Citation
Critical periods of maternal stress exposure and early childhood obesity : exploring risk and protective factors in New Zealand

Material Information

Title:
Critical periods of maternal stress exposure and early childhood obesity : exploring risk and protective factors in New Zealand
Creator:
Farewell, Charlotte V.
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Doctor of philosophy)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
School of Education and Human Development
Degree Disciplines:
Health and behavioral sciences
Committee Chair:
Tracer, David
Committee Members:
Thayer, Zaneta
Scandlyn, Jean
Puma, Jini

Notes

Abstract:
Exposure to environmental stressors during development alters human biology. Importantly, stress experienced by mothers during both the pre- and post-natal periods of growth can have negative impacts on offspring development. The primary objective of this dissertation was to use mixed-methods to explore the relationship between pre- and post-natal maternal stress and early childhood obesity among a nationally representative sample of New Zealand mothers. The Growing Up in New Zealand longitudinal study provided information on 5,839 pregnant women and their children to assess the quantitative objectives. Exposure to one additional objective stressor during pregnancy was significantly associated with a .06 increase in BMI z-score at 54-months (p<.01), after controlling for covariates. Exposure to maternal stress during either the pre- or early post-natal period was associated with higher childhood BMI at 54-months of age relative to children of women not exposed to stress (p<.01). Individuals who experienced stress both prenatally and at 24-months had children with significantly higher BMI at 54-months than individuals who experienced stress at neither or only one time point (p<.01). Structural equation modeling and qualitative methods (n=74) revealed ethnic variations in the lived experience of maternal stress and risk and protective pathways between stress and early childhood obesity. This study informs our understanding of sociocultural influences on risk exposures, protective factors and stress responses early in life, and resulting impacts on offspring obesity risk. Findings may help identify strategies that decrease early life predisposition to chronic disease.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Charlotte Farewell. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
CRITICAL PERIODS OF MATERNAL STRESS EXPOSURE AND EARLY CHILDHOOD
OBESITY:
EXPLORING RISK AND PROTECTIVE FACTORS IN NEW ZEALAND
by
CHARLOTTE V. FAREWELL B.S., University of Richmond, 2004 MPH, Tulane University, 2012
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences
2019


This thesis for the Doctor of Philosophy degree by Charlotte V. Farewell has been approved for the Health and Behavioral Sciences Program by
David Tracer, Chair Zaneta Thayer Jean Scandlyn Jini Puma
Date: May 18th, 2019
ii


Farewell, Charlotte (PhD, Health and Behavioral Sciences)
Critical Periods of Maternal Stress Exposure and Early Childhood Obesity: Exploring Risk and Protective Factors in New Zealand Thesis directed by: Professor David Tracer
ABSTRACT
Exposure to environmental stressors during development alters human biology. Importantly, stress experienced by mothers during both the pre- and post-natal periods of growth can have negative impacts on offspring development. The primary objective of this dissertation was to use mixed-methods to explore the relationship between pre- and post-natal maternal stress and early childhood obesity among a nationally representative sample of New Zealand mothers. The Growing Up in New Zealand longitudinal study provided information on 5,839 pregnant women and their children to assess the quantitative objectives. Exposure to one additional objective stressor during pregnancy was significantly associated with a .06 increase in BMI z-score at 54-months (p<01), after controlling for covariates. Exposure to maternal stress during either the pre- or early post-natal period was associated with higher childhood BMI at 54-months of age relative to children of women not exposed to stress (p<01). Individuals who experienced stress both prenatally and at 24-months had children with significantly higher BMI at 54-months than individuals who experienced stress at neither or only one time point (p<01). Structural equation modeling and qualitative methods (n=74) revealed ethnic variations in the lived experience of maternal stress and risk and protective pathways between stress and early childhood obesity. This study informs our understanding of sociocultural influences on risk exposures, protective factors and stress responses early in life, and resulting impacts on offspring obesity risk. Findings may help identify strategies that decrease early life predisposition to chronic disease.


The form and content of this abstract are approved. I recommend its publication.
Approved: David Tracer
IV


To Sean, for demanding that I balance hard work with self-care and play, supporting my adventures across the world, bolstering my self-esteem when I needed it most, and keeping me
well-fed. I love you.
v


ACKNOWLEDGEMENTS
I acknowledge first and foremost my dissertation chair, Dr. David Tracer, and advisor, Dr. Zaneta Thayer, as well as my two additional committee members, Dr. Jini Puma and Dr. Jean Scandlyn, for their invaluable support, investment, and expertise throughout the course of this project.
I acknowledge the children and the families who are part of the Growing Up in New Zealand study, and the women who participated in the focus groups and interviews for this project. This includes the staff at the Central Plunket clinics for their support in promoting this research project and supporting recruitment of participants.
I acknowledge the multiple government agencies that fund and support Growing Up in New Zealand, in particular the Ministry of Social development and the former Social Policy Evaluation and Research Unit (also formerly the Families Commission) for their management of the contract on behalf of the Crown, as well as the ongoing support from Auckland UniServices and the University of Auckland. I thank all the members of the Growing Up in New Zealand research team for their invaluable work in interviewing participants and managing the data used in this analysis, as well as the members of Growing Up’s Kaitiaki Group and Executive Scientific Advisory Group.
Finally, I acknowledge the National Science Foundation and specifically the Biological Anthropology Program for funding this dissertation project through their Doctoral Dissertation Improvement Grant award.
VI


TABLE OF CONTENTS
I. INTRODUCTION.....................................................................1
1.1 Introduction to Study.......................................................1
1.2 Problem Statement...........................................................4
1.3 Research Objective..........................................................5
1.4 Aims and Hypotheses.........................................................5
II. THEORETICAL AND EMPIRICAL BACKGROUND............................................8
2.1 What is Stress?.............................................................8
2.2 Measures of Stress..........................................................8
2.3 Intergenerational Transmission of Stress...................................11
2.4 Maternal Stress and Early Childhood Obesity through a DOHAD lens...........12
2.5 Maternal Stress and Early Childhood Obesity through a Life Course Epidemiology
lens............................................................................13
2.6 Proposed Pathways between Maternal Stress and Early Childhood Obesity......15
2.6.1 Biological Pathways...................................................15
2.6.2 Behavioral Pathways...................................................18
2.7 Lazarus Theory of Stress and Coping........................................21
2.8 Culture and Resilience.....................................................21
2.9 Maternal Stress and Childhood Obesity in a New Zealand Context.............24
2.10 Literature Review Summary.................................................27
III. QUANTITATIVE METHODS..........................................................28
3.1 Study Design Overview......................................................28
3.2 Quantitative Participant Selection.........................................31
3.3 Quantitative Data Collection Methods.......................................31
vii


3.4 Quantitative Data Measures
33
3.4.1 Independent Variables................................................34
3.4.2 Dependent Variables..................................................37
3.4.3 Moderating and Mediating Variables...................................38
IV. QUANTITATIVE ANALYSES AND FINDINGS............................................41
4.1 Data Cleaning.............................................................41
4.2 Aim #1....................................................................45
4.3 Aim #2....................................................................47
4.4 Aim #3....................................................................50
4.5 Aim #4....................................................................53
4.6 Summary of Quantitative Findings..........................................64
V. QUALITATIVE METHODS...........................................................66
5.1 Study Design Overview.....................................................66
5.2 Participant Selection.....................................................67
5.3 Data Collection Methods...................................................68
5.4 Data Collection Instruments...............................................69
VI. QUALITATIVE ANALYSES AND RESULTS..............................................72
6.1 Participants..............................................................71
6.2 Data Analyses.............................................................72
6.3 Results of Analyses.......................................................74
Domain 1: Sources of Maternal Stress.......................................74
Domain 2: Maternal Stress and Risk Behaviors...............................82
Domain 3: Maternal Stress and Protective Factors...........................89
viii


Domain 4: Maternal Stress and Early Childhood BMI: Patterns of risk and protective
factors......................................................................100
VII. DISCUSSION.....................................................................104
7.1 Justification of research topic.............................................104
7.2 Recap of literature review and methodology..................................105
7.3 Brief recap of results......................................................106
7.4 AIM 1 Discussion............................................................108
7.5 AIM 2 Discussion............................................................109
7.6 AIM 3 Discussion............................................................Ill
7.7 AIM 4 and Aim 5 (Mixed-Methods) Discussion..................................113
7.7.1 Sources of Maternal Stress............................................113
7.7.2 Risk Behavior Pathways between Maternal Stress and Early Childhood BMI... 113
7.7.3 Protective Factor Pathways between Maternal Stress and Early Childhood
BMI.............................................................118
VIII. CONCLUSION AND IMPLICATIONS............................................132
8.1 Final Summary of Research Question and Overall Findings...............132
8.2 Reliability, Validity and Generalizability...........................133
8.3 Limitations..........................................................136
8.4 Contributions of the Study...........................................137
8.5 Recommendations and Implications for Future Research.................139
REFERENCES...................................................................142
IX


LIST OF TABLES
TABLE
3.1 Mixed- Methods Research Timeline (Implementation Time Frame: February 2017- March
2019......................................................................................30
3.2 Demographic Characteristics of the Quantitative Sample (n=5,839)....................32
3.3 Variables included in Quantitative Analyses..........................................33
3.4 Objective Stressors included in GUiNZ Vulnerability Scale...........................35
4.1 Descriptive Statistics before and after Multiple Imputation..........................44
4.2 Correlations between all continuous variables........................................46
4.3 Hierarchical Linear Regression Models: BMI at 24-months of age.......................48
4.4 Hierarchical Linear Regression Models: BMI at 54-months of age.......................49
4.5 Differences between Maternal Stress Groups and Childhood BMI at 54-months of age after
controlling for covariates................................................................52
4.6 Sidak Corrected Post Hoc Comparisons for Childhood BMI at 54-months of age among four
Maternal Stress Groups....................................................................52
6.1 Qualitative Sample Characteristics...................................................73
6.2 Percentages of coding references and quotes related to the most commonly mentioned
sources of stress among a diverse group of New Zealand mothers...........................80


LIST OF FIGURES
FIGURE
1.1 DOHaD and Life Course Epidemiology models of early life stress and BMI at 54-
months...............................................................................6
2.1 Proposed Biological and Behavioral Pathways between Pre- and Post-natal Maternal Stress
and Early Childhood Obesity.........................................................15
4.1 Maternal stress from prenatal to 24-months of age and average childhood BMI at 54-months
of age..............................................................................51
4.2 Hypothesized Path Analysis Model of Prenatal Stress, Childhood BMI at 54-months, and
mediators and moderators............................................................55
4.3 Mediation Model.................................................................56
4.4 Overall Path Model..............................................................58
4.5 Final Path Analysis Model among European Mothers in New Zealand.................59
4.6 Final Path Analysis Model among Maori Mothers in New Zealand....................60
4.7 Final Path Analysis Model among Pacifika Mothers in New Zealand.................61
4.8 Total Effects of Objective Prenatal Stress on Childhood BMI among Pacifika Women
experiencing Low, Medium, and High levels of Protective Factors.....................62
4.9 Final Path Analysis Model among Asian Mothers in New Zealand....................63
5.1 Mixed-Methods Design............................................................67
XI


LIST OF ABBREVIATIONS
Developmental Origins of Health and Disease (DOHaD) Body Mass Index (BMI)
Growing Up in New Zealand (GUiNZ)
Perceived Stress Scale (PSS)
Hypothalamus-Pituitary-Adrenal (HPA) Corticotropin-Releasing Hormone (CRH)
Computer Assisted Personal Interview (CAPI)
Food Frequency Questionnaire (FFQ)
Variance Inflation Score (VIF)
Statistical Package for the Social Sciences (SPSS) Analysis of Covariance (ANCOVA)
Structural Equation Modeling (SEM)
Comparative Fit Index CFI)
Root Mean Square (RMSEA)
Akaike information Criterion (AIC)
International Obesity Task Force (IOTF)
Standardized Root Mean Square Residual (SRMR) Confidence Interval (Cl)
Standard Deviation (SD)
New Zealand Dollar (NZD)
Intimate Partner Violence (IPV)
XII


CHAPTER 1
INTRODUCTION
1.1 Introduction to Study
Exposure to environmental stressors alters human biology (Kuzawa & Quinn, 2009; Kuzawa & Sweet, 2009; Wells, Chomtho, & Fewtrell, 2007). Importantly, stress experienced by mothers during both the pre- and post-natal periods can have impacts on offspring development, including emotional, neurodevelopmental, and physical consequences (Gluckman, Hanson, Cooper, & Thornburg, 2008; Rice et al., 2010). Previous research examining early experience of maternal stress and its impact on children’s biology can be framed by two models: the Developmental Origins of Health and Disease (DOHaD) model (Gluckman, 2008) and Life Course Epidemiology (Kuh, 2003).
The DOHaD model explores the biological impacts of stress primarily during the prenatal and early postnatal period of development (Gluckman, 2008). Research has found associations between exposure to prenatal stress and the prevalence of metabolic syndrome, diabetes, cardiovascular disease, and obesity in adulthood/later life (Benyshek, 2007; Cao-Lei et al., 2015; DJP, 2004; K.N. et al., 2012; Liu, Dancause, Elgbeili, Laplante, & King, 2016a; McMillen et al., 2008). However, a focus on stress exposures solely during the perinatal period, or prenatal and early postnatal period, ignores the potential cumulative impacts of postnatal stress on human biology.
The life course approach to childhood adversity expands the DOHaD model to incorporate an epidemiological life history perspective arguing that cumulative stress from the prenatal period through childhood and adulthood plays a critical role in the development of poor health (Kuh, 2003). The life course model proposes that both the environment and genes
1


influence health and development throughout the life course. Studies have found associations between accumulation of stress throughout the first years of life and a multitude of mental and physical health outcomes in adulthood, including post-traumatic stress disorder, major depressive disorder, schizophrenia, diabetes, obesity, and cardiovascular disease (Cameron & Demerath, 2002; Daskalakis, Bagot, Parker, Vinkers, & de Kloet, 2013).
A common limitation of prior research exploring impacts of maternal stress on child development outcomes is reliance on a single measure of stress. Stress is a multidimensional, translational concept (Lazarus & Folkman, 1987). Transactions between external and internal demands and resources impact individual’s experiences of stress during the pre- and post-natal periods. Stress-provoking factors (e.g., external objective stressors), stress-mediating or moderating factors (e.g. coping, social support) and stress-resulting factors (e.g., perceived stress) capture different aspects of the human stress response (Lazarus & Folkman, 1987). Measures of objective stress (i.e. objective stressors) and perceived stress therefore may or not be highly correlated (Kingston, Sword, Krueger, Hanna, & Markle-Reid, 2012; Laplante, Brunet, Schmitz, Ciampi, & King, 2008; Turner & Avison, 2003). Multiple measures of stress are needed to best understand correlations between these dimensions.
Early childhood body mass index (BMI) is a useful outcome to explore the impacts of maternal stress during critical periods of development as offspring weight and height are extremely sensitive to environmental influences (Entringer, Buss, & Wadhwa, 2010). Early childhood obesity, defined by BMI scores at or above the 95th percentile, is reaching epidemic proportions on a global level (World Health Organization, 2013). Further, childhood obesity leads to adult obesity, which is a major risk factor for the development of chronic diseases, many of which have been separately associated with early life factors such as low birth weight (Biro &
2


Wien, 2010). Consideration of early life factors that may play a role in the onset of obesity requires further analyses of both the timing and duration of different stress exposures.
Not all children whose mothers experience early life stress develop obesity (Walton, Simpson, Darlington, & Haines, 2014). An integration of biological and sociocultural perspectives using mixed-methods can be employed to explore these complex relationships (Rodney & Mulligan, 2014). A biocultural approach is useful for understanding the lived experience of stress and why there is variability in early childhood obesity in response to early life stress. More specifically, this approach allows an investigation of resiliency, or sociocultural protective factors early in life that may buffer mothers and children from the detrimental impacts of external stressors (Huang, Lee, & Lu, 2007; Panter-brick & Eggerman, 2012). For example, variation in mothers’ perceptions of stress may be partially influenced by cultural resources (Dressier, Balieiro, Ribeiro, & dos Santos, 2016; Koolhaas, de Boer, & Buwalda, 2006). Dressier et al. (2007) theorizes that culture shapes the meaning attributed to stressors and thus promotes individual variation in maternal coping behaviors through group norms, practices and criteria of collective social esteem and accomplishment. Sociocultural protective factors may buffer the intergenerational transmission of stress and confer positive adaptation, or resilience, among mothers and their offspring (Ager, Stark, Akesson, & Boothby, 2010; Dressier, Balieiro, Ribeiro, & Santos, 2007; Panter-brick & Eggerman, 2012). Identifying the factors across diverse cultural groups that exacerbate or promote resiliency to stress response can facilitate development of interventions to reduce the detrimental impact of maternal stress on child development.
This study integrates qualitative and quantitative methods to explore associations and pathways between maternal stress and early childhood obesity. Quantitative data analyses were conducted using the Growing up in New Zealand (GUiNZ) data set. The GUiNZ study is a
3


prospective longitudinal cohort study that began in 2009 with the recruitment of 6,822 pregnant women in Auckland, New Zealand. This sample represents 11% of all infants bom in New Zealand during the study period (Morton et al., 2013). This study is unique due to its capacity to provide a comprehensive picture of contemporary child development over time for children bom in New Zealand, and for its inclusion of significant numbers of ethnic minorities. Qualitative data were collected from a diverse, convenience sample of women recruited through Plunket Centers in Auckland, New Zealand. Plunket is a national not-for-profit organization, community-owned and governed and is the largest provider of free support services for the development, health and wellbeing of children under five in New Zealand. The organization provides services for more than 90% of newborns in New Zealand each year (Plunket, 2018). The mixed-methods approach used in this study allows for triangulation of data to better understand the complexities surrounding maternal stress experiences, early childhood obesity, and mediators and moderators that may explain these associations in diverse contexts.
1.2 Problem Statement
Maternal stress is detrimental to maternal and child health outcomes. However, research exploring the impacts of maternal stress is limited by the use of interchangeable objective and subjective measures (Kingston et al., 2012; Liu et al., 2016a). A lack of prospective longitudinal studies that allow for the analysis of critical periods of stress exposure and transitions in stress exposure from the prenatal period through the postnatal period has also made it difficult to understand these associations (Collins & Manolio, 2007). The unique population of New Zealand is a particularly useful sociocultural context in which to explore these associations. The Maori (indigenous people of New Zealand) and Pacifika (non- Maori people of Polynesian descent) communities are exposed to external risks, such as poverty, unemployment, and overcrowding,
4


that are two to three times higher than for other ethnic groups and experience the highest rates of early childhood obesity compared to European and Asian families in New Zealand (Perry 2015; Turner and Lloyd 2004; Ministry of Health 2012). This diverse setting allows for exploration of associations between maternal stress and early childhood obesity within and between ethnic groups. Additional qualitative methods are needed to explore variations in individual responses to stress throughout early development and risk and protective factors that may explain pathways linking maternal stress and early childhood obesity (Shonkoff, Boyce, & McEwen, 2009).
1.3 Research Objective
The primary objective of this study was to use quantitative and qualitative methods to explore associations and hypothesized pathways between pre- and post-natal maternal stress and early childhood obesity among an ethnically diverse and representative sample of New Zealand mothers.
1.4 Aims and Hypotheses
The following aims were investigated to further explore the potential impacts of maternal stress exposure on early childhood obesity in a New Zealand context:
Aim 1: Determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy.
Hypothesis 1: Objective and subjective dimensions of stress will be moderately correlated within the study sample.
Aim 2: Analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24- and 54-months of age.
Hypothesis 2: Objective and subjective measures of prenatal stress are independently associated with childhood BMI at 24- and 54- months after controlling for covariates.
5


Aim 3: Examine associations between the timing and duration of maternal stress exposure during the pre-and post-natal periods of development and early childhood BMI at 54-months of age.
Hypothesis 3: Cumulative exposure to maternal stress is more strongly positively associated with early childhood BMI at 54- months of age compared to stress exposure during solely the pre- or post-natal period.
The DOHaD model proposes that maternal stress during the prenatal period, regardless of postnatal experiences, is positively associated with BMI. Therefore, we predict high BMI in children experiencing prenatal stress regardless of postnatal environment.
The Life Course Epidemiology model proposes that accumulation of maternal stress from the prenatal period through the postnatal period of development is positively associated with BMI.
Therefore, BMI is predicted to be higher in children exposed to cumulative maternal stress.
Figure 1.1 DOHaD and Life Course Epidemiology models of early life stress and BMI at 54-months
Aim 4: Analyze risk and protective factors that may mediate or moderate associations between prenatal stress and early childhood BMI at 54-months of age using structural equation modeling.
Hypothesis 4a: Risk factors, including length of exclusive breastfeeding, maternal eating behaviors, and maternal activity levels, will mediate associations between prenatal stress and early childhood BMI.
Hypothesis 4b: Protective Factors, including external support, family support, cultural identity, neighborhood integration, and household cohesiveness will moderate associations between prenatal stress and early childhood BMI.
DOHaD
Life Course Epidemiology
Birth 2 years 5 years Period of stress exposure
Birth 2 years 5 years Period of stress exposure
------ High prenatal stress, high postnatal stress
------ High prenatal stress, low postnatal stress
...... Low prenatal stress, high postnatal stress
-----Low prenatal stress, low postnatal stress
6


Aim 5: Explore qualitatively the lived experience of stress and associations with childhood BMI among a diverse group of New Zealand mothers.
7


CHAPTER 2
THEORETICAL AND EMPIRICAL BACKGROUND
2.1 What is Stress?
Perspectives on human growth and development are shifting from a nature versus nurture dualism to a biocultural perspective requiring consideration of biology, environmental experiences, and cultural interactions. The complex interactions between biology (nature) and experience (nurture) collectively impact human development (Kuzawa and Sweet 2009). Among research exploring developmental outcomes associated with early life adversity, the influences of environmental experiences, or external stressors, on the physical body and individual variation in responses to these stressors are becoming well recognized (Lock 2013; McDade 2002). The experience of stress can loosely be defined as a non-specific response of the body to any demand for change (Selye, 1955). Stress represents a state in which homeostasis, or internal balance, is actually threatened or perceived to be threatened (Johnson, Kamilaris, Chrousos, & Gold, 1992). It is well recognized that the definition and experience of stress represents a multifaceted system and varies by individual.
2.2 Measures of Maternal Stress
Stress can be conceptualized as biological, environmental, and psychological (Sheldon Cohen, Kessler, & Underwood, 1995). There is currently a lack of a widespread understanding of the relationship between these differing dimensions. Measuring biological stress captures the activation of specific physiological systems and is most often measured through the stress hormone cortisol. Defining stress in an environmental context accounts for external events or experiences that are objectively associated with physical stress. Psychosocial stress primarily captures individuals’ abilities to cope with demands and unique responses, including subjective perceptions, of events (Sheldon Cohen et al., 1995). Stress research is complex due to these
8


interconnected but differing facets of the human stress response.
A common limitation of prior research exploring maternal stress is the use of a single measure of stress. The majority of studies to date have relied on self-report retrospective measures of stress (Wadhwa, Entringer, Buss, & Lu, 2011). These measures have numerous biases and are often neither valid or reliable (Entringer, Buss, & Wadhwa, 2015). Additionally, measures of objective stress (i.e., objective stressors) and subjective responses to those stressors (i.e., perceived stress measures) capture different aspects of the human stress response and may not be highly correlated (Kingston et al., 2012; Laplante et al., 2008; Turner & Avison, 2003).
Objective stress measures include stressful event scales, specific risk factors such as exposure to a natural disaster, exposure to external objective stressors, such as poverty, and stress-related behaviors (Gundersen, Mahatmya, Garasky, & Lohman, 2011; Huang et al., 2007). The Number of Stressful Life Events Index is the most commonly used objective stressful life events scale and asks participants about experiences of stressful life events in the past 12 months. The outcome is measured via a checklist of items including, “A close family member was very sick and had to go to the hospital,” and “I got separated or divorced from my husband or partner” (Cohen et al., 1995).
Alternative studies have operationalized stress using specific objective risk factors including bereavement and stress-related behaviors. Huang, Lee, and Lu (2007) measured stress defined as maternal bereavement related to recent death of a loved one and utilized measures of stressful behaviors (e.g., drinking, smoking) to explore impacts on physical health outcomes. Gunderson et al. (2011) quantified stress as independent risk factors including divorce, chronic physical health conditions and domestic violence. These objective measures are useful because they collect information about external events that are easy to quantify and require simple
9


measurement procedures (Cohen, Kamarck, & Mermelstein, 1983). However, risk factors used to create composite stress indices vary significantly and are often minimally correlated, making generalizability a challenge (Dole, 2003).
Although women may experience similar stressful events during pregnancy, secondary appraisal, or assessing ones coping options and available resources, impacts sources and perceptions of stress (Lazarus & Folkman, 1987) and women vary in their subjective responses to external stressors (Koolhaas et al., 2006). Studies have found subjective perceptions of stress were related to increased risk of preterm birth, birth weight, and cortisol levels in offspring (Dole, 2003; Rondo et al., 2003; Wadhwa, Sandman, & Garite, 2001). The Perceived Stress Scale (PSS) is often used to measure these subjective responses in an attempt to better explain individual variation in stress experience (Cohen et al., 1983). The PSS consists of 10 questions measured with a 5-item likert scale and has been shown to have high reliability in a variety of populations (Cohen et al., 1983). Dole et al. (2003) found that increased subjective perceptions of stressful life events were related to increased risk of preterm birth. An alternative study used self-report of mood as a marker for stress during pregnancy and found positive significant associations between subjective reports of stress and both birth weight and premature birth (Copper et al., 1996). Wadhwa et al. (2001) found strong correlations between subjective stress and cortisol levels suggesting that measuring subjective perception of stressors may accurately inform biological measures of stress. There is a need to provide clarity on the specific aspects and domains of maternal stress that may be particularly important with respect to child health outcomes (Wadhwa et al., 2011b). In order to best understand the intergenerational impacts of maternal stress on child health outcomes it may therefore be helpful to explore multiple measures of stress.
10


2.3 Intergenerational Transmission of Stress
Prenatal stress exposure is associated with poor birth outcomes, including preterm birth (Dole, 2003; Staneva, Bogossian, Pritchard, & Wittkowski, 2015) and low birth weight (Sable & Wilkinson, 2000). Both pre- and post-natal stress exposure impact postnatal health outcomes, including poor cognitive ability and a variety of affective disorders (Talge, Neal, & Glover, 2007), behavioral problems (Zijlmans, Riksen-Walraven, & de Weerth, 2015), poor academic performance (Pearson et al., 2016), and elevated BMI in their offspring (Cameron & Demerath, 2002; Farewell, Thayer, Puma, & Morton, 2018; Farewell, Thayer, Tracer, & Morton, 2018; Koch, Sepa, & Ludvigsson, 2008; Liu et al., 2016a; Wu et al., 2017).
Elevated BMI in childhood is a particularly critical outcome of interest due to the developmental origins of behavioral and biological obesity risk factors (Entringer et al., 2010). Additionally, BMI at age five predicts subsequent negative developmental outcomes including diabetes, heart disease and stroke in later life (Biro & Wien, 2010; Young-Hyman, Schlundt, Herman, De Luca, & Counts, 2001). Early life exposure to maternal stress may impact development of early childhood obesity through complex and intertwined physiological and behavioral pathways (Barker, 2007; Dong et al., 2004; Gillman et al., 2006).
Health and disease susceptibility is determined by the dynamic interplay between genes and the environment particularly during early critical periods of development (Entringer et al., 2015). Although stress experienced during the prenatal and early postnatal period may not cause early childhood obesity, it may predispose children to experiencing elevated BMIs throughout childhood and into adulthood. Two primary pathways have been proposed; 1) high levels of maternal stress may trigger the offspring’s own biological response to stress leading to chronically high cortisol levels (Gundersen, Lohman, Garasky, Stewart, & Eisenmann, 2008)
11


which in turn has shown to be linked with elevated levels of early childhood BMI (Marniemi et al., 2002) and 2) maternal stress may lead to maladaptive coping behaviors which impact healthy eating and physical activity behaviors of the child (Laessle, Uhl, & Lindel, 2001). These biological and behavioral pathways often interact in ways that increase obesity risk and have lifelong impacts on child health and development (Mikkila, Rasanen, Raitakari, Pietinen, & Viikari, 2005; Nader et al., 2006).
2.4 Maternal Stress and Early Childhood Obesity through a DOHaD Lens
DOHaD research suggests that the stress experienced during the perinatal period of development is critical for laying the foundations for child growth and development (Ben-Shlomo, 2002; Gluckman & Hanson, 2004). DOHaD literature proposes ways in which environmental experiences can impact human biology and development (Holliday, 2006). This theory has been developed over the last 25 years and was initially derived from the “Barker hypothesis” (Barker et al., 1993). The Barker Hypothesis underscored the genetic response to the physical environment and the origination of disease in sensitive periods of developmental plasticity, primarily in utero (Barker, 2007). Longitudinal studies have found that beginning in utero, determinants of energy imbalance and physiological responses to external stressors can impact later adoption of obesogenic behaviors (Bauer & Boyce, 2004). Li et al. (2010) found that elevated prenatal stress was associated with higher childhood BMI, although this relationship was not significant until the age of 10. Another study exploring prenatal programming of early childhood obesity found that stress-induced behaviors in pregnancy, i.e. malnutrition and smoking, were significantly positively associated with child obesity at age 5 (Huang et al., 2007). Finally, two additional studies found that prenatal exposure to objective hardship, defined as exposure to a natural disaster, was significantly associated with childhood BMI at age 5 (K.N. et
12


al., 2012; Liu, Dancause, Elgbeili, Laplante, & King, 2016b).
While the physiological mechanisms linking prenatal stress to childhood BMI development are becoming more clear, the potential effects of post-natal behavioral factors also need to be considered (Entringer et al., 2010).
2.5 Maternal Stress and Early Childhood Obesity through a Life Course Epidemiology
Lens
The DOHaD model has been expanded to incorporate environmental exposures accumulating over a longer period of development, from the prenatal period through to adulthood (Bogin, 1999). Contextualizing the cumulative impact of stress on child health outcomes allows for an analysis of individual’s lives that encompasses biological, environmental, and behavioral influences. The interplay of environmental risk and protective factors influence patterns of human biology throughout one’s lifetime and mounting evidence points to the “long arm of childhood” (Umberson, Crosnoe, & Reczek, 2010). Accumulation of stress exposures during early life translates into disease trajectories into adulthood (Katz, Sprang, & Cooke, 2012). Life Course Epidemiology incorporates DOHaD principles but reinforces Lazarus and Folkman’s (1987) view of stress as transactional and multidimensional. Stress is neither solely an internal nor external state.
Life course epidemiology builds off the DOHaD hypothesis by contending that stress exposure from the prenatal period throughout the life course impacts biological, behavioral, and psychosocial processes that contribute to adult health and disease risk. Stress early in life predisposes individuals to be more vulnerable to stress exposures throughout the life course, which leads to allostatic loading of cumulative stress (Nederhof & Schmidt, 2012). Allostatic load results from the presence of excessive stress and vulnerability over an individual’s life
13


course or the inefficient operation of this stress hormone response system(McEwen & Gianaros, 2010). Cumulative exposure to maternal stress hormones has been linked to high levels of child cortisol (Gundersen et al., 2008) and high levels of cortisol are associated with elevated BMI in early childhood (Abraham, Rubino, Sinaii, Ramsey, & Nieman, 2013).
Cross-sectional studies have found significant associations between concurrent maternal stress and early childhood obesity, suggesting that maternal stress during the preschool years is correlated with early childhood BMI (Parks et al., 2012; Walton et al., 2014). However, the cross-sectional nature of these studies limits their findings. Using a longitudinal design, one study found that cumulative stress among mothers with children 5 to 10 years of age resulted in an increase in childhood BMI over time, highlighting the cumulative impact of stress (Shankardass et al., 2014). An additional study supporting the life course epidemiology framework discovered cumulative stress exposure early in life was associated with concurrent childhood obesity and subsequent obesity into adulthood (Evans, Fuller-Rowell, & Doan, 2012). These studies expand past research focused on perinatal effects by accounting for the additive impact of post-natal stress experiences.
Pathways of association between early exposure to maternal stress and early childhood obesity have been researched although the findings are still limited. Exploring how stress translates into obesity requires a development and life course approach since stress can impact both biological and behavioral responses that are associated with early childhood obesity development (Hackman, Farah, & Meaney, 2010; Shonkoff et al., 2012). In addition, testing the timing and duration of maternal stress exposures is necessary to best understand these relationships (Buss et al., 2007).
14


2.6 Proposed Pathways through which Maternal Stress may influence Early Childhood
Obesity
Maternal stress may impact early childhood obesity through both biological and behavioral pathways. These hypothesized pathways are presented in Figure 2.1.
Figure 2.1 Proposed Biological and Behavioral Pathways between Pre- and Post-natal Maternal Stress and Early Childhood Obesity
2.6.1 Biological Pathways
Prenatal stress may impact childhood obesity through various biological pathways.
Stress leads to prolonged activation of physiological systems which increases the risk for development of physical and physiological diseases. The physiological stress response has evolved to provide necessary energy for fight or flight situations. However, this system can cause damaging health effects in the long term when chronically activated. The hypothalamus-pituitary-adrenal (HPA) axis is a central biological stress-regulation system that is responsible
15


for producing cortisol, the primary hormone released in response to stress (Bjomtorp, 2001; Incollingo Rodriguez et al., 2015). When stress is acute, feedback loops signal the HPA axis to stop the production of cortisol. Under periods of prolonged stress, there is initial excessive cortisol production followed by dysregulated cortisol levels (Pervanidou & Chrousos, 2016). Specifically among preschool aged children, chronic stress exposure was found to be associated with blunted cortisol levels, emotional overeating, and higher BMI (Miller, Clifford, et al.,
2013). Cortisol may also directly influence appetite and cravings by modulating other hormones and stress responsive factors that stimulate appetite (Epel, Lapidus, McEwen, & Brownell, 2001; Lumeng et al., 2014).
The 41-amino acid hypothalamic peptide, Corticotropin-Releasing Hormone (CRH) is the main regulator of HPA axis activity during stress. Prenatal stress may impact fetal growth and later obesity risk through overexposure to CRH in pregnancy. This hormone plays a key role throughout pregnancy and levels naturally rise as pregnancy progresses. However, experiencing stress during pregnancy leads to elevated levels of CRH (Sandman et al., 1994). Higher CRH levels during pregnancy are associated with poor birth outcomes, including preterm birth, which is associated with later development of early childhood obesity (Stout, Espel, Sandman, Glynn, & Davis, 2015). Gillman et al., (2006) conducted a prospective cohort study with pregnant women and found that maternal CRH levels during pregnancy were positively associated with BMI among their three-year old children. These findings suggest CRH levels may explain some of the biological pathways between exposure to maternal stress during pregnancy and early childhood obesity.
CRH during pregnancy stimulates cortisol levels as well as glucose levels. Consistently high blood glucose levels combined with insulin suppression leads to cells that are starved of
16


glucose due to de-sensitization of glucose receptors (A. Harris & Seckl, 2011). The energy requirements of these cells may send hunger signals to the brain, thereby resulting in overeating and maternal weight gain. Excessive gestational weight gain is associated with early childhood obesity (Fraser et al., 2013).
Exposure to stress throughout early life may also disrupt leptin levels, a hormone that is secreted by fat cells. Leptin supports energy regulation by balancing hunger and energy expenditure. Higher levels of leptin are associated with increased risk for early childhood and adult obesity (Miller, Lumeng, et al., 2013). When stress exposure causes a disruption in the HPA axis, leptin cannot reach the brain and suppress hunger hormones. This can cause higher levels of leptin secretion, leading to higher fat storage, and increased levels of hunger hormones which may lead to overeating (Jeanrenaud & Rohner-Jeanrenaud, 2001; Miller, Lumeng, et al., 2013). Leptin is also secreted in response to stress (Tomiyama et al., 2012). One retrospective study found that higher adversity scores during childhood were associated with higher leptin levels (Tomiyama et al., 2012). The specific mechanisms linking early life stress, leptin responses, and childhood obesity are still unclear.
Chronic stress exposure throughout the first five years of life can negatively impact the developing brain of a child (Shonkoff et al., 2012). Specifically, the prefrontal cortex may be impacted by stress. The prefrontal cortex is responsible for executive functions including the regulation of behavior and restraint (Blair et al., 2011; Blair & Raver, 2012). Elevated levels of stress results in decreased activity of the prefrontal cortex which can impact a child’s ability to react and adapt appropriately to stressful situations. Exposure to stress early in life may also alter the development and functioning of regions of the brain that are responsible for reward systems including the intake of food which may promote increased fat and sugar intake (Dillon et al., 2009;
17


Hanson et al., 2016; Mehta et al., 2010). Disruptions in these reward systems can result in poor eating behaviors, leading to the onset of obesity (Shonkoff JP, 2009).
One prospective, longitudinal study that explored associations between children exposed to chronic stressors throughout early childhood found that such children had lower self-regulation abilities and higher BMI from 9 to 13 years of age (Evans et al., 2012). Another study found that among a group of 1500 fourth grade children, increased levels of executive functioning skills were positively associated with healthy food consumption, specifically fruit and vegetable intake, and negatively associated with intake of unhealthy food items (Riggs, Spruijt-Metz, Chou, & Pentz,
2012). Pieper and Laugero (2013) explored associations between executive functioning skills and calories consumed among a group of preschool-aged children and found that these variables were inversely related; lower executive functioning skills were associated with a higher number of total calories consumed. Finally, two additional studies found that poor self-regulation among a group of toddler-aged children was associated with being overweight (Graziano, Calkins, & Keane, 2010; Miller, Rosenblum, Retzloff, & Lumeng, 2016).
2.6.2 Behavioral Pathways
Prenatal stress is positively associated with unhealthy eating behaviors and inactivity during pregnancy (Lobel et al., 2008a). Oken et al. (2007) found that greater weight gain throughout pregnancy was associated with higher child BMI scores at three years of age. A longitudinal study of 4,234 mothers found that gestational weight gain was positively associated with BMI at all ages, from birth to 42 years of age (Schack-Nielsen, Michaelsen, Gamborg, Mortensen, & Sorensen, 2010). These findings suggest that healthy eating and physical activity behaviors during pregnancy may significantly impact the BMI of offspring throughout early childhood.
Perinatal maternal stress is also associated with increased unhealthy eating behaviors and sugar sweetened beverage consumption among young children (Adam & Epel, 2007). Poor
18


maternal mental health is associated with a lower likelihood of being present or involved in meals (McCurdy, Gorman, Kisler, & Metallinos-Katsaras, 2014), a higher likelihood of serving their children sugary drinks, a higher likelihood of eating outside the home, a lower likelihood of modeling healthy eating behaviors, and a lower likelihood of setting limits on consumption (Elias et al., 2016; Hughes, Power, Liu, Sharp, & Nicklas, 2015).
Exposure to stress early in life may also be correlated with unhealthy food environments and access to healthy foods which impact eating behaviors. Leung et al. (2014) found that low-income preschool children who were living in stressful and chaotic home environments were at greater risk for obesogenic eating behaviors. An alternative study found that among 4,320 school-aged children, higher levels of self-reported stress in the home were associated with lower consumption of fruits and vegetables, and a higher consumption of high fat foods (Cartwright et al., 2003). Bowman et al., (2004) explored associations between the numbers of stressors experienced by mothers and fast food intake of their preschool-aged children and found significantly positive correlations. One study found that low-income mothers living in stressful environments do not have the resources or access to purchase fruits and vegetables. Additionally, maternal concerns about food waste if children have not been exposed to the novel fruits and vegetables leads to the purchasing of energy-dense and nutrient-poor foods (Daniel, 2016).
Perinatal stress is also associated with decreased levels of physical activity and increased sedentary levels among young children. The built environment may play a significant role (Ding & Gebel, 2012; Sallis, Floyd, Rodriguez, & Saelens, 2012). Mothers experiencing high stress environments often live in low income neighborhoods which are less likely to have parks and recreation areas where children participate in physical activity (Ding & Gebel, 2012). Families living in high poverty areas are also exposed to higher rates of crime and unsafe conditions, thus minimizing opportunities for outdoor play (Ding & Gebel, 2012). A primary risk factor related to
19


maternal stress is overcrowding which also is inversely associated with active play indoors (Gary W. Evans & English, 2002). Self-report of high maternal stress was found to be associated with increased levels of physical inactivity and fewer limits on screen time among a group of 110 mothers and their preschool-aged children (Walton et al., 2014). A systematic literature review of 168 studies utilizing varied measures of stress exposure found that both objective and subjective measures of stress were associated with reduced physical activity across diverse ages and ethnic groups (Stults-Kolehmainen & Sinha, 2014).
Eligh maternal stress exposure may also be associated with an earlier introduction of complimentary foods compared to mothers living in lower stress environments. This is correlated with shorter periods of breastfeeding. Li et al. (2008) found that experiencing stressful life events during pregnancy increased the odds for early cessation of breastfeeding independent of maternal socio-demographic and biomedical factors. Shorter durations of breastfeeding are also associated with increased risk for development of early childhood obesity above and beyond associations between maternal BMI and early childhood obesity (Li et al., 2005). Therefore, length of exclusive breastfeeding duration may also partially explain associations between perinatal maternal stress and early childhood obesity.
Finally, studies show that families reporting high levels of stress sleep less (El-Sheikh, Buckhalt, Mize, & Acebo, 2006; El-Sheikh, Kelly, Buckhalt, & Benjamin Hinnant, 2010;
Mezick et al., 2008). This may be associated with overcrowding, increased screen time, a lack of limits and bedtime routines, and anxiety and worry. Decreased sleep duration is associated with unhealthy eating behaviors and lack of physical activity, thus increasing the risk for developing early childhood obesity (Cappuccio et al., 2008; Dev, McBride, Fiese, Jones, & Cho, 2013).
Risk behaviors, including eating behaviors, physical activity, breastfeeding, and sleep, are associated with perinatal maternal stress and early childhood obesity and therefore may
20


mediate these pathways. Exploring the direct and indirect effects of these factors can help to identify behavioral intervention opportunities to prevent the onset of early childhood obesity.
2.7 Lazarus’s Theory of Stress and Coping Perspectives on biomedical health in the western world are shifting from Cartesian dualism to a transactional understanding of health and disease. The French philosopher Rene Descartes argued that the workings of the soul, or the mind, are unimportant to consider when evaluating health and disease (Lovallo, 2015). This Cartesian perspective is based on the dualism between mind and body and supports the biomedical model of disease defining health as a property of the physical body. There has been movement away from this model, emphasizing instead an ecological approach. The theory of transactionism stipulates that interactions between individuals and their environments shape the human stress response (Lazarus & Folkman, 1987). Transactions and the interplay between external and internal demands and stressors can impact maternal and child health. Isolating different measures and experiences of stress may yield the most accurate understanding of how stress impacts the physical body. To explore these associations, stress-provoking factors (e.g., external risk factors), stress-mediating or moderating factors (e.g. social support) and stress-resulting factors (e.g., perceived stress) were investigated independently using mixed-methods in this study.
2.8 Protective Factors and Resilience
The lived experience of stress is rooted in meanings, interpretations, activities and interactions which are shaped by accessibility of social and cultural resources (Desjarlais &
Jason Throop, 2011). The impacts of maternal stress on child health outcomes cannot be separated from sociocultural contexts. These sociocultural contexts can foster resilience and inform external protective coping factors that buffer the negative impacts of early life stress on
21


child health and development (Dressier, Oths, Balieiro, Ribeiro, & Dos Santos, 2012; Shonkoff et al., 2012).
Social support is an example of a protective coping factor that may buffer the association between maternal stress and child health outcomes. The social buffering hypothesis contends that social support, defined as the perceived availability of interpersonal resources as well as the degree of integration into a network, is positively associated with decreased stress and increased well-being (Sheldon Cohen & Wills, 1985). Past studies have found that social support significantly modifies the association between pre- and post-natal stress and maternal mental and physical health as well as child cortisol levels (Cohen, S., Gottlieb, B. H., & Underwood, 2004; Horton & Wallander, 2001; Thoits, 1995, 2011; Wadhwa et al., 2001). However, it has been hypothesized that different types of social support, such as instrumental, emotional, and informational, as well as varying sources of social support, including family, friends, and partners, may differentially moderate perceptions of stress and resulting health outcomes (Thoits, 1995).
Cultural identity, or the perception of cohesion, commonality, and belongingness with other group members, may also increase individuals’ capability to positively cope with external stressors (Cameron, 2004; Clauss-Ehlers, Yang, & Wan-Chun Chen, 2006; Dressier, Balieiro, Ribeiro, & Santos, 2007). The theoretical construct of “cultural consonance” refers to the degree to which individuals’ behaviors and beliefs align with shared cultural models across five domains: lifestyle, social support, family life, national identity and food ((Dressier et al., 2007). Higher levels of cultural consonance, when measured as two latent variables across all domains, has been found to be associated with lower levels of psychological stress (Dressier et al., 2007) and with lower risk of obesity (Dressier et al., 2012). Dressier et al. (2005) found that low
22


cultural consonance is often a stressor. Because cultural norms and beliefs are so widely accepted within societal groups, an inability to incorporate or adapt these behaviors into one’s own life is innately stressful. Elevated exposure to stressors coupled with low cultural consonance may result in unhealthy risk behaviors leading to greater food intake and/or consumption of a high fat diet which negatively impacts both mothers and their young children (Dressier, Oths, Ribeiro, Balieiro, & Dos Santos, 2008).
Finally, neighborhood integration and cohesion may decrease stress and promote positive coping. Lower levels of neighborhood cohesion are associated with reduced social support and an increased occurrence of negative life events and neighborhood disorder; neighborhood social disorder is associated with poor maternal mental and physical health (M. Franco, J. Pottick, & Huang, 2010). Low neighborhood cohesion with disadvantaged communities has been shown to compound the effect of maternal stress and depression (Kohen, Leventhal, et al., 2008).
Increased social contact with neighbors has also been found to reduce fear in low-income neighborhoods (Kruger, Reischl, & Gee, 2007). However, the protective role of neighborhood integration may vary by community. One study found that lower frequency of interaction with neighborhoods in high stressed communities may actually provide a protective effect on mental health (Dupere & Perkins, 2007)
Factors that promote positive coping can promote resilience, which may explain variations in the lived experience of stress between diverse groups. Panter-Brick and Eggerman (2012) contend that resilience is a function of both biological and behavioral attributes as well as programmed responses to environmental factors. Coping factors may confound the impacts of maternal stress on development of early childhood obesity (Koolhaas et al., 2006). Exploration of the sociocultural context of stress will inform external factors that can be mobilized to prevent
23


the negative impacts of early life stress on child development (Shonkoff et al., 2012). These protective factors may buffer the intergenerational transmission of stress and confer resilience among mothers and their offspring (Ager et al., 2010).
2.9 Maternal Stress and Childhood Obesity in a New Zealand Maternal Stress
The Maternal Health study found that among a group of 1507 moms in New Zealand and Australia, 7.3% of women reported anxiety occasionally or often during pregnancy, 15.7% in the first 3-months postpartum, 10.9% at 6-months and 8.5% at 9-months postpartum (S. J. Brown, Lumley, McDonald, & Krastev, 2006). An additional study found that among a group of 6703 mothers in New Zealand and Australia, 10.7% of women experienced anxiety during their first antenatal clinic visit and 9.1% reported experiencing anxiety 6-months after birth (Najman et al., 2005). New Zealand mothers reported that being a mother in itself is inherently stressful; a longitudinal birth cohort study found that as New Zealand mothers have more children, they experience significant increases in stress and reductions in life satisfaction (Boden, Fergusson, & John Horwood, 2007). New Zealand and Australian moms reported severe and chronic stress after giving birth was related to acute stressful life events, lack of social support and neonatal risk. These factors were also associated with a higher rate of major depression (Hammen, Kim, Eberhart, & Brennan, 2009). Researchers have identified objective stressors that are most prominent among New Zealand mothers. In the Maternal Health Study, mothers who experienced exposure to high external stressors were under the age of 24, unmarried, receiving government assistance, less educated compared to the average population, and not receiving paid employment during pregnancy, thus ineligible to receive paid maternity leave. Mothers who reported experiencing physical violence were also more likely to report high stress (Gao et al.,
24


2010). A systematic review of eight longitudinal studies found that high rates of stress among New Zealand and Australian moms were associated with low socio-economic status, low social support, poor partner relationship and unwanted pregnancy (Schmied et al., 2013).
Ethnicity and Stress
More than 200 ethnic groups reside in Auckland, the largest city in New Zealand (MacPherson et al. 2011). The four majority ethnic groups include Europeans, Asians, Maori (indigenous people of New Zealand) and Pacifika (non- Maori people of Polynesian descent). Maori and Pacifika mothers experience significantly higher levels of stress compared to Europeans and Asians in New Zealand (R. Harris et al., 2006).
Ethnic discrimination during the pre- and post-natal periods is associated with numerous poor maternal and child health outcomes including hypertension, low self-reported health, increased health risk behaviors, adverse birth outcomes and childhood obesity (Dixon et al., 2012; R. Harris et al., 2012, 2006; Thayer & Kuzawa, 2015). Ethnic and racial discrimination may contribute to higher stress exposure in minority populations. Harris et al. (2012) found that the reported lifetime discrimination prevalence among the New Zealand population varied among ethnic groups; 35% among Asians, 29.5% among Maori, 23% among non-Maori Polynesians, and 13.5% among Europeans. One study found that Maori and Pacifika families experience significantly more ethnic discrimination than New Zealand Europeans and that Maori were almost 10 times more likely to experience multiple types of discrimination when compared to NZ Europeans (Harris et al., 2006). Another study found that women who reported experiencing ethnic discrimination in pregnancy had poorer self-rated health and gave birth to infants with higher stress reactivity at six weeks of age, which is associated with poor health and development outcomes (Thayer & Kuzawa, 2015). However, potentially because of a small
25


sample size, this study found no differences in these relationships by ethnic group.
Objective contextual factors may also explain higher stress levels among Maori and Pacifika families. Maori families experience high rates of mental health disorders with 19.4% of Maori reporting anxiety. Families with the lowest income and education levels are at greatest risk for mental health disorders (Baxter et al., 2006). Minority ethnic groups in New Zealand are exposed to high rates of external risks, such as poverty and living in high deprivation neighborhoods (Perry, 2015). Sixty five percent (65%) of Maori and 78% of Pacifika people live in the most deprived neighborhoods and these groups tend to experience objective stressors two to three times higher than other ethnic groups (New Zealand Living Standard Survey, 2008). The Pacific Islander Family Study found that among a sample of 1590 Pacific Islander mothers, 76% reported experiencing verbal aggression and 23% reported physical violence within six weeks after giving birth. Twenty four months after giving birth, 86% reported having experienced verbal aggression and 27% reported having experienced physical violence (Gao et al., 2010). Living in high deprivation neighborhoods across all ethnic groups is associated with higher cortisol levels during pregnancy and higher offspring cortisol levels after birth (Thayer & Kuzawa, 2014).
Early Childhood Obesity
Twelve percent (12%) of New Zealand children between the ages of 2 and 14 are obese and a total of 31% of children are overweight or obese (Ministry of Health, 2017). The prevalence of early childhood overweight and obesity in New Zealand is higher in higher deprivation neighborhoods throughout the country. Children living in the most deprived areas are 2.5 times more likely to be obese compared to children living in the least deprived areas after adjusting for age, sex and ethnicity (Ministry of Health, 2017). Ethnic group differences in the
26


prevalence of overweight and obesity also exist. The Maori and Pacifika populations in New Zealand experience significantly higher rates of early childhood obesity compared to the general population. Among children aged three to five, 18% of Maori children are obese and 29% of Pacifika children are obese (Statistics New Zealand & Ministry of Pacific Island Affairs, 2011).
2.10 Literature Review Summary
Measuring maternal stress is a challenge due to the multiple dimensions of stress and interplay of genes and experience that impact the biological stress response. Use of both objective and subjective measures can provide insight into the impacts of objective stressors and perceptions of stress on maternal and child health. Due to the intergenerational impact of maternal stress and detrimental impacts on child health and development, multiple measures are needed. Maternal stress during the pre- and post-natal period impact child health, and specifically development of early childhood obesity. However, analyses of the specific timing and duration of stress exposures during these critical periods on early childhood BMI is unclear. The DOHad model contends that the prenatal and early post-natal periods are the most critical. The Life Course Epidemiology model proposes that cumulative maternal stress exposure from the prenatal period throughout the first five years of life and beyond impacts child health outcomes more than a specific critical period. Both models identify potential behavioral and biological pathways that may explain associations between early exposure to maternal stress and childhood obesity. Lazarus’s Theory of Stress and Coping emphasizes the need for consideration of stress-provoking, stress-mediating or moderating, and stress- resulting factors in stress research. Past studies have found that risk behaviors, including unhealthy eating, lack of physical activity, lack of exclusive breastfeeding, and lack of sleep, may mediate associations between maternal stress and early childhood obesity. Protective factors, such as social support,
27


cultural identity, and neighborhood integration and cohesiveness, may also moderate these associations to confer resilience. These relationships are particularly relevant in New Zealand due to the high rates of maternal stress and early childhood obesity, socioeconomic and ethnic disparities, and diverse cultural setting which allows for the exploration of mediating and moderating factors that may explain individual variation in maternal stress responses and early childhood obesity.
28


CHAPTER 3
QUANTITATIVE METHODS
3.1 Study Design Overview Two-Phase Simultaneous Explanatory Design
The data for this study were collected in multiple waves. I explored my first three aims
from February 2017 to August 2017 by conducting secondary data analyses with the Growing Up in New Zealand (GUiNZ) data set. The final two aims were explored simultaneously from March 2018 to November 2018. First, I prepared for the qualitative data collection phase of my study by submitting my ethics application to the University of Colorado Institutional Review Board and the local ethics review board in New Zealand (Plunket Ethics). I developed drafts of my qualitative protocols as well as a sampling plan and recruitment strategy. Concurrently, I explored the hypothesized mediation and moderation models with the GUiNZ data set to see if risk behaviors and protective factors were significantly related to prenatal stress and early childhood BMI at 54-months. I analyzed my final path analyses models once the qualitative data collection phase was completed. Findings from the qualitative data were used to understand variations in my final path analyses model by ethnicity and generate potential explanations for the observed differences. I then disseminated my findings via publications, presentations, and community cafe events. A timeline of all activities is displayed in Table 3.1.
29


Table 3.1. Mixed- Methods Research Timeline (Implementation Time Frame: February 2017- March 2019)
Goals and Activities F ‘17 M A M J J A s 0 N D J ‘18 F M A M J J A s 0 N D J ‘19 F M
Meet with GUiNZ team; prepare data X X
Quantitative Analyses-Hl, H2, H3 X X X X X X X
Ethics Approval X X
Finalize Qualitative Tools X X X
Quantitative Analyses-H4 X X X X X X X X X
Recruit Qualitative Participants X X
Conduct Focus Groups X X
Conduct Interviews X X
Qualitative Analyses- H5 X X X X X X X X X
Disseminate Findings X X X X X X X
30


3.2 Quantitative Participant Selection
The GUiNZ study is based out of the Center for Longitudinal Research- He Ara Ki Mua at the University of Auckland. This longitudinal study began in 2009 with the recruitment of 6,822 pregnant women in the last 12 weeks of pregnancy in the Manukau and Waikato District Health Board Regions in Auckland, New Zealand and is designed to continue until the children are 21 years of age. The children bom into this study reprszXesent 11% of all infants bom in New Zealand during the study period and the ethnicity and socio-demographic characteristics are generalizable to those of children being bom in New Zealand today (Morton et al., 2014). Approximately 24% of the sample identified as Maori, 21% as Pacifika, 16% as Asian, and 66% as European or Other.
3.3 Quantitative Data Collection Methods
A variety of data collection methods were used in the GUiNZ study, including face-to-
face Computer Assisted Personal Interviews (CAPI), direct observations, developmental and anthropometric assessments, telephone interviews, and routine linkages to clinical records from the three District Health Boards. Data were collected from the mothers and children during the prenatal period, 9-months, 24-months, and 54-months after birth and were related to child’s health and wellbeing, whanau (family) life, education, psychological development, neighborhood and environment, and culture and identity. For my quantitative analyses, GUiNZ data sets from four data collection waves, prenatal, 9-months, 24-months, and 54-months, were merged using the unique child identifier based on non-missing data for the childhood BMI at 54-months variable (n=5,839). Table 3.2 displays characteristics of the quantitative sample.
31


Table 3.2 Demographic Characteristics of the Quantitative Sample (n=5,839)
Age of Child
Antenatal 24-months 54-months
Variables m sd m sd m sd
Child ZBMI Score - - 0.84 1.27 0.78 1.12
Objective Stress Scores .91 1.38 0.83 1.29
Subjective Stress scores 12.98 6.35
Maternal Age 30.42 5.86
n %
Ethnicity (n=5692)
European 3336 58.6
Maori 741 13.0
Pacific 690 12.1
Asian 743 13.1
MELAA 98 1.7
Other 11 0.2
New Zealander Education (n=5695) 73 1.3
No sec school qualification 352 6.2
Sec school/NCEA 1-4 1267 22.2
Diploma/Trade cert/NCEA 5-6 1743 30.6
Bachelor’s degree 1388 24.4
Higher degree Household Income (n=4470) 945 16.6
<=20K 156 3.5
>20K <=30K 214 4.8
>3 OK <=50K 571 12.8
>5 OK <=70K 718 16.1
>70K <=100K 1062 23.8
>100K <=150K 1054 23.6
>150K 695 15.5
*Due to low attrition rates, sample characteristics remained fairly stable over time and were not collected at the 2-year data collection time point
32


3.4 Quantitative Data Measures
Table 3.3 displays the variables included in all quantitative analyses. Table 3.3 Variables included in Quantitative Analyses
Variable Type of Variable M/SD (continuous) or N/% (categorical)
Primary Independent and Dependent Variables
Childhood BMI z-scores (Dependent Variable) Continuous (IOTF Z-Scores) 24 months: Mean=.85, SD= 1.27 54 months: Mean= .78, SD=1.12
Subjective Stress (Independent Variable) Continuous (Perceived Stress Scale (PSS) Prenatal: Mean=12.99, SD=6.35
Objective Stress (Independent Variable) Continuous (Objective Vulnerability Scale) Prenatal: Mean=1.0, SD= 1.37 9 months: Mean=.90, SD=1.28 2 years: Mean=.91, SD= 1.27
Covariates
Parity (Cohort Child Order) Categorical l=first: n=2407, 42.2% 2=subsequent: n=3296, 57.8%
Maternal Race/Ethnicity Categorical l=European: n=3336, 58.6% 2=Maori: n=741, 13.0% 3=Pacific: n=690, 12.1% 4=Asian: n=743, 13.1% 5=Other; n=182, 3.2%
Maternal Age Continuous Mean=30.38, SD=5.86
Maternal Education Categorical l=no secondary school qualification: n=352, 6.2% 2=secondary school/NCEA 1-4: n=1267, 22.2% 3=diploma/trade certificate/NCEA 5-6: n=1743, 30.6% 4=bachelor’s degree: n=1388, 24.4% 5=higher degree: n=945, 16.6%
Total Household Income Categorical 1=< 20k: n=156, 3.5% 2=20k to <3Ok: n=214, 4.8% 3=3Ok to <5Ok: n=571, 12.8% 4=5Ok to <70k: n=718, 16.1% 5=70kto <100k: 1062,23.8% 6= 100k to <150k: n=1054, 23.6% 7=> 150k: n=695, 15.5%
Birthweight z-score Adjusted for Gestational Age Continuous Mean=.98, SD= 1.28
Pre-Pregnancy Maternal BMI Continuous Mean=25.5, SD=5.81

33


Table 3.2 cont’d Hypothesized Stress-Mediating Variables
Maternal Healthy Eating Behaviors Continuous (Aggregate of 4 dichotomous Variables: Meeting NZ MOH nutrition guideline for Fruit and Vegetable Consumption, Breads and Cereal, Milk and Dairy, and Meat) Mean=1.30, SD=1.02
Maternal Physical Activity Behaviors Continuous (Average number of days women participate in moderate to vigorous physical activity (MVPA) during their first and second trimester) Mean=1.61, SE=1.40
Maternal Exclusive Breastfeeding Continuous (Length of exclusive breastfeeding in months) Mean=5.08 SD=1.22
Hypothesized Stress-Moderating Variables
Family Support Continuous (Family Support Scale, Dunst et al., 1984) Mean=23.99, SD=5.38
External Support Continuous (sources of external (outside of family) support) Mean=22.96, SD=5.16
Neighborhood Integration Continuous (Neighborhood Integration Scale, Kavanagh, Turrell & Subramanian, 2006) Mean=34.53, SD=4.8
Cultural Identity Continuous (20 Questions related to ethnic identity, language, cultural identity and belonging, and perception of ethnic discrimination (Morton et al., 2010) Mean=12.54, SD=2.42
Household Cohesiveness Continuous (Country-specific scale of 9 questions related to family connectedness) Mean= 30.64, SD=4.19
3.4.1 Independent Variables
Objective Stress: Objective stress can be defined as stress-related vulnerability, or external factors that impact life stress and individual risk trajectories (McEwen & Gianaros, 2010). Environmental risk factors derived from the GUiNZ maternal vulnerability scale were used as the structural measure of stress and are presented in Table 3.3 (Morton et al., 2013). These measures were collected via self-report during the last trimester of pregnancy and 9-months and
34


24-months after birth.
Table 3.4 Objective Stressors included in GUiNZ Vulnerability Scale
Risk Factor Definition
Maternal Smoking Continuing to smoke after first trimester of
pregnancy or continuing to smoke regularly/every
_______________________day____________________________________________
Maternal Age___________Teenage mother at time of pregnancy____________
Relationship Status Mother with no current partner
Maternal Education Mother with no formal secondary school qualifications
Deprivation Area Living in NZDEP2006 area deciles 9 or 10
Unemployment Mother not on leave, actively seeking work but not currently working
Tenure-public rental Living in social housing
Income tested benefit In receipt of an income tested government benefit
Overcrowding Having 2 or more persons per bedroom
This is a population specific index of structural vulnerability measured during the prenatal period. Risk factors were dichotomized with a score of 0 or 1 depending on whether they were experienced or not and aggregated into a continuous index. These risk factors have been informed by international studies as well as pilot work to create a population-relevant vulnerability scale (Morton et al., 2014). They also are routinely available and consistently measured sources of life stress. The vulnerability scale used in this study extends previous research, much of which has utilized a single risk factor to quantify external stress, for example poverty or teen pregnancy. Due to the complexity of stress and multiple factors impacting stress, the scale used in this study comprising 9 structural risk factors is a more effective measure of vulnerability at the population level (Chittleborough, Lawlor, & Lynch, 2011; Tallman, 2016). This study recognizes stress as a transactional process (Lazarus & Folkman, 1987). Therefore, both real and perceived environmental demands will be assessed as independent stress measures.
35


Prenatal Perceived Stress Scale: Perceived stress during pregnancy was operationalized with the Cohen Perceived Stress Scale which is a validated tool and is designed to measure “the degree to which situations in one's life are appraised as stressful” (Cohen et al., 1983).The perceived stress scale consists of 10 questions measured with a 5-item likert scale. Examples of questions asked include, “In the last month, how often have you been upset because of something that happened unexpectedly?” and “In the last month, how often have you felt that you were unable to control the important things in your life?” The Perceived Stress Scale score for this sample ranged from 0 to 40. This measure was used as a single continuous variable in analyses.
Maternal age: Maternal age at pregnancy interview was included as a confounder in all analyses. Younger age at pregnancy has been found to be positively associated with high pre- and postnatal stress and depression levels (J. D. Brown, Harris, Woods, Buman, & Cox, 2012). Maternal age is coded as a continuous variable in all analyses (Mean=30.38, SD=5.86).
Parity: Parity, or cohort child order, has been found to be negatively associated with elevated levels of BMI in childhood (Ong, Preece, Emmett, Ahmed, & Dunger, 2002). Parity was coded as a dichotomous categorical variable in the GUiNZ data set and as such, was used as a dichotomous variable in all analyses (l=first: n=2407, 42.2%; 2=subsequent: n=3296, 57.8%). Race/Ethnicity: Race/ethnic variation in objective and subjective stress experiences has been found among pregnant women (Lu & Chen, 2004; Woods, Melville, Guo, Fan, & Gavin, 2010). Significant associations between race/ethnicity and childhood BMI suggest that Pacifika and Maori children experience the highest levels of BMI in New Zealand (Utter, Scragg, Schaaf, & Fitzgerald, 2006a). Race/ethnicity was coded as a categorical variable in all analyses (l=European: n=3336, 58.6%; 2=Maori: n=741, 13.0%; 3=Pacifika: n=690, 12.1%; 4=Asian: n=743, 5= Other: n=182).
36


Maternal Education: Populations with lower education levels are more likely to be exposed to stress and experience higher rates of childhood obesity compared to more educated and wealthier populations (Cohen & Janicki-Deverts, 2012). Maternal education was coded as a categorical variable in all analyses (l=no secondary school qualification: n=352, 6.2%; 2=secondary school/NCEA 1-4: n=1267, 22.2%; 3=diploma/trade certificate/NCEA 5-6: n=1743, 30.6%; 4=bachelor’s degree: n=1388, 24.4%; 5=higher degree: n=945, 16.6%).
Household Income: Populations with lower income levels are more likely to be exposed to stress and experience higher rates of childhood obesity compared to more educated and wealthier populations (Cohen & Janicki-Deverts, 2012). Household income was coded as a categorical variable in all analyses (1=< 20k:n=156, 3.5%; 2=20k to <30k: n=214, 4.8%; 3=30k to <50k: n=571, 12.8%; 4=50kto<70k: n=718, 16.1%; 5=70k to <100k: 1062, 23.8%; 6=100k to <150k: n=1054, 23.6%; 7=> 150k: n=695, 15.5%).
Birth weight (adjustedfor gestational age): Prenatal stress is associated with low birth weight in offspring (Staneva et al., 2015). Low birth weight also is positively associated with development of early childhood obesity (Danielzik, Pust, Landsberg, & Muller, 2005). Birth weight as coded as a continuous variable in all analyses (Mean=.98, SD= 1.28).
Maternal Pre-Pregnancy BMI\ Maternal Pre-Pregnancy BMI is associated with childhood BMI. Past studies suggest the maternal obesity during the prenatal period is associated with obesity in childhood (Benyshek, 2007; Levin, 2006). Maternal pre-pregnancy BMI was coded as a continuous variable in all analyses (Mean=25.5, SD=5.81).
3.4.2 Dependent Variable
Childhood Body Mass Index (BMI): BMI is calculated as weight in kilograms divided by height in meters squared collected and validated by two trained GUiNZ researchers during study visits
37


at 24-months and 54-months of age. The continuous measure of gender- and age-specific BMI z-scores is based on the International Obesity Task Force Growth Standards (Cole, 2000). BMI was coded as a continuous variable in all analyses (24 months: Mean=.85, SD= 1.27; 54 months: Mean= .78, SD=1.12).
3.4.3 Mediation and Moderation Variables
Maternal Healthy Eating Behaviors: This variable represents an aggregate of 4 dichotomous items related to meeting the New Zealand Ministry of Health’s nutrition guidelines. These guidelines for pregnant women in New Zealand contain recommendations focused on the daily intake of the four major food groups: vegetables and fruit; bread and cereals; milk and milk products; and lean meat, meat alternatives and eggs. The self-reported measure was assessed using a semi-quantitative, forty-four item food frequency questionnaire (FFQ). Previous studies have found that approximately 24% of pregnant women in the GUiNZ study did not meet the Ministry of Health recommendations for daily servings for any of the four main food groups (Morton et al., 2013). The continuous measure was collected prenatally (scale: 0-4;
Mean=1.30, SD=1.02).
Maternal Physical Activity Behaviors: This variable is a continuous variable operationalized as the average number of days women participate in moderate or vigorous physical activity (scale: 0-7 days; Mean=1.61, SE=1.40). Moderate activity was defined as intense walking and vigorous activity was defined as activities that noticeably increase one’s breathing and energy expenditure. The GUiNZ participants were asked about their usual physical activity levels during face-to-face interviews at pregnancy (Morton et al., 2013).
Exclusive Breastfeeding: This variable is measured continuously in the GUiNZ data set as the length of exclusive breastfeeding in months. Although the New Zealand Ministry of Health
38


recommends six months of exclusive breastfeeding, on average, mothers in the GUiNZ sample exclusively breastfed for five months (Morton et al., 2013). This variable was collected via face-to-face interviews at 9-months post-natally (scale: 0-9 months; Mean=5.08 SD=1.22).
Family Support: This is a continuous variable in the GUiNZ data set defined using the Family Support Scale (Dunst, Jenkins, Trivette, 1984). Support sources included partner, parents, partner’s parents, extended family, partner’s extended family and friends. This variable was collected prenatally and a higher score reflected higher expected helpfulness (scale: 0-36; Mean=30.64, SD=4.19).
External support: Parents were asked during the last trimester of pregnancy to report what sources of external (outside of family) support they expected to have available, and how helpful they expected each source to be, once their baby was bom. These support sources included family doctor, professionals (such as Plunket nurse or kaiawhina), early parenting programs (such as Parents as First Teachers), books and the internet. This variable was collected prenatally and a higher score reflected higher expected helpfulness (scale: 0-36; Mean=22.96, SD=5.16).
Neighborhood Integration: This is a continuous variable comprised of 13 questions about connections to one’s neighborhoods, commonalities with neighbors, and respect (Neighborhood Integration Scale, Kavanagh, Turrell & Subramanian, 2006). This variable was collected prenatally and a higher score reflected higher neighborhood integration (scale: 0-40; Mean=34.53, SD=4.8).
Cultural Identity. This is a continuous variable comprised of 20 questions related to ethnic identity, language, cultural identity and belonging, and perception of ethnic discrimination (Morton et al., 2010). This variable was collected prenatally and a higher score reflected higher
39


cultural identity (scale: 0-20; Mean=12.54, SD=2.42).
Household Cohesiveness: This is a continuous variable comprised of 9 questions related to family connectedness and interaction (Morton et al., 2010). This variable was collected prenatally and a higher score reflects higher family cohesiveness (scale: 9-36, Mean=30.64, SD=4.19).
40


CHAPTER 4
QUANTITATIVE ANALYSES AND FINDINGS Section 4.1 Cleaning, Diagnosing, and Treating the Data
Independent and dependent variables of interest in the Growing Up in New Zealand (GUiNZ) data set were screened to explore linear regression assumptions. First, multivariate normality was explored. Multivariate normality refers to the normality of the error terms related to the independent and dependent variables. Violations of normality may bias the parameter estimates and standard errors. However, normality is a larger concern among small sample sizes and, therefore, is not as important an assumption when running linear regression models with the GUiNZ data set (n=5,839). Histograms and normal p-p plots of regression standardized residuals were analyzed using SPSS software version 25 (IBM SPSS Inc., 2012). The residuals of the variables of interest did not violate the assumption of multivariate normality.
Next, the assumption of homoscedasticity, or constant variance, was evaluated. The disturbance term should not differ for varying values of the independent variable. If heteroscedasticity exists, parameter estimates may be inefficient and observed standard errors may be biased which could lead to inaccurate conclusions about significant associations between variables. Homoscedasticity in the GUiNZ data set was evaluated by running a scatterplot of residuals to ensure that there is equal variability across the x-axis from the bottom to the top of the y-axis. The scatterplot revealed that the data met this assumption.
Since the assumptions of normality and homoscedasticity were met, the assumption of linearity was also met. Multi-collinearity between all variables included in the regression models was evaluated. When independent variables are too highly correlated it may lead to instability of the model coefficients. The standard errors of the regression coefficients were reviewed to ensure
41


they weren’t too large which would suggest multi-collinearity may be an issue. Correlations between all variables of interest were analyzed to ensure none were considered high, or above .6. Next, the tolerance and variance inflation factor (VIF) scores for each independent variable was reviewed. The VIF measures the impact of collinearity among the variables in a regression model and is equal to 1/tolerance. Cohen et al (2003) suggests that a problem exists if the tolerance values are less than .1 or the variance inflation factors are above 10. All variables of interest met these criteria. It was concluded that no problems related to multi-collinearity exist in the GUiNZ data set.
To explore the presence of extreme cases in my variables of interest, distance and leverage which dictate the influence of extreme cases were explored. Distance refers to the distance of estimates from the regression line. These outliers have an extreme case on the outcome variable. Plotting the studentized residuals, or standardized residuals, can reveal if the distance of any data points are a concern. Leverage refers to the distance from the mean of X.
To assess the overall influence of extreme cases, distance is multiplied by the leverage. Cooks D, which is the most common measure of influence, was used to identify any maximum values that may exceed Cooks D in this data set (Cooks D=.014). The Cooks D values were plotted along the x-axis in a scatterplot with the ID’s along the y-axis. There were three points that were far from zero and exceeded the maximum Cooks D Value. I explored these outliers and determined that they were extreme potentially due to data entry or measurement error. I ran my basic regression model with and without these three outliers; the coefficients did not change.
Therefore, I dropped these three outliers from the data set for analyses.
Due to the large sample size represented in the GUiNZ data set, there was a large percentage of missing on specific variables on interest. Missing data ranged from .2% to 17.6%
42


across all variables of interest. However, only 1.8% of data on the primary outcome measure (BMI z-score at 54-months of age) was missing. Of the 5,839 maternal-child dyads included in the analyses, 3,363 cases (57.6%) had complete data for all variables. Data was missing at random for the primary dependent and independent variables included in this analysis. Since more than 5% of data was missing on key variables, multiple imputation was used to create imputed values for missing data across all variables (White, Royston, & Wood, 2011). The imputation model included all the variables in the multivariate analysis model. The observed cases and the imputed cases were analyzed separately to compare results before and after imputation. Findings were consistent and are displayed in Table 4.1. Pooled data was used for all regression analyses.
43


Table 4.1 Descriptive Statistics before and after Multiple Imputation
Before Imputation After Imputation
Variables N M SD N M SD
Predictor Variables
Objective Stress-Prenatal 5130 0.91 1.38 5839 0.94 1.39
Objective Stress-9 months 5139 0.83 1.28 5839 0.86 1.30
Objective Stress-24 months 5244 0.83 1.29 5839 0.85 1.30
Subjective Stress 5189 12.98 6.35 5839 12.97 6.37
Outcome Variables
Childhood BMI 24-months 4814 0.84 1.27 5839 0.85 1.27
Childhood BMI 54- months 5732 0.78 1.12 5839 0.78 1.12
Stress Mediator Variables
Maternal Healthy Eating Behaviors 5189 1.30 1.02 5839 1.31 1.02
Days Moderately to Physically Active-Prenatal 5302 1.61 1.40 5839 1.62 1.39
Exclusive Breastfeeding 5373 5.88 2.15 5839 5.89 2.14
Stress Moderator Variables
Family Support- Prenatal 5189 23.99 5.38 5839 24.03 5.36
External Support- Prenatal 5189 22.96 5.16 5839 23.00 5.16
Family Cohesiveness-Prenatal 5702 30.64 4.19 5839 30.63 4.20
Cultural Identity- Prenatal 5653 12.54 2.42 5839 12.52 2.42
Neighborhood Integration-Prenatal 5185 34.53 4.89 5839 34.53 4.89
Covariates
Maternal Age 5709 30.42 5.86 5839 30.41 5.85
Parity 5702 1.58 0.49 5839 1.58 0.49
Race/Ethnicity 5692 1.93 1.37 5839 1.93 1.37
Birthweight Adjusted for Gestational Age 5827 0.98 1.29 5839 0.98 1.29
Maternal BMI 5081 25.32 5.78 5839 25.53 5.82
Maternal Education 5695 2.23 1.15 5839 2.23 1.15
Household Income 4470 4.85 1.58 5839 4.71 1.63
Due to the large sample size of this data set (n=5839), too much statistical power was a concern. Sufficient statistical power is critical to ensure that one avoids making a Type I error (rejecting a true null hypothesis) or Type II error (failing to reject a false null hypothesis). The sample size, significance criterion, and population effect size are involved in determining statistical power. I was careful about interpreting significance and set the alpha to .01 because of
44


this large sample size. I also was careful to ensure findings were practically significant, in addition to statistically significance.
Section 4.2 Aim #1
Aim 1. Determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy.
Analyses
Bivariate correlations between all variables were analyzed using SPSS software version 25 (IBM SPSS Inc., 2012).
Results
Bivariate analyses were conducted between prenatal objective stress and prenatal subjective stress. All bivariate correlates are displayed in Table 4.2. Correlations between prenatal subjective stress and objective stress were low but significant and positive (r = .27,
P< 01).
45


Table 4.2 Correlations between all continuous variables
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1. BMI 24-Months -
2. BMI 54-Months .48** -
3. Objective Stress_AM 77** -
4. Objective Stress_9Mos j 1** .22** .83** -
5. Objective Stress_24Mos j j** 7j** .81** .86** -
6. Perceived Stress .04** .08** 77## 79## -
7. Highest Education -.08** 17** - 47** - 47** . 47** 17** -
S. Household Income -.03* 13** _ 49*# -.46** -.46** _ 71 ** -.38** -
9. Birthvvdght Z-Sc ore 19** 73** -.05** -.05** -.04* .03 -.01 .01 -
10. Maternal BMI .16** .30** 20** .20** .18** 17** 19#* 13** .16** -
11. Nutrition Guid dines .02 -.07** . 13 ** -.12** 17*# .03 -.05** .Os** .03 -.01 -
12. Average Exerdse Days .03 .02 .04* .03* .05** -.06** .01 -.03* -.02 -.05** .08“ -
13. Exdusive Breastfeeding -.04** -.04** -.09** -.11“ -.11** -.08** .10** .02 .04** -.09** .02 .05** -
14. External Support .02 .00 .01 .01 .00 .00 -.04** .04* -.06** .04** .04** .00 -.06** -
15. Family Support .00 -.01 -.15** -.14** -.14** -.12** .07** .17** -.01 .01 .03* .02 .02 .29** -
16. Family Cohesiveness -.02 .02 .00 -.02 -.02 -.12** .01 .01 .00 .02 .08** .04** .01 .12** .31** -
17. Neighbourhood Integration .01 -.01 -.12** -.12** -.11** -.14** .05** .07** .02 -.02 .05** .08** .04** .05“ .11** .13“ -
18. Cultural Identity7 .02 .03* .02 .00 .00 .03 .12** -.01 .00 .05** .06** 07#* .03* .09** .09** .09** .07**
**0.01 level * 0.05 level
46


Section 4.3 Aim #2
Aim 2: Analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24- and 54-months of age.
Analyses
Bivariate correlations were conducted between prenatal objective and subjective stress variables and childhood BMI at 24- and 54-months. Greater prenatal subjective stress was significantly and positively correlated with greater BMI at 24-months although the correlations were very low (r = .04) and 54-months of age (r = .08) (both p<01). Greater prenatal objective stress was significantly and positively associated with greater BMI at 24-months (r = .11) and 54-months (r = .22) (both p<01). Multivariate regression models were used to explore associations between the primary independent variables (prenatal objective stress and prenatal subjective stress) and primary dependent variables (BMI z-scores at 24-months and 54-months) using SPSS software version 25 (IBM SPSS Inc., 2012). First, hierarchical linear regression models were used to explore associations between objective stress and BMI z-scores at 24-months of age. Covariates were entered in Block 1. Next, the objective stress variable was entered in Block 2. The model building process was then repeated with the subjective stress variable entered in Block 2. The only difference between the objective stress and subjective stress regression models was the inclusion of additional covariates, including maternal education, income, and age, in the subjective stress models. These demographic variables are part of the objective stress variable and are, therefore, not controlled for in these models. Both sets of models were re-run with BMI z-scores at 54-months of age as the primary outcome variable. Coefficients, standard errors, p-values, R2and adjusted R2 values are reported.
Results
Table 4.3 shows R2 and adjusted R2 values, coefficients, standard errors, t-statistics and
47


p-values for hierarchical linear regression models with BMI z-scores at 24-months of age as the primary outcome variable. First, associations between objective prenatal stress and BMI at 24-months of age were analyzed. Covariates were entered first into block 1 and the objective stress variable was added in Block 2. Exposure to one additional risk factor during pregnancy was significantly associated with a .06 increase in BMI z-score at 24-months of age (t-statistic = 5.45; p<01). Next, the model was run with subjective stress as the primary independent variable. Subjective stress was not significantly associated with BMI z-scores at 24-months of age after controlling for covariates (t-statistic = -1.43; p = 0.15).
Table 4.3 Hierarchical Linear Regression Models: BMI at 24-months of age
Outcome: BMI Z-scores at 24-Months of Age
R2 Adj R2 p B SE B T Statistic P-Value
Model 1 (Confounded) .07 .07 <001
Parity -0.07 0.04 -1.91 .06
Ethnicity
Maori v. European 0.12 0.06 1.99 .05*
Pacifka v. European 0.31 0.07 4.51 .00**
Asian v. European -0.31 -0.09 -5.82 .00**
Other v. European 0.03 0.00 0.29 .78
Pre-Pregnancy BMI 0.02 0.00 5.78 .00**
Birthweight (z-score) Adj 0.16 0.01 10.84 .00**
Model 2 (Full Model) .08 .08 <001
Parity -0.08 0.04 -2.00 .05*
Ethnicity
Maori v. European 0.02 0.07 0.34 .77
Pacifka v. European 0.28 0.08 3.75 .00**
Asian v. European -0.33 0.06 -5.90 .00**
Other v. European -0.05 0.10 -0.47 .64
Pre-Pregnancy BMI 0.02 0.00 5.41 .00**
Birthweight (z-score) Adj 0.16 0.02 10.43 .00**
Objective Stress 0.06 0.01 5.45 .00**
Table 4.4 displays R2 and adjusted R2 values, coefficients, standard errors, t-statistics and p-values for hierarchical linear regression models with BMI z-scores at 54-months of age as the
48


primary outcome variable. The model building process was repeated beginning with analyzing objective stress as the primary predictor. In the full model, 17.1% of the variation in BMI z-scores at age 54-months is accounted for and an increase in exposure to one risk factor during pregnancy is significantly associated with a .06 increase in BMI z-score at age 54-months (t=5.95, p<01). Subjective stress was not significantly associated with BMI z-scores at 54-months of age (t=-l .59, p=. 11).
Table 4.4 Hierarchical Linear Regression Models: BMI at 54-months of age____________________
Outcome: BMI Z-scores at 54-Months of Age
R2 Adj R2 P B SE B T Statistic P-Value
Model 1 (Confounders) .17 .16 <001
Parity -0.05 0.03 -1.81 .07
Ethnicity
Maori v. European 0.34 0.05 7.31 .00**
Pacifka v. European 0.58 0.05 11.00 .00**
Asian v. European -0.27 0.04 -6.34 .00**
Other v. European 0.11 0.08 1.43 .15
Pre-Pregnancy BMI 0.04 0.00 12.80 .00**
Birthweight (z-score) Adj 0.15 0.01 12.14 .00**
Model 2 (Full Model) .17 .17 <001
Parity -.05 0.03 -1.58 .11
Ethnicity
Maori v. European 0.23 0.05 4.45 .00**
Pacifka v. European 0.49 0.06 8.13 .00**
Asian v. European -0.31 0.05 -6.84 .00**
Other v. European 0.09 0.08 1.14 .26
Pre-Pregnancy BMI 0.04 0.00 12.80 .00**
Birthweight (z-score) Adj 0.15 0.01 12.14 .00**
Objective Stress 0.06 0.01 5.95 .00**
49


Section 4.4 Aim #3
Aim 3. Examine associations between the timing and duration of maternal stress exposure during the pre-and post-natal periods of development and early childhood BMI at 54-months of age. Analyses
To explore the timing and duration of objective stress exposure, the objective stress measure was dichotomized into no exposure to stress or exposure to stress (No Stress = 0 risk factors; Stress = 1 or more risk factors). Just over half of the sample (51.9%) experienced no stress during the prenatal period and the 24-month period of data collection. There were no other natural cut-points in the number of risk factors mothers experienced from the prenatal period to 24-months. Stably low stress was defined as mothers experiencing no stress during the prenatal and 24- month data collection period. Stably high stress was defined as mothers experiencing stress during the prenatal and 24-month data collection period. Increased stress was defined as mothers experiencing no stress during the prenatal period but experiencing stress during the 24-month data collection period. Decreased stress was defined as mothers experiencing high stress during the prenatal period and no stress during the 24-month data collection period. The stably low group was used as the reference group in all analyses. A one-way between subjects ANCOVA was conducted to compare the effect of stress on childhood BMI among stably low stress, stably high stress, increased stress, and decreased stress conditions, controlling for covariates. Since statistically significant differences were found, post hoc contrast tests were conducted. All analyses were conducted using SPSS software version 25 (IBM SPSS Inc., 2012). Results
Figure 4.1 depicts the average childhood BMI at 54-months of age among the four maternal stress transition groups: stably low stress (n = 3928), increased stress (n = 299),
50


decreased stress (n = 438) and stably high stress (n = 1174). The average childhood BMI z-scores were 0.61 (n=3928), 0.97 (n=299), 0.98 (n=438), and 1.28 (n=1174) standard deviations above the IOTF reference mean for the stably low, increased stress, decreased stress, and stably high stress groups, respectively.
Maternal Stress Groups and Childhood BMI z-scores at 54-months of Age
Maternal Stress Transition Groups
Error Bars: 95% Cl
Figure 4.1 Maternal stress from prenatal to 24-months of age and average childhood BMI at 54-months of age
Table 4.5 highlights the significant main effect of maternal stress on childhood BMI at 54-months of age among all four stress groups after controlling for covariates [F(3, 5833) =53.3, p < 0.01], Sidak corrected post hoc comparison analyses are presented in Table 4.6. These planned contrasts revealed that mothers experiencing cumulative high stress, p < 0.01, 95% Cl [1.04, 1.17], increased stress, p < 0.01, 95% Cl [0.76, 0.98], or decreased stress p < 0.01, 95% Cl [0.76, 1.03], from the pre- to post-natal period is significantly associated with higher early
51


childhood BMI at 54-months of age compared to mothers who experience stably low stress. Mothers experiencing transitions in stress exposure from the pre- to the post-natal period, regardless of direction, is significantly associated with lower early childhood BMI at 54-months of age compared to mothers who experience stably high stress (p < 0.01). However, the mean early childhood BMI score for the increased stress group did not differ from the mean childhood BMI score for the decreased stress group (p= 1.0).
Table 4.5 Differences between Maternal Stress Groups and Childhood BMI at 54-months of age after controlling for covariates
Sum of Squares df Mean Square F
Between Groups 175.95 3 58.65 53.5**
Within Groups 6412.85 5833 1.10
Total 7315.1 5839
**p < 0.01
Table 4.6 Sidak Corrected Post Hoc Comparisons for Childhood BMI at 54-months of age among four Maternal Stress Groups
(I) Transition Group CO Transition Group Mean Diff (I-J) Std. Error 95% Confidence Interval Lower Upper Bound Bound
Stably Low Increased -.18* .06 -.35 -.02
Decreased _ 2i** .05 -.35 -.07
Stably High _ 44** .04 -.54 -.35
Increased Stably Low .18* .06 .02 .35
Decreased -.03 .08 -.24 .18
Stably High . 26** .07 -.44 -.08
Decreased Stably Low .21** .05 .07 .35
Increased .03 .08 -.18 .24
Stably High _ 23** .06 -.39 -.08
Stably High Stably Low 44** .04 .35 .54
Increased .26** .07 .08 .44
Decreased 23** .06 .08 .39
*p<05 **p<01
52


Section 4.5 Aim #4
Aim 4. Analyze risk behaviors and protective factors that may mediate or moderate associations between maternal stress and early childhood BMI at 54-months of age using structural equation modeling.
Analyses
Structural equation modeling (SEM) was used to examine relationships between the objective stress measure, risk behaviors and protective factors, and childhood BMI at 54-months. SEM is a unique method to explore the complex relationships between pre- and post-natal stress measures and childhood BMI for multiple reasons including: 1) it allows for analyses of multiple dependent variables; 2) it permits variables to be examined as independent and dependent variables; and 3) it is beneficial for testing both moderation and mediation models (Bowen & Guo, 2011). There are four key steps involved in running SEM analyses. These include: 1. Specification - form explicit hypothesis using regression and factor analysis concepts to form structural restrictions, 2. Estimation - use a SEM software package to estimate coefficients and standard errors and various statistical indicators, 3. Evaluation - compare alternative structural restrictions in a series of statistical tests, and 4. Re-Evaluation - reconsider alternative model and suggest ways to deal with new concepts (R. Kline, 2015).
Path analysis, which is a special case of SEM where all variables are measured, was used to explore relationships between the measured variables. The overall goal of path analysis is to estimate causal verse non-causal aspects of observed correlations between variables. Analyses convey how the hypothesized model accounts for the data-observed correlations (i.e. standardized variables) or covariances (i.e. unstandardized variables). Path analysis is an extension of multiple regression whereby several regression relationships can be estimated
53


simultaneously (e.g., a variable that is an outcome of one variable may be simultaneously examined as a predictor of another). Assumptions of path analyses models include: (a) variables are measured without error; (b) measures are at the interval level; (c) residuals are normally distributed with zero mean and constant variance; (d) residuals are uncorrelated with one another and with the predictor variables in the equation in which each residual appears; (e) relationships among variables are unidirectional, thereby ruling out reciprocal relationships and feedback loops; and (f) relationships among variables are additive and linear (P. Kline, 2014).
Path analyses were conducted using SPSS Version 25 for descriptive and correlational analyses and Mplus Version 7.2 (Muthen and Muthen. 1998) for path model estimation. Prior to model estimation, the zero-order correlations among all model constructs were examined. Model relationships on early childhood BMI were estimated. Models are evaluated in terms of how well the model, as a whole, fit the data, and in terms of the significance of each of the specific proposed relationships. This method provides a way to examine both direct and indirect predictors of early childhood BMI. Figure 4.2 depicts the estimated model, where prenatal objective stress was estimated as a direct predictor of early childhood BMI at 54-months of age and an indirect predictor of early childhood BMI at 54-months of age through unhealthy risk behaviors, including unhealthy eating and inactivity during pregnancy, and shorter periods of exclusive breastfeeding. Protective factors including social support, family support, neighborhood integration, family cohesiveness, and cultural identity were estimated as moderators in the model and predicted to moderate the direct effect between prenatal stress and early childhood BMI. In other words, associations between prenatal stress and early childhood BMI were hypothesized to vary based on the level of protective factors mothers experienced during pregnancy. Mediation and moderation models were explored with each of these variables
54


analyzed independently, as well as in aggregate form (e.g. aggregate risk variable, aggregate protective variable).
Non-hierarchical models based on a priori hypotheses were evaluated in terms of the magnitude, direction, and significance of the estimated path coefficients by several standard measures of overall model fit: a chi-square p-value > 0.05 for the difference between the theoretical and the empirical model, comparative fit index (CFI) > .90, root mean square error of approximation (RMSEA) < 0.08 and standardized root mean square residual (SRMR) <.08. Missing data was addressed using multiple imputation techniques as described above.
Figure 4.2 Hypothesized Path Analysis Model of Prenatal Stress, Childhood BMI at 54-months, and mediators and moderators
Results
All path coefficients are standardized for ease of interpretation of the magnitude of the effect. First, the mediation model was explored. Each of the mediating variables was included in the model. Figure 4.3 depicts the mediation model.
55


Figure 4.3 Mediation Model of Prenatal Stress, Childhood BMI at 54-months, and mediators
Figure 4.3 Mediation Model
The number of nutrition guidelines met during pregnancy and length of exclusive breastfeeding significantly mediated the relationship between prenatal stress and early childhood BMI at 54 months (both p<01); however, the number of days mothers exercised during pregnancy did not mediate this relationship. Therefore, the exercise variable was dropped from the model and only the nutrition and breastfeeding variables were included as mediators.
To explore the direct moderation of the hypothesized protective variables (moderation of direct path between prenatal stress and early childhood BMI), moderation models were run with all five variables measured independently and subsequently as one aggregate variable. The model with protective factors as an aggregate variable had better model fit (AIC=67057.90, AIC= 50505, respectively). Based on these findings, the protective variables were included as one aggregated protective variable for all path analyses.
Next, this direct moderation model was compared to a model with both direct and indirect moderation to see which model had better fit. This model was based on the a priori hypothesis that protective factors may mitigate the significant association between objective stress and unhealthy eating behaviors and lower levels of exclusive breastfeeding in addition to the direct pathways between objective stress and early childhood BMI at 54-months. The model
56


with both direct and indirect moderation exhibited the best fit (AIC=50497.21). Since the model was just-identified, additional fit statistics were not available. A just-identified, or saturated model, is an identified model in which the number of free parameters exactly equals the number of known values; the model has zero degrees of freedom. To further explore the model fit, the pathway between protective factors and early childhood BMI was constrained. This path was chosen because the relationship was not significant and close to zero (fi =.01). The resulting path analysis model is displayed in Figure 4.4.
Figure 4.4 Overall Path Model
This model most accurately fit the data (j2 (1) = .826, p=.3634; AIC= 50496.03; CFI=
1.0; RMSEA = .00, 95% CI=0.00, 0.03; SRMS= .002). Prenatal stress was positively associated with early childhood BMI at 54 months of age (/? =.23, p<001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy (/? =-.14, p<001) and length of exclusive breastfeeding (/? =-.08, p<001). The number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding were negatively associated with early childhood BMI at 54 months (fi = -.04, /? = -.03, both p<001). There was a significant, but low,
57


positive correlation between the two mediating variables (ft =.04, p<001). Protective factors, including social support, family support, family cohesiveness, neighborhood integration and cultural identity, did not moderate the direct pathway between prenatal stress and early childhood BMI (ft = .02, p=.10). The protective variable was also positively and significantly associated with the number of nutrition guidelines met during pregnancy (ft = .09, p<001) which suggests this mediation pathway is moderated by protective factors. Protective factors may moderate the mediated effects of prenatal stress on early childhood BMI transmitted primarily through nutrition behaviors during pregnancy. The protective variable was not associated with length of exclusive breastfeeding in this model. This suggests that although protective factors may have an impact on nutrition behaviors health, they do not moderate the relationship between prenatal objective stress and early childhood BMI at 54-months of age in this overall model.
This final model was explored with European mothers (n=3336), Maori mothers (n=741), Pacifika mothers (n=690), and Asian mothers (n=743) in New Zealand to understand variations between prenatal stress and early childhood BMI by ethnicity.
European Mothers
First, the model was explored among mothers who identified as European (n=3336). Among Europeans, the protective path between protective factors and early childhood BMI was similarly not significant and close to zero so this path was constrained to display model fit statistics. This model accurately fit this subset of the sample (j2 (1) = 1.08, p=.2997; AIC= 34180.35; CFI= .999; RMSEA = .005, 95% CI=0.00, 0.05; SRMR= .004) and is displayed in Figure 4.5.
58


Figure 4.5 Final Path Analysis Model among European Mothers in New Zealand
Among European mothers, prenatal stress was positively associated with early childhood BMI at 54 months of age (fi =. 11, p<001). Nutrition behaviors and length of exclusive breastfeeding were not significant mediators in this model. However, prenatal stress was negatively associated with the length of exclusive breastfeeding (fi = -.09, p<001). There was a significant and positive correlation between the two mediating variables (fi =.08, p<001). Protective factors did not significantly moderate the direct pathway between prenatal stress and early childhood BMI. Protective factors were positively and significantly associated with the number of nutrition guidelines met during pregnancy (fi = .13, p<001).
Maori Mothers
Next, the model was explored within mothers who identified as Maori (n=741). Among Maori women, the protective path between protective factors and early childhood BMI was significant so this path was not constrained. The correlation between the mediators was not statistically significant and close to zero so this path was constrained. This model accurately fit
59


this subset of the sample (x2 (1) = .686, p= 4075; AIC= 8957.76; CFI= 1.00; RMSEA = .00, 95% CI=0.00, 0.08; SRMR= .007) and is displayed in Figure 4.6.
Figure 4.6 Final Path Analysis Model among Maori Mothers in New Zealand
Among Maori mothers, Prenatal stress was positively associated with early childhood BMI at 54 months of age (fi =.14, p<001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy (fi = -.14, p<001) and length of exclusive breastfeeding (fi = -.11, p<001). The length of exclusive breastfeeding was negatively associated with early childhood BMI at 54 months (fi = -.06, p<001), therefore the length of exclusive breastfeeding was a significant mediator between prenatal stress and early childhood BMI. The protective factor variable was negatively and significantly associated with the number of nutrition guidelines met during pregnancy (fi = .1, p<001). Protective factors were also significantly and positively associated with early childhood BMI (fl = .13, p<001). Protective factors do not significantly moderate the direct pathway between prenatal stress and early childhood BMI among Maori mothers.
60


Pacifika Mothers
The model was next explored within mothers who identified as Pacifika (n=690). Among Pacifika women, the protective path between protective factors and early childhood BMI was significant so this path was not constrained. The correlation path between the mediators was not statistically significant and close to zero so this path was constrained. This model accurately fit this subset of the sample (x2 (1) = .091, p= 7629; AIC= 7893.96; CFI= 1.00; RMSEA = .00, 95% CI=0.00, 0.07; SRMR= .003) and is displayed in Figure 4.7.
Figure 4.7 Final Path Analysis Model among Pacifika Mothers in New Zealand
Among Pacifika mothers, prenatal stress was positively associated with early childhood BMI at 54 months of age (fi =.08, p<001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy (fi = -.16, p<001) and length of exclusive breastfeeding (fi = -.08, p<001). The number of nutrition guidelines met during pregnancy was negatively associated with early childhood BMI at 54 months (fi = -.08, p<001), therefore, the number of nutrition guidelines met during pregnancy was a significant mediator between prenatal stress and early childhood BMI. The protective factor variable was also positively and
61


significantly associated with the length of exclusive breastfeeding (fi = .13, p<001). Protective factors were also significantly and positively associated with early childhood BMI (fi = . 13, p<001). Protective factors also significantly and negatively moderate the direct pathway between prenatal stress and early childhood BMI (/? = -.12, p<001) suggesting direct moderation exists between these variables among Pacifika mothers.
To explore how the total effects of prenatal stress on early childhood BMI vary at different levels of protective factors, the model was evaluated under low protective factors (-1 SD from the mean= -2.75), medium protective factors (mean=0), and high protective factors (1 SD from the mean=2.75) conditions. Results are displayed in Figure 4.8. Among Pacifika mothers experiencing low levels of protective factors, the slope between prenatal stress and early childhood BMI was . 12 (SE=.04, t =3.07, p=.002). Among Pacifika mothers experiencing average levels of protective factors, the slope between prenatal stress and early childhood BMI was .07 (SE=.03, t = 2.69, p=.007). Among Pacifika mothers experiencing high levels of protective factors, the slope between prenatal stress and early childhood BMI was .02 (SE=.03, t = .77, p=44).
Figure 4.8 Total Effects of Objective Prenatal Stress on Childhood BMI among Pacifika Women experiencing Low, Medium, and High levels of Protective Factors
62


Asian Mothers
Finally, the model was explored within mothers who identified as Asian (n=743). Among Asian women, the protective path between protective factors and early childhood BMI was significant so this path was not constrained. The correlation between the mediators was not statistically significant and close to zero so this path was constrained. This model accurately fit this subset of the sample (x2 (1) = .115, p= 7342; AIC= 8168.68; CFI= 1.00; RMSEA = .00, 95% CI=0.00, 0.07; SRMR= .003) and is displayed in Figure 4.9.
Figure 4.9 Final Path Analysis Model among Asian Mothers in New Zealand
Among Asian mothers, prenatal stress was not associated with early childhood BMI at 54-months of age. Nutrition behaviors and length of exclusive breastfeeding were not significant mediators in this model. However, prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy (fi = -.07, p<001) and the length of exclusive breastfeeding was positively associated with early childhood BMI at 54-months of age (fi = .07, p<001). Protective factors did not significantly moderate the direct pathway between prenatal stress and early childhood BMI. However, the protective variable was positively associated with
63


the number of nutrition guidelines met during pregnancy (fi = . 11, p<001).
Section 4.6 Summary of Quantitative Findings
Correlations between prenatal subjective stress and objective stress were low, but positive and statistically significant. Greater prenatal subjective and objective stress were significantly and positively correlated with greater BMI at 24-months and 54-months of age. However, after controlling for confounding variables, only objective stress was significantly associated with early childhood BMI at both time points. Mothers experiencing transitions in objective stress exposure from the pre- to the post-natal period, regardless of direction, was associated with statistically significantly lower early childhood BMI at 54-months of age when compared to mothers who experience stably high stress, and statistically significantly higher early childhood BMI at 54-months of age when compared to mothers who experienced no stress. However, mothers who experienced increased stress or decreased stress from the pre- to post-natal period did not differ from each other with respect to early childhood BMI suggesting that the direction of stress transition is not predictive of early childhood BMI at 54-months. Risk and protective pathways between objective prenatal stress and early childhood BMI at 54-months were explored. In the overall sample, the number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding significantly mediated associations between objective prenatal stress and early childhood BMI at 54-months; the number of days women exercised during pregnancy did not significantly mediate this relationship. Higher objective prenatal stress was significantly associated with a lower number of nutrition guidelines met and shorter length of exclusive breastfeeding. Higher levels of both these mediators were significantly associated with lower early childhood BMI at 54-months. Five protective factors were explored as moderators. A final path analysis model with an aggregate protective moderating variable and the two
64


significant mediators was fit to the data. The model with both direct and indirect moderation best fit the sample. Protective factors were positively and significantly associated with nutrition guidelines but were not significantly associated with breastfeeding behaviors or the direct path between prenatal stress and early childhood BMI at 54-months in the overall sample. However, this full model was explored among European mothers, Maori mothers, Pacifika mothers, and Asian mothers to understand differences in risk and protective pathways between prenatal stress and early childhood BMI and results varied by ethnic group. Higher levels of protective factors were positively and significantly associated with healthy eating behaviors among European and Asian mothers; this association was negative and significant among Maori and Pacifika mothers. Among Maori and Pacifika mothers, protective factors were significantly and positively associated with childhood BMI. The interaction between protective factors and stress was only significant among Pacifika mothers suggesting that protective factors mitigate the experience of maternal stress and buffer the direct pathway between maternal stress and early childhood BMI at 54-months for this ethnic group.
65


CHAPTER 5
QUALITATIVE METHODS
This chapter describes the qualitative methods utilized in this project. The qualitative phase began after completion of the first three quantitative aims: 1) to determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy, 2) to analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24- and 54-months of age, and 3) to examine associations between the timing and duration of maternal stress exposure during the pre-and post-natal periods of development and early childhood BMI at 54-months of age. Aim #4 (to analyze risk and protective factors that may mediate or moderate associations between prenatal stress and early childhood BMI at 54-months of age using structural equation modeling) and Aim #5 (to explore qualitatively the lived experience of stress and associations with childhood BMI among a diverse group of New Zealand mothers) were explored concurrently. This chapter presents findings related to Aim #5. The qualitative phase took approximately six months from development through data collection and analysis. Details of the methods, analyses and results follow.
5.1 Study Design Overview
The primary purpose of the qualitative phase is to expand on the understanding of the lived experience of maternal stress among ethnically diverse mothers in New Zealand and risk and protective factors that may explain associations between maternal stress and early childhood BMI. The joint use of qualitative and quantitative methods allows for the triangulation of data. The methods serve as a check on one another and allows a researcher to gain a clearer understanding of the issues and topics under investigation (Maxwell, 2013). Mixed-methods also
66


result in data related to different aspects of maternal stress and variations within and between ethnic groups. The current chapter will focus on the highlighted sections below.
QUAN Data Collection and Analyses
AIM 1 Quantitative Analyses - AIM 2 Quantitative Analyses - AIM 3 Quantitative Analyses AIM 4 Quantitative Analyses


QUAL Data Collection
AIM 5 QUAL Data Collection AIM 5 QUAL Data Analyses
and Analyses
Figure 5.1 Mixed-Methods Design|
5.2 Participant Selection
The targeted population for the qualitative component of this study was mothers who met the following criteria: 1) living in Auckland, New Zealand, 2) at least one child five years of age or younger, and 3) currently experiencing objective stress (e.g. living in highest deprivation quantile). Carpenter (1999) suggests that the sample size in phenomenological research should be small so that each experience can be examined in depth. The purpose is not necessarily to generalize the findings but to understand the complexity of meaning attached to stress.
Therefore, my goal was to recruit a purposive sample of mothers with young children who were experiencing at least one objective stressor and conduct a combination of focus groups and in-depth interviews. Additionally, I recruited an equal sample of mothers from the four primary ethnic groups in New Zealand: European, Maori, Pacifika, and Asian.
Recruitment
Recruitment occurred using a purposive, non-probabilistic sampling method. In February of 2018,1 began recruitment for focus groups. First, I compiled a list of high-deprivation suburbs in the Auckland region. The deprivation scores represent a single and simple index of
67


socioeconomic deprivation derived from national census data in New Zealand. The deprivation index is the most commonly used measure of an individual’s socioeconomic position in New Zealand (Salmond & Crampton, 2012). Living in a neighborhood that falls into the highest quintile of deprivation is a risk factor on the Growing Up in New Zealand (GUiNZ) objective stress scale. I then overlaid demographic information to target diverse communities with high numbers of Maori, Pacifika, European, and/or Asian families, respectively. I recruited from Plunket, which are family support and resource clinics, because approximately 90% of mothers with young children utilize some aspects of Plunket services in New Zealand (Plunket, 2018). I met with the Central Director of the Plunket offices, and we discussed opportunities to attend currently scheduled Plunket sessions, including Maori and Pacifika playgroups, coffee groups, and mother support groups comprised of mothers residing in these high deprivation and ethnically-diverse neighborhoods. Because the director felt that the topic of maternal stress aligned with the focus of many of these sessions, Plunket facilitators allowed me to attend some of their activities and conduct focus groups during and/or after their sessions.
One of the Plunket staff members introduced me to a community member who became my New Zealand consultant. She works with a faith-based organization which is a community-affiliate of Plunket. Her main roles were to review my focus group and in-depth interview protocols, provide feedback from a local perspective, and share her insights into the cultural and contextual factors that may be related to the maternal stress experience in Auckland. She also helped to recruit mothers for in-depth interviews in the neighborhoods identified above. We worked to recruit an equal number of participants from each of the primary ethnic groups.
5.3 Data Collection Methods
I collected qualitative data using both focus groups and in-depth interviews from
68


March 6th to April 20th, 2018. The objectives of the focus groups were to explore societal experiences of stress among mothers with young children residing in Auckland and to understand patterns of risk behaviors and protective factors that might explain the association between stress and early childhood obesity. Since stress is a broad topic, I felt that focus groups would encourage mothers to share their experiences and views on stress and coping behaviors in the context of their communities. I also collected data using in-depth interviews to explore individual lived experiences of maternal stress and variations in risk and protective behaviors. Stress can be a sensitive topic and mothers may feel more comfortable sharing their experiences in a one-on-one format. I found that these two methods were complementary and provided a unique understanding of maternal stress in this diverse community. I conducted focus groups and interviews within the same two-month period. I facilitated 7 one-hour focus groups with 46 mothers and 28 one-hour in-depth interviews in individual’s homes, community centers, parks, churches, coffee shops, and libraries. Data were collected from a total of 74 participants. On completion of the interview or focus group, participants received a $20 gift card. Ethical approval for this study was obtained from the Colorado Multi-Institutional Review Board in the United States and the Plunket Ethics Review Committee in New Zealand.
5.4 Data Collection Instruments
With the help of my advisors and my New Zealand consultant, focus group questions were adapted from Dressier et al. (2005) and Abdou et al. (2010) to ensure cultural competence and relevance for the purposive sample. Semi-structured focus group questions addressed two primary categories: 1) community values and cultural norms and 2) stress, coping resources and health behaviors. The first set of open-ended questions were based on Dressier’s cultural consonance methodology. Participants were asked how they define their community, what their
69


community means to them, and expectations and norms around health and well-being within their community. The second category included open-ended questions asking the perspectives of individuals and their community related to health and wellness. Sample questions include, “What helps people stay healthy in your community? What sorts of things cause stress for people in your community? What barriers does your community face related to healthy eating and active living?” Emergent themes from the group interviews were used to inform subsequent development of in-depth interviews.
In-depth interviews were conducted with mothers within public and private spaces to allow for observational analysis of environmental and objective factors that may impact individual variations in sources of stress, coping responses and early childhood obesity. A variety of participant observation and elicitation techniques were used in conjunction with in-depth interviews. In-depth interview participants partook in a concept mapping exercise. This exercise involved participants using notecards to visually portray their primary sources of external objective stressors (e.g. financial, family, relationship), perceived links between these stressors and risk behaviors, and protective factors and behaviors that help them cope with stress. These individualized concept maps illustrated how mediating and moderating risk and protective factors related to stress and obesity are connected. I interviewed participants throughout this activity using a semi-structured approach to facilitate interaction and discussion of their concept maps.
A one-page demographic questionnaire was administered to parents at the start of the focus groups and in-depth interviews. This form asked about ethnicity, education, neighborhood of residence, and average levels of stress. It also asked child-specific questions including the gender, age, height, and weight of all children under five years of age. This information was used
70


to analyze similarities and differences in themes among mothers reporting children with normal and high body mass indices (BMI).
71


CHAPTER 6
QUALITATIVE ANALYSES AND FINDINGS
This study’s qualitative analyses focused on four primary domains. These are the identification of: 1) unique sources of maternal stress, 2) maternal stress and risk behaviors, 3) maternal stress and protective factors, and 4) risk and protective pathways between maternal stress and early childhood BMI.
6.1 Participants
All participants were mothers living in high deprivation neighborhoods in Auckland with at least one child under the age of five. An ethnically diverse sample of mothers participated in the data collection; 17 mothers identified as European, 17 identified as Asian, 15 identified as Pacifika, 21 identified as Maori, and 4 identified as other. Approximately half of the mothers held a bachelor’s degree. More than half of this sample of mothers reported moderate or high levels of stress (55.4%). Approximately two thirds of mothers (n=50) reported children under five with low or normal childhood BMI (BMI for age <85th percentile) and 1/3 of mothers (n=24) reported at least one child under five with high childhood BMI (BMI for age>85th percentile).
72


Table 6.1 Qualitative Sample Characteristics (n=74)
Variables n %
Race/Ethnicity
European 17 22.9
Maori 21 28.4
Pacific 15 20.3
Asian 17 22.9
Other 4 5.4
Education
No sec school qualification 5 7.2
Sec school/NCEA 1-4 26 37.7
Diploma/Trade cert/NCEA 5-6 3 4.3
Bachelor's degree 23 33.3
Higher degree 10 14.5
Level of Stress
No Stress 6 8.1
Low Stress 25 33.8
Moderate Stress 30 40.5
High Stress 11 14.9
Child BMI
<85th Percentile 50 67.6
>85th Percentile 24 32.4
6.2 Data Analyses
After each focus group or interview, I made note of interesting observations and reflections from the session and re-read any memos made during the session. Next, I listened to all recorded focus groups and interviews and then transcribed each recording. Each transcription was read multiple times to become familiar with the data. Qualitative analysis of the transcripts from focus group discussions and in-depth interviews was completed with NVIVO software (Leech & Onwuegbuzie, 2007; NVIVO, 2012). Sources of maternal stress were coded inductively to allow for exploration and discovery of unique sources of stress. I looked for similarities and differences across ethnic groups. Based on underlying theoretical frameworks and hypotheses related to risk and protective factors associated with maternal stress, deductive coding also included a priori codes such as unhealthy eating and family support. The analysis
73


was iterative; interviews were continuously transcribed and analyzed to inform the construction and addition of future interview questions. I worked on constructing narrative summaries on an ongoing basis to integrate context into my analyses. I re-organized and aggregated the codes into themes and highlighted selected quotes that spoke to some of the social and cultural contexts surrounding maternal stress in New Zealand. This was also an iterative process, and I continued to rearrange my data in this way as new themes and patterns surfaced.
6.3 Results of Analyses Domain 1: Sources of Maternal Stress
Theme 1.1: Financial stress is the most prominent source of stress for mothers with children under the age of five from all ethnic groups residing in Auckland. New Zealand.
It is so expensive to live here now. The mental strain of living here is just hard.
Financial stress was the most frequently discussed source of stress across all ethnic
groups. Financial stress was most often tied to housing and overcrowding. Many mothers shared that they receive community housing benefits, live with their extended families, or are forced to live in neighborhoods where they do not feel safe. One mother said, “The thing that stresses me out about Auckland the most are the prices. It’s so expensive here, and we only have a two-bedroom and didn’t think that we were going to have twins. So now we need like one million dollars in five years to move out.” Another mother lived with her in-laws outside of downtown Auckland. She said, “We don’t have to pay rent, but we are where we are because of financial pressures and worries. We couldn’t get by living anywhere else.” One mother discussed her financial stressors related to dependency on community housing and overcrowding. She shared, “Housing is also a big stress for me-1 am on a list of community housing but haven’t heard anything yet. My daughter and partner live with me and my three grandkids and my partner and we only have a two-bedroom house.”
74


Financial stress was also related to work-stress. One mother said, “Sometimes I worry about my mortgage, and I have to go back to work after just five months [after the birth] to pay the bills.” Another mother shared, “I think in general, financial stresses are big here. You might not want to go back to work after you have a baby, but you have to because you both need to work.” Another said, “Financial stress is also big for us. We saved a lot for the baby and I don’t want to have to go back to work. My husband doesn’t want me to either but it’s hard. It is always a trade-off between staying at home and being able to raise our kids versus having to go back to work but having more money.” A final mother said, “I try to work a couple cleaning jobs under the table and then I am on the benefit but it’s not enough for four kids.”
Mothers also shared how financial stress is often interrelated with other stressors including their ability to take care of their children and provide for them. One mother said, “It is so expensive and that causes family stress and relationship stress and problems with my health and my child’s health.” Another said, “Financial stress is huge and the price of everything like power, food, everything is going up and up. It’s hard to do anything with the kids for that reason. And the older kids want to do sports and all of that costs money, so you can’t do that much and you have to really balance.” A mother shared, “Here, clothing is really expensive and even baby supplies have a really high tax.”
Cultural-norms and expectations were also related to financial stress among some mothers. One mother shared a story about her husband’s family and their Samoan culture. She said,
My husband’s family is huge and big into the Samoan culture and they are always calling for money. They complain about it if we don’t give them money and I’ve learned it’s just something that I have to deal with. We’ve had quite big arguments about this. My husband’s mom will ring and say there is a family at church that needs money, and I’m like, “Well we don’t even know them.” But his mom gives because Polynesians believe that what you get in the future is based on past giving so it always comes back to you.
75


Another mother who identified as Pacifika shared a story that highlighted this challenge. She said,
Last year my dad passed away, he was a high chief (matai) and all 14 siblings were expected to give $10,000 towards the funeral arrangements. All money goes to the food and cultural traditions that we have around funerals. Since he was a chief all the townspeople come and villagers and they give money and canned beef and livestock and there is a person that writes down everything that people give. That’s how they determine what they will give to their families at the next funeral. That is how they know what to give back.
Although financial stress was a common theme across all ethnic groups, Maori and European mothers shared experiences of financial stress more often than Asian and Pacifika mothers.
Theme 1.2: Maori mothers reported disproportionately higher rates of overall stress compared to Pacifika. European and Asian mothers.
Family stressors are big -with both sets of relatives. His family lives on the peninsula too, but he is disconnectedfrom them. I in like, ‘That’s your family, what are you up to? ’ Even though they have their share of contributing to my stress as well and now I had had to cut them off. Growing up I was surrounded by so many people and they violated so many boundaries so I started filtering people out. With my extendedfamily too, my circle now is quite tight —just immediate
family which is why I like it.
More than half of Maori mothers (>50%) discussed experiencing seven or more sources of stress; the experience of stress in this ethnic group was higher than stress reported by European, Pacifika and Asian mothers. Sources of maternal stress included: family stress, financial stress, relationship stress, general mom stress, work stress, mother’s physical health-related stress, and child’s physical health-related stress. Family stress was the most prominent source of stress for Maori mothers, followed by, and often linked to, financial stress and overcrowding. One Maori mother remarked, “Family stress is big because it’s annoying, but my family members are always trying to move in and it seems like when one moves out another moves in. They pretty much all live with us, both my family and my in-laws.” Another Maori
76


mother talked about the overlap between family stress and financial stress and said, “"My main
stress is really financial because money is not available. I have my three grandchildren that I
have to feed and keep them clothed and housed.” Another Maori mother shared,
Family stress is big for me. A lot of people can relate I think, because like you love your family but everyone in the family has their own issues and you get stuck in a cycle sometimes with your family. I live with my mom and her partner, my sister and brother, and my two kids and we are all in our 20s, my siblings I mean, and the way my mom’s life is, we all make it comfortable for her but it’s stressful and you don’t want to affect the family.
Mothers discussed aspects of Maori culture that may be associated with experiences of family stress. One mother said,
Now I sometimes have a problem with my culture because I don’t always understand it. We all rent three-bedrooms. My mom and her partner and her mom share a room, and my sister has a room, and my kids and I have a room, and my brother sleeps in the living room and our culture is Maori which means you never leave family behind, so I don’t know what’s happening with my culture now. It sometimes feels like my family isn’t strong and supporting each other. My family support will still always be something that I know I have and I will always carry that mindset to remind myself that I know I will never have no support but these days it’s harder and harder... When one stresses, the lot stresses.
Some Maori mothers talked about Whanau, which is the familial unit in Maori society and consists of immediate and extended family. One Maori mother shared that, “When I was induced the whole family wanted to be in the room during the birth. That is normal in Maori families for everyone to be in the space together and even to pass around the baby right when it is born before even having skin to skin contact with the mom.” Maori mothers talked about how traditionally their culture believes that the baby belongs to the family, not just to the mother and father. This is one reason why generations of families often live together which can lead to overcrowding and family stress. However, it is important to note that discussions about family stress were often intertwined with family support. Findings related to this theme are presented in Domain 3, Theme 3.1.
77


Challenges related to relationship stress were also commonly shared by Maori mothers. Relationship stress was discussed most often among Maori mothers compared to all other ethnic groups and was most often connected to the challenges of balancing children and partnerships. One mother said, “My kids put a stress on my relationship and they put so much on it that he left and had an affair when I was pregnant. He still hasn’t moved back in, but we are together now. So, we are still sorting our shit out.” Another talked about prioritizing children and grandchildren over her husband which puts a strain on their relationship. She shared, “I love him to the max; we have been together for many moons. We do have stress problems. We always travelled on holidays before and now we can’t do that no more because of the grandchildren... I don’t want to talk to him like that because it makes him feel like he’s not important, but my grandchildren are my priority.”
Theme 1.3: European mothers reported disproportionately higher rates of general mom-stress compared to Pacitika. Maori and Asian mothers.
I think the biggest health issue in my community is fatigue and that sense that you have to do everything and specially as mothers you must have a great life and have it all put together like
the house, the partner, your Instagr am...
European mothers talked about general mom-related stress more than any other source of stress. Many mothers talked about the, “generalized stress of having a baby and juggling everything” as the greater source of stress in their lives. A subtheme of general mom-related stress was the expectation that mothers juggle many different things and “do it all”. One European mother said, “I think burn out and pressure is a big health problem, like both physical and social strain.” Another said, “I think there is some competitive pressure within my social group, that’s a big one in my circle. Pressure to be social and around having kids and raising kids.” Finally, a mother shared, “it’s so hard like keeping the household running and making
78


sure the kids have food and are clothed in the mornings and are being nice people and doing the right things. Because that is so important. And even like toilet training is stressful. There is just so much to do.”
Many European mothers talked about the experience of social strain and “mom guilt”. One mother said, “Mom guilt is such as stressor like working or not working, like prioritizing baby over family and friends, the balance is hard.” Another talked about societal expectations around being a mother and raising children. She said, “Society expects us to be mothers, like mothers and working mothers and having more kids... .why are boys and girls expected to be a certain way, too? There are too many expectations” and “I experience general mom stress the most and that’s from different viewpoints and pressures related to how to raise your child.” Some mothers with multiple children also talked about guilt related to raising their children and challenges around equitability of care and attention. One mother said, “And I have three kids and the youngest one isn’t getting as much attention as the older ones did when they were little so that makes me feel guilty.” Finally, one mother shared, “You also feel guilty all the time as a mum. I mean that’s everyone.”
Domain 1 Summary: Although financial stress was the most commonly cited source of stress across all mothers, diverse sources of stress were shared and there were significant variations among ethnic groups. Overall, Maori mothers reported exposure to the highest unique number of stressors compared to European, Pacifika, and Asian mothers. Family stress, financial stress, and relationship stress were the most frequently mentioned sources of stress among Maori mothers. European mothers reported the most stress related to general-mom stress, which included social strain and mom guilt. Table 6.1 displays percentage of sources of stress referenced by each ethnic group.
79


Table 6.1 Percentages of coding references and quotes related to the most commonly mentioned sources of stress among a diverse group of New Zealand mothers (n=70)
Most Common Sources of Stress Themes European (n=17) Asian (n=17) Maori (n=21) Pacific (n=15)
Financial Stress 80% "it's so expensive. Ipay $580 a week to rent our house. It is just expensive to live here in general. There has be more perks for paying ridiculous amounts of money." 57.14% "Like financial pressure, keeping up with the bills and taking care of kids is hard." 81.25% "My main stress is really financial because money is not available.I have my 3 grandchildren that I have to feed and keep them clothed and housed." 66.67% "I try to work a couple cleaning jobs under the table and then I am on the benefit but it’s not enough for 4 kids."
Family Stress 40% "A Plunket nurse told me that I had to wake him up to feed him every three hours and my mother in law came over and said that I shouldn't be doing that." 28.57% "Not having family on either side is stressful." 93.75% "It's annoying but mv family members are always trying to move in and it seems like when one moves out another moves in." 75% "Family stress is big for me. The childrens ’ father is in jail and we see him frequently but it’s stressful. "
General Mom Stress 90% "There is some competitive pressure within my social group... Pressure to be social and around having kids and raising kids." 57.14% "Generalized stress of having a baby and juggling evenPhing." 62.50% "I think it’s really stressful being a mom in Auckland, there Is just so much to know and do." 50% "I’m a single mom and I end up thinking so much and getting anxious."
80


Table 6.1 cont’d
Work Stress 70% " I don’t want to have to go hack to work. My husband doesn’t want me to either but it’s hard. It is always a trade-off between staying at home versus having to go hack to work hut having more money. ” 71.43% "Now I can’t work overtime because I have a baby so that makes it harder because I can’t get my work done." 50% "I realize my job was the ultimate stress and everything snowballed around it. I’m learning how to calm down. " 16.67% "Work is stressful."
Time-Related Stress 70% "I never have enough arms, legs or hands and never enough time." 57.14% "I think time management is my biggest stress and if I don’t manage my time then I am stressed." 37.50% "I don’t have any time to be healthy, No time to rest and just not enough time to heal when you ’re sick." 58.33% "I'm heavily involved in church and I’m part of a women’s life group...wearing so many hats can be stressful. "
Relationship Stress 30% "My partner works long hours and by the time we have eaten and gotten the kids to bed we have like 30-minutes to have a cup of tea and some biscuits to talk about our days and then we are so tired." 28.57% "Now with baby brain, my English is harder I always feel like I misunderstood otherwise- that is part of why I argue with my husband." 62.50% "My kids put a stress on my relationship and they put so much on it that he left." 33.33% "Domestic violence is common here and there really aren ’t many supports... Organizations often just separate you from your partner. It’s like do I go back to the devil I know, or the devil I don’t know."
81


Domain 2: Maternal Stress and Risk Behaviors
Theme 2.1 Maternal Stress and Eating Behaviors
Stress makes it hard to eat healthy.
Overall, New Zealand mothers reported that stress impacts their eating behaviors more than any other health behaviors including physical activity, sleep, breastfeeding, drinking alcohol, and smoking. Maori and European mothers talked about the negative impacts of stress on healthy eating more frequently than Asian and Pacifika mothers. The feeling of being overwhelmed linked with guilt and worry was a commonly cited explanation for links between stress and unhealthy eating behaviors. One mother said, “But there is so much mom guilt and I am trying to make sure they eat different food groups and their vitamins and it’s just a lot.” Another mother shared, “I was so worried about my baby and my partner, I felt guilty and I would forget to eat, but I realized that that is not ok.” Finally, a mother said, “Yes, stress suppressed my appetite, I wasn’t hungry when I was worried about other things and I just couldn’t find time to eat.”
Lack of time was another factor that impacted stress and unhealthy eating behaviors among New Zealand mothers. One mother shared, “I love to eat healthy but when it comes to money and stress and sleep, you eat what is available because you have no time; you just try to stay sane.” Another mother said, “It’s so often hard to find time to eat even, like whenever you sit down for a quick bite he starts screaming and then I realized that if he’s crying I can let him cry a little bit which helped.” Some mothers also talked about how time affects their ability to cook meals at home for their family. One mother shared, “Sometimes at night when it’s busy it’s so easy to just get takeaways after I pick up my kids from school and have a quick meal especially when I’m stressed.” Another said, “I just stop cooking so when I don’t cook it’s just because of time management and then we don’t eat healthy so it’s just hard.” Finally, a mother
82


said, “Stress definitely affects my eating; we usually don’t eat dinner until like 10 pm. When friends and family bring food over it is so helpful.”
Mothers shared that stress impacted increased consumption of specific foods and takeaways. Sugar, and specifically chocolate and ice cream, were the most commonly mentioned foods. One mother said, “I eat more food now, my husband says you eat more than me! And more sugar now, I mean that’s what your body wants when your stressed and tired. Now I need ice cream all the time.” Another mother said, “Quite often I turn to chocolate when I’m stressed. I have emergency chocolate bars. I’ll cook dinner for the kids but then forget to eat myself or I’ll just eat whatever is in the house like biscuits. Definitely eating is affected by stress- I’ll just go buy takeaways or just eat chocolate... candy, cakes, muffins, it doesn’t matter as long as it’s chocolate.” Mothers also discussed an increased likelihood to binge eat when they are feeling stressed. One mother shared, “For me, if I’m stressed I just want to chill and like watch Netflix and snack, it’s a way for me to unwind. I usually binge eat when I’m stressed.” During periods of stress, mothers shared that they often resort to grabbing takeaways because they are quick, cheap and accessible. A mother shared that her family is working on cooking more at home because she knows it’s healthier but she said, “This week we have done more takeaways but it’s been a really stressful week.” This emphasizes the connection between stress and behavior. Buying takeaways is also associated with money-related stressors and receiving government assistance. One mother said, “Mondays and Tuesdays are hard for my family because money is low and we don’t have money for food. So, you can’t look after your physical well-being. We get takeaways a lot because of this.” Another mother said, “It is so expensive to eat healthy. I don’t make my son special meals because that is too stressful so we often eat what he likes since he is a picky eater.. .like pasta Bolognese and takeaways.. .not the
83


healthiest, but I don’t really have a choice. Certainty I turn to unhealthier items when I’m really stressed which is hard because I want to set a good example and eat right for the kids.”
Many of the discussions with New Zealand mothers emphasized their awareness of the importance of eating healthy and modeling good behaviors for their children as well as the cyclical links between stressors and eating behaviors. A mother shared, “Stress impacts our eating and eating crappy foods definitely impacts our moods and makes the kids tired. I am much more likely to be short with the kids too.” However, several mothers talked about the importance of prioritizing the health of their children over their own. There was a disconnect in understanding the link between maternal health and child health. For example, a mother said, “But for me, my healthy eating doesn’t really matter much to me. What matters is the health of our baby.” This highlights a gap in knowledge in this population and a need to focus on the associations between maternal and child health during antenatal and postnatal visits. One Maori mother shared an interesting story that suggests how familial norms and values may impact associations between stress and healthy eating. She said, “I started noticing one of my girls was becoming obese, and she was displaying real consciousness about it because she was getting bullied. I’m really supportive of her making healthier behaviors but her dad is not and doesn’t want to eat healthy which is hard because I’m cooking- like he sabotages that so trying to manage that can be difficult because he doesn’t like this vegetable or that vegetable.”
84


Case study: A Maori mothers experience of stress, loss, and health behaviors
Hana is in her late twenties, has three children under the age of ten, and identifies as Maori and European. She lives with her three children in a two-bedroom house on the far west side of Auckland. She shared her experience of stress and the recent loss of her partner, as well as impacts on behaviors and coping.
Healthy eating is affected because before my partner passed, I would always cook but now since he has passed its different; I have just now started to try to get back to a routine. I normally try to cook now, but last week, not as much. When I feel like I’m not being a good mom that really stresses me out even more. Going through grief is so challenging. I have never experienced it before. All healthy eating and physical activity helps me cope with stress though. So, when everything happened, I dropped a lot of weight, but I still made sure my kids ate well but it was hard for me. I also have a lot of friends that pray for me. I’m not as creative right now but in the past that has helped and will come back to me in time. I believe in being strong in life, I thought after he passed I would only need a month and that’s what I told my friends. But it takes time. I like the challenges of life. There has to be a rainbow at the end.
Theme 2.2 Maternal Stress and Physical Activity Behaviors
I can feel the difference after physical activity, but I just can 7 all the time and it’s just like a
spiral.
Maori and European mothers talked about the negative impacts of stress on physical activity more frequently than Asian and Pacifika mothers. However, across all ethnic groups, mothers reported that stress impacted other behaviors, including healthy eating, sleep and breastfeeding more than it impacted physical activity. Mothers with new babies and very young children talked about the challenges of finding time to exercise. One mother said, “After three months I wanted to pick up running and yoga because that is so important for my confidence but I still haven’t. I think activity is really quite important to coping with stress.” Another said, “Normally, I make an intention about being active when I am stressed. It’s much harder now and that is kind of out the window now with a baby. I know I need to get out there, it’s just hard.” Finally, a mother shared, “I almost forget sometimes I have a baby and then I’m like, Oh wait, I can’t be active.”
85


Some mothers talked about the cyclical relationships between stress and healthy lifestyles. One mother said, “If I have had a stressful day, I want to go to the gym but then thinking about it makes me more stressed. I mean, how do you exercise with having kids? I can’t do both.” Another mother shared, “Stress also impacts my physical activity, I just let it go because I’d rather just sleep or eat junk food.” When asked about how stress and activity influence her children, a mother said, “And I feel bad because my iPAD has become my new best friend and like my iPAD is my babysitter too but I don’t have a choice because I need to get dinner ready and everything.” A final mother said, “My physical activity is affected by stress, once I get in a depressive mood I don’t want to do anything and get lazy. But I don’t want my kids to think they are getting me down so I try not to show it.”
Theme 2.3 Maternal Stress and Breastfeeding
There is a lot of just general mom pressure I think and especially around breastfeeding. Hospitals are not helpful and now you have to sign a waiver if you want to bottle feed your child.
So, I think ‘Fed is best ’ not ‘Breast is best. ’
European and Pacifika mothers talked about the impacts and associations of stress and breastfeeding more frequently than Maori and Asian mothers. Often mothers shared that stress around breastfeeding impacted their ability to breastfeed which made them more stressed. One mother said, “I think there is a huge link between breastfeeding and stress. Breastfeeding was so hard for me and made me so stressed... It is really a giant logistical pain in my ass.” Many mothers talked about the direct impacts of stress on their milk supply. One mother said, “When I’m stressed my milk supply is lower which is really hard.” Another mother shared similar challenges, “I find breastfeeding difficult and when I get really really stressed my milk won’t let down and then she gets grumpy and that makes me more stressed.” A final mother said, “My hormone levels change a lot especially after breastfeeding and sometimes they change drastically during the day. Hormones are a big part of stress. Stress definitely impacts milk supply.
86


Sometimes I get stressed that he’s not getting enough and then I can’t breastfeed for a while.”
Societal expectations around breastfeeding were a big source of stress. One mother
shared, “People’s opinions are quite stressful. People are really judgmental about it like not
getting your milk through breastfeeding. It’s stressful and people make you feel guilty.” Another
mother said, “There is so much pressure to breastfeed and not being able too is stressful. I feel
guilty when I can’t feed my child.” Mothers talked about pressures to breastfeed within the
health care system in New Zealand and frustrations related to postnatal care. One mother shared,
“Breastfeeding was also really stressful, and my midwife made me even though it was so painful,
and I was bleeding and everything. She said it was important for the baby. I gave up after two
months because they weren’t latching, and she could tell that I was bottle feeding because of the
color on their tongue. I think my stress was worse for the baby than bottle feeding.” Another
mother discussed a similar concern that her stress levels were more detrimental than beneficial
for her baby and challenges related to breastfeeding information and support. She said,
I’ve had horrendous experience with breastfeeding and it caused me massive anxiety but I refused to stop because of the benefits of breastfeeding for the baby. Like what’s the best for my baby? I know my anxiety was not good for my babies- the vibrations and stress messages are sent to them. But I just inundated myself with information about breastfeeding and even from support groups there was always a lack of consistent messaging.
Lack of a supportive environment for breastfeeding was also a commonly mentioned stressor. One mother talked about the challenges of breastfeeding her baby when her preschooler is around due to the chaotic environment. She said, “I find it’s more difficult for me to breastfeed since I have a preschooler, when he is around it makes it really hard because it’s loud and I don’t think she [baby] wants to breastfeed as much when it is chaos like that.” Another shared, “The silliest thing for me is worry about my clothing for breastfeeding. To be appropriate in public it is hard to find the proper thing to wear and sometimes I spend hours deciding what to wear
87


before going out with her. I do try to avoid feeding when I am out in public.”
Theme 2,4 Maternal Stress and Sleep
If you don 7 sleep you don 7 really have energy to be active and sometimes my mind is so busy that I can 7 sleep because I am thinking and worrying about too many things.
Maori mothers talked about the impact and associations between stress and lack of sleep more than all other ethnic groups. Mothers talked about stress leading to worry and concern about their children which prevented them from getting enough sleep. A lack of sleep also seemed to be normalized; mothers appeared to accept minimal sleep as part of being a mother. One mother shared, “I just don’t sleep that much. I usually wake up at 5 am or 6 am at the latest and I stay up until 3 am or even 4.1 just don’t feel comfortable, but sleep isn’t a big stress for me because this just is how it is.” Another shared, “Obviously, young kiddies muck up your sleep. I don’t think I really have slept for the past 5 years. I felt like I couldn’t sleep because of my sons’ seizures at first and that was massive and didn’t help my stress or sleep.”
Mother also shared that they don’t prioritize sleep when they are stressed. One mother said, “And I would sleep a lot less [when I was stressed] and stay home watching movies all night and fall asleep in front of the TV because I don’t focus on sleep when I’m stressed.” A lack of sleep also impacts other health behaviors, specifically eating and physical activity. A mother said, “The sleep definitely goes when you are stressed and you can never get enough because your mind is always racing. For example, this morning I wanted to go to boot camp but because baby didn’t sleep well last night I work up so tired with a really sore back so I didn’t go this morning.”
Finally, mothers talked about the need for therapy and sleeping pills due to their high stress levels. One mother said, “When I am stressed, thoughts just keep going round and round my head and I can’t sleep. That’s why I’m in therapy and sometimes take sleeping pills.”
88


Full Text

PAGE 1

CRITICAL PERIODS OF MATERNAL STRESS EXPOSURE AND EARLY CHILDHOOD OBESITY: EXPLORING RISK AND PROTECTIVE FACTORS IN NEW ZEALAND by CHARLOTTE V. FAREWELL B.S., University of Richmond, 2004 MPH, Tulane University, 2012 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences 2019

PAGE 2

ii This thesis for the Doctor of Philosophy degree by Charlotte V. Farewell h as been approved for the Health and Behavioral Sciences Program by David Tracer , Chair Zaneta Thayer Jean Scandlyn Jini Puma Date: May 18 th , 201 9

PAGE 3

iii Farewell, Charlotte (PhD, Health and Behavioral Sciences) Critical P eriods of M aternal S tress E xposure and E arly C hildhood O besity: E xploring R isk and P rotective F actors in N ew Z ealand Thesis directed by: Professor David Tracer ABSTRACT Exposure to environmental stressors during development alters hum an biology. Importantly, stress experienced by mothers during both the pre and post natal periods of growth can hav e negative impacts on offspring development . The primary objective of this dissertation wa s to use mixed methods to explore the relationship between pre and post natal materna l stress and early childhood obesity among a nationally representative sample of New Zealand mothers. The Growing Up in New Zealand longitudinal study provided informa tion on 5,839 pregnant women and their children to assess the quantitative objectives . Exposure to one additional objective stressor during pregnancy was sig nificantly associated with a .06 increase in BMI z score at 54 months (p<.01), after controlling for covariates. Exposure to maternal stress during either the p re or early post natal period wa s associated with higher childhood BMI at 54 months of age relative to children of women not exposed to stress (p<.01). Indivi duals who experienced stress b oth prenatally and at 24 months had children with significantly higher BMI at 54 months than individuals who experienced s tress at neither or only one time point (p<.01). Structural equation modeling and qualitative methods (n =74) revealed ethnic variations in the lived experience of maternal stress and risk and protective pathways between stress and early childhood obesity. This study informs our understanding of sociocultural influences on risk exposures, protective factors a nd stress responses early in life , and resulting impacts on offspring obesity risk. Findings may help identify strategies that decrease early life predisposition to chronic disease.

PAGE 4

iv The form and content of this abstract are approved. I recommend its public ation. Approved: David Tracer

PAGE 5

v To Sean , for demanding that I balance hard work with self care and play , supporting my adventures across the world, bolstering my self esteem when I needed it most, and keeping me well fed. I love you.

PAGE 6

vi ACKNOWLEDGEMENTS I acknowledge first and foremost my dissertation chair, Dr. David Tracer, and advisor, Dr. Zaneta Thayer, as well as my two additional committee members, Dr. Jini Puma and Dr. Jean Scandlyn , for their invaluable support, investment, and expertise throughout the course of this project. I acknowledge the children and the families who are part of the Growing Up in New Zealand study , and the women who participated in the focus groups and inter views for this project . This includes the staff at the Central Plunket clinics for their support in promoting this research project and supporting recruitment of participants. I acknowledge the multiple government agencies that fund and support Growing Up in New Zealand , in particular the Ministry of Social development and the former Social Policy Evaluation and Research Unit ( also formerly the Families Commission) for their management of the contract on behalf of the Crown, as well as the ongoing support from Auckland UniServices and the University of Auckland. I thank all the members of the Growing Up in New Zealand research team for their invaluable work in interviewing participants and managing the data used in this analysis, as well as the members of Kaitiaki Group and Executive Scientific Advisory Group. Finally, I acknowledge the National Science Foundation and specifically the Biological Anthropology Program for funding this dissertation project thr ough their Doctoral Dissertation Improvement Grant award.

PAGE 7

vii TABLE O F CONTENT S I. INTRODUCTION ....... ......... .... 1 1.1 Introduction to Study ...1 1.2 Problem Statement ...4 1.3 Research Objective ..5 1.4 A ims and Hypotheses ..5 II. THEORETICAL AND EMPIRICAL BACKGROUND ................... .8 2.1 What is Stress? 8 2.2 Measures of Stress ...8 2.3 Intergenerational Transmission of Stress 1 1 2.4 Maternal Stress and Early Childhood Obesity through a DOHAD lens 1 2 2.5 Maternal Stress and Early Childhood Obesity through a Life Course Epidemiology lens 2.6 Proposed Pathways between Maternal Stress and Early Childhood Obesity 2.6 .1 Biological Pathways . . 15 2.6 .2 Behavioral Pathways .18 2.7 Lazarus Theory of Stress and Coping 2.8 Culture and Resilience 2.9 Maternal Stress and Childhood Obesity in a New Zealand Context 2.10 Literature Review Summary 27 III. QUANTITATIVE METHODS ................. . 28 3.1 Study Design Overview .28 3.2 Quantitative Participant Selection . . 31 3.3 Quantitative Data Collection Methods

PAGE 8

viii 3.4 Quantitative Data Measures 3 3.4.1 Independent Variables 4 3.4.2 Dependent Variables 7 3.4.3 Moderating and Mediating Variables 8 IV. QUANTITATIVE ANALYSES AND FINDINGS ................. ...4 1 4.1 Data Cleaning 4 1 4.2 Aim #1 ...4 5 4.3 Aim #2 7 4.4 Aim #3 ... 50 4.5 Aim #4 ...5 3 4.6 Summary of Quantitative Findings 6 4 V. QUALITATIVE METHODS .................. 6 6 5.1 Study Design Overview . 6 6 5.2 Participant Selection .. 6 7 5.3 Data Collection Methods .. . 6 8 5.4 Data Collection Instruments ..6 9 VI. QUALITATIVE ANALYSES AND RESULTS .................. ...7 2 6.1 Participants 6.2 Data Analyses 72 6.3 Results of Analyses 4 Domain 1 : Sources of Maternal Stress ...7 4 Domain 2 : Maternal Stress and Risk Behaviors 2 Domain 3 : Maternal Stress and Protective Factors 9

PAGE 9

ix Domain 4 : Maternal Stress and Early Childhood BMI: Patterns of risk and protective factors ..100 VII. DISCUSSION ............... 4 7.1 Justification of research topic ..10 4 7.2 Recap of literature review and methodology ...10 5 7.3 Brief recap of results 10 6 7.4 AIM 1 Discussion 8 7.5 AIM 2 Discussion 10 9 7.6 AIM 3 Discussion 11 1 7.7 AIM 4 and Aim 5 (Mixed Methods) Discussion 3 7.7.1 Sources of Maternal Stress ...11 3 7.7.2 Risk Behavior Pathways between Maternal Stress and Early Childhood BMI ...11 3 7.7.3 Protective Factor Pathways between Maternal Stress and Early Childhood BMI 8 VIII. CONCLUSION AND IMPLICATIONS ............... 13 2 8.1 Final Su mmary of Research Question and Overall F indings ...13 2 8.2 Reliability , Validity and Generalizability 13 3 8.3 Limitations ...13 6 8.4 Contributions of the Study ...13 7 8.5 Recommendations and Implications for Future Research ...13 9 REFERENCES 2

PAGE 10

x LIST OF T ABLES TABLE 3.1 Mixed Methods Research Timeline (Implementation Time Frame: February 2017 March 2019 .. ............ .30 3.2 Demographic Characteristics of the Quantitative Sample (n=5,839) ......................................32 3.3 Variables included in Quantitative Analyses .......... .3 3 3. 4 Objective Stressors included in GUiNZ Vulnerability Scale .......... 5 4.1 Descriptive Statistics before and after Multiple Imputation .......... ..4 4 4.2 Correlations between all continuous variables .......... 6 4.3 Hierarchical Linear Regression Models: BMI at 24 months of age .......... 8 4.4 Hierarchical Linear Regression Models: BMI at 54 months of age .............. ..4 9 4.5 Differences between Maternal Stress Groups and Childhood BMI at 54 months of age after controlling for covariates ........ .5 2 4.6 Sidak Corrected Post Hoc Comparisons for Childhood BMI at 54 months of age among four Maternal Stress Groups .................... .5 2 6.1 Qualitative Sample Characteristics .......... 7 3 6.2 Percentages of coding references and quotes related to the most commonly mentioned sources of stress among a diverse gr oup of New Zealand mothers ... 80

PAGE 11

xi LIST OF FIGURES FIGURE 1 .1 DOHaD and Life Course Epidemiology models of early life stress and BMI at 54 months 2.1 Proposed Biological and Behavioral Pathways between Pre and Post natal Maternal Stress and Early Childhood Obesity ........... .15 4.1 Maternal stress from prenatal to 24 months of age and average childhood BMI at 54 months of age ............. 5 1 4.2 Hypothesized Path Analysis Model of Prenatal Stress, Childhood BMI at 54 months, and mediators and moderators ....... .55 4.3 Mediation Model ............ 56 4.4 Overall Path Model .. ........... 58 4 .5 Final Path Analysis Model among European Mothers in New Zealand .. ........... 59 4.6 Final Path Analysis Model among Mothers in New Zealand ........... 60 4.7 Final Path Analysis Model among Pacifika Mothers in New Zealand ............ 61 4.8 Total Effects of Objective Prenatal Stress on Childhood BMI among Pacifika Women experiencing Low, Medium, and High levels of Protective Factors ..............................................62 4. 9 Final Path Analysis Model among Asian Mothers in New Zealand ............ 6 3 5.1 Mixed Methods Design ........... 6 7

PAGE 12

xii LIST OF ABBREVIATIONS Developmental Origins of Health and Disease (DOHaD) Body Mass Index (BMI) Growing Up in New Zealand (GUiNZ) Perceived Stress Scale (PSS) Hypothalamus Pituitary Adrenal (HPA) Corticotropin Releasing Hormone (CRH) Computer Assisted Personal Interview (CAPI) Food Frequency Questionnaire (FFQ) Variance Inflation Score (VIF) Statistical Package for the Social Sciences ( SPSS ) Analysis of Covariance ( ANCOVA ) Structural Eq uation M odeling (SEM) Comparative Fit Index CFI) Root Mean Square (RMSEA) Akaike information Criterion ( AIC ) International Obesity Task Force ( IOTF ) Standardized Root Mean Square Residual ( SRMR ) Confidence Interval ( CI ) Standard Deviation ( SD ) New Zealand Dollar ( NZD ) Intimate Partner Violence ( IPV )

PAGE 13

1 CHAPTER 1 INTRODUCTION 1.1 Introduction to Study Exposure to environmental stressors alters human biology (Kuzawa & Quinn, 2009; Kuzawa & Sweet, 2009; Wells, Chomtho, & Fewtrell, 2007) . Importantly, stress experienced by mothers during both t he pre and post natal periods can have impacts on offspring development, including emotional, neurodevelopment al, and physical consequences (Gluckman, Hanson, Cooper, & Thornburg, 2008; Rice et al., 2010) . Previous research examining early experience of Developmental Origins of Heal th and Disease (DOHaD) model (Gluckman, 2008) and Life Course Epidemiology (Kuh, 2003) . The DOHaD model explores the biological impacts of stress primarily during the prenata l and early postnatal period of development (Gluckman, 2008) . Research has found associations between exposure to prenatal stress and the prevalence of metabolic syndrome, diabetes, cardiovascular disease, and obesity in adulthood/later life (Benyshek, 2007; Cao Lei et al., 2015; DJP, 2004; K.N. et al., 2012; Liu, Dancause, Elgbeili, Laplante, & King, 2016a; McMillen et al., 2008) . However, a focus on stress exposures solely during the perinatal period, or prenatal and early postnatal period, ignores the potential cumulative impacts of postnatal stress on human biology. The life course approach to childhood adversity expands the DOHaD model to inco rporate an epidemiological life history perspective arguing that cumulative stress from the prenatal period through childhood and adulthood plays a critical role in the development of poor health (Kuh, 2003) . The life course model proposes that both the environment and genes

PAGE 14

2 influence health and development throughout the life course. Studies have found associations between accumulation of stress throughout the first years of li fe and a multitude of mental and physical health outcomes in adulthood, including post traumatic stress disorder, major depressive disorder, schizophrenia, diabetes, obesity, and cardiovascular disease (Cameron & Demerath, 2002; Daskalakis, Bagot, Parke r, Vinkers, & de Kloet, 2013) . A common limitation of prior research exploring impacts of maternal stress on child development outcomes is reliance on a single measure of stress. Stress is a multidimensional, translational concept (Lazarus & Folkman, 1987) . Transactions between external and internal demands and resources impact experiences of stress during the pre and post natal periods. Stress provoking factors (e.g. , external objective stressors ), stress mediating or moderating factors (e.g. coping, social support) and stress resulting factors (e.g. , perceived stress) capture different aspects of the human stress response (Lazarus & Folkman, 1987) . Measures of objective stress (i.e. objective stressors) and perceived stress therefore may or not be highly correlated (Kingston, Sword, Krueger, Hanna, & Markle Reid, 2012; Laplante, Brunet, S chmitz, Ciampi, & King, 2008; Turner & Avison, 2003) . Multiple measures of stress are needed to best understand correl ations between these dimensions. Early childhood body mass index (BMI) is a useful outcome to explore the impacts of maternal stress during critical periods of development as offspring weight and height are extremely sensitive to environmental influences (Entringer, Buss, & Wadhwa, 2010) . Early childhood obesity, defined by BMI scores at or above the 95 th percentile, is reaching epidemic proportions on a global level (World Health Organization , 2013) . Further, childhood obesity leads to adult obesity, which is a major risk factor for the development of chronic diseases, many of which have been separately associated with early life factors such as low birth weight (Biro &

PAGE 15

3 Wien, 2010) . Consideration of early life factors that may play a role in the onset of obesity requires further analyses of both the timing and duration of different stress exposures. Not all children whose m others experience early life stress develop obesity (Walton, Simpson, Darlington, & Hai nes, 2014) . An integration of biological and sociocultural perspectives using mixed methods ca n be employed to explore these complex relationships (Rodney & Mulligan, 2014) . A b iocultural approach is useful for understanding the lived experience of stress and why there is variability in early childhood obesity in response to early life stress. More specifically, this approach allows an investigation of resiliency, or sociocultura l protective factors early in life that may buffer mothers and children from the detrimental impacts of external stressors (Huang, Lee, & Lu, 2007; Panter brick & Eggerman, 2012) . For example, (Dressler, Balieiro, Ribeiro, & dos Santos, 2016; Koolhaas, de Boer, & Buwalda, 2006) . Dressler et al. (2007) theorizes that culture shapes the meaning attributed to stress ors and thus promotes individual variation in maternal coping behaviors through group norms, practices and criteria of collective social esteem and accomplishment. Sociocultural protective factors may buffer the intergenerational transmission of stress and confer positive adaptation, or resilience, among mothers and their offspring (Ager, Stark, Akesson, & Boothby, 2010; Dressler, Balieiro, Ribeiro, & Santos, 2 007; Panter brick & Eggerman, 2012) . Identifying the factors across diverse cultural groups that exacerbate or promote resiliency to stress response can facilitate development of interventions to reduce the detrimental impact of maternal stress on child development. This study integrates qualitative and quantitative methods to explore associations and pathways between maternal stress and early childhood obesity. Quantitative data analyses were conducted using the Growing up in New Zealand (GUiNZ) data s et. The GUiNZ study is a

PAGE 16

4 prospective longitudinal cohort study that began in 2009 with the recruitment of 6,822 pregnant women in Auckland, New Zealand. This sample represents 11% of all infants born in New Zealand during the study period ( Morton et al., 2013) . This study is uniq ue due to its capacity to provide a comprehensive picture of contemporary child development over time for children born in New Zealand, and for its inclusion of significant numbers of ethnic minorities. Qualitative data were collected from a diverse, conve nience sample of women recruited through Plunket Centers in Auckland, New Zealand . Plunket is a national not for profit organiz ation , community owned and governed and is the largest provider of free support services for the development, health and wellbeing of children under five in New Zealand. The organization provides services for more than 90% of newborns in New Zealand each year (Plunket, 2018) . The mixed methods approach used in this st udy allows for triangulation of data to better understand the complexities surrounding maternal stress experiences, early childhood obesity, and mediators and moderators that may explain these associations in diverse contexts. 1.2 Problem Statement Matern al stress is detrimental to maternal and child health outcomes. However, research exploring the impacts of maternal stress is limited by the use of interchangeable objective and subjective measures (Kingston et al., 2012; Liu et al., 2016a) . A lack of prospective longitudinal studies that allow for the analysis of critical periods of stress exposure and transitions in stress exposure from the prenatal period through the postnatal period has also made it difficult to understand these associations (Collins & Manolio, 2007) . The unique population of New Zealand is a particularly useful sociocultural context in which to explore these associations. The (indigenou s people of New Zealand) and Pac ifika (non people of Polynesian descent) communities are exposed to ext ernal risks , such as poverty, unemployment, and overcrowding,

PAGE 17

5 that are two to three times higher than for other e thnic groups and experience the highest rates of early child hood obesity compared to European and Asian families in New Zealand (Perry 2015; Tu rner and Lloyd 2004; Ministry of Health 2012). This diverse setting allows for exploration of associations between maternal stress and early childhood obesity within and between ethnic groups. Additional qualitative methods are needed to explore variations in individual responses to stress throughout early development and risk and protective factors that may explain pathways linking maternal stress and early childhood obesity (Shonkoff, Boyce, & McEwen, 2009) . 1.3 Research Objective The primary objective of this study wa s to use quantitative and qualitative methods to explore associations and hypothesized pathways between pre and post natal materna l stress and early childhood obesity among an ethnically diverse and representative sam ple of New Zealand mothers. 1.4 Aims and Hypotheses The following aims were investigated to further explore the potential impacts of maternal stress exposure on early childhood obesity in a New Zealand context: Aim 1: Determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy. Hypothesis 1: Objective and subjective dimensions of stress will be moderately correlated within the study sample . Aim 2: A nalyze the associatio ns between objective and subjective prenatal stress and early childhood BMI at 24 and 54 months of age. Hypothesis 2 : Objective and subjective measures of prenatal stress are independently associated with childhood BMI at 24 and 54 months after controll ing for covariates.

PAGE 18

6 Aim 3: Examine associations between the timing and duration of maternal stress exposure during the pre and post natal periods of development and early childhood BMI at 54 months of age. Figure 1 .1 DOHaD and Life Course Epidemiology models of early life stress and BMI at 54 months Aim 4 : Analyze risk and protective factors that may mediate or moderate associations between prenatal stress and early childhood BMI at 54 months of age using structural equation modeling. Hypothesis 4a: Risk factors, including l ength of exclusive breastfeeding, maternal eating behaviors , and maternal activity levels , will mediate associations between prenatal stress and early childhood BMI. Hypothesis 4b: Protective Factors, including external support, family support, cultural identit y, neighborhood integration , and household cohesiveness will moderate associations between prenatal stress and early ch ildhood BMI. Hypothesis 3: Cumulative exposure to maternal str ess is more strongly positively associated with early childhood BMI at 54 months of age compared to stress exposure during solely the pre or post natal period. The DOHaD model proposes that maternal stress during the prenatal period, regardless of postnatal experiences, is positively associated with BMI. Therefore, we predict high BMI in children experiencing prenatal stress regardless of postnatal environment. The Life C ourse Epidemiology model proposes that accumulation of maternal stress from the prenatal period through the postnatal period of development is positively associated with BMI. Therefore, BMI is predicted to be higher in children exposed to cumulative matern al stress. DOHaD Life Course Epidemiology

PAGE 19

7 Aim 5: Explore qualitatively the lived experience of stress and associations with childhood BMI among a diverse group of New Zealand mothers .

PAGE 20

8 CHAPTER 2 THEORETICAL AND EMPIRICAL BACKGROUND 2.1 What is Stress? Perspectives on human growth and development are shifting from a nature versus nurture dualism to a biocultural perspective requiring consideration of biology, environmental experiences, and cultural interactions. The complex interactions between biology ( nature) and experience (nurture) collectively impact human development (Kuzawa and Sweet 2009). Among research exploring developmental outcomes associated with early life adversity, the influences of environmental experiences, or external stressors, on the physical body and individual variation in responses to these stressors are becoming well recognized (Lock 2013; McDade 2002). The experience of s tress can loosely be defined as a non specific response of the body to any demand for change (Selye, 1955) . Stress represents a state in which homeostasis , or internal balance, is actually threatened or perceived to be threatened (Johnson, Kamilaris, Chrousos, & Gold, 1992) . It is well recognized that the definition and experience of stress represents a multifaceted system and v aries by individual. 2.2 Measures of Maternal Stress Stress can be conceptualized as biological, environmental, and psychological (Sheldon Cohen, Kessler, & Underwood, 1995) . There is currently a lack of a wides pread understanding of the relationship between these differing dimensions. Measuring biological stress captures the activation of specific physiological systems and is most often measured through the stress hormone cortisol . Defining stress in an environm ental context accounts for external events or experiences that are objectively associated with physical stress. Psychosocial stress primarily perceptions, of ev ents (Sheldon Cohen et al., 1995) . Stress research is complex due to these

PAGE 21

9 interconnected but dif fering facets of the human stress response. A common limitation of prior research exploring maternal stress is the use of a single measure of stress. The majority of studies to date have relied on self report retrospective measures of stress (Wadhwa, Entringer, Buss, & Lu, 2011) . These measures ha ve numerous biases and are often neither valid or reliable (Entringer, Buss, & Wadhwa, 2015) . Additionally, measures of objective stress (i.e. , objective stressors) and s ubjective responses to those stressors (i.e. , perceived stress measures) capture different aspects of the human stress response and may not be highly correlated (Kingston et al., 2012; Laplante et al., 2008; Turner & Avison, 2003) . Objective stress measures include stressful event scales, specific risk factors such as exposure to a natural disaster, exposure to external objective stressors, such as poverty, and stress related behaviors (Gundersen, Mahatmya, Garasky, & Lohman, 2011; Huang et al., 2007) . The Number of Stressful Life Events Index is the most commonly used objective stressful life events scale and asks participants about experiences of stressful life events in the past 12 months. The outcome is measured via a checklist of items sick and had to go rced from my husband or partner (Cohen et al., 1995) . Alternative studies have operationalized stress using specific objective risk factors including bereavement and stress related behaviors. Huang, Lee, and Lu (2007) measured stress defined as maternal bereavement related to recent death of a loved one and utilized measures of stressful behaviors (e.g. , drinking, smoking) to explore impacts on physical health outcomes. Gunderson et al. (2011) quantified stress as independent risk factors including divorce, chronic physical health conditions and domestic violence. These objective measures are useful because they collect information about external events that are easy to quantify and require simple

PAGE 22

10 measurement procedures ( Cohen, Kamarck, & Mermelstein, 1983) . Howe ver, risk factors used to create composite stress indices vary significantly and are often minimally correlated, making generalizability a challenge (Dole, 2003) . Although women may experience similar stressful events during pregnancy, secondary appraisal, or assessing ones coping options and available resources, impacts sources and perceptions of stress (Lazarus & Folkman, 1987) a nd w omen vary in their subjective responses to external stressors (Koolhaas et al., 2006) . Studies have found su bjective perceptions of stress were related to increased risk of preterm birth, birth weight, and cortisol levels in offspring ( Dole, 2 003; Rondó et al., 2003; Wadhwa, Sandman, & Garite, 2001) . The Perceived Stress Scale (PSS) is often used to measure these subjective responses in an attempt to better explain individual variation in stress experience ( Cohen et al., 1983) . The PSS consists of 10 questions measured with a 5 item likert scale and has been shown to have high reliability in a variety of populations ( Cohen et al., 1983) . Dole et al. (2003) found that increased subjective perceptions of stressful life events were related to increased risk of preterm birth. An alternative study used self report of mood as a marker for stress durin g pregnancy and found positive significant associations between subjective reports of stress and both birth weight and premature birth (Copper et al., 1996). Wadhwa et al. (2001) found strong correlations between subjective stress and cortisol levels suggesting that measuring subjective perception of stressors may accurately inform biological measures of stress. There is a need to provide clarity on the specific aspects and domains of maternal stress that may be particularly important with respect to child health outcomes ( Wadhwa et al., 2011b) . In order to best understand the intergenerational impacts of maternal stress on child health outcomes it may therefore be helpful to explore multiple measures of stress .

PAGE 23

11 2.3 Intergenerational Transmission of Stress Prenatal stress exposure is associated with poor birth outcomes, including preterm birth (Dole, 2003; Staneva, Bogossian, Pritchard, & Wittkows ki, 2015) and low birth weight (Sable & Wilkinson, 2000) . Both pre and post natal stress exposure impact postnatal health outcomes, including poor cognitive ability and a variety of affective disorders (Talge, Neal, & Glover, 2007) , behav ioral problems (Zijlmans, Riksen Walraven, & de Weerth, 2015) , poor academic performance (Pearson et al., 2016) , and elevated BMI in their offspring (Cameron & Demerath, 2002; Farewell, Thayer, Puma, & Morton, 2018; Farewell, Thayer, Tracer, & Morton, 2018; Koch, Sepa, & Ludvigsson, 2 008; Liu et al., 2016a; Wu et al., 2017) . Elevated BMI in childhood is a particularly critical outcome of interest due to the developmental origins of behavioral and biological obesity risk factors (Entringer e t al., 2010) . Additionally, BMI at age five predicts subsequent negative developmental outcomes including diabetes, heart disease and stroke in later life (Biro & Wien, 2010; Young Hyman, Schlundt, Herman, De Luca, & Counts, 2001) . Early life exposure to maternal stress may impact development of early childhood obesity through complex and intertwined physiological and behavioral pathways ( Barker, 2007; Dong et al., 2004; Gillman et al., 2006) . Health and disease susceptibility is determined by the dynamic interplay between genes and the environment particularly during early critical periods of development (Entringer et al., 2015) . Although stress experienced during the prenatal and early postnatal period may not cause early childhood obesity, it may predispose children to experiencing elevated BMIs throughout childhood and into adulthood. Two primary pathways have been proposed; 1) high levels of maternal stress may trigger the response to stress leading to chronical ly high cortisol levels (Gundersen, Lohman, Garasky, Stewart, & Eisenmann, 2008)

PAGE 24

12 which in turn has shown to be linked with elevated levels of early childhood BMI (Marniemi et al., 2002) and 2) maternal stress may lead to maladaptive coping behaviors which impact healthy eating and physical activity behaviors of the child (Laessle, Uhl, & Lindel, 2001) . These biological and behavioral pathways often interact in ways that increase obesity risk and have lifelong impacts on child health and development (Mikkilä, Räsänen, Raitakari, Pietinen, & Viikari, 2005; Nader et al., 2006) . 2.4 Maternal Stress and Early Childhood Obesity through a DOHaD Lens DOHaD research suggests that the stress experienced during the perinatal period of development is critical for laying the foundations for child growth and development (Ben Shlomo, 2002; Gluckman & Hanson, 2004) . DOHaD literature proposes way s in which environmental experiences can impact human biology and development (Holliday, 2006) . This theory h as been developed over the last 25 years and was initially derived from the Barker hypothesis (Barker et al., 1993) . The Barker Hypothesis underscored the genetic response to the physical environment and the origination of disease in sensitive periods of developm ental plasticity, primarily in utero ( Barker, 2007) . Longitudinal studies have found that beginning in utero, determinants of energy imbalance and physiological responses to external stressors can impact later adoption of obesogenic behaviors (Bauer & Boyce, 2004) . Li et al. (2010) found t hat elevated prenatal stress was associated with higher childhood BMI, although this relationship was not significant until the age of 10. Another study exploring prenatal programming of early childhood obesity found that stress induced behaviors in pregna ncy, i.e. malnutrition and smoking, were significantly positively associated with child obesity at age 5 (Huang et al., 2007) . Finally, two additional studies found that prenatal exposure to objec tive hardship , defined as exposure to a natural disaster , was significantly associated with childhood BMI at age 5 (K.N. et

PAGE 25

13 al., 2012; Liu, Dancause, Elgbeili, Laplante, & King, 2016b) . While the physiological mechanisms linking prenatal stress to childhood BMI development are becoming more clear, the potential effects of post natal behavioral factors also need to be considered (Entringer et al., 2010) . 2.5 Maternal Stress and Early Childhood Obesity through a Life Cou rse Epidemiology Lens The DOHaD model has been expanded to incorporate environmental exposures accumulating over a longer period of development, from the prenatal period through to adulthood (Bogin, 1999) . Contextualizing the cumulative impact of stress on child health environmental, and behavioral influences. The interplay of environmental risk and protective factors influenc (Umberson, Crosnoe, & Reczek, 2010) . Accumulation of stress ex posures during early life translates into disease trajectories into adulthood (Katz, S prang, & Cooke, 2012) . Life Course Epidemiology incorporates DOHaD principles but reinforces neither solely an internal nor external state. Life course epidem iology builds off the DOHaD hypothesis by contending that stress exposure from the prenatal period throughout the life course impacts biological, behavioral, and psychosocial processes that contribute t o adult health and disease risk. Stress early in life predisposes individuals to be more vulnerable to stress exposures throughout the life course, which leads to allostatic loading of cumulative stress (Nederhof & Schmidt, 2012) . Allostatic s life

PAGE 26

14 course or the inefficient operation of this stress hormone response system (McEwen & Gianaros, 2010) . Cumulative exposure to maternal stress hormones has been linked to high levels of child cortisol (Gundersen et al., 2008) and high levels of cortisol are associated with elevated BMI in early childhood (Abraham, Rubino, Sinaii, Ramsey, & Nieman, 2013) . Cross sectional studies have found significant associations between concurrent maternal stress and earl y childhood obesity, suggesting that maternal stress during the preschool years is correlated with early childhood BMI (Parks et al., 2012; Walton et al., 2014) . However, the cross sectional nature of these studies limits their findings. Using a longitudinal design, one study found that cumulative stress among mothers with children 5 to 10 years of age resulted in an increase in childhood BMI over time, highligh ting the cumulative impact of stress (Shankardass et al., 2014) . An additional study supporting the life course epidemiology framework discovered cumulative stress exposure early in life was ass ociated with concurrent childhood obesity and subsequent obesity into adulthood ( Evans, Fuller Rowell, & Doan, 2012) . These studies expand past research focused on perinatal effects by accounting for the additive impact of post natal stress experiences. Pathways of association between early exposure to maternal stress and early ch ildhood obesity have been researched although the findings are still limited. Exploring how stress translates into obesity requires a development and life course approach since stress can impact both biological and behavioral responses that are associated with early childhood obesity development (Hackman, Farah, & Meaney, 2010; Shonkoff et al., 2012) . In addition, testing the timing and duration of mat ernal stress exposures is necessary to best understand these relationships (Buss et al., 2007) .

PAGE 27

15 2.6 Proposed Pathways through which Maternal Stress may influence Early Childhood Obesity Maternal stress may impact early childhood obesity thr ough both biologica l and behavioral pathways. These hypothesized pathways are presented in Figure 2.1. Figure 2.1 Proposed Biological and Behavioral Pathways betwee n Pre and Post natal Maternal S tress and Early Childhood Obesity 2 .6 .1 Biological Pathways Prenatal stress may impact childhood obesity through various biological pathways. Stress leads to prolonged activation of physiological systems which increases the risk for development of physical and physiological diseases. The physiological stress response has evolved to provide necessary energy for fight or flight situations. However, this system can cause damaging health effects in the long term when chronically activated. The hypothalamus pituitary adrenal (HPA) axis is a cen tral biological stress regulation system that is responsible

PAGE 28

16 for producing cortisol, the primary hormone released in response to stress (Bjorntorp, 2001; Incollingo Rodriguez et al., 2015) . When stress is acute, feedback loops signal the HPA axis to stop the production of cortisol. Under periods of prolonged stress, there is initial excessive cortisol production followed by dysregulated cortisol levels (Pervanidou & Chrousos, 2016) . Specifically among preschool aged children, chronic stress exposure was found to be associated with blunted cortisol levels, emotional overeating, and higher BMI (Miller, Clifford, et al., 2013) . C ortisol may also directly influence appetite and cravings by modulating other hormones and stress responsive factors that stimulate appetite (Epel, Lapidus, McEwen, & Brownell, 2001; Lumeng et al., 2014) . The 41 amino acid hypothalamic peptide, Corticotropin Releasing Hormone (CRH) is the main regulator of HPA a xis activity during stress. Prenatal stress may impact fetal growth and later obesity risk through overexposure to CRH in pregnancy. This hormone plays a key role throughout pregnancy and levels naturally rise as pregnancy progresses. However, experiencin g stress during pregnancy leads to elevated levels of CRH (Sandman et al., 1994) . Higher CRH levels during pregnancy are associated with poor birth outcomes, including preterm birth, which is associated with later development of early childhood obesity (Stout, Espel, Sandman, Glynn, & Davis, 2015) . Gillman et al., ( 2006) conducted a pr ospective cohort study with pregnant women and found that maternal CRH levels during pregnancy were positively associated with BMI among their three year old children. These findings suggest CRH levels may explain some of the biological pathways between ex posure to maternal stress during pregnancy and early childhood obesity. CRH during pregnancy stimulates cortisol levels as well as glucose levels. Consistently high blood glucose levels combined with insulin suppression leads to cells that are starved of

PAGE 29

17 glucose due to de sensitization of glucose receptors (A. Harris & Seckl, 2011) . The energy requirements of these cells may send hunger signals to the brain, thereby resulting in overeating and maternal weight gain. Excessive g est ational weight gain is associated with early childhood obesity (Fraser et al., 2013) . Exposure to stress throughout early life may also disrupt leptin levels, a hormone that is secreted by fat cells. Leptin supports energy re gulation by balancing hunger and energy expenditure. Higher levels of leptin are associated with increased risk for early childhood and adult obesity (Miller, Lumeng, et al., 2013) . When stress exposure causes a disruptio n in the HPA axis, leptin cannot reach the brain and suppress hunger hormones. This can cause higher levels of leptin secretion, leading to higher fat storage, and increased levels of hunger hormones which may lead to overeating (Jeanrenaud & Rohner Jeanrenaud, 2001; Miller, Lumeng, et al., 2013) . Leptin is also secreted in response to stress (Tomiyama et al., 2012) . One retrospective study found that higher adversity scores during childhood were associated with higher leptin levels (Tomiyama et al., 2012) . The specific mechanisms linking early life stress, leptin responses, and childhood obesity are still unclear. Chronic stress exposure throughout the first five years of life can negatively impact the developing brain of a child (Shonkoff et al., 2012) . Specifically, the prefrontal cortex may be impacted by stress. The prefrontal cortex is responsible for executive functions including the regulation of behavior and restraint (Blair et al., 2011; Blair & Raver, 2012) . Elevated levels of stress results in decr adapt appropriately to stressful situations. Exposure to stress early in life may also alter the development and functioning of regions of the brain that are responsibl e for reward systems including the intake of food which may promote increased fat and sugar intake (Dillon et al., 2009;

PAGE 30

18 Hanson et al., 2016; Mehta et al., 2010) . Disruptions in these reward systems can result in poor eating behaviors, leading to the onset of obesity (Shonkoff JP, 2009) . One prospectiv e, longitudinal study that explored associations between children exposed to chronic stressors throughout early childhood found that such children had lower self regulation abilities and higher BMI from 9 to 13 years of age (Evans et al., 2012) . Another study found that among a group of 1500 fourth grade children, increased levels of executive functioning skills were positively associated with healthy food consumption, specifically frui t and vegetable intake, and negatively associated with intake of unhealthy food items (Riggs, Spruijt Metz, Chou, & Pentz, 2012) . Pieper and Laugero (2013) explored associations between executive functioning skills and calories consumed among a group of preschool aged children and found that these variables were inversely related; lower executive functioning skills were a ssociated with a higher number of total calories consumed. Finally, two additional studies found that poor self regulation among a group of toddler aged children was associated with being overweight (Graziano, Calkins, & Keane, 2010; Miller, Rosenblum, Retzloff, & Lumeng, 2016) . 2.6 .2 Behavioral Pathways Pre natal stress is positively associated with unhealthy eating behaviors and inactivity during pregnancy (Lobel et al., 2008a) . Oken et al. ( 2007) found that greater weight gain throughout pregnancy was associated with higher child BMI scores at three years of age. A longitudinal study of 4,234 mothers found that gestational weight gain was positively associated with BMI at all ages, from birth to 42 years of age (Schack Nielsen, Michaelsen, Gamborg, Mortensen, & Sørensen, 2010) . These findings suggest that healthy eating and physical activity behaviors during pregnancy may significantly impact the BMI of offsprin g throughout early childhood. Perinatal maternal stress is also associated with increased unhealthy eating behaviors and sugar sweetened beverage consumption among young children ( Adam & Epel, 2007) . Poor

PAGE 31

19 maternal mental health is associated with a lower likelihood of being present or involved in meals (McCurd y, Gorman, Kisler, & Metallinos Katsaras, 2014) , a higher likelihood of serving their children sugary drinks, a higher likelihood of eating outside the home, a lower likelihood of modeling healthy eating behaviors, and a lower likelihood of setting limi ts on consumption (Elias et al., 2016; Hughes, Power, Liu, Sharp, & Nicklas, 2015) . Exposure to stress early in life may also be correlated with unhealthy food environments and access to healthy foods which impact eating behav iors. Leung et al. (2014) found that low income preschool children who were living in stressful and chaotic home environments were at greater risk for obesogenic eating behaviors. An alternative study found that among 4,320 school aged children, higher levels of self reported stre ss in the home were associated with lower consumption of fruits and vegetables, and a higher consumption of high fat foods (Cartwright et al., 2003) . Bowman et al., (2004) explored associations between the numbers of stressors experienced by mothers and fast f ood intake of their preschool aged children and found significantly positive correlations. One study found that low income mothers living in stressful environments do not have the resources or access to purchase fruits and vegetables. Additionally, materna l concerns about food waste if children have not been exposed to the novel fruits and vegetables leads to the purchasing of energy dense and nutrient poor foods (Daniel, 2016) . Perinatal stress is also associated with decreased levels of physical activity and increased sedentary levels among young children. The built environment may play a significant role (Ding & Gebel, 2012; Sallis, Floyd, Rodríguez, & Saelens, 2012) . Mothers experiencing high stress environments often live in low income ne ighborhoods which are less likely to have parks and recreation areas where children participate in physical activity (Ding & Gebel, 2012) . Families living in high poverty areas are also exposed to higher rates of crime and unsafe conditions, thus minimizing opportunities for outdoor play (Ding & Gebel, 2012) . A primary risk fa ctor related to

PAGE 32

20 maternal stress is overcrowding which also is inversely associated with active play indoors (Gary W. Evans & English, 2002) . Self report of high maternal stress was found to be associated with increased levels of physical inactivity and fewer limits on screen time among a group of 110 mothers and their preschool aged children (Walton et al., 2014) . A systematic literature review of 168 studies utilizing varied measures of stress exposure found that both objective and subjective measures of stress were associated with reduced physical activity across diverse ages and ethnic groups (Stults Kolehmainen & Sinha, 2014) . High maternal stress exposure may also be associated with an earlier introduction of complimentary foods compared to mothers living in lower stress environments . This is cor related with shorter periods of breastfeeding. Li et al. ( 2008) found that experiencing stressful life events during pregnancy increased the odds for early cessation of breastfeeding independent of maternal socio demographic and biomedical factors. Shorter durations of breastfeed ing are also associated with increased risk for development of early childhood obesity above and beyond associations between maternal BMI and early childhood obesity ( Li et al., 2005) . Therefore, length of exclusive breastfeeding duration m ay also partially explain associations between perinatal maternal stress and early childhood obesity. Final ly, studies show that families reporting high levels of stress sleep less (El Sheikh, Buckhalt, Mize, & Acebo, 2006; El Sheikh, Kelly, Buckhalt, & Benjamin Hinnant, 2010; Mezick et al., 2008) . This may be associated with overcrowding, increased screen time, a lack of limits and bedtime routines, and anxiety and worry. Decreased sleep duration is associated with unhealthy eating behaviors and lack of physical ac t ivity, thus increasing the risk for developing early childhoo d obesity (Cappuccio et al., 2008; Dev, McBride, Fiese, Jones, & Cho, 2013 ) . Risk behaviors , including eating behaviors, physical activity, breastfeeding, and sleep, are associated with perinatal maternal stress and early childhood obesity and therefore may

PAGE 33

21 mediate these pathways. Exploring the direct and indirect effects of these factors can help to identify behavioral intervention opportunities to prevent the onset of early childhood obesity. 2.7 Lazarus Perspectives on biomedical health in the western world are shifting from Cartesian dualism to a transactional understanding of health and disease. The French philosopher Rene Descartes argued that the workings of the soul, or the mind, are unimportant to consider when evaluating health and disease (Lovallo, 2015 ) . This Cartesian perspective is based on the dualism between mind and body and supports t he biomedical model of disease defining health as a property of the physical body. There has been m ovement away from this model , emphasiz ing instead an ecological approach. The theory of transactionism stipulates that interactions between individuals and t heir environments shape the human stress response (Lazarus & Folkman, 1987) . Transactions and the interplay between ext ernal and internal demands and stressors can impact maternal and c hild health. Isolating different measures and experiences of stress may yield the most accurate understanding of how stress impacts the physical body. To explore these associations, s tress provoking factors (e.g. , external risk factors), stress mediating or moderating factors (e.g. social support) and stress resulting factors (e.g. , perceived stress) were investigated independently using mixed methods in this study. 2.8 Protective Factors and Resilience The lived experience of stress is rooted in meanings, interpretations, activities and interactions which are shaped by accessibility of social and cultural resources (Desjarlais & Jason Throop, 2011) . The impacts of mate rnal stress on child health outcomes cannot be separated from sociocultural contexts. These sociocultural contexts can foster resilience and inform external protective coping factors that buffer the negative impacts of early life stress on

PAGE 34

22 child health and development ( D ressler, Oths, Balieiro, Ribeiro, & Dos Santos, 2012; Shonkoff et al., 2012) . Social support is an example of a protective coping factor that may buffer the association between maternal stress and child health outcomes . The social buffering hypothesis contends that social support, defined as the perceived availability of interpersonal resources as well as the degree of integration into a network, is positively associated with decreased stress and increased well being (Sheldon Cohen & Wills, 1985) . Past studies have found that social support significantly m odifies the association between pre and post natal stress and maternal mental and physical health as well as child cortisol levels (Cohen, S., Gottlieb, B. H., & Underwood, 2004; Horton & Wallander, 2001; Thoits, 1995, 2011; Wadhwa et al., 2001) . However, it has been hypothesized that different types of social support, such as instrumental, emotional, and informational, as well as varying sour ces of social support, including family, friends, and partners, may differentially moderate perceptions of stress and resulting health outcomes (Thoits, 1995) . Cu ltural identity , or the perc eption of cohesion, commonality, and belongingness with other group members, may also increase individuals capability to positively cope with external stressors (Cameron, 2004; Clauss Ehlers, Yang, & Wan Chun Chen, 2006; Dressler, Balieiro, Ribeiro, & Santos, 2007) . The theoretica domains: lifestyle, social support, family life, national identity and food ( (Dressler et al., 2007) . Higher levels of cultural consonance, when measured as two latent variables across all domains, has been found to be associated with lower levels of psychological stress (Dressler et al., 2007) and with lower risk of obesity (Dressler et al., 2012) . Dressler et al. (2005) found that low

PAGE 35

23 cultural consonance is often a stressor. Because cultural norms and beliefs are so widely accepted within societal groups , an inability to incor own life is innately stressful. Elevated exposure to stressors coupled with low cultural consonance may result in unhealthy risk behaviors leading to greater food intake and/or consumption of a high fat diet which negatively impacts both mothers and their young children (Dressler, Oths, Ribeiro, Balieiro, & Dos Santos, 2008) . Finally, neighborhood integration and cohesion may decrease stress and promote positive coping . Lower levels of neighborhood cohesion are associated with reduced social support and an increased occurrence of negative life events and neighborhood disorder; neighborhood social disorder is associated with poor maternal mental and physical health (M. Franco, J. Pottick, & Huang, 2010) . Low neighborhood cohesion with disadvantaged communities has been shown to compound the eff ect of maternal stress and depression (K ohen, Leventhal, et al., 2008 ) . Increased social contact with neighbors has also been found to reduce fear in low income neighborhoods (Kruger, Reischl, & Gee, 2007) . However, the protective role of neighborhood integration may vary by com munity. One study found that lower frequency of interaction with neighborhoods in high stressed communities may actually provide a protective effect on mental health (Dupéré & Perkins, 2007) Factor s that promote positive coping can promote resilience, which may explain variations in the lived experience of stress between diverse groups. Panter B rick and Eggerman ( 2012) contend that resilience is a function of both biological and behavioral attributes as well as programmed responses to environmental factors. Coping factors may confound the impacts of maternal stress on development of early childhood obesity (Koolhaas et al., 2006) . Exploration of the sociocultural context of stress will inform external factors that can be mobilized to prevent

PAGE 36

24 the negative impacts of early life stress on child development (Shonkoff et al., 2012) . These protective factors may buffer the intergenerational transmiss ion of stress and confer resilience among mothers and their offspring (Ager et al., 2010) . 2.9 Maternal Stress and Childhood Obesity in a New Zealand Maternal Stress The Maternal Health study found that among a group of 1507 moms in New Zealand and Australia, 7.3% of women reported anxiety occasionally or often during pregnancy, 15.7% in the first 3 months postpartum, 10.9% at 6 months and 8.5% at 9 months postpartum (S. J. Brown, Lumley, McDonald, & Krastev, 2006) . An additional study found that among a group of 6703 mothers in New Zealand and Australia, 10.7% of women experienc ed anxiety during their first antenatal clinic visit and 9.1% reported experiencing anxiety 6 months after birth (Najman et al., 2005) . New Zea land mothers reported that being a mother in itself is inherently stressful; a longitudinal birth cohort study found that as New Zealand mothers have more children, they experience significant increases in stress and reductions in life satisfaction (Boden, Fergusson, & John Horwood, 2007) . New Zealand and Australia n moms reported severe and chronic stress after giving birth was related to acute stressful life events, lack of social support and neonatal risk. These factors were also associated with a higher rate of major depression (Hammen, Kim, Eberhart, & Brennan, 2009) . Researchers have identified objective stressors that are most prominent among New Zealand mothers. In the Maternal Health Study, mothers who experienced exposure to high external stressors were under the age of 24, unmarried, receiving government assistance, less educated compared to the average population, and not receiving paid employment during pregn ancy, thus ineligible to receive paid maternity leave. Mothers who reported experiencing physical violence were also more likely to report high stress (Gao et al.,

PAGE 37

25 2010) . A systematic review of eight longitudinal studies found that high rates of stress among New Zealand and Australian moms were associated with low socio economic status, low social support , poor partner relationship and unwanted pregnancy (Schmied et al., 2013) . Ethnicity and Stress More than 200 ethnic groups reside in Auckland, the largest city in New Zealand (MacPherson et al. 2011). The four majority ethnic groups include Europeans, Asians, (indigeno us people of New Zealand) and Pac ifika (non people of Polynesian descent). and Pacifika mothers experience significantly higher levels of stress compared to Europeans and Asians in New Zealand (R. Harris et al., 2006) . Ethnic discrimination during the pre and post natal periods is associated with n umerous poor maternal and child health outcomes including hypertension, low self reported health, increased health risk behaviors, adverse birth outcomes and childhood obesity (Dixon et al., 2012; R. Harris et al., 2012, 2006; Thayer & Kuzawa, 2015) . Ethnic and racial discrimination may contribut e to higher stress exposure in minority populations. Harris et al. ( 2012) found that the reported lifetime discrimination prevalence among the New Zealand population varied Polynesians, and 13.5% among Europeans. One stu dy found that and Pacifika families experience significantly more ethnic discrimination than New Zealand Europeans and that were almost 10 times more likely to experience multiple types of discrimination when compared to NZ Europeans ( Harris et al., 2006) . Another study found that women who reported experiencing ethnic discrimination in pregnancy had poorer self rated health and gave birth to infants with higher stress reactivity at six weeks of age, whi ch is associated with poor health and development outcomes (Thayer & Kuzawa, 2015) . However , potentially because of a small

PAGE 38

26 sample size, this study found no differences in these relationships by ethnic group. Objective contextual factors may a lso explain higher stress levels among and Pacifika families. families experience high rates of mental health disorders with 19.4% of reporting anxiety. Families with the lowest income and education levels are at greatest risk for mental health disorders (Baxter et al., 2006) . Minori ty ethnic groups in New Zealand are exposed to high rates of external risks, such as poverty and living in high deprivation neighborhoods (Perry, 2015). Sixty five percent (65%) of and 78% of Pacifika people live in the most deprived neighborhoods an d these groups tend to experience objective stressors two to three times higher than other ethnic groups (New Zealand Living Standard Survey, 2008). The Pacific Islander Family Study found that among a sample of 1590 Pacific Islander mothers, 76% reported experiencing verbal aggression and 23% reported physical violence within six weeks after giving birth. Twenty four months after giving birth, 86% reported having experienced verbal aggression and 27% reported having experienced physical violence (Gao et al., 2010) . Living in high deprivation neighborhoods across all ethnic groups is associated with higher cortisol levels during pregnancy and higher offspring cortisol levels after birth (Thayer & Kuzawa, 2014) . Early Childhood Obesity Twelve percent (12%) of New Zealand children between the ages of 2 and 14 are obese and a total of 31% of children are overweight or obese (Ministry of Health, 2017) . The prevalence of early childhood overweight and obesity in New Zealand is higher in higher deprivation neighborhoods throughout the country. Children living in the most deprived areas are 2.5 times more likely to be obese compared to children living in the least deprived areas after adjusting for age, sex and ethnicity (Ministry of Health, 2017) . Ethnic group differences in the

PAGE 39

27 prevalence of overweight and obesity also exist. The and Pac ifika populations in New Zealand experience significantly higher rates of early childhood obesity compared to the general population . Among children aged three to five, 18% of children are obese and 29% of Pacifika children are obese (Statistics New Zealand & Ministry of Pacific Island Affairs, 2011) . 2.10 Literature Review Summary Measuring m aternal stress is a challenge due to the multiple dimensions of stress and interplay of genes and experience that impact the biological stress response. Use of both objective and subjective measures can provide insight into the impacts of objective stressors and perceptions of stress on maternal and child health. Due to the intergenerational impact of maternal stress and detrimental impacts on child health and development, multiple measures are needed. Maternal stress during the pre and post natal period impact child health, and specifically development of early childhood obesity. However, analyses of the specific timing and duration of stress exposures during these critical periods on early childhood BMI is unclear. The DOHad model contends that th e prenatal and early post natal periods are the most critical. The Life Course Epidemiology model proposes that cumulative maternal stress exposure from the prenatal period throughout the first five years of life and beyond impacts child health outcomes mo re than a specific critical period. Both models identify potential behavioral and biological pathways that may explain associations between early exposure to maternal stress and childhood obesity. for consideration of stress provoking, stress mediating or moderating, and stress resulting factors in stress research. Past studies have found that risk behaviors, including unhealthy eating, lack of physical activity, lack of exclusive breastfeeding, an d lack of sleep, may mediate associations between maternal stress and early childhood obesity. Protective factors, such as social support,

PAGE 40

28 cultural identity, and neighborhood integration and cohesiveness , may also moderate these associations to confer resi lience. These relationships are particul arly relevant in New Zealand due to the high rates of maternal stress and early childhood obesity, socioeconomic and ethnic disparities, and diverse cultural setting which allows for the exploration of mediating and moderating factors that may explain individual variation in maternal stress responses and early childhood obesity.

PAGE 41

29 CHAPTER 3 QUANTITATIVE METHODS 3.1 Study Design Overview Two Phase Simultaneous Explanatory Design The data for this study were collected in multiple waves. I explored my first three aims from February 2017 to August 2017 by conducting secondary data analyses with the Growing Up in New Zealand (GUiNZ) data set. The final two aims were explored simultaneously fro m March 2018 to November 2018. First, I prepared for the qualitative data collection phase of my study by submitting my ethics application to the University of Colorado Institutional Review Board and the local ethics review board in New Zealand (Plunket Et hics). I developed drafts of my qualitative protocols a s well as a sampling plan and recruitment strategy. Concurrently, I explored the hypothesized mediation and moderation models with the GUiNZ data set to see if risk behaviors and protective factors wer e significantly related to prenatal stress and early childhood BMI at 54 months. I analyzed my final path analyses models once the qualitative data collection phase was completed. Findings from the qualitative data were used to understand variations in my final path analyses model by ethnicity and generate potential explanations for the observed differences. I then disseminated my findings via publications, presentations, and community café events. A timeline of all activities is displayed in Table 3.1.

PAGE 42

30 Table 3.1 . Mixed Methods Research Timeline (Implementation Time Frame : February 2017 March 2019 ) Go als and Activities F 17 M A M J J A S O N D J 18 F M A M J J A S O N D J F M Meet with GUiNZ team; prepare data X X Quantitative Analyses H1, H2, H3 X X X X X X X Ethics Approval X X Finalize Qualitative Tools X X X Quantitative Analyses H4 X X X X X X X X X Recruit Qualitative Participants X X Conduct Focus Groups X X Conduct Interviews X X Qualitative Analyses H5 X X X X X X X X X Disseminate Findings X X X X X X X

PAGE 43

31 3.2 Quantitative Participant Selection The GUiNZ study is based out of the Center for Longitudinal Research He Ara Ki Mua at the University of Auckland. This longitudinal study began in 2009 with the recruitment of 6,822 pregnant women in the last 12 weeks of pregnancy in the Manukau and Waikato District Health Board Regions in Auckland, New Zealand and is designed to contin ue until the children are 21 years of age. The children born into this study repr szX esent 11% of all infants born in New Zealand during the study period and the ethnicity and socio demographic characteristics are generalizable to those of children being bo rn in New Zealand today (Morton et al., 2014) . Approximately 24% of the sample iden tified as , 21% as Pacifika, 16% as Asian, and 66% as European or Other. 3.3 Quantitative Data Collection Methods A variety of data collection methods were used in the GUiNZ study, including face to face Computer Assisted Personal Interviews (CAPI), direct observations, developmental and anthropometric assessments, telephone interviews, and routine linkages to clinical records from the three District Health Boards. Data were collected from the mothers and children during the prenatal period, 9 months, 24 months, and 54 health and wellbeing, whanau (family) life, education, psychological development, neighborhood and environment, and culture and identity. For my quantitative analyses, GUiNZ data sets from four data collection waves, prenatal, 9 months, 24 months, and 54 months, were merged using the unique child identifier based on non missing data for the childhood BMI at 54 months variable (n=5,839). Table 3.2 displays characteristics of t he quantitative sample.

PAGE 44

32 Table 3.2 Demographic Characteristics of the Quantitative Sample (n=5,839) Age of Child Antenatal 24 months 54 months Variables m sd m sd m sd Child ZBMI Score 0.8 4 1.27 0.78 1.12 Objective Stress Scores .91 1.3 8 0. 83 1.2 9 Subjective Stress scores Maternal Age 12.98 30. 42 6.35 5.86 n % Ethnicity (n=5692) European 3336 58.6 Maori 741 13.0 Pacific 690 12.1 Asian 743 13.1 MELAA 98 1.7 Other 11 0.2 New Zealander 73 1.3 Education (n=5695) No sec school qualification 352 6.2 Sec school/NCEA 1 4 1267 22.2 Diploma/Trade cert/NCEA 5 6 1743 30.6 1388 24.4 Higher degree 945 16.6 Household Income (n=4470) <=20K 156 3.5 >20K <=30K 214 4.8 >30K <=50K 571 12.8 >50K <=70K 718 16.1 >70K <=100K 1062 23.8 >100K <=150K 1054 23.6 >150K 695 15.5 *Due to low attrition rates, sample characteristics remained fairly stable over time and were not collected at the 2 year data collection time point

PAGE 45

33 3.4 Quantitative Data Meas u res Table 3. 3 displays the variables included in all quantitative analyses. Table 3. 3 Variables included in Quantitative Analyses Variable Type of Variable M/SD (continuous) or N/% (categorical) Primary Independent and Dependent Variables Childhood BMI z s cores (Dependent Variable) Continuous (IOTF Z Scores) 24 months: Mean=.85 , SD= 1.27 54 months: Mean= .78, SD=1.12 Subjective Stress (Independent Variable) Continuous (Perceived Stress Scale (PSS) Prenatal: Mean=12.99, SD=6.35 Objective Stress (Independent Variable) Continuous ( Objective Vulnerability Scale) Prenatal: Mean=1.0, SD= 1.37 9 months: Mean=.90 , SD=1.28 2 years: Mean = .91, SD= 1.27 Covariates Parity (Cohort Child Order) Categorical 1=first: n=2407, 42.2% 2=subsequent : n=3296, 57.8% Maternal Race/Ethnicity Categorical 1= European: n=3336, 58.6% 2= 3= Pacific: n=690, 12.1% 4= Asian: n=743, 13.1% 5= Other; n=182, 3.2% Maternal Age Continuous Mean=30.38, SD= 5.86 Maternal Education Categorical 1=no secondary school qualification: n=352, 6.2% 2= secondary school/NCEA 1 4: n=1267, 22.2% 3=dip loma/trade certificate/NCEA 5 6: n=1743, 30.6% 4= 5=higher degree : n=945, 16.6% Total Household Income Categorical 1=< 20k: n=156, 3.5% 2=20k to < 30k: n=214, 4.8% 3=30k to < 50k: n=571, 12.8% 4=50k to < 70k: n=718, 16.1% 5=70k to < 100k: 1062, 23.8% 6=100k to < 150k: n=1054, 23.6% 7=> 150k : n=695, 15.5% Birthweight z s core Adjusted for Gestational Age Continuous Mean=.98, SD= 1.28 Pre Pregnancy Maternal BMI Continuous Mean=25.5, SD=5.81

PAGE 46

34 Hypothesized Stress Mediating Variables Maternal Healthy Eating Behaviors Continuous (Aggregate of 4 dichotomous Variables: Meeting NZ MOH nutrition guideline for Fruit and Vegetable Consumption, Breads and Cereal, Milk and Dairy, and Meat) Mean=1.30, SD=1.02 Maternal Physical Activity Behaviors Continuous (Average number of days women participate in moderate to vigorous physical activity (MVPA) during their first and second trimester) Mean=1.61, SE=1.40 Maternal Exclusive Breastfeeding Continuous (Length of exclusive breastfeeding in months) Mean=5.08 SD=1.22 Hyp othesized Stress Moderating Variables Family Support Continuous ( Family Support Scale, Dunst et al., 1984) Mean=23.99, SD=5.38 External Support Continuous (sources of external (outside of family) support) Mean=22.96, SD=5.16 Neighborhood Integration Continuous (Neighborhood Integration Scale, Kavanagh, Turrell & Subramanian, 2006) Mean=34.53, SD=4.8 Cultural Identity Continuous (20 Questions related to ethnic identity, language, cultural identity and belonging, and perception of ethnic discrimination ( Morton et al., 2010) Mean=12. 54, SD=2.42 Household Cohesiveness Continuous (Country specific scale of 9 questions related to family connectedness) Mean= 30.64, SD= 4.19 3.4.1 Independent Variables Objective Stress: Objective stress can be defined as stress related vulnerability, or external factors that impact life stress and individual risk trajectories (McEwen & Gianaros, 2010) . Environmental risk factors derived from the GUiNZ maternal vulnerability scale were used as the structural measure of stress and are presented in Table 3.3 (Morton et al., 2013) . These measures were collected via self report during the last trimester of pregnancy and 9 months and

PAGE 47

35 24 months after birth . Table 3. 4 Objective Stressors included in GUiNZ Vulnerability Scale Risk Factor Definition Maternal Smoking Continuing to smoke after first trimester of pregnancy or continuing to smoke regularly/every day Maternal Age Teenage mother at time of pregnancy Relationship Status Mother with no current partner Maternal Education Mother with no formal secondary school qualifications D eprivation Area Living in NZDEP2006 area deciles 9 or 10 Unemployment Mother not on leave, actively seeking work but not currently working Tenure public rental Living in social housing Income tested benefit In receipt of an income tested government benefit Overcrowding Having 2 or more persons per bedroom This is a population specific index of structural vulnerability measured during the prenatal period. Risk factors were dichotomized with a score of 0 or 1 depending on whether they were experienced or not and aggregated into a continuous index. These risk factors have been informed by international studi es as well as pilot work to create a population relevant vulnerability scale ( Morton e t al., 2014) . They also are routinely available and consistently measured sources of life stress. The vulnerability scale used in this study extends previous research, much of which has utilized a single risk factor to quantify external stress, for exam ple poverty or teen pregnancy. Due to the complexity of stress and multiple factors impacting stress, the scale used in this study comprising 9 structural risk factors is a more effective measure of vulnerability at the population level (Chittleborough, Lawlor, & Lynch, 2011; Tallman, 2016) . This study recognizes stress as a transactional process (Lazarus & Folkman, 1987) . Therefore, both real and perceived environmental demands will be assessed as independent stress measures.

PAGE 48

36 Prenatal Perceived Stress Scale: Perceived stress during pregnancy was operationalized with t he Cohen Perceived S tress Sc (Cohen et al., 1983) . The perceived stress scale consists of 10 questions measured with a 5 item likert scale. Examples of questions asked In the last month, how often have you been upset bec ause of something that happened control P erceived S tress S cale score for this sample ranged from 0 to 40. This measure was used as a single continuous variable in analyses. Maternal age: Maternal age at pregnancy interview was included as a confounder in all analyses. Younger age at pregnancy has been found to be positively associated with high pre and post nata l stress and depression levels (J. D. Brown, Harris, Woods, Buman, & Cox, 2012) . Maternal age is coded as a continuous variable in all analyses ( Mean=30.38, SD=5.86). Parity: Parity, or cohort child order, has been found to be negatively associated with elevated levels of BMI in childhood (Ong, Preece, Emmett, Ahmed, & Dunger, 2002) . Parity was coded as a dichotomous catego rical variable in the GUiNZ data set and as such, was used as a dichotomous variable in all analyses ( 1=first: n=2407, 42.2%; 2=subsequent: n=3296, 57.8%). Race/Ethnicity : Race/ethnic variation in objective and subjective stress experiences has been found among pregnant women (Lu & Chen, 2004; Woods, Melville, Guo, Fan, & Gavin, 2010) . Significant associations between race/ethnicity and childhood BMI suggest that Pacifika and children experience the highest levels of BMI in New Zealand (Utter, Scragg, Schaaf, & Fitzgerald, 2006a) . Race/ethnicity was coded as a categorical variable in all analyses ( 1=European: n=3336, 58.6%; 2= : n=741, 13.0%; 3=Pacifi ka : n=690, 12.1%; 4=Asian: n=743, 5= Other: n=182).

PAGE 49

37 Maternal Education: Populations with lower education levels are more likely to be exposed to stress and experience higher rates of childhood obesity compared to more educated and wealthier populations (Cohen & Janicki Deverts, 2012) . Maternal education was coded as a categorical variable in all analyses ( 1=no secondary school qualification: n=352, 6.2%; 2=secondary school/NCEA 1 4: n=1267, 22.2%; 3=diploma/trade certificate/NCEA 5 6: n=1743, 30.6%; 5=higher degree: n=945, 16.6%). Household Income : Populations with lower income levels are more likely to be exposed to stress and experience higher rates of childhood obesity compared to more educated and wealthier populations (Cohen & Janicki Deverts, 2012) . Household income was coded as a categorical variable in all analyses ( 1=< 20 k:n=156, 3.5%; 2=20k to <30k: n=214, 4.8%; 3=30k to <50k: n=571, 12.8%; 4=50k to <70k: n=718, 16.1%; 5=70k to <100k: 1062, 23.8%; 6=100k to <150k: n=1054, 23.6%; 7=> 150k: n=695, 15.5%). Birth weight (adjusted for gestational age): Prenatal stress is associated with low birth weight in offspring (Staneva et al., 2015) . Low birth weight also is positively associated with development of early childhood obesity (Danielzik, Pust, Landsberg, & Müller, 2005) . Birth weight as coded as a continuous variable in all analyses ( Mean=.98, SD= 1.28). Maternal Pre Pregnancy BMI : Maternal Pre Pregnancy BMI is associated with childhood BMI. Past studies suggest the maternal obesity during the prenatal period is associated with obesity in childhood (Benyshek, 2007; Levin, 2006) . Maternal pre pregnancy BMI was coded as a continuous variable in all analyses ( Mean=25.5, SD=5.81). 3.4.2 Dependent Variable Childhood Body Mass Index (BMI): BMI is calculated as weight in kilograms divi ded by height in meters squared collected and validated by two trained GUiNZ researchers during study visits

PAGE 50

38 at 24 months and 54 months of age. The continuous measure of gender and age specific BMI z scores is based on the International Obesity Task Force Growth Standards (Cole, 2000) . BMI was coded as a continuous variable in all analyses ( 24 months: Mean=.85, SD= 1.27; 54 mont hs: Mean= .78, SD=1.12). 3.4.3 Mediation and Moderation Variables Maternal Healthy Eating Behaviors: This variable represents an aggregate of 4 dichotomous items related to guidelines for pregnant women in New Zealand contain recommendations focused on the daily intake of the four major food groups: vegetables and fruit; bread and cereals; milk and milk products; and lean meat, meat alternatives and eggs. The self reported measure was assessed using a semi quantitative, forty four item food frequency questionnaire (FFQ). Previous studies have found that approximately 24% of pregnant women in the GUiNZ st udy did not meet the Ministry of Health recommendations for daily servings for any of the four main food groups (Morton et al., 2013) . The continuous measure was collected prenatally (scale: 0 4; Mean=1.30, SD=1.02). Maternal Physical Activity Behaviors: This variable i s a continuous variable operationalized as the average number of days women participate in moderate or vigorous physical activity (scale: 0 7 days; Mean=1.61, SE=1.40). Moderate activity was defined as intense walking and vigorous activity was defined as a ctivities that expenditure. The GUiNZ participants were asked about their usual physical activity levels during face to face interviews at pregnancy (Morton et al., 2013) . Exclusive Breastfeeding: This variable is measured c ontinuously in the GUiNZ data set as the length of exclusive breastfeeding in months. Although the New Zealand Ministry of Health

PAGE 51

39 recommends six months of exclusive breastfeeding, on average, mothers in the GUiNZ sample exclusively breastfed for five month s (Morton et al., 2013) . This variable was collected via face to face interviews at 9 months post natally (scale: 0 9 months; Mean=5.08 SD=1.22). Family Support: This is a continuous variable in the GUiNZ data set defined using the Family Support Scale ( Dunst, Jenkins, Trivette, 1984). Support sources included partner, parents, collected prenatally and a higher score reflected higher expected helpfulness (scale: 0 36; Mean=30.64, SD=4.19). External support: Parents were asked during the last tr imester of pregnancy to report what sources of external (outside of family) support they expected to have available, and how helpful they expected each source to be, once their baby was born . These support sources included family doctor, professionals (such as Plunket nurse or kaiawhina ), early parenting programs (such as Parents as First Teachers), books and the interne t. This variable was collected prenatally and a higher score reflected h igher expected helpfulness (scale: 0 36; Mean=22.96, SD=5.16 ). Neighborhood Integration: This is a continuous variable comprised of 13 questions about (Neighborhood Integration Scale, Kavanagh, Turrell & Subramanian, 2006). This variable was collected prenatally and a higher score reflected higher neighborhood integration (scale: 0 40; Mean=34.53, SD=4.8). Cultural Identity : This is a continuous variable comprised of 20 questions related to ethnic identity, language, cultural identity and belonging, and perception of ethnic discrimination (Morton et al., 2010) . This va riable was collected prenatally and a higher score reflected higher

PAGE 52

40 cultural identity (scale: 0 20; Mean=12.54, SD=2.42). Household Cohesiveness: This is a continuous variable comprised of 9 questions related to family connectedness and interaction (Morton et al., 2010) . This variable was collected prenatally and a higher score reflects higher family cohesiveness (scale: 9 36, Mean=30.64, SD=4.19).

PAGE 53

41 CHAPTER 4 QUANTITATIVE ANALYSES AND FINDINGS Section 4.1 Cleaning, Diagnosing, and Treating the Data Independent and dependent variables of interest in the Growing Up in New Zealand (GUiNZ) data set were screened to explore linear regression assumptions. First, multivariate normality was explored. Multivariate normality refers to the normality of the error terms related to the independent and dependent variables. Violations of normality may bias the parameter estimates and standard errors. However, normal ity is a larger concern among small sample sizes and, therefore, is not as important an assumption when running linear regression models with the GUiNZ data set (n=5,839). Histograms and normal p p plots of regression standardized residuals were analyzed using SPSS software version 25 (IBM SPSS Inc., 2012) . The residuals of the variables of interest did not violate the assumption of multivariate normality. Next, the assumption of homoscedasticity, or constant variance, was evaluated. The disturbance term should not differ for varying values of the independent variable. If heteroscedasticity exists, parameter estimates ma y be inefficient and observed standard errors may be biased which could lead to inaccurate conclusions about significant associations between variables. Homoscedasticity in the GUiNZ data set was evaluated by running a scatterplot of residuals to ensure th at there is equal variability across the x axis from the bottom to the top of the y axis. The scatterplot revealed that the data met this assumption. Since the assumptions of normality and homoscedasticity were met, the assumption of linearity was also me t. M ulti collinearity between all variables included in the regression models was evaluated . When independent variables are too highly correlated it may lead to instability of the model coefficients. The standard errors of the regression coefficients were reviewed to ensure

PAGE 54

42 collinearity may be an issue. Correlations between all variables of interest were analyzed to ensure none were considered high, or above .6. Next, the tolerance and variance inflation fact or (VIF) scores for each independent variable was reviewed. The VIF measures the impact of collinearity among the variables in a regression model and is equal to 1/tolerance. Cohen et al (2003) suggests that a problem exists if the tolerance values are les s than .1 or the variance inflation factors are above 10. All variables of interest met these criteria. It was concluded that no problems related to multi collinearity exist in the GUiNZ data set. To explore the presence of extreme cases in my variables of interest, distance and leverage which dictate the influence of extreme cases were explored. Distance refers to the distance of estimates from the regression line. These outliers have an extreme ca se on the outcome variable. Plotting the studentized residuals, or standardized residuals, can reveal if the distance of any data points are a concern. Leverage refers to the distance from the mean of X. To assess the overall influence of extreme cases, d istance is multiplied by the leverage. Cooks D, which is the most common measure of influence, was used to identify any maximum values that may exceed Cooks D in this data set (Cooks D=.014). The Cooks D values were plotted along the x axis in a scatterplo axis. There were three points that were far from zero and exceeded the maximum Cooks D Value. I explored these outliers and determined that they were extreme potentially due to data entry or measurement error. I ran my basic reg ression model with and without these three outliers; the coefficients did not change. Therefore, I dropped these three outliers from the data set for analyses. Due to the large sample size represented in the GUiNZ data set, there was a large percentage o f missing on specific variables on interest. Missing data ranged from .2% to 17.6%

PAGE 55

43 across all variables of interest. However, o nly 1.8% of data on the primary outcome measure (B MI z score at 54 months of age) w as missing . Of the 5,839 maternal child dyads included in the analyses, 3,363 cases (57.6%) had complete data for all variables. Data was missing at random for the primary dependent and independent variables included in this analysis. Since more than 5% of data was miss ing on key variables, multiple imputation was used to create imputed values for missing data across all variables (White, Royston, & Wood, 2011) . The imputation model included all the variables in the multivariate analysis mode l. The ob served cases and the imputed cases were analyzed separately to compare results before and after imputation. Findings were consistent and are displayed in Table 4.1. Pooled data was used for all regression analyses.

PAGE 56

44 Table 4.1 Descriptive Statistics before and after Multiple Imputation Before Imputation After Imputation Variables N M SD N M SD Predictor Variables Objective Stress Prenatal 5130 0.91 1.38 5839 0.94 1.39 Objective Stress 9 mo nths 5139 0.83 1.28 5839 0.86 1.30 Objective Stress 24 mo nths 5244 0.83 1.29 5839 0.85 1.30 Subjective Stress 5189 12.98 6.35 5839 12.97 6.37 Outcome Variables Childhood BMI 24 months 4814 0.84 1.27 5839 0.85 1.27 Childhood BMI 54 months 5732 0.78 1.12 5839 0.78 1.12 Stress Mediator Variables Maternal Healthy Eating Behaviors 5189 1.30 1.02 5839 1.31 1.02 Days Moderately to Physically Active Prenatal 5302 1.61 1.40 5839 1.62 1.39 Exclusive Breastfeeding 5373 5.88 2.15 5839 5.89 2.14 Stress Moderator Variables Family Support Prenatal 5189 23.99 5.38 5839 24.03 5.36 External Support Prenatal 5189 22.96 5.16 5839 23.00 5.16 Family Cohesiveness Prenatal 5702 30.64 4.19 5839 30.63 4.20 Cultural Identity Prenatal 5653 12.54 2.42 5839 12.52 2.42 Neighborhood Integration Prenatal 5185 34.53 4.89 5839 34.53 4.89 Covariates Maternal Age 5709 30.42 5.86 5839 30.41 5.85 Parity 5702 1.58 0.49 5839 1.58 0.49 Race/Ethnicity 5692 1.93 1.37 5839 1.93 1.37 Birthweight Adjusted for Gestational Age 5827 0.98 1.29 5839 0.98 1.29 Maternal BMI 5081 25.32 5.78 5839 25.53 5.82 Maternal Education 5695 2.23 1.15 5839 2.23 1.15 Household Income 4470 4.85 1.58 5839 4.71 1.63 Due to the large sample size of this data set (n=5839), too much statistical power was a concern. Sufficient statistical power is critical to ensure that one avoids making a Type I error (rejecting a true null hypothesis) or Type II error (failing to rejec t a false null hypothesis). The sample size, significance criterion, and population effect size are involved in determining statistical power. I was careful about interpreting significance and set the alpha to .01 because of

PAGE 57

45 this large sample size. I also was careful to ensure findings were practically significant, in addition to statistically significance. Section 4.2 Aim #1 Aim 1. Determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy. Analyses Bivariate correlations between all variables were analyzed using SPSS software version 25 (IBM SPSS Inc., 2012) . Results Bivariate analyses were conducted between prenatal objective stress and prenatal subjective stress. All bivariate correlates are displayed in Table 4.2 . Correlations between prenatal subjective stress and objective stress were low but significant and positive ( r = .27 , p<.01).

PAGE 58

46

PAGE 59

47 Section 4.3 Aim #2 Aim 2: Analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24 and 54 months of age. Analyses Bivariate correlations were conducted between prenatal objective and subjective stress variables and childhood BMI at 24 and 54 months. Greater prenatal subjective stress was s ignificantly and positively c orrelated with greater BMI at 24 month s although the correlations were very low ( r = .04 ) and 54 month s of age (r = .08) ( both p<.01). Greater prenatal objective stress was significantly and positively associated with greater BMI at 2 4 months ( r = .11) and 54 months ( r = .22 ) ( both p<.01). Multivariate regression models were used to explore associations between the primary independent variables (prenatal objective stress and prenatal subjective stress) and primary depe ndent variables (BMI z scores at 24 months and 54 months ) using SPSS software version 25 (IBM SPSS Inc., 2012) . First, hierarchical linear regression models were used to explore associations between objective stress and BMI z scores at 24 months of age. Covariates were entered in Block 1. Next, the objective stress variable was entered in Block 2. The model building proce ss was then repeated with the subjective stress variable entered in Block 2. The only difference between the objective stress and subjective stress regression models w as the inclusion of additional covariates, including maternal education, income, and age , in the subjective stress models. These demographic variables are part of the objective stress variable and are, therefore, not controlled for in these models. Both sets of models were re run with BMI z scores at 54 months of age as the primary outcome va riable. C oefficien ts, standard errors, p values, R 2 and adjusted R 2 values are reported. Results Table 4.3 shows R 2 and adjusted R 2 values, coefficients, standard errors, t statistics and

PAGE 60

48 p values for hierarchical linear regression models with BMI z scores at 2 4 months of age as the primary outcome variable. First, associations between objective prenatal stress and BMI at 24 months of age were analyzed. Covariates were entered first into block 1 and the objective stress variable was added in Block 2 . Exposure to one additional risk factor during pregnancy wa s sig nificantly associated with a .06 i ncrease in BMI z score at 24 months of age ( t statistic = 5.45; p<.01). Next, the model was run with subjective stress as the primary independent variable. S ubjective stress was not significantly as sociated with BMI z scores at 24 months of age after controlling for covariates (t statistic = 1.43; p = 0.15). Table 4.3 Hi erarchical Linear Regression Models: BMI at 24 months of age Table 4.4 displays R 2 and adjusted R 2 values, coefficients, standard errors, t statistics and p values for hierarchical linear regression models with BMI z scores at 54 months of age as the B B

PAGE 61

49 primary outcome variable. The model building process was repeated beginning with analyzing objective stress as the primary predictor. In the full model, 17.1 % of the vari ation in BMI z scores at age 54 months is accounted for and an increase in exposure to one risk factor during pregnancy is sig nificantly associated with a .06 increase in BMI z score at age 54 months ( t=5.95, p<.01). Subjective str ess was not significantly ass ociated with BMI z scores at 54 months of age (t= 1.59, p=.11). Table 4.4 H ierarchical Linear Regression Models: BMI at 54 months of age B B

PAGE 62

50 Section 4.4 Aim #3 Aim 3. Examine associations between the timing and duration of maternal stress exposure during the pre and post natal periods of development and early childhood BMI at 54 months of age. Analyses To explore the timing and duration of objective stress exposure, the objective stress measure was dichotomized into no exposure to stress or exposure to stress (No Stress = 0 risk factors; Stress = 1 or more risk factors). Just over half of the sample (51.9%) experienced no stress during the prenatal period and the 24 mont h period of data collection. There were no other natural cut points in the number of risk factors mothers experienced from the prenatal period to 24 months. Stably low stress was defined as mothers experiencing no stress during the prenatal and 24 month d ata collection period. Stably high stress was defined as mothers experiencing stress during the prenatal and 24 month data collection period. Increased stress was defined as mothers experiencing no stress during the prenatal period but experiencing stress during the 24 month data collection period. Decreased stress was defined as mothers experiencing high stress during the prenatal period and no stress during the 24 month data collection period. The stably low group was used as the reference group in all an alyses. A one way between subjects ANCOVA was conducted to compare the effect of stress on childhood BMI among stably low stress, stably high stress, increased stress, and decreased stress conditions , controlling for covariates . Since statistically signifi cant differences were found, post hoc contrast tests were conducted. All analyses were conducted using SPSS software version 25 (IBM SPSS Inc., 2012) . Results Figure 4.1 depicts the average childhood BMI at 54 months of age among the four maternal stress transition groups: stably low stress (n = 3928), increased stress (n = 299),

PAGE 63

51 decreased stress (n = 438) and stably high stress (n = 1174). The average childhood BMI z scores were 0.61 (n=3928) , 0.97 (n=299) , 0.98 (n=438) , and 1.28 (n=1174) standard deviations above the IOTF reference mean for the stably low, increased stress, decreased stress, and stably hi gh stress groups, respectively. Figure 4.1 Mate rnal stress from prenatal to 24 months of age and average childhood BMI at 54 months of age Table 4.5 highlights the significant main effect of maternal stress on childhood BMI at 54 months of age among all four stress groups after controlling for covariates [F(3, 5833) =53.3, p 0.01]. Sidak corrected post hoc comparison analyses are presented in Table 4 .6. These planned contrasts revealed that mothers experiencing cumulative high stress, p < 0.01, 95% CI [1.04, 1.17], increased stress, p < 0.01, 95% CI [0.76, 0.98], or decreased stress p < 0.01, 95% CI [0.76, 1.03], from the pre to post natal period is significantly associated with higher early

PAGE 64

52 childhood BMI at 54 months of age compared to mothers who experience stably low stress. Mothers experiencing transitions in stress exposure from the pre to the post natal period, regardless of direction, is signi ficantly associated with lower early childhood BMI at 54 months of age compared to mothers who experience stably high stress (p < 0.01). However, the mean early childhood BMI score for the increased stress group did not differ from the mean childhood BMI s core for the decreased stress group (p= 1.0). Table 4.5 Differences between Maternal Stress Groups and Childhood BMI at 54 months of age after controlling for covariates Sum of Squares df Mean Square F Between Groups 175.95 3 58.65 53.5 ** Within Groups 6412.85 5833 1. 10 Total 7315.1 5839 **p < 0.01 Table 4.6 Sidak Corrected Post H oc Comparisons for Childhood BMI at 54 months of age amon g four Maternal Stress Groups 95% Confidence Interval (I) (J) Mean Std. Lower Upper Transition Group Transition Group Diff (I J) Error Bound Bound Stably Low Increased .18* .06 .35 .02 Decreased .21 ** .05 .35 .07 Stably High .44 ** .04 .54 .35 Increased Stably Low .18* .06 .02 .35 Decreased .03 .08 .24 .18 Stably High .26 ** .07 .44 .08 Decreased Stably Low .21 ** .05 .07 .35 Increased .03 .08 .18 .24 Stably High .23 ** .06 .39 .08 Stably High Stably Low .44 ** .04 .35 .54 Increased .26 ** .07 .08 .4 4 Decreased .23 ** .06 .08 .39 *p<.05 **p<.01

PAGE 65

53 Section 4.5 Aim #4 Aim 4. Analyze risk behaviors and protective factors that may mediate or moderate associations between maternal stress and early childhood BMI at 54 months of age using structural equation modeling. Analyses Structural equation modelin g (SEM) was used to examine relationships between the objective stress measure, risk behaviors and protective factors, and childhood BMI at 54 months. SEM is a unique method to explore the complex relationships between pre and post natal stress measures and childhood BMI for multiple reasons including: 1) it allows for analyses of multiple dependent variables; 2) it permits variables to be examined as independent and dependent variables; and 3) it is beneficial for testing both moderation and mediation mo dels (Bowen & Guo, 2011). There are four key steps involved in running SEM analyses. These include: 1. Specification form explicit hypothesis using regression and factor analysis concepts to form structural restrictions, 2. Estimation use a SEM softwar e package to estimate coefficients and standard errors and various statistical indicators, 3. Evaluation compare alternative structural restrictions in a series of statistical tests, and 4. Re Evaluation reconsider alternative model and suggest ways to deal with new concepts (R. Kline, 2015) . Path analysis, which is a special case of SEM where all variables are measured, was used to explore relationships between the measured variables. The overall goal of path analysis is to estimate causal verse non causal aspects of observed correlations betw een variables. Analyses convey how the hypothesized model accounts for the data observed correlations (i.e. standardized variables) or covariances (i.e. unstandardized variables). Path analysis is an extension of multiple regression whereby several regres sion relationships can be estimated

PAGE 66

54 simultaneously (e.g., a variable that is an outcome of one variable may be simultaneously examined as a predictor of another). Assumptions of path analyses models include: (a) v ariables are measured without error; (b) me asures are at the interval level; (c) residuals are normally distributed with zero mean and constant variance; (d) residuals are uncorrelated with one another and with the predictor variables in the equation in which each residual appears; (e) relationship s among variables are unidirectional, thereby ruling out reciprocal relationships and feedback loops; and (f) relationships among var iables are additive and linear (P. Kline, 2014) . Path a nalyses were conducted using SPSS Version 25 for descriptive and correlational analyses a nd Mplus Version 7.2 ( Muthén and Muthén, 1998 ) for path model estimation. Prior to model estimation, the zero order correlations among all model constructs were examined. Model relationships on early childhood BMI were estimated. Models are evaluated in terms of how well the model , as a whole , fit the data, and in terms of the significance of each of the specific proposed relationships. This method provides a way to examine both direct and indirect predictors of early childhood BMI. Figure 4.2 depicts the estimated model, where prenatal objective stress was estimated as a direct predictor of early childhood BMI at 54 months of age and an indirect predictor of early childho od BMI at 54 months of age through unhealthy risk behaviors, including unhealthy eating and inactivity during pregnancy, and shorter periods of exclusive breastfeeding. Protective factors including social support, family support, neighborhood integ ration, family cohesiveness , and cultural identity were estimated as moderators in the model and predicted to moderate the direct effect between prenatal stress and early childhood BMI. In other words, associations between prenatal stress and early childh ood BMI were hypothesized to vary based on the level of protective factors mothers experienced during pregnancy. Mediation and moderation models were explored with each of these variables

PAGE 67

55 analyzed independently, as well as in aggregate form (e.g. aggregate risk variable, aggregate protective variable). Non hierarchical models based on a priori hypotheses were evaluated in terms of the magnitude, direction, and significance of the estimated path coefficients by several standard me asures of overall model fit: a c hi square p value > 0.05 for the difference between the theoretical and the empirical model, comparative fit index (CFI) > . 90, root mean square error of approximation (RMSEA) < 0.08 and standardized root mean square residual (SRMR) <.08. Missing d ata was addressed using multiple imputation techniques as described above. Figure 4.2 Hypothesized Path Analysis Model of Prenatal Stress, Childhood BMI at 54 months, and mediators and moderators Results All path coefficients are standardized for ease of interpretation of the magnitude of the effect. First, the mediation model was explored. Each of the mediating variables was included in the model. Figure 4.3 depicts the mediation model.

PAGE 68

56 Figure 4. 3 Mediation Model of Prenatal Stress, Childhood BMI at 54 months, and mediators Figure 4.3 Mediation Model The number of nutrition guidelines met during pregnancy and length of exclusive breastfeeding significantly mediated the relationship between prenat al stress and early childhood BMI at 54 months (both p<.01); however, the number of days mothers exercised during pregnancy did not mediate this relationship. Therefore, the exercise variable was dropped from the model and only the nutrition and breastfeed ing variables were included as mediators. To explore the direct moderation of the hypothesized protective variables (moderation of direct path between prenatal stress and early childhood BMI), moderation models were run with all five variables measured in dependently and subsequently as one aggregate variable. The model with protective factors as an aggregate variable had better model fit (AIC=67057.90, AIC= 50505, respectively). Based on these findings, the protective variables were included as one aggrega ted protective variable for all path analyses. Next, this direct moderation model was compared to a model with both direct and indirect moderation to see which model had better fit. This model was based on the a priori hypothesis that protective factors ma y mitigate the significant association between objective stress and unhealthy eating behaviors and lower levels of exclusive breastfeeding in addition to the direct pathways between objective stress and early childhood BMI at 54 months. The model

PAGE 69

57 with both direct and indirect moderation exhibited the best fit (AIC = 50497.21 ). Since the model was just identified, additional fit statistics were not available. A just identified, or saturated model, is an identified model in which the number of free parameters exactly equals the number of known values; the model has zero degrees o f freedom. To further explore the model fit, the pathway between protective factors and early childhood BMI was constrained. This path was chosen because the relationship was not significant and close to zero ( =.01). The resulting path analysis model is displayed in Figure 4.4. Figure 4.4 Overall Path Model This model most accurately fit the data ( ² (1 ) = .826, p=.3634 ; AIC= 50496.03 ; CFI= 1.0 ; RMSEA = .00, 95% CI=0.00, 0.03; SRMS= .002 ). Prenatal stress was positively associated with early childhood BMI at 54 months of age ( =.23, p<.001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy ( = .14, p<.001) and length of exclusive breastfeeding ( = .08, p<.001). The number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding were negatively associated with early childhood BMI at 54 months ( = .04, = .03, both p<.001). There was a significant, but low,

PAGE 70

58 positive correlation between the two mediating variables ( =.04, p<.001). Protective factors, including social support, family support, family cohesiveness, neighborhood integration and cultural identity, did not moderate the direct pathway between prenatal stress and early childhood BMI ( = .02, p=.10). The pro tective variable was also positively and significantly associated with the number of nutrition guidelines met during pregnancy ( = .09, p<.001) which suggests this mediation pathway is moderated by protective factors. Protective factors may moderate the m ediated effects of prenatal stress on early childhood BMI transmitted primarily through nutrition behaviors during pregnancy. The protective variable was not associated with length of exclusive breastfeeding in this model. This suggests that although prot ective factors may have an impact on nutrition behaviors health, they do not moderate the relationship between prenatal objective stress and early childhood BMI at 54 months of age in this overall model. This final model Pacifika mothers (n=690), and Asian mothers (n=743) in New Zealand to understand variations between prenatal stress and early childhood BMI by ethnicity. European Mothers First, the model was explored among mothers who identified as European (n=3336). Among Europeans, the protective path between protective factors and early childhood BMI was similarly not significant and close to zero so this path was constrained to display model fit stati stics. This model accurately fit this subset of the sample ( ² (1 ) = 1.08, p=.2997 ; AIC= 34180.35 ; CFI= .999 ; RMSEA = .005, 95% CI=0.00, 0.05; SRM R = .004) and is displayed in Figure 4.5.

PAGE 71

59 Figure 4.5 Final Path Analysis Model among European Mothers in New Zealand Among European mothers, prenatal stress was positively associated with early childhood BMI at 54 months of age ( =.11, p<.001). Nutrition behaviors and length of exclusive breastfeeding were not significant mediators in this model. However, pr enatal stress was negatively associated with the length of exclusive breastfeeding ( = .09, p<.001). There was a significant and positive correlation between the two mediating variables ( =.08, p<.001). Protective factors did not significantly moderate the direct pathway between prenatal stress and early childhood BMI. Protective factors were positively and significantly associated with the number of nutrition guidelines met during pregnancy ( = .13, p<.001). Next, the model was explored significant so this path was not constrained. The correlation between the mediators was not statistically significant and close to zero so this path was constrained. This model accurately fit

PAGE 72

60 this subset of the sample ( ² (1 ) = .686, p=.4075 ; AIC= 8957.76 ; CFI= 1.00 ; RMSEA = .00, 95% CI= 0.00, 0.08; SRM R = .007) and is displayed in Figure 4.6. BMI at 54 months of age ( =.14, p<.001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy ( = .14, p<.001) and length of exclusive breastfeeding ( = .11, p<.001). The length of exclusive breastfeeding was negatively associated with early c hildhood BMI at 54 months ( = .06, p<.001), therefore the length of exclusive breastfeeding was a significant mediator between prenatal stress and early childhood BMI. The protective factor variable was negatively and significantly associated with the nu mber of nutrition guidelines met during pregnancy ( = .1, p<.001). Protective factors were also significantly and positively associated with early childhood BMI ( = .13, p<.001). Protective factors do not significantly moderate the direct pathway between prenatal stress and early

PAGE 73

61 Pacifika Mothers The model was next explored within mothers who identified as Pacifika (n=690). Among Pacifika women, the protective path between protective factors and early childhood BMI was significant so this path was not constrained. The correlation path between the mediators was not statistically significant and close to zero so this path was constrained. This model accurately fit this subset of the sample ( ² (1 ) = .091, p=.7629 ; AIC= 7893.96 ; CFI= 1.00 ; RMSEA = .00, 95% CI=0.00, 0.07; SRM R = .003) and is displayed in Figure 4.7. Figure 4.7 Final Path Analysis Model among Pacifika Mothers in New Zealand Among Pacifika mothers, prenatal stress was positively associated with early childh ood BMI at 54 months of age ( =.08, p<.001). Prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy ( = .16, p<.001) and length of exclusive breastfeeding ( = .08, p<.001). The number of nutrition guidelines met during pregnancy was negatively associated with early childhood BMI at 54 months ( = .08, p<.001), therefore, the number of nutrition guidelines met during pregnancy was a significant mediator between prenatal stress and early childhood BM I. The protective factor variable was also positively and

PAGE 74

62 significantly associated with the length of exclusive breastfeeding ( = .13, p<.001). Protective factors were also significantly and positively associated with early childhood BMI ( = .13, p<.001) . Protective factors also significantly and negatively moderate the direct pathway between prenatal stress and early childhood BMI ( = .12, p<.001) suggesting direct moderation exists between these variables among Pacifika mothers. To explore how the tot al effects of prenatal stress on early childhood BMI vary at different levels of protective factors, the model was evaluated under low protective factors ( 1 SD from the mean= 2.75), medium protective factors (mean=0), and high protective factors (1 SD fr om the mean=2.75 ) conditions. Results are displayed in Figure 4.8. Among Pacifika mothers experiencing low levels of protective factors, the slope between prenatal stress and early childhood BMI was .12 (SE=.04, t =3.07, p=.002). Among Pacifika mothers exp eriencing average levels of protective factors, the slope between prenatal stress and early childhood BMI was .07 (SE=.03, t = 2.69, p=.007). Among Pacifika mothers experiencing high levels of protective factors, the slope between prenatal stress and early childhood BMI was .02 (SE=.03, t = .77, p=.44). Figure 4.8 Total Effect s of Objective Prenatal Stress on Childhood BMI among Pacifika Women experiencing Low, Medium, and High levels of P rotective F actors

PAGE 75

63 Asian Mothers Finally, the model was explored within mothers who identified as Asian (n=743). Among Asian women, the protective path between protective factors and early childhood BMI was significant so this path was not constrained. The correlation between the mediator s was not statistically significant and close to zero so this path was constrained. This model accurately fit this subset of the sample ( ² (1 ) = .115, p=.7342; AIC = 8168.68 ; CFI= 1.00 ; RMSEA = .00, 95% CI=0.00, 0.07; SRM R = .003) and is displayed in Figure 4.9. Figure 4.9 Final Path Analysis Model among Asian Mothers in New Zealand Among Asian mothers, prenatal stress was not associated with early childhood BMI at 54 months of age. Nutrition behaviors and length of exclusive breastfeeding were not significant mediators in this model. However, prenatal stress was negatively associated with the number of nutrition guidelines met during pregnancy ( = .07, p<.001) and the length of exclusive breastfeeding was positively associated with early childhood BMI at 54 months of age ( = .07, p<.001). Protective factors did not significantly moderate the direct pathway between prenatal stress and early childhood BMI. However, the protective variable was positively associated with

PAGE 76

64 the number of nutrition guidel ines met during pregnancy ( = .11, p<.001). Section 4.6 Summary of Quantitative Findings Correlations between prenatal subjective stress and objective stress were low , but positive and statistically significant. Greater prenatal subjective and objective stress were significantly and positively c orrelated with greater BMI at 24 months and 54 month s of age . However, after controlling for confounding variables, only objective stress was significantly associated with early childhood BMI at both time points. Mothers experiencing transitions in objective stress exposure from the pre to the post natal pe riod, regardless of direction, wa s associated with statistically significantly lower early childhood BMI at 54 months of age when compared to mothers who experi ence stably high stress , and statistically significantly higher early childhood BMI at 54 months of age when compared to mothers who experienced no stress . However, mothers who experienced increased stress or decreased stress from the pre to post natal pe riod did not differ from each other with respect to early childhood BMI suggesting that the direction of stress transition is not predictive of early childhood BMI at 54 months. Risk and protective pathways between objective prenatal stress and early childhood BMI at 54 months were explored. In the overall sample, the number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding significantly mediated associations between objective prenatal stress and e arly childhood BMI at 54 months; the number of days women exercised during pregnancy did not significantly mediate this relationship. Higher objective prenatal stress was significantly associated with a lower number of nutrition guidelines met and shorter length of exclusive breastfeeding. Higher levels of both these mediators were significantly associated with lower early childhood BMI at 54 months. Five protective factors were explored as moderators. A final path analysis model with an aggregate protectiv e moderating variable and the two

PAGE 77

65 significant mediators was fit to the data. The model with both direct and indirect moderation best fit the sample. Protective factors were positively and significantly associated with nutrition guidelines but were not sign ificantly associated with breastfeeding behaviors or the direct path between prenatal stress and early childhood BMI at 54 months in the overall sample. However, Asia n mothers to understand differences in risk and protective pathways between prenatal stress and early childhood BMI and results varied by ethnic group. Higher levels of protective factors were positively and significantly associated with healthy eating beh aviors among European and associated with childhood BMI. The interaction b etween protective factors and stress was only significant among Pacifika mothers suggesting that protective factors mitigate the experience of maternal stress and buffer the direct pathway between maternal stress and early childhood BMI at 54 months for th is ethnic group.

PAGE 78

66 CHAPTER 5 QUALITATIVE METHODS This chapter describes the qualitative methods utilized in this project. The qualitative phase began after completion of the first three quantitative aims: 1) to determine whether objective and subjective measures of stress are correlated among a diverse sample of New Zealand women during pregnancy, 2) to analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24 and 54 months of age, and 3) to e xamine associations between the timing and duration of maternal stress exposure during the pre and post natal periods of development and early childhood BMI at 54 months of age. A im #4 (to a nalyze risk and protective factors that may mediate or moderate as sociations between prenatal stress and early childhood BMI at 54 months of age using structural equation modeling) and Aim #5 (to e xplore qualitatively the lived experience of stress and associations with childhood BMI among a diverse group of New Zealand mothers) were explored concurrently. This chapter presents findings related to Aim #5. The qualitative phase took approximately six months from development through data collection and analysis. Details of the methods, analyses and results follow. 5.1 Study Design Overview The primary purpose of the qualitative phase is to expand on the understanding of the lived experience of maternal stress among ethnically diverse mothers in New Zealand and risk and protective factors that may explain associations be tween maternal stress and early childhood BMI. The joint use of qualitative and quantitative methods allows for the triangulation of data. The methods serve as a check on one another and allows a researcher to gain a clearer understanding of the issues and topics under investigation (Maxwell, 2013) . Mixed methods also

PAGE 79

67 result in data related to different aspects of maternal stress and variations within and between ethnic groups. The current chapter will focus on the highlighted sections below. 5.2 Participant Selection The targeted population for the qualitative component of this study was mothers who met the following criteria: 1) living in Auckland, New Zealand, 2) at least one child five years of age or younger, and 3) currently experiencing ob jective stress (e.g. living in highest deprivation quantile). Carpenter (1999) suggests that the sample size in phenomenol ogical research should be small so that each experience can be examined in depth. The purpose is not necessarily to generalize the fi ndings but to understand the complexity of meaning attached to stress. Therefore, my goal was to recruit a purposive sample of mothers with young children who were experiencing at least one objective stressor and conduct a combination of focus group s and in depth interviews. Additionally, I recruited an equal sample of mothers from the four primary et hnic groups in New Zealand: European, Recruitment Recruitment occurred using a purposive, non probabilistic sampling method. In February of 2018, I began recruitment for focus groups. First, I compiled a list of high deprivation suburbs in the Auckland region. The deprivation scores represent a single and simple index of

PAGE 80

68 socioeconomic deprivation derived from national census dat a in New Zealand. The deprivation Zealand (Salmond & Crampton, 2012) . Living in a neighborhood that falls into the highest quin tile of deprivation is a risk factor on the G rowing Up in New Zealand (GUiNZ) objective stress scale. I then overlaid demographic information to target diverse communities with high ited from Plunket, which are family support and resource clinics, because approximately 90% of mothers with young children utilize some aspects of Plunket services in New Zealand (Plunket, 2018) . I met with the Central Director of the Plunket offices , and we discuss ed opportunities to attend currently and mother support groups comprised of mothers residing in these high deprivation and e thnically diverse neighborhoods. Because the director felt that the topic of maternal stress aligned with the focus of many of these sessions, Plunket facilitators allowed me to attend some of their activities and conduct focus groups during and/or after t heir sessions. One of the Plunket staff members introduced me to a community member who became my New Zealand consultant. She works with a f aith based organization which i s a community affiliate of Plunket. He r main roles were to review my focus group a nd in depth interview protocols, provide fee dback from a local perspective, and share her insight s into the cultural and con textual factors that may be related to the maternal stress experience in Auckland. She also helped to recruit mothers for in depth i nterviews in the neighborhoods identified above. We worked to recruit an equal number of participants from each of the primary ethnic groups. 5. 3 Data Collection Methods I collected qualitative data using both foc us groups and in depth interviews from

PAGE 81

69 March 6 th to April 20 th , 2018. The objectives of the focus groups were to explore societal experiences of stress among mothers with young children residing in Auckland and to understand patterns of risk behaviors and protective factors that might explain t he association between stress and early childhood obesity. Since stress is a broad topic, I felt that focus groups would encourage mothers to share their experiences and views on stress and coping behaviors in the context of their communities. I also colle ct ed data using in depth interviews to explore individual lived experiences of maternal stress and variations in risk and protective behaviors. Stress can be a sensitive topic and mothers may feel more comfortable sharing their experiences in a one on one format. I found th at these two methods were comple mentary and provided a unique understanding of maternal stress in this diverse community. I conducted focus groups and interviews within the same two month period. I facilitated 7 one hour focus groups wit h 46 mothers and 28 one hour in churches, coffee shops, and libraries. Data were collected from a total of 74 participants. On completion of the interview or focus group, participants received a $20 gift card. Ethical approval for this study was obta ined from the Colorado Multi Institutional Review Board in the United States and the Plunket Ethics Review Committee in New Zealan d. 5.4 Data Collection Instruments With the help of my advisors and my New Zealand consultant, focus group questions were adapted from Dressler et al. (2005) and Abdou et al. (2010) to ensure cultural competence and relevance for the purposive sample. Semi stru ctured focus group questions addressed two primary categories: 1) community values and cultural norms and 2) stress, coping resources and health behaviors. The first set of open consonance methodology. Participants were asked how they define their community, what their

PAGE 82

70 community means to them, and expectations and norms around health and well being within their community. The second category included open ended questions asking the perspectives of indi helps people stay healthy in your community? What sorts of things cause stress for people in your community? What barriers does your community face related to healt hy eating and active development of in depth interviews. In depth interviews were conducted with mothers within public and private spaces to allow for observational analysi s of environmental and objective factors that may impact individual variations in sources of stress, coping responses and early childhood obesity. A variety of participant observation and elicitation techniques were used in conjunction with in depth interv iews. In depth interview participants partook in a concept mapping exercise . This exercise involved participants using note cards to visually portray their primary sources of external objective stress ors (e.g. financial, family, relationship) , perceived links between these stressors and risk behaviors, and protective factors and behaviors that help them cope with stress. These individualized concept maps illustrated how mediating and moderating risk and protective factors related to stress and obesity are connected. I interviewed participants throughout this activity using a semi structured approach to facilitate interaction and discussion of their concept maps. A one page demographic questionnaire was administered to parents at the start of the focus groups and in depth interviews. This form asked about ethnicity, education, neighborhood of residence, and average levels of stress. It also asked child specific questions including the gender, age, height, and weight of all children under five y ears of age. This information was used

PAGE 83

71 to analyze similarities and differences in themes among mothers reporting children with normal and high body mass indices (BMI).

PAGE 84

72 CHAPTER 6 QUALITATIVE ANALYSES AND FINDINGS of: 1) unique sources of maternal stress, 2) maternal stress and risk behaviors, 3) maternal stress and protective factors, and 4) risk and protective pathways between maternal stress and early childhood BMI. 6.1 Participants All participants were mothers living in high deprivation neighborhoods in Auckland with at least one child under the age of five. An ethnically diverse sample of mothers participated in the data collection; 1 7 mothers identified as European, 17 identified as Asian, 1 5 identified as Pacifika, 2 1 rs reported moderate or high levels of stress (55.4%). Approximately two thirds of mothers (n=50) reported children under five with low or normal childhood BMI (BMI for age <85 th percentile) and 1/3 of mothers (n=24) reported at least one child under five with high childhood BMI (BMI for age>85 th percentile).

PAGE 85

73 Table 6.1 Qualitative Sample Characteristics (n=7 4 ) Variables n % Race/Ethnicity European 17 22.9 2 1 2 8.4 Pacific 15 20.3 Asian 17 22.9 Other 4 5.4 Education No sec school qualification 5 7.2 Sec school/NCEA 1 4 26 37.7 Diploma/Trade cert/NCEA 5 6 3 4.3 23 33.3 Higher degree 10 14.5 Level of Stress No Stress 6 8.1 Low Stress 25 33.8 Moderate Stress 30 40.5 High Stress 11 14.9 Child BMI < 85 th Percentile 50 67.6 >85 th Percentile 24 32.4 6.2 Data Analyses After each focus group or interview , I made note of interesting observations and reflections from the session and re read any memos made during the session. Next, I listened to all recorded focus groups and interviews and then transcribed each recording. Each transcription was read multiple times to become familiar with the data. Qualitative analysis of the transcripts from focus group discussions and in depth interviews was completed with NVIVO software (Leech & Onwuegbuzie, 2007; NVIVO, 2012 ) . Sources of maternal stress were coded inductively to allow for exploration and discovery of unique sources of stress. I looked for similarities and differences across ethnic groups. Based on underlying theoretical frameworks and hypotheses related to risk and protective factors associated with maternal stress, deductive coding also included a priori codes such as unhealthy eating and family support. The analysis

PAGE 86

74 was iterative; i nterviews were continuously transcribed and analyzed to inform the construction and addition of future interview questions. I worked on constructing narrative summaries on an ongoing basis to integrate context into my analyses. I re organized and aggregated the codes into themes and highlighted selected quotes that spoke to some of t he social and cultural contexts surrounding maternal stress in New Zealand. This was also an iterative process, and I continued to rearrange my data in this way as new themes and patterns surfaced. 6.3 Results of Analyses Domain 1: Sources of Maternal Stress Theme 1.1: Financial stress is the most prominent source of stress for mothers with children under the age of five from all ethnic groups residing in Auckland, New Zealand. It is so expensive to live here now. The mental strain of living here is just hard . Financial stress was the most frequently discussed source of stress across all ethnic groups. Financial stress was most often tied to housing and overcrowding. Many mothers shared that they receive community housing benefits, live with their ext ended families, or are forced to T he thing that stress es me out about Auc kland the most are the prices. I nsive here, and we only have a two e going to have twins. So no w we need like one million dollars Another mother lived with her in laws outside of downtown pressures and financial stressors related to dependency on community housing and overcrowding. She shared, Housing is also a big stress for me eard anything yet. My daughter a nd partner live with me and my three grandkids and my partner and we only have a two

PAGE 87

75 Financial stress was also related to work about my mortgage , and I have to go back to work after just five months [after the birth] to pay I think in general, financial stresses are big here. You might not want to go back to work after you have a baby , but you have t o because you both need to a trade off between staying at home and being able to r aise our kids versus having to go bac k to I try to work a couple cleaning jobs under the table and then I am on the benefit Mothers also shared how financial stress is o ften interrelated with other stressors I t is so expensive and that causes family stress and relationship stress and problems with my health Financial stress is huge and the price of everything like And the older kids want to do sports and all of that costs money , so you can t d o that much and Here, cloth ing is really expensive and even baby supplies have a really high tax Cultural norms and expectations were also related to financial stress among some mothers. One mother shared a said, M something But his mom gives because Polynesians believe that what you get in the future is based on past giving so it always comes back to you.

PAGE 88

76 Another mother who identified as Pacifika shared a story that highlig hted this challenge. She said, L ast year my dad passed away, he was a high chief ( matai ) and all 14 siblings were expected to give $10,000 towards the funeral arrangements. All money goes to the food and cultural traditions that we have around funerals. Since he was a chief all the townspeople come and villagers and they give money and canne d beef and livestock and what they will give to their families at the next funeral. That is how they know what to give back. Although financial stress was a common theme across all ethnic groups, and European mothers shared experiences of financial stress more often than Asian and Pacifika mothers. compared to Pacifika, European and Asian mothers. Family stressors are big with both sets of relatives. His family lives on the peninsula too, but he have the ir share of contributing to my stress as well and now I had had to cut them off. Growing up I was surrounded by so many people and they violated so many boundaries so I started filtering people out. With my extended family too, my circle now is quite tigh t just immediate family which is why I like it. of stress; the experience of stress in this ethnic group was higher than stress reported by European, Pacifika and Asian m others. Sources of maternal stress included: family stress, related stress. Family stress was the most prominent so overcrowding. One remarked but my family members are always trying to move in and it seems like when one moves out another moves i n. They pretty much all live with us, both my family and my in

PAGE 89

77 mother talked about the overlap between family stress and financial stress and said stress is really financial because money is not avai lable . I have my three grandchildren that I have to feed an Another mother shared, Family stress is big for me. A lot of people can relate I think , because like you love your family but everyone in the family has thei r own issues and you get stuck in a cycle sometimes with your family. I live with my mom and her partner, my sister and brother, and my two kids and we are all in our 20s, my siblings I mean , and the way my mom s life is, we all make it comfortable for her but it s stressful and you the family. Mothers discussed aspects of culture that may be associated with experiences of f amily stress. One mother said, erstand i t. We all rent three bedrooms. M y mom and her partner and her mom share a room, and my sister has a room, and my kids and I have a room, and my brother sleeps in the living room and our culture is which means you never leave family behind , s strong and supporting each other. My family support will still always be something that I know I have and I will always carry that mindset to remind myself that I know I will When one str esses, the lot stresses. Whanau and consists of immediate and extended family. One When I was induced the whole family wanted to be in the room during the birth. That is normal in families for everyone to be in the space together and even to pass around the baby right when it is born before even having ski n to mothers talked about how traditionally their culture believe s that the baby belongs to the family, not just to the mother and father. This is one reason why generations of families often live together which can lead to ov ercrowding and family stress . However, it is important to note that discussions about family stress were often intertwined with family support. Findings related to this theme are presented in Domain 3, Theme 3.1.

PAGE 90

78 Challenges related to relationship stress Relationship stress was discussed most often among mothers compared to all other ethnic groups and was most often connected to the challenges of balancing children and partnerships . One mother ds put a stress on my relationship and they put so much on it that he left , but we are together now. So , we Another talked about prioritizing children an d I love him to the max; we have been together for many moons. We do have stress problems. We always e of the grandchildren... I , but m y Theme 1.3: European mothers reported disproportionately higher rates of general mom stress I think the biggest health issue in my community is fatigue and that sense that you have to do everything and speciall y as mothers you must have a great life and have it all put together like the hous European mothers talked about general mom related stress more than any other source of stress. Many mothers eneralized stress of having a baby and juggling relate d stress was the expectation that mothers ju . One out and pressure is a big health problem, like b oth physical d having kids and raising kids

PAGE 91

79 sure the kids have food and are clothed in the mornings and are being nice people and doing the right things. Because that is so important. And even like toilet train ing is stressfu l. There is just Many European mothers talked about the experience of social strain and mom guilt . baby over family an d friends Another talked about societal expectations mothers , like mothers and working mothers certa in way , Some mothers with multiple children also talked about guilt related to raisi ng their children and challenges around equitability of care and attention And I have three kids and le so ou also feel guilty all the time as a Domain 1 Summary : Although financial stress was the most commonly cited source of stress across all mothers, diverse sources of stress were shared and there were significant variations r of stressors compared to European, Pacifika, and Asian mothers. Family stress, financial stress, and European mothers reported the most stress related to genera l mom stress, which included social strain and mom guilt. Table 6.1 displays percentage of sources of stress referenced by each ethnic group .

PAGE 92

80 Table 6.1 Percentages of coding references and quotes related to the most commonly mentioned sources of stress amo ng a diverse group of New Zealand mothers (n= 70 ) Most Common Sources of Stress Themes European (n=17) Asian (n=17) (n=2 1 ) Pacific (n=15) Financial Stress 80% 57.14% 81.25% 66.67% $580 a week to rent our house. It is just expensive to live here in general. There has be more perks for paying ridiculous amounts of money . " "Like financial pressure, keeping up with the bills and taking care of kids is hard." "My main stress is really financial because money is not available . I have my 3 grandchildren that I have to feed and keep them clothed and housed." "I try to work a couple cleaning jo bs under the table and then I am on the benefit kids . " Family Stress 40% 28.57% 93.75% 75% "A Plunket nurse told me that I had to wake him up to feed him every three hours and my mother in law came over and said that I "Not having family on either side is stressful . " "It's annoying but my family members are always trying to move in and it seems like whe n one moves out another moves in ." "Family stress is big for me. The childrens father is in jail and we see him General Mom Stress 90% 57.14% 62.50% 50% "There is some competitive pressure within my social group... Pressure to be social and around having kids and raising kids." "Generalized stress of having a baby and juggling everything . " "I think it s really stressful being a mom in Auckland, there Is just so much to know and do . " "I m a single mom and I end up thinking so much and getting anxious."

PAGE 93

81 Table 6.1 Work Stress 70% 71.43% 50% 16.67% back to work. My husband trade off between staying at home versus having to go back to work but having more money. "Now I can t work overtime because I have a baby so that makes it harder because "I realize my job was the ultimate stress and everything snowballed around it. I m learning how to calm down." "Work is stressful . " Time Related Stress 70% 57.14% 37.50% 58.33% "I never have enough arms, legs or hands and never enough time . " "I think time management is my biggest stress and if I am stressed." t have any time to be healthy, N o time to rest and just not enough time to heal "I'm hea vily involved in part of a women s life group...wearing so many hats can be stressful." Relationship Stress 30% 28.57% 62.50% 33.33% "My par t ner works long hours and by the time we have eaten and gotten the kids to bed we have like 30 minutes to have a cup of tea and some biscuits to talk about our days and then we are so tired." "Now with baby brain, my English is harder I always feel like I misunderstood otherwise that is part of why I argue with my husband." " My kids put a stress on my relationship and they put so much on it that he left . " " D omestic violence is common here and there support often just separate you from o I go back to the devil I know , or

PAGE 94

82 Domain 2 : Maternal Stress and Risk Behaviors Theme 2.1 Maternal Stress and Eating Behaviors Stress makes it hard to eat healthy. Overall, New Zealand mothers reported that stress impacts their eating behaviors more than any other health behaviors including physical activity, sleep, breastfeeding, drinking alcohol, and smoking. mothers talked about the negative impacts of stress on healthy eating more frequently than Asian and Pacifika mothers . The feeling of being overwhelmed linked with guilt and worry was a commonly cited explanation for links between stress and unhealthy eating behaviors. O am trying to make sure they eat different food groups and th Another mother shared, I was so worried about my baby and my partner, I felt guilty and I would forget to eat, but Lack of time was another factor that impacted stress and u nhealthy eating behaviors among New Zealand mothers. One money and stress and sleep, you eat what is available because you have no time ; you just try to ten hard to find time to eat even, like whenever you Some moth ers also talked about how time affects their ability to cook me als at home for their family. One m so easy to just get takeaways after I pick up my kids from school and have a quick meal Finally, a mother

PAGE 95

83 en friends and family br Mothers shared that stress impacted increased consumption of specific foods and take a ways. Sugar, and specifically chocolate and ice cream, were the most commonly mentioned foods. One mother said, Now Another mother often I turn to chocolate when m stressed. I have emergency chocolate bars. I ll cook dinner for the kids bu t then forget to eat ll just eat whatever is in the house like biscuits. Definitely eating is affected by stress tter Mothers also discussed an increased likelihood to binge eat when they are feeling stressed. One m stressed I just want to chill and like me to unwind. I usuall During periods of stress, mothers shared that they often resort to grabbing takeaways because they are quick, cheap and accessible. A mother shared that her family is working on cooking more at home because she knows it T his week we have done emphasizes the co nnection between stress and behavior. Buying takeaways is also associated with money related stressors and receiv ing government assistance. One , physical well being. We get takeaways a lot because of this moth

PAGE 96

84 healthiest , stressed which is hard because I want to set a good exam Many of the discussions with New Zealand mothers emphasized their awareness of the importance of eating healthy a nd modeling good behaviors for their children as well as the cyclical links between stressors and eating behaviors. A mother eating and eating cra ppy foods definitely impacts our moods and makes the kids tired. I am much more li kely mothers talked about the importance of prioritizing the health of their children over their own . There was a disconnect in understanding the link between maternal health and child health. For exa mple, a mother said, the health of This highlights a gap in knowledge in this population and a need to focus on the associations between maternal and child health during antenatal and postnatal visits. One mother shared an interesting story that suggests how familial norms and values may impact irls was becoming obese, and she w as displaying real consciousness about it because she was getting like he sabotages that so trying to man

PAGE 97

85 Theme 2.2 Maternal Stress and Physical Activity Behaviors spiral. and European mothers talked about the negative impacts of stress on physical activity more frequently than Asian and Pacifika mothers . However, across all ethnic groups , mothers reported that stress impacted other behavi ors , including healthy eating, sleep and breastfeeding more than it impa cted physical activity . Mothers with new babies and very young children talked about the challenges of finding time to exercise. One After three mo nths I wanted to pick u p running and yoga because that is so important for my confidence but Another said, ow and that is kind of out the window now with a baby. I know I need t Finally, a , O h wait, I Case study: A Maori mothers experience of stress, loss, and health behaviors Hana is in her late twenties, has three children under the age of ten, and identifies as Maori and European. She lives with her three children in a two bedroom house on the far west side of Auckland. She shared her experience of stress and the recent loss of her partner, as well as impacts on behaviors and coping. Healthy eating is affected because before my partner passed , I would always cook but now since he has passed its different; I have just now started to try to get back to a routine. I normally try to cook now , but last week , not as much. When I feel like m not being a good mom that really stresses me out even more . Going through grief is so challenging. I have never experienced it before. All healthy eating and physical activity helps me cope with stress though . So , when everything happened, I dropped a lot of weight, but I still made sure my kids ate well but it w as hard for me. I also have a lot of friends that pray for me. I m not as creative right now but in the past that has helped and will come back to me in time. I believe in being strong in life, I thought what I told my friends. But it takes time. I like the challenges of life. There has to be a rainbow at the end.

PAGE 98

86 Some mothers talked about the cyclic al relationships between stress and healthy , I want to go to the gym but then thinking about it makes me more stressed. I mean, how do you exercise wi Another mother S tress also impacts my physical activity, I just let it go When asked about how stress and activity i PAD has become my new be st friend and like my i dinner ready and everyth kids to think they are getting Theme 2.3 Maternal Stress and Breastfeeding There is a lot of just general mom pressure I think and especially around breastfeeding. Hospitals are not helpful an d now you have to sign a waiver if you want to bottle feed your child. European and Pacifika mothers talked about the impacts and associations of stress and breastfeeding more frequently than and Asian mothers. Often mothers shared that stress around breastfeeding impacted their ability to breastfeed which made them more stressed. One hard for me and made m e so stressed... It is really a g Many mothers talked about the direct impacts of stress on their milk supply. One mother A nother mother shared simila r down and then she gets grumpy a A final mother hormone levels change a lot especially after breastfeeding and so metimes they change drastically during the day. Hormones are a big part of stress. Stress definitely impacts milk supply.

PAGE 99

87 Societal expec tations around breastfeeding were a big source of stress. One mother getting your milk through breastfeeding. I mother s o o is stressful. I feel gui Mothers talked about pressures to breastfeed within the health care system in New Zealand and frustrations related to postnatal ca re. One mother shared, and I was bleeding and everything. She said it was important for the baby. I gave up after two she could tell that I was bottle feeding because of the color on their tongue. I think my stress was worse fo Another mother discussed a similar concern that her stress levels were more detrimental than beneficial for her ba by and challenges related to breastfeeding information and support. She said, e had horrendous experience with breastfeeding and it caused me massive anxiety but b est for my baby? I know my anxiety was not good for my babies the vibrations and stress messages are sent to them. But I just inundated myself with information about breastfeeding and even from support groups there was always a lack of consistent messagi n g. Lack of a supportive environment for breastfeeding was also a commonly mentione d stressor. One mother talked about the challenges of breastfeeding her baby when her preschooler is around due to the c cult for me to breastfeed think she [baby] wants to breastfeed as mu silliest thing for me is worry abou t my clothing for breastfeeding. To be appropriate in public it is hard to find the proper thing to wear and sometimes I spend hours deciding what to wear

PAGE 100

88 before going out with her. I do try to avoid f Theme 2.4 Maternal St ress and Sleep mothers talked abou t the impact and associations between stress and lack of sleep more than all other ethnic groups. Mothers talked about stress leading to worry and concern about their children which prevented them from getting enough sleep. A lack of sleep also seemed to be normalized; mothers appeared to accept min imal sleep as part of being a mother . because this just Another shared, up your slee p. I seizures at first and that was massive and d One mother A lack of sleep also impacts other health behaviors, specifically eating and physical activity. A The sleep definitely goes when you are stressed and you can never get enough because your mind is always racing. For e xample, this morning I wanted to go to boot camp but Finally, mothers talked about the need for therapy and sleeping pills due to their high stress levels. One mother round

PAGE 101

89 onth and I thought he was going to be helpful but h Domain 2 Summary: New Zealand mothers shared that stress most significantly impacts their diet and that increased stress results in forgetting to eat, eati ng junk food and sugar sweetened beverages, eating takeaways, and less cooking at home. Stress also makes it more difficult to be active on a regular basis, breastfeed, and sleep. These effects are not as pronounced as the effects of stress on eating behav iors in this population. Domain 3: Maternal Stress and Protective Factors Theme 3.1 External Support (Family, Social, and Partner Support) Almost all mothers discussed the importance of external support as a coping mechanism for dealing with stress; howe and Pacifika mothers reported family as their primary source of support, European mothers reported social networks and friendships as their primary sources of support, and Asian mother s reported partners as their primary source of support. and Pacifika m others . I have lots of family here. If I talk about other support groups they are the last place I would ever I have my family. Family s upport was most commonly mentioned among and Pacifika women. and Pacifika mothers often shared that most of their support network is their family because of co habitation or their family members all living within walk ing distance. One y mother in my sis comes around once a week to check on Family support

PAGE 102

90 Another is amazing. Anything I need she s always saying you need to let me watch the kids so you can do something for yourself but I and Pacifika women also talked about the interconnectedness of their families as a source of support. A common theme was a shared sense of well being among all family members; if one is struggling, the whole family bonds together to support that individual. One sisters are all close, and they are like my arms and legs. When o mother shared a story that my family. I supported it but some of my sisters and children did no t. So , we all had to sit down as a family and talk about it. At the end of the day everything got sorted out though. Whenever family is stressed out we all band together and support each other . Another F amily support will still alway s be something that I know I have and I will always carry We grew up so close and continued on s Grandparents were mentioned as a common source of support for and Pacifika women. One like my immediate family is there. They continue to help extensively and are my childcare they take all my kids whenever I want to do something for me. They cook and care for my children helping each other and filling in , reall y encour A Pacifika mother parents are younger so they are healthy and can still work which means they can help support us

PAGE 103

91 financially, so we can live with them and they can help with the kids. Sometimes my mom will do grocery drop offs or my dad always helps with like my car and when I need new tires he helps with that and support European mothers. My external social group, like my friends and coffee groups on Facebook and Instagram are my major sources of support. Support outside of the family unit was the most commonly mentioned source of support for European mothers with young children. European mothers often shared that coffee groups, which are facilitated social group s organized by Plunket that bring mothers together on a weekly basis, were a large source of support for them and helped to mitigate their stress. One European through my coffee group and I still keep in touch with all of them. My friends also live quite s a go European mothers also talked about social int eraction as a way to combat a feeling of isolation and promote well being; particularly as a new mother or me. They are just a finger touch away and I talk every day to a handful of my friends many of which have kids and b ant and helps me stay rtant or it can be so isolat Asian m others . My husband really helps me cope. Among Asian mothers, partner support was the most frequently mentioned source of

PAGE 104

92 support. Many talked about having a good husband as the most important support factor that because I have a efinitely my partner gives me support and I found having him around to help is so great Theme 3.2 Neighborhood Integration and Cohesiveness My connection to my neighborhood is a support for me because I think of all the local friends who I see regularly Over half of New Zealand mothers talked about neighborhood integration and connection to their community as sources of support. Mothers shared that having neighbors who are mothers My neighbor is in the same situation a and another said , wonderful neighbors on either side with little ones which is amazing and they help me fee l connected to t and Pacifika mothers felt most integrated and connected to their neighborhoods compared to European and Asian mothers. It was common for mothers from these ethnic groups to have lived in their house or neighborhood their whole lives in my home for 19 years , so I grew up knowing my neighbors all around me and like everyone on my street. Also , my dad is involved in community patrol with the oth er dads, so they are all ve also grown up Pacifika mother said everyone in this community I feel really safe. I grew up here. I feel connected to my neighbor A mother

PAGE 105

93 to my neighbors. We have lived here for a long time, we grew up h ere and our kids go to the same school that my husband went to and so we are very f families in Auckland. Mothers also talked about tra nsitions in Auckland neighbor hoods based on cost of living increases in the city. These neighborhood changes are often associated with new and unique challenges . One My family home used to be like that but now the neighborhood has c hanged an d is very transitional . Another I did feel connected to my community Mothers talked about the importance of knowing their neighbors and how neighborhood t Knowing my neighbors is another thing things have changed so much around Auckland I think our neighborhood i socioeconomically and houses are now getting so expensive. There is so much development going on. But I know most of my neighbors and we support each other. Like if I hear a neighbors alarm go off I will definitely go make sure everything is ok. And we have a community pa trol group because there was some dodgy stuff happening at the end of our stress in an empty lot; gang members were camping there and it was bad. So , we rely on our neighbors to make sure everything is safe when we are out of town and things. We always tex t each other about thin gs going on in our neighborhood. Mothers residing in one neighborhood on the west side of Auckland were particularly defined geographically so th e c ve got extreme poverty and affluence which is obvious by the physical appearance of the houses, we have social housing

PAGE 106

94 and homelessness and affluent houses but the thing about our community is genuine care for others no matter what eighborhood connection is so apparent in my community not much you can there is a sense of being taken care of and general connectedness with all my neighbors. Being connected to people is somethin Another mother and this is such a great little village. Everyone knows each other and you call upon others to help like pick up your kids and such. Just recently with the power outages ve been like , hey can we come take showers at your house and that i Women also shared that a variety of local community resources support them and their well being. One example was a . One mother ave lots of pay it forward pages throughout this area where people take , you can get lots of free stuff from the community and otherwise you would have to buy it. Lots of page s like this out in W est Auckland like , and I needed something for him and put up a post on Facebook and I found one free through a neighborhood f riend online Other mothers talked about community based exercise classes and healthy meal programs as sources of support. One mother se to connect with my community I feel very connected. There is a kaitahe once a month which is a shared meal provided by the community center. They serv Finally, mothers talked about neighborhood activities and resources to support mothers specifically with young children. One mother e how friendly my community is. I

PAGE 107

95 activities for little kids, clean Theme 3.3 Cultural Identity and Norm s I love Auckland because it is really culturally diverse. Auckland is actually one of the most culturally diverse places in the world which brings like, open mindedness. The theme of cultural identity was most prominently discussed as a source of support that fostered a feeling of connectedness among mothers. One mother I have a lot of cultural identity, the best I can describe to you is like how Native Americ ans feel in your country. We are all very connected. Not all s are connected, but I was brought up very old school so I Another mother big because it all comes down to what you and that is a bi Two other way does support me .. . I am part and my family has a strong sense of whanau. I feel really lucky to have time and I do think it could be Finally, on e mother you grow up your cultural connectio Cultural i dentity was also discussed as a source of stress due to the integration of cultures that exists in Auckland. Mothers who identified as more than one ethni city expressed specific challenges related to cultural clashes. One mother I also feel like I have been raised i n a white world because my father is pakeha [European] . I never thought culture was a problem and always felt like I had a strong culture, but my kids father is Tongan and his father was an activist for the Polynesian panthers. He was really affected by culture and differenc h y do you act white and I never understood that.

PAGE 108

96 so someti mother challenge. My son will be raised in a school so he will speak the language and our family wants him to speak it at home which means that I have to learn. So , I think I will take classes once he goes to daycare. There is a A dditional challenges related specifically to language considerations were common in thes e conversations. One mother said, I speak Chinese , and Chinese raise kids differently than kiwi kids so that is sometimes hard, but Plunket nurses sometimes speak Chinese which helps me so much. Now with baby brain, my English is harder I always feel like I misunderstood otherwise that is part cultural identity. I was raised as language and more about when I came from and my c ulture. I was raised as warriors and A loss of cultural identity, particularly among mothers , was discussed as a source . It separates our people. Man , we are all going to die at the end of the day. W that different. Even in terms of language, a lot of my peers say I have to speak ther cultures and that allows you t o Another mother talked about the challenges of exposing her children to T he first t ime my children have been to a M arae was actually last weekend though. It was hard. We stayed two nights and they really struggled. Everyone was up late until 11 or 11:30 and my kids

PAGE 109

97 very differ Case study: A Pacifika mothers experience of family stress and cultural identity Lulu is in her late twenties, has three children under the age of eight, and identifies as Niuean and Tongan. She lives with her three children, younger sister, and her parents in a three bedroom house on the west side of Auckland. Her family has lived in this same house for many generations. She shared her perception of stressors related to family and cultural identity in her culture. But yeah , sleeping with babies after you give birth is more important than sleeping with your partner for the first month and clothing after giving birth; you are always supposed to keep yourself warm and dress warmly after you have a baby. Also, they will use hot water e or sit on a pot of hot water so the steam goes up there. This is really important to clean the body. And yes, we bury the placenta and since this was my nanas home, all the placentas of all the babies b orn in our family are buried here. Also, when you are carrying you cannot wear a necklace. We believe that if you do the placenta will wrap around the babys neck and actually , that happened to one of my friends. You can t drink red stuff or eat red stuff. Which was hard because I really like raspberry flavor. If you do, your kids will come out with rashes and eczema. I have also seen that happen to some of my family members. But there are so many rules and sometimes they are really annoying. I am Niuean an d Tongan and there are so many traditions. What my kids eat is also a stress for my family because like I started introducing food at 4 use a dummy [pacifier] because that wil l also mess with their teeth. The older generation was always sayin g I had to breastfeed. And like, another one is when boys them have fat under their arms like on thei r sides. So , you always have to put their arms down by their sides.

PAGE 110

98 Theme 3.4 Coping Behaviors spirit are in check. This impacts everything and I can multitask better and just live my life better. I asked New Zealand mothers what behaviors help them to cope with stress. Across all ethnic groups the most commonly mentioned behaviors included physical activity, time away from their child/ren, communication with partners and friends, drinking alcohol or smoking cigarettes, hobbies including reading and cooking, and humor. Activity was the most frequently discussed method of coping across all mothers. One mother me more active because I know it will help. My main coping mechanism Another mother talked about how activity decreased stress and depression and did medications, so I increased my activity and took magnesium. A nd it helped so much I got s o much better. So, I know that physical activity is so important for me. And then be bothered but I try to force myself because Taking a walk and leaving the house was the primary way mothers mentioned coping with stress. Mothers talked about community parks and the benefits of just being outdoors. One mother The re are a lot of parks and with that is just fresh air and going fo Many mothers said that it is stressful being in the I alw ays go outside to the park. It is better to be out. The baby tries to do everything the big boy

PAGE 111

99 and go for a walk it helps a lot to burn their energy off. And going to the parks. Just getting out of the house . A small number of mothers also talked about activity related to yoga and mindfulness. One mother space a p p I have heard abo ut mindfulness and yoga and that really helped during pregnancy. Although Taking time away from their child/ren was also a commonly mentioned supportive behavior among all mothers. One mother Daycare and school helps me cope so I have time getting away One talked about the importance of separating herself from her children when stress levels are high. remove myself from the situation Another mother when my child is screaming, I take time awa Mothers also expressed that communication incre ased their ability to cope with stress. Mothers said that realizing others are experiencing the same challenges was helpful. One shared with group so it normalizes it . Another mother experiences and communication like an exchan Domain 3 Summary: New Zealand mothers reported that support helped them cope with stress cifika mothers said family support was most helpful, European mothers said social support outside of the family unit was most helpful, and Asian mothers said partner support was most helpful in effectively coping with stress. Neighborhood integration and cohesiveness or connection to a community was also a prominent source of support mentioned by mothers. Cultural identity was supportive but also

PAGE 112

100 stressful and these associations varied by individual and ethnic group. Finally, mothers felt that exercise and being outdoors, taking time away from their children, and communication were behaviors that supported their ability to cope with stress. Domain 4 : Maternal Stress and Early Childhood BMI : Patterns of risk and protective factors Theme 4.1: Ethnicity and E arly Childhood BMI Mothers were asked to self children under the age of five. BMI percentiles for age were calculated and if at least one child had a BMI higher than the 85 th percentile, mothers were included in the high BMI group. Child mothers (n=17) identified as European, 33% of mothers identified as Asian (n=16), 19% of 9), and 10% of mothers identified as Pacifika (n=5). In the group 45% of mothers identified as Pacifika (n=10), 4.5% of mothers identified as Asian (n=1), and 0% o f mothers identified as European. The analyses below present findings on the differences in patterns of risk and protective factors reported in the normal and high groups; however, it is important to acknowledge that these findings may be highly confounded by ethnicity due to the unequal representation of ethnicities in each group. Theme 4.2 Patterns of Stressors and Risk Behaviors I explored differences in patterns of stressors and risk behaviors between mothers who reported children under five years of ag overweight or obese. The purpose of this was to better inform risk pathways between the experience of maternal stress and early childhood obesity.

PAGE 113

101 Mothers reporting children with high BMIs shared experiences of stress related to culture and relationships more often than mothers reporting children with normal BMIs. Some mothers talked about the cultural norms around eating and events that often have u nhealthy foods, which makes it particularly challenging to adopt healthy eating behaviors. One mother physical health is a stressor. I am doing boot camps and new diets and longing for my older body before having kids. This is hard for going to family feeds and n all like KFC. These are usually a mix of pigs and homemade taro and other things as well as KFC. This is huge in the Polynesian community This group of mothers also talked more about relationship stres s and shared different ways that tension within a partnership impacts health behaviors. One mother share the same values about our children She felt that her partner did not prioritize eating healthy which made it challenging for her cook healthy meals for the family. Mothers also talked about relationship stress and time management in relation to eating behaviors. One mother and by the time w e have eaten and gotten the kids to bed we have like 30 minutes to have a cup of tea and some biscuits to talk about our There were no differences in risk behaviors, including physical activity, healthy eating, breastfeeding , and sleep, between mothers reporting children with normal BMIs and mothers reporting children with high BMIs. Theme 4.3: Patterns of Stressors and Protective Factors I explored differences in patterns of protective factors between mothers who reported ch or obese. The purpose of this was to better inform protective pathways between the experience of

PAGE 114

102 maternal stress and early childhood obesity. Mothers reporting children with normal BMIs discussed social support, social services including Plunket services, and activity as support more often than mothers reporting children with high BMIs. This group of mothers shared that social support helps them cope with stress and social activities are often healthy activities such as walking and spending time outside. One mother s a go to w This group of mothers also talked about Plunket services and the support that Plunket provides related to making healthy choices. One mother great services and resources for new mothers like the Plun Another talked visits you in the first few weeks after the baby is born and they have lactation consultants who lso has a drop in at family centers so you can drop in and sleep for a bit w hile they watch your baby. The P lunket centers are really amazing. I have also met with the karitane nurses to talk about his s Finally, the group o as a coping mechanism more frequently than mothers mother e class (PRAM classes) that are r These types of classes were m stressed I want to be more active. Sort of comes from my husband he is trainin g me for a 12k la Having a social group that supports health behaviors like physical activity helps this group of mothers cope with stress.

PAGE 115

103 Domain 4 Summary: Early childhood BMI varied by ethnicity with and Pacifika groups groups. Mothers reporting children with high er relationships more frequently than mothers reporting children wi no differences in risk behaviors such as inactivity, unhealthy eating, lack of breastfeeding, or lack of sleep. There were differences in protective factors between these two groups. Mothers reporting children with normal BMI

PAGE 116

104 CHAPTER 7 DISCUSSION 7.1 Justification of Research T opic The p rimary objective of this study wa s to employ quantitative and qualitative metho ds to explore associations and pathways between pre and post natal maternal stress and early childhood obesity among an ethnically diverse and representative sample of New Zealand mo thers. Maternal stress is detrimental t o maternal and child health outcomes. However, research exploring the impacts of maternal stress is often limited by the use of differing objective and subjective measures (Kingston et al., 2012; Liu et al., 2016a) . A lack of prospective longitudinal studies that allow for the analysis of critical periods of stress exposure and transitions in stress exposure from the prenatal period through the postnatal period has also made it difficult to understand how timing of s tress exposure can affect these associations (Collins & Manolio, 2007) . New Zealand is a particularly useful sociocultural context in which to explore these associations as well . The (indigenous people of New Zealand) and Pa c ifika (non people of Polynesian descent) communities are exposed to external stressors , such as poverty, unemployment, and overcrowding, that are two to three times higher than members of other ethnic groups, and experience the highest rates of early childhood obesi ty compared to European and Asian families in New Zealand (Perry 2015; Turner and Lloyd 2004; Ministry of Health 2012). This diverse setting allows for exploration of relations hips between maternal stress and early childhood obesity within and between ethn ic groups. The value of this approach is that findings will inform our understanding of variation in risk and protective pathways between maternal stress and early childhood BMI, as well as sociocultural factors that may differentially confer resilience.

PAGE 117

105 7.2 Recap of literature review and methodology Measuring maternal stress is a challenge due to the multiple dimensions of stress and interplay of genes and experience that impact biology and health . Use of both objective and subjective measures can provide in sight into the impacts of objective external stressful events and subjective perceptions of stress on maternal and child health. Due to the intergenerational impact of maternal stress and detrimental impacts on child health and development (Pearson et al., 2016; Talge et al., 2007; Zijlmans et al., 2015) , multiple measures of stress are needed. Maternal stress during the pre and post natal period are known to impact child health, and specifically development of early childhood obesity (Cameron & Demerath, 2002; Farewell, Thayer, Puma, et al., 2018; Farewell, Thayer, Tracer, et al., 2018; Koch et al., 2008; Liu et al., 2016 a; Wu et al., 2017) . However , analyses of the specific timing and duration of stress exposures on early childhood body mass index ( BM I) suggest critical periods of development are unclear. The D evelopmental Origins of Health and Disease (DOHaD) model co ntends that the prenatal and early post natal periods are the most critical in the development of poor child health outcomes (Gluckman et al., 2008) . The Life C ourse Epidemiology model , in contrast, proposes that cumulative maternal stress exposure from the prenatal period throughout the first five years of life and beyond equally impacts health outcomes (Kuh, 2003) . Both models identify potential behavioral and biological pathways that may explain associations between early exposure to maternal stress and childhood obesity. Past studies have found that risk behaviors, including unhealthy eating, lack of physical activity, lack of exclusive breastfeeding, and lack of sleep, may mediate associations between maternal stress and early childhood obesity (Adam & Epel, 2007; Cameron, 2004; Clauss Ehlers et al., 2006; Ding & Gebel, 2012; Dressler et al., 2007; El -

PAGE 118

106 Sheikh et al., 2010; Lobel et al., 2008b; Franco et al., 2010; Mezick et al., 2008; Sallis et al., 2012; Stults Kolehm ainen & Sinha, 2014) . Conversely protective factors, such as social support, cultural identity, neighborhood integration, and household cohesiveness, may moderate these associations and confer resilience (Cohen, S., Gottlieb, B. H., & Underwood, 2004; Horton & Wallander, 2001; Thoits, 1995, 2011; Wadhwa et al., 2001) . These relationships are particularly relevant in New Zealand due to t he high rates of maternal stress and early childhood obesity, socioeconomic and ethnic disparities child health outcomes, and diverse cultural setting which allows for the exploration of mediating and moderating factors that may explain individual variatio n in maternal stress responses and early childhood obesity. I used mixed methods to explore associations and pathways between maternal stress and early childhood obesity. Quantitative data analyses were conducted using the Growing up in New Zealand (GUiNZ ) data set. The GUiNZ study is a prospective longitudinal cohort study that began in 2009 with the recruitment of 6,822 pregnant women in the North Island of New Zealand. This sample represents 11% of all infants born in New Zealand during the study period ( Morton et al., 2 013) . I collected qualitative data from a diverse, convenience sample of women recruited through Plunket Centers in Auckland, New Zealand (N=74). Plunket is a national not for profit organiz ation , community owned and governed and is the largest provider of free support services for the development, health and wellbeing of children under five in New Zealand. The mixed methods approach used in this study allowed for triangulation of data to better understand the complexities surrounding maternal stress exp eriences, early childhood obesity, and mediators and moderators that may explain these associations in this context. 7.3 Brief recap of results There was a moderate , statistically significant positive correlation ( r = .27, p<.01)

PAGE 119

107 between prenatal objective and subjective stress in this sample. Prenatal objective and subjective stress were independently and significantly associated with childhood BMI at 24 and 54 months of age. However, after controlling for covariates including maternal pre pregnancy BMI and gestational weight, subjective stress was no longer significantly associated with early childhood BMI at either time point. The associations between prenatal stress and childhood BMI were stronger at 54 months compared to 24 months. Both the timing an d duration of objective stress from the prenatal through the first 24 at 54 months. Mothers experiencing objective stress during both the pre and post natal period, or at least at one of these time point s wa s significantly associated with higher early childhood BMI at 54 months of age compared to mothers who experience low objective stress at both time points. Among mothers experiencing transitions in objective stress exposure from the prenatal to the postnatal period, t he mean early childhood BMI score for the increased stress group did not differ from the mean childhood BMI score f or the decreased stress group. Next, I explored mediating and moderating pathway s between prenatal objective stress and early childhood BMI at 54 months using structural equation modeling. Risk behaviors, including the number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding, significantly mediated relationships between prenatal objective stress and early childhood BMI; physical activity levels during pregnancy did not. Protective factors, including family support, external support, neighborhood integration, household cohesiveness, and cultural iden tity were explored. Neighborhood integration and family support were significantly and negatively correlated with maternal objective stress and all protective variables were significantly and positively correlated with the number of nutrition guidelines me t during pregnancy. Protective variables were aggregated for path analyses. The aggregated variable did not significantly moderate the direct

PAGE 120

108 path between prenatal objective stress and early childhood BMI at 54 months in the full sample . Path analysis mod els and qualitative data highlight patterns of risk and protective pathways between prenatal stress and early childhood BMI. Integration of quantitative and qualitative methods highlight risk and protective pathways between maternal stress and early childh ood BMI and potential explanations for ethnic variations in these pathways. 7.4 AIM 1 Di scussion Aim 1: Determine whether objective and subjective measures of prenatal stress are correlated among a diverse sample of New Zealand women during pregnancy. There was a positive correlation (r=.27) between objective stress scores and subjective stress scores assessed in this analysis. As individuals experience increased exposure to objective stress during pregnancy, they also report higher subjective stress. H owever, the correlation was low (r=.27), potentially highlighting the discrepancies between these measures. Past studies have found that differences in stress outcomes vary significantly based on the measure of stress employed (Turner & A vison, 2003) . Interestingly, LaPlante et al. ( 2008) found a similar correlation coefficient (.28) between objective and subjective measures of prenatal maternal stress as found in my study. The objective measure used in the LaPlante study related to stress of exposure to an acute disaster, while this study used exposure to objective risk factors. Although the objective measures in these studies vary, the low correlation supports the need for a transactional view of stress that includes measures of both real and perceived environmental demands since objective and subjective measures are potentially measuring differing fa cets of maternal stress (Dressler, Oths, & Gr avlee, 2005) . Responses to external stressors are shaped by individual attitudes and accessibility of social and cultural resources (Desjarlais & Jason Throop, 2011) . Discrepancies between objective risk exp osures and subjective accounts of stress may be

PAGE 121

109 shaped by variation in sociodemographic factors such as ethnicity or age, as well as access to coping resources (Gundersen et al., 2011; Kingston et al., 2012) . 7.5 AIM 2 Discussion Aim 2: Analyze the associations between objective and subjective prenatal stress and early childhood BMI at 24 and 54 months of age. Objective prenatal stress, defined as exposure to nine external risk factors, was positively associated with childhood BMI z scores at 24 and 54 months of age, after controlling for covariates including maternal pre pregnancy BMI and birthweight adjusted for gestational age. These findings align with past research that has found significant associations between exposure to o bjective stress during pregnancy and childhood BMI at age five (King, Dancause, Turcotte Tremblay, Veru, & Laplante, 2012; Liu et al., 2016a) . The association between prenatal objective stress and childhood BMI was stronger at 54 months compared to 24 months. The full model with objective stress and maternal factors explained 8% of variation in BMI at 24 months of age and 17% of BMI at 54 months of age. Objective stress experienced during pregnancy seems to be more strongly associated with BMI as children age. Similarly, Liu et al. (2016) found that the association between prenatal risk and childhood BMI increases with child age. Accumulated exposure to chronic stress throughout the postnatal period of development, which is likely correlated with prenatal stress experience, may explain the stronger associations observed between prenatal stress and childhood BMI at 54 months of age. Alternatively, associations between prenatal stress and BMI may only emerge in later childhood since BMI growth trajectories are so variable in the first few year s of life (Li et al., 2007). Maternal pre pregnancy BMI and birthweight adjusted for gestational age were significantly associated with childhood BMI in the final model . Past studies have found that

PAGE 122

110 maternal BMI and birthweight significantly impact offspr ing overweight and obesity development (Benyshek, 2007; Levin, 2006) . However, exposure to objective stress during pregnancy predicts additional vari ation in childhood BMI levels after accounting for pre pregnancy weight status and birthweight . Exposure to external risks in pregnancy is potentially more impactful on early childhood obesity than elevated maternal pre pregnancy BMI levels and birthweight . Prenatal subjective stress was not significantly associated with childhood BMI at 2 4 or 54 months of age, after controlling for covariates . Similarly, Laplante et al. (2008) found that although objective prenatal stress was associated with child develo pmental outcomes, subjective stress was not significantly associated with developmental outcomes at 24 or 54 months of age. Alternative studies c ontradict these results and found significant associations between perceived stress in pregnancy and developm ental risk factors in early childhood, including risk of preterm birth and low birth weight (Dole, 20 03) . Wadhwa, Sandman and Garite (2001) found strong correlations between perceived stress in pregnancy and child cortisol levels highlighting the importance of measuring the appraisal of external stressors . The minimal associations between subjective me asures of stress in pregnancy and childhood BMI at 24 and 54 months of age in the GUiNZ sample suggests that the external environment may shape childhood obesity independent of the appraisal of stress. These findings support the DOHaD hypothesis that pr enatal exposures impact child growth and development. Multiple pathways linking prenatal stress to early childhood obesity have been proposed . Prenatal stress may impact childhood obesity through heightened glucose levels that impact metabolism (Harris & Seckl, 2011) . Consistently high blood glucose levels combined with insulin suppression leads to cells that are starved of glucose due to de -

PAGE 123

111 sensitizatio n of glucose receptors . The energy requirements of these cells may send hunger si gnals to the brain, thereby resulting in overeating. Overeating leads to increased gestational weight gain, which is associated with early childhood obesity (Fraser et al., 2010). Finally , cortisol may directly influence appetite and cravings by modulating other hormones and stress responsive factors that stimulate appetite (Epel et al., 2001; Lumeng et al., 2014). While it is clear that physiological mechanisms can link prenatal stress to childhood BMI , the potential effects of post natal behavioral factors also need to be considered (Entringer, Buss and Wadhwa, 2010). 7.6 AIM 3 Discussion Aim 3: Examine associations between the timing and duration of maternal stress exposure during the pre and post natal p eriods of development and early childhood BMI at 54 months of age. I conducted regression analyses to explore the impact of both timing and duration of objective stressors on early childhood development. Exposure to objective stress during either the pre or post natal period of development was significantly and positively associated with childhood BMI at 54 months of age ( =.06, p<.01) . Children who experienced maternal pre or post natal objective stress were more likely to have higher BMI at 54 months of age compared to children who experienced stably low maternal objective stress in this sample (p<.01). One proposes that a fetus exposed to maternal stress, such as living in a high deprivation area, might adapt by developing an energy sparing metabolism, which can be maladaptive in a Westernized environment characterized by energy abundance. A mismatch between the expected later environment and actual circumstances experienced may increase chances for disease thus

PAGE 124

112 decreasing chance s for survival (Brakefield, Pijpe, & Zwaan, 2007) . Alternatively, these findings signify that cumulative exposure to maternal stress during the pre and post natal period has the greatest impact on BMI in early childhood. Whi le both pre and post natal stress independently predicted child BMI at 54 months of age, the timing of stress exposure did not lead to significant differences in childhood BMI. More specifically, there were no significant differences in childhood BMI at 5 4 months of age between mothers whose stress exposure increased compared to those whose stress exposure decreased from the prenatal period to the 24 month period. Exposure to prenatal stress and postnatal stress at 24 months of age may have the same impact on childhood BMI at 54 months. These findings contradict studies that have hypothesized that the prenatal period is the most critical period of development (Amugongo & Hlusko, 2013; Cao Lei et al., 2015; Heindel & vom Saal, 2009) . However, pre and post natal objective stress scores in this sample, and likely others, are highly correlated (in this study: r=.81, p<.01). Further examination of the impacts of timing of stress on childhood BMI requires a sample of mothers who experience higher rates of trans itions in stress exposure. This may be found in large samples of migrants or refugees, as well as following natural or human induced disasters. The duration of stress exposure was significantly positively associated with early childhood BMI, consistent with the cumulative stress model (Nederhof & Schmidt, 2012) . Children whose mothers experienced objective stress during both the prenatal and 24 month period of data collection were signifi cantly more likely to have higher BMI compared to children who experienced no maternal stress at both time points or whose mothers experienced stress at only one time point. This suggests that chronic exposure to stressors during the pre and post natal per iods of development have greater impacts on childhood BMI than stressors experienced

PAGE 125

113 at either time point in isolation. Contrary to these findings, Suglia et al. ( 2012) found that while stress experienced at one and three years of age independently predicted childhood ob esity at age five, cumulative stress was not associated with childhood obesity at age five. The measures of stress varied between these studies and included bereavement, stressful life events, and exposure to a natural disaster, which may explain these dis similar findings. Further exploration of the impacts of accumulated stress exposure on early life development, particularly with respect to childhood BMI, is needed. 7.7 AIM 4 and AIM 5 Discussion Aim 4: Analyze risk and protective factors that may mediate or moderate associations between prenatal stress and early childhood BMI at 54 months of age using structural equation modeling. Aim 5: Explore the lived experience of stress and associations with childhood BMI among a diverse group of New Zealand mothers This section summarizes and integrates the findings from AIM 4 (quantitative findings) and AIM 5 (qualitative findings). 7.7.1 Sources of Maternal Stress Q ualitative findings inform our understan ding of the lived experience of stress and unique sources of stress exposure among mothers with young children in New Zealand. Financial stress was the most commonly mentioned source of stress among mothers across all ethnic groups and was often tied to h ousing concerns and overcrowding. I conducted many interviews in and saw multiple examples of mothers living with their partners, parents, siblings, and children in one or two bedroom houses. The average cost of rent for a two bedroom house in West Auckland, which is one area with a large concentration of

PAGE 126

114 higher deprivation neighborhoods , is $430/week or $1720 NZD/month (~$1200 USD/month) (Carroll, Witten, & Kearns, 2011; ENZ, 2018) . This is an extreme cost for low and middle income families with young children, particularly in situations where only one parent can work outside the home. Extended families often live together to support each other and share the rent. These findings al ign with a previous qualitative study that found financial stress, and specifically affordability and availability of housing, was a considerable challenge and source of stress for families living in Auckland (Carroll et al., 2011) . Another frequently mentioned source of maternal stress was work and government assistance related concerns. In New Zealand, m ost jobs allow for up to 12 months of parental leave; the government provides up to 5 months of parental leave. However, many mothers felt r that the government stipend was not enough to support living in Auckland. Additionall y, there are specific requirements for receiving the benefit which many mothers mentioned were barriers and align with previous studies (Kahu & Morgan, 2007) . These barriers include: difficulty completing the paperwor k; training and coursework requirements when receiving the benefit, making it challenging to juggle childcare and other responsibilities; restrictions on additional income; and societal pressure to work outside the home. Additionally, poor employment condi tions and partner unemployment were discussed as sources of stress and have previously been found to be associated with postnatal depression among mothers in New Zealand (Schmied et al., 2013) . Work and government assistance related stress were often interrelated with financial stress and were experienced by most mothers in the qualitat ive sample, regardless of ethnicity. Mothers shared stories related to cultural norms and traditions that impacted their experience of stress. Pacifika mothers talked about some of their cultural traditions in the context

PAGE 127

115 of financial stress. For example , a few mothers shared norms and challenges around gift giving, which exacerbate the experiences of financial stress in their communities. In many Pacifika cultures, it is a significant insult to turn down gifts, whereas accepting gifts often means accepti ng future debt (Tracer, 2004) . Particularly in Samoan communities, reciprocal gift giving arrangements between extended f amily members are a fundamental tradition of the , or chief system . A recent study found that these traditions are changing as families choose to family (Thornton, Kerslake, & Binns, 2010) . This conflict between traditional and modern customs may lead to familial challenges and additional stressors surrounding money. A lthough financial stress was a common theme across all ethnic groups in this qualitative sample , and European mothers shared experiences of financial stress more often than Asian and Pacifika mothers. This suggests that it is likely that a constellation of factors contribute to the experience of financial stress among mothers with young children livi ng in Auckland, New Zealand. ethnic group. Stress related to family, financial concerns, and partner relationships were the most commonly discussed sources of stress among this group. These results align with previous findings that suggest overcrowding, which is linked to family and financial stress, compared to European and Asian groups in New Zealand (Schluter, Carter, & Kokaua, 2007 ) . Extended families oft en live responsibility of grandparents is to guide the children in the Whanau unit. Elders, or tupuna , are seen as educators, mentors and role models (Edwards, Mccreanor, & Moewaka Barnes, 2010) . Moko typically means a facial image and

PAGE 128

116 puna refers to a spring or a pond. Moko puna signifies an image reflected in a pond and believe that the grandchild is a reflection of the grandparent (Edwards et al., 2010) . Multiple generations of families living together is not only financially beneficial but ali gns with these cultural norms and beliefs. These cultural norms may partially explain the stress experience of however, these norms may also lead to enhanced social su pport. This theme is discussed in more detail below. Relationship intimate partner violence (IPV). For example, a large longitudinal study in Ne w Zealand found that there is an uneven distribution of exposure to IPV, which is related to relationship stress, by the national average (Marie, Fergusson, & Boden, 2008 .; MSD, 2018 ) . One hypothesis that may sadvantage more than most other ethnic groups in New Zealand, and rates of IPV are higher among low socioeconomic communities (Marie, Fergusson, & Boden, 2008 .; MSD, 2018 ) . The qualitative sample of women in this study were from socioeconomic disadvantaged neighborhoods, so the elevated this hypothesis. Another possible explanation is an inter generational relationship in which rates of experiencing IPV in adulthood (Marie, Fergusson, & Boden, 2008 .; MSD, 2018 ) women experience higher exposure to all forms of violence in childhood compared to any other ethnic group in New Zealand (Ministry of Health, 2017; MSD, 2018 ) . Finally, it has been

PAGE 129

117 Past and present colonization in New Zealand has contributed to the loss of land, language, beliefs and cultural identity of as well as the breakdown of traditional structures within . Poverty and decreased family and community support accentuated by the impacts of colonization may explain higher repo rted levels of relationship stress, as well as family and financial stress , experienced in this community (Penehira & Doherty, 2013) . Although and Pacifika mothers experienced similar levels of objective stress in the GUiNZ data set, mothers in the qualitative sample discussed exposures and sources of stress more frequently than Pacifika mothers. Appraisal of health reflects cultural values and norms surrounding health and wellness (Kandula, Lauderdale, & Baker, 2007) . Exposure to external thus decreasing the negative appraisal of stress . Pacifika mothers may also have increas ed access to coping resources which mitigates their stress experience. Finally a cculturation, or the process of change that takes place because of interactions between two or more cultural groups, may partially explain some of these findings (Berry, 2005) . Past studies suggest that among individuals who identify as Pacifika in New Zealand, those who retain strong alignment to Pacific culture ha ve better infant and maternal health outcomes compared to individuals who integrate or assimilate into the dominant culture (Borrows, Williams, Schluter, Paterson, & Helu, 2010) , perhaps due to less acculturative stress. While not assessed here, it is possible that the Pacifika mothers in the qualitative data set retain stronger links to their cultural norms and as a protective factor and decrease their subjective experience of stress. The role of cultural identity and associa tions between stress and early childhood obesity is discussed in more detail in Section 7.7.3.

PAGE 130

118 European mothers shared experiences of stress related to and societal pressure s more than all women from the other ethnic groups in this qualitativ e sample . Prior qualitative studies conducted with European mothers in New Zealand discovered a similar theme (Kahu & Morgan, 2007) . Mothers fe lt that the societal pressure surrounding being a mother is extreme and th at being a good mother is compatible with also working outside the home. These studies also found that mothers felt societal guilt whether they chose to stay home with their children or to seek paid employment when their children are under the age of five. There is a perception in New Zealand that stay at home mothers are not valued ( Kahu & Morgan, 2007) . societal pressures related to motherhood by race or ethnicity in this cultural context. European mothers may feel more general mom related stress and guilt due to their social networks and norms within those networks. In the qualitative sample, many Eur opean mothers shared feeling pressures from within their own support groups. The cultural norms of Europeans may prioritize working outside the home while raising children more than other ethnic groups in New Zealand. European mothers may also feel that th ey are held to higher expectations including balancing social life, work life, home life, and family life, and that these expectations are unattainable. 7.7.2 Risk Behavior Pathways between Maternal Stress and Early Childhood BMI The previously discussed regression models (Aims 2 and 3) suggest that the associations between objective measures of stress and BMI at 54 months are strongest compared to subjective stress measures and BMI at 24 months of age. Therefore, the primary independent and dependent variables for all path analysis models were objective prenatal stress and childhood BMI at 54 months, respectively. The path analysis model with the full quantitative sample

PAGE 131

119 illustrates that prenatal stress was positively and significantly associated with early childhood BMI at 54 months of age. Mediation pathways were explored. The numbe r of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding significantly mediated this relationship; higher levels of prenatal stress were associated with lower numbers of nutrition guidelines met and shorter lengths of exclus ive breastfeeding, and these risk behaviors were negatively associated with early childhood BMI at 54 months in the full quantitative sample. There was a significant, but low, positive correlation (r=.04, p<.05) between these two mediating variables. The number of days mothers reported being moderately to vigorously physically active during pregnancy did not significantly mediate this relationship. The qualitative findings somewhat align with the quantitative results and provide additional insight related to risk behavior pathways between prenatal stress and early childhood BMI. New Zealand mothers shared that eating behaviors were the most commonly affected health behaviors when experiencing stress. Mothers talked about their likelihood of cooking less an d purchasing takeaways (take out food), consuming high sugar and high fat foods, and eating at irregular hours during periods of high stress. These data support the significant mediation of stress and BMI by eating behaviors. Maternal eating behaviors impa ct the consumption behaviors of young children, which may translate to elevated BMIs in early childhood and obesity. A study of New Zealand families with young children found that healthy eating guidance and monitoring by parents was related to the consump tion of fewer unhealthy foods and that a lack of parental control was related to a higher intake of unhealthy foods (Ha szard, Skidmore, Williams, & Taylor, 2015) . Parental knowledge and modeling of healthy eating pathway.

PAGE 132

120 However, there was inconsistencies between quantit ative and qualitative findings when stratifying the sample by ethnicity. Although in the qualitative sample the impact of stress on the analyses were explored within each ethnic group, the mediation pathway was significant among guidelines met significantly mediated associations between maternal stress and early childhood BMI am ong Pacifika mothers. A previous study found that across all ethnic groups in New takeaway shops, which is associated with higher BMIs (Utter, Scragg, Schaaf, & Fit zgerald, 2006b) . The elevated experiences of external stressors may further promote these unhealthy eating behaviors, findings do not entirely align with respect to ethnic differences in this mediation pathway, both methods suggest that in the overall sample, healthy eating behaviors partially explain associations between maternal stress and early childhood BMI. Most mothers shared that stress significantly impacte d their eating behaviors; however, there were no significant differences in these behaviors between mothers who had children with normal BMIs compared to children with high BMIs. Biological and alternative behavioral influences, such as breastfeeding and s leep behaviors, may mediate associations between maternal stress and early childhood BMI and help to explain this relationship. The length of exclusive breastfeeding significantly mediated pathways between maternal stress and early childhood BMI in the q uantitative sample. Qualitative results supported these findings. In the qualitative sample, many mothers shared how stress, and specifically societal

PAGE 133

121 pressures, impacts their ability to produce milk and breastfeed exclusively for extended periods of time. Multiple barriers and stressors related to breastfeeding were shared across all ethnic groups. Studies exploring associations between maternal stress and milk production confirm these data. One study found that poor maternal mental health correlated negat ively with milk production and that these associations varied by ethnicity (C. Lau, Hurst, Smith, & Schanler, 2007) . Mothers also talked about stress surrounding breastfeeding guidance and pressure from healthcare providers. Conflicting advice from health care workers and a lack of professionalism may contribute to maternal stress and lower rates of breastfeeding in New Zealand (Manhire, Hagan, & Floyd, 2007) . New Zealand public hospitals are working on implementing Baby Friendly Hospital Initiatives . However, one study found significant variation in the implementation and promotion of these policies. This may contribute to the stress mothers feel related to conflicting information about breastfeeding from healthcare professionals (Moore, Gauld, & Williams, 2007) . Past studies that have explored factors associated with breastfeeding in New Zealand have found that maternal deprivation is associated with lower rates of breastfeeding (Landy, Sword, & Valaitis, 2009) . Mothers living in high deprivation neighborhoods are more likely t o smoke (Schaap & Kuns t, 2009) and less likely to receive antenatal care (Essex, Counsell, & Geddis, 1992) which may explain associations between stress and lack of breastfeeding found in this study. Past studies also found that maternal stress related to milk supply, painful breastfeeding, press ure and conflicting advice from healthcare professionals, the obligations of paid employment, and societal stigma around breastfeeding in public may contribute to a lack of breastfeeding among New Zealand mothers further validating the findings from this q ualitative

PAGE 134

122 sample (FORD et al., 1994; Landy et al., 2009) . Quantitative fin dings suggest that the relationship between stress, breastfeeding and BMI shared experiences of stress and impacts on breastfeeding most frequently. A recent stud y among New Zealand mothers found that Pacifika, and Asian mothers are least likely to exclusively breastfeed compared to Europeans (Castro et al., 2017) . These findings suggest that cultural influences surrounding breastfeeding may impact variations of stress, breastfeeding, and early childhood BMI across all mothers live with their extended families and older, more traditional generations may have contrasting beliefs about breastfeeding compared to the younger, contemporary generation. Past studies have they have trouble exclusively breastfeeding due to multiple barriers including lack of support, lack of timely information, confusion about smoking during breastfeeding, uncert ainty about the safety of bed sharing, and a perceived lack of acceptability of breastfeeding in public (Glover, Waldon, Manaena Biddle, Holdaway, & Cunningham, 2009) . Nevertheless, across all ethnic groups in both the quantitative and qualitative samples, higher levels of maternal stress result in shorter lengths of exclusive breastfeeding, suggesting that all mothers experience barriers making it difficult to exclusi vely breastfeed for 6 months. The role of physical activity as a possible mediator between maternal stress and early childhood obesity was also explored. In the quantitative models, the number of days moderately or vigorously active during pregnancy was not a significant mediator. As a result, this variable was dropped from the model. Qualitative findings support this finding; New Zealand mothers discussed the impact of stress on physical activity levels with less frequency than all other risk

PAGE 135

123 behaviors. Although one study conducted in Australia found significant positive associations between high levels of life stress experienced by mothers with young children and physical activity levels (Craike, Coleman, & MacMahon, 2010) , no studies to date have been conducted with a large sample of New Zealand mothers of young children. These mixed methods findin gs suggest that although experiencing stressors as a mother of young children may detrimentally impact physical activity levels, alternative health behaviors such as healthy eating and breastfeeding are more severely affected. Maternal sleep behaviors wer e not available in the quantitative data set. However, qualitative findings suggest that higher levels of maternal stress lead to shorter sleep duration. Shorter sleep duration is associated with obesity which may be explained by the increased consumption of unhealthy foods and inactivity that often co occur with less sleep (Nielsen, Danielsen, & Sørensen, 2011) . Lack of sleep detrimentally impacts the physical health behaviors of the mother which may also lead to role modeling of unhealthy behaviors and negative consequences for young children. Asso in the qualitative sample. The associations between stress and lack of sleep may be more prominent among mothers d ue to specific sources of stress previously discussed, i ncluding family stress and overcrowding, relationship stress, and work related stress. One study in New groups in New Zealand (Gander, Marshall, Harris, & Reid, 2005) . If a family is co habituating and sharing sleeping spaces, sleep may be more difficult. The higher rates of overcrowding milies may explain the associations between stress and sleep being (Schluter et al., 2007 ) .

PAGE 136

124 The mixed methods findings indicate that among New Zealand mothers, higher exposures to stress are associated with lower numbers of nutrition guidelines met and shorter periods of exclusive breastfeeding. These behavio rs significantly impact early childhood BMI at 54 months of age. Maternal physical activity levels do not significantly mediate the associations between maternal stress and childhood BMI in these data. Although there are variations in these behavioral path ways according to maternal ethnicity, the findings in this study do not provide evidence that these pathways are unique to any one ethnic group. Across both quantitative and qualitative samples, mediation of stress and BMI by eating behaviors and exclusive to Asian mothers. However, Asian mothers experience the lowest rates of stress, lowest rates of adult obesity, and the lowest early childhood BMI levels in New Zeala nd. The direct association between stress and BMI is weakest within this ethnic group which may explain the insignificant direct and indirect pathways observed among Asian mothers. 7.7.3 Protective Factor Pathways between Maternal Stress and Early Childho od BMI Protective factors, including social support, family support, household cohesiveness, neighborhood integration and cultural identity, were explored as an aggregate variable for path analysis models based on model fit criteria. The protective factor variable did not significantly moderate the direct pathway between prenatal stress and early childhood BMI in the full sample. However, these protective factors were positively and significantly associated with the number of nutrition guidelines met during pregnancy. These protective factors may moderate the mediated effects of prenatal stress on early childhood BMI, transmitted primarily through nutrition behaviors during pregnancy. These protective factors were not significantly associated with the lengt h of exclusive breastfeeding in the overall path model. Although exposure to more

PAGE 137

125 protective factors may result in healthier eating behaviors, in turn buffering the impact of prenatal stress on early childhood BMI, the overall association between prenatal stress and early childhood BMI does not vary by level of protective factors. These protective factors differentially impact the direct and mediation pathways between maternal stress and early childhood BMI by ethnic group. Among European and Asian mothers , protective factors did not significantly impact early childhood BMI . mothers, protective factors did significantly impact early childhood BMI; higher levels of protective factors were associated with higher early childhood BMI at 54 months of age. However, protective factors significantly interacted with maternal stress and moderated the direct path between stress and BMI solely among Pacifika mothers. This suggests that although protective factors are positively and significantly associated with child BMI independently, these factors also help to buffer the experience of maternal stress and resulting impact on child BMI. Associations between protective factors and healthy eating behaviors were also significant among women from al l ethnic groups. Among European and Asian mothers, higher levels of protective factors were significantly associated with more nutrition guidelines met significant ly associated with less nutrition guidelines met during pregnancy. These findings contradict the hypothesis that more protective factors promote the adoption of healthy behaviors and buffer the impact on early childhood BMI among all New Zealand mothers. I t also provides insight into the relationships between stress, healthy eating behaviors, protective factors and BMI specifically among Pacifika mothers: Protective factors negatively impact healthy eating behaviors which translates to high childhood BMI. H owever, they also can buffer the experience of maternal stress, thus mitigating the direct pathway between stress and obesity.

PAGE 138

126 Qualitative methods allowed for the identification of possible explanations for these quantitative findings as well as explorati on of patterns within and between ethnic groups related to individual protective factors. To further understand pathways between protective factors, maternal stress, risk behaviors, and BMI, mothers were asked about their experiences with each individual p rotective variable. Mothers in New Zealand talked about social support, including family and external support, and neighborhood integration as primary sources of support. Cultural identity was discussed equally as a source of support and a source of stress . about family support as their primary source of support for helping them cope with stress. These findings align with research that has explored the role of family within groups. culture. Ratima and Grant ( 2007) perspective is s own whanau , and in a broader sense it emphasizes s connection to all peoples and all The focus on the familial unit and supporting the family is paramount within culture . also place high significance on whanaungatanga , or maintaining high quality social connections , as well as identification with their family background such as waka (canoe), iwi (tribe), and hapu (sub tribe) (Brougham & Haar, 2013) . Among Pacifika women, strong family and church networks that prioritize the continuation of traditional beliefs and practices throughout the extended family unit is a cornerstone of Pacifika life in New Zealand (Abel, Park, Tipene Leach, Finau, & Lennan, 2001; Podsiadlowski, 2011) . Th ese collectivist, family centered cultural models are fundamental ly differen t from the concept of individuality often found in European cultures. Pakehas , or Europeans, in New

PAGE 139

127 Zealand prioritize individuality more than indigenous and Pacific Islander groups; this is often rooted in autonomy and self interest. This individualism may result in decreased access or use of family support and connection (Hook, Waaka, & Raumati, 2007) . One study of European mothers in New Zealand found that social support (external to the family) moderated maternal stress and attenuated negative child development outcomes related to academic achie vement (Slykerman et al., 2005) . Although this study did not explore these associations in other ethnic groups, these findings substantiate the protective role of external social support found among European mothers in New Zealand. A large research team in New Zealand collected data to understand social capital and social support differences between ethnic groups (Social Capital Programme Team, 2001) . They found that the European perspective of social capital i s primarily capital as created within the community made up by the extended family network. This aligns with findings from this qualitative sample because Europea n mothers discussed external support, primarily support from friends, more than family support. T hese variations within cultures cou ld Pacifka mothers compared to Euro pean and Asian mothers. Within the qualitative sample, Asian mothers talked about the role of partner support more than support from their families or external social groups. Previous studies have found that among Asian mothers, family harmony is of utmost importance and that clo se family ties are highly valued (Lee, Pomeroy, & Bohman, 2007; Yamashiro & Matsuoka, 1997) . There may also be an ideology focused on self reliance among Asian cultures in New Zealand. A study conducted in the United States proposed that help seeking behaviors differ in American and Asian cu ltures; self reliance and mistrust towards outside institutions may be more common

PAGE 140

128 among Asian mothers compared to American mothers (Yamashiro & Matsuoka, 1997) . These studies may explain the reliance on partner support over external social support and support services in this sample. The role of neighborhood integration in mitigating the experience of stress among New Zealand mothers was also explore d qualitatively. integrated and connected to their neig hborhoods compared to European and Asian mothers. (Witten, Kearns, McCreanor, Penney, & Faalau, 2009) . Studies in New Zealand have found that neighborhood connection matters for mothers with young children and that communities in Auckland are a significant nexus for social support for all ethnic groups (Witten et al., 2009) . Accessible public meeting spaces were found to be a primary source of community cohesiveness. These findings suggest that neighborhood integration and cohesiveness can help to buffer the negative impacts of stress on risk behaviors. others may benefit most significantly from this connectedness to their neighborhoods in Auckland. New Zealand mothers shared that cultural identity helped them cope with stress while identity as a source of support more than all other ethnic groups. These findings align with past studies that have explored the role of cultural identity in mental health and well being among diverse groups. Cultural identity, including cultural knowledge and language skills, buffers maternal stress and (Brougham & Haar, 2013; Keddell, 2007) . For these women, abiding by cultural norms and understanding their cultural heritage may lead to an increased identification and comfort level with their cultural group, which may enhance well -

PAGE 141

129 being (Oishi & Diener, 2009; Social Capital Programme Team, 2001) ) . Suh et al. (1998) contends that both awareness of social norms leads to higher levels of happiness. Curr ently in New Zealand, there is a rapid evolution of cultural identities influenced by globalization and migration, resulting in a unique mixing of cultures and expression of new cultural forms (Keddell, 2007) . This leads to new cultural identities and senses of self, which may increase stress and decrease the protective effect of culture. Due to the complexity of cultural identity and sh ifting cultural landscape in New Zealand, cultural identity can serve as both a source of stress and as a source of support simultaneously for mothers with young children. Mothers who identified as more than one ethnicity expressed specific challenges rela ted to cultural clashes. Acculturation, and stress experienced by mothers because of acculturation related factors, provides insight into the complexities of cultural identity and maternal stress, particularly among mothers who identify with multiple ethni cities. Mothers who maintain ties with their original cultural identit y may report higher levels of cultural identity . As mentioned previously, this may serve as a protective factor and buffer the associations between maternal stress and early childhood o besity. This may be particularly pronounced among Pacifika mothers, which would explain the buffering effect of high levels of protective factors on maternal stress in this group. Pacifika mothers in this sample may have retained more traditional values co mpared to the other ethnic groups, which could help them to cope with stress. This aligns with findings in section 7.7.1 that suggest Pacifika mothers perceive lower levels of stress s. Mothers who identify as having multiple ethnicities may report lower cultural identity due to attempts to assimilate into one culture or integrate multiple cultures, thus causing more stress and leading to

PAGE 142

130 poorer maternal and child health outcomes (Berry, 2005; Borrows et al., 2011 ) It is also likely that acculturation may be a risk factor for some individ uals and ethnic groups, while a protective factor for others. Further research is needed to understand the direct and indirect effects of culture on maternal stress and child health outcomes (Fox, Thayer, & Wadhwa, 2017) . The five protective factors included in these analyses may be significant sources of both support and stress for New Zealand mothers. Additionally, these factors may be associated with an increased exposure to unhealthy eating behaviors, resulting in highe r childhood BMIs, among may also explain some of these findings. A small sample of mothers shared aspects of and Pacifika culture and community gatherings tha t may be associated with unhealthy eating behaviors and early childhood obesity. For example, one mother talked about a personal challenge surrounding eating healthy foods because of the frequent family events and the typical food, such as Kentucky Fried C hicken, often served at these events. Higher levels of cultural identity, family support, and neighborhood integration in these ethnic groups may lead to unhealthier consumption behaviors. populations may also Pacifika women may value a fuller figure body as it can represent high status, power, authority and wealth compared to European and Asian women who place more value on low body weight and thinness (DIJK & N., 1991; Metcalf, Scragg, Willoughby, Finau, & Tipene Leach, 2000; Pollock, 1995) acifika women perceive themselves at a healthy weight, suggesting that cultural factors influence perception of body size and health behaviors within this population (Metcalf et al., 2000) .

PAGE 143

131 Finally, generational and gender di fferences in eating behaviors may also partially explain the unique relationships between protective factors, eating, and early childhood BMI family conflict related t o child consumption behaviors. Many mothers talked about challenges with grandparents feeding their children sweets and refusing to modify these behaviors. As Paci fika families (Pinson Millburn, Fabian, Schlossberg, & Pyle, 1996; Worrall, 2009) . A qualitative study found that the feeding practices of grandparents in these communities are influenced by objective elements including income, cultural norms and customs, and government policies (Tapera, Harwood, & Ander son, 2017) . Although grandparents may be knowledgeable being, they often feel limited by economic and material factors, including financial struggles. This may explain the significant associations between protective factors, unhealthy eating behaviors, and early childhood obesity. Pathways between prenatal stress and early childhood obesity are complex and the sociocultural context shapes risk behaviors and protective factors among New Zealand mo thers. Prenatal stress may result in unhealthy maternal eating behaviors and shorter periods of exclusive breastfeeding, leading to elevated levels of childhood BMI early in life. Social support, neighborhood integration and cultural identity may buffer th ese relationships; however, these buffering effects vary by ethnic group. The mixed methods findings of this study suggest that maternal stress is detrimental for child health and development and that interventions must consider cultural norms and values r elated to the experience of stress, health behaviors, and supportive factors that can confer resilience among mothers with young children in New Zealand.

PAGE 144

132 CHAPTER 8 CONCLUSIONS AND IMPLICATIONS 8.1 Final Summary of Research Questions and Overall Findings Th e primary objec tive of this study wa s to use quantitative and qualitative methods to explore associations and pathways between pre and post natal materna l stress and early childhood obesity among an ethnically diverse and representative sample of New Zealand mothers. There was a moderate , positive correlation between prenatal objective and subjective stress in this sample. Prenatal objective and subjective stress were independently and significantly associated with childhood BMI at 24 and 54 months of age. However, after controlling for covariates including maternal pre pregnancy BMI and gestational weight, subjective stress was no longer significantly associated with early childhood BMI at either time point. The associations between prenatal stress and childhood BMI were stronger at 54 months compared to 24 months. Both the timing and duration of objective stress from the prenatal through the first 24 months. Mothers experiencing objective str ess during both the pre and post natal period, or at least at one of these time points wa s significantly associated with higher early childhood BMI at 54 months of age compared to mothers who experience low objective stress at both time points. Among moth ers experiencing transitions in objective stress exposure from the prenatal to the postnatal period, t he mean early childhood BMI score for the increased stress group did not differ from the mean childhood BMI score f or the decreased stress group. Mediatin g and moderating pathways between prenatal objective stress and early childhood BMI at 54 months were explored. Risk behaviors, including the number of nutrition guidelines met during pregnancy and the length of exclusive breastfeeding, significantly media ted relationships between prenatal objective stress

PAGE 145

133 and early childhood BMI; physical activity levels during pregnancy did not. Protective factors, including family support, external support, neighborhood integration, household cohesiveness, and cultural i dentity were explored. Neighborhood integration and family support were significantly and negatively correlated with maternal objective stress and all protective variables were significantly and positively correlated with the number of nutrition guidelines met during pregnancy. Protective variables were aggregated for path analyses. The aggregated variable did not significantly moderate the direct path between prenatal objective stress and early childhood BMI at 54 months in the full sample . Path analysis models suggest that risk and protective pathways between prenatal stress and early childhood BMI vary by ethnicity. Qualitative data fu rther supports these findings and implies significant variations in the lived experience of stress 8.2 Reliability , Validity and Generalizability Reliability The quantitative findings are highly reliable and replicable within the GUiNZ data set. I was not involved in the quantitative data collection phase ensuring internal researcher bias did not impact the data. I worked to recruit participants for the qualitative component of this study through my New Zealand contacts, including a New Zealand consultant, to improve the rapport with participants and quality of the data collected. mothers may limit or alter what they choose to share based on my background. Although I used a purposive sampling design to recruit participants, a similar sample size was collected from each of the four primary ethnic groups in New Zealand. Since the goal was to explore patterns within and between these specific ethnic groups, this representative sample was appropriate. My data collection procedures and protocols were similar across participants and I analyzed data within

PAGE 146

134 48 hours of collection using consistent coding methods in NVIVO. I also reviewed the findings with my New Zealand consultant to minimize the impact of any internal biases that may have check for internal consistency or reliability (Jick, 1979) . Validity Most of the variables used in the GUiNZ data set (e.g. perceived stress scale, childhood BMI) are widely used and therefore have high construct validity; however, the objective stress measure (GUiNZ vulnerab ility scale) is context specific. This scale is not validated with other objective stress measures (e.g. number of stressful life events scales) but this set of risk factors have utility because they are routinely available and measured in a standard way a cross the diverse New Zealand population. Although this may impact validity and generalizability of the findings, the use of these vulnerability risk factors in many other studies suggests that they are important indicators of objective and objective stress in varying contexts. A purposive sampling plan for the qualitative component was an appropriate sampling strategy because the study was based on a priori findings and hypotheses from prior literature and the quantitative data (L. Leung, 2015) . While conducting the focus groups and interviews, I checked for respondent validity by repeati ng and summarizing responses and asking for comments and clarification (Noble & Smith, 2015) . I also reviewed themes with my New Zealand consultant throughout the analysis phase. She helped to interpret the findings through triangulation of my quantitat ive and qualitative findings suggest that my qualitative findings are valid and reflect true patterns related to the lived experience of stress among mothers in New Zealand.

PAGE 147

135 Generalizability The findings from this study align with past research suggesting that objective stress is associated with early childhood BMI. These studies have used varying techniques and scales to measure objective stress (e.g. number of stressful life events, bereavem ent, exposure to a natural disaster). This suggests that exposure to external risks early in life is significantly associated with early childhood obesity across diverse populations. employed to recruit the GU iNZ cohort participants. The advantage of this sampling method was the engagement with participants which translated into a high retention rate as well as cost and time savings. GUiNZ researchers analyzed the alignment of the cohort to New Zealand births d uring the same period and found that birth parameters were closely aligned to the full population of New Zealand from 2007 to 2010. The researchers concluded that data from the cohort can be extrapolated to the general population and inform policy developm ent relevant to the diversity of New Zealand (Susan M B Morton et al., 2015) . Generalizability is often not the primary goal of qualitative research, particularly when the specific sociocultural context impacts the primary variables of interest (L. Leung, 2015) . The purpose of the qualitative component of this study was to understand the lived experience of maternal stress and variations between individuals and ethnic groups specific to New Zealand. I recognize that because of the sampling methodology, the findings may not be generalizable to mothers in New Zealand who are more isolated and/or who choose not to access government funded services. Triangulation of data using qualitative and quantitative methods improved the generalizability of these findings (L. Leung, 2015) .

PAGE 148

136 8.3 Limitatio ns The use of a secondary data set for quantitative analyses can be a limitation because of the distance of the researcher from the data collection procedures. This presents unique challenges which may impact the validity of the data and reliability of the measurements. For example, exploring and understanding outliers in the data set can pose a challenge. However, due to the low number of outliers and the large sample size in the GUiNZ data set, this was not a significant concern in this study. Additionally, I acknowledge that exposure to pre and post natal maternal stress is not necessarily causally related to BMI scores in early childhood. Nevertheless, these findings of having higher BMI in the first five years of life. Fu ture research is needed to continue exploring the biological and behavioral pathways that may create these associations. Both prenatal stress and early childhood BMI are complex factors. A variety of different stressors and appraisal of stressors may influence the observed pathways and additional measures are important to explore in future analyses. The qualitative sample is limited by the non random sampling methods used in this research. As previou sly mentioned, the goal of the qualitative component of this study was to provide additional details related to the lived experience of maternal stress and patterns of risk and protective factors within and between ethnic groups. The integration of quantit ative and qualitative methods to understand these phenomena in New Zealand was a strength of this study.

PAGE 149

137 8.4 Contributions of the Study This study has significant public health and anthropological implications. The findings contribute to our understa nding of the origins of childhood obesity. The mechanisms explaining how external factors impact biology are complex. The current epidemic of child obesity is impacting communities on a global scale, and behavioral interventions are not always effective du obesity. Addressing negative exposures to stress during early developmental periods may prove to be more effective than attempting to alter health behaviors and t rajectories in adulthood (Dancause et al., 2015; Entringer et al., 2010; E. Y. Lau, Liu, Archer, McDonald, & Liu, 2014; Liu et al., 2016b; Monasta et al., 2010) . The findings from this study suggest that objective measures of stress and subjective responses to stress are not highly correlated and may be measuring different characteristics of the stress response. Transactional models of stress hig hlight the importance of measuring both real and perceived environmental demands independently (Lazarus & Folkman, 1987) . However, objective stress was a stronger predictor of childhood BMI compared to subjective stress measures in this sample . It is possible that subjective stress could be more strongly predictive of maternal and offspring phenotypic outcomes separate from offspring BMI . Stress experienced during both the pre and post natal periods results in higher childhood BMI at 54 months of age. The timing of stress exposure (pre versus post ) was not significantly associated with the outcome, whereas the duration of stress exposure was significantly associated with childhood BMI. Transitions in stress exposure from the pre through the post natal period (decrease or increase) did not differ in their impact on early childhood BMI. Cumulative

PAGE 150

138 maternal stress from the pre through the p ost natal period was most detrimental to early childhood BMI in the GUiNZ data set. A longer duration of stress exposure from the prenatal period throughout early childhood may be the most damaging to child development and the subsequent onset of early chi ldhood and adult obesity (Katz et al., 2012; Nederhof & Schmidt, 2012; Umberson et al., 2010) . Appl ication of the Life Course Epidemiology model to understand the impacts of maternal stress on early childhood BMI may be more appropriate compared to the Developmental Origins of Health and Disease model. This suggests that interventions that target both t he pre and post natal periods of early childhood will be most impactful. Nevertheless, interventions targeting either the pre or post natal period will be beneficial and decrease the likelihood of childhood obesity. Risk behaviors, including maternal eating behaviors during pregnancy and length of exclusive breastfeeding partially mediate these relationships; physical activity behaviors during pregnancy do not. Maternal health behaviors during the pre and post natal periods are critical to the healthy development of a child (Lobe l et al., 2008b; Oken et al., 2007; Schack Nielsen et al., 2010) . Although partial mediation suggests that additional biological and behavioral factors are at play, promoting healthy behaviors throughout these sensitive periods can be protective against the development of early childhood obesity for New Zealand mothers and their offspring. These findings also highlight the influence of the sociocultural context on human health and development and demonstrates the need for interdisciplinary collaborati on to better understand dimensions of maternal stress, protective factors that impact stress experiences, and associations with the development of early childhood obesity (Dressler et al., 2012; Shonkoff et al., 2012) . Social support, neighborhood integration, household cohesiveness, and cultural identity may confer resilience differentially between ethnic groups. For example, primary

PAGE 151

139 sources of social support vary by ethnic group among New Zealand mothers. Further insight into variation of these factors within and between cultures may inform shared patterns of protective cultural resources across communities throughout New Zealand. Th e unique sources of stress and variations between ethnic groups that were identified from the qualitative work in this study can be quantitatively analyzed in subsequent studies to understand which sources of stress may be the most impactful on biological and behavioral pathways linking maternal stress and early childhood BMI in this population. Stress reduction and obesity prevention interventions must be tailored to the cultural norms and values of the specific target audience to promote maternal and chil d resiliency. 8.5 Recommendations and Implications for Future Research Developing a better understanding of the mechanisms that link pre and post natal maternal stress experience and early childhood obesity is needed, including articulation of sociocult ural variations between individuals and ethnic groups. Many of the risk behaviors and protective factors linking maternal stress and early childhood obesity likely interact and are modifiable. This is particularly relevant for designing interventions that reduce maternal stress, combat the early childhood obesity epidemic and diminish health disparities. The following recommendations are based on the findings from this study: It is essential that early childhood obesity prevention efforts target the first f ew years of life, particularly during the pre and early post natal periods of development. Biological systems, stress regulation and coping mechanisms, and health behaviors are shaped during these critical periods of development. Given the plasticity in e arly development, targeting specific mechanisms that impact obesity risk during early childhood can be highly effective. These findings suggest that interventions will be most impactful if both

PAGE 152

140 the pre and early post natal periods are targeted. However, n egative child health outcomes related to obesity are buffered even when a mother experiences lower levels of stress during either developmental period compared to high stress during both periods. Interventions that target only one sensitive period of devel opment can still be impactful and reduce the onset of early childhood obesity. A two generation approach is needed to tackle the early childhood obesity epidemic. Maternal stress and associated behaviors during pregnancy, including eating behaviors, as we ll as during the postnatal period, such as length of exclusive breastfeeding, have negative impacts related to the onset of early childhood obesity. Behavioral interventions including adult education related to the importance of healthy lifestyles, job tra ining, and prenatal care are critical to support the healthy development of the child. These efforts need to be freely available and accessible for low income pregnant women to help counteract the socioeconomic disparities that exist with relation to mater nal stress experience and early childhood obesity. Stress reduction and obesity prevention interventions should address multiple relationship contexts and include the extended family, specifically for collectivist groups such as and Pacifika communi ties. In some cases, grandparents may have even greater influence over health related behaviors in early childhood compared to parents because of co habitation, communal living, and cultural values around the grandparent grandchild relationship common in M Coping strategies must start during pregnancy and involve supports at the individual and community level. to effectively cope with stress and make health promoting behavio r choices through

PAGE 153

141 enhancing social support, integration and connection, and cultural identity . Addressing caregiver needs first may be critical for facilitating parenting that supports healthy child development. Community and policy level changes, such a s improvements in neighborhood safety, free or low cost family activities and cultural events, changes to government assistance based maternity leave requirements, and improvements in the availability and access of high quality child care may foster matern al coping and resilience. Interventions need to be tailored to meet the needs of unique cultures and ethnic groups. Complex sociocultural factors, including normative perceptions of stress and cultural norms surrounding food intake, breastfeeding, and wei ght, are associated with the lived experience of stress and health promoting behaviors linked to obesity. Considering individual cultural aspects surrounding maternal stress and health should be a key component of early intervention.

PAGE 154

142 REFEREN C ES Abel, S., Park, J., Tipene Leach, D., Finau, S., & Lennan, M. (2001). Infant care practices in New Zealand: a cross cultural qualitative study. Social Science & Medicine , 53 (9), 1135 1148. https://do i.org/10.1016/S0277 9536(00)00408 1 Abraham, S. B., Rubino, D., Sinaii, N., Ramsey, S., & Nieman, L. K. (2013). Cortisol, obesity, and the metabolic syndrome: A cross sectional study of obese subjects and review of the literature. Obesity , 21 (1), E105 E117 . https://doi.org/10.1002/oby.20083 Adam, T. C., & Epel, E. S. (2007). Stress, eating and the reward system. Physiology & Behavior , 91 (4), 449 458. https://doi.org/10.1016/j.physbeh.2007.04.011 Ager, A., Stark, L., Akesson, B., & Boothby, N. (2010). Defini ng best practice in care and protection of children in crisis affected settings: A Delphi study. Child Development , 81 (4), 1271 1286. https://doi.org/10.1111/j.1467 8624.2010.01467.x Amugongo, S. K., & Hlusko, L. J. (2013). Impact of Maternal Prenatal Stre ss on Growth of the Offspring. Aging and Disease , 5 (1), 1 16. https://doi.org/10.14336/AD.2014.05001 Barker, D. J. P. (2007). The origins of the developmental origins theory. In Journal of Internal Medicine (Vol. 261, pp. 412 417). https://doi.org/10.1111/j.1365 2796.2007.01809.x Barker, D. J. P., Godfrey, K. M., Gluckman, P. D., Harding, J. E., Owens, J. A., & Robinson, J. S. (1993). Fetal nutrition and cardiovascular disease in adult life. The Lancet , 341 ( 8850), 938 941. environments and biological propensities affect the health of populations. International Journal of Behavioral Medicine , 11 (3), 164 175. https://doi.org/10. 1207/s15327558ijbm1103_5 Baxter, J., Kani Kingi, T., Tapsell, R., Durie, M., Mcgee, M. A., & Team, F. T. N. Z. M. H. S. R. Zealand Mental Health Survey. Australian & New Zealand Journal of Psychiatry , 40 (10), 914 923. https://doi.org/10.1080/j.1440 1614.2006.01911.x Ben Shlomo, Y. (2002). A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives. Inte rnational Journal of Epidemiology , 31 (2), 285 293. https://doi.org/10.1093/ije/31.2.285 Benyshek, D. C. (2007). The developmental origins of obesity and related health disorders -prenatal and perinatal factors. Collegium Antropologicum , 31 (1), 11 17.

PAGE 155

143 Berr y, J. W. (2005). Acculturation: Living successfully in two cultures. International Journal of Intercultural Relations , 29 (6), 697 712. https://doi.org/10.1016/J.IJINTREL.2005.07.013 Biro, F. M., & Wien, M. (2010). Childhood obesity and adult morbidities. T he American Journal of Clinical Nutrition , 91 (Suppl 1), 1499S 1505S. https://doi.org/10.3945/ajcn.2010.28701B.1 Bjorntorp, P. (2001). Do stress reactions cause abdominal obesity and comorbidities? Obesity Reviews , 2 (2), 73 86. https://doi.org/10.1046/j.146 7 789x.2001.00027.x Blair, C., Granger, D. A., Willoughby, M., Mills Fortunato, C. K. (2011). Salivary Cortisol Mediates Effects of Poverty and Parenting on Executive Functions in Early Childhood. Child Development , 82 (6), 1970 1984. https://doi.org/10.1111/j.1467 8624.2011.01643.x Blair, C., & Raver, C. C. (2012). Child development in the context of adversity: Experiential canalization of brain and behavior. American Psychologist , 67 (4), 309 318. https://doi.org/10. 1037/a0027493 Boden, J. M., Fergusson, D. M., & John Horwood, L. (2007). Early motherhood and subsequent life outcomes. Journal of Child Psychology and Psychiatry , 49 (2), 151 160. https://doi.org/10.1111/j.1469 7610.2007.01830.x Bogin, B. (1999). Patterns of Human Growth. Cambridge University Press , (1995), 472. https://doi.org/10.1017/CBO9780511600456 Borrows, J., Williams, M., Schluter, P., Paterson, J., & Helu, S. L. (n.d.). Pacific Islands Families Study: The Association of In fant Health Risk Indicators and Acculturation of Pacific Island Mothers Living in New Zealand. Journal of Cross Cultural Psychology , 42 (5), 699 724. https://doi.org/10.1177/0022022110362750 Bowman, S. A., Gortmaker, S. L., Ebbeling, C. B., Pereira, M. A., Ludwig, D. S., & Kazak, A. (2004). Effects of Fast Food Consumption on Energy Intake and Diet Quality Among Children in a National Household Survey. PEDIATRICS , 113 (1), 112 118. https://doi.org/10.1542/peds.113.1.112 Brakefield, P. M., Pijpe, J., & Zwaan, B. J. (2007). Developmental plasticity and acclimation both contribute to adaptive responses to alternating seasons of plenty and of stress in Bicyclus butterflies. Journal of Biosciences , 32 (3), 465 475. https://doi.org/10.1007/S12038 007 0046 8 Brougham, D., & Haar, J. M. (2013). Collectivism, Cultural Identity and Employee Mental Social Indicators Research , 114 (3), 1143 1160. https://doi.org/10.1007/s11205 012 0194 6

PAGE 156

144 Brown, J. D., Harris, S. K., Woods, E. R., Buman, M. P., & Cox, J. E. (2012). Longitudinal Study of Depressive Symptoms and Social Support in Adolescent Mothers. Maternal and Child Health Journal , 16 (4), 894 901. https://doi.org/10.1007/s10995 011 0814 9 Brown, S. J., Lumley, J. M., McDonald, E. A., & Kr astev, A. H. (2006). Maternal health study: a prospective cohort study of nulliparous women recruited in early pregnancy. BMC Pregnancy and Childbirth , 6 (1), 12. https://doi.org/10.1186/1471 2393 6 12 Buss, C., Lord, C., Wadiwalla, M., Hellhammer, D. H., L upien, S. J., Meaney, M. J., & Pruessner, J. C. (2007). Maternal Care Modulates the Relationship between Prenatal Risk and Hippocampal Volume in Women But Not in Men. Journal of Neuroscience , 27 (10). Retrieved from http://www.jneurosci.org/content/27/10/25 92.short Cameron, J. E. (2004). A Three Factor Model of Social Identity. Self and Identity , 3 (3), 239 262. https://doi.org/10.1080/13576500444000047 Cameron, N., & Demerath, E. W. (2002). Critical periods in human growth and their relationship to diseases of aging. Yearbook of Physical Anthropology , 45 , 159 184. https://doi.org/10.1002/ajpa.10183 Cao DNA methylation mediates the impact of exposure to prenatal maternal stress on BMI and central adiposity in children at age 13 1/2 years: Project Ice Storm. Epigenetics , 10 (8), 749 761. https://doi.org/10.1080/15592294.2015.1063771 Cappuccio, F. P., Taggart, F. M., Kandala, N. B., Currie, A., Peile, E., Stranges, S., & Miller, M. A. (2008). Meta Analysis of Short Sleep Duration and Obesity in Children and Adults. Sleep , 31 (5), 619 626. https://doi.org/10.1093/sleep/31.5.619 Carroll, P., Witten, K., & Kearns, R. (2011). Housing Intensification in Auckland, New Zealand: Implications for Children and Families. Housing Studies , 26 (3), 353 367. https://doi.org/10.1080/02673037.2011.542096 Cartwright, M., Wardle, J., Steggles, N., Simon, A. E., Croker, H., & Jarvis, M. J. (2003). Stress and dietary practices in adolescents. Health Psychology , 22 (4), 362 369. https://doi.org/10.1037/0278 6133.22.4.362 Bre astfeeding indicators among a nationally representative multi ethnic sample of New Zealand children. The New Zealand Medical Journal , 130 (1466), 34 44. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/29197899 Chittleborough, C. R., Lawlor, D. A., & Lynch, J. W. (2011). Young maternal age and poor child development: predictive validity from a birth cohort. Pediatrics , 127 (6), e1436 44. https://doi.org/10.1542/peds.2010 3222

PAGE 157

145 Clauss Ehlers, C. S., Yang, Y. T. T., & Wan Chun Chen, M. J. (2006). Resilience from Childhood Stressors The Role of Cultural Resilience, Ethnic Identity, and Gender Identity. Journal of Infant, Child, and Adolescent Psychotherapy , 5 (1), 124 138. Retrieved from http://www.tandfonline.com/doi/p df/10.2513/s15289168jicap0501_7?needAccess=true Cohen, S., Gottlieb, B. H., & Underwood, L. G. (2004). Social relationships and health. S. Cohen, L. G. Underwood & B. H. Gottlieb (Eds.), Social Support Measurement and Intervention: A Guide for Health and S ocial Scientists , 3 25. Retrieved from http://kungfu.psy.cmu.edu/~scohen/CohGotUnd2000.pdf COHEN, S., & JANICKI Stress in the United States in Probability Samples from 1983, 2006, and 20091 . Journal of Applied Social Psychology , 42 (6), 1320 1334. https://doi.org/10.1111/j.1559 1816.2012.00900.x Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior , 24 (4), 385 396. https: //doi.org/10.2307/2136404 Cohen, S., Kessler, R. C., & Underwood, L. G. (1995). social scientists . Oxford University Press. Cohen, S., & Wills, T. A. (1985). Stress, Social Support, and the Buffering Hypothesis . Ps ychologkal Bulletin (Vol. 98). Retrieved from https://s3.amazonaws.com/academia.edu.documents/40586205/PsychBull1985.pdf?AWSAc cessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1540157742&Signature=YUGBhZA Io%2Bmvl%2Ba7N4t2nWtAER4%3D&response content disposition=inline %3B filename%3DStress_Social_Support_and_the_Buffering.pdf Cole, T. J. (2000). Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ , 320 (7244), 1240 1240. https://doi.org/10.1136/bmj.320.7244.1240 Collins , F. S., & Manolio, T. A. (2007). Merging and emerging cohorts: Necessary but not sufficient. Nature 2007 445:7125 . Craike, M. J., Coleman, D., & MacMahon, C. (2010). Direct and Buffering Effects of Physical Activity on Stress Related Depression in Mothers of Infants. Journal of Sport and Exercise Psychology , 32 (1), 23 38. https://doi.org/10.1123/jsep.32.1.23 (2015). Prenatal stress due to a natural disaster predicts adiposity in childhood: The iowa flood study. Journal of Obesity , 2015 . https://doi.org/10.1155/20 15/570541 Daniel, C. (2016). Economic constraints on taste formation and the true cost of healthy eating. Social Science & Medicine , 148 , 34 41. https://doi.org/10.1016/J.SOCSCIMED.2015.11.025

PAGE 158

146 Danielzik, S., Pust, S., Landsberg, B., & Müller, M. J. (2005). First lessons from the Kiel Obesity Prevention Study (KOPS). International Journal of Obesity , 29 , S78 S83. https://doi.org/10.1038/sj.ijo.0803099 Daskalakis, N. P., Bagot, R. C., Parker, K. J., Vinkers, C. H., & de Kloet, E. R. (2013). The three hit conc ept of vulnerability and resilience: Toward understanding adaptation to early life adversity outcome. Psychoneuroendocrinology , 38 (9), 1858 1873. https://doi.org/10.1016/j.psyneuen.2013.06.008 Desjarlais, R., & Jason Throop, C. (2011). Phenomenological App roaches in Anthropology. Annual Review of Anthropology , 40 (1), 87 102. https://doi.org/10.1146/annurev anthro 092010 153345 Dev, D. A., McBride, B. A., Fiese, B. H., Jones, B. L., & Cho, on behalf of the STRONG Kids R, H. (2013). Risk Factors for Overweigh t/Obesity in Preschool Children: An Ecological Approach. Childhood Obesity , 9 (5), 399 408. https://doi.org/10.1089/chi.2012.0150 DIJK, van, & N. (1991). The Hensel and Gretel syndrome adaptation hypothesis and an alternative model. New Zealand J. Archaeology , 13 , 65 89. Retrieved from https://ci.nii.ac.jp/naid/10003562739/ Dillon, D. G., Holmes, A. J., Birk, J. L., Brooks, N., Lyons Ruth, K., & Pizzagalli, D. A. (2009). Childhood Adversity Is Associated with Left Basal Ganglia Dysfunction During Reward Anticipation in Adulthood. Biological Psychiatry , 66 (3), 206 213. https://doi.org/10.1016/j.biopsych.2009.02.019 Ding, D., & Gebel, K. (2012). Built environment, physical activity, and obesity: What have we learned from reviewing the literature? Health & Place , 18 (1), 100 105. https://doi.org/10.1016/J.HEALTHPLACE.2011.08.021 Dixon, B., Rifas Shiman, S. L., James Taveras, E. M. (2012). Maternal experiences of racial discriminati on and child weight status in the first 3 years of life. Journal of Developmental Origins of Health and Disease , 3 (06), 433 441. https://doi.org/10.1017/S2040174412000384 DJP, B. (2004). The developmental origins of chronic adult disease. Acta Paediatrica , 93 (0), 26 33. https://doi.org/10.1080/08035320410022730 Dole, N. (2003). Maternal Stress and Preterm Birth. American Journal of Epidemiology , 157 (1), 14 24. https://doi.org/10.1093/aje/kwf176 Dong, M., Giles, W. H., Felitti, V. J., Dube, S. R., Williams, J. E., Chapman, D. P., & Anda, R. F. (2004). Insights into causal pathways for ischemic heart disease: Adverse childhood experiences study. Circulation . https://doi.org/10.1161/01.CIR.0000143074.54995.7F

PAGE 159

147 Dressler, W. W., Balieiro, M. C., Ribeiro, R. P., & dos Santos, J. E. (2016). Culture and the Immune System: Cultural Consonance in Social Support and C reactive Protein in Urban Brazil. Medical Anthropology Quarterly , 30 (2), 259 277. https://doi.org/10.11 11/maq.12213 Dressler, W. W., Balieiro, M. C., Ribeiro, R. P., & Santos, J. E. Dos. (2007). Cultural consonance and psychological distress: examining the associations in multiple cultural domains. Culture, Medicine and Psychiatry , 31 (2), 195 224. https://d oi.org/10.1007/s11013 007 9046 2 Dressler, W. W., Oths, K. S., Balieiro, M. C., Ribeiro, R. P., & Dos Santos, J. E. (2012). How culture shapes the body: Cultural consonance and body mass in urban brazil. American Journal of Human Biology , 24 (3), 325 331. https://doi.org/10.1002/ajhb.22207 Dressler, W. W., Oths, K. S., & Gravlee, C. C. (2005). RACE AND ETHNICITY IN PUBLIC HEALTH RESEARCH: Models to Explain Health Disparities. Annual Review of Anthropology , 34 (1), 231 252. https://doi.org/10.1146/annurev.ant hro.34.081804.120505 Dressler, W. W., Oths, K. S., Ribeiro, R. P., Balieiro, M. C., & Dos Santos, J. E. (2008). Cultural consonance and adult body composition in urban Brazil. American Journal of Human Biology , 20 (1), 15 22. https://doi.org/10.1002/ajhb.20 674 Dupéré, V., & Perkins, D. D. (2007). Community types and mental health: a multilevel study of local environmental stress and coping. American Journal of Community Psychology , 39 (1 2), 107 119. https://doi.org/10.1007/s10464 007 9099 y Edwards, S., Mccr eanor, T., & Moewaka Barnes, H. (2010). family culture: A context of youth development in Counties/Manukau. https://doi.org/10.1080/1177083X.2007.9522420org/10.1080/1177083X.2007.9522420 El Sheikh, M., Buckhalt, J. A., Mize, J., & Acebo, C. (2006). Marital Conflict and Disruption of Child Development , 77 (1), 31 43. https://doi.org/10.1111/j.1467 8624.2006.00854.x El Adjustment Over Time: T he Role of Socioeconomic Context. Child Development , 81 (3), 870 883. https://doi.org/10.1111/j.1467 8624.2010.01439.x Elias, C. V., Power, T. G., Beck, A. E., Goodell, L. S., Johnson, S. L., Papaioannou, M. A., & Hughes, S. O. (2016). Depressive Symptoms a nd Perceptions of Child Difficulty Are Associated with Less Responsive Feeding Behaviors in an Observational Study of Low Income Mothers. Childhood Obesity , 12 (6), 418 425. https://doi.org/10.1089/chi.2016.0125

PAGE 160

148 Entringer, S., Buss, C., & Wadhwa, P. D. (2010). Prenatal stress and developmental programming of human health and disease risk: concepts and integration of empirical findings. Current Opinion in Endocrinology, Diabetes, and Obesity , 17 (6), 507 516. https: //doi.org/10.1097/MED.0b013e3283405921 Entringer, S., Buss, C., & Wadhwa, P. D. (2015). Prenatal stress, development, health and disease risk: A psychobiological perspective 2015 Curt Richter Award Paper. Psychoneuroendocrinology , 62 , 366 375. https://doi. org/10.1016/j.psyneuen.2015.08.019 ENZ. (2018). House Rents West Auckland, NZ How much should you pay? Retrieved November 21, 2018, from https://www.enz.org/house rents auckland west.html Epel, E., Lapidus, R., McEwen, B., & Brownell, K. (2001). Stress m ay add bite to appetite in women: A laboratory study of stress induced cortisol and eating behavior. Psychoneuroendocrinology , 26 (1), 37 49. https://doi.org/10.1016/S0306 4530(00)00035 4 Essex, C., Counsell, A. M., & Geddis, D. C. (1992). The Demographic C haracteristics of Early and Late Attenders for Antenatal Care. The Australian and New Zealand Journal of Obstetrics and Gynaecology , 32 (4), 306 308. https://doi.org/10.1111/j.1479 828X.1992.tb02839.x Evans, G. W., & English, K. (2002). The Environment of P overty: Multiple Stressor Exposure, Psychophysiological Stress, and Socioemotional Adjustment. Child Development , 73 (4), 1238 1248. https://doi.org/10.1111/1467 8624.00469 Evans, G. W., Fuller Rowell, T. E., & Doan, S. N. (2012). Childhood Cumulative Risk and Obesity: The Mediating Role of Self Regulatory Ability. PEDIATRICS , 129 (1), e68 e73. https://doi.org/10.1542/peds.2010 3647 Farewell, C. V., Thayer, Z. M., Puma, J. E., & Morton, S. (2018). Exploring the timing and duration of maternal stress exposure: Impacts on early childhood BMI. Early Human Development , 117 . https://doi.org/10.1016/j.earlhumdev.2017.12.001 Farewell, C. V., Thayer, Z. M., Tracer, D. P., & Morton, S. (2018). Prenatal stress exposure and early childhood BMI: Exploring associations in a New Zealand context. American Journal of Human Biology . https://doi.org/10.1002/ajhb.23116 FORD, R. P. K., MITCHELL, E. A., SCRAGG, R., STEWART, A. W., TAYLOR, B. J., & ALLEN, E. M. (1994). Factors adversely associated with breast feeding in New Zealand. Journal of Paediatrics and Child Health , 30 (6), 483 489. https://doi.org/10.1111/j.1440 1754.1994.tb00717.x Fox, M., Thayer, Z., & Wadhwa, P. D. (2017). Assessment of acculturation in minority health research. Social Science & Medicine , 176 , 123 132. http s://doi.org/10.1016/J.SOCSCIMED.2017.01.029

PAGE 161

149 Fraser, A., Macdonald Lawlor, D. A. (2013). Cohort Profile: The Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort. Internati onal Journal of Epidemiology , 42 (1), 97 110. https://doi.org/10.1093/ije/dys066 Gander, P. H., Marshall, N. S., Harris, R., & Reid, P. (2005). The Epworth Sleepiness Scale: Influence of Age, Ethnicity, and Socioeconomic Deprivation. Epworth Sleepiness Scor es of Adults in New Zealand. Sleep , 28 (2), 249 254. https://doi.org/10.1093/sleep/28.2.249 Gao, W., Paterson, J., Abbott, M., Carter, S., Iusitini, L., & McDonald Sundborn, G. (2010). Impact of Current and Past Intimate Partner Violence on Maternal Mental Health and Behaviour at 2 Years After Childbirth: Evidence from the Pacific Islands Families Study. Australian & New Zealand Journal of Psychiatry , 44 (2), 174 182. https://doi.org/10.3109/00048670903487126 Gillman, M. W., Rich Edwards, J. W., Huh, S., Majzoub, J. A., Oken, E., Taveras, E. M., & Rifas Shiman, S. L. (2006). Maternal corticotropin releasing hormone levels during pregnancy and offspring adiposity. Obesity (Silver Spring, Md.) , 14 (9), 1647 1653. https://doi.org/10.1038/oby.2006.189 Glover, M., Waldon, J., Manaena Biddle, H., Holdaway, M., & Cunningham, C. (2009). Barriers and Services. Journal of Human Lactation , 25 (3), 307 316. https://doi.org/10.1177/0890334409332436 Gluckman, P. D. (2008). Effect of In Utero and Early Life Conditions on Adult Health and Disease. The New England Journal of Medicine , 359 (1), 61 73. https://doi.org/10.1056/NEJMra0708473.Effect Gluckman, P. D., Hanson, M. A., Cooper, C., & Thornburg, K. L. (2008). Effect of In Utero and Early Life Conditions on Adult Health and Disease. New England Journal of Medicine , 359 (1), 61 73. https://doi.org/10.1056/NEJMra0708473 Gluckman, P. D., & Hanson , M. a. (2004). Living with the past: evolution, development, and patterns of disease. Science (New York, N.Y.) , 305 (5691), 1733 1736. https://doi.org/10.1126/science.1095292 Graziano, P. A., Calkins, S. D., & Keane, S. P. (2010). Toddler self regulation s kills predict risk for pediatric obesity. International Journal of Obesity , 34 (4), 633 641. https://doi.org/10.1038/ijo.2009.288 Gundersen, C., Lohman, B. J., Garasky, S., Stewart, S., & Eisenmann, J. (2008). Food security, maternal stressors, and overweig ht among low income US children: Results from the National Health and Nutrition Examination Survey (1999 2002). Pediatrics , 122 (3), e529 e540. https://doi.org/10.1542/peds.2008 0556

PAGE 162

150 Gundersen, C., Mahatmya, D., Garasky, S., & Lohman, B. (2011). Linking psy chosocial stressors and childhood obesity. Obesity Reviews , 12 (501). https://doi.org/10.1111/j.1467 789X.2010.00813.x Hackman, D. A., Farah, M. J., & Meaney, M. J. (2010). Socioeconomic status and the brain: mechanistic insights from human and animal resea rch. Nature Reviews Neuroscience , 11 (9), 651 659. https://doi.org/10.1038/nrn2897 Hammen, C., Kim, E. Y., Eberhart, N. K., & Brennan, P. A. (2009). Chronic and acute stress and the prediction of major depression in women. Depression and Anxiety , 26 (8), 718 723. https://doi.org/10.1002/da.20571 Hanson, J. L., Albert, D., Iselin, A. M. R., Carré, J. M., Dodge, K. A., & Hariri, A. R. (2016). Cumulative stress in childhood is associated with blunted reward related brain activity in adulthood. Social Cognitive a nd Affective Neuroscience , 11 (3), 405 412. https://doi.org/10.1093/scan/nsv124 Harris, A., & Seckl, J. (2011). Glucocorticoids, prenatal stress and the programming of disease. Hormones and Behavior , 59 (3), 279 289. https://doi.org/10.1016/J.YHBEH.2010.06.0 07 Harris, R., Cormack, D., Tobias, M., Yeh, L. C., Talamaivao, N., Minster, J., & Timutimu, R. (2012). The pervasive effects of racism: Experiences of racial discrimination in New Zealand over time and associations with multiple health domains. Social Sci ence & Medicine , 74 (3), 408 415. https://doi.org/10.1016/J.SOCSCIMED.2011.11.004 Harris, R., Tobias, M., Jeffreys, M., Waldegrave, K., Karlsen, S., & Nazroo, J. (2006). Racism and health: The relationship between experience of racial discrimination and health in New Zealand. Social Science & Medicine , 63 (6), 1428 1441. https://doi.org/ 10.1016/J.SOCSCIMED.2006.04.009 Haszard, J. J., Skidmore, P. M., Williams, S. M., & Taylor, R. W. (2015). Associations between parental feeding practices, problem food behaviours and dietary intake in New Zealand overweight children aged 4 8 years. Public Health Nutrition , 18 (06), 1036 1043. https://doi.org/10.1017/S1368980014001256 Heindel, J. J., & vom Saal, F. S. (2009). Role of nutrition and environmental endocrine disrupting chemicals during the perinatal period on the aetiology of obesity. Molecular a nd Cellular Endocrinology . https://doi.org/10.1016/j.mce.2009.02.025 Holliday, R. (2006). Epigenetics: A historical overview. Epigenetics . https://doi.org/10.4161/epi.1.2.2762 Hook, G. R., Waaka, T., & Raumati, L. P. (2007). framework . Target Article (Vol. 3). Retrieved from http://www.review.mai.ac.nz

PAGE 163

151 Horton, T. V., & Wallander, J. L. (2001). Hope and social support as resilience factors against psychological distress of mothers who care for children with chronic physical c onditions. Rehabilitation Psychology , 46 (4), 382 399. https://doi.org/10.1037/0090 5550.46.4.382 Huang, J. S., Lee, T. A., & Lu, M. C. (2007). Prenatal programming of childhood overweight and obesity. Maternal and Child Health Journal . https://doi.org/10.1 007/s10995 006 0141 8 Hughes, S. O., Power, T. G., Liu, Y., Sharp, C., & Nicklas, T. A. (2015). Parent emotional distress and feeding styles in low income families. The role of parent depression and parenting stress. Appetite , 92 , 337 342. https://doi.org/ 10.1016/J.APPET.2015.06.002 IBM SPSS Inc. (2012). SPSS Statistics for Windows. IBM Corp. Released 2012 , Version 20 , 1 8. Incollingo Rodriguez, A. C., Epel, E. S., White, M. L., Standen, E. C., Seckl, J. R., & Tomiyama, A. J. (2015). Hypothalamic pituitary adrenal axis dysregulation and cortisol activity in obesity: A systematic review. Psychoneuroendocrinology , 62 , 301 318. https://doi.org/10.1016/j.psyneuen.2015.08.014 Jeanrenaud, B., & Rohner Jeanrenaud, F. (2001). Effects of Neuropeptides and Leptin on N utrient Partitioning: Dysregulations in Obesity. Annual Review of Medicine , 52 (1), 339 351. https://doi.org/10.1146/annurev.med.52.1.339 Jick, T. D. (1979). Mixing Qualitative and Quantitative Methods: Triangulation in Action. Administrative Science Quarte rly , 24 (4), 602. https://doi.org/10.2307/2392366 Johnson, E. O., Kamilaris, T. C., Chrousos, G. P., & Gold, P. W. (1992). Mechanisms of stress: A dynamic overview of hormonal and behavioral homeostasis. Neuroscience & Biobehavioral Reviews , 16 (2), 115 130. https://doi.org/10.1016/S0149 7634(05)80175 7 K.N., D., D.P., L., S., F., A., B., A., C., N., S., & S., K. (2012). Prenatal exposure to a natural disaster increases risk for obesity in 5 year old children. Pediatric Research . Retrieved from http://ovidsp. ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=emed10&NEWS=N&AN =2011693549 Kahu, E., & Morgan, M. (2007). A critical discourse analysis of New Zealand government policy: Women as mothers and workers. , 30 (2), 134 146. https://doi.org/10.1016/J.WSIF.2007.01.003 Kandula, N. R., Lauderdale, D. S., & Baker, D. W. (2007). Differences in Self Reported Health Among Asians, Latinos, and Non Hispanic Whites: The Role of Language and Nativity. Annals of Epidemiology , 17 (3), 191 1 98. https://doi.org/10.1016/J.ANNEPIDEM.2006.10.005

PAGE 164

152 Katz, D. A., Sprang, G., & Cooke, C. (2012). The Cost of Chronic Stress in Childhood: Understanding and Applying the Concept of Allostatic Load. Psychodynamic Psychiatry , 40 (3), 469 480. https://doi.org/1 0.1521/pdps.2012.40.3.469 Keddell, E. (2007). Cultural Identity and the Children, Young Persons, and their Families Act 1989: Ideology, Policy and Practice . Retrieved from https://www.msd.govt.nz/documents/about msd and our work/publications resources/jour nals and magazines/social policy journal/spj32/32 pages 49 71.pdf King, S., Dancause, K., Turcotte Tremblay, A. M., Veru, F., & Laplante, D. P. (2012). Using Natural Disasters to Study the Effects of Prenatal Maternal Stress on Child Health and Development . Birth Defects Research Part C: Embryo Today: Reviews , 96 (4), 273 288. https://doi.org/10.1002/bdrc.21026 Kingston, D., Sword, W., Krueger, P., Hanna, S., & Markle Reid, M. (2012). Life Course Pathways to Prenatal Maternal Stress. JOGNN: Journal of Obstet ric, Gynecologic & Neonatal Nursing , 41 (5), 609 626. https://doi.org/10.1111/j.1552 6909.2012.01381.x Kline, P. (2014). An Easy Guide to Factor Analysis . Routledge. https://doi.org/10.4324/9781315788135 Kline, R. (2015). Principles and practice of structur al equation modeling . Retrieved from https://books.google.com/books?hl=en&lr=&id=Q61ECgAAQBAJ&oi=fnd&pg=PP1&ots= jEhn xA7rh&sig=P8Xg_DKHkMEy7_DVHpJpQpxvBlY Koch, F. S., Sepa, A., & Ludvigsson, J. (2008). Psychological Stress and Obesity. The Journal of Pedi atrics , 153 (6), 839 844.e3. https://doi.org/10.1016/j.jpeds.2008.06.016 C., & 2008, undefined. (n.d.). Neighborhood disadvantage: Pathways of effects for young children. Wiley Online Library . Retrieved from https://online library.wiley.com/doi/abs/10.1111/j.1467 8624.2007.01117.x Koolhaas, J. M., de Boer, S. F., & Buwalda, B. (2006). Stress and Adaptation. Current Directions in Psychological Science , 15 (3), 109 112. https://doi.org/10.1111/j.0963 7214.2006.00417.x Kruger, D . J., Reischl, T. M., & Gee, G. C. (2007). Neighborhood Social Conditions Mediate the Association Between Physical Deterioration and Mental Health. American Journal of Community Psychology , 40 (3 4), 261 271. https://doi.org/10.1007/s10464 007 9139 7 Kuh, D. (2003). Life course epidemiology. Journal of Epidemiology & Community Health , 57 (10), 778 783. https://doi.org/10.1136/jech.57.10.778 Kuzawa, C., & Quinn, E. (2009). Developmental origins of adult function and health: evolutionary hypotheses. Annua l Review of Anthropology . Retrieved from http://annualreviews.org/doi/abs/10.1146/annurev anthro 091908 164350

PAGE 165

153 Kuzawa, C. W., & Sweet, E. (2009). Epigenetics and the embodiment of race: Developmental origins of US racial disparities in cardiovascular healt h. American Journal of Human Biology , 21 (1), 2 15. https://doi.org/10.1002/ajhb.20822 Laessle, R. G., Uhl, H., & Lindel, B. (2001). Parental influences on eating behavior in obese and nonobese preadolescents. International Journal of Eating Disorders , 30 (4 ), 447 453. https://doi.org/10.1002/eat.1106 Landy, C. K., Sword, W., & Valaitis, R. (2009). The Experiences of Socioeconomically Disadvantaged Postpartum Women in the First 4 Weeks at Home. https://doi.org/10.1177/1049732308329310 Laplante, D. P., Brunet, A., Schmitz, N., Ciampi, A., & King, S. (2008). Project Ice Storm: prenatal maternal stress affects cognitive and linguistic functioning in 5 1/2 year old children. Journal of the American Academy of Child and Adolescent Psychiatry , 47 (9), 1063 1072. http s://doi.org/10.1097/CHI.0b013e31817eec80 Lau, C., Hurst, N. M., Smith, E. O., & Schanler, R. J. (2007). Ethnic/racial diversity, maternal stress, lactation and very low birthweight infants. Journal of Perinatology , 27 (7), 399 408. https://doi.org/10.1038/s j.jp.7211770 Lau, E. Y., Liu, J., Archer, E., McDonald, S. M., & Liu, J. (2014). Maternal weight gain in pregnancy and risk of obesity among offspring: A systematic review. Journal of Obesity , 2014 . https://doi.org/10.1155/2014/524939 Lazarus, R. R. S., & Folkman, S. (1987). Transactional theory and research on emotions and coping. European Journal of Personality , 1 (May), 141 169. https://doi.org/10.1002/per.2410010304 Lee, J., Pomeroy, E. C., & Bohman, T. M. (2007). Intimate Partner Violence and Psychologi cal Health in a Sample of Asian and Caucasian Women: The Roles of Social Support and Coping. Journal of Family Violence , 22 (8), 709 720. https://doi.org/10.1007/s10896 007 9119 6 Leech, N. L. N. L. & O. A. J. N. L., & Onwuegbuzie, A. A. J. A. J. J. (2007). An array of qualitative data analysis tools: A call for data analysis triangulation. School Psychology Quarterly , 22 (4), 557 584. https://doi.org/10.1037/1045 3830.23.4.587 Leung, C. Y. Y., Lumeng, J. C., Kaciroti, N. A., Chen, Y. P., Rosenblum, K., & Mil ler, A. L. (2014). Surgency and negative affectivity, but not effortful control, are uniquely associated with obesogenic eating behaviors among low income preschoolers. Appetite , 78 , 139 146. https://doi.org/10.1016/J.APPET.2014.03.025 Leung, L. (2015). Va lidity, reliability, and generalizability in qualitative research. Journal of Family Medicine and Primary Care , 4 (3), 324 327. https://doi.org/10.4103/2249 4863.161306

PAGE 166

154 Levin, B. E. (2006). Metabolic imprinting: critical impact of the perinatal environment on the regulation of energy homeostasis. Philosophical Transactions of the Royal Society B: Biological Sciences , 361 (1471), 1107 1121. https://doi.org/10.1098/rstb.2006.1851 Li, C., Kaur, H., Choi, W. S., Huang, T. T. K., Lee, R. E., & Ahluwalia, J. S. (2005). Additive Interactions of Maternal Prepregnancy BMI and Breast feeding on Childhood Overweight. Obesity Research , 13 (2), 362 371. https://doi.org/10.1038/oby.2005.48 Li, J., Kendall, G., Henderson, S., Downie, J., Landsborough, L., & Oddy, W. (2008). Maternal psychosocial well being in pregnancy and breastfeeding duration. Acta Paediatrica , 97 (2), 221 225. https://doi.org/10.1111/j.1651 2227.2007.00602.x Liu, G. T., Dancau se, K. N., Elgbeili, G., Laplante, D. P., & King, S. (2016a). Disaster related prenatal maternal stress explains increasing amounts of variance in body composition through childhood and adolescence: Project Ice Storm. Environmental Research . https://doi.or g/10.1016/j.envres.2016.04.039 Liu, G. T., Dancause, K. N., Elgbeili, G., Laplante, D. P., & King, S. (2016b). Disaster related prenatal maternal stress explains increasing amounts of variance in body composition through childhood and adolescence: Project Ice Storm. Environmental Research , 150 . https://doi.org/10.1016/j.envres.2016.04.039 Lobel, M., Cannella, D. L., Graham, J. E., DeVincent, C., Schneider, J., & Meyer, B. A. (2008a). Pregnancy specific stress, prenatal health behaviors, and birth outcomes. Health Psychology , 27 (5), 604 615. https://doi.org/10.1037/a0013242 Lobel, M., Cannella, D. L., Graham, J. E., DeVincent, C., Schneider, J., & Meyer, B. A. (2008b). Pregnancy specific stress, prenatal health behaviors, and birth outcomes. Health Psychology , 27 (5), 604 615. https://doi.org/10.1037/a0013242 Lovallo, W. R. (n.d.). . Retrieved from https://books.google.co.nz/books?hl=en&lr=&id=1KGXBgAAQBAJ&oi=fnd&pg=PT17& dq=lovallo+&ots=QUaxHExi x&s ig=ix1vNIuzoFJW7xTu7aKjdTxMnpY#v=onepage&q=lovallo&f=false Lu, M. C., & Chen, B. (2004). Racial and ethnic disparities in preterm birth: The role of stressful life events. American Journal of Obstetrics and Gynecology , 191 (3), 691 699. https://doi.org/10.1 016/j.ajog.2004.04.018 Lumeng, J. C., Miller, A., Peterson, K. E., Kaciroti, N., Sturza, J., Rosenblum, K., & Vazquez, D. M. (2014). Diurnal cortisol pattern, eating behaviors and overweight in low income preschool aged children. Appetite , 73 , 65 72. https ://doi.org/10.1016/j.appet.2013.10.016

PAGE 167

155 M. Franco, L., J. Pottick, K., & Huang, C. C. (2010). Early parenthood in a community context: neighborhood conditions, race ethnicity, and parenting stress. Journal of Community Psychology , 38 (5), 574 590. https://d oi.org/10.1002/jcop.20382 Manhire, K. M., Hagan, A. E., & Floyd, S. A. (2007). A descriptive account of New Zealand ended questions on their breast feeding experiences. Midwifery , 23 (4), 372 381. https://doi.org/10.1016/J.MIDW.20 06.01.002 Marie, D., Fergusson, D. M., & Boden, J. M. (n.d.). eThNic iDeNTiTy aND iNTimaTe ParTNer ViOLeNce iN a NeW ZeaLaND BirTh cOhOrT . Retrieved from https://www.otago.ac.nz/christchurch/otago014505.pdf Marniemi, J., Kronholm, E., Aunola, S., Toikka, T., Mattlar, C. E., Koskenvuo, M., & Rönnemaa, T. (2002). Visceral fat and psychosocial stress in identical twins discordant for obesity. Journal of Internal Medicine , 251 (1), 35 43. https://doi.org/10.1046/j .1365 2796.2002.00921.x Maxwell, J. A. (2013). Qualitative Research Design: An Interactive Approach Joseph A. Maxwell Google Books . Thousand Oaks, Califordina. Retrieved from https://books.google.com/books?hl=en&lr=&id=xAHCOmtAZd0C&oi=fnd&pg=PR5&dq= qua litative+research+design+maxwell&ots=Y1DTwpwfj0&sig=b6XuULG1h1VJBh0M04 DLAn7209Y#v=onepage&q=qualitative research design maxwell&f=false McCurdy, K., Gorman, K. S., Kisler, T., & Metallinos Katsaras, E. (2014). Associations between family food behaviors, ma ternal depression, and child weight among low income children. Appetite , 79 , 97 105. https://doi.org/10.1016/J.APPET.2014.04.015 McEwen, B. S., & Gianaros, P. J. (2010). Central role of the brain in stress and adaptation: Links to socioeconomic status, hea lth, and disease. Biology of Disadvantage: Socioeconomic Status and Health , 1186 , 190 222. https://doi.org/DOI 10.1111/j.1749 6632.2009.05331.x McMillen, I. C., MacLaughlin, S. M., Muhlhausler, B. S., Gentili, S., Duffield, J. L., & Morrison, J. L. (2008). Developmental origins of adult health and disease: the role of periconceptional and foetal nutrition. Basic & Clinical Pharmacology & Toxicology , 102 (2), 82 89. https://doi.org/10.1111/j.1742 7843.2007.00188.x Mehta, M. A., Gore Langton, E., Golembo, N., Colvert, E., Williams, S. C. R., & Sonuga Barke, E. (2010). Hyporesponsive Reward Anticipation in the Basal Ganglia following Severe Institutional Deprivation Early in Life. Journal of Cognitive Neuroscience , 22 (10), 2316 2325. https://doi.org/10.1162/jocn .2009.21394 Metcalf, P., Scragg, R., Willoughby, P., Finau, S., & Tipene Leach, D. (2000). Ethnic differences in perceptions of body size in middle aged European, and Pacific People living in New Zealand. International Journal of Obesity , 24 (5), 593 599. https://doi.org/10.1038/sj.ijo.0801202

PAGE 168

156 S. E. (2008). Influence of race and socioeconomic status on sleep: Pittsburgh SleepSCORE project. Psychosomatic Me dicine , 70 (4), 410 416. https://doi.org/10.1097/PSY.0b013e31816fdf21 Mikkilä, V., Räsänen, L., Raitakari, O. T., Pietinen, P., & Viikari, J. (2005). Consistent dietary patterns identified from childhood to adulthood: The Cardiovascular Risk in Young Finns Study. British Journal of Nutrition , 93 (06), 923. https://doi.org/10.1079/BJN20051418 Miller, A. L., Clifford, C., Sturza, J., Rosenblum, K., Vazquez, D. M., Kaciroti, N., & Lumeng, J. C. (2013). Blunted cortisol response to stress is associated with highe r body mass index in low income preschool aged children. Psychoneuroendocrinology , 38 (11), 2611 2617. https://doi.org/10.1016/j.psyneuen.2013.06.014 Miller, A. L., Lumeng, C. N., Delproposto, J., Florek, B., Wendorf, K., & Lumeng, J. C. (2013). Obesity Rel ated Hormones in Low Income Preschool Age Children: Implications for School Readiness. Mind, Brain, and Education , 7 (4), 246 255. https://doi.org/10.1111/mbe.12034 Miller, A. L., Rosenblum, K. L., Retzloff, L. B., & Lumeng, J. C. (2016). Observed self regu lation is associated with weight in low income toddlers. Appetite , 105 , 705 712. https://doi.org/10.1016/J.APPET.2016.07.007 Ministry of Health. (2017). New Zealand Health Survey. Retrieved April 11, 2018, from https://minhealthnz.shinyapps.io/nz health survey 2016 17 annual data explorer/_w_ae780c68/#!/home Monasta, L., Batty, G. D., Cattaneo, A., Lutje, V., Ronfani, L., Van Lenthe, F. J., & Brug, J. (2010). Early life determinants of overweight and obesity: a review of systematic reviews. for the Study of Obesity , 11 (10), 695 708. https://doi.org/10.1111/j.1467 789X.2010.00735.x Moore, T., Gauld, R., & Williams, S. (2007). Implementing Baby Friendly Hospital Initiative policy: the case of New Zealand public hospitals. International Breastf eeding Journal , 2 (1), 8. https://doi.org/10.1186/1746 4358 2 8 Morton, S. M. B., Atatoa Carr, P. E. (Polly E., Bandara, D. K. (Dinusha K., Grant, C. C., Ivory, (2010). Before we are born . http://www.growingup.co.nz/media/12254/growing up in new zealand before we are born nov 2010.pdf . Growing Up in New Zealand. Retrieved from https://researchspace.auckland.ac.nz/handle/2292/6120 Morton, S. M. B., Atatoa Carr, P. E., Gran Wall, C. (2013). Cohort Profile: Growing Up in New Zealand. International Journal of Epidemiology , 42 (1), 65 75. https://doi.org/10.1093/ije/dyr206

PAGE 169

157 Morton, S. M. B., Carr, P. E. A., Berry, S. D., Grant Bandara, C. C. (2014). Growing Up in New Zealand: A longitudinal study of New Zealand children and their families Residential Mobility Report 1: Moving house in the first 1000 days. Retrieved from https://cdn.auckland. ac.nz/assets/growingup/research findings impact/report05.pdf (2015). Growing Up in New Zealand cohort alignment with all New Zealand births. Australian and New Zealand Journal of Public Health , 39 (1), 82 87. https://doi.org/10.1111/1753 6405.12220 MSD. (n.d.). Ethnic Identity and Intimate Partner Violence in a New Zealand Birth Cohort Ministry of Social Development. Retrieved from https://www.msd.govt.nz/about msd and our work/publications resources/journals and magazines/social policy journal/spj33/33 ethnic identity and intimate partner violence in a new zealand birth cohort p126 145.html Muthén, L. K., & Muthén, B. (1998). Mplus. The comprehensive modelling p rogram for applied , 5 . of Child Health and Human Development Early Child Care Research Network. (2006). Identifying risk for obesity in early childhood. Pediatrics , 118 (3), e594 601. https://doi.org/10.1542/peds.2005 2801 Profile: The Mater University of Queensland Study of Pregnancy (MUSP). International Journal of Epidemiology , 34 (5), 992 997. https://doi.org/10.1093/ije/dyi119 Nederhof, E., & Schmidt, M. V. (2012). Mismatch or cumulative stress: Toward an integrated hypothesis of programming effects. Physiology and Behavior . https://doi.org/10.1016/j.physbeh.2011.12.008 Nielsen, L. S., Dan ielsen, K. V., & Sørensen, T. I. A. (2011). Short sleep duration as a possible cause of obesity: critical analysis of the epidemiological evidence. Obesity Reviews , 12 (2), 78 92. https://doi.org/10.1111/j.1467 789X.2010.00724.x Noble, H., & Smith, J. (2015 ). Issues of validity and reliability in qualitative research. Evidence Based Nursing , 18 (2), 34 35. https://doi.org/10.1136/eb 2015 102054 NVivo qualitative data analysis software | QSR International. (n.d.). Retrieved March 24, 2018, from https://www.qsr international.com/nvivo/home Oishi, S., & Diener, E. (2009). Goals, Culture, and Subjective Well Being (pp. 93 108). Springer, Dordrecht. https://doi.org/10.1007/978 90 481 2352 0_5

PAGE 170

158 Oken, E., Taveras, E. M., Kleinman, K. P., Rich Edwards, J. W., & Gillman , M. W. (2007). Gestational weight gain and child adiposity at age 3 years. American Journal of Obstetrics and Gynecology , 196 (4). https://doi.org/10.1016/j.ajog.2006.11.027 Ong, K. K. L., Preece, M. A., Emmett, P. M., Ahmed, M. L., & Dunger, D. B. (2002). Size at Birth and Early Childhood Growth in Relation to Maternal Smoking, Parity and Infant Breast Feeding: Longitudinal Birth Cohort Study and Analysis. Pediatric Research , 52 (6), 863 867. https://doi.org/10.1203/00006450 200212000 00009 Panter brick, C., & Eggerman, M. (2012). Understanding culture, resilience, and mental health: The production of hope. In The social ecology of resilience: A handbook of theory and practice (pp. 369 386). https://doi.org/10.1007/978 1 4614 0586 3 Parks, E. (2012). Influence of stress in parents on child obesity and related behaviors. Pediatrics , 130 (5), e1096 104. https://doi.org/10.1542/peds.2012 0895 Pearson, R. M., Bor (2016). Maternal perinatal mental health and offspring academic achievement at age 16: the mediating role of childhood executive function. Journal of Child Psychology and Psychiatry , 57 (4), 491 501. https://doi.org/10.1111/jcpp.12483 Aotearoa, New Zealand. Pimatisiwi n: A Journal of Aboriginal and Indigenous Community Health Journal of Aboriginal and Indigenous Community Health , 10 (103). Retrieved from http://www.mellowparenting.org/wp content/uploads/2016/03/Adapting Mellow Parenting for mothers in Aotearoa NZ. pdf Pervanidou, P., & Chrousos, G. P. (2016). Stress and Pediatric Obesity: Neurobiology and Behavior. Family Relations , 65 (1), 85 93. https://doi.org/10.1111/fare.12181 Pieper, J. R., & Laugero, K. D. (2013). Preschool children with lower executive functi on may be more vulnerable to emotional based eating in the absence of hunger. Appetite , 62 , 103 109. https://doi.org/10.1016/J.APPET.2012.11.020 Pinson Millburn, N. M., Fabian, E. S., Schlossberg, N. K., & Pyle, M. (1996). Grandparents Raising Grandchildre n. Journal of Counseling & Development , 74 (6), 548 554. https://doi.org/10.1002/j.1556 6676.1996.tb02291.x Plunket. (2018). Who we are Plunket. Retrieved April 8, 2018, from https://www.plunket.org.nz/what we do/who we are/ Podsiadlowski, A. (2011). Collectivist Value Orientations among Four Ethnic Groups: Collectivism in the New Zealand Context . New Zealand Journal of Psychology (Vol. 40). Retrieved from http://www.psychology.org.nz/wp content/uploads/Padsiadlowski.pdf

PAGE 171

159 Pollock, N. J. (1995). Cultural elaborations of obesity fattening practices in Pacific Societies. Asia Pacific Journal of Clinical Nutrition . Retrieved from http://apjcn.nhri.org.tw/server/APJCN.../4/4/357.pdf Prepared by Anne Spellerberg (Social Policy) assisted by the Social Capital P rogramme team Framework for the Measurement of Social Capital in New Zealand . (n.d.). Retrieved from http://unpan1.un.org/intradoc/groups/public/documents/APCITY/UNPAN011127.pdf Ratima, M., & Barbara, G. (2007). Peer Commentary 1 Thinking about difference across and within mentoring . Retrieved from https://www.researchgate.net/publication/26500233 Rice, F., Harold, G. T., Boivin, J., van den Bree, M., Hay, D. F., & Thapar, A. (2010). The links between prenatal stress and offspring development and psychopath ology: disentangling environmental and inherited influences. Psychological Medicine , 40 (02), 335. https://doi.org/10.1017/S0033291709005911 Riggs, N. R., Spruijt Metz, D., Chou, C. P., & Pentz, M. A. (2012). Relationships between executive cognitive functi on and lifetime substance use and obesity related behaviors in fourth grade youth. Child Neuropsychology , 18 (1), 1 11. https://doi.org/10.1080/09297049.2011.555759 Rodney, N. C., & Mulligan, C. J. (2014). A biocultural study of the effects of maternal stress on mother and newborn health in the democratic republic of Congo. American Journal of Physical Anthropology , 155 (2). https://doi.org/10.1002/ajpa.22568 Rondó, P. H. C., Ferreira, R. F., Nogueira, F., Ribeiro, M. C. N., Lobert, H., & Artes, R. (2003). Maternal psychological stress and distress as predictors of low birth weight, prematurity and intrauterine growth retardation. European Journal of Clinical Nutrition , 57 (2), 266 272. https://doi.org/10.1038/sj.ejcn.1601526 Sable, M. R., & Wilkinson, D. S. (2000). Impact of perceived stress, major life events and pregnancy attitudes on low birth weight. Family Planning Perspectives , 32 (6), 288 294. https://doi.org/10.23 07/2648197 Sallis, J. F., Floyd, M. F., Rodríguez, D. A., & Saelens, B. E. (2012). Role of built environments in physical activity, obesity, and cardiovascular disease. Circulation , 125 (5), 729 737. https://doi.org/10.1161/CIRCULATIONAHA.110.969022 Salmond (NZDep) and Its Uptake as a National Policy Tool. Canadian Journal of Public Health / Revue Canadienne de Santé Publique . Canadian Public Health Association. https://doi.org/10. 2307/41995682

PAGE 172

160 Sandman, C. A., Wadhwa, P. D., Dunkel Schetter, C., Chicz DeMet, A., Belman, J., Porto, M., human fetus and infant birth outcomes. Annals of the New Y ork Academy of Sciences , 739 , 198 210. https://doi.org/10.1111/j.1749 6632.1994.tb19822.x Schaap, M. M., & Kunst, A. E. (2009). Monitoring of socio economic inequalities in smoking: Learning from the experiences of recent scientific studies. Public Health , 123 (2), 103 109. https://doi.org/10.1016/J.PUHE.2008.10.015 Schack Nielsen, L., Michaelsen, K. F., Gamborg, M., Mortensen, E. L., & Sørensen, T. I. A. (2010). Gestational weight gain in relation to offspring body mass index and obesity from infancy throug h adulthood. International Journal of Obesity , 34 (1), 67 74. https://doi.org/10.1038/ijo.2009.206 Schluter, P., Carter, S., & Kokaua, J. (1248). Indices and perception of crowding in Pacific households domicile within Auckland, New Zealand: findings from t he Pacific Islands Families Study . THE NEW ZEALAND MEDICAL JOURNAL (Vol. 120). Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.502.1998&rep=rep1&type=pdf Schmied, V., Johnson, M., Naidoo, N., Austin, M. Yeo, A. (2013). Maternal mental health in Australia and New Zealand: A review of longitudinal studies. Women and Birth , 26 (3), 167 178. https://doi.org/10.1016/J.WOMBI.2013.02.006 Selye, H. (1955). Stress and disease. The Laryngoscope , 65 (7), 500 514. http s://doi.org/10.1288/00005537 195507000 00002 (2014). Parental stress increases body mass index trajectory in pre adolescents. Pediatric Obesity , 9 (6), 435 442. https://doi.org/10.1111/j.2047 6310.2013.00208.x Shonkoff, J. P., Boyce, W. T., & McEwen, B. S. (2009). Neuroscience, molecular biology, and the childhood roots of health disparities: building a new framework for health promotion and disease prevention. Ja ma , 301 (21), 2252 2259. https://doi.org/10.1001/jama.2009.754 Wood, D. L. (2012). The Lifelong Effects of Early Childhood Adversity and Toxic Stress. PEDIATRICS , 129 (1), e232 e246. https://doi.org/10.1542/peds.2011 2663 Shonkoff JP, B. W. (2009). Neuroscience, molecular biology, and the childhood roots of health disparities: Building a new framework for health promotion and disease prevention. JAMA , 301 (21), 2252 2 259. https://doi.org/10.1001/jama.2009.754 Slykerman, R. F., Thompson, J. M. D., Pryor, J. E., Becroft, D. M. O., Robinson, E., Clark, P. intelligence. Early Human Devel opment , 81 (10), 815 821. https://doi.org/10.1016/J.EARLHUMDEV.2005.05.005

PAGE 173

161 Staneva, A., Bogossian, F., Pritchard, M., & Wittkowski, A. (2015). The effects of maternal depression, anxiety, and perceived stress during pregnancy on preterm birth: A systematic review. Women and Birth . https://doi.org/10.1016/j.wombi.2015.02.003 Statistics New Zealand, & Ministry of Pacific Island Affairs. (2011). Health and Pacific Peoples in New Zealand . Pacific Progress . Retrieved from http://stats.govt.nz/browse_for_stats/peo ple_and_communities/pacific_peoples/pacific progress health.aspx Stout, S. A., Espel, E. V, Sandman, C. A., Glynn, L. M., & Davis, E. P. (2015). Fetal Psychoneuroendocrinology , 53 , 29 39. https://doi.org/10.1016/j.psyneuen.2014.12.009 Stults Kolehmainen, M. A., & Sinha, R. (2014). The effects of stress on physical activity and exercise. Sports Medicine (Auckland, N.Z.) , 44 (1), 81 121. https://doi.org/10.1007/s40279 013 0090 5 Suglia, S. F., Duarte, C. S., Chambers, E. C., & Boynton Jarrett, R. (2012). Cumulative Social Risk and Obesity in Early Childhood. Pediatrics , 129 (5). Retrieved from http://pediatrics.aappublications.org/content/129/5/e1173.short J. of pers onality and, & 1998, undefined. (n.d.). The shifting basis of life satisfaction judgments across cultures: Emotions versus norms. Psycnet.Apa.Org . Retrieved from http://psycnet.apa.org/fulltext/1997 38975 015.html Talge, N. M., Neal, C., & Glover, V. (200 7). Antenatal maternal stress and long term effects on child neurodevelopment: How and why? Journal of Child Psychology and Psychiatry and Allied Disciplines . https://doi.org/10.1111/j.1469 7610.2006.01714.x Tallman, P. S. (2016). The Index of Vulnerabilit y: An anthropological method linking social ecological systems to mental and physical health outcomes. Social Science & Medicine , 162 , 68 78. https://doi.org/10.1016/j.socscimed.2016.06.016 Tapera, R., Harwood, M., & Anderson, A. (2017). A qualitative Kaup Auckland, New Zealand. Public Health Nutrition , 20 (06), 1090 1098. https://doi.org/10.1017/S1368980016002950 Thayer, Z. M., & Kuzawa, C. W. (2014). Early origins of health disparities: Material deprivation predicts maternal evening cortisol in pregnancy and offspring cortisol reactivity in the first few weeks of life. American Journal of Human Biology , 26 (6), 723 730. https://doi.org/10.1002/ajhb.22532 Thayer, Z. M., & Kuzawa, C. W. (2015). Ethnic discrimination predicts poor self rated health and cortisol in pregnancy: Insights from New Zealand. Social Science & Medicine , 128 , 36 42. https://doi.org/10.1016/j.soc scimed.2015.01.003

PAGE 174

162 Thoits, P. A. (1995). Stress, Coping, and Social Support Processes: Where Are We? What Next? Journal of Health and Social Behavior , 35 , 53. https://doi.org/10.2307/2626957 Thoits, P. A. (2011). Mechanisms Linking Social Ties and Support to Physical and Mental Health. Journal of Health and Social Behavior , 52 (2), 145 161. https://doi.org/10.1177/0022146510395592 Thornton, A., Kerslake, M. T., & Binns, T. (2010). Alienation and obligation: Religion and social change in Samoa. Asia Pacific V iewpoint , 51 (1), 1 16. https://doi.org/10.1111/j.1467 8373.2010.01410.x Tomiyama, A. J., Schamarek, I., Lustig, R. H., Kirschbaum, C., Puterman, E., Havel, P. J., & Epel, E. S. (2012). Leptin concentrations in response to acute stress predict subsequent in take of comfort foods. Physiology & Behavior , 107 (1), 34 39. https://doi.org/10.1016/J.PHYSBEH.2012.04.021 Tracer David P. ( 2004 ) . Market Integration, Reciprocity and Fairness in Rural Papua New Guinea: Results from a Two Village Ultimatum Game Study. In J. Henrich, R. Boyd, S. Bowles, C. Camerer, E. Fehr and H. Gintis (eds.) Foundations of Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen Small Scale Societies. Oxford University Press. pp. 232 259. Turner, R. J., & Avison, W. R. (2003). Status variations in stress exposure: implications for the interpretation of research on race, socioeconomic status, and gender. Journal of Health and Social Behavior , 44 (4), 488 505. https://doi.org/10.2307/1519795 Umberson, D., Crosnoe, R., & Reczek, C. (2010). Social Relationships and Health Behavior Across Life Course. Annual Review of Sociology , 36 , 139 157. https://doi.org/10.1146/annurev soc 070308 120011 Utter, J., Scragg, R., Schaaf, D., & Fitzgerald, E. (200 6a). Nutrition and physical activity behaviours among Mäori, Pacific and NZ European children: identifying opportunities for population based interventions. Australian and New Zealand Journal of Public Health , 30 (1), 50 56. https://doi.org/10.1111/j.1467 8 42X.2006.tb00086.x Utter, J., Scragg, R., Schaaf, D., & Fitzgerald, E. (2006b). Nutrition and physical activity behaviours among Mäori, Pacific and NZ European children: identifying opportunities for population based interventions. Australian and New Zeala nd Journal of Public Health , 30 (1), 50 56. https://doi.org/10.1111/j.1467 842X.2006.tb00086.x Wadhwa, P. D., Entringer, S., Buss, C., & Lu, M. C. (2011). The contribution of maternal stress to preterm birth: issues and considerations. Clinics in Perinatolo gy , 38 (3), 351 384. https://doi.org/10.1016/j.clp.2011.06.007 Wadhwa, P. D., Sandman, C. A., & Garite, T. J. (2001). The neurobiology of stress in human pregnancy: Implications for prematurity and development of the fetal central nervous

PAGE 175

163 system. In Progres s in Brain Research (Vol. 133, pp. 131 142). https://doi.org/10.1016/S0079 6123(01)33010 8 Walton, K., Simpson, J. R., Darlington, G., & Haines, J. (2014). Parenting stress: a cross sectional analysis of associations with childhood obesity, physical activi ty, and TV viewing. BMC Pediatrics , 14 , 244. https://doi.org/10.1186/1471 2431 14 244 [doi] Wells, J. C. K., Chomtho, S., & Fewtrell, M. S. (2007). Programming of body composition by early growth and nutrition. The Proceedings of the Nutrition Society , 66 ( 3), 423 434. https://doi.org/10.1017/S0029665107005691 White, I. R., Royston, P., & Wood, A. M. (2011). Multiple imputation using chained equations: Issues and guidance for practice. Statistics in Medicine , 30 (4), 377 399. https://doi.org/10.1002/sim.4067 WHO (World Health Organization). (2013). WHO obesity and overweight fact sheet no 311. Obesity and Oveweight Fact Sheet , (March), 4 7. https://doi.org/http://www.who.int/mediacentre/factsheets/fs311/en/ Witten, K., Kearns, R., McCreanor, T., Penney, L., & Faalau, F. (2009). Connecting place and the everyday practices of parenting: Insights from Auckland, New Zealand. Environment and Planning A . https://doi.org/10.1068/a41377 Woods, S. M., Melville, J. L., Guo, Y., Fan, M. Y., & Gavin, A. (2010). Psychosocia l stress during pregnancy. American Journal of Obstetrics and Gynecology , 202 (1), 61.e1 61.e7. https://doi.org/10.1016/j.ajog.2009.07.041 Worrall, J. (2009). Acknowledgements . Retrieved from https://www.grg.org.nz/site/grg/GRG 2009 Research Report A view over time.pdf (2017). Prenatal stress, methylation in inflammation related genes, and adiposity measures in early childhood. Psychosomatic Medicine , 1. https: //doi.org/10.1097/PSY.0000000000000517 Yamashiro, G., & Matsuoka, J. K. (1997). Help Seeking among Asian and Pacific Americans: A Multiperspective Analysis. Social Work , 42 (2), 176 186. https://doi.org/10.1093/sw/42.2.176 Young Hyman, D., Schlundt, D. G., Herman, L., De Luca, F., & Counts, D. (2001). Evaluation of the insulin resistance syndrome in 5 to 10 year old overweight/obese African American children. Diabetes Care , 24 (8), 1359 1364. https://doi.org/10.2337/diacare. 24.8.1359 Zijlmans, M. A. C., Riksen Walraven, J. M., & de Weerth, C. (2015). Associations between maternal prenatal cortisol concentrations and child outcomes: A systematic review. Neuroscience & Biobehavioral Reviews , 53 , 1 24.

PAGE 176

164 https://doi.org/10.1016/J. NEUBIOREV.2015.02.015