Women's childhood Characteristics. During the women's childhood, the majority of their parents were married and almost half of their fathers and mothers had some post-secondary education. Sixteen percent of women recalled their elementary years as more stressful or extremely more stressful than others their age, while one-quarter of the sample recalled experiencing this level of stress during their teen years. Over 30% of women recalled their childhood as moderately to extremely stressful, and almost half reported scores of recalled childhood anxiety above the screening cutoff point of 9 on the Generalized Anxiety Disorder subscale (GAD).

The Final Model

The final model is shown in Figure 2. All pathway estimates (β) are reported in standardized format. Pathway estimates and associated significance levels are noted on Figure 2, and pathways are numbered (e.g., pathway 1) to correspond with the description of the findings. Model statistics indicated that the final model adequately fit the data (CFI = .93; RMSEA = .048). The model chi-square value was statistically significant, X2 (706, N = 441) = 1407.5, p < .001, suggesting that there were significant differences between observed and implied covariance matrices of the model. However, this is not unusual for large, complex models (Kline, 2005). Calculation of the normed chi-square (NC) resulted in a value of 2.0, indicating a reasonable fit (Kline). The correlation matrix is found in Table 4.

image

Figure 2. Final Life Course Prenatal Maternal Stress Model. Standardized pathway coefficients are reported. Pathway numbers are in square brackets following the coefficient. Pathways 2, 4, 5, 7, and 8 are non−significant (NS). SEP = socioeconomic position.

***p < .001. **p < .01. *p < .05.

Download figure to PowerPoint

Table 4. Implied Correlation Matrix
 Child objective SEPChild supportChild subjective SEPAdult supportAdult objective SEPAdult subjective SEPAdult family stressAdult stress past monthAdult financial stressAdult stressChild family financial stressChild overall stressChild family stressSCAREDStress teen yearsChild deprivationStress elementary yearsAdult stress past yearAdult stress past 3 yearsPSS
Note
  1. Correlations are not reported for individual family cohesion items. SEP = socioeconomic position. PSS = percieved stress scale.

  2. **p < .01. *p < .05.

Child objective SEP1.000                   
Child support.0771.000                  
Child subjective SEP.482**.267**1.000                 
Adult support.050.358**.133**1.000                
Adult objective SEP.345**.027.166**.0171.000               
Adult subjective SEP.189**.128**.424**.178**.509**1.000              
Adult family stress−.063−.139**−.096−.247**−.157**−.260**1.000             
Adult stress past month−.070−.155**−.107*−.275**−.174**−.289**.376**1.000            
Adult finance stress−.081−.180**−.125**−.320**−.203**−.336**.438**.487**1.000           
Adult deprivation−.072−.160**−.110**−.227**−.180**−.298**.388**.431**.502**1.000          
Child family financial stress−.162**−.264**−.328**−.138**−.056−.079.172**.191**.223**.197**1.000         
Child overall stress−.206**−.336**−.418**−.175**−.071.−101*.219**.243**.284**.251**.645**1.000        
Child family stress−.192**−.313**−.390**−.163**−.066−.094.204**.227**.264**.234**.601**.765**1.000       
SCARED−.120**−.195**−.243**−.102*−.041−.059.127**.142**.165*.146**.375**.478**.656**1.000      
Stress teen years−.169**−.276**−.344**−.144**−.058−.083.180**.200**.233**.207**.530**.675**.645**.540**1.000     
Child deprivation−.160**−.262**−.326**−.137**−.055−.079.171**.190**.221**.309**.503**.558**.667**.559**.550**1.000    
Stress elementary years−.165**−.269**−.336**−.141**−.057−.081.176**.195**.228**.202**.518**.659**.656**.550**.541**.559**1.000   
Adult stress past year−.072−.160**−.111*−.284**−.180**−.298**.389**.432**.503**.446**−.198**.252**−.255**−.213**−.210**−.217**−.213**1.000  
Adult stress past 3 years−.061−.135**−.093−.240**−.153**−.252**.329**.365**.426**.377**.167**.213**−.215**−.180**−.177**−.184**−.181**.626**1.000 
PSS−.075−.166**−.115**−.295**−.187**−.310**.403**.660**.522**.463**.205**.261**−.264**−.221**−.218**−.225**−.222**.464**.392**1.000

Pathways to Prenatal Stress

Impact of Recalled Childhood Stress on Prenatal Stress [pathway 1]. As hypothesized, recalled childhood stress was significantly related to prenatal stress (β = .54), after adjustment for other child and adult factors

(Fig. 2). As such, a woman who recalled experiencing high stress during childhood also tended to perceive that she had high stress in pregnancy. In mediational analyses, adult subjective SEP and adult family cohesion partially mediated this relationship. This finding suggested that recalled childhood stress influenced prenatal stress, in part, because it had a negative impact on women's family cohesion and subjective SEP.

Impact of Recalled Childhood and Adult Family Cohesion on Prenatal Stress [pathways 2 and 3]. Recalled family cohesion in childhood was indirectly associated with prenatal stress through its effect on recalled childhood stress, current family cohesion, and current subjective SEP. That is, a woman who reported that her childhood family was characterized by lack of closeness was more likely to recall experiencing stress during her childhood, and this was related to prenatal stress. Similarly, a woman who reported having low childhood family cohesion was more likely to have impaired family functioning in her current family, and this contributed to prenatal stress. Unlike childhood family cohesion, adult family cohesion was directly associated with prenatal stress (β = −.25). In other words, a pregnant woman with impaired adult family cohesion was more likely to experience prenatal stress.

Impact of recalled childhood and adult subjective SEP on prenatal stress [pathways 4 and 6]. Recalled childhood subjective SEP was indirectly associated with prenatal stress through its effect on adult subjective SEP and recalled childhood stress. As such, a woman who recalled that her family had low social standing in childhood tended to have this same perception during pregnancy, and this was related to prenatal stress. We also found that women who reported having low subjective SEP in childhood recalled higher levels of stress in childhood and higher levels of prenatal stress. Adult subjective SEP was directly related to prenatal stress, having a moderate, negative effect (β = −.44).

Impact of recalled childhood and adult objective SEP on prenatal stress [pathways 5 and 7]. Objective SEP reported during childhood or adulthood was not directly related to prenatal stress once controlled for by subjective SEP. We found that women who reported having low objective SEP in childhood were more likely to experience low objective SEP in adulthood, and to report low subjective status in both childhood and adulthood.

Impact of Recalled Social Support During Childhood and Adulthood on Prenatal Stress [pathways 8 and 9]. The effect of recalled child social support on prenatal stress was indirect, mediated by its effect on recalled childhood stress and adult social support. In other words, women who recalled having low social support in childhood also reported experiencing greater stress in childhood and low support during pregnancy, with both of these factors playing a role in prenatal stress. We found that adult social support had a direct, negative effect on prenatal stress (β = –.23). This relationship was mediated to a small degree by both adult family cohesion and subjective SEP. As such, a woman with high social support during pregnancy experienced lower prenatal stress. In addition, her high level of social support enhanced her family cohesion and sense of subjective SEP, which in turn reduced her level of prenatal stress.

Additional Factors. None of the additional factors that we assessed were significantly related to prenatal stress, including parity, current marital status, maternal age, weeks gestation at the time of survey completion, pregnancy complications, ethnicity, and whether born in Canada or not.

Findings indicate that women's pre-conception experiences, including those recalled from childhood, have an influence on their levels of prenatal stress.

Discussion

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

Few studies have explored the impact of adult factors on prenatal stress, and research on child factors is even more limited. As such, our study makes a significant contribution to the small body of existing literature on predictors of stress in pregnancy, and provides new evidence regarding the association between recalled stress in childhood and prenatal stress.

Our finding that recalled childhood stress was associated with prenatal stress is consistent with a previous study that demonstrated a linkage between retrospectively measured childhood stress and physiologic stress dysregulation in pregnant women (Shea et al., 2007). Although the mechanisms underlying the association between childhood stress and prenatal stress are unclear, some have proposed that stress during the critical, formative childhood years may interact with a genetic predisposition (Luecken & Lemery, 2004; Steptoe, 2008) or a biological vulnerability to psychological disorders in adulthood (Luecken, Kraft, Appelhans, & Enders, 2009; Nemeroff, 2004). Others have found that adverse outcomes of early childhood stress may be due to ineffective coping mechanisms (Olff, Langeland, & Gersons, 2005), lack of perceived control over stressors (Chorpita & Barlow, 1998), a tendency to view the world as a threatening place (Chen & Matthews, 2003; Chen, Langer, Raphaelson, & Matthews, 2004; Luecken & Lemery), poor school performance that impacts life opportunities, or impaired interpersonal relationships (Hazell, 2007).

The role of recalled childhood family cohesion in the perception of prenatal stress is consistent with others who have found that lack of family cohesion or a high level of family conflict has an adverse effect on adult psychological health (Kuh, Hardy, Rodgers, & Wadsworth, 2002; Luecken et al., 2005). Most research has focused on the impact of poor family relationships on the psychological health of a child (rather than the child as an adult). Some studies have found that the childhood family influences family functioning in adulthood through negative interpersonal processes and family communication patterns (Huurre, Junkkari, & Aro, 2006; Whitton et al., 2008), impaired parent-child attachment (Cassidy, Lichtenstein-Phelps, Sibrava, Thomas, Jr., & Borkovec, 2009), and the intergenerational transmission of parenting practices (Belsky, Jaffee, Hsieh, & Silva, 2001; Capaldi, Pears, Kerr, & Owen, 2008). This study extends this body of research by suggesting that other mechanisms underlying the association between impaired childhood family cohesion and adult family cohesion may involve increased stress in childhood and low adult family cohesion.

In our study, low adult family cohesion was one of the main factors associated with prenatal stress. Given that our sample comprised childbearing women, a major contributor to family cohesion most likely is the quality of the spousal/partner relationship. Other studies have found that poor partner relationships are significantly related to increased perceived stress (Stancil, Hertz-Picciotto, Schramm, & Watt-Morse, 2000) and physiological markers of stress (Robles & Kiecolt-Glaser, 2003).

Our findings are consistent with the few studies have explored the influence of adult perceived SEP on psychological stress (Adler et al., 2000; Ghaed & Gallo, 2007) and found that subjective SEP was a more important predictor of stress than objective SEP. It may also be that the lack of relationship between objective SEP and our outcomes is in part due to the high SEP that the majority women in the sample enjoyed. Among adolescents, low levels of subjective SEP have been associated with perceived stress (Goodman et al., 2005) and greater physiological reactivity to stress (Chen et al., 2004). This study extends this work by suggesting that the perception of social status that is formed during childhood may persist into adulthood and play a role in prenatal stress.

Others also have observed that childhood objective SEP plays an indirect role in adult psychological health by influencing adult objective SEP and related life opportunities (Sacker, Schoon, & Bartley, 2002). This study adds to this body of research by suggesting that low objective childhood SEP recalled by women is associated with prenatal stress through its effect on perceptions of social status in childhood and adulthood.

Women who reported having low childhood social support tended to have higher prenatal stress, a relationship that was mediated by recalled child stress and adult social support. Children with strong social support may possess a more adaptive coping style that reduces their stress (Shulman, 1993), or may have relationship skills, personality traits, or advantageous life circumstances that transfer into adulthood and facilitate the garnering of social support as adults.

Strengths and Limitations

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

Our study has several strengths. The sample was drawn from two community hospitals in two cities, which enhances the generalizability of the findings. Using structural equation modeling, we described direct and indirect pathways linking women's recalled childhood stressors and those experienced in adulthood to prenatal stress, which can inform potential prevention and intervention strategies. Our study also provided insight into the comparative effects of objective and subjective SEP on stress reported in childhood and pregnancy.

The limitations of this study must be considered. We cannot eliminate the possibility of selection bias by systematically excluding women who did not attend pre-birth clinics, although our sample characteristics reflect the community profiles. Selection bias also may have been introduced through the relatively low participation rate; however, our rates are comparable with other studies involving pregnant women (Daniels et al., 2006). The childhood measures were retrospective and women's perceptions of their childhood may be altered by their adult circumstances or mood. We did not control for current mood, and were unable to assess its impact on the relationships between subjective childhood factors and prenatal stress. However, the influence of current mood on recall has been found to be a greater issue with short-term than long-term recall (Blaney, 1986; Matt, Vazquez, & Campbell, 1992), and a recent study comparing the effect of prospectively- and retrospectively-collected data of childhood adversity found no difference in the level of risk of psychological adjustment in middle-aged women (Hardt, Vellaisamy, & Schoon, 2010).

Conclusion and Implications

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

Overall, the findings of this study indicate that women's pre-conception experiences, including their recollections from childhood, have an influence on their levels of prenatal stress. As such, the implications of this study are two-fold. Firstly, the findings highlight the importance of early assessment and intervention in girls’ mental health in order to reduce the consequences of poor mental health across the life course. Secondly, the results can be used to inform the development of psychosocial assessment instruments and interventions that target women pre-conceptually and during pregnancy. Given that most women do not seek care for mental health issues during the perinatal period (Woolhouse, Brown, Krastev, Perlen, & Gunn, 2009), routine well-woman visits for women of childbearing age and prenatal care provide windows of opportunity for risk assessment and intervention. Applying the Theory of Stress Appraisal (Lazarus & Folkman, 1984) to prenatal stress, risk assessment and intervention may be targeted at a mother's source of distress (i.e., stressors), protective factors, perception or appraisal of the stressor, or coping strategies. The modifiable factors (e.g., current family cohesion, stress appraisal) represent opportunities for intervention. The non-modifiable stressors (e.g., recalled childhood stress, poor childhood family cohesion) may lend themselves to a strengths-based approach whereby providers assist women to identify and understand influences on their levels of stress, gain insight into and mobilize their stress-reducing coping strategies, develop and engage healthy coping processes, and bring resolution to past adversity.

Understanding preconception and pregnancy-related factors associated with prenatal stress can inform psychosocial risk assessment, referral, and care across women's lives.

Based on the role that family cohesion plays in women's stress, our findings also substantiate the need for a family-centered approach to the care of women's psychosocial needs. Future studies should seek to use prospective designs to evaluate the effects of stressors across women's lives on their pre-conception and prenatal mental health and incorporate life course predictors of prenatal stress into prenatal psychosocial risk assessment instruments.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies

Biographies

  1. Top of page
  2. ABSTRACT
  3. Methods
  4. Results
  5. Discussion
  6. Strengths and Limitations
  7. Conclusion and Implications
  8. REFERENCES
  9. Biographies
  • Dawn Kingston, RN, PhD is an assistant professor in the Faculty of Nursing, Edmonton Clinic Health Academy, University of Alberta, Edmonton, Alberta, Canada.

  • Wendy Sword, RN, PhD, is a professor and assistant dean of research in the School of Nursing, McMaster University, Hamilton, ON, Canada.

  • Paul Krueger, PhD, is an associate professor and associate director of research in the Department of Family and Community Medicine, University of Toronto, Toronto, ON, Canada.

  • Steve Hanna, PhD, is an associate professor and assistant dean for the Health Research Methods Program in the Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, ON, Canada.

  • Maureen Markle-Reid, RN, MScN, PhD, is an associate professor in the School of Nursing, McMaster University, Hamilton, ON, Canada.