Pain-Associated Stressor Exposure and Cortisol Values at

37 Postmenstrual Weeks for Premature Infants in Neonatal Intensive Care

Annie J. Rohan

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy under Executive Committee of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2013

© 2013 Annie Jill Rohan All rights reserved

ABSTRACT

Pain-Associated Stressor Exposure and Cortisol Values at

37 Postmenstrual Weeks for Premature Infants in Neonatal Intensive Care

Annie J. Rohan

Background: Ongoing stress has costs in brain and body adaptations. Recurrent and prolonged stress taxes the adaptive capacity of the premature infant and has been postulated as a risk factor for the development of later disease (Sullivan, Hawes,

Winchester, & Miller, 2008). Cortisol appears to be an important mediator in the relationship between prolonged stress and subsequent disease and disability. Neonatal

Intensive Care Unit (NICU) stress may give rise to early recognizable patterns of adrenal axis adaptation in prematurely born infants. These adaptations may underlie difficulties in learning and development for a population of children that already bears a high burden of disease. Premature infants who spend their first weeks of life in the NICU have been shown to have altered glucocorticoid values at approximately one year of age (Grunau et al., 2007; Grunau, Weinberg, & Whitfield, 2004; Haley, Weinberg, & Grunau, 2006). It is hypothesized that adaptations of the adrenal axis may begin to emerge even earlier - during NICU hospitalization - and in relation to recurrent pain-associated procedural exposure or the prolonged stress of assisted ventilation.

Objective: The primary purpose of this research was to examine the relationship between random cortisol values at 37 postmenstrual weeks of age and pain-associated stressor exposure in prematurely born infants who received extended care in the NICU.

Because random cortisol values in premature infants have been shown to be relatively dissociated from clinical conditions, this research also explored the relationship between values of the cortisol precursor, 17-hydroxyprogesterone (17-OHP), and pain-associated stressor exposure in prematurely born infants.

Methods: A descriptive, cross-sectional design was used to examine relationships between two major types of pain-associated stressor exposures - number of recurrent skin-breaking procedures and hours of assisted ventilation - and random serum cortisol and 17-OHP values at 37 postmenstrual weeks of age. Subjects were 59 infants born at under 35 completed weeks of gestation, who were at least two weeks of age, and who had been cared for in in the NICU since birth. Infants with intrauterine growth restriction, major anomalies, or exposure to postnatal steroids or surgery were excluded.

Results: During this study, the 59 prematurely born infants were collectively exposed to over 3350 pain-associated procedures and over 17,000 hours of assisted ventilation. The greatest number of individual pain-associated exposures was 238, and no infant was exposed to less than 20 procedures (mean = 57). Over half of these exposures were in the first week after birth. There was a significant negative correlation (Spearman rank, one- tailed) between recurrent pain-associated stressor exposure in premature infants and their values of 17-OHP at 37 postmenstrual weeks of age (r= -0.232, p=0.039). A trend was

also identified for differences in 17-OHP values between infants who had the highest and lowest recurrent pain-associated stressor exposure (U=68.5, p=0.068). No significant relationships were identified between cortisol values and any of the study variables. Due to multicollinearity, it was not possible to examine the independent effects of recurrent pain-associated stressor exposure, gestational age, chronological age or hours of ventilation on 17-OHP values. One-tailed correlation analyses, however, failed to identify any significant relationships between the collinear variables or birthweight and

17-OHP values.

Conclusions: Premature infants are exposed to a high number of recurrent skin-breaking procedures while receiving care in the NICU, especially during their first week of age.

There is a statistically significant relationship between the incidence of these recurrent pain-associated procedures and an aspect of the adrenocorticoid profile at 37 postmenstrual weeks of age. Recurrent pain-associated stressor exposure may be a more important factor in explaining the variance of 17-OHP values at 37 postmenstrual weeks of age than birthweight, gestational age, chronological age or hours of assisted ventilation.

Table of Contents

List of Tables iv

List of Figures vii

Acknowledgements ix

Dedication xi

CHAPTER 1: Introduction 1

Problem Statement 4

Research Purpose and Study Objectives 5

Research Question, Specific Aims and Hypotheses 6

Conceptual Framework 7

Assumptions 8

Significance 9

CHAPTER 2: Review of the Literature 11

Overview of Neurologic Development and the Infant HPA System 11

Stress and Pain in Infancy 28

Perinatal Events, Stress Adaptation and Developmental Sequelae 38

Pain-Associated NICU Stress and the Adrenocorticoid Phenotype 43

Overall Summary 49

CHAPTER 3: Methods 51

Overview 51

i

Operational Definitions 51

Setting 53

Subjects 53

Procedures 55

Measures 62

Summary of Variables 68

Data Management and Analysis Plan 70

Protection of Human Subjects 73

CHAPTER 4: Results 76

Description of Sample 76

Inter-rater Reliability Testing 87

Stressor Variables 89

Glucocorticoid Variables 98

Correlation Analyses 103

Hypothesis Testing 115

Exploratory Analyses 119

CHAPTER 5: Discussion 130

Introduction 130

Cortisol and 17-OHP Values 132

Recurrent Exposure to Pain-Associated Procedures 136

Recent Pain-Associated Procedures 138

Days of Age to Attain Birthweight 139

ii

Male/Female Differences 139

Multicollinearity 141

Relationship between Study Variables and Adrenocorticoid Values 142

Recruitment and Sampling Issues 145

Study Strengths 146

Challenges and Limitations 147

Implications for Practice 150

Implications for Research 151

Conclusions 153

References 155

Appendices

A. Data Collection Form 179

B. Nursing Standards of Care Document 182

C. Columbia University IRB Approval 184

D. Stony Brook University IRB Approval 185

E. Research Permission Form 187

iii

List of Tables

Table 3.1. Operational Definitions 52

Table 3.2. Summary of Study Variables 69

Table 4.1. Demographic Characteristics of Study Sample 80

Table 4.2. Breakdown of Subjects by Gestational Age and Distribution of Twins 81

Table 4.3. Distribution Descriptors for Independent Variables 83

Table 4.4. Characteristics of Recurrent Pain-Associated Procedures 90

Table 4.5. Recurrent Stressor Exposure in First Week of Age 91

Table 4.6. Characteristics of Hours of Assisted Ventilation Exposure 92

Table 4.7. Assisted Ventilation Exposure in First Week of Age 92

Table 4.8. Distribution Descriptors for Stressor Variables 95

Table 4.9. Bivariate Analyses Testing for Differences in Recurrent Procedural Exposure 97

Table 4.10. Bivariate Analyses Testing for Differences in Assisted Ventilation Exposure 98

Table 4.11. Characteristics of Cortisol and 17-OHP at 37 Postmenstrual Weeks of Age 99

Table 4.12. Characteristics of Cortisol and 17-OHP: Gestational Age Subgroups 100

Table 4.13. Distribution Descriptors for Cortisol and 17-OHP 100

Table 4.14. Bivariate Analyses Testing for Differences in Cortisol Values 102

iv

Table 4.15. Bivariate Analyses Testing for Differences in 17-OHP Values 103

Table 4.16. Identification of Collinearity 114

Table 4.17. Hypothesis Testing: Correlations between Stressor Variables and Cortisol 116

Table 4.18. Correlations between Ventilation Variables and Cortisol 117

Table 4.19. Hypothesis Testing: Correlations between Stressor Variables and 17-OHP 118

Table 4.20. Correlations between Ventilation Variables and 17-OHP 118

Table 4.21. Summary of Simple Linear Regression Model for 17-OHP 120

Table 4.22. Summary of Multiple Linear Regression Model for 17-OHP 121

Table 4.23. Correlations between 17-OHP and Collinear Independent Variables 122

Table 4.24. Bivariate Analyses Testing for Differences in Glucocorticoid Values after Recent Pain-Associated Procedures 123

Table 4.25 Bivariate Analyses Testing for Differences in Weight between Male and Female Infants 124

Table 4.26 Bivariate Analyses Exploring Days to Attain Birthweight and Exposure to Stressors 125

Table 4.27. Bivariate Analyses Testing for Differences in Glucocorticoid Values between Infants who Attain Birthweight Earlier and Later 126

Table 4.28. Bivariate Analyses Testing for Differences in Cortisol Values between Infants with Extremes of Stress Exposure 127

Table 4.29. Bivariate Analyses Testing for Differences in 17-OHP Values between Infants with Extremes in Stress Exposure 128

v

Table 4.30. Bivariate Analyses Testing for Differences in Glucocorticoid Values between Larger-Born and Smaller-Born Twins 129

Table 5.1. Characteristics of Cortisol (ug/dl) at 37 Postmenstrual Weeks of Age: Comparisons with Other Researchers 134

Table 5.2. Characteristics of 17-OHP (ng/dl) at 37 Postmenstrual Weeks of Age: Comparisons with Other Researchers 135

vi

List of Figures

Figure 1.1. The Life Cycle Model of Stress 8

Figure 3.1. Flow Chart of Research Procedures 59

Figure 4.1. Flow Diagram of Recruitment and Inclusion 78

Figure 4.2. Distribution of Gestational Age at Birth 83

Figure 4.3. Distribution of Birthweight 84

Figure 4.4. Distribution of Days of Age to Attain Birthweight 84

Figure 4.5. Distribution of Recent Pain-Associate Procedures 85

Figure 4.6. Distribution of Weight at 37th Postmenstrual Week of Age 85

Figure 4.7. Distribution of Days of Age at Lab Testing 86

Figure 4.8. Distribution of Exposures to Pain-Associated Procedures 93

Figure 4.9. Distribution of Hours of Assisted Ventilation 94

Figure 4.10. Distribution of Hours of Invasive Assisted Ventilation 94

Figure 4.11. Distribution of Hours of Non-invasive Assisted Ventilation 95

Figure 4.12. Distribution of Cortisol Values 101

Figure 4.13. Distribution of 17-OHP Values 101

Figure 4.14. Scatterplot: Recurrent Pain-Associated Procedures versus Hours of Assisted Ventilation 104

Figure 4.15. Scatterplot: Cortisol versus 17-OHP 104

Figure 4.16. Scatterplot: Recurrent Pain-Associated Procedures versus Cortisol 105

vii

Figure 4.17. Scatterplot: Hours of Assisted Ventilation versus Cortisol 106

Figure 4.18. Scatterplot: Recurrent Pain-Associated Procedures versus 17-OHP 107

Figure 4.19. Scatterplot: Hours of Assisted Ventilation versus 17-OHP 107

Figure 4.20. Phi Correlation Matrix 109

Figure 4.21. Point-Biserial Correlation Matrix 110

Figure 4.22. Spearman Rank Correlation Matrix 111

Figure 4.23. Pearson Correlation Matrix 113

viii

Acknowledgements

Just as it takes a village to raise a child, Ph.D. students do not arrive to graduation without years of support by a small army of enthusiasts.

First, I would like to thank my dissertation sponsor, Dr. Mary Byrne, whose guidance was integral to the successful completion of this project. During a time of extraordinary circumstance, she has maintained meticulous attention to detail and relentless commitment to the highest research standards. By demanding clarity in my thoughts and writing, she has also accomplished the most difficult task of harnessing the chaotic intelligence of an established front-line ICU clinician.

I am also very appreciative of the wisdom and support of my committee members, Dr. Joan

Kearney and Dr. William Dan Roberts. Despite very busy schedules, they gave me the time and confidence needed to complete this endeavor. In addition, Dr. Nancy Reame and Dr. John

Lorenz - both renowned experts in their fields – so kindly agreed to serve as readers. Their generous contributions considerably transformed this project into an example of interdisciplinary and translational research.

My deepest gratitude goes to Donald and Barbara Jonas and the Jonas Center for Nursing

Excellence, who have provided unprecedented funding for nursing education and who supported me as one of their first Jonas Nursing Scholars. I am determined to see that their vision for the future of nursing is a keystone for the next stages in my career.

ix

Special thanks are due to my cherished friends who have seen me along this path. Nurse or not, you are an exceptional group of caretakers. I have been inspired by the enthusiasm, compassion, effort, ingenuity and generosity that each of you brings to our world.

Finally, and most significantly, an exceptional acknowledgement is owed to my family, who has unconditionally supported me on this journey. My incredible children - Alexa, Jack and Arielle - have been magnificent both in endorsing my studies and failing to allow me to become lost in them. My husband and best friend, Jimmie, has as much – if not more – earned this degree by keeping me upbeat, focused, nourished and caffeinated. He has tirelessly maintained the mundane, so that I could immerse myself in the fanciful. More so, he has managed to love me despite the many days I never changed from my pajamas.

Almost thirty years ago, I had a conversation with my dad regarding the practicality of pursuing a degree in nursing. Given the cost of college and loss of my cashier wages, he calculated that I would need to work seven years beyond graduation to reach a “break-even point” with a nursing career. Twenty-five years past that first graduation, and now with eager anticipation of at least twenty-five more, I am pleased to say that we made a VERY good choice.

--Annie

x

Dedication

I dedicate this research to neonatal intensive care nurses. There is perhaps no patient more vulnerable in our healthcare system as the one who is born prematurely and ill, to a mother who is sick, overwhelmed and intimidated. The tireless work of neonatal nurses, who relentlessly advocate in the face of politics, economics, traineeships and defensive medicine, assures dignity and balance for these smallest, youngest, voiceless patients. At times, the commitment of these special nurses falls nothing short of heroic.

And certainly, to the likes of Baby Paul, Baby Brown, Baby Lawson and Baby Harding: know that each of you has made an impact.

xi

1

CHAPTER 1: Introduction

The rate of premature birth continues to rise in most countries (World Health

Organization, 2012). Recent reports indicate that the United States now has a premature birth rate exceeding 12 percent of all live births, and is faced with over 543,000 prematurely born infants each year (National Center for Chronic Disease Prevention and

Health Promotion - Division of Reproductive Health, 2010). In addition, although the rate of premature birth continues to rise worldwide, preterm birth rates in the United

States are higher than 130 other countries who report these statistics, including many poorer nations (World Health Organization, 2012). Most premature infants will require specialized hospital care. For many, this care will be carried out over several weeks or months in a Neonatal Intensive Care Unit (NICU).

Stress is a condition of the human experience, and also a factor in the expression of disease. It is not just the toll of dramatic life events, but rather the many events of daily life that elevate activities of physiologic systems and cause some measure of wear and tear on these systems (McEwen & Seeman, 1999). An agent that produces a demand on the body is known as a stressor. Stressors include agents such as light, noise, cold, tissue trauma and infection. Some neuroscientists are beginning to transform the way we think about stress by proposing that stress refers to conditions whereby an agent’s demand for adaptation exceeds the natural regulatory capacity of the organism (Koolhaas et al.,

2011).

Care in the NICU is stressful for infants. As a consequence of caretaker efforts to support survival and treat disease, premature infants face numerous stimuli that challenge

2 their ability to maintain physiologic homeostasis. These interventions include diagnostic testing, nursing procedures, and treatments that have therapeutic intent. Painful stimuli are stressful, particularly for infants (Gunnar, Talge, & Herrera, 2009; Whitfield &

Grunau, 2000). Skin-breaking procedures and mechanical ventilation impart a significant burden of pain and stress on premature hospitalized infants (Aranda et al., 2005;

Cignacco et al., 2009; Porter, Wolf, Gold, Lotsoff, & Miller, 1997).

For many years, researchers have been interested in the long-term effect of stress on health and disease (Selye, 1976; Selye & Horava, 1953). The term allostasis was introduced by Sterling and Eyer in 1988 to describe the active process by which the body predicts, responds and adapts to daily events and stressors to maintain homeostasis

(Sterling & Eyer, 1988). The term allostatic load was introduced in 1993 by McEwen and Stellar to describe the cumulative physiologic toll of repeated attempts at stress adaptation. These researchers proposed that allostatic load was “the strain on the body produced by repeated ups and downs of physiologic response, as well as by the elevated activity of physiologic systems under challenge…” (McEwen & Stellar, 1993). Early life elevations in allostatic load are postulated to increase vulnerability to physical disease and psychosocial disorders (Gunnar, Frenn, Wewerka, & Van Ryzin, 2009; S. M. Korte,

Koolhaas, Wingfield, & McEwen, 2005; Lupien, McEwen, Gunnar, & Heim, 2009;

McEwen, 2001; McEwen & Wingfield, 2003).

The hypothalamic-pituitary-adrenal (HPA) axis is the major physiologic system responsible for regulating hormonal stress response. Recurrent and prolonged activation of the HPA axis has been implicated as a mechanism by which allostasis modifies health

3 outcomes. Stress activates the HPA axis, exposing tissues to elevated concentrations of glucocorticoids. Chronically high levels of these hormones damage tissues or can give rise to adrenal axis adaptations. These adaptations shield the body from excessive glucocorticoid exposure by modifying adrenal hormone response or function. While initially protective, these developmental adaptations may compromise the ability of the organism to adequately modulate stress later in life. It is postulated that early alterations in adrenocorticoid production is a mechanism that impedes the return of normal long- term psychological, social and physical functioning when environmental conditions improve (Dowd, Simanek, & Aiello, 2009; Phillips, 2001, 2007; Sullivan, Hawes, et al.,

2008; Ward et al., 2004).

An increasing body of evidence supports that stressful events in fetal and early life permanently alter the adrenocorticoid phenotype (Clark, 1998; Cutfield et al., 2004; Seckl

& Holmes, 2007; Ward, Syddall, Wood, Chrousos, & Phillips, 2004; Xita & Tsatsoulis,

2010). While elevated levels of infant stressor exposure during NICU hospitalization is reasonably accepted for extended survival, high cumulative exposure to stressors may nevertheless modify HPA axis activity and predispose premature infants to additional disease, compromised functioning, or homeostatic dysregulation throughout their lives.

This may be particularly manifest in the smallest and sickest patients, in whom diagnostic and life-sustaining therapies are associated with the highest overall exposure to pain- associated stressors (D. P. Barker & Rutter, 1995).

Despite a literature supporting a link between allostatic load, adrenal axis adaptation and subsequent disease, there is a paucity of research examining the relationship between

4 chronic or repetitive stress in the neonatal period and early adrenal axis adaptation. Little is known about the early adrenocorticoid phenotype that evolves following exposure by premature infants to prolonged and recurrent stressors in the NICU environment.

Examination of adrenal axis adaptations in prematurely born infants is particularly compelling, as this is a relatively large pediatric population with a significant lifetime burden of cognitive, psychological and physical disease.

Problem Statement

Ongoing stress has costs in brain and body adaptations. Recurrent and prolonged stress taxes the adaptive capacity of the premature infant and can result in biological and physiological changes that have been postulated as a risk factor for the development of later disease (Cutfield, et al., 2004; Grunau, et al., 2007; Sullivan, Hawes, et al., 2008).

Adrenal axis adaptations may predispose a population of medically fragile infants to superimposed health problems.

Chronic or repeated stress in infancy may give rise to early recognizable alterations of the adrenocorticoid phenotype. There is increasing evidence that very premature infants, who spend their first weeks in the NICU, have altered cortisol secretion subsequent to hospital discharge (Grunau, et al., 2007; Grunau et al., 2010; Grunau, et al., 2004; Haley, et al., 2006). There is reason to believe that adrenocorticoid adaptations emerge even earlier, and in relation to recurrent exposure to painful stimuli or to the stress of mechanical ventilation during NICU care.

The effect of stress in the neonatal period, and the extent to which the infant’s adrenal axis modifies following repeated attempts to maintain physiologic homeostasis is poorly

5 understood. Examining the relationship between major types of stressor exposures and glucocorticoid hormone values during NICU hospitalization adds to our understanding of the relationship between stress, allostasis and developmental adaptation in prematurely born infants.

Research Purpose and Study Objectives

The primary purpose of this research was to examine the relationship between pain- associated stressor exposure and the activity of adrenocortical hormones – the mediators of stress responses - at 37 postmenstrual weeks of age in prematurely born infants. The broad-acting hormone, cortisol, has been important in stress research, and was the primary adrenocortical hormone identified for examination in this study. The literature, however, suggests that cortisol values in premature infants may not be predictably influenced by clinical events, and that cortisol precursors may better reflect stressor exposure in this population. Therefore, in addition to cortisol, this research also explored the relationship between 17-hydroxyprogesterone (17-OHP) values at 37 postmenstrual weeks of age and pain-associated stressor exposure in prematurely born infants.

The examination of 17-OHP as a potential biomarker for stress, was more appealing than the examination of any other cortisol precursor or HPA axis hormone because all newborn infants have a random 17-OHP level drawn as part of hospital state screening processes for congenital adrenal hyperplasia. By choosing 17-OHP for examination, this research capitalized on a potential opportunity to use a broadly existing piece of clinical data for another purpose.

6

Research Question, Specific Aims and Hypotheses

This research study was developed to answer the following question:

What is the relationship between recurrent exposure to pain-associated procedures

or hours of assisted ventilation and random cortisol and 17-OHP values at 37

postmenstrual weeks of age in premature infants who have been cared for in the

NICU?

The specific aims associated with this question were as follows:

1) To characterize the nature of recurrent pain-associated procedures for premature

infants in the NICU from birth up until the 37th postmenstrual week of age

2) To examine relationships between recurrent exposure to pain-associated

procedures, hours of assisted ventilation, and random cortisol and 17-OHP values

at 37 postmenstrual weeks of age in a sample of premature infants who had

received care for a minimum of two weeks in the NICU

There were two hypotheses associated with the second research aim:

1) Recurrent exposure to pain-associated procedures and hours of assisted

ventilation will be negatively correlated with random cortisol values at 37

postmenstrual weeks of age.

2) Recurrent exposure to pain-associated procedures and hours of assisted

ventilation will be negatively correlated with random 17-OHP values at 37

postmenstrual weeks of age.

7

Conceptual Framework

The life cycle model of stress developed by Lupien, McEwen, Gunnar and Heim was incorporated as the conceptual framework for this study (Figure 1.1) (Lupien, et al.,

2009).

Lupien, McEwen, Gunnar and Heim (2009) developed a model for explaining the emergence of different disorders as they are temporally related to stress exposure. This

“life cycle model of stress” postulates that stressor exposure has enduring effects on the brain through activation of the HPA axis and release of glucocorticoids, with maximum neurodevelopmental impact at the time of stressor exposure. Stress during developmentally sensitive periods has particular capacity to alter neural development and produce enduring changes in stress responsivity.

This research focuses on the examination of pain-associated stressors as a possible foundation for early differentiation effects in the HPA axis of premature infants. Specific to this research, the model identifies that brain structures are vulnerable to change by significant stressor exposures as early as the first few weeks of age. Depending on the exposure, there can be changes in developing brain structures, evidenced by early increases or decreases in glucocorticoid production.

Lupien’s model guided development of this research’s directional hypothesis by identifying that traumatic stressful exposures in the prenatal and postnatal periods may reprogram the HPA axis toward diminished glucocorticoid production. Pain-associated procedural stress can be likened to early trauma, and was therefore hypothesized to be associated with early diminished glucocorticoid production. Accordingly, the life cycle

8 model of stress provided an appropriate conceptual framework for this cross sectional exploration of the relationship between NICU pain-associated stressor exposure and glucocorticoid values.

Figure 1.1. The Life Cycle Model of Stress.

The life cycle model of stress explains the impact of chronic or repeated stressor exposures at different life stages. Stress in the postnatal period can be associated with elevated or depressed levels of glucocorticoid production, depending on the exposure. For stress related to early trauma, the sensitive period of postnatal brain development that is relative to this study is circled, along with the proposed HPA axis outcome (Lupien, et al., 2009).

Assumptions

Assumptions underlying this study have included:

1) Skin breaking procedures that are characterized as painful in the clinical setting by infant caretakers, or described as painful by adults and older children who

9 experience similar procedures, are inherently painful and considered to be painful in infants.

2) Assisted ventilation, which involves painful procedures such as suctioning and intubation, is stressful in infants, consistent with evidence in older populations.

3) Premature infants respond to painful procedures and stressful conditions with an adrenocortical stress response.

Significance

This study examines the relationship between cortisol and 17-OHP values at 37 postmenstrual weeks of age and pain-associated stressor exposure from two major sources in premature infants during hospitalization in the NICU. This study is the first to describe both cortisol and 17-OHP values in relation to recurrent pain-associated stressor exposure and hours of assisted ventilation at 37 postmenstrual weeks of age.

The information gained by this research also adds to our ability to objectively describe the experience of stress in premature infants who receive care in the NICU. Few studies have quantified recurrent pain-associated procedural exposure in premature infants during NICU hospitalization, although it is commonly described as high. In the course of this research, using counts of skin-breaking procedures, a more accurate representation of pain-associated stressor exposure has emerged. This information can be used to further develop methods for pain and stress measurement in premature infants, or to develop exposure benchmarks for neonatal intensive care.

10

Finally, this study extends what is known about factors influencing the adrenocorticoid phenotype of premature infants at 37 postmenstrual weeks of age. Few studies have described the adrenocorticoid phenotype in premature infants in relation to pain- associated stressors. The information gained by this research can inform future work examining the impact of stress on premature infants and help us to understand the trajectory of stress adaptation in this vulnerable population.

11

CHAPTER 2: Review of the Literature

This chapter presents a review of the literature relevant to this study. This review will begin with a summary of research exploring the main variables pertinent to the study, and move to a review of the literature in which the main variables are examined in relationships. The first sections of this chapter present an overview of neurologic development and introduce cortisol and 17-OHP as part of the stress response system in infants. This overview is followed by an examination of stress and pain in infancy, with attention to literature describing the measurement of these variables. The next section presents a brief overview of research examining the relationship between stressful early life exposures and neurodevelopmental outcomes. The final section of this chapter will present what is known of the relationship between pain or pain-associated stress and the adrenocorticoid phenotype of premature infants who have been cared for in the NICU.

Overview of Neurologic Development and the Infant HPA System

The nervous system is divided into peripheral and central components. The central components comprise the brain and spinal cord, while the peripheral components consist of cranial and spinal nerves, and a network of somatic, visceral and motor neuronal structures that connect the integrating central components to the rest of the body.

The peripheral nervous system can be further organized into the somatic nervous system and the autonomic motor system. The somatic nervous system primarily serves the frame of the body, conducting impulses from skeletal muscle to central components, while the autonomic nervous system serves the involuntary visceral muscles and glands. It is the somatic nervous system that is charged with the perception of environmental stimuli,

12 while the autonomic nervous system, and its sympathetic and parasympathetic divisions, that are responsible for regulating homeostasis. In general, the sympathetic division of the autonomic nervous system regulates homeostasis by responding to stress, while the parasympathetic division regulates homeostasis through the promotion of growth, energy conservation, and system restoration.

The embryology of pain and nociception. For pain to be perceived, nociceptive pathways between peripheral receptors, the spinal cord, the thalamus and the cerebral cortex must be intact. Peripheral sensory receptors have been identified in the human fetus within the first two months of gestation, and have spread to all cutaneous surfaces by the 20th week of gestation (Fernandez & Watterberg, 2009; Lecanuet & Schaal,

1996). By the eighth week of gestation, connections from the periphery to the spinal cord have begun to develop (Glover & Fisk, 1999). These synaptic interconnections are completed between 20 and 24 weeks gestation (Anand & Hickey, 1987).

The cerebral cortex starts to form at ten weeks of gestation and develops as a layered structure (Glover & Fisk, 1999). One layer - the cortical plate - will eventually form the deep structures of the mature cortex. By 28 weeks of gestation, thalamic fibers penetrate this cortical plate. Synaptogenesis of the cortical plate suggests a fully integrated cortical pathway (Kostovic & Judas, 2010; Lagercrantz & Changeux, 2009). The development of thalamocortical connectivity is a major neurodevelopmental event in late gestation, allowing for sensory experiences, such as pain (Vanhatalo & van

Nieuwenhuizen, 2000). Studies have confirmed that pain-associated stressors activate cortical responses in premature infants as young as 25 postmenstrual weeks of age

13

(Lagercrantz & Changeux, 2009), and it is generally acknowledged that thalamocortical pathways allowing for pain perception are intact by 28 weeks of gestation (Anand &

Hickey, 1987; Kostovic & Judas, 2010; S. Lee, Ralston, Drey, Partridge, & Rosen, 2005).

Cortical development, however, continues long after birth (Glover & Fisk, 1999).

Myelination of neuronal tissue is similarly not complete at the time of full-term birth, and continues through the first year of life (Deoni et al., 2011). Prior to the 1980s, it was a common assumption that infants were unable to perceive pain due to the immaturity of their nervous system and deficiency of myelinated nerve fibers (Unruh, 1992).

Researchers have since identified that nociceptive impulses can travel through unmyelinated peripheral nerve fibers, and have suggested that lack of myelination, in terms of capacity for pain perception, is offset by short inter-neuron distances in infants, as well as a relative density of cutaneous nerve endings (Anand & Carr, 1989; Lecanuet

& Schaal, 1996).

The stress response system. Autonomic responses to environmental or internal stressors that are identified by peripheral receptors are coordinated through the central nervous system. When activated, the sympathetic division of the autonomic nervous system can affect changes in heart rate and bronchial tone to support organism defense, as well as adrenergic activities that stimulate metabolism.

There are two interrelated sympathetic systems of stress response: the sympathetic- adrenomedullary (SAM) system, and the hypothalamic-pituitary-adrenocortical (HPA) system. The SAM system effects rapid mobilization of catecholamines from the adrenal medulla and stimulates gluconeogenesis in the liver. In effect, the SAM system ensures

14 swift blood supply and energy to the brain and muscles to rapidly fuel defensive responses. In contrast, the HPA system affects the release of glucocorticoids, including cortisol from the adrenal cortex, which promotes bodily adaptations through changes in gene expression (Greenspan & Gardner, 2004; Gunnar & Quevedo, 2007).

A cascade of biochemical events precedes the release of the cortisol glucocorticoid from the adrenal cortex after HPA axis activation. In response to stress, integrated central nervous system activity cause corticotrophin-releasing hormone (CRH) and arginine vasopressin (AVP) to be released from the paraventricular nuclei of the hypothalamus.

These hormones are carried to the anterior pituitary gland, where they effect adrenocorticotropic hormone (ACTH) release (Greenspan & Gardner, 2004; Stansbury &

Gunnar, 1994).

ACTH interacts with receptors of the adrenal cortex and cholesterol substrate, stimulating biosynthesis of cortisol. Early in this cascade, pregnenolone is produced and partly converted to progesterone with the enzymatic influence of 3β-hydroxysteroid dehydrogenase (3β-HSD). These substances are then converted to either 17-OH pregnenolone or 17-OH progesterone (17-OHP) under the influence of the enzyme 17-α – hydroxylase, and then converted to 11-deoxycortisol under the influence of 3β-HSD and

21-hydroxylase. Finally, enzymatic activity of 11β-hydroxylase facilitates the conversion of 11-deoxycortisol to cortisol. This end product is released into the circulation where it is carried to target organs through the bloodstream. A negative feedback loop to the hypothalamus and pituitary supports the regulation of cortisol production (Greenspan &

Gardner, 2004; Stansbury & Gunnar, 1994). The hippocampus is a center of

15 corticosteroid receptors and serves as a major regulator of HPA axis sensitivity to negative feedback (Jacobson & Sapolsky, 1991). This hippocampal region has additionally received attention for its particular vulnerability to disruption following stress (Carrion, Weems, & Reiss, 2007; McEwen, 2007).

Cortisol is the primary hormonal glucocorticoid product of HPA axis activation. Once circulating, most of this glucocorticoid binds to cortisol binding globulin (CBG) to form cortisone. It is the unbound circulating cortisol that is biologically active and can enter the cells of target tissues. Upon entering these tissues, cortisol interacts with receptors, crosses into the cell nucleus, and regulates the transcription of genes. This process exerts effects on physiology and behavior for extended periods of time (Gunnar & Quevedo,

2007; Stansbury & Gunnar, 1994).

Cortisol has a role in glucose homeostasis, electrolyte balance, and the facilitation of free water excretion, as well as having potent anti-inflammatory effects and immune- regulatory properties. Cortisol and its precursor hormones affect central nervous system activity, and influence emotion, memory and cognition (Fernandez & Watterberg, 2009;

Stansbury & Gunnar, 1994).

The HPA system is highly responsive to conditions that threaten homeostasis. The cascade of events that occur in response to stressors, and that result in the release of catabolic adrenal glucocorticoids, allow the body to increase levels of energy substrates such as glucose, free fatty acids and amino acids. In addition, these glucocorticoids suppress immunologic responses such as inflammation at a time when mobility is important to the organism. Ongoing exposure to glucocorticoids, however, can present a

16 risk for the organism by leading to adaptations that present as insulin resistance, hypertension, hyperlipidemia, impaired growth, muscle atrophy and immunosuppression.

To protect against chronic excessive exposure to glucocorticoids, the organism has a propensity for adaptation. Cortisol resistance is one such adaptation that has been supported by the literature. In this state, higher basal levels of circulating hormone are indicators of target tissue insensitivity. Hypocortisolism is another such adaptation, whereby the pituitary does not vigorously respond to CRH, and lower basal levels of circulating hormone are consequently measured. These changes are advantageous if adaptations are toward a “defensive phenotype,” that is, modified for adversity (Phillips

& Matthews, 2011). While such adaptations to stress have been suggested to have evolutionary value, for humans in our modern world, there may be a mismatch between the adrenocorticoid phenotype that evolves of significant early life stress and the subsequent less demanding environment in which the child will develop and remain in later life (Glover, O'Connor, & O'Donnell, 2010).

Three major pathways can mediate an adrenocorticoid stress response and release of cortisol: (1) visceral and sensory stimulation, such as that produced by pain or temperature changes, (2) direct activation of the pituitary and hypothalamus by chemical stimulation, and (3) transmission of psychological stimuli from the cerebral cortex via the limbic circuits (Stansbury & Gunnar, 1994). For infants, in whom the study of psychological stimuli is precluded by early developmental stage, adrenocorticoid stress responses are primarily examined after activations by the first two of these pathways.

17

Development of the HPA axis. The development of the hypothalamic-pituitary axis is a complex process which depends upon fetal hypothalamic and pituitary function, and is influenced by placental hormones and other elements.

The fetal adrenal cortex develops as two structural zones –a large fetal zone that produces dehydroepiandrosterone sulfate (DHEAS), and a smaller definitive zone that produces mineralocorticoids. A functional third zone produces glucocorticoids. Enzymatic activity of 11β-hydroxysteroid dehydrogenase Type II (11β-HSD Type II) in the fetal zone of the developing adrenal cortex has a role in cortisol regulation. This enzyme causes conversion of cortisol to cortisone in the developing fetus, resulting in relatively high fetal levels of this biologically inactive compound (Bolt, van Weissenbruch,

Lafeber, & Delemarre-van de Waal, 2002).

Different from adult cortisol production processes, the fetal adrenal cortex has low expression of 3β-hydroxysteroid dehydrogenase, preventing significant conversion of pregnenolone into progesterone. The placenta is thought to be largely responsible for this conversion during gestation, until the expression of this enzyme increases after birth.

The placenta, however, lacks expression of 17-α-hydroxylase, so placental progesterone is transported to the fetal adrenal cortex for hydroxylation into 17-OH pregnenolone and

17-OHP to take place (Bolt, van Weissenbruch, Lafeber, et al., 2002). During fetal development, the placenta also synthesizes several other hormones, including CRH and

ACTH, which are important regulators of fetal cortisol.

Regulation of cortisol concentrations in the fetus is affected by placental estrogens which, in addition to the activity of 11β-HSD Type II in the fetal zone, facilitate the conversion

18 of active cortisol into biologically inactive compound. Reduced cortisol concentration in the fetus impacts the negative feedback loop to the fetal hypothalamus and pituitary.

Inhibition of this loop causes an increase in fetal CRH and ACTH secretion, which results in extensive growth of the adrenal cortex during the latter part of gestation (Bolt, van Weissenbruch, Lafeber, et al., 2002; Midgley et al., 2001).

The adrenal gland displays structural and functional changes during the transition from intrauterine to extrauterine life when the feto-placental unit ceases to exist. Structurally, the fetal zone of the adrenal cortex declines rapidly in size after term birth (Midgley, et al., 2001). Functional development continues after birth as the newborn adapts to extrauterine life. This structural and functional development varies between premature infants and full-term infants. The fetal zone, for example, persists in premature infants, with continued function believed to be related to maturity from conception, rather than days from birth (Midgley et al., 1998).

Characterization of adrenocorticoid phenotype in term and premature infants.

Studies of homeostasis and stress response have been facilitated by biochemical assays of the hormones or metabolites of the HPA system. These assays have been widely available for many years (Esteban & Yergey, 1990; Gaskell & Sieckmann, 1986).

Studies of adrenocorticoid function in term and premature infants accordingly date back several decades.

Beginning in the late 1980’s, Gunnar and colleagues conducted numerous studies examining the strength and stability of adrenocortical response in term newborns

(Gunnar, 1989, 1992; Gunnar, Brodersen, Krueger, & Rigatuso, 1996; Gunnar, Connors,

19

& Isensee, 1989; Gunnar, Connors, Isensee, & Wall, 1988; Gunnar, Fisch, Korsvik, &

Donhowe, 1981; Gunnar, Fisch, & Malone, 1984; Gunnar, Hertsgaard, Larson, &

Rigatuso, 1991; Gunnar, Isensee, & Fust, 1987; Gunnar, Larson, Hertsgaard, Harris, &

Brodersen, 1992; Gunnar, Malone, Vance, & Fisch, 1985; Hertsgaard, Gunnar, Larson,

Brodersen, & Lehman, 1992; Larson, Gunnar, & Hertsgaard, 1991; Larson, White,

Cochran, Donzella, & Gunnar, 1998; Malone, Gunnar, & Fisch, 1985; Stang, Gunnar,

Snellman, Condon, & Kestenbaum, 1988). From the work of Gunnar and colleagues, it was learned that the adrenocortical system in newborns was responsive to stimulation, and that infant behavioral stability was largely independent of stability in adrenocortical activity. In addition, both basal cortisol levels and cortisol response levels were found to decline in the first year of life in healthy infants who were born at term.

Wittekind and colleagues conducted one of the earliest studies examining adrenal steroid levels longitudinally in premature infants (Wittekind, Arnold, Leslie, Luttrell, & Jones,

1993). Using a sample of 25 infants with a mean gestational age of 28 weeks, researchers examined the influence of clinical state and glucocorticoid administration on plasma

ACTH and cortisol levels. Random morning serum values were drawn at 1, 2, 4 and 8 weeks of age in these infants. Mean cortisol values significantly decreased from week one to week eight, without correlation between plasma cortisol and ACTH values.

Cortisol values were highly variable and at no time were significantly different between sick and well infants. Postnatal glucocorticoid exposure was significantly correlated with lower ACTH values at 8 weeks of age, while exposure to antenatal steroids did not alter

ACTH or cortisol values.

20

Other research has also identified that cortisol levels overlap between infants who are experiencing significant illness and those who are not. In 1989, Lee and colleagues compared the cortisol and 17-OHP levels of 9 sick premature infants (mean gestational age 32.7 weeks, requiring mechanical ventilation) to 13 well premature infants (mean gestational age 33.5 weeks, not requiring mechanical ventilation) (M. Lee, Rajagopalan,

Berg, & Moshang, 1989). Cortisol levels were low in both groups. Sick premature infants were found to have higher mean 17-OHP values than well premature infants, although there were wide ranges and many overlapping values.

Several explanations for relative cortisol insufficiency in premature infants have been proposed. There may be a rate-limiting step in the synthesis of cortisol that is exacerbated under chronic stress (C. Korte et al., 1996). Alternatively, an immaturity, deficit or deactivation of adrenal enzyme activity in stressed premature infants may result in labile early cortisol levels despite high levels of stress and elevated cortisol precursor levels (Arnold et al., 1997; Huysman, Hokken-Koelega, De Ridder, & Sauer, 2000;

Nykanen, Heinonen, Riepe, Sippell, & Voutilainen, 2010; Wallace et al., 1987). Low levels of circulating proteins in sick premature infants is another possible etiology for blunted cortisol activity in premature infants (Langer, Modi, & Agus, 2006).

Other research teams have reported labile cortisol and 17-OHP values in premature infants that are widely ranging and independent of degree of illness (Kaye et al., 2006;

Lacey et al., 2004; J. Lee et al., 2008a; Martins & Procianoy, 2007; Nordenstrom,

Wedell, Hagenfeldt, Marcus, & Larsson, 2001). Using SNAP scores and other measures of acuity, numerous studies report values of cortisol in premature infants that are

21 disproportionately low given their high burden of disease (Fernandez, Montman, &

Watterberg, 2008, 2010; Hanna et al., 1997; Heckmann et al., 2005; Merce Fernandez-

Balsells et al., 2010; Wittekind, et al., 1993). Cortisol response to surgical stress has additionally been shown to be sluggish in premature infants (Fernandez, et al., 2010).

After considering the labile cortisol values found in premature infants, and higher variability in precursor values, some researchers have proposed the examination of precursor, or precursor to cortisol ratios in research (Franck & Miaskowski, 1997;

Watterberg, Gerdes, & Cook, 2001).

Chronic intrauterine stress appears to have a directional effect on 17-OHP levels. In

2008, Ersch and colleagues examined 17-OHP values on the first day of age in 90 premature (25-33 weeks) infants and identified endocrinologic alterations in infants who had been exposed to chronic - but not acute - intrauterine stress (Ersch, Beinder,

Stallmach, Bucher, & Torresani, 2008). For infants born to mothers with chronic stress

(evidenced by preeclampsia), 17-OHP values measured in first 3 hours of age were significantly higher than in infants born to mothers with acute stress (evidenced by chorioamnionitis). In this study, antenatal steroid exposure was not found to impact 17-

OHP values. The presence of intrauterine growth restriction (IUGR) exacerbated the elevation of 17-OHP values under the condition of chronic stress.

Bolt and colleagues proposed that the maturation of the adrenal cortex is more closely linked to age from conception than to age at birth (Bolt et al., 2002a). In research testing the hypothesis that adrenal function is programmed from the time of conception rather than from the time from birth, Wallace and colleagues conducted a series of studies

22 examining the values of 17-OHP and steroid metabolites in premature and term infants in the first two months of age (Wallace, et al., 1987). Serum 17-OHP levels were first examined in 285 infants (142 premature and 143 term) between days 6 and 10 of age.

Researchers concluded that serum 17-OHP levels were inversely related to gestational age. Next, researchers examined 17-OHP levels at eight points in the first month of age in a group of eight infants born between 33 and 35 weeks of gestation, and compared them to levels from a group of infants born between 26 and 32 weeks of gestation.

Researchers identified differences in the patterns of 17-OHP after birth between these groups. The 17-OHP levels in less mature infants fell in the first week of age, but then increased progressively to nearly birth levels by day 25 of age. In the more mature infants, an initial drop in 17-OHP levels was noted; however a secondary rise was not evident.

Subsequent studies describing adrenocorticoid values for premature infants have generally examined these values during the first week or two of age, or report values in terms of chronological age (CA) since birth (J. Lee et al., 2008b; Ng et al., 2002;

Nykanen, et al., 2010; Sari et al., 2012). Despite the evidence that postmenstrual age is important, only a small number of studies have examined adrenocorticoid values in premature infants at a common postmenstrual age, beyond the first two weeks after birth.

Most studies have also used cross-sectional designs, although some exceptions are noted.

In 1995, Al Saedi and colleagues conducted a longitudinal study in which weekly adrenocorticoid values were measured, up until 37 postmenstrual weeks of age, in infants who were under 31 weeks of gestation at birth. A total of 39 infants without acute illness

23 were sampled, yielding the construction of reference ranges for cortisol and 17-OHP for premature infants from 30 to 37 postmenstrual weeks of age. Arnold and colleagues also examined adrenocorticoid values longitudinally and developed their data relative to postmenstrual age (Arnold, et al., 1997). In Arnold’s study, 26 premature infants

(median gestational age 28 weeks) were followed to 16 weeks of age. Infants were characterized as unwell (requiring mechanical ventilation, FiO2 >25%, or intravenous fluids) or well. A total of 95 plasma cortisol and 78 plasma 17-OHP samples were collected, with progressive loss of weekly sample size throughout the investigation.

Researchers concluded that plasma cortisol and 17-OHP levels were influenced by both postnatal and postmenstrual age. Cortisol levels were not influenced by illness at any point; however 17-OHP levels were significantly higher in unwell infants between five and eight weeks of age.

In 1999, Linder and colleagues measured 17-OHP values in a sample of 66 premature infants at eight points between birth and 90 days of age, and compared the results to birth samples of 100 term newborns (Linder et al., 1999). Researchers identified that by 90 days of age, for infants born at over 29 weeks of gestation, the average 17-OHP concentrations fell to those of the term infants. For infants born at under 29 weeks of gestation, the concentration at 90 days of age was almost double that of full term infants.

Only infants who were born at more than 33 weeks of gestation achieved term levels by

60 days of age. Researchers concluded that post-conceptual age was the most important factor determining 17-OHP concentration.

24

In 2002, Weller and colleagues conducted a longitudinal study of 87 larger premature infants (1500-2000 grams) who were randomized to receive kangaroo care or conventional care after birth (Weller et al., 2002). Thyroid hormone and 17-OHP values were measured at birth, two weeks of age, and at 41 postmenstrual weeks of age. For all infants, the final measurement was obtained after discharge from the hospital. Post- discharge values were highly variable but approached normal full-term values at 41 postmenstrual weeks of age, independent of whether kangaroo care was applied.

Differences in adrenocorticoid profiles between small and average weight prematures has been explored. In 1999, Heckmann and colleagues sampled 37 appropriate-for- gestational-age (AGA) and 8 small-for-gestational-age (SGA) infants born at under 30 weeks of gestation to identify resting serum cortisol values (Heckmann, Wudy, Haack, &

Pohlandt, 1999). Serum cortisol values were measured once during the first two weeks of age for each infant. SGA infants were found to have significantly higher cortisol values compared to AGA infants.

The evidence for differences in adrenocorticoid values between premature male and female infants is mixed. While a number of researchers do not identify differences in adrenocorticoid values between male and female infants (Grunau, et al., 2007; Grunau et al., 2005; Heckmann, Wudy, Haack, & Pohlandt, 2000; Scott & Watterberg, 1995), some report otherwise. In a sample of 38 preterm and 319 full term infants, Ballerini and colleagues found higher 17-OHP values in boys compared to girls during the first year of age, but no differences for cortisol (Ballerini et al., 2010). More recently, Grunau and colleagues identified that cortisol values were relatively depressed at four months

25 corrected conceptual age (CCA) in formerly premature male infants compared to their female counterparts (Grunau, et al., 2010).

Adrenocorticoid levels appear to be consistently impacted by SGA and IUGR in premature infants, although the direction of alteration is variable. Disproportionately small size has been associated with both higher (Heckmann, et al., 1999) and lower

(Doerr, Versmold, Bidlingmaier, & Sippell, 1989) cortisol values in premature infants, as well as various alterations in other HPA axis hormones (Doerr, et al., 1989; Ersch, et al.,

2008; Heckmann, et al., 1999).

Studies examining the impact of antenatal and postnatal glucocorticoids on the neonatal

HPA axis are well represented in the literature. There is consensus that antenatal glucocorticoid exposure has only a minimal effect on infant glucocorticoid levels after two weeks of age (Bolt et al., 2002b; Ersch, et al., 2008; Huysman, et al., 2000; Midgley, et al., 1998; Ng et al., 1999; Tegethoff, Pryce, & Meinlschmidt, 2009), and that they do not have prolonged impact on infant adrenocorticoid activity (Ng, et al., 2002; Ng, et al.,

1999). The evidence is quite different for postnatally administered glucocorticoids.

Numerous studies have found that postnatal glucocorticoid exposure significantly impacts infant adrenocorticoid activity. Postnatal glucocorticoid exposure in premature infants is associated with lower basal cortisol levels during hospitalization (Bettendorf et al., 1998; Cole et al., 1999; Kari, Heinonen, Ikonen, Koivisto, & Raivio, 1993; Kari,

Raivio, Stenman, & Voutilainen, 1996; Ng et al., 1997; Wittekind, et al., 1993). In addition, administration of postnatal glucocorticoids appears to be associated with

26 ongoing HPA axis suppression in long-term follow-up studies of premature infants

(Glover, Miles, Matta, Modi, & Stevenson, 2005; Grunau, et al., 2007).

There is also consensus in the literature that, in contrast to the circadian pattern reported in older children and adults, a free running rhythm is the hallmark of cortisol secretion in both premature and term neonates. Circadian rhythms in the first three months of age have been found to be absent using both serum cortisol measures (Price, Close, &

Fielding, 1983; Vermes, Dohanics, Toth, & Pongracz, 1980) and salivary cortisol measures (Price, et al., 1983). In 1993, as part of a larger investigation, Economou and colleagues examined patterns of cortisol in sick term and sick premature infants, and compared them to well term and well premature infants (Economou, Andronikou, Challa,

Cholevas, & Lapatsanis, 1993). Sixty infants were sampled at six-hour intervals on days of age 1, 3, 5, 10, 15 and 30. Researchers were unable to identify increases or decreases in cortisol within a 24 hour cycle in any of the groups of sick or well infants.

Even in the absence of clear circadian patterns in young infants, there is a pulsatility feature of hormone secretion, whereby cortisol is secreted in 15 to 30 shorts bursts throughout the day (Metzger, Wright, Veldhuis, Rogol, & Kerrigan, 1993). Low birthweight infants, however, show little variability in their plasma cortisol levels over time. As such, random serum cortisol measurements have been proposed to be adequate in reflecting the adrenal status of these infants (Jett et al., 1997).

Recently, with widespread use of 17-OHP in newborn state screening programs for congenital adrenal hyperplasia, there has been increased interest in explaining the variability of 17-OHP values in premature infants. Using a sample of 762 premature

27 infants, Ryckman and colleagues found that respiratory distress syndrome in premature infants was associated with higher 17-OHP, lower gestational age, and lower birthweight

(Ryckman et al., 2012). Authors proposed incorporating birthweight and gestational age data into the interpretation of 17-OHP values in infants to decrease the false positive rates of testing.

Summary of neurologic development and the infant HPA system. The HPA axis activates in response to stress, affecting the release of cortisol. Cortisol reaches target tissues and stimulates hypertension, immunosuppression, glucose release, and other survival responses. It would be difficult to thrive under these conditions if they were persistent over time. The efficacy with which the chronically stressed organism promotes negative feedback is considered critical for survival.

The literature examining adrenocorticoid values in infants is broad, and has generally evolved from characterization of the basal state and the stress response, to the examination of factors that alter these conditions. The distribution of cortisol values in prematurely born infants is marked by low variability. Cortisol values overlap between premature and term infants, between sick and well infants, and across a range of clinical diagnoses. Adrenocorticoid data for premature infants is typically presented in terms of

CA, and uncommonly according to postmenstrual age. Postmenstrual age, however, has been found to impact the interpretation of these values.

Cortisol levels do not always correlate well with their precursor levels. Cortisol precursors may better represent stress responses in premature infants. Because of its role in screening for congenital adrenal hyperplasia, the precursor 17-OHP has been more

28 commonly examined than other precursors in both premature and term infants. Premature infants generally have higher 17-OHP levels than term infants at similar postmenstrual ages. Infants who have been exposed to chronic in-utero stress also appear to have elevated 17-OHP levels. With widespread use of 17-OHP as a biomarker for congenital adrenal hyperplasia in newborns, there has been increased interest in explaining its variability in premature infants.

Infants who have had restricted growth prior to birth have different adrenocorticoid phenotypes compared to appropriately grown infants. It is less clear whether there are differences in phenotype between male and female infants. Postnatal steroid exposure has consistently been shown to alter the adrenocorticoid phenotype, while the effect of prenatal adrenocorticoid exposure is transient. In contrast to the circadian pattern found in children and adults, a free running rhythm is the hallmark of cortisol secretion in both premature and term infants in the first months after birth.

Stress and Pain in Infancy

Researchers have long been interested in the effect of stress on health and disease (Selye,

1976; Selye & Horava, 1953). Stress has been viewed from both psychological and biological perspectives. From the biological perspective, stress is a term applied to physiologic and endocrinologic responses generated by some external stimuli (Wong,

McIntosh, Menon, & Frank, 2003). Stress has been defined as the nonspecific response of the body to any demand, and a stressor as the agent that produces that demand (Selye,

1976). Stress responses can be produced by numerous agents that make adaptability

29 demands on the organism. These stressors include agents such as light, infection, hypoxia, noise, hemorrhage or pain.

Pain, nociception and HPA activation. Since Gunnar’s work in the 1980s, other researchers have demonstrated that the neonate experiences a physiologic response to presumably painful procedures, which can be described in terms of HPA activation

(Anand & Hickey, 1987, 1992). Numerous studies have described adrenocorticoid values in premature infants following such pain-associated exposures (Boyer, Johnston, Walker,

Filion, & Sherrard, 2004; Cignacco, Denhaerynck, Nelle, Buhrer, & Engberg, 2009;

Hughes et al., 1987). These studies largely demonstrate that the infant mounts a hormonal stress response to skin-breaking procedures, and that while analgesia alters the behavioral responses to these procedures, the effect of analgesia is limited on the adrenocortical stress response (Boyer, et al., 2004; Franck, Ridout, Howard, Peters, &

Honour, 2011; Grunau, et al., 2005; Stang, et al., 1988).

Pain-associated procedures can be considered on a continuum of stressors that have varying impact on HPA activation (Grunau, et al., 2005). In 2009, Gunnar and colleagues conducted a review of the literature to identify stressor paradigms used to produce HPA activation in pediatric research, as measured by mean elevations in salivary cortisol in (Gunnar, Talge, et al., 2009). While handling and novel stimulation appeared to affect hormonal stress responses in certain groups of infants, minor skin-breaking procedures, such as blood draws and inoculations, were identified to most consistently provoke significant elevations in mean cortisol in infants. Of the 18 relevant studies

30 identified in this review, 17 studies (94%) reported elevations in mean cortisol values with this stressor.

There is a differentiation between pain and nociception in the literature. Pain is typically defined with attention to the shared nature of this phenomenon as both sensory and emotional. According to the International Association for the Study of Pain (IASP), pain is “an unpleasant sensory and emotional experience associated with actual or potential tissue damage or described in terms of such damage” (Loeser & Treede, 2008). The

AAP Neonatal Pain Control Group has addressed the inherent weakness of emotional or self-reported components as necessary conditions for pain, and has proposed that “the inability to communicate verbally or nonverbally does not negate the possibility that an individual is experiencing pain…” (Kanwaljeet et al., 2006).

Because mature thalamocortical connection are not present until mid to late gestation, there is controversy as to whether the very premature infant can experience pain until then. Pain requires both nociception and interpretation (Lowery et al., 2007).

Nociception is a term used to describe the activation of nervous tissue by tissue damage, or the metabolic or neurobiological effects of a noxious stimulus, independent of any judgment of higher consciousness, memory, emotional effects or suffering (Wong, et al.,

2003). Like the human response to pain, nociception is associated with an increase in catecholamines, cortisol, and other stress hormones.

Procedure-associated stress responses are not necessarily the result of pain (Grunau,

Holsti, & Peters, 2006). It has been suggested that the lack of differentiation between pain and nociception has obscured evidence that nociception – distinct from pain – may

31 be biologically important in the neonate (Anand & Hickey, 1987). Several researchers suggest that physiologic responses to pain-associated stressors in premature infants should be described in terms of nociceptive activity, rather than in terms of pain, to avoid implicating an unexamined psychological experience (Anand & Hickey, 1987; Simons &

Tibboel, 2006).

Procedural pain and pain-associated stressors. In the absence of self-report, painful

NICU stressors have been established through expert provider rating of procedural painfulness, and by using instruments that develop proxy pain scores from a composite of infant response measurements following various stimuli. From these research efforts, a set of pain-associated stressors has been identified.

In 2009, Cignacco and colleagues, as part of a larger study on the frequency of procedural pain in NICU, surveyed 431 NICU caregivers to identify perceived procedural painfulness and to establish a pain intensity rating for many of the routine procedures conducted in the NICU (Cignacco, Hamers, et al., 2009). Using median pain intensity ratings for 27 acknowledged procedures, Cignacco identified that 70% were rated as painful, and 44% were judged as very painful by providers. Among the very painful procedures identified by this study were thoracic drain insertion and removal, tracheal intubation, lumbar puncture, bladder puncture, peripheral intravenous catheter insertion, intramuscular injection, endotracheal suctioning, heel lance, and venipuncture.

In 1999, also as part of a larger pain study, Porter and colleagues attempted to define procedural painfulness by examining the relationship between clinician-perceived invasiveness/intensity and the magnitude of an infant responses to these procedures

32

(Porter, Wolf, & Miller, 1999). Significant correlation between clinician-perceived invasiveness/intensity and magnitude of infant response was identified for arterial puncture, venipuncture, heel lance, circumcision and eye examination. Researchers found that there were considerable overlaps between perceived invasiveness gradations and concluded that there may not be an assessment threshold to clearly mark the presence of pain in infants. In addition, inadequate distinction between the concepts of

“invasiveness” and “painfulness” may have compromised the validity of study findings.

Other studies have supported the measure of caretaker perceived painfulness. In 2003,

Dodds surveyed 21 NICU nurses and found strong agreement for the painfulness of infant thoracic drain insertion, circumcision, central venous catheter insertion, arterial catheter insertion, lumbar puncture, venipuncture, intramuscular injection, heel lance, tracheal intubation, and airway suction (Dodds, 2003). In 2003, Rohrmeister and colleagues conducted a survey of pain practices in Austrian NICUs during which commonly performed NICU procedures were rated for perceived painfulness by respondents (Rohrmeister et al., 2003). Heel lance, intramuscular injection, venipuncture, arterial and central line placement, and thoracic drain insertion were judged by over 90% of respondents to be painful.

In the absence of an empirical way of assigning a pain intensity weight to painful exposures, Grunau operationalized NICU procedural pain exposure as the sum of a composite of skin-breaking procedures (Grunau, et al., 2004; Grunau et al., 2009).

Furthermore, she suggested that this measure more likely reflected cumulative pain- related stress, than pain per se, and probably underestimateed true cumulative pain

33 exposure (Grunau, et al., 2009). The use of counting procedures as a marker for pain exposure, or as measurement of pain-related stress, has been used by other researchers as well (Holsti, Weinberg, Whitfield, & Grunau, 2007; Porter, et al., 1999; Stevens et al.,

2003).

Difficulty in obtaining an accurate count of skin-breaking procedures is also described.

In 2008, Carabjal and colleagues used direct observation techniques to examine exposure to painful procedures in NICU patients. Researchers identified that over 35% of procedures performed were supplemental attempts at primary procedures, with some painful procedures needing as many as 10 to 15 attempts for completion.

Pain and stress of assisted ventilation. Mechanical ventilation is empirically recognized as a pain-associated stressor in the NICU (Anand et al., 1999; Aranda, et al.,

2005). In their review of pain management for neonates receiving assisted ventilation,

Hall, Boyle and Young noted that “assisted ventilation in neonates is presumed to be associated with chronic repetitive pain, which in turn is associated with adverse long- term sequelae” (Hall, Boyle, & Young, 2007).

Sick, premature infants often require assisted ventilation for prolonged periods of time.

Several studies in older children and adults have supported this experience as stressful

(Granja et al., 2005; Grossbach, Chlan, & Tracy, 2011; Novaes et al., 1999; Tracy &

Chlan, 2011), but few have examined assisted ventilation as a pain-associated stressor in premature infants.

It has been suggested that this gap in evidence is related to the lack of validated pain assessment tools for this purpose (Hall, et al., 2007; van Dijk et al., 2009). The difficulty

34 in measuring pain during mechanical ventilation in premature infants was noted by

Anand and colleagues during the NEOPAIN (Neurological Outcomes and Pre-emptive

Analgesia in Neonates) Multicenter Trial (Anand et al., 2004; Boyle, Freer, Wong,

McIntosh, & Anand, 2006). The goal of this trial was to understand whether morphine analgesia decreased the rate of death or significant neurologic events in preterm infants.

As part of this study, factors that could be useful in assessing persistent pain in premature infants who were receiving mechanical ventilation were examined. These factors included general activity, response to handling, respiratory effort, ventilator synchrony, resting posture, quality of spontaneous movements, blood pressure and heart rate, amount of sleep, and response to noise. Problems in differentiating between acute pain and persistent pain were identified, with nearly one-third of nursing assessments incorrectly identifying whether or not an infant subject was randomized to analgesia. Researchers recommended an interim approach of presuming that mechanical ventilation was nevertheless painful and distressing, while persisting in the investigation of analgesic approaches for premature infants receiving this therapy.

In 1992, Quinn and colleagues examined the effect of morphine and pancuronium on noradrenaline stress responses in ventilated premature infants (Quinn et al., 1992). While the study was limited by subject withdrawal, morphine was demonstrated to reduce overall noradrenaline levels compared with pancuronium. Researchers concluded that, compared to paralysis, analgesia may have attenuated the stress of NICU care.

Despite lack of validation of assisted ventilation per se as painful, the ongoing procedures associated with assisted ventilation in the NICU, such as suctioning, tube securement and

35 intubation, have been recognized as painful (Cignacco et al., 2008; Porter, et al., 1997;

Porter, et al., 1999). In addition, to further complicate its examination as an independent variable, measures of exposure to assisted ventilation have been found to be collinear with other infant variables, such as gestational age and birthweight (Grunau, et al., 2007;

Grunau, et al., 2010; Grunau, et al., 2004; Yamada et al., 2007).

In 2007, Yamada and colleagues examined the association between hair cortisol values and illness in infants who had received NICU care to determine if hair cortisol could be a surrogate measure for cumulative stress (Yamada, et al., 2007). Cumulative stress was measured in terms of days of mechanical ventilation. Researchers did not find significant differences in hair cortisol values between premature and term infants, however identified a statistically significant correlation between hair cortisol values and days of assisted ventilation. While this study was limited by recruitment difficulties and a lack of sufficient hair samples in several subjects, results supported the examination of hair cortisol as a potential biomarker for cumulative stress.

Accumulation of pain-associated NICU stress. Advancement in NICU survival generally entails the need for a greater number diagnostic and therapeutic procedures, and greater exposure to assisted ventilation (Lorenz, Paneth, Jetton, den Ouden, & Tyson,

2001). Consequently, the early experience of premature infants is markedly different from that of normal newborns. The reasons for these differences have been summarized as repeated procedural exposures, tissue damage related to surgery, and invasive devises used in mechanical ventilation (Kanwaljeet, et al., 2006).

Several studies have demonstrated that infants cared for in the NICU can undergo

36 hundreds of pain-associated procedures during their initial hospitalization, with the highest exposures in the youngest, smallest and most vulnerable patients (D. P. Barker &

Rutter, 1995; Carbajal et al., 2008; Cignacco, Hamers, et al., 2009).

Distinguishing stress response from pain response was an interest of Holsti and colleagues in 2005 (Holsti, Grunau, Oberlander, Whitfield, & Weinberg, 2005). In addition, these researchers were interested in examining the impact of prior pain on infant responses. In a sample of 54 premature infants at 32 postmenstrual weeks of age, behavioral and physiologic responses to heelstick (pain) and clustered care (stress) were measured and compared. Researchers identified that changes in facial activity and heart rate were the most sensitive indicators of pain in premature infants, and that facial activity became diminished after recurrent prior pain exposure. Both changes in facial activity and heart rate occurred with stress, although these changes were less intense and more protracted.

In 2009, Newnham and colleagues set out to develop a tool to measure cumulative infant stress based upon the presumptions of health care providers (Newnham, Inder, &

Milgrom, 2009). Using survey methods, 137 nurses and doctors used a Likert scale to rate the stressfulness of acute activities and chronic living conditions for premature infants. In addition, respondents rated the stressfulness of acute activities for the provider. Individual responses ranged across the five-point-scale for nearly all items.

Researchers found that nearly all of the acute events were perceived as moderately, very, or extremely stressful for infants of all gestational ages. Skin-breaking procedures were most consistently judged as extremely stressful. Living conditions of hospitalized infants

37 were also judged to be at least a little or moderately stressful by respondents. Many individual items were correlated with clinician’s own stress perception, notably gavage feeding at all gestational ages and breast feeding at older gestational ages. The scale was not validated with any physiologic measures of stress. It was also not clear how researchers chose items for the survey. Conditions known in the literature to be associated with endocrinologic stress responses, such as temperature instability, maternal separation, and light/noise exposure were not specifically included among the items in the survey. Newnham’s work reveals some of the difficulties in measuring cumulative stress in the premature infant population, such as construct validity and high response variability.

Summary of stress and pain in early life. In an examination of stressor paradigms that produce cortisol elevations, skin-breaking procedures have been identified as most effective in causing immediate elevations in mean cortisol in infants. Skin-breaking procedures and procedures associated with assisted ventilation are recognized as painful and stressful for infants cared for in the NICU. Premature infants may undergo hundreds of painful procedures during their initial hospitalization. Proxy assessment methods have been developed to measure painful and stressful conditions and procedures in the NICU, although these measurements have limited validity.

Pain requires both nociception and interpretation. Mature thalamocortical connections are not present until mid to late gestation, which has led some researchers to believe that very premature infants do not experience pain until then. Nociception, however, is associated with a physiologic stress response. While pain is stressful, it is important to

38 recognize that the procedure-associated stress response in premature infants is not necessarily the result of pain. Researchers caution that physiologic responses to skin- breaking procedures in premature infants should therefore not ordinarily be described as painful. Some evidence suggests that there are differences in bodily movements between infants who are experiencing stress and pain, although this has not been validated using hormonal measures.

Perinatal Events, Stress Adaptation and Developmental Sequelae

Prematurity is a major public health problem in the United States, contributing substantially both to infant mortality and to childhood disability. Infants born with low birthweight account for nearly seventy percent of infant deaths (Heron et al., 2010). In addition, premature birth is an important factor in lifelong disability. An estimated 50 to

70 percent of infants weighing under 1500 grams at birth will have long-term disabilities

(Aylward, 2003). The increasing capacity of neonatal intensive care has resulted in a demographic shift from neonatal death to infant survival (Mathews, MacDorman, &

MacDorman, 2008) with a corresponding escalation in the prevalence of prematurity- associated disability (Wilson-Costello, Friedman, Minich, Fanaroff, & Hack, 2005).

Premature birth has long been associated with motor impairment, academic achievement difficulties, psychological and emotional issues, and attention and behavioral problems

(P. Anderson & Doyle, 2003; P. J. Anderson & Doyle, 2008; Botting, Powls, Cooke, &

Marlow, 1997, 1998; Edwards et al., 2011; Hack et al., 2002; Hack et al., 2005; S Saigal et al., 1994; S. Saigal, Rosenbaum, Szatmari, & Campbell, 1991; S. Saigal, Szatmari,

Rosenbaum, Campbell, & King, 1991; Sullivan & Hawes, 2007; Szatmari, Saigal,

39

Rosenbaum, Campbell, & King, 1990; Whitfield, Grunau, & Holsti, 1997; Winchester et al., 2009). Survivors of premature birth have also been identified with more subtle neurobehavioral sequelae, including difficulties with executive functions, self-regulation, temperament, pain sensitivity, arousal and social interaction, as well as disorders of growth (Aarnoudse-Moens, Weisglas-Kuperus, van Goudoever, & Oosterlaan, 2009; P.

Anderson & Doyle, 2003; Grunau, Whitfield, & Petrie, 1994; Jongbloed-Pereboom,

Janssen, Steenbergen, & Nijhuis-van der Sanden, 2012; McGrath et al., 2005; Sullivan,

McGrath, Hawes, & Lester, 2008).

Medical complications during the perinatal period seem to account for only a moderate amount of variance in developmental outcomes for premature infants (Aarnoudse-Moens,

Smidts, Oosterlaan, Duivenvoorden, & Weisglas-Kuperus, 2009; Johnson, Wolke,

Hennessy, & Marlow, 2011; van Baar, Vermaas, Knots, de Kleine, & Soons, 2009). A shift in attention from methods aimed at supporting neonatal survival to those aimed at identifying and modifying the multiple factors that may contribute to altered neurobehavioral outcomes and lifelong disability is accordingly identified in the literature.

Adverse early life events and developmental sequelae. Above the impact of prematurity, exposure to early life stress is being increasingly recognized as a factor that negatively affects developmental outcomes. Fetal exposure to intrauterine stress marked by IUGR and low birthweight, has been associated with a constellation of diseases and disabilities in later life. Independent of prematurity, IUGR and intrauterine stress have been associated with neurocognitive impairment, hypertension, coronary artery disease,

40 glucose intolerance and non-insulin dependent diabetes mellitus (D. J. Barker, 2002;

Henry et al., 2011; Thompson & Regnault, 2011; Zhang, 2005).

Like IUGR, poor postnatal growth has been used as a proxy for an adverse infant environment (Phillips & Jones, 2006). Poor postnatal growth has been associated with impaired neurocognitive development (Ehrenkranz et al., 2006; Latal-Hajnal, von

Siebenthal, Kovari, Bucher, & Largo, 2003; Stephens et al., 2009). Infants with neonatal morbidities tend to regain birth weight later. Researchers have reported an association between slower growth rates in extremely premature infants in the NICU and adverse long-term neurodevelopmental outcomes (Franz et al., 2009).

Pain exposure in the NICU has been proposed as another specific stressor that contributes to neurodevelopmental alterations in this population (Anand, 2000; Grunau, 2002;

Grunau, et al., 2006) although only one human study has been identified in which this hypothesis was tested. In 2009, Grunau and colleagues examined Bayley developmental scores from 116 premature at 8 and/or 18 months of age as part of a larger longitudinal study (Grunau, et al., 2009). Retrospective chart review was used to develop a count of pain-associated NICU procedures and days of assisted ventilation. Researchers identified a relationship between high procedural pain exposure and impaired cognitive and motor function at 8 and 18 months of age. Collinearity between predictor variables made it difficult to examine the independent effect of pain exposure on neurodevelopment, although gestational age at birth was not associated with either cognitive or motor outcomes once there was control for illness severity, dexamethasone use and morphine

41 exposure. Researchers speculated that neonatal pain and ventilation exposure was a more specific indicator for neurodevelopmental outcome than gestational age at birth.

Adrenocorticoid adaptations and developmental sequelae. Early life elevations in allostatic load have been postulated to impact stress adaptation, thereby increasing vulnerability to certain physical diseases and psychosocial disorders (Gunnar, Frenn, et al., 2009; S. M. Korte, et al., 2005; Lupien, et al., 2009; McEwen, 2001; McEwen &

Wingfield, 2003).

The Developmental Origins Theory has been the basis for numerous studies, proposing that early stress provokes adaptive changes in endocrine processes that impact later health

(D. J. Barker, 2007; D. J. Barker, Eriksson, Forsen, & Osmond, 2002). In an extension of this theory, adrenocorticoid alterations associated with infant stress, such as those which results from an adverse perinatal environment, have been proposed as the mechanism underlying many of the chronic diseases and disabilities in low birth weight survivors (Sullivan, Hawes, et al., 2008).

Studies are beginning to test this extension of the Developmental Origins Theory. In

2006, Haley and colleagues examined infant learning in relation to basal and stress cortisol levels in premature and term infants who were three months CCA (Haley, et al.,

2006). Researchers identified that premature infants had lower cortisol levels and smaller cortisol responses than full-term infants, and that even when corrected for prematurity, lower cortisol was associated with poorer memory for accomplishing a mobile task.

In another application, researchers demonstrated that the rate of catch-up growth in IUGR infants was associated with alterations in the HPA axis (Cianfarani, Geremia, Scott, &

42

Germani, 2002). After examining the cortisol profiles of 49 IUGR children, Cianfarani and colleagues identified that poor catch-up growth in IUGR children was associated with higher concentrations of serum cortisol in childhood. They postulated that intrauterine reprogramming of the HPA axis resulted in modification of the adrenocorticoid response to stress, thereby impacting postnatal growth rates of IUGR children.

In 2004, Hofman and Cutfield examined 72 children and identified that those who were born prematurely had reduced insulin sensitivity patterns similar to the patterns seen in

SGA, term-born children (Hofman et al., 2004). Prematurely born infants who were

AGA and SGA did not, however, have significant differences in insulin sensitivity.

Researchers concluded that, like growth restricted term infants, children who are born prematurely have adrenocortical adaptations resulting in reduced insulin sensitivity in childhood, which may be a risk factor for type 2 diabetes mellitus.

Summary of perinatal events, stress adaptation and developmental sequelae.

Premature delivery affects approximately 12% of all births in the United States and is a leading cause of death and disability among children. Premature birth is associated with lifetime cognitive, emotional and physical disabilities. The degree of lifetime disability experienced by premature infants, however, appears to be disproportionate to their experience of medical complications.

A growing body of literature supports the association of early life stress with the development of subsequent disease. HPA axis adaptations associated with early life stress have been implicated as a mediator of many of the relationships between early

43 adversity and adult disease. HPA axis adaptations associated with NICU care are also being proposed as a mechanism underlying many of the diseases and disabilities in survivors of prematurity.

Pain-Associated NICU Stress and the Adrenocorticoid Phenotype

Five studies have been identified that examine the relationship between adrenocorticoid values and cumulative pain-associated procedural exposure in infants who have been cared for in the NICU. These studies, which looked at premature infants during NICU hospitalization and in their early months following discharge, were all conducted under the direction of the Canadian researchers, Grunau and Holsti.

NICU Studies. Two studies examining adrenocorticoid function in premature infants within the NICU are identified. The results from the first of these studies suggests that high cumulative exposure to painful stressors predicts dampened adrenocortical reactivity at 32 postmenstrual weeks of age in infants who are born at the lowest gestational ages.

The other study demonstrates that recent exposure to painful stressors (within 24 hours) is associated with elevated resting adrenocorticoid activity.

In 2005, Grunau and colleagues examined the relationships between cumulative painful exposures, morphine exposure and subsequent physiologic, behavioral and adrenocorticoid responses to pain and stress in premature infants (Grunau, et al., 2005).

Eighty-seven premature infants who had not been exposed to postnatal corticosteroids were studied at 32 post-conceptual weeks of age. Infant responses to heel lance (pain) and to clustered nursing care (stress) were examined in a repeated measures randomized crossover design. Comparisons were made between earlier born (<28 weeks of gestation

44 at birth) and later born (29-32 weeks of gestation at birth) infants. Cumulative painful exposures were defined as the total number of skin breaking procedures since birth, including heel lance, venipuncture, arterial or venous line insertion, lumbar puncture and thoracotomy. Researchers identified that higher cumulative pain was significantly negatively correlated with cortisol reactivity and facial activity to handling stress in the younger gestational age group, but not in the older gestational age group. Researchers postulated that there may be a “cutoff” of stressful exposures that must be reached before seeing dampened cortisol reactivity. Researchers also identified that morphine exposure was not related to cortisol response. After adjusting for illness severity, higher cumulative painful exposures also predicted lower cortisol reactivity to handling stress in the earlier-born premature infants. Regression analysis was limited by multicollinearity: the number of skin-breaking procedures in NICU was highly correlated with gestational age, birthweight, illness severity, and number of days of mechanical ventilation.

In 2007, Holsti and colleagues examined adrenocorticoid values (ACTH and cortisol) at rest, as well as adrenocorticoid values and facial activity in response to clustered nursing care, in a sample of 90 premature infants at 32 post-conceptual weeks of age using a within subjects crossover design (Holsti, et al., 2007). Comparisons were again made between earlier born (<28 weeks of gestation at birth) and later born (29-32 weeks of gestation at birth) infants. Researchers failed to identify differences in resting or stress adrenocorticoid values between earlier-born and later-born infants and concomitantly between those with higher and lower cumulative pain exposure. A higher number of recent skin breaking procedures (within 24 hours) was correlated with elevated resting cortisol values in both groups. Researchers also looked at the relationship between

45 hormones and identified a correlation between stress ACTH and cortisol in the older infants, but no relationship between stress values in the younger gestational age group, or between resting values in either age group. Facial responses to handling were dissociated from hormone values in both groups. Researchers speculated that basal hormone values remain relatively well preserved in the first few weeks of age, and that the lack of change in the relationship between ACTH and cortisol under stress conditions in the younger gestational age group might represent an early neuroprotective adaptation in infants exposed to greater cumulative stress.

NICU follow-up studies. Three studies were identified in which the relationship between stressful NICU exposures and adrenocorticoid activity in the first months of age following NICU discharge in prematurely born infants was examined. While studies were limited by multicollinearity, salivary sampling issues, and participant withdrawal, researchers in two of these studies identified that cumulative exposure to skin-breaking procedures and mechanical ventilation in the NICU were important predictors of adrenocorticoid alterations in the early months following NICU discharge.

In 2004, Grunau and colleagues used a cross sectional design to examine the relationship between the incidence of painful procedures and morphine exposure in NICU and basal cortisol values and cortisol responsivity to visual novelty at 8 months CCA (Grunau, et al., 2004). Cortisol outcomes from earlier born premature infants (<28 weeks of gestation at birth) and later born premature infants (29-32 weeks of gestation at birth) were compared to the cortisol outcomes of full term infants. Salivary cortisol values from 76 infants (54 preterm, 22 term) were measured before the introduction of a novel

46 toy (basal), 20 minutes after introduction of the toy (post 1), and 30-40 minutes after introduction of the toy and following developmental assessment (post 2). Researchers identified that earlier-born premature infants had significantly higher cortisol at all phases than later-born premature infants or full-term controls, and that the pattern of cortisol response differed significantly between earlier-born premature infants (a drop, rather than an increase, in cortisol from post 1 to post 2) compared to later-born premature infants and the controls. Basal cortisol values in premature infants overall were significantly correlated to days on oxygen, gestational age at birth, and the number of skin breaking procedures in NICU. Morphine exposure was not associated with cortisol values in any phase. After controlling for early illness and days on oxygen, a higher number of skin breaking procedures independently predicted basal and post 1 cortisol at 8 months CCA.

Gestational age, days of mechanical ventilation and number of skin breaking procedures were highly correlated, so researchers repeated the analysis using gestational age as the predictor. After controlling for early illness and days on oxygen, gestational age was not related to basal cortisol at 8 months CCA in the premature infants, although it was associated with post 1 cortisol to the same extent as number of skin-breaking procedures.

Researchers concluded that early pain exposure predicted basal cortisol values at 8 months CCA, and that cortisol reactivity at this age was equally well predicted by early pain exposure and gestational age.

In 2007, Grunau examined salivary cortisol values at 3, 6, 8 and 18 months CCA in earlier born premature infants (<28 weeks of gestation at birth) and later born premature infants (29-32 weeks of gestation at birth), and compared cortisol outcomes to those of full term infants, to clarify developmental patterns of basal cortisol following NICU

47 discharge (Grunau, et al., 2007). Using a cross-sectional design with retrospective NICU chart review, salivary cortisol was examined in a sample of 225 infants (145 preterm, 80 term). Researchers identified that at three months CCA, cortisol values were significantly lower in premature infants compared to term infants. At 6 months there were no differences among the groups. At 8 and 18 months, earlier-born premature infants had higher basal cortisol values when compared to later-born premature infants or term infants. At 18 months, basal cortisol values were significantly correlated with gestational age, number of days of mechanical ventilation, and higher number of skin breaking procedures in the NICU. Multicollinearity between the number of skin breaking procedures, days of mechanical ventilation, and gestational age at birth limited analyses, and researchers were unable to examine the independent relationship between these factors and cortisol. In addition, researchers identified that they were unable to isolate pain-related stress from the stress of mechanical ventilation.

In 2010, Grunau and colleagues examined salivary cortisol, facial actions and heart rate with immunization at approximately four months CCA in earlier born premature infants

(<28 weeks of gestation at birth) and later born premature infants (29-32 weeks of gestation at birth), and made comparisons with full term infants, to identify patterns of pain response following NICU discharge. A total of 138 infants (99 preterm, 39 term) had salivary samples collected at three points during a nurse home immunization visit: before vaccinations (basal), 20 minutes after injection (reactivity) and 30 minutes after injection (recovery). With regard to cortisol, there were significant changes for all groups across events. Researchers also identified that premature boys had lower overall cortisol values and lower reactivity cortisol when compared to term boys. Results may

48 have been confounded by post-NICU surgery, which was carried out on 7 boys (6 premature, 1 term) and no girls. No significant differences between boys and girls were identified for number of skin breaking procedures in the NICU, days of mechanical ventilation, or early illness severity. Researchers concluded that premature infants have altered adrenocorticoid activity during immunization compared to term infants, but that these differences were not exaggerated in premature infants who were younger, sicker or exposed to more skin-breaking procedures in the NICU.

Summary of pain-associated NICU stress and adrenocorticoid phenotype. Many studies implicate NICU care as an adverse early life experience with enduring impact, although few have measured this adversity and studied it in relation to adrenocorticoid consequence. Research is beginning to demonstrate a relationship between the incidence of exposures to pain-associated procedures during the NICU care and HPA axis adaptations in premature infants.

Both cortisol resistance and cortisol sensitivity have been found in relation to NICU care.

High exposure to pain-associated procedures appears to be associated with dampened cortisol reactivity at 32 post-conceptual weeks of age in infants who are born at the lowest gestational ages. At four months CCA, premature infants appear to have lower cortisol levels than term controls, and then go on to develop elevated resting cortisol levels between 8 and 18 months of age. Male infants who are born at the lowest gestational ages may have different cortisol patterns than their female counterparts, as well as different patterns than older-born premature male or female infants.

49

Most of the studies examining relationships between pain-associated NICU stressors and adrenocorticoid values have been limited by multicollinearity. The number of stressor exposures in NICU is highly correlated with gestational age, birthweight, illness severity, and number of days of mechanical ventilation, making it difficult to identify the independent effect of each of these factors on the adrenocorticoid phenotype.

Overall Summary

Stress has been associated with alterations in long-term psychological, social and physical functioning and in the development of disease and disability. Cortisol appears to be an important factor in the relationship between stress exposure and subsequent disease and disability. NICU stressor exposure presents as a plausible explanation for HPA axis adaptation and some of the neurodevelopmental disability found in children who were born prematurely.

The study of pain and stress in infancy dates back many decades. The distribution of cortisol and cortisol precursor values at many postmenstrual and postconceptual ages is now partially described. Our understanding has also evolved from an early view that infants were unable to perceive pain, to current acceptance that exposure to painful stimuli in infancy affects short-term behavioral, physiologic and endocrinologic outcomes. In addition, pain-associated procedural exposure in infancy is currently being examined as a risk factor for adverse long-term neurodevelopmental outcomes in children.

There remains much to be learned about the early developmental trajectory of adrenocorticoid adaptations in premature infants, and the influence of early life exposures

50 on this trajectory. Little is known of the evolution of HPA axis adaptations in premature infants in relation to their exposure to recurrent or prolonged pain-associated stressors, such as skin breaking procedures and assisted ventilation. Both cortisol resistance and cortisol sensitivity have been identified in prematurely born infants during their first year of age. It is hypothesized that altered cortisol phenotypes begin to emerge even earlier, and in relation to pain-associated stressor exposure. Two studies have been identified that examined this relationship at 32 post-conceptual weeks of age, and three studies have been identified that examined this relationship at between four and 18 months CCA. The approximate six month period between 32 post-conceptual weeks of age and four months

CCA in this population remains relatively unexplored in this regard.

51

CHAPTER 3: Methods

Overview

A descriptive, cross-sectional design was proposed to examine the association between recurrent exposure to pain-associated procedures, hours of assisted ventilation, and random cortisol values at 37 postmenstrual weeks of age in a sample of infants who were born prematurely and received care from birth in the NICU. Cortisol is a well- understood and commonly measured hormone in the clinical setting, and is relatively inexpensive to analyze. Because there was concern that random cortisol values would not adequately reflect exposure to the multiple stressful stimuli, the exploration of associations between recurrent exposure to pain-associated procedures, hours of assisted ventilation, and random17-OHP values at 37 postmenstrual weeks of age was also conducted. As previously indicated, the examination of 17-OHP as a potential stress biomarker capitalized on an opportunity to use a ordinarily present piece of infant clinical data – 17-OHP levels drawn after birth for congenital adrenal hyperplasia screening – in another way.

Operational Definitions

The following operational definitions clarify terminology that was used in the research:

52

Table 3.1

Operational Definitions

Term Operational Definition

Infant A live born fetus from birth through the first year of age.

Neonate An infant in the first 28 days after birth. For this study, neonates are included when referring to infants in the study sample.

Premature Infant A fetus born at less than 37 completed weeks of gestation. Gestational Age (GA) The age of a fetus, expressed in completed weeks, calculated from the first day of the last normal menstrual period until the day of delivery. Chronological Age (CA) The time elapsed since birth, described in terms of days, weeks, months or years (also known as “postnatal age”).

Postmenstrual Age Gestational Age + Chronological Age, usually described in terms (PMA) of weeks. Corrected Conceptual The age of a premature infant from the expected date of delivery; Age (CCA) it is the chronological age reduced by the number of weeks born before 40 weeks of gestation.

Neonatal Intensive Care An inpatient hospital care unit specializing in complex care of ill Unit (NICU) or premature infants, usually from birth through hospital discharge.

Clustered Care An approach to nursing care where all “hands on” caregiving activities (diapering, feeding, suctioning, repositioning, etc.) are conducted in quick sequence, allowing for an extended period of patient rest before repeating the caregiving activities. Medical Record An account of a patient’s history, care activities and test results in a particular health care setting, including both electronic and physical forms of documentation.

Medical Record Within the medical record, a discrete subset of documentations Encounter related to a specific patient visit or event.

Assisted Ventilation The use of mechanical or other devices to help maintain or supplement spontaneous breathing.

Medical Team A multidisciplinary group of health care provider responsible for a patient’s medical care.

Heel Lance A method for blood sampling in infants whereby a lancet is used to prick the skin of the heel.

Specimen Receiving The department of a hospital where biological samples are Department received from patient care areas, prepared, and distributed to appropriate laboratories for analysis.

53

Setting

The study was conducted at a university-affiliated Level III, 46-bed NICU in Suffolk

County, New York. The NICU has approximately 1000 admissions annually, of which typically one-third are related to prematurity. The NICU is a tertiary care center, employs neonatal nurse practitioners, and provides training for pediatric residents and neonatal fellows. The NICU was renovated in 2011 to enhance provision of family- centered care. This included the establishment of single-patient rooms and elimination of twin co-bedding, which allowed for a measure of control of environmental disturbances to premature infants. For example, infant exposure to noise and light are minimized under this arrangement. In addition, there are no outside windows in the designated

NICU setting that could impose variations in lighting in the patient care areas.

Premature infants in this NICU receive blood sample analysis on at least a weekly basis.

Subjects

Sixty infants were recruited for this study. Infants who had been admitted to the NICU within 24 hours of birth, who were under 35 completed weeks of gestation at birth, and who had received care in the NICU for a minimum of two weeks were eligible for inclusion.

Several exclusionary criteria were identified to remove from this study those situations already established as influencing adrenocorticoid values. These exclusionary criteria included infant diagnosis with a major endocrinologic disorder or a major congenital anomaly, the presence of IUGR, prolonged administration of postnatal systemic glucocorticoids, or any event of a major surgical procedure. Infants were additionally

54 excluded if there was incomplete medical record data, such as resulting from a temporary inter-institutional transfer.

Diagnosis with a major endocrinologic disorder or a major congenital anomaly was included as an exclusionary criterion to ensure control for the influence of intrinsic disease on adrenocorticoid values and is commonly exclusionary in studies examining cortisol activity in infants.

The presence of IUGR was also exclusionary criterion, as IUGR has been identified to influence adrenocorticoid values in infancy. In the absence of direct prenatal measurements of intrauterine growth, the diagnosis of a newborn as SGA has suggested the presence IUGR (Ogundipe, Wolfe, Seed, & Gamsu, 2000). IUGR was considered to be present if an infant’s weight at birth was below the 10th percentile for gestational age based upon the Fenton growth curves (Fenton, 2003).

Prolonged administration of postnatal systemic glucocorticoids has been reported to alter resting cortisol values in premature infants before the 37th postmenstrual week of age and was therefore exclusionary criterion. Prolonged systemic glucocorticoid administration was defined as the use of intravenous or enteric glucocorticoids (but not inhaled glucocorticoids) for a continuous period of greater than 48 hours prior to cortisol measurement during the 37th postmenstrual week of age.

Major surgical procedures often involve numerous discrete procedures such as endotracheal intubation, skin punctures and injections, and suture placements. A count of these individual procedures would have been extremely difficult to obtain given the format in which operating room procedures are documented. Furthermore, aggregating a

55 surgical procedure into a single pain-associated event was not felt to correctly represent such an event. Common surgical procedures such as hernia repair, shunt or tracheostomy placement, and bowel re-anastomosis are commonly deferred until late in a premature infant’s NICU course, and have been associated with short term alterations in hormonal stress responsivity. This potentially affects measured adrenocorticoid values in the 37th postmenstrual week of age. For these reasons, the decision was made to include major surgical procedures as an exclusionary criterion. A major surgical procedure was defined as one in which the infant was transferred to the operating room and received general anesthesia.

Incomplete medical record data precluded collection of pertinent information. In particular, this was true of infants who were treated in another facility for greater than 24 hours at any time prior to cortisol measurement. Such infants were not eligible for recruitment into this study.

Procedures

Subject identification. The investigator met with a NICU charge nurse weekly to review the census and identify newly admitted premature infants. As all infants who were eligible for study inclusion were to be receiving care in the NICU for a minimum of two weeks, this one-week interval supported enough time to identify and obtain consent from all potential subjects without burdening the NICU charge nurse with recurrent review.

Upon identification of a potential subject, a brief review of charted admission demographics was used to confirm the date of birth and gestational age of each infant.

56

When the potential premature infant subject reached the end of the 36th postmenstrual week of age, a brief review of the medical record was again conducted to identify any exclusionary criteria. After this review, for infants who met criteria for study inclusion, the investigator approached a parent and requested consent for infant enrollment into the study. A recruitment log (compliant with preparatory to research HIPAA standards) was maintained to keep track of participant enrollment.

Data collection. During the 37th postmenstrual week of age, following study enrollment, the investigator communicated with the primary medical team and ascertained plans for routine laboratory testing of each infant subject. Arrangements were then made to have the study samples drawn by the nursing staff in conjunction with routine labs.

Data for the derivation of measures of recurrent exposure to pain-associated procedures, recent exposure to pain-associated procedures, and hours of assisted ventilation were obtained through a complete chart audit by the researcher in the week following blood collection for the measurement of cortisol and 17-OHP values at 37 postmenstrual weeks of age. Data was recorded on a data collection sheet (Appendix A) prior to entry into the

IBM SPSS Statistics 19 software program (Release 19.0; August, 2010) which was used for all data analysis procedures.

Blood collection. Identification of the 37th postmenstrual week of age was calculated from the gestational age, as previously described, and recorded on the data collection sheet. The investigator coordinated blood sampling for cortisol and 17-OHP values during this week so that it was aligned with routine blood sampling to minimize patient burden. Blood sampling was coordinated only with routine labs, and not with incident

57 related labs, to improve likelihood that the values obtained were random. Time of day for sample collections was not specified. Circadian rhythms have not been conclusively demonstrated as present in neonates and so time was not expected to modify study results.

Blood collection using heel lance or vessel puncture was done at the beginning of clustered care interventions. Consistent with previous research (Holsti, et al., 2007), it was assured that the infant had at least 30 minutes of undisturbed rest prior to laboratory sampling. The process of performing laboratory testing at the beginning of clustered care interventions and after a 30 minute period of undisturbed rest was reviewed by the investigator at the beginning of the shift with each nurse who was to be drawing blood, and the investigator made episodic observations of the infants during this 30-minute period to assure compliance with the period of undisturbed rest. Single-patient rooms facilitated minimization of noise and light during this rest period.

Blood handling. At the time of routine blood sampling, an additional single aliquot of infant blood totaling two milliliters was collected in additive-free, green-topped micro- containers by the infant’s nurse. Blood samples were labeled by the collector in accordance with policy, and either given to the investigator or forwarded directly by the nurse to Stony Brook Hospital’s specimen receiving department. There, blood samples were centrifuged by laboratory staff to separate serum. Serum was transferred into appropriate transport tubes and expressed shipped by the specimen receiving department to ARUP National Reference Laboratories (www.aruplab.com), at a temperature of 2-

8ºC, where analysis was completed.

58

A separate medical record research encounter was generated for each subject for ordering and reporting the results of blood specimens, as is customary for the separation of the clinical record and the research record. This separate encounter also assured that there was no cost to the patient or to his/her payor for participation in this research. Laboratory costs generated under the research encounter were billed directly to an account funded by the investigators. Laboratory results generated by the research are in this way also made discoverable by the infant’s health care providers.

For problems related to laboratory error or specimen handling issues in the measurement of cortisol and 17-OHP, there was an opportunity to repeat testing if another routine blood sample was indicated for the infant within the 37th postmenstrual week of age.

Arrangements were made with the specimen receiving department and with ARUP to report such problems as soon as possible to the investigator, using email or cellular phone, so that opportunities for repeat specimen collection could be identified.

59

Figure 3.1. Flow Chart of Research Procedures

60

Staff instruction and support. As resources for their role in clinical investigations, bedside nursing staff in the NICU received the support of two NICU nurse educators and the investigator. Nurse educators were available to provide shift-based support for protocol adherence, and were prepared to assist in problem-solving with the staff. NICU staff members are additionally required to be certified in the responsible conduct of research and had consequently received education on human subjects protection and research compliance through the Collaborative Institutional Training Initiative

(www.citiprogram.org).

Consistent blood sampling procedures and comprehensive documentation by NICU staff members was necessary to ascertain the validity and reliability of study results. The investigator provided instruction to staff regarding the study purpose and protocols, and provided both in-person and on-call support for protocol adherence. To address staff questions or concerns when drawing blood samples, the investigator was available by cellular phone when not available to address these issues in person. The research study was also reviewed by the investigator with the NICU nurse educators on both day and night shifts to improve the adequacy of resources available to staff for compliance with study protocols.

As required by nursing administration, a binder of study protocols and a “Nursing

Standards of Care” document (Appendix B), was placed in the NICU in an area dedicated to the keeping of resource manuals. Shortly after IRB approval, small group in-services were conducted for the bedside nurses, in the NICU patient pods, to assure awareness of the study and to direct staff to study protocol references and support materials. In

61 addition, following IRB approval, the investigator used notices on the NICU bulletin boards to ascertain study awareness and remind staff of study resources and reference materials. The investigator personally assured individual nurse knowledge regarding the study protocols and resources for those who were responsible to draw study labs during their shift by meeting with these individuals at the beginning of the nursing shift.

Prior to pursuing IRB approval at Stony Brook Hospital, the study was presented to both the department chair of Pediatrics, the chief of Neonatology, the neonatal attending physician staff, and the appropriate division head of Nursing. Following acceptance by these key personnel, the study was also formally presented to the medical and nurse practitioner staff of the NICU.

The procedures developed in this study ensured fidelity of the implementation as well as assurance in the ethical conduct of the project. This study was conducted in a setting where the investigator has served in the clinical role of a neonatal nurse practitioner. In this role, she had been regularly involved in study recruitment, consent procedures, coordination of labs, and the collection of research data. This experience strengthened the prospect of these processes progressing smoothly for the research. For example, the investigator’s prior acquaintance with staff eased the process of coordinating the timing of research labs with routine clinical labs, and her familiarity with the clinical recording system enhanced the accuracy of data collected from the medical records. Additionally, an existent mutual respect between nursing staff and the investigator augmented subject identification and adherence to the research protocols by the nursing staff. Because of the investigator’s prior role as a nurse practitioner, there was concern that there may be

62 confusion by staff that the labs ordered for this study were for clinical purposes. This distinction was emphasized by the investigator with the individual nurses who were responsible for drawing the labs at the beginning of the nursing shift. Establishment of a separate medical record research encounter assured accuracy in billing and reporting of these labs even where misunderstanding may have existed.

Measures

This study employed the following measures to assess the major variables of interest, which were (a) random cortisol values in the 37th postmenstrual week of age, (b) random

17-OHP values in the 37th postmenstrual week of age, (c) recurrent exposure to pain- associated procedures and (d) hours of assisted ventilation.

Quantitative chemiluminescent immunoassay was used to measure random cortisol in the 37th postmenstrual week of age (Advia Centaur, Siemens Inc., Cortisol test kit, analytical sensitivity 0.2 ug/dl). A random cortisol value was defined as the level of cortisol in serum from a single resting blood draw, conducted by nursing staff, during the

37th postmenstrual week of age. For this study, cortisol values were described as

“random,” rather than as “basal,” to reflect the free running rhythm (versus diurnal variation) that has been shown to be the hallmark of cortisol secretion in neonate

(Economou, et al., 1993; Price, et al., 1983; Vermes, et al., 1980).

Even in the absence of clear circadian patterns in young infants, the pulsatility feature of hormone secretion needed to be considered, whereby cortisol is secreted in 15 to 30 shorts bursts throughout the day (Metzger, et al., 1993). Low birthweight infants, however, have been shown to have little variability in their plasma cortisol levels over

63 time. Therefore, a single random serum cortisol measurement has been proposed to be adequate in reflecting the adrenal status of these infants (Jett, et al., 1997).

Test specimens were not batched, but instead analyzed as they became available, typically on separate days at ARUP Laboratories. This is an appropriate method of laboratory collection and analysis for translational research, in that it emulates standard clinical procedures and standard laboratory analysis methods.

Immunoassay is a method of testing biological liquids for the presence of an analyte (in this case, cortisol) using the analyte’s unique ability to bind to a specific group of molecules (usually an antibody or antigen). In chemiluminescent immunoassay, a chemiluminescent dye is used that allows for the detection and quantification of cortisol binding events. Serum cortisol values were reported in micrograms per deciliter units.

Quantitative radioimmunoassay was used to measure random 17-OHP in the 37th postmenstrual week of age (Wallac Wizard Gamma Counter, Siemens Inc., Coat-A-

Count 17A-OH progesterone test kit, analytical sensitivity 7 ng/dl). A random17-OHP value was defined as the level of 17-OHP in serum from a single resting blood draw, conducted by nursing staff, during the 37th postmenstrual week of age.

Radioimmunoassay is a relatively inexpensive analytic technique in which a known concentrations of an antigen (in this case, 17-OHP) is radiolabeled and mixed with a known amount of antibody for that antigen. After the two substances bind to one another, a sample of serum from a patient with an unknown quantity of antigen is added, displacing the radiolabeled antigen. The radioactivity of the free antigen is then

64 measured, and the amount of antigen in the patient’s serum can be derived. Serum 17-

OHP values were reported in nanograms per deciliter units.

As was the case with cortisol specimens, 17-OHP specimens were not batched, and instead analyzed as they became available. This emulated standard clinical and laboratory analysis procedures, such as is used for generating 17-OHP values from infants for the purpose of newborn state screening.

A simple count of the total number of painful procedures sustained by each subject from birth up until the time of the serum cortisol measurement in the 37th postmenstrual week of age was used to measure recurrent exposure to pain- associated procedures. The number of painful procedures, as reported in the medical record from birth up until the time of serum cortisol measurement in the 37th postmenstrual week of age, was counted by the investigator to obtain this measurement.

Procedures that had been categorized in previous research by Cignacco as “very painful” when carried out in the NICU were used in this count of events (Cignacco, Hamers, et al., 2009). These procedures include: endotracheal intubation, heel lance, arterial puncture, venipuncture (for any purpose), bladder puncture, lumbar puncture, thoracic needle or tube placement, suture placement, and subcutaneous or intramuscular injection.

Porter’s study of procedural painfulness, as well as similar studies examining pain exposure in the NICU, supported that a count of intubations and/or skin-breaking procedures provides a reasonable measure of recurrent pain (Dodds, 2003; Porter, et al.,

1997; Rohrmeister, et al., 2003). Furthermore, Grunau has suggested that a count of

NICU skin-breaking procedures provides a marker of acute pain exposure in this

65 population, and that counting the total number of skin-breaking procedures from birth provides a marker of overall stress exposure (Grunau, et al., 2004).

A simple count of the number of hours, up until the time of serum cortisol measurement in the 37th postmenstrual week of age, in which a subject received assisted ventilation, was used to measure hours of assisted ventilation. The number of hours in which there was any application of assisted ventilation, as reported in the medical record, was counted by the investigator to obtain this measurement. An infant was considered to be receiving assisted ventilation if requiring high-frequency ventilation

(HFV), intermittent mandatory ventilation (IMV), or continuous positive airway pressure

(CPAP), including the use of high flow nasal cannula (> 1 liter/minute flow ).

In the NICU, assisted ventilation implies the need for endotracheal intubation or placement of CPAP/cannula prongs, recurrent device adjustment and securing to the face, regular placement and removal of adhesive oxygen saturation and/or transcutaneous carbon dioxide monitors, frequent repositioning of devices to avoid pressure-related skin breakdown, and repeated endotracheal and/or nasopharyngeal aspiration (Bonner &

Mainous, 2008; Haney & Allingham, 1992). Many of these procedures have been identified by Cignacco as a painful (Cignacco, et al., 2008).

Various measures reflecting exposure to assisted ventilation have been described in infant adrenocorticoid studies. Investigators have categorized infants as having required or not required assisted ventilation, or have operationalized this variable as a count of days from birth with mechanical ventilation or with supplemental oxygen. Multicollinearity of this latter measure with gestational age, days of age, or number of painful procedures has

66 proven to be limitation in several studies (Grunau, et al., 2007; Grunau, et al., 2005;

Grunau, et al., 2010; Grunau, et al., 2004; Yamada, et al., 2007). The correlation between days of mechanical ventilation and procedural pain exposure is plausible - sicker patients, such as those requiring assisted ventilation, typically need more diagnostic assessments and therapeutic interventions. By collecting the number of total hours of assisted ventilation, rather than the number of days, the measurement of this variable is more precise.

In addition to the measures used to assess the major variables in this study, other measurements were collected by the investigator to answer the research question:

A simple count of the total number of painful procedures sustained by each subject in the 24 hours prior the time of serum cortisol measurement in the 37th postmenstrual week of age was used to measure recent exposure to pain-associated procedures. The same procedures used in the measurement of recurrent exposure to pain-associated procedures were used in the measurement of this variable, but limited to a

24 hour interval. A count of painful procedures, as reported in the medical record for the

24 hour period prior to cortisol measurement, was conducted.

Identification of any occurrence of blood pH (including cord blood pH) less than 7.1, or any event of cardiopulmonary resuscitation requiring chest compressions was used to measure history of severe academia. Measurement of this dichotomous variable was carried out using review of the medical record.

Identification of any exposure to corticosteroids that had been administered prenatally to a mother because of threatened premature delivery was used to

67 measure exposure to antenatal steroids. Measurement of this dichotomous variable was carried out using review of the medical record.

The measurement of gestational age was conducted for each infant by the obstetric staff using last menstrual period or prenatal ultrasound (where prenatal care had been obtained) or by the neonatal staff using the Ballard Score (where prenatal care had not been obtained). The use of different methods of measurement to determine gestational age is common because of varied circumstances surrounding pregnancy and delivery. These different methods have been found to produce reasonably comparable measurements (Ananth, 2007; Ballard et al., 1991). The investigator collected the gestational age measurement from the NICU admission note. Gestational age was recorded in fully completed weeks of gestation.

The measurement of birthweight was conducted for each infant by nursing staff using a bedscale upon admission to the NICU. The investigator collected this measurement from the NICU admission note. Birthweight was recorded in grams.

Identification of the day of age in which the infant had laboratory testing for cortisol and 17-OHP measurement was used to measure days of age at 37 postmenstrual weeks of age. The investigator collected this measurement from the medical record and record the measurement in completed days of age, where the birth date was day of age

#1.

68

Identification of the interval number of days between the birth day and the day in which the infant had a subsequent weight at or above the birth weight was used to measure days of age to attain birthweight. Days of age to attain birthweight was used as a measure of extrauterine growth velocity. The investigator collected this measurement from the medical record and recorded the measurement in completed days of age, where the birth date was day of age #1.

The measurement of weight at the 37th postmenstrual week of age was conducted for each infant by nursing staff using a bedscale. Infant weight was used in the daily calculation of medications, calorie intake and output in the NICU. Daily weight measurement was therefore a clinical preference, although some clinical situations required that an estimated weight be used. Weight at the 37th postmenstrual week of age was defined as the last recorded measured weight prior to cortisol measurement in the

37th postmenstrual week of age. The investigator collected this measurement from the medical record. Weight at the 37th postmenstrual week of age was recorded in grams.

The assignment (measurement) of sex was conducted for each infant by visual inspection of the infant by staff members that were in attendance at the infant’s delivery. The investigator collected this categorical measurement from the delivery room record, which was part of the medical record.

Summary of Variables

The following table is presented to summarize the variables:

69

Table 3.2

Summary of Study Variables

Variable Data type Measurement units (categories) Random cortisol value Continuous micrograms per deciliter

Random 17-OHP value Continuous nanograms per deciliter

Recurrent exposure to pain-associated Continuous number of events procedures

Hours of assisted ventilation Continuous number of hours

Recent exposure to pain-associated Continuous number of events procedures

History of severe acidemia Categorical (yes/no)

Exposure to antenatal steroids Categorical (any/none)

Gestational age at birth Continuous completed weeks

Birthweight Continuous grams

Day of age Continuous number of days

Days of age to attain birthweight Continuous number of days

Weight at the 37th postmenstrual week of Continuous grams age

Sex Categorical (male/female)

70

Data Management and Analysis Plan

Data analysis. Data analysis proceeded in a stepwise fashion, moving from descriptive examination to inferential procedures as indicated and supported by the data. Since the sample size was small, emphasis was placed on univariate and bivariate analyses.

Limited multivariable analyses were planned as a final step to explore the data where such analyses could be supported. Statistical tests included correlational analyses, tests of group differences, and simple linear regression.

Data management. The IBM SPSS Statistics 19 program was used for all data analysis procedures. Prior to analysis, data management included a plan for the identification of missing values. Because infants were not eligible for study recruitment if there was a gap in medical record data, missing values were expected to primarily relate to the measurements of cortisol and 17-OHP. If only one of these two measurements was obtained, the infant was to remain in the study and analyses proceed for where there was complete data. If neither cortisol nor 17-OHP measurements were able to be obtained in the 37th postmenstrual week of age, then the infant data would not have been included in the analysis. There was not to be imputation of missing data.

The first phase of data analysis was descriptive. Means, standard deviations and ranges were reported for continuous data (recurrent exposure to pain-associated procedures, hours of assisted ventilation, cortisol values, 17-OHP values, gestational age at birth, day of age, days of age to attain birthweight, birthweight, weight at the 37th postmenstrual week of age and recent exposure to pain-associated procedures). Relative frequencies were reported for categorical data (sex, exposure to antenatal steroids and history of

71 severe acidemia). As appropriate, continuous variables were categorized and relative frequencies reported. For example, the measurement of recurrent exposure to pain- associated procedures was presented in quartiles. Visual inspection of distributions on continuous variables was conducted to assess for normality.

Scatterplots were constructed for general inspection of the variables. This included scatterplots for the examination of relationships between independent and dependent variables, as well as relationships among independent and dependent variables.

Scatterplot inspection was followed by construction of a correlation matrix of all variables to identify statistically significant relationships. This matrix was developed using Pearson or Spearman rank correlation in the case of continuous data, point-biserial correlation in the case of mixed data, and phi correlation in the case of dichotomous data.

The major independent variables (recurrent exposure to pain-associated procedures, hours of assisted ventilation, gestational age at birth, sex, history of severe acidemia, and recent exposure to pain-associated procedures) were then examined for intercorrelations. In the case of high intercorrelations ( >0.7), an opportunity existed to either collapse across highly correlated variables by deriving a secondary score or to drop less clinically significant variables from the predictive phase of the analysis, where necessary.

Bivariate analyses (t-testing, paired sample t-testing, and non-parametric Mann Whitney

U testing) was applied to examine differences in cortisol and 17-OHP values with respect to categories of age, sex, history of severe acidemia, and exposure to antenatal steroids.

Grunau, Holsti and colleagues have examined differences in cortisol values in infants at

32 weeks PCA using gestational age cutoffs of 24-28 weeks versus 29-32 weeks at birth.

72

The use of similar cutoffs for age was considered, in addition to considering the use of any naturally occurring cutoffs that presented in the data.

The next phase of data analysis addressed the research hypotheses. The first study hypothesis to be tested was as follows:

Recurrent exposure to pain-associated procedures and hours of assisted ventilation will be negatively correlated with random cortisol values at 37 postmenstrual weeks of age.

Correlational analysis was proposed to test this research hypothesis. Because a directional hypothesis was established, one-tailed statistical tests were to be employed.

Separate Pearson correlations between each of the two independent variables (recurrent exposure to pain-associated procedures and hours of assisted ventilation) and the dependent variable (random cortisol values) were planned to examine this relationship. If data was not normally distributed based upon visual inspection of distributions, then non- parametric Spearman rank correlation was to be applied.

The second study hypothesis to be tested was as follows:

Recurrent exposure to pain-associated procedures and hours of assisted ventilation will be negatively correlated with random 17-OHP values at 37 postmenstrual weeks of age.

Correlational analysis was proposed to test this research hypothesis. Separate Pearson correlations between each of the two independent variables (recurrent exposure to pain- associated procedures and hours of assisted ventilation) and the dependent variable

(random17-OHP values) were planned, again using one-tailed statistical tests to reflect the directional hypothesis. If data was not normally distributed based upon visual

73 inspection of distributions, then non-parametric Spearman rank correlation was to be applied.

After testing these research hypotheses, exploratory analyses would be conducted based upon their support by the data. For example, multiple regression analyses or stepwise multiple regression analyses would be considered to evaluate the change of variance in cortisol or 17-OHP values at 37 postmenstrual weeks of age that is explained by recurrent exposure to pain-associated procedures and hours of assisted ventilation after controlling for other factors (gestational age, sex, history of severe acidemia and recent exposure to pain-associated procedures).

Power analysis. Power analysis was based upon the potential use of regression analysis.

The power of the regression analysis was calculated using a sample size of 60 subjects, alpha set at 0.05, and two-tailed significance testing. With two predictor variables

(recurrent exposure to pain-associated procedures and hours of assisted ventilation) added to a set of up to four control variables (e.g.: gestational age at birth, sex, history of severe acidemia and recent exposure to pain-associated procedures), there is power of 0.75 to detect a moderate (f2 = 0.15) effect size for the outcome variable (random cortisol values or random17-OHP values) in each of the regression analyses.

Protection of Human Subjects

This research was conducted only after Institutional Review Board (IRB) protocol review and approval at Columbia University, where the study originated, and at Stony Brook

Hospital, where the subjects were recruited and studied (Appendix C, Appendix D).

74

Informed consent. Informed consent was obtained in accordance with IRB guidelines for investigator review of medical records and for the collection of an additional aliquot of infant blood at 37 postmenstrual weeks of age during routine blood sampling. An

IRB-approved consent form was used (Appendix E). The IRB required the consent of only one parent for this study based upon minimal risk in this research. Consistent with local practice, which has been developed in the context of New York State paternity law and the federal Health Insurance Portability and Accountability Act (HIPAA) privacy rules, paternal consent could be obtained by the investigator only if the condition of marriage to the infant’s birth mother was met, or if there was documentation of paternity.

A parent who consented to have his/her infant participate in the study was given a copy of the consent form. No incentive to participate was offered. Consenting parents were informed of their right to withdraw their infant from the study at any time prior to laboratory sampling. In concordance with IRB policies of Columbia University and

Stony Brook University, appropriate consent translation and/or the use of interpretation services were provided in cases where English is not the primary language of the parent.

The population of parents of subjects in this study was expected to include a number of those who were not fluent in English, but who were fluent in Spanish. The long form consent document was translated into Spanish. Consistent with Stony Brook Hospital policy, Pacific Interpreters Inc. (www.pacificinterpreters.com) provided Spanish translation and interpretation services in the NICU. For parents who were not fluent in

English or Spanish, but who were fluent in another language or who were illiterate, appropriate interpretation services would have been similarly provided, and the short form consent document used and witnessed in accordance with IRB policy.

75

Confidentiality. To maintain confidentiality, data files were maintained in a locked file and on a password-protected laptop computer that was maintained and safeguarded by the investigator and not accessible for unauthorized review. Access to research files was limited to the investigators and the IRB. Infant identifiers were used only for research purposes and will never be used in publication or presentation.

In addition to providing standards for how identifiable health data is collected, stored and utilized in the course of research, the HIPPA impacts how protected health information can be used preparatory to research. In concordance with these standards, any information recorded during the process of identifying infants who were eligible for inclusion in this study was deidentified. Alphanumeric coding was used to protect the identity of the participants during this preparatory to research phase.

76

CHAPTER 4: Results

This chapter presents the data analysis in detail. Presentation begins with a description of the sample, as well as the results of analyses used to examine the stress variables and the glucocorticoid variables. Bivariate correlation analyses of the major variables are next presented. This is followed by a summary of the results of the statistical analyses used to test the research hypotheses. The chapter concludes with the results of several exploratory analyses.

Description of Sample

Recruitment and inclusion. Sixty subjects were recruited from a pool of 94 infants admitted to the NICU at less than 35 completed weeks of gestation between May 1, 2012 and September 12, 2012. From the pool of 94 infants, 30 (31.9%) infants were ineligible to participate. The reasons for ineligibility were IUGR (n=5, 5.3%), postnatal steroid use

(n=1, 1.1%), transfer or surgery (n=4, 4.3%), early discharge (n=16, 17.0%) and death

(n=4, 4.3%).

Closer examination was made of the infants who were excluded because of death, transfer, surgery and postnatal steroid administration (n=9). This group was comprised primarily of male infants (77.8%, 7/9). All infants in this group had a birthweight under

1000 grams (mean 791, range 640-960).

A parent was able to be reached for consent for all eligible infant subjects. Consent was refused in the case of four potential subjects (two sets of twins). In these cases, there was

77 disagreement regarding participation between the mother and the father. The final rate of consent was 93.8% among the eligible subjects.

Data from one subject was excluded because of late identification of IUGR. This resulted in a final sample of 59 subjects.

Laboratory testing was completed on all consented subjects. There were no laboratory or sampling errors resulting in incomplete data collection. All enrolled subjects had complete medical records, allowing for collection of all planned charted data.

Recruitment and inclusion results are summarized in Figure 4.1.

78

Figure 4.1.

Flow Diagram of Recruitment and Inclusion

79

Characteristics of study sample. Subjects had a mean gestational age of approximately

31 weeks at birth and a mean birthweight of 1743 grams. Males and females were similarly represented (47.5% female, 52.5% male). The average length of time in the

NICU prior to laboratory testing (equivalent to total days of age) was approximately 31 days, and mean weight at the time of testing was 2342 grams. A vast majority (94.9%) of subjects had been exposed to at least one dose of antenatal steroid, while only a minority

(6.8%) had sustained severe acidosis. These results are summarized in Table 4.1.

Self-reported race and ethnicity data was collected from each infant’s mother. Bi-racial and bi-ethnic infants were noted to be common in this NICU, however paternal data was not collected. The sample’s mothers self-defined as African-Americans in 13.6% of cases, as Caucasian non-Hispanics in 62.7% of cases, and as Caucasian Hispanics in

23.7% of cases.

80

Table 4.1

Demographic Characteristics of Study Sample (n=59)

Variable Mean Median n % (+/-SD) (range) Gestational Age in completed weeks 30.8 31 (+ 2.24) (25-34)

Birthweight in grams 1743 1690 (+ 513) (640-2880)

37-Week Weight in grams 2342 2305 (+ 372) (1566-3170)

Day of Age at Lab Testing 31.3 27 (+ 16.4) (14-69)

Days of Age at to Attain Birthweight 11.9 12 (+ 3.7) (4-20)

Recent Number of Procedural Exposures 0.39 0 (+ 0.965) (0-5)

Sex Male 31 52.5 Female 28 47.5

Antenatal Steroids Exposure Any 56 94.9 None 3 5.1

History of Severe Acidemia Any 4 6.8 None 55 93.2

Mother’s Reported Race/Ethnicity African-American 8 13.6 Asian 0 0 Caucasian, Hispanic 14 23.7 Caucasian, non-Hispanic 37 62.7 Other 0 0 ______

81

The sample had more representation of infants born at older gestational ages. Only two infants in the sample were born at 25-26 completed weeks of gestation, but 19 infants were born at 33-34 completed weeks of gestation. Twinning affected 25.5% (12/47) of the pregnancies represented by the data such that 39% (23/59) of the subjects in the study were a twin. There was one subject who was not enrolled with her twin because of IUGR in the sibling. Twins were also overrepresented among infants born between 29 and 30 completed weeks of gestation, contributing to higher enrollment in this category (Table

4.2).

Table 4.2

Breakdown of Subjects by Gestational Age and Distribution of Twins

Completed Weeks of n % twin Gestation at Birth subjects

25-26 2 3.4 0

27-28 5 8.5 2

29-30 19 32.2 10

31-32 14 23.7 4

33-34 19 32.2 7 ______

Total: 59 100 23 ______

82

Distributions of demographic/potentially confounding variables. The distributions of the independent variables of gestational age, birthweight, days of age to attain birthweight, number of recent pain-associated procedures, weight at testing and day of age at testing were examined for normality. These distributions are described in Table

4.3 and illustrated in Figures 4.2 through 4.7.

The distribution of birthweight and days of age to attain birthweight both approached the normal distribution upon visual inspection. The distributions for gestational age at birth and days of age at testing were expectedly imaged in opposite directions – fewer infants born at lower gestational ages corresponded with fewer chronologically older infants at the 37th postmenstrual week of age. In addition, while birthweight was relatively normally distributed (skew = 0.15), there was a mild shift to the left in infant weight at 37 post menstrual weeks of age (skew=0.50) suggesting a differential in weight gain over time among infants.

83

Table 4.3

Distribution Descriptors for Independent Variables

Variable Skewness Kurtosis

Gestational Age in completed weeks -0.48 -0.46

Birthweight in grams 0.15 -0.26

Days of Age to Attain Birthweight -0.14 -0.26

Number of Recent Pain-Associated Procedures 3.3 11.7

Weight at 37th Postmenstrual Week in grams 0.50 -0.30

Day of Age at Lab Testing 0.98 0.13 ______Figure 4.2

Figure 4.2

Distribution of Gestational Age at Birth

84

Figure 4.3

Distribution of Birthweight

Figure 4.4

Distribution of Days of Age to Attain Birthweight

85

Figure 4.5

Distribution of Recent Pain-Associated Procedures

Figure 4.6

Distribution of Weight at 37th Postmenstrual Week of Age

86

Figure 4.7

Distribution of Days of Age at Lab Testing

Addressing gestational age as a potential confounder. An important potentially confounding variable in this study was gestational age. The distribution for gestational age approached normal with a clustering at the upper gestational age range, consistent with the tendency for infants to be born closer to term (Figure 4.2). Both the mean and median gestational age at birth in the study sample was approximately 31 weeks.

In addition to its examination as a continuous variable, gestational age was recoded as a dichotomous variable to create comparison groups of infants born at a higher and lower gestational ages.

The sample was divided in two ways for bivariate analyses. First, the sample was divided at the median, creating a group of infants born at the younger gestational age of

87 under 31 completed weeks of gestation (“earlier-born,” n=26, 44.1%), and a group of infants born at the higher gestational age of at or over 31 completed weeks of gestation

(“later-born,” n=33, 55.9%).

Next, the sample was divided where a natural break in the data appeared, between 28 and

29 completed weeks of gestation. This created a group of infants born at the younger gestational age of under 29 completed weeks of gestation (“extremely premature,” n=7,

11.9%), and a group of infants born at the higher gestational age of at or over 29 completed weeks of gestation (“moderately premature,” n=52, 88.1%).

Inter-rater Reliability Testing

The following section presents analyses related to the key stressor variables. Because counts of the key stressor variables were based only upon retrospective medical record review by the researcher, their accuracy was assessed through inter-rater reliability testing. Inter-rater reliability provides a score of how much conformity there is between different raters. Ten percent of the medical records were selected to be independently reviewed by an experienced neonatal nurse practitioner who was not otherwise involved in this investigation. This reviewer was trained on the systematic process used for chart review by the investigator using a chart from the first subject in the study who had elements of assisted ventilation. Each chart review was conducted to obtain a count of the number of hours that a subject received assisted ventilation (hours of assisted ventilation) and a count of the number of painful procedures sustained by each subject

(recurrent exposure to pain-associated procedures) up until the time of serum cortisol measurement in the 37th postmenstrual week of age. The sections of each chart that were

88 examined comprised the clinical flowsheets and the results flowsheets. In addition, for each chart, a review of the immunization history and procedural time-out sheets was conducted.

Following training completion, the charts of the first two subjects completing laboratory analysis each week, for three consecutive weeks, were selected for inter-rater reliability testing. At the end of the three week period, six charts (representing ten percent of the target number of subjects) were reviewed.

Inter-rater reliability was assessed in two ways. First, consistency between raters was assessed by calculating the Pearson correlations between raters for each of the data sets.

For the count of number of hours that a subject received assisted ventilation, the overall

Pearson correlation between raters was 0.999. For the count of number of painful procedures, the overall Pearson correlation between raters was 0.966.

Cohen’s Kappa can also be used in inter-rater reliability analysis to quantify the level of agreement between raters. This statistic improves upon the Pearson correlation by taking into account the amount of agreement that could be expected to occur by chance between two raters. The Kappa statistic indicates the proportion of agreement beyond that expected by chance, and can thereby verify that agreement between two raters exceeds chance levels. For the count of painful procedures, the proportion of agreement between raters (the likelihood that the second reviewer agreed with the investigator about each exposure) was 0.775. For the count of hours that a subject received assisted ventilation, the proportion of agreement between raters was 0.916. Both of these Kappa values were

89 statistically significant (p<0.001). Landis and Koch have proposed that a Kappa statistic of 0.61-0.80 indicates substantial agreement between raters (Landis & Koch, 1977).

Stressor Variables

The first aim of this study was to characterize the nature of recurrent pain-associated procedures for infants in the NICU from birth up until the 37th postmenstrual week of age. Univariate and bivariate analyses were applied to provide a comprehensive characterization of this stressor variable, as well as characterization of exposure to assisted ventilation in this sample of infants.

Recurrent exposure to pain-associated procedures. A total of 3354 pain-associated procedures were identified for the 59 subjects (Table 4.4). Premature infants were exposed to an average of 56.8 of procedures by their 37th postmenstrual week of age.

Heel lance, followed by venipuncture, was the predominant procedure to which infants were recurrently exposed. Heel lance accounted for 80.8% of all procedures, while venipuncture accounted for 9.7% of all procedures.

Over half of the procedures (56.5%) were carried out during the first week of age, and nearly one-third of the procedures (32.6%) occurred in the first three days of age (Table

4.5). Heel lance accounted for nearly two-thirds (65%, 717/1094) of procedures in the first three days of age, and three-quarters (74.4%, 1409/1895) of procedures in the first week of age.

90

Table 4.4

Characteristics of Recurrent Pain-Associated Procedures

Recurrent Frequency (%) Mean (+/- SD) Median (range) Stressor (n=3354) (n=59) (n=59)

Tracheal Intubation 49 (14.6) 0.8 + 1.2 0 (0-5)

Heel Lance 2710 (80.8) 45.9 + 37.7 33 (15-204)

Arterial Puncture 103 (3.1) 1.8 + 2.1 1 (0-14)

Venipuncture 324 (9.7) 5.5 + 3.9 5 (1-18)

Lumbar Puncture 2 (0.06) 0.03 + 0.3 0 (0-2)

Thoracic Puncture 7 (0.2) 0.1 + 0.7 0 (0-5)

Suture Placement * 68 (2.0) 1.1 + 1.3 1 (0-7)

IM/SQ Injection 91 (2.7) 1.5 + 0.9 1 (1-4)

Bladder Puncture 0 (0) 0 + 0 0 (0) ______

Total 3354 (100) 56.8 + 43.4 42 (20-238) ______*sutures were placed for the securement of central catheters and thoracic tubes

91

Table 4.5

Recurrent Stressor Exposure in First Week of Age (n=59)

Stressor First 3-Day First 3-Day First Week First Week Freq (% of Mean Freq (% of Mean like procedure) (range) like procedure) (range)

Tracheal Intubation 32 (65.3) 0.5 (0-2) 41 (83.7) 0.7 (0-4)

Heel Lance 717 (26.5) 12.2 (0-25) 1409 (52.0) 23.9 (10-44)

Arterial Puncture 71 (68.9) 1.2 (0-3) 74 (71.8) 1.3 (0-3)

Venipuncture 156 (48.1) 2.6 (0-12) 250 (77.2) 4.2 (0-17)

Lumbar Puncture 0 (0) 0 (0) 0 (0) 0 (0)

Thoracic Puncture 0 (0) 0 (0) 0 (0) 0 (0)

Suture Placement 59 (86.8) 1 (0-3) 62 (91.2) 1.1 (0-3)

IM/SQ Injection 59 (64.8) 1 (1-1) 59 (64.8) 1 (1-1)

Bladder Puncture 0 (0) 0 (0) 0 (0) 0 (0) ______

Total 1094 18.5 1895 32.1 (32.6) (8-33) (56.5) (16-55) ______

Hours of assisted ventilation. A total of 17,328 hours of assisted ventilation were identified (Table 4.6). Premature infants were exposed to an average of 293 hours of assisted ventilation up until their 37th postmenstrual week of age of which 16.6% were in the form of invasive ventilation. As expected, there was high exposure (31.8% of all hours) to assisted ventilation during the first week of age. In addition, half of the hours

(50.1%) of exposure to invasive ventilation occurred during this week (Table 4.7).

92

Table 4.6

Characteristics of Hours of Assisted Ventilation Exposure

Hours (%) Mean (+/- SD) Median (range) (n=17328) (n=59) (n=59)

All Assisted 17328 (100) 293.2 + 397.2 142 (0-1583) Ventilation

Invasive 2883 (16.6) 48.9 + 145.5 0 (0-962) Ventilation

Non-invasive 14445 (83.4) 237.2 + 293.1 142 (0-1060) Ventilation ______

Table 4.7

Assisted Ventilation Exposure in First Week of Age (n=59)

First 3-Day First 3-Day First Week First Week Total Mean Total Mean (% of total) (range) (% of total) (range)

Hours of 2376 40.3 5517 93.5 Assisted (13.7) (0-69) (31.8) (0-165) Ventilation (n=17328)

Hours of 841 14.3 1444 24.5 Invasive (29.2) (0-69) (50.1) (0-165) Ventilation (n=2883)

Hours of 1535 26.0 4073 69.0 Non-invasive (10.6) (0-68) (28.2) (0-158) Ventilation (n=14445) ______

93

Distributions of stress variables. The distributions of exposures to pain-associated procedures and hours of assisted ventilation were non-normal upon visual inspection

(Figures 4.8 to 4.11). All distributions were right-tailed, meaning that the highest numbers of exposures were limited to only few numbers of infants. Skew and kurtosis measurements identified non-normality to be most marked for total hours of invasive ventilation (most infants were not exposed to invasive ventilation), and least marked for total hours of non-invasive ventilation (many infants received some exposure to non- invasive ventilation) (Table 4.8).

Figure 4.8

Distribution of Exposures to Pain-Associated Procedures

94

Figure 4.9

Distribution of Hours of Assisted Ventilation

Figure 4.10

Distribution of Hours of Invasive Assisted Ventilation

95

Figure 4.11

Distribution of Hours of Non-invasive Assisted Ventilation

Table 4.8

Distribution Descriptors for Stressor Variables

Variable Skewness Kurtosis

Number of Pain-Associated Procedures 2.5 6.8

Total Hours of Assisted Ventilation 1.7 2.3

Total Hours of Invasive Ventilation 5.1 29.0

Total Hours of Non-invasive Ventilation 1.5 1.2 ______

96

Bivariate analyses. Bivariate analysis using Mann Whitney U testing was used to examine if there were any significant differences in levels of recurrent exposure to pain- associated procedures or hours of assisted ventilation based upon the demographic variable of gestational age (dichotomized in two ways) or sex. Exposure to antenatal steroids was not examined as a in this type of bivariate analysis because nearly all subjects (94.9%, n=56) had this exposure. Similarly, severe acidemia was not examined in this type of analysis because very few subjects (6.8%, n=4) had this exposure.

Parametric and non-parametric testing was considered in light of the sample distributions and sample size. Mann Whitney U testing was applied for these bivariate analyses because the distributions of the stressor variables were considered to be non-normal. It was decided that parametric testing was to be applied where the skew of the distribution was <1, and nonparametric testing was to be applied where the skew of the distribution was >1.

When comparing exposure to recurrent stressors by gestational age, the U value was significant (p < 0.001) indicating that there was a statistically significant difference in procedural exposure between the earlier-born and later-born infants, and between extremely premature and moderately premature infants, with a greater number of exposures among infants born at the lower gestational ages (Table 4.9). For sex, the U value was not significant (p =0.298) indicating that there was not a statistically significant difference in procedural exposure between the male and female infants.

When comparing exposure to assisted ventilation by gestational age, the U value was significant (p < 0.001) indicating that there was a statistically significant difference in

97 assisted ventilation exposure between the earlier-born and later-born infants, and between extremely premature and moderately premature infants (Table 4.10). Again, as anticipated, there was a significantly higher exposure to ventilation for infants born at the lower gestational ages. There was not a statistically significant difference in ventilation exposure between the male and female infants.

Table 4.9

Bivariate Analyses Testing for Differences in Recurrent Procedural Exposure

Variable n mean U-statistic P-Value* rank

Gestational Age Earlier-born 26 42.6 110.0 0.000† Later-born 33 20.1

Gestational Age Extremely Premature 7 50.2 40.5 0.000† Moderately Premature 52 27.3

Sex Male 31 27.8 365.5 0.298 Female 28 32.45 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples † Results significant at the 0.01 level

98

Table 4.10

Bivariate Analyses Testing for Differences in Assisted Ventilation Exposure

Variable n mean U-statistic P-Value* rank

Gestational Age Earlier-born 26 43.4 82 0.000† Later-born 33 19.5

Gestational Age Extremely Premature 7 54.1 13.0 0.000† Moderately Premature 52 26.8

Sex Male 31 29.3 412.5 0.743 Female 28 30.8 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples † Results significant at the 0.01 level

Glucocorticoid Variables

As previously described, quantitative chemiluminescent immunoassay was used to measure random cortisol values in the 37th postmenstrual week of age, and radioimmunoassay was used to measure random17-OHP values in the 37th postmenstrual week of age. Cortisol was reported in micrograms per deciliter units and 17-OHP was reported in nanograms per deciliter units.

Characteristics of cortisol and 17-OHP. There was great variability in cortisol and 17-

OHP values for the premature infants at 37 postmenstrual weeks. Random cortisol values ranged from 1.3 to 31.4 while random 17-OHP values ranged from 56 to 650 (Table

4.11).

99

Table 4.11

Characteristics of Cortisol and 17-OHP Values at 37 Postmenstrual Weeks of Age (n=59)

Parameter Cortisol 17-OHP (ug/dl) (ng/dl)

Range 1.3-31.4 56-650 Mean 7.0 238 Median 4.4 215 SD 6.0 118

1st Quartile 2.6 157 3rd Quartile 9.4 296

10th Percentile 1.8 110 90th Percentile 15.5 411 ______

As previously described, other researchers have examined cortisol and 17-OHP values at

37 postmenstrual weeks of age for infants born at the lowest gestational ages (Al Saedi,

Dean, Dent, & Cronin, 1995; Arnold, et al., 1997). To compare our findings with these study results, data from subgroups of earlier-born infants in our study has been extracted and analyzed. The characteristics of cortisol and 17-OHP at 37 postmenstrual weeks of age for infants born at under 30 weeks of gestation and at under 31 weeks of gestation are presented in Table 4.12.

100

Table 4.12

Characteristics of Cortisol and 17-OHP: Gestational Age Subgroups ______<31 Weeks Gestation <30 Weeks Gestation Parameter Cortisol 17-OHP Cortisol 17-OHP (ug/dl) (ng/dl) (ug/dl) (ng/dl) n=26 n=26 n=18 n=18

Range 1.4-24.6 57-650 1.4-15.6 57-495 Mean 6.6 250 5.2 215 Median 4.6 231 3.9 179 SD 5.4 144 4.5 93

1st Quartile 2.4 143 2.0 129 3rd Quartile 8.8 318 8.3 298

10th Percentile 1.8 77 1.8 57 90th Percentile 15.2 447 9.7 430 ______

The distribution of cortisol values was non-normal upon visual inspection (Figure 4.12) while the distribution of 17-OHP values more closely approached the normal distribution

(Figure 4.13). Skew and kurtosis measurements supported these observations (Table

4.13).

Table 4.13

Distribution Descriptors for Cortisol and 17-OHP Values

Variable Skewness Kurtosis

Cortisol 1.9 4.4

17-OHP 1.1 1.7 ______

101

Figure 4.12

Distribution of Cortisol Values

Figure 4.13

Distribution of 17-OHP Values

102

Bivariate analyses. Bivariate analysis using Mann Whitney U testing was used to examine if there were any significant differences in values of cortisol or 17-OHP based upon the demographic variables of gestational age or sex. Nonparametric testing was applied, again using the criteria that both sampling distributions were non-normal (skew

>1).

There were no significant difference in either cortisol or 17-OHP values with respect to either gestational age at birth or sex (Tables 4.14 and 4.15).

Table 4.14

Bivariate Analyses Testing for Differences in Cortisol Values

Variable n mean U-statistic P-Value* rank

Gestational Age Earlier-born 26 29.7 420 0.891 Later-born 33 30.3

Gestational Age Extremely Premature 7 35.4 144.0 0.387 Moderately Premature 52 29.3

Sex Male 31 29.1 405 0.2660 Female 28 31.0 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples

103

Table 4.15

Bivariate Analyses Testing for Differences in 17-OHP Values

Variable n mean U-statistic P-Value* rank

Gestational Age Earlier-born 26 30.3 422.5 0.0921 Later-born 33 29.8

Gestational Age Extremely Premature 7 24.9 146.0 0.413 Moderately Premature 52 30.7

Sex Male 31 30.8 409.0 0.704 Female 28 29.1 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples

Correlation Analyses

Scatterplots. Scatterplots were developed to inspect for patterns of linearity among the major variables of interest.

First, scatterplots were developed to examine for linear relationships between the two major stressor variables (recurrent exposure to pain-associated procedures and hours of assisted ventilation) and to examine for linear relationships between two adrenocorticoid variables (random cortisol values and random 17-OHP values) (Figures 4.14 and 4.15).

Inspection of these scatterplots suggests that there is a strong positive linear relationship between the two major stressor variables and a weak positive linear relationship between the two adrenocorticoid variables.

104

Figure 4.14

Scatterplot: Recurrent Pain-Associated Procedures versus Hours of Assisted Ventilation

Figure 4.15

Scatterplot: Cortisol versus 17-OHP

105

Next, scatterplots were developed to examine for linear relationships between the two major stressor variables and cortisol (Figures 4.16 and 4.17). Inspection of these scatterplots suggests that there is not a linear relationship between either of the major stressor variables and cortisol.

Figure 4.16

Scatterplot: Recurrent Pain-Associated Procedures versus Cortisol

106

Figure 4.17

Scatterplot: Hours of Assisted Ventilation versus Cortisol

Finally, scatterplots were developed to examine for linear relationships between the two major stressor variables and 17-OHP (Figures 4.18 and 4.19). Inspection of these scatterplots suggests that there may be a weak negative linear relationship between recurrent pain-associated procedures and 17-OHP, but no relationship between hours of assisted ventilation and 17-OHP.

107

Figure 4.18

Scatterplot: Recurrent Pain-Associated Procedures versus 17-OHP

Figure 4.19

Scatterplot: Hours of Assisted Ventilation versus 17-OHP

108

Correlation matrices. Correlation matrices were constructed using all of the variables to explore these relationships and examine for confounding. The first matrix was developed using phi correlation for the cases of dichotomous data (Figure 4.20). The second matrix was developed using point-biserial correlation for the cases of mixed data

(Figure 4.21). The third matrix was developed using Spearman rank correlation, for the cases of continuous data where there was a skew >1 in the distribution of at least one variable (cortisol, 17-OHP, recurrent pain-associated procedures, all ventilation variables, recent pain-associated procedures) (Figure 4.22). The final matrix was developed using

Pearson correlation in the cases of continuous data with more normal distributions

(Figure 4.23).

As expected, severe acidemia was significantly negatively correlated (p<0.01) with earlier gestational age (r= -0.493), lower birthweight (r= -0.440), age at testing (r=

-0.489), greater procedural pain exposure (r= -0.695), and greater exposure to ventilation

(r= -0.559). Additionally, male sex was found to be significantly correlated (p<0.05) with higher birthweight (r= -0.292), higher weight at testing (r= -0.383) and lower incidence of severe acidosis (r= 0.284).

109

Figure 4.20

Phi Correlation Matrix

Antenatal History of Severe Steroid

Acidemia Exposure Sex (any/none) (any/none) Sex Phi Correlation 1 .284* .220 Sig. (2-tailed) .029 .094 N 59 59 59 History of Severe Acidemia Phi Correlation 1 .062 (any/none) Sig. (2-tailed) .639 N 59 59 Antenatal Steroid Exposure Phi Correlation 1 (any/none) Sig. (2-tailed) N 59 * Correlation is significant at the 0.05 level (2-tailed).

110

Figure 4.21

Point-Biserial Correlation Matrix

Antenatal Steroid Severe Acidemia Exposure Exposure (any/none) (any/none) Sex Cortisol Value(ug/dl) Point-Biserial Correlation .071 .013 .107 Sig. (2-tailed) .593 .921 .421 N 59 59 59 17-OHP Value (ng/dl) Point-Biserial Correlation -.148 -.218 -.079 Sig. (2-tailed) .264 .097 .550 N 59 59 59 Number of Pain-Associated Point-Biserial Correlation .094 .695** .236 Procedures Sig. (2-tailed) .478 .000 .072 N 59 59 59 Total Hours of Assisted Point-Biserial Correlation .135 .559** .058 Ventilation Sig. (2-tailed) .309 .000 .663 N 59 59 59 Total Hours of Invasive Point-Biserial Correlation .074 .705** .210 Assisted Ventilation Sig. (2-tailed) .577 .000 .110 N 59 59 59 Total Hours of Non-invasive Point-Biserial Correlation .140 .319* -.051 Assisted Ventilation Sig. (2-tailed) .290 .014 .704 N 59 59 59 Gestational Age at Birth Point-Biserial Correlation -.193 -.493** -.195 (completed weeks) Sig. (2-tailed) .143 .000 .138 N 59 59 59 Birthweight (grams) Point-Biserial Correlation -.005 -.440** -.292* Sig. (2-tailed) .972 .000 .025 N 59 59 59 Days of Age to Attain Point-Biserial Correlation .137 .082 .053 Birthweight Sig. (2-tailed) .301 .535 .690 N 59 59 59 Weight at Testing (grams) Point-Biserial Correlation .149 -.305* -.383** Sig. (2-tailed) .261 .019 .003 N 59 59 59 Day of Age at Testing Point-Biserial Correlation .133 .489** .124 Sig. (2-tailed) .317 .000 .348 N 59 59 59 Number of Recent Pain- Point-Biserial Correlation .094 .242 .180 Associated Procedures Sig. (2-tailed) .478 .064 .172 (prior 24 hours) N 59 59 59 * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

111

Figure 4.22

Spearman Rank Correlation Matrix (Part A)

Total Hours Number of Total Hours of Non- Cortisol 17-OHP Pain- Total Hours of Invasive invasive Value Value Associated of Assisted Assisted Assisted (ug/dl) (ng/dl) Procedures Ventilation Ventilation Ventilation Cortisol Value (ug/dl) Spearman Coefficient 1.000 Sig. (2-tailed) . N 59 17-OHP Value (ng/dl) Spearman Coefficient .247 1.000 Sig. (2-tailed) .059 . N 59 59 Number of Pain- Spearman Coefficient -.044 -.232 1.000 Associated Procedures Sig. (2-tailed) .742 .077 . N 59 59 59 Total Hours of Assisted Spearman Coefficient -.047 -.132 .824** 1.000 Ventilation Sig. (2-tailed) .723 .318 .000 . N 59 59 59 59 Total Hours of Invasive Spearman Coefficient -.070 -.071 .685** .819** 1.000 Assisted Ventilation Sig. (2-tailed) .598 .592 .000 .000 . N 59 59 59 59 59 Total Hours of Non- Spearman Coefficient -.041 -.127 .820** .991** .792** 1.000 Invasive Assisted Sig. (2-tailed) .760 .336 .000 .000 .000 . Ventilation N 59 59 59 59 59 59

112

Spearman Rank Correlation Matrix (Part B)

Number of Recent Gestational Pain- Age at Associated Birth Days of Age Weight at Procedures (completed Birthweight To Attain Testing Day of Age (prior 24 weeks) (grams) Birthweight (grams) at Testing hours) Cortisol Value (ug/dl) Spearman Coefficient .071 .088 .083 .042 -.008 .156 Sig. (2-tailed) .591 .507 .531 .750 .953 .238 N 59 59 59 59 59 59 17-OHP Value (ng/dl) Spearman Coefficient .045 .064 .028 -.080 -.140 -.148 Sig. (2-tailed) .738 .631 .831 .547 .290 .265 N 59 59 59 59 59 59 Number of Pain- Spearman Coefficient -.753** -.672** .407** -.182 .801** .443** Associated Procedures ( Sig. (2-tailed) .000 .000 .001 .167 .000 .000 N 59 59 59 59 59 59 Total Hours of Assisted Spearman Coefficient -.762** -.598** .512** -.031 .777** .279* Ventilation Sig. (2-tailed) .000 .000 .000 .816 .000 .032 N 59 59 59 59 59 59 Total Hours of Invasive Spearman Coefficient -.721** -.632** .334** -.095 .728** .236 Assisted Ventilation Sig. (2-tailed) .000 .000 .010 .472 .000 .072 N 59 59 59 59 59 59 Total Hours of Non- Spearman Coefficient -.753** -.599** .551** -.019 .771** .256 Invasive Assisted Sig. (2-tailed) .000 .000 .000 .886 .000 .051 Ventilation N 59 59 59 59 59 59 Gestational Age at Spearman Coefficient 1.000 -.223 Birth Sig. (2-tailed) . P P P P .089 (completed weeks) N 59 59 Birthweight (grams) Spearman Coefficient 1.000 -.243 Sig. (2-tailed) . P P P .064 N 59 59 Day of Age When Spearman Coefficient 1.000 .279* Attains Birthweight Sig. (2-tailed) . P P .032 N 59 59 Weight at Testing Spearman Coefficient 1.000 -.251 (grams) Sig. (2-tailed) . P .055 N 59 59 Day of Age at Testing Spearman Coefficient 1.000 .281* Sig. (2-tailed) . .031 N 59 59 Number of Recent Pain- Spearman Coefficient 1.000 Associated Procedures Sig. (2-tailed) . (prior 24 hours) N 59 * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). (P = see Pearson correlation matrix)

113

Figure 4.23

Pearson Correlation Matrix

Gestational Age at Birth Day of Age

(completed Birthweight When Attains Day of Age at Weight at Testing weeks) (grams) Birthweight Testing (grams) Gestational Age at Pearson Correlation 1 .858** -.262* -.936** .211 Birth Sig. (2-tailed) .000 .045 .000 .109 (completed weeks) N 59 59 59 59 59 Birthweight (grams) Pearson Correlation 1 -.139 -.796** .566** Sig. (2-tailed) .295 .000 .000 N 59 59 59 59 Day of Age When Pearson Correlation 1 .293* -.037 Attains Birthweight Sig. (2-tailed) .025 .783 N 59 59 59 Day of Age at Testing Pearson Correlation 1 -.054 Sig. (2-tailed) .684 N 59 59 Weight at Testing Pearson Correlation 1 (grams) Sig. (2-tailed) N 59 *Correlation is significant at the 0.05 level (2-tailed). **Correlation is significant at the 0.01 level (2-tailed).

Identification of collinearity. Collinearity among the study variables can impede statistical analyses and required identification prior to the exploratory regression analyses. For this research, collinearity was defined as a highly significant correlation of

> 0.7 or < -0.7 (p<0.001). Multicollinearity existed between the two major stressor variables (r= 0.824), as well as between each of the major stressor variable and several of the other study variables (Table 4.16). Collinearity also existed between hours of invasive and non-invasive ventilation (r=0.792), and gestational age at birth was collinear with birthweight (r=0.858), day of age at testing (r= -0.936), and total hours of non- invasive (r= -0.753), and invasive ventilation (r= -0.721). Of note, birthweight was not collinear with the major stressor variables.

114

Table 4.16

Identification of Collinearity

Variables Recurrent Hours of Pain-Associated Assisted Procedures Ventilation (r) (r)

Gestational Age -0.753 -0.762 Days of Age at Testing 0.801 0.777 Birthweight (n/a) (n/a)

Hours of Invasive Ventilation (n/a) 0.819 Hours of Non-invasive Ventilation 0.820 0.991 ______Collinearity is defined as a highly significant correlation of > 0.7 or < -0.7 All results are significant at < 0.001 level

The lack of collinearity between recurrent pain-associated procedures and hours of invasive ventilation warranted further exploration. It was anticipated that infants requiring the greatest ventilatory support would have had the highest incidence of procedural pain exposure. The early placement of central arterial and venous catheters in our sample of infants requiring assisted ventilation was postulated as the factor that precluded collinearity between procedural pain exposure and hours of invasive ventilation. To test this hypothesis, Spearman rank correlation (1-tailed) was conducted to examine the relationship between invasive ventilation in the first week of age and suture placement in the first week of age. A single suture was used in this NICU during three procedures - umbilical arterial catheter placement, umbilical venous catheter placement, and thoracic tube placement. As no thoracic tubes were placed in this sample of infants during the first week of age, all sutures during this time were related to a central umbilical catheter placement. During this first week, infants were exposed to

115

1444 hours of invasive ventilation (50.1% of all hours) and 62 suture placements (91.2% of all sutures). The correlation between these variables was 0.783 (p<0.001), indicating a significant positive relationship between hours of invasive ventilation and suture placement (or, in this case, a significant positive relationship between hours of invasive ventilation and umbilical line placement).

Hypothesis Testing

The second aim of this study was to examine relationships between recurrent exposure to pain-associated procedures, hours of assisted ventilation, and random cortisol and 17-

OHP values at 37 postmenstrual weeks of age in a sample of premature infants who had received care for a minimum of two weeks in the NICU.

Correlation between stressor exposure and cortisol values. The first hypothesis associated with this study aim was:

Recurrent exposure to pain-associated procedures and hours of assisted ventilation will be negatively correlated with random cortisol values at 37 postmenstrual weeks of age.

Correlation analysis was proposed to test this research hypothesis. Spearman rank correlation was applied (one-tailed) because the distributions of cortisol, recurrent pain- associated procedures and hours of assisted ventilation were determined to be non-normal based upon visual inspection and examination of skew (skew >1).

There was not a significant linear relationship between recurrent exposure to pain- associated procedures and random cortisol values at 37 postmenstrual weeks of age. In addition, there was not a significant linear relationship between hours of assisted

116 ventilation and random cortisol values at 37 postmenstrual weeks of age. The first research hypothesis was not supported (Table 4.17).

Table 4.17

Hypothesis Testing: Correlations between Stressor Variables and Cortisol ______Cortisol ______

Stressor Variable n=59 P-value* ______

Number of Pain-Associated -0.044 0.371 Procedures

Total Hours of Assisted -0.047 0.362 Ventilation ______* Analyses were performed using one-tailed Spearman rank correlation

Correlations between cortisol values and the ventilation subgroups (hours of invasive and non-invasive ventilation) were additionally analyzed but failed to reveal significant findings (Table 4.18).

117

Table 4.18

Correlations between Ventilation Variables and Cortisol ______Cortisol ______

Ventilation Variable n=59 P-value* ______

Hours of Invasive -0.70 0.299 Assisted Ventilation

Hours of Non-invasive Assisted Ventilation -0.041 0.380 ______* Analyses were performed using one-tailed Spearman rank correlation

Correlation between stressor exposure and 17-OHP values. The second hypothesis associated with this study aim was:

Recurrent exposure to pain-associated procedures and hours of assisted ventilation will be negatively correlated with random 17-OHP values at 37 postmenstrual weeks of age.

Correlation analysis was proposed to test this research hypothesis. Spearman rank correlation was applied (one-tailed) because the distributions of 17-OHP, recurrent pain- associated procedures and hours of assisted ventilation were determined to be non-normal based upon visual inspection and examination of skew (skew >1).

There was a significant negative correlation between recurrent exposure to pain- associated procedures and random17-OHP values at 37 postmenstrual weeks. There was not a significant relationship between hours of assisted ventilation and random17-OHP values at 37 postmenstrual weeks (Table 4.19).

118

Table 4.19

Hypothesis Testing: Correlations between Stressor Variables and 17-OHP ______17-OHP ______

Stressor Variable n=59 P-value* ______

Number of Pain-Associated -0.232 0.039† Procedures

Total Hours of Assisted -0.132 0.159 Ventilation ______* Analyses were performed using one-tailed Spearman rank correlation † Correlation is significant at the 0.05 level

Correlations between 17-OHP and the ventilations subgroups were additionally analyzed but failed to reveal significant findings (Table 4.20).

Table 4.20

Correlations between Ventilation Variables and 17-OHP ______17-OHP ______

Ventilation Variable n=59 P-value* ______

Hours of Invasive -0.071 0.296 Assisted Ventilation

Hours of Non-invasive Assisted Ventilation -0.127 0.168 ______* Analyses were performed using one-tailed Spearman rank correlation

119

Exploratory Analyses

Since there was not a correlation between 17-OHP and hours of assisted ventilation, regression analysis for this aspect of the second research hypothesis could not be supported.

A regression model for 17-OHP, however, was developed using the predictor variable of recurrent exposure to pain-associated procedures. Correlation analyses were reviewed for other significant predictor variables to include in this analysis, and the use of birthweight was considered. Birthweight was significantly correlated with recurrent exposure to pain-associated procedures (r= -0.672) however did not meet the definition for collinearity.

A simple linear regression was created of recurrent pain-associated procedures on 17-

OHP values. In this univariable model, the coefficient for recurrent pain-associated procedures was -0.502 indicating that for each additional procedural exposure, there was a decrease in the 17-OHP value by 0.5 ng/dl (Table 4.21). The R2 of this model indicates that only 1.7% of the variance in 17-OHP values can be explained by recurrent pain- associated procedures. This is not a statistically significant relationship (p=0.161).

120

Table 4.21

Summary of Simple Linear Regression Model for 17-OHP

______

Independent Variable Adj β SE t Sig. Adj. R2

______

Recurrent Pain- -0.502 0.354 -1.421 0.161 Associated Stress 0.017 ______

A multiple linear regression model was next created of recurrent pain-associated procedures on 17-OHP values after controlling for birthweight. The percent of variability accounted for in the prediction of 17-OHP increased from 1.8% to 2.7% with the addition of the stressor variable to a model already including birthweight (Table 4.22). Neither the first model (birthweight) nor the second model (birthweight plus pain-associated procedures) predicted 17-OHP values to a statistically significant degree (p=0.984, p=0.172).

121

Table 4.22

Summary of Multiple Linear Regression Model for 17-OHP

______

Independent Variable Adj β SE t p-value Adj. R2 ______

Birthweight -0.001 0.030 -0.020 0.984

Model 0.984 -0.018 ______

Birthweight -0.050 0.039 -1.264 0.211

Recurrent Pain- -0.887 0.465 -1.907 0.062 Associated Procedures

Model 0.172 0.027 ______

Testing of collinear variables. Because of collinearity between recurrent pain- associated procedures and hours of assisted ventilation, hours of invasive ventilation, hours of non-invasive ventilation, gestational age, and day of age, these variables could not be used together in the regression analysis. The extent to which these variables and birthweight were related to 17-OHP, in comparison the recurrent pain-associated procedures, was examined in another way. Using a one-tailed model, the relationship between each of these variables and 17-OHP was examined for the strength of a directional (negative) correlation. These analyses failed to identify any other significant relationships (Table 4.23).

122

Table 4.23

Correlations between 17-OHP and Collinear Study Variables ______17-OHP ______Variable n=59 P-value* ______

Recurrent Pain-Associated Procedures -0.232 0.039†

Days of Age -0.140 0.145

Hours of Assisted Ventilation -0.132 0.159

Hours of Non-invasive Ventilation -0.127 0.168

Hours of Invasive Ventilation -0.071 0.296

Gestational Age 0.045 0.369 ______* Analyses were performed using one-tailed Spearman rank correlation † Correlation is significant at the 0.05 level

Recent pain-associated procedures. Previous research by Holsti has described alterations in ACTH and cortisol values in premature infants at 32 postmenstrual weeks of age who had been exposed to skin-breaking procedures in the previous 24 hours

(Holsti, et al., 2007). Bivariate analysis using Mann Whitney U testing was conducted to examine for differences in cortisol values relative to recent procedures in this data set.

The recent pain-associated procedure variable was recoded as a dichotomous variable

(any/none) to create comparison groups of infants who had any procedural pain exposure in the previous 24 hours (13/59, 22%) and no procedural pain exposure in the previous 24 hours (46/59, 78%).

123

There were no differences in either cortisol or 17-OHP values between infants who had and had not been exposed to any pain-associated procedures in the previous 24 hours

(Table 4.24). For this sample, it does not appear that adrenocorticoid values remain transiently altered in the 24 hours following skin breaking procedures.

Table 4.24

Bivariate Analyses Testing for Differences in Glucocorticoid Values after Recent Pain- Associated Procedures

Variable N mean U-statistic P-Value* rank

Cortisol Any recent pain 13 35.4 229 0.200 No recent pain 46 28.5

17-OHP Any recent pain 13 24.7 230 0.207 No recent pain 46 31.5 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples

Birthweight and weight gain. As previously described, male sex was correlated with higher birthweight, as well as higher weight at 37 postmenstrual weeks of age. Further analysis using independent samples t-testing revealed significant differences in weight between the male and female infants in this sample (Table 4.25).

124

Table 4.25

Bivariate Analyses Testing for Differences in Weight between Male and Female Infants

Variable N mean t-statistic P-Value* (SD)

Birthweight Male 31 1885 (487) 2.31 0.025† Female 28 1587 (504)

37-Week Weight Male 31 2476 (371) 3.13 0.003†† Female 28 2193 (317) ______*Bivariate analyses were performed using independent samples t-testing † Results significant at the 0.05 level †† Results significant at the 0.01 level

Correlation analysis also revealed significant relationships between days of age to attain birthweight and all of the stress variables (recurrent pain-associated procedures, hours of assisted ventilation, hours of invasive ventilation and hours of non-invasive ventilation).

This data was further examined in two ways. First, the two major stressor variables were divided at the median and the days of age to attain birthweight were examined for significant differences. Next, days of age to attain birthweight was divided at the median and the two major stressor variables were examined for significant differences.

Mann Whitney U-Testing revealed significant differences in both recurrent pain- associated procedures and hours of assisted ventilation with respect to days of age to attain birthweight, as well as significant differences in days of age to attain birthweight with respect to both recurrent pain-associated procedures and hours of assisted ventilation

(Table 4.26).

125

Table 4.26

Bivariate Analyses Exploring Days to Attain Birthweight and Exposure to Stressors

Variable N mean U-statistic P-Value* rank (95% CI)

Recurrent Pain-Associated Procedures Attain Birthweight < 12Days 35 24.4. 224 0.002† Attain Birthweight > 12 Days 24 38.2

Hours of Assisted Ventilation Attain Birthweight < 12 Days 35 22.9 171 0.000† Attain Birthweight > 12 Days 24 40.4

Days of Age to Attain Birthweight Recurrent Pain < 42 Events 30 24.2 262 0.008† Recurrent Pain > 42 Events 29 36.0

Days of Age to Attain Birthweight Assisted Ventilation < 142 Hours 30 22.4 206 0 .000† Assisted Ventilation > 142 Hours 29 37.9 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples †Results significant at the 0.01 level

Extrauterine growth velocity is an important indicator of infant health, just as the rate of

“catch-up” growth following early life disease has been used as a proxy for premature infant resilience. Using independent sample t-testing (2-tailed) the data were examined to see if there were differences in adrenocorticoid values between infants who had earlier attainment of their birthweight and infants who had later attainment of their birthweight.

There were no differences in cortisol or 17-OHP at 37 postmenstrual weeks of age between infants who had earlier attainment of their birthweight and infants who had later attainment of their birthweight (Table 4.27).

126

Table 4.27

Bivariate Analyses Testing for Differences in Glucocorticoid Values between Infants who Attain Birthweight Earlier and Later

Variable N mean t-statistic P-Value* (SD)

Cortisol Early (< 12 days) 24 6.7 (5.7) -0.207 0.837 Late (> 12 days) 35 7.1 (6.3)

17-OHP Early (< 12 days) 24 222 (108) -0.822 0.415 Late (> 12 days) 35 248 (125) ______*Bivariate analyses were performed using independent samples t-testing

The 90th percentile of stress. Grunau and colleagues suggested that there may be a critical threshold of stressful exposures that must be reached before seeing an adaptation of the HPA axis (Grunau, et al., 2005). To examine this hypothesis, adrenocorticoid values in infants whose stressful exposures were above the 75th percentile for this sample distribution were compared to infants whose stressful exposures were below the 25th percentile for this sample distribution. Non-parametric Mann Whitney U testing was applied after the appropriate sampling quartiles were identified.

There were no significant differences in cortisol values between infants who had experienced the highest stressor exposure compared with those who had experienced the lowest stressor exposure for both recurrent pain-associated procedures and for hours of assisted ventilation (Table 4.28). In addition, there was no significant difference in 17-

OHP values between infants who had experienced the highest stressor exposure

127 compared with those who had experienced the lowest stressor exposure for assisted ventilation. The difference in 17-OHP values between infants who had been exposed to the greatest number of recurrent pain-associated procedures compared with those who had been exposed to the lowest number of recurrent pain-associated procedures, however, approached significance (p=0.068) (Table 4.29).

Table 4.28

Bivariate Analyses Testing for Differences in Cortisol Values between Infants with Extremes of Stress Exposure

Variable N mean U-statistic P-Value* rank

Number of Pain-Associated Procedures

> 75th % (> 67 procedures) 15 16.9 92 0.395 < 25th % (< 31 procedures) 15 14.1

Total Hours of Assisted Ventilation

> 75th % (> 355 hours) 15 15.2 108 0.852 < 25th % (< 10 hours) 15 15.8 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples

128

Table 4.29

Bivariate Analyses Testing for Differences in 17-OHP Values between Infants with Extremes in Stressor Exposure

Variable N mean U-statistic P-Value* rank

Number of Pain-Associated Procedures

> 75th % (> 67 procedures) 15 18.4 68.5 0.068 < 25th % (< 31 procedures) 15 12.6

Total Hours of Assisted Ventilation

> 75th % (> 355 hours) 15 16.4 98.5 0.581 < 25th % (< 10 hours) 15 14.6 ______*Bivariate analyses were performed using Mann Whitney U test for independent samples

Twins. There were eleven sets of twins with complete data represented in the database.

To this point in the analysis, the data from the infant subjects was considered to be independent. The availability of twin data, however, presented a unique opportunity to examine differences in adrenocorticoid values with respect to birthweight. While it is well-accepted that IUGR is associated with long-term alterations in cortisol, there may also be alterations in adrenocorticoid values for infants who are small at birth as a result of relative intrauterine stress, but who do not fit the strictest criteria for IUGR.

Bivariate analysis using paired sample t-testing was conducted to examine for differences in cortisol and 17-OHP values between the larger-born and smaller-born of twins. One set of twins was excluded in the analysis because of identical birthweights.

129

There were no differences in either cortisol values or 17-OHP values between the larger- born and smaller-born of twins although the results are cautionary given the small sample size (Table 4.30).

Table 4.30

Bivariate Analyses Testing for Differences in Glucocorticoid Values between Larger- Born and Smaller-Born Twins

Variable N mean mean t-statistic P-Value* (SD) difference (SD)

Cortisol Larger-born 10 4.5 (3.9) -1.2 (6.8) -0.554 0.593 Smaller-born 10 5.7 (3.9)

17-OHP Larger-born 10 235 (104) 2.8 (166) 0.053 0.959 Smaller-born 10 233 (174) ______*Bivariate analyses were performed using paired sample t-testing

130

CHAPTER 5: Discussion

Introduction

This study examined the relationship between cortisol and 17-OHP values at 37 postmenstrual weeks of age and two major types of pain-associated stressor exposure in premature infants during hospitalization in the NICU. This study is the first to describe both cortisol and 17-OHP values in relation to pain-associated procedural exposure and hours of assisted ventilation in a sample of premature infants who had received at least two weeks of NICU care.

Of importance, this study establishes that there is a statistically significant relationship between recurrent pain-associated procedural exposure in premature infants and an aspect of their adrenocorticoid profile at 37 postmenstrual weeks of age. This research supported the hypothesis that recurrent exposure to pain-associated procedures was negatively correlated with random 17-OHP values at 37 postmenstrual weeks of age (r=

-0.232, p=0.039). A trend was also identified for differences in 17-OHP values between infants who had the highest and lowest pain-associated procedural exposure (U=68.5, p=0.068).

Dissociation between stress and cortisol values in early infancy has been identified across numerous previous studies (Holsti, et al., 2007; M. Lee, et al., 1989). In concordance with these earlier results, this research also finds no significant relationship between recurrent pain-associated procedures (r= -0.044, p=0.742) or hours of assisted ventilation

(r= -0.047, p=0.723) and cortisol values in premature infants at 37-postmenstrual weeks of age.

131

The findings of this research does not support that recurrent pain-associated procedural exposure is predictive of 17-OHP values at 37 postmenstrual weeks (adj R2= 0.017, p=0.161) and does not establish a relationship between length of assisted ventilation and

17-OHP values at 37 postmenstrual weeks (r= -0.132, p=0.318). While support of the correlational hypothesis gives reason to pursue this line of inquiry, inability to identify a predictor of hormonal values at 37 postmenstrual weeks of age emphasizes that there remains much to be learned about the factors that influence the adrenocorticoid profile of prematurely born infants.

This study is also important in that it successfully tackled some of the difficult tasks associated with clinical translational research. In particular, the study was not impeded by lack of affordable in-house laboratory testing, successfully utilizing an outside clinical laboratory for specimen analysis. The separation of the clinical and research records was successfully accomplished through novel admitting department mechanisms. In addition, coordination of research activities with clinical activities in this intensive care unit was found to be feasible, diminished patient burden, and was likely a factor that supported

IRB approval. Success in conducting the research in this patient-oriented way, rather than using bench science methodologies, has important implications for promoting the integration of research findings into practice. Should future research support 17-OHP as a physiologic indicator for cumulative NICU stress, then support for its use as a practical biomarker is strengthened by knowledge of its feasible collection using routine clinical and laboratory procedures.

132

The following sections of this discussion will address specific findings of this study.

This will include an examination of specific variables, relationships that were identified between variables, and the comparison of this study’s findings to those of other investigators. Finally, strengths, challenges and limitations are discussed, as well as implications for practice and future research.

Cortisol and 17-OHP Values

Previous research has suggested that glucocorticoid values in both term and premature infants vary in relation to postmenstrual age (Bolt, Van Weissenbruch, et al., 2002b;

Holsti, et al., 2007; Linder, et al., 1999). While, normative data for random cortisol and

17-OHP values are widely available for term infants at standardized postmenstrual ages

(e.g.: at term birth), comparatively little data is available for premature infants in whom standardization for postmenstrual age is made difficult by varying gestational ages at birth. Consequently, there is speculation regarding the clinical significance of cortisol and 17-OHP values in NICU practice, particularly cases of low serum cortisol and high serum 17-OHP values.

Nevertheless, we compare the results obtained in this sample of infants to results obtained in similar samples by other researchers. In 1995, Al Saedi and colleagues reported weekly cortisol and 17-OHP values for 39 infants born at under 31 weeks of gestation up until the 37th postmenstrual week of age (Al Saedi, et al., 1995). In 1997, Arnold and colleagues followed cortisol and 17-OHP values to 16 weeks of age in premature infants born at under 30 weeks of gestation (Arnold, et al., 1997). Cortisol and 17-OHP values at approximately 37 postmenstrual weeks of age were reported in both of these studies.

133

These findings can be compared to cortisol and 17-OHP values from subgroups of infants in the current study (see Table 4.12). To facilitate comparison, hormone values were standardized across studies and presented in micrograms per deciliter units.

The median cortisol values in the current study sample were higher than the results reported by both Al Saedi and by Arnold (Table 5.1). These values, however, are within the broad accepted range for infants in their first year of age (Soldin, Brugnara, & Wong,

2003). One explanation for this difference may relate to nutritional status differences of the premature infants in the studies, which were conducted approximately 15 years apart.

In each of these studies, total cortisol was measured, rather than biologically active

(unbound) free cortisol. Values of nutrient transport proteins are recognized as an etiology for cortisol variance in hospitalized patients (Beishuizen, Thijs, & Vermes,

2001; Langer, et al., 2006). Improvements in parenteral and enteral nutrition for premature infants over the years have targeted circulating values of albumin transport proteins (van den Akker et al., 2007), which were not measured as part of the adrenocorticoid hormone studies. Reported cortisol values in these studies do not accurately reflect the actual secretion and action of unbound cortisol and may be different across studies related to the hormone component that is bound to transport proteins.

In 2010, Ballerini and colleagues examined 38 premature infants who were divided by gestational age at birth (Ballerini, et al., 2010). Among infants born at < 32 weeks of gestation (age range 7-90 days), a median cortisol value of 4.8 was reported (range 1.0-

13.8), and among infants born at > 32 weeks of gestation (age range 3-45 days), a median cortisol value of 4.2 was reported (range 1.0-9.1). While Ballerini did not correct for

134 postmenstrual age, the cortisol values reported in her study serve to broadly affirm our values.

Table 5.1

Characteristics of Cortisol (ug/dl) at 37 Postmenstrual Weeks of Age: Comparisons with Other Researchers ______<31 Weeks Gestation <30 Weeks Gestation 0-24 months

Parameter Rohan Al Saedi Rohan Arnold Soldin n=26 n= 30 n=18 n=12

Range 1.4-24.6 nr 1.4-15.6 nr 1.0-34 Mean 6.6 nr 5.2 nr Median 4.6 2.8 3.9 1.2 SD 5.4 nr 4.5 nr

1st Quartile 2.4 nr 2.0 1.0 3rd Quartile 8.8 nr 8.3 3.2

10th Percentile 1.8 1.1 1.8 nr 90th Percentile 15.2 11.5 9.7 nr ______nr: not reported Conversion factor for cortisol: ug/dl cortisol x 27.59 = nmol/l cortisol (Al Saedi, et al., 1995; Arnold, et al., 1997; Soldin, et al., 2003)

The 17-OHP values in the current study sample were higher and with a greater spread than the results reported by Al Saedi, but lower than those reported by Arnold (Table

5.2). Small sample sizes and differences in the methods of serum analysis (Ballerini, et al., 2010) may have given rise to the differences in these reported values.

135

Table 5.2

Characteristics of 17-OHP (ng/dl) at 37 Postmenstrual Weeks of Age: Comparisons with Other Researchers ______<31 Weeks Gestation <30 Weeks Gestation

Parameter Rohan Al Saedi Rohan Arnold n=26 n=30 n=18 n=11

Range 57-650 nr 57-495 nr Mean 250 nr 215 nr Median 231 170 179 284 SD 144 nr 93 nr

1st Quartile 143 nr 129 238 3rd Quartile 318 nr 298 601

10th Percentile 77 97 57 nr 90th Percentile 447 292 430 nr ______nr: not reported Conversion factor for 17-OHP: ng/dl 17-OHP x 33 = nmol/l 17-OHP (Al Saedi, et al., 1995; Arnold, et al., 1997)

The findings from this study also support that cortisol values may be low in premature infants without signifying acute disease. In the current study sample, 15/59 (25.4%) infants had cortisol values under 3 ug/dl, and 8/59 (13.6%) had values under 2 ug/dl.

While acuity was not measured as a discrete variable, there were no infants in this study receiving invasive ventilation at 37 postmenstrual weeks and all were without intravenous devices at that time. This suggests a lack of acute disease at the time of laboratory testing.

136

Recurrent Exposure to Pain-Associated Procedures

The first aim of this study was to characterize the nature of recurrent pain-associated procedures for premature infants in the NICU from birth up until the 37th postmenstrual week of age. Relatively little research has examined the frequency of exposures to recurrent invasive procedures in premature infants during their NICU hospitalization.

This study adds to a growing body of work that examines the infant experience of recurrent pain-associated exposures relative to NICU hospitalization.

During this study, 59 infants were collectively exposed to 3354 pain-associated procedures, for an average of approximately 57 invasive procedures per infant (see Table

4.4). The greatest number of individual exposures was 238, and no infant was exposed to fewer than 20 procedures. The overwhelming burden of these exposures was during the first few days of age. Over 56% of exposures were in the first week after birth, and over

32% of exposures were during the first 48-72 hours of age.

The majority of the recurrent procedural exposures for this sample of premature infants were from heel lance and venipuncture (see Table 4.4). Heel lances are often considered relatively benign procedures in the NICU, however collectively comprised over 80% of the pain-associated events. The premature infants averaged exposure to approximately

24 heel lances during the first week of age. Venipunctures also comprised a substantial portion of the pain-associated exposures. The conservative method of retrospective chart review identified over 15 venipuncture events in the first week of age for some of the infants.

137

Previous studies have reported exposure to pain-associated procedures for premature infants during the course of hospitalization. Using retrospective chart review in which secondary attempts at primary procedures were routinely documented, Grunau and colleagues reported procedural exposure for three samples of infants. In these studies, there was a mean exposure during NICU hospitalization of 193, 204 and 129 procedures for infants born at a gestational age of 22 to 28 weeks, and a mean exposure of 72, 61 and

52 procedures for infants born at a gestational age of 29 to 32 weeks (Grunau, et al.,

2007; Grunau, et al., 2010; Grunau, et al., 2004).

Grunau’s findings provide data for reasonable comparisons. Although the current study examines procedural exposures only up until 37 postmenstrual weeks of age, and

Grunau’s study examines data from the entire NICU hospitalization, it appears that procedural exposure greatly diminishes with hospitalization and approaching discharge. It might be expected that there are relatively few procedures represented in Grunau’s data past the 37th postmenstrual week of age. In addition, it appears that many premature infants are discharged close to the 37th postmenstrual week of age. After removing infants who were born at over 32 weeks of gestation, comparisons were made between

Grunau’s study findings and the current study findings.

In the current study there was a mean total exposure to 138 procedures for infants born at a gestational age of 25 to 28 weeks (n=7), and a mean total exposure to 54 procedures for infants born at a gestational age of 29 to 32 weeks (n=33). These values are comparable to the exposures identified by Grunau. In the current study, however, counts of painful procedures did not routinely recognize secondary attempts at primary procedures. From

138

Carbajal, we know that secondary attempts at primary procedures considerably augment the overall number of exposures to painful procedures (Carbajal, et al., 2008). It is therefore suspected that the actual exposure in the study institution greatly exceeded the exposures reported by Grunau, suggesting differences in invasive procedure use across institutions. Presuming that the modern institutions in which these studies were conducted had similar infant outcomes, variability in practice implies that there may be different ways to achieve the similar outcome endpoints. There may be an opportunity to adjust practices to reduce painful exposures for premature infants in the NICU without affecting standard infant outcomes.

Recent Pain-Associated Procedures

Holsti and colleagues identified a correlation between cortisol values and the number of recent skin-breaking procedures in premature infants at 32 postmenstrual weeks of age, as well as a difference in ACTH values in infants who had been exposed to higher versus lower numbers of recent painful stressors (Holsti, et al., 2007). To follow up on these observations, the relationship between both cortisol and 17-OHP values and the number of recent exposures to pain-associated procedures was examined. This study did not identify a significant correlation between the number of recent exposures to pain- associated procedures and either cortisol (r=0.156; p=0.238) or 17-OHP (r= -0.148; p=0.265) values. In addition, after recoding the recent pain-associated procedures variable to a dichotomous variable, the cortisol and 17-OHP values of infants who had been exposed to any pain-associated procedures in the previous 24 hours were compared to those of infants who had not been exposed to any such procedures (see Table 4.24).

No differences were found in either cortisol values (p=0.200) or 17-OHP values

139

(p=0.207) between these groups of infants. This research was unable to support Holsti’s proposition that pain exposure induces a proximate “down-regulation” of the HPA axis.

Only a small number of infants in the current study sample, however, had been exposed to recent pain-associated procedures (n=13/59), limiting the generalizability of findings.

Days of Age to Attain Birthweight

The correlation between days of age to attain birthweight and each of the stress variables

(recurrent pain-associated procedures, hours of assisted ventilation, hours of invasive ventilation and hours of non-invasive ventilation) was significant at the 0.01 level.

Further longitudinal studies would be necessary to understand if premature infants grow slowly when they are exposed to numerous pain-associated procedures, or if altered metabolism and slow growth is part of the illness condition which predisposes premature infants to higher levels of intensive care.

To this point, the association between extrauterine growth velocity and adrenocorticoid activity of premature infants has not been described. In our research, there was not a significant correlation between days of age to attain birthweight and either cortisol values

(r=0.083; p=0.531) or 17-OHP values (r=0.028; p=0.831). There remains a lack of evidence to support a relationship between this marker of infant resilience and these aspects of the adrenocorticoid profile in premature infants.

Male/Female Differences

For the subjects in this study, male sex was significantly correlated with higher birthweight, higher weight at 37 postmenstrual weeks of age, and lower incidence of severe acidosis. There was also a statistically significant difference in both birthweight

140 and 37-week weight between male and female infants, with males having a higher weight at both points. These findings prompted further exploration, as premature male infants have been reported to have higher morbidity (Tyson, Parikh, Langer, Green, & Higgins,

2008).

Further analysis revealed that the exclusionary criteria of transfer/surgery, postnatal steroid exposure and death, predominantly disqualified extremely low birthweight, male infants (77.8%, 7/9) from study participation. When the birthweights of the nine excluded infants are included in a bivariate analysis (n=68), there was no longer a statistically significant difference between males and females for birthweight (t=1.023, p=0.310). Data was not available to conduct similar bivariate analyses using 37-week weight and incidence of acidosis. While male infants included in the study have indicators of superior health when compared to the female infants (higher weight at birth, higher weight at 37 weeks, and lower incidence of severe acidosis), this must be considered in the larger context: Among the pool of infants who were cared for in the

NICU for over two weeks, there was primarily exclusion of male infants with complex health issues.

Previous research has both identified (Ballerini, et al., 2010; Grunau, et al., 2010) and failed to identify (Grunau, et al., 2007; Grunau, et al., 2005; Heckmann, et al., 2000;

Scott & Watterberg, 1995) differences in cortisol and 17-OHP values between male and female infants. In our sample we failed to identify differences in cortisol values

(p=0.660) or in 17-OHP values (p=0.704) between the male and female infants. Because

141 of different exclusionary criteria among these studies, it continues to be difficult to interpret these results in a clinically meaningful way.

Multicollinearity

Collinearity between study variables impacted certain statistical analyses. Many of these relationships were predictable. For example, there was multicollinearity between gestational age at birth, total hours of ventilation and procedural pain exposure. More immature infants have a higher incidence of disease at birth necessitating treatment with assisted ventilation and more diagnostic testing. Similarly, there was multicollinearity between gestational age at birth, birthweight and day of age at testing. More immature infants have a lower birthweight and a longer interval of NICU care between birth and 37 postmenstrual weeks of age. Except for birthweight (which only approached collinearity with recurrent pain-associated procedures, r= -0.672), this study’s findings of collinearity were similar to the findings of Grunau and colleagues, who identified multicollinearity between number of skin breaking procedures, birthweight, days of mechanical ventilation and gestational age (Grunau, et al., 2005; Grunau, et al., 2010; Grunau, et al., 2004;

Grunau, et al., 2009).

The relationship between procedural pain exposure and hours of invasive ventilation

(r= 0.685) was noted for a lack of collinearity and explored. Premature infants receiving assisted ventilation typically require more frequent laboratory monitoring, and are commonly restricted on enteral feedings, necessitating the maintenance of intravenous access. For infants requiring the highest levels of respiratory assistance - invasive ventilation - added assessments are indicated. While multicollinearity between

142 procedural pain exposure, total hours of ventilation and hours of non-invasive ventilation existed, the lack of collinearity between recurrent pain-associated procedures and hours of invasive ventilation was unexpected. It was anticipated that invasive ventilation would be highly collinear with procedural pain exposure.

A collinear relationship was instead identified between hours of invasive ventilation in the first week of age and suture placement in the first week of age (r=0.783; p<0.001).

The majority of hours of invasive ventilation occurred during the first week of age in this sample (see Table 4.7). In addition, all sutures placed in the first week of age were for the securement of central arterial and venous catheters (see Table 4.5). These deep catheters were used for blood draws and fluid infusions, precluding the need for many heelsticks and venipunctures. This significant relationship between suture placement and hours of invasive ventilation suggests that the use of central umbilical catheters during the first week of age in infants who require invasive ventilation serves to attenuate their exposure to recurrent pain-associated procedures.

Relationship between Study Variables and Adrenocorticoid Values

Hypothesis testing for this study did not identify a significant correlation between recurrent exposure to pain-associated procedures or between hours of assisted ventilation and cortisol values at 37 post-conceptual weeks of age. Similarly, a significant correlation could not be identified between hours of assisted ventilation and 17-OHP values at 37 post-conceptual weeks of age.

An important finding of this study was support of the hypothesis that recurrent exposure to pain-associated procedures was significantly negatively correlated with random 17-

143

OHP values at 37 postmenstrual weeks of age (r= -0.232; p=0.039). In addition, there was a trend for differences in 17-OHP values between infants with the highest and lowest recurrent pain-associated procedural exposure.

Due to multicollinearity, it was not possible to examine the independent effects of recurrent pain-associated procedures, gestational age, hours of ventilation and days of age on 17-OHP values. This was concerning, as gestational age (and birthweight) had been previously identified as important factors in the variability of 17-OHP values (J. Lee, et al., 2008a; Ryckman, et al., 2012). The extent to which the collinear variables and birthweight were related to 17-OHP, relative to the relationship between recurrent pain- associated procedures and 17-OHP, was instead examined using one-tailed correlation analyses. These analyses identified the strength of the negative linear relationships between the infant variables and 17-OHP at 37 postmenstrual weeks of age, and provided an opportunity to compare these relationships to the strength of the significant negative correlation between recurrent pain-associated procedures and 17-OHP (Table 4.23).

These exploratory analyses failed to identify any other significant linear relationships between the infant variables and 17-OHP. The strength of the linear relationship (and the level of significance) between 17-OHP and each of the study variables diminished successively after recurrent pain-associated procedures as follows: days of age (r= -0.140; p=0.145), hours of assisted ventilation (r= -0.132; p=0.159), hours of non-invasive ventilation (r= -0.127; p=0.168), hours of invasive ventilation (r= -0.071; p=0.296), gestational age (r= 0.045; p=0.369), and birthweight (r= 0.064; p=0.631). While limited to linear relationships, these one-tailed correlation analyses suggest that the factors of

144 birthweight and gestational age may not necessarily be more important than recurrent procedural exposure in explaining the variability of 17-OHP values. Although further research using larger samples is needed, exposure to recurrent pain-associated procedures has the potential for being an even more specific indicator of 17-OHP values at 37 postmenstrual weeks of age compared to other infant variables.

The extent to which birthweight was related to 17-OHP values was also examined in an exploratory regression analysis (see Table 4.22). In this multiple linear regression analysis of recurrent pain-associated procedures on 17-OHP values, the independent effect of birthweight on 17-OHP values was examined when it was used as a control variable in the model. The percent of variability accounted for in the prediction of 17-

OHP by birthweight was 1.8% which was not statistically significant (t= -0.020, p=0.984).

In 2004, Grunau examined the effect of neonatal factors on cortisol values at eight month of age and encountered a similar problem of collinearity between gestational age and procedural exposure (Grunau, et al., 2004). After controlling for illness severity, and days of supplemental oxygen, she identified that a higher number of neonatal procedures independently predicted basal cortisol values at 8 months of age. To evaluate whether gestational age rather than procedures was the operative factor, regression analysis was repeated with substitution of gestational age for procedural exposure. This analysis did not identify a relationship between gestational age and basal cortisol at 8 months of age.

Researchers concluded that procedural exposure appeared to be the key factor in influencing the adrenocorticoid phenotype, over the other neonatal variables.

145

It is important to consider that the relationship between the stressor variables and 17-

OHP may be quadratic, or otherwise nonlinear. The scatterplots depicting the relationships between the stressor and outcome variables were examined in this regard.

There were not any nonlinear relationships detected in this examination.

Recruitment and Sampling Issues

The overall consent rate for this study was 93.3%. Taking twinning into account, where one parent was able to consent for two subjects, a total of 96% of parents who were approached agreed to study enrollment for their infants. A prospective, multi-center

PICU study examining surrogate consent rates for pediatric critical care research reported a consent rate of 79.4% for non-therapeutic studies (Menon et al., 2012). Differences in consent processes, parental perceptions of illness and illness severity make it difficult to extrapolate findings to research in other settings.

Our high rate of consent for this study may be related to a pre-established parental acceptance of blood sampling as part of the existing NICU routine, or a function of the high levels of trust that parents report for nurses (and in this case, a nurse-investigator).

In addition, as previously described, this study was conducted in a setting where the investigator had served in a clinical role as a neonatal nurse practitioner. This long, positive clinical relationship may have enhanced staff support of parents who were considering participation in the research investigation.

The sampling methodologies and exclusion criteria used in this study resulted in an overrepresentation of infants born at higher gestational ages (see Table 4.2). It was hoped that conducting this study at a Regional Perinatal Center would result in a higher

146 representation of infants with younger gestational ages through the mechanism of infant transport. Instead of similar enrollment across gestational age categories, the gestational age distribution is more reflective of the data from the National Vital Statistics Report in which there is a negative correlation between birth rate and gestational age (Martin,

2011).

Twins are reported to represent 3.3% of all live births nationally (Kochanek, Kirmeyer,

Martin, Strobino, & Guyer, 2012), however are overrepresented among premature births.

Twinning affected 25.5% (12/47) of the pregnancies represented by our data such that

39.0% (23/59) of the subjects in our study were a twin (see Table 4.2).

Study Strengths

While the overall sample size was relatively small, the reported results represent one of the larger cohorts of infants examined for a relationship between stressor exposure and subsequent adrenocorticoid production, as well as one of the few NICU cohorts in which recurrent pain-associated stressor exposure has been quantified.

As previously mentioned, this study also has value because of its translational - rather than bench science - approach. Should future research support 17-OHP as a physiologic indicator for cumulative NICU stress, then support for its use as a practical biomarker is strengthened by knowledge of its feasible collection using routine clinical and laboratory procedures.

This study also represents an initial attempt to validate the life cycle model of stress

(Lupien, et al., 2009) for explaining the influence of stressful exposures on differentiation in glucocorticoid production at the developmentally-sensitive period of early infancy.

147

This model is important because it turns attention from the study of immediate biological and physiological responses to stress, which are relatively well described in infancy, and instead provides a framework for understanding the more lasting impact of stress adaptation.

Challenges and Limitations

Obtaining an accurate count of pain-associated procedures using a review of the medical record was limited by a deficiency of data describing supplemental attempts at primary skin-breaking procedures, likely reducing the count of recurrent pain-associated procedures to significantly less than the actual occurrence. The need for supplemental attempts at primary procedures was recognized by Carabjal and colleagues’ using direct observation techniques (Carbajal, et al., 2008). In their research, it was identified that over 35 percent of procedures performed in the NICU were supplemental attempts at a primary procedure, with some painful procedures needing as many 10 to 15 attempts for completion. These supplemental attempts were not found to be typically recorded in the medical record. Researchers have noted that these repeated attempts are therefore difficult to capture using chart review methods (Simons et al., 2003).

At the time of this study, records were not routinely maintained in the NICU regarding secondary attempts at primary procedures, although this information was occasionally available in a nursing “free text” note. The number of times that such documentation of secondary attempts at primary procedures occurred was in less than 0.5% of all procedural notations. In a setting with similar nursing documentation practices, chart

148 audit likely under-represents the actual number of recurrent pain-associated procedural exposures.

Medical record review is complicated by other documentation practices as well. All studies using medical record review are limited because of the potential for recording errors, and because records are only maintained for selected factors. The medical records reviewed for this study were primarily electronic. The study site converted to electronic charting in 2011, and training was still incomplete. In addition, for the electronic records, there were ongoing documentation process changes being implemented by the hospital’s

Computer Informatics Department throughout the course of the study. These changes, while aimed at improving the quality of the nursing documentation, may have resulted in a higher incidence of charting errors as nurses transcended a learning curve.

There are many methodological challenges in adrenal hormone research, particularly in studies of premature infants, in whom sampling can be difficult, and frequently yields inadequate amount of specimen. While the use of standard clinical and laboratory procedures may have enhanced the translational value of this study, the use of these methods may have resulted in a loss of precision in laboratory values. Similarly, while research has proposed that a single measurement of cortisol is adequate in reflecting the adrenal status of a premature infant, the results obtained in this study may have in some way been affected by the pulsatile nature of cortisol. Caution needs to be exercised when drawing conclusions about adrenal axis function based upon the premise of single measurement reliability.

149

The results of the analyses also need to be considered in the context of overall stress exposure for premature infants who are cared for in the NICU. While recurrent skin- breaking procedures and assisted ventilation comprise a considerable burden of stress for premature infants, there was not control for exposures to many other stressor variables.

Light, noise and hypothermia have all been identified as stressors for this population

(Bissinger & Annibale, 2010; Bremmer, Byers, & Kiehl, 2003; Perlman, 2001) and may contribute to the adrenocorticoid phenotype that evolves after NICU care. Prenatal stressors have additionally been implicated to impact offspring, even by contributing to premature delivery (Hobel, Goldstein, & Barrett, 2008). Cocaine exposure is an example of one such potentially HPA-axis altering prenatal exposure which was not examined in this study (Bauer et al., 2011; Scafidi et al., 1996). Conversely, the buffering effects of maternal caregiving, kangaroo holding and careful attention to developmental interventions may have selectively modified the impact of stressors for certain infants.

These variables were not measured.

The study sample was one of convenience, and results may for that reason not be generalizable to other NICU infant populations. For example, the regional transport center where this study was conducted is different from many other centers in that it covers a broad county with a sparse public transportation system. Premature birth at this hospital leaves many parents with NICU visitation difficulties. The center is also a state facility where a disproportionately high percentage of premature infant hospitalization costs are covered under the Medicaid system. A scarcity of financial resources exacerbates transportation difficulties for many families. It is plausible that differences in prenatal and family stressors, or in parental visitation practices, between the sample of

150 infants who were enrolled in this study and the general population of premature infants who are cared for in the NICU may have impacted the study results.

Finally, while a power analysis was conducted, it was based upon a moderate to large effect size for exploratory research. This resulted in a relatively small sample size, and raises concern for Type II error.

Implications for Practice

The information gained by this research adds to our ability to quantitatively describe the cumulative exposure to stressors experienced by premature infants in the NICU. In the course of this research, using counts of invasive procedures and hours of assisted ventilation, a more accurate representation of pain-associated stressor exposure has emerged. This information can potentially be used to develop exposure benchmarks for neonatal intensive care. As indicated, there may be differences in exposure rates between the study institution and other research centers. Examination of variability in practices across institutions provides a potential opportunity to reduce pain-associated procedural exposure for premature infants without compromising other outcome benchmarks.

Difficulty in obtaining an accurate count of pain-associated exposures points to an opportunity for improved documentation practices in the NICU. Incorporating fields into the electronic medical records to capture primary and secondary attempts at skin- breaking procedures for NICU patients may enhance quantification of this variable, and potentially improve upon our ability to identify the impact of these recurrent procedures.

Direct-observation techniques continue to be the gold standard for collecting clinical data such as hours of ventilation and procedural exposure for research purposes. These

151 methods are expensive and often not feasible. In addition to improving upon the methods of provider documentation, exploration of video or other newer electronic technologies for research purposes may improve reliability of such clinical measurements.

This study provides evidence that the use of central umbilical catheters during the first week of age in infants who require invasive ventilation may serve to attenuate their exposure to recurrent pain-associated procedures. The first week of age has been demonstrated to be a time when almost every premature infant in this study sustained the highest burden of recurrent pain-associated procedures. The risks of umbilical catheter placement have traditionally been balanced by factors such as provider convenience and data accuracy. The findings from this study suggest that, outside of the obligation to provide humane care, these catheters may serve to reduce an exposure which may have endocrinologic implications for premature infants. Guidelines for the placement of these catheters warrant review in the context of this new information.

Implications for Research

The results from this study provide reason to pursue further examination of the relationship between cumulative stressor exposure in premature infants and adaptations of their adrenocorticoid profile. There is a need to examine recurrent stressor exposure using measures affording greater accuracy, in larger samples of premature infants. More research is needed in samples including younger and sicker infants – particularly male infants - and ones with high stressor exposure if we are to gain meaningful knowledge about the effect of recurrent stressor exposure on the adrenocorticoid phenotype of

152 premature infants. The use of longitudinal studies to examine variability in these values through the first year of age would also be important.

As previously mentioned, using similar measures of procedural pain exposure, Grunau and colleagues identified a relationship between high procedural pain in premature infants who were cared for in the NICU and basal cortisol values at 8 and 18 months of age, as well as a relationship between high procedural pain exposure and diminished neuro-cognitive outcomes at 8 and 18 months of age. A relationship between high procedural pain exposure and 17-OHP values at 37 postmenstrual weeks in this population has now been identified. Research examining the relationship between 17-

OHP values at 37 postmenstrual weeks and neuro-cognitive outcomes at 8 and 18 months would be an important next step to consider. Newborn state screening data, which includes 17-OHP values, as well as data from early intervention programs and NICU follow-up clinics, are promising sources of information for such studies.

Reducing future health risks for premature infants is important. Early work seems to indicate that understanding the factors that can lead to altered adrenocortical resting and responsivity patterns in premature infants may be important for understanding their future problems in learning and development. Beyond the demonstration that one of the two major and prevalent types of stressors in premature infants is related to 17-OHP values at

37 postmenstrual weeks of age, future research is needed to explore the relationship between 17-OHP values and cumulative stressor exposure at 37 postmenstrual weeks of age and the long-term health and development of prematurely born children. The

153 enduring implications of early 17-OHP values and exposure to recurrent pain-associated procedures in the population of infants who are born prematurely remain to be elucidated.

Conclusions

This research extends what is known about factors influencing the adrenocorticoid phenotype of premature infants at 37 postmenstrual weeks of age. Few studies have examined cortisol values in the general population of premature infants in relation to pain-associated procedural exposure and none have looked at these values at 37 postmenstrual weeks of age in relation to procedural exposure. Furthermore, no studies could be identified that have examined 17-OHP values in relation to pain-associated procedural exposure. This study succeeded in showing that there was a relationship between recurrent pain-associated procedural exposure and 17-OHP values in premature infants, although multicollinearity limited analysis and exposure measurements may have been underestimated. The information gained by this research can nevertheless be used to inform future work on pain-associated stress and HPA axis adaptations in premature infants.

Recurrent exposure to pain-associated procedures may diminish the production of 17-

OHP in premature infants, and has previously been shown to be negatively correlated with early developmental milestones. Even though recurrent stressful exposures in NICU reflect efforts to bring premature infants to health, these stressful procedures may contribute to some of the developmental difficulties observed in later childhood. While more research is needed to clarify these early findings, it is important to consider that recurring pain-associated procedures, which are a customary aspect of intensive care,

154 may cause lasting changes in the endocrinologic function of some of the patients in whom we are aiming to restore health.

Until further evidence emerges, the impact of recurrent skin breaking procedures should not be presumed as transient by caretakers in the NICU. In order to do more for our smallest and youngest patients, we are beginning to find ourselves examining how we can do less to these patients, to attenuate their high cumulative experience of pain-associated stress.

155

References

Aarnoudse-Moens, C. S., Smidts, D. P., Oosterlaan, J., Duivenvoorden, H. J., & Weisglas-Kuperus, N. (2009). Executive function in very preterm children at early school age. J Abnorm Child Psychol, 37(7), 981-993. doi: 10.1542/peds.2008-2816

Aarnoudse-Moens, C. S., Weisglas-Kuperus, N., van Goudoever, J. B., & Oosterlaan, J. (2009). Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics, 124(2), 717-728.

Al Saedi, S., Dean, H., Dent, W., & Cronin, C. (1995). Reference ranges for serum cortisol and 17-hydroxyprogesterone levels in preterm infants. J Pediatr, 126(6), 985-987.

Anand, K. J. (2000). Effects of perinatal pain and stress. Prog Brain Res, 122, 117-129.

Anand, K. J., Barton, B. A., McIntosh, N., Lagercrantz, H., Pelausa, E., Young, T. E., & Vasa, R. (1999). Analgesia and sedation in preterm neonates who require ventilatory support: results from the NOPAIN trial. Neonatal Outcome and Prolonged Analgesia in Neonates. Arch Pediatr Adolesc Med, 153(4), 331-338.

Anand, K. J., & Carr, D. B. (1989). The neuroanatomy, neurophysiology, and neurochemistry of pain, stress, and analgesia in newborns and children. Pediatr Clin North Am, 36(4), 795-822.

Anand, K. J., Hall, R. W., Desai, N., Shephard, B., Bergqvist, L. L., Young, T. E., . . . Barton, B. A. (2004). Effects of morphine analgesia in ventilated preterm neonates: primary outcomes from the NEOPAIN randomised trial. Lancet, 363(9422), 1673-1682. doi: 10.1016/S0140-6736(04)16251-X

Anand, K. J., & Hickey, P. R. (1987). Pain and its effects in the human neonate and fetus. N Engl J Med, 317(21), 1321-1329.

Anand, K. J., & Hickey, P. R. (1992). Halothane-morphine compared with high-dose sufentanil for anesthesia and postoperative analgesia in neonatal cardiac surgery. N Engl J Med, 326(1), 1-9.

156

Ananth, C. V. (2007). Menstrual versus clinical estimate of gestational age dating in the United States: temporal trends and variability in indices of perinatal outcomes. Paediatr Perinat Epidemiol, 21 Suppl 2, 22-30. doi: 10.1111/j.1365- 3016.2007.00858.x

Anderson, P., & Doyle, L. W. (2003). Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. JAMA, 289(24), 3264-3272.

Anderson, P. J., & Doyle, L. W. (2008). Cognitive and educational deficits in children born extremely preterm. Semin Perinatol, 32(1), 51-58.

Aranda, J., Carlo, W., Hummel, P., Thomas, R., Lehr, V., & Anand, K. (2005). Analgesia and sedation during mechanical ventilation in neonates. Clin Ther, 27(6), 877- 899.

Arnold, J., Leslie, G., Bowen, J., Watters, S., Kreutzmann, D., & Silink, M. (1997). Longitudinal study of plasma cortisol and 17-hydroxyprogesterone in very-low- birth-weight infants during the first 16 weeks of life. Biol Neonate, 72(3), 148- 155.

Aylward, G. P. (2003). Cognitive function in preterm infants: no simple answers. JAMA, 289(6), 752-753. doi: jed30000 [pii]

Ballard, J. L., Khoury, J. C., Wedig, K., Wang, L., Eilers-Walsman, B. L., & Lipp, R. (1991). New Ballard Score, expanded to include extremely premature infants. J Pediatr, 119(3), 417-423.

Ballerini, M. G., Chiesa, A., Scaglia, P., Gruneiro-Papendieck, L., Heinrich, J. J., & Ropelato, M. G. (2010). 17alpha-hydroxyprogesterone and cortisol serum levels in neonates and young children: influence of age, gestational age, gender and methodological procedures. J Pediatr Endocrinol Metab, 23(1-2), 121-132.

Barker, D. J. (2002). Fetal programming of coronary heart disease. Trends Endocrinol Metab, 13(9), 364-368.

Barker, D. J. (2007). The origins of the developmental origins theory. J Intern Med, 261(5), 412-417. doi: 10.1111/j.1365-2796.2007.01809.x

157

Barker, D. J., Eriksson, J. G., Forsen, T., & Osmond, C. (2002). Fetal origins of adult disease: strength of effects and biological basis. Int J Epidemiol, 31(6), 1235- 1239.

Barker, D. P., & Rutter, N. (1995). Exposure to invasive procedures in neonatal intensive care unit admissions. Arch Dis Child Fetal Neonatal Ed, 72(1), F47-48.

Bauer, C. R., Lambert, B. L., Bann, C. M., Lester, B. M., Shankaran, S., Bada, H. S., . . . Higgins, R. D. (2011). Long-term impact of maternal substance use during pregnancy and extrauterine environmental adversity: stress hormone levels of preadolescent children. Pediatr Res, 70(2), 213-219. doi: 10.1203/PDR.0b013e3182226e26

Beishuizen, A., Thijs, L. G., & Vermes, I. (2001). Patterns of corticosteroid-binding globulin and the free cortisol index during septic shock and multitrauma. Intensive Care Med, 27(10), 1584-1591. doi: 10.1007/s001340101073

Bettendorf, M., Albers, N., Bauer, J., Heinrich, U. E., Linderkamp, O., & Maser-Gluth, C. (1998). Longitudinal evaluation of salivary cortisol levels in full-term and preterm neonates. Horm Res, 50(6), 303-308.

Bissinger, R. L., & Annibale, D. J. (2010). Thermoregulation in very low-birth-weight infants during the golden hour: results and implications. Adv Neonatal Care, 10(5), 230-238. doi: 10.1097/ANC.0b013e3181f0ae63

Bolt, R. J., van Weissenbruch, M. M., Lafeber, H. N., & Delemarre-van de Waal, H. A. (2002). Development of the hypothalamic-pituitary-adrenal axis in the fetus and preterm infant. J Pediatr Endocrinol Metab, 15(6), 759-769.

Bolt, R. J., van Weissenbruch, M. M., Popp-Snijders, C., Sweep, C. G., Lafeber, H. N., & Delemarre-van de Waal, H. A. (2002a). Fetal growth and the function of the adrenal cortex in preterm infants. J Clin Endocrinol Metab, 87(3), 1194-1199.

Bolt, R. J., Van Weissenbruch, M. M., Popp-Snijders, C., Sweep, F. G., Lafeber, H. N., & Delemarre-van de Waal, H. A. (2002b). Maturity of the adrenal cortex in very preterm infants is related to gestational age. Pediatr Res, 52(3), 405-410.

158

Bonner, K. M., & Mainous, R. O. (2008). The nursing care of the infant receiving bubble CPAP therapy. Adv Neonatal Care, 8(2), 78-95; quiz 96-77. doi: 10.1097/01.ANC.0000317256.76201.72

Botting, N., Powls, A., Cooke, R. W., & Marlow, N. (1997). Attention deficit hyperactivity disorders and other psychiatric outcomes in very low birthweight children at 12 years. J Child Psychol Psychiatry, 38(8), 931-941.

Botting, N., Powls, A., Cooke, R. W., & Marlow, N. (1998). Cognitive and educational outcome of very-low-birthweight children in early adolescence. Dev Med Child Neurol, 40(10), 652-660.

Boyer, K., Johnston, C., Walker, C. D., Filion, F., & Sherrard, A. (2004). Does sucrose analgesia promote physiologic stability in preterm neonates? Biol Neonate, 85(1), 26-31. doi: 10.1159/000074954

Boyle, E. M., Freer, Y., Wong, C. M., McIntosh, N., & Anand, K. J. (2006). Assessment of persistent pain or distress and adequacy of analgesia in preterm ventilated infants. Pain, 124(1-2), 87-91. doi: 10.1016/j.pain.2006.03.019

Bremmer, P., Byers, J. F., & Kiehl, E. (2003). Noise and the premature infant: physiological effects and practice implications. J Obstet Gynecol Neonatal Nurs, 32(4), 447-454.

Carbajal, R., Rousset, A., Danan, C., Coquery, S., Nolent, P., Ducrocq, S., . . . Breart, G. (2008). Epidemiology and Treatment of Painful Procedures in Neonates in Intensive Care Units. JAMA, 300(1), 60-70. doi: 10.1001/jama.300.1.60

Carrion, V. G., Weems, C. F., & Reiss, A. L. (2007). Stress predicts brain changes in children: a pilot longitudinal study on youth stress, posttraumatic stress disorder, and the hippocampus. Pediatrics, 119(3), 509-516. doi: 10.1542/peds.2006-2028

Cianfarani, S., Geremia, C., Scott, C. D., & Germani, D. (2002). Growth, IGF system, and cortisol in children with intrauterine growth retardation: is catch-up growth affected by reprogramming of the hypothalamic-pituitary-adrenal axis? Pediatr Res, 51(1), 94-99. doi: 10.1203/00006450-200201000-00017

Cignacco, E., Denhaerynck, K., Nelle, M., Buhrer, C., & Engberg, S. (2009). Variability in pain response to a non-pharmacological intervention across repeated routine

159

pain exposure in preterm infants: a feasibility study. Acta Paediatr, 98(5), 842- 846. doi: 10.1111/j.1651-2227.2008.01203.x

Cignacco, E., Hamers, J., van Lingen, R. A., Stoffel, L., Buchi, S., Muller, R., . . . Nelle, M. (2009). Neonatal procedural pain exposure and pain management in ventilated preterm infants during the first 14 days of life. Swiss Med Wkly, 139(15-16), 226- 232.

Cignacco, E., Hamers, J. P., Stoffel, L., van Lingen, R. A., Schutz, N., Muller, R., . . . Nelle, M. (2008). Routine procedures in NICUs: factors influencing pain assessment and ranking by pain intensity. Swiss Med Wkly, 138(33-34), 484-491. doi: 2008/33/smw-12147

Clark, P. M. (1998). Programming of the hypothalamo-pituitary-adrenal axis and the fetal origins of adult disease hypothesis. Eur J Pediatr, 157 Suppl 1, S7-10.

Cole, C. H., Shah, B., Abbasi, S., Demissie, S., MacKinnon, B., Colton, T., & Frantz, I. D., 3rd. (1999). Adrenal function in premature infants during inhaled beclomethasone therapy. J Pediatr, 135(1), 65-70. doi: S0022-3476(99)70329-X [pii]

Cutfield, W. S., Regan, F. A., Jackson, W. E., Jefferies, C. A., Robinson, E. M., Harris, M., & Hofman, P. L. (2004). The endocrine consequences for very low birth weight premature infants. Growth Horm IGF Res, 14 Suppl A, S130-135. doi: 10.1016/j.ghir.2004.03.028

Deoni, S. C., Mercure, E., Blasi, A., Gasston, D., Thomson, A., Johnson, M., . . . Murphy, D. G. (2011). Mapping infant brain myelination with magnetic resonance imaging. J Neurosci, 31(2), 784-791. doi: 10.1523/JNEUROSCI.2106- 10.2011

Dodds, E. (2003). Neonatal procedural pain: a survey of nursing staff. Paediatr Nurs, 15(5), 18-21.

Doerr, H. G., Versmold, H. T., Bidlingmaier, F., & Sippell, W. G. (1989). Adrenocortical steroids in small-for-gestational-age term infants during the early neonatal period. Pediatr Res, 25(2), 115-118.

160

Dowd, J. B., Simanek, A. M., & Aiello, A. E. (2009). Socio-economic status, cortisol and allostatic load: a review of the literature. Int J Epidemiol, 38(5), 1297-1309. doi: 10.1093/ije/dyp277

Economou, G., Andronikou, S., Challa, A., Cholevas, V., & Lapatsanis, P. D. (1993). Cortisol secretion in stressed babies during the neonatal period. Horm Res, 40(5- 6), 217-221.

Edwards, J., Berube, M., Erlandson, K., Haug, S., Johnstone, H., Meagher, M., . . . Zwicker, J. G. (2011). Developmental coordination disorder in school-aged children born very preterm and/or at very low birth weight: a systematic review. J Dev Behav Pediatr, 32(9), 678-687. doi: 10.1097/DBP.0b013e31822a396a

Ehrenkranz, R. A., Dusick, A. M., Vohr, B. R., Wright, L. L., Wrage, L. A., & Poole, W. K. (2006). Growth in the neonatal intensive care unit influences neurodevelopmental and growth outcomes of extremely low birth weight infants. Pediatrics, 117(4), 1253-1261. doi: 10.1542/peds.2005-1368

Ersch, J., Beinder, E., Stallmach, T., Bucher, H. U., & Torresani, T. (2008). 17- Hydroxyprogesterone in premature infants as a marker of intrauterine stress. J Perinat Med, 36(2), 157-160. doi: 10.1515/JPM.2008.013

Esteban, N. V., & Yergey, A. L. (1990). Cortisol production rates measured by liquid chromatography/mass spectrometry. Steroids, 55(4), 152-158. doi: 0039- 128X(90)90103-I [pii]

Fenton, T. R. (2003). A new growth chart for preterm babies: Babson and Benda's chart updated with recent data and a new format. BMC Pediatr, 3, 13. doi: 10.1186/1471-2431-3-13

Fernandez, E. F., Montman, R., & Watterberg, K. L. (2008). ACTH and cortisol response to critical illness in term and late preterm newborns. J Perinatol, 28(12), 797-802. doi: 10.1038/jp.2008.190

Fernandez, E. F., Montman, R., & Watterberg, K. L. (2010). Adrenal function in newborns undergoing surgery. J Perinatol, 30(12), 814-818. doi: 10.1038/jp.2010.44

161

Fernandez, E. F., & Watterberg, K. L. (2009). Relative adrenal insufficiency in the preterm and term infant. J Perinatol, 29 Suppl 2, S44-49. doi: 10.1038/jp.2009.24

Franck, L. S., & Miaskowski, C. (1997). Measurement of neonatal responses to painful stimuli: a research review. J Pain Symptom Manage, 14(6), 343-378.

Franck, L. S., Ridout, D., Howard, R., Peters, J., & Honour, J. W. (2011). A comparison of pain measures in newborn infants after cardiac surgery. Pain. doi: 10.1016/j.pain.2011.03.017

Franz, A. R., Pohlandt, F., Bode, H., Mihatsch, W. A., Sander, S., Kron, M., & Steinmacher, J. (2009). Intrauterine, early neonatal, and postdischarge growth and neurodevelopmental outcome at 5.4 years in extremely preterm infants after intensive neonatal nutritional support. Pediatrics, 123(1), e101-109. doi: 10.1542/peds.2008-1352

Gaskell, S. J., & Sieckmann, L. (1986). Mass-spectrometric determination of serum cortisol: comparison of data from two independent laboratories. Clin Chem, 32(3), 536-538.

Glover, V., & Fisk, N. M. (1999). Fetal pain: implications for research and practice. Br J Obstet Gynaecol, 106(9), 881-886.

Glover, V., Miles, R., Matta, S., Modi, N., & Stevenson, J. (2005). Glucocorticoid exposure in preterm babies predicts saliva cortisol response to immunization at 4 months. Pediatr Res, 58(6), 1233-1237. doi: 10.1203/01.pdr.0000185132.38209.73

Glover, V., O'Connor, T. G., & O'Donnell, K. (2010). Prenatal stress and the programming of the HPA axis. Neurosci Biobehav Rev, 35(1), 17-22. doi: 10.1016/j.neubiorev.2009.11.008

Granja, C., Lopes, A., Moreira, S., Dias, C., Costa-Pereira, A., & Carneiro, A. (2005). Patients' recollections of experiences in the intensive care unit may affect their quality of life. Crit Care, 9(2), R96-109. doi: 10.1186/cc3026

Greenspan, F. S., & Gardner, D. G. (2004). Basic & clinical endocrinology (7th ed.). New York: McGraw-Hill.

162

Grossbach, I., Chlan, L., & Tracy, M. F. (2011). Overview of mechanical ventilatory support and management of patient- and ventilator-related responses. Crit Care Nurse, 31(3), 30-44. doi: 10.4037/ccn2011595

Grunau, R. E. (2002). Early pain in preterm infants. A model of long-term effects. Clin Perinatol, 29(3), 373-394, vii-viii.

Grunau, R. E., Haley, D. W., Whitfield, M. F., Weinberg, J., Yu, W., & Thiessen, P. (2007). Altered basal cortisol levels at 3, 6, 8 and 18 months in infants born at extremely low gestational age. J Pediatr, 150(2), 151-156. doi: 10.1016/j.jpeds.2006.10.053

Grunau, R. E., Holsti, L., Haley, D. W., Oberlander, T., Weinberg, J., Solimano, A., . . . Yu, W. (2005). Neonatal procedural pain exposure predicts lower cortisol and behavioral reactivity in preterm infants in the NICU. Pain, 113(3), 293-300. doi: 10.1016/j.pain.2004.10.020

Grunau, R. E., Holsti, L., & Peters, J. (2006). Long-term consequences of pain in human neonates. Semin Fetal Neonatal Med, 11(4), 268-275.

Grunau, R. E., Tu, M. T., Whitfield, M. F., Oberlander, T. F., Weinberg, J., Yu, W., . . . Scheifele, D. (2010). Cortisol, behavior, and heart rate reactivity to immunization pain at 4 months corrected age in infants born very preterm. Clin J Pain, 26(8), 698-704. doi: 10.1097/AJP.0b013e3181e5bb00

Grunau, R. E., Weinberg, J., & Whitfield, M. F. (2004). Neonatal procedural pain and preterm infant cortisol response to novelty at 8 months. Pediatrics, 114(1), e77- 84.

Grunau, R. E., Whitfield, M. F., Petrie-Thomas, J., Synnes, A. R., Cepeda, I. L., Keidar, A., . . . Johannesen, D. (2009). Neonatal pain, parenting stress and interaction, in relation to cognitive and motor development at 8 and 18 months in preterm infants. Pain, 143(1-2), 138-146.

Grunau, R. E., Whitfield, M. F., & Petrie, J. H. (1994). Pain sensitivity and temperament in extremely low-birth-weight premature toddlers and preterm and full-term controls. Pain, 58(3), 341-346.

163

Gunnar, M. R. (1989). Studies of the human infant's adrenocortical response to potentially stressful events. New Dir Child Dev(45), 3-18.

Gunnar, M. R. (1992). Reactivity of the hypothalamic-pituitary-adrenocortical system to stressors in normal infants and children. Pediatrics, 90(3 Pt 2), 491-497.

Gunnar, M. R., Brodersen, L., Krueger, K., & Rigatuso, J. (1996). Dampening of adrenocortical responses during infancy: normative changes and individual differences. Child Dev, 67(3), 877-889.

Gunnar, M. R., Connors, J., & Isensee, J. (1989). Lack of stability in neonatal adrenocortical reactivity because of rapid habituation of the adrenocortical response. Dev Psychobiol, 22(3), 221-233. doi: 10.1002/dev.420220304

Gunnar, M. R., Connors, J., Isensee, J., & Wall, L. (1988). Adrenocortical activity and behavioral distress in human newborns. Dev Psychobiol, 21(4), 297-310. doi: 10.1002/dev.420210402

Gunnar, M. R., Fisch, R. O., Korsvik, S., & Donhowe, J. M. (1981). The effects of circumcision on serum cortisol and behavior. Psychoneuroendocrinology, 6(3), 269-275.

Gunnar, M. R., Fisch, R. O., & Malone, S. (1984). The effects of a pacifying stimulus on behavioral and adrenocortical responses to circumcision in the newborn. J Am Acad Child Psychiatry, 23(1), 34-38.

Gunnar, M. R., Frenn, K., Wewerka, S. S., & Van Ryzin, M. J. (2009). Moderate versus severe early life stress: associations with stress reactivity and regulation in 10-12- year-old children. Psychoneuroendocrinology, 34(1), 62-75. doi: 10.1016/j.psyneuen.2008.08.013

Gunnar, M. R., Hertsgaard, L., Larson, M., & Rigatuso, J. (1991). Cortisol and behavioral responses to repeated stressors in the human newborn. Dev Psychobiol, 24(7), 487-505. doi: 10.1002/dev.420240704

Gunnar, M. R., Isensee, J., & Fust, L. S. (1987). Adrenocortical activity and the Brazelton Neonatal Assessment Scale: moderating effects of the newborn's biomedical status. Child Dev, 58(6), 1448-1458.

164

Gunnar, M. R., Larson, M. C., Hertsgaard, L., Harris, M. L., & Brodersen, L. (1992). The stressfulness of separation among nine-month-old infants: effects of social context variables and infant temperament. Child Dev, 63(2), 290-303.

Gunnar, M. R., Malone, S., Vance, G., & Fisch, R. O. (1985). Coping with aversive stimulation in the neonatal period: quiet sleep and plasma cortisol levels during recovery from circumcision. Child Dev, 56(4), 824-834.

Gunnar, M. R., & Quevedo, K. (2007). The neurobiology of stress and development. Annu Rev Psychol, 58, 145-173. doi: 10.1146/annurev.psych.58.110405.085605

Gunnar, M. R., Talge, N. M., & Herrera, A. (2009). Stressor paradigms in developmental studies: What does and does not work to produce mean increases in salivary cortisol. Psychoneuroendocrinology, 34(7), 953-967. doi: 10.1016/j.psyneuen.2009.02.010

Hack, M., Flannery, D. J., Schluchter, M., Cartar, L., Borawski, E., & Klein, N. (2002). Outcomes in young adulthood for very low birth-weight infants. New England Journal of Medicine, 346(3), 149-157.

Hack, M., Taylor, H. G., Drotar, D., Schluchter, M., Cartar, L., Andreias, L., . . . Klein, N. K. (2005). Chronic conditions, functional limitations, and special health care needs of school-aged children born with extremely low birth weight in the 1990s. JAMA, 294(3), 318-325.

Haley, D. W., Weinberg, J., & Grunau, R. E. (2006). Cortisol, contingency learning, and memory in preterm and full-term infants. Psychoneuroendocrinology, 31(1), 108- 117. doi: 10.1016/j.psyneuen.2005.06.007

Hall, R. W., Boyle, E., & Young, T. (2007). Do ventilated neonates require pain management? Semin Perinatol, 31(5), 289-297. doi: 10.1053/j.semperi.2007.07.002

Haney, C., & Allingham, T. M. (1992). Nursing care of the neonate receiving high- frequency jet ventilation. J Obstet Gynecol Neonatal Nurs, 21(3), 187-195.

Hanna, C. E., Jett, P. L., Laird, M. R., Mandel, S. H., LaFranchi, S. H., & Reynolds, J. W. (1997). Corticosteroid binding globulin, total serum cortisol, and stress in

165

extremely low-birth-weight infants. Am J Perinatol, 14(4), 201-204. doi: 10.1055/s-2007-994127

Heckmann, M., Hartmann, M. F., Kampschulte, B., Gack, H., Bodeker, R. H., Gortner, L., & Wudy, S. A. (2005). Cortisol production rates in preterm infants in relation to growth and illness: a noninvasive prospective study using gas chromatography- mass spectrometry. J Clin Endocrinol Metab, 90(10), 5737-5742. doi: 10.1210/jc.2005-0870

Heckmann, M., Wudy, S. A., Haack, D., & Pohlandt, F. (1999). Reference range for serum cortisol in well preterm infants. Arch Dis Child Fetal Neonatal Ed, 81(3), F171-174.

Heckmann, M., Wudy, S. A., Haack, D., & Pohlandt, F. (2000). Serum cortisol concentrations in ill preterm infants less than 30 weeks gestational age. Acta Paediatr, 89(9), 1098-1103.

Henry, S. L., Barzel, B., Wood-Bradley, R. J., Burke, S. L., Head, G. A., & Armitage, J. A. (2011). The developmental origins of obesity-related hypertension. Clin Exp Pharmacol Physiol. doi: 10.1111/j.1440-1681.2011.05579.x

Heron, M., Sutton, P., Xu, J., Ventura, S., Strobino, D., & Guyer, B. (2010). Annual summary of vital statistics: 2007. Pediatrics, 125(1), 4-15.

Hertsgaard, L., Gunnar, M., Larson, M., Brodersen, L., & Lehman, H. (1992). First time experiences in infancy: when they appear to be pleasant, do they activate the adrenocortical stress response? Dev Psychobiol, 25(5), 319-333. doi: 10.1002/dev.420250503

Hobel, C. J., Goldstein, A., & Barrett, E. S. (2008). Psychosocial stress and pregnancy outcome. Clin Obstet Gynecol, 51(2), 333-348. doi: 10.1097/GRF.0b013e31816f2709

Hofman, P. L., Regan, F., Jackson, W. E., Jefferies, C., Knight, D. B., Robinson, E. M., & Cutfield, W. S. (2004). Premature birth and later insulin resistance. N Engl J Med, 351(21), 2179-2186. doi: 10.1056/NEJMoa042275

Holsti, L., Grunau, R. E., Oberlander, T. F., Whitfield, M. F., & Weinberg, J. (2005). Body movements: an important additional factor in discriminating pain from

166

stress in preterm infants. Clin J Pain, 21(6), 491-498. doi: 00002508-200511000- 00005 [pii]

Holsti, L., Weinberg, J., Whitfield, M. F., & Grunau, R. E. (2007). Relationships between adrenocorticotropic hormone and cortisol are altered during clustered nursing care in preterm infants born at extremely low gestational age. Early Hum Dev, 83(5), 341-348.

Hughes, D., Murphy, J. F., Dyas, J., Robinson, J. A., Riad-Fahmy, D., & Hughes, I. A. (1987). Blood spot glucocorticoid concentrations in ill preterm infants. Arch Dis Child, 62(10), 1014-1018.

Huysman, M. W., Hokken-Koelega, A. C., De Ridder, M. A., & Sauer, P. J. (2000). Adrenal function in sick very preterm infants. Pediatr Res, 48(5), 629-633.

Jacobson, L., & Sapolsky, R. (1991). The role of the hippocampus in feedback regulation of the hypothalamic-pituitary-adrenocortical axis. Endocr Rev, 12(2), 118-134.

Jett, P. L., Samuels, M. H., McDaniel, P. A., Benda, G. I., Lafranchi, S. H., Reynolds, J. W., & Hanna, C. E. (1997). Variability of plasma cortisol levels in extremely low birth weight infants. J Clin Endocrinol Metab, 82(9), 2921-2925.

Johnson, S., Wolke, D., Hennessy, E., & Marlow, N. (2011). Educational outcomes in extremely preterm children: neuropsychological correlates and predictors of attainment. Dev Neuropsychol, 36(1), 74-95. doi: 10.1080/87565641.2011.540541

Jongbloed-Pereboom, M., Janssen, A. J., Steenbergen, B., & Nijhuis-van der Sanden, M. W. (2012). Motor learning and working memory in children born preterm: a systematic review. Neurosci Biobehav Rev, 36(4), 1314-1330. doi: 10.1016/j.neubiorev.2012.02.005

Kanwaljeet, J. S., Anand, K. J., Aranda, J. V., Berde, C. B., Buckman, S., Capparelli, E. V., . . . Walco, G. A. (2006). Summary proceedings from the neonatal pain- control group. Pediatrics, 117(3 Pt 2), S9-S22. doi: 10.1542/peds.2005-0620C

Kari, M. A., Heinonen, K., Ikonen, R. S., Koivisto, M., & Raivio, K. O. (1993). Dexamethasone treatment in preterm infants at risk for bronchopulmonary dysplasia. Arch Dis Child, 68(5 Spec No), 566-569.

167

Kari, M. A., Raivio, K. O., Stenman, U. H., & Voutilainen, R. (1996). Serum cortisol, dehydroepiandrosterone sulfate, and steroid-binding globulins in preterm neonates: effect of gestational age and dexamethasone therapy. Pediatr Res, 40(2), 319-324.

Kaye, C. I., Accurso, F., La Franchi, S., Lane, P. A., Hope, N., Sonya, P., . . . Michele, A. L. (2006). Newborn screening fact sheets. Pediatrics, 118(3), e934-963. doi: 10.1542/peds.2006-1783

Kochanek, K. D., Kirmeyer, S. E., Martin, J. A., Strobino, D. M., & Guyer, B. (2012). Annual summary of vital statistics: 2009. Pediatrics, 129(2), 338-348. doi: 10.1542/peds.2011-3435

Koolhaas, J. M., Bartolomucci, A., Buwalda, B., de Boer, S. F., Flugge, G., Korte, S. M., . . . Fuchs, E. (2011). Stress revisited: a critical evaluation of the stress concept. Neurosci Biobehav Rev, 35(5), 1291-1301. doi: 10.1016/j.neubiorev.2011.02.003

Korte, C., Styne, D., Merritt, T. A., Mayes, D., Wertz, A., & Helbock, H. J. (1996). Adrenocortical function in the very low birth weight infant: improved testing sensitivity and association with neonatal outcome. J Pediatr, 128(2), 257-263.

Korte, S. M., Koolhaas, J. M., Wingfield, J. C., & McEwen, B. S. (2005). The Darwinian concept of stress: benefits of allostasis and costs of allostatic load and the trade- offs in health and disease. Neurosci Biobehav Rev, 29(1), 3-38. doi: 10.1016/j.neubiorev.2004.08.009

Kostovic, I., & Judas, M. (2010). The development of the subplate and thalamocortical connections in the human foetal brain. Acta Paediatr, 99(8), 1119-1127. doi: 10.1111/j.1651-2227.2010.01811.x

Lacey, J. M., Minutti, C. Z., Magera, M. J., Tauscher, A. L., Casetta, B., McCann, M., . . . Matern, D. (2004). Improved specificity of newborn screening for congenital adrenal hyperplasia by second-tier steroid profiling using tandem mass spectrometry. Clin Chem, 50(3), 621-625. doi: 10.1373/clinchem.2003.027193

Lagercrantz, H., & Changeux, J. P. (2009). The emergence of human consciousness: from fetal to neonatal life. Pediatr Res, 65(3), 255-260. doi: 10.1203/PDR.0b013e3181973b0d

168

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159-174.

Langer, M., Modi, B. P., & Agus, M. (2006). Adrenal insufficiency in the critically ill neonate and child. Curr Opin Pediatr, 18(4), 448-453. doi: 10.1097/01.mop.0000236397.79580.85

Larson, M. C., Gunnar, M. R., & Hertsgaard, L. (1991). The effects of morning naps, car trips, and maternal separation on adrenocortical activity in human infants. Child Dev, 62(2), 362-372.

Larson, M. C., White, B. P., Cochran, A., Donzella, B., & Gunnar, M. (1998). Dampening of the cortisol response to handling at 3 months in human infants and its relation to sleep, circadian cortisol activity, and behavioral distress. Dev Psychobiol, 33(4), 327-337. doi: 10.1002/(SICI)1098- 2302(199812)33:4<327::AID-DEV4>3.0.CO;2-S [pii]

Latal-Hajnal, B., von Siebenthal, K., Kovari, H., Bucher, H. U., & Largo, R. H. (2003). Postnatal growth in VLBW infants: significant association with neurodevelopmental outcome. J Pediatr, 143(2), 163-170. doi: 10.1067/S0022- 3476(03)00243-9

Lecanuet, J. P., & Schaal, B. (1996). Fetal sensory competencies. Eur J Obstet Gynecol Reprod Biol, 68(1-2), 1-23.

Lee, J., Moon, Y., Lee, M., Jun, Y., Oh, K., & Choi, J. (2008a). Corrected 17-alpha- hydroxyprogesterone values adjusted by a scoring system for screening congenital adrenal hyperplasia in premature infants. Ann Clin Lab Sci, 38(3), 235-240.

Lee, J., Moon, Y., Lee, M., Jun, Y., Oh, K., & Choi, J. (2008b). Corrected 17-alpha- hydroxyprogesterone Values Adjusted by a Scoring System for Screening Congenital Adrenal Hyperplasia in Premature Infants. Annals of Clinical and Laboratory Science, 38(3), 235-240.

Lee, M., Rajagopalan, L., Berg, G., & Moshang, T. (1989). Serum adrenal steroid concentrations in premature infants. J Clin Endocrinol Metab, 69(6), 1133-1136.

169

Lee, S., Ralston, H., Drey, E., Partridge, J., & Rosen, M. (2005). Fetal pain: a systematic multidisciplinary review of the evidence. JAMA, 294(8), 947-954. doi: 10.1001/jama.294.8.947

Linder, N., Davidovitch, N., Kogan, A., Barzilai, A., Kuint, J., Mazkeret, R., & Sack, J. (1999). Longitudinal measurements of 17alpha-hydroxyprogesterone in premature infants during the first three months of life. Arch Dis Child Fetal Neonatal Ed, 81(3), F175-178.

Loeser, J. D., & Treede, R. D. (2008). The Kyoto protocol of IASP Basic Pain Terminology. Pain, 137(3), 473-477. doi: 10.1016/j.pain.2008.04.025

Lorenz, J., Paneth, N., Jetton, J., den Ouden, L., & Tyson, J. (2001). Comparison of management strategies for extreme prematurity in New Jersey and the Netherlands: outcomes and resource expenditure. Pediatrics, 108(6), 1269-1274.

Lowery, C. L., Hardman, M. P., Manning, N., Hall, R. W., Anand, K. J., & Clancy, B. (2007). Neurodevelopmental changes of fetal pain. Semin Perinatol, 31(5), 275- 282. doi: 10.1053/j.semperi.2007.07.004

Lupien, S. J., McEwen, B. S., Gunnar, M. R., & Heim, C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nat Rev Neurosci, 10(6), 434-445. doi: 10.1038/nrn2639

Malone, S. M., Gunnar, M. R., & Fisch, R. O. (1985). Adrenocortical and behavioral responses to limb restraint in human neonates. Dev Psychobiol, 18(5), 435-446. doi: 10.1002/dev.420180508

Martin, J. A. (2011). Preterm births - United States, 2007. MMWR Surveill Summ, 60 Suppl, 78-79.

Martins, P. G., & Procianoy, R. S. (2007). Cortisol and 17-alpha-hydroxy-progesterone levels in infants with refractory hypotension born at 30 weeks of gestation or less. Braz J Med Biol Res, 40(4), 577-582.

Mathews, T., MacDorman, M., & MacDorman, M. (2008). Infant Mortality Statistics from the 2005 Period Linked Birth/Infant Death Data Set. Retrieved from http://www.cdc.gov/nchs/data/nvsr/nvsr57/nvsr57_02.pdf.

170

McEwen, B. S. (2001). Plasticity of the hippocampus: adaptation to chronic stress and allostatic load. Ann N Y Acad Sci, 933, 265-277.

McEwen, B. S. (2007). Physiology and neurobiology of stress and adaptation: central role of the brain. Physiol Rev, 87(3), 873-904. doi: 10.1152/physrev.00041.2006

McEwen, B. S., & Seeman, T. (1999). Protective and damaging effects of mediators of stress. Elaborating and testing the concepts of allostasis and allostatic load. Ann N Y Acad Sci, 896, 30-47.

McEwen, B. S., & Stellar, E. (1993). Stress and the individual. Mechanisms leading to disease. Arch Intern Med, 153(18), 2093-2101.

McEwen, B. S., & Wingfield, J. C. (2003). The concept of allostasis in biology and biomedicine. Horm Behav, 43(1), 2-15.

McGrath, M. M., Sullivan, M., Devin, J., Fontes-Murphy, M., Barcelos, S., DePalma, J. L., & Faraone, S. (2005). Early precursors of low attention and hyperactivity in a preterm sample at age four. Issues Compr Pediatr Nurs, 28(1), 1-15. doi: 10.1080/01460860590913945

Menon, K., Ward, R. E., Gaboury, I., Thomas, M., Joffe, A., Burns, K., & Cook, D. (2012). Factors affecting consent in pediatric critical care research. Intensive Care Med, 38(1), 153-159. doi: 10.1007/s00134-011-2412-0

Merce Fernandez-Balsells, M., Muthusamy, K., Smushkin, G., Lampropulos, J. F., Elamin, M. B., Abu Elnour, N. O., . . . Murad, M. H. (2010). Prenatal dexamethasone use for the prevention of virilization in pregnancies at risk for classical congenital adrenal hyperplasia because of 21-hydroxylase (CYP21A2) deficiency: a systematic review and meta-analyses. Clin Endocrinol (Oxf), 73(4), 436-444. doi: 10.1111/j.1365-2265.2010.03826.x

Metzger, D. L., Wright, N. M., Veldhuis, J. D., Rogol, A. D., & Kerrigan, J. R. (1993). Characterization of pulsatile secretion and clearance of plasma cortisol in premature and term neonates using deconvolution analysis. J Clin Endocrinol Metab, 77(2), 458-463.

171

Midgley, P. C., Holownia, P., Smith, J., Moore, M., Russell, K., Oates, N., . . . Honour, J. W. (2001). Plasma cortisol, cortisone and urinary glucocorticoid metabolites in preterm infants. Biol Neonate, 79(2), 79-86.

Midgley, P. C., Russell, K., Oates, N., Holownia, P., Shaw, J. C., & Honour, J. W. (1998). Adrenal function in preterm infants: ACTH may not be the sole regulator of the fetal zone. Pediatr Res, 44(6), 887-893.

National Center for Chronic Disease Prevention and Health Promotion - Division of Reproductive Health. (2010). Premature Birth. Retrieved September 1, 2011, from http://www.cdc.gov/Features/PrematureBirth/

Newnham, C. A., Inder, T. E., & Milgrom, J. (2009). Measuring preterm cumulative stressors within the NICU: the Neonatal Infant Stressor Scale. Early Hum Dev, 85(9), 549-555. doi: 10.1016/j.earlhumdev.2009.05.002

Ng, P. C., Lam, C. W., Lee, C. H., Ma, K. C., Fok, T. F., Chan, I. H., & Wong, E. (2002). Reference ranges and factors affecting the human corticotropin-releasing hormone test in preterm, very low birth weight infants. J Clin Endocrinol Metab, 87(10), 4621-4628.

Ng, P. C., Wong, G. W., Lam, C. W., Lee, C. H., Fok, T. F., Wong, M. Y., & Ma, K. C. (1999). Effect of multiple courses of antenatal corticosteroids on pituitary-adrenal function in preterm infants. Arch Dis Child Fetal Neonatal Ed, 80(3), F213-216.

Ng, P. C., Wong, G. W., Lam, C. W., Lee, C. H., Fok, T. F., Wong, M. Y., . . . Chan, D. C. (1997). Pituitary-adrenal suppression and recovery in preterm very low birth weight infants after dexamethasone treatment for bronchopulmonary dysplasia. J Clin Endocrinol Metab, 82(8), 2429-2432.

Nordenstrom, A., Wedell, A., Hagenfeldt, L., Marcus, C., & Larsson, A. (2001). Neonatal screening for congenital adrenal hyperplasia: 17-hydroxyprogesterone levels and CYP21 genotypes in preterm infants. Pediatrics, 108(4), E68.

Novaes, M. A., Knobel, E., Bork, A. M., Pavao, O. F., Nogueira-Martins, L. A., & Ferraz, M. B. (1999). Stressors in ICU: perception of the patient, relatives and health care team. Intensive Care Med, 25(12), 1421-1426.

172

Nykanen, P., Heinonen, K., Riepe, F. G., Sippell, W. G., & Voutilainen, R. (2010). Serum concentrations of adrenal steroids and their precursors as a measure of maturity of adrenocortical function in very premature newborns. Horm Res Paediatr, 74(5), 358-364. doi: 10.1159/000314970

Ogundipe, E. M., Wolfe, C. D., Seed, P., & Gamsu, H. R. (2000). Does the antenatal detection of small-for-gestational-age babies influence their two-year outcomes? Am J Perinatol, 17(2), 73-81.

Perlman, J. M. (2001). Neurobehavioral deficits in premature graduates of intensive care- -potential medical and neonatal environmental risk factors. Pediatrics, 108(6), 1339-1348.

Phillips, D. I. (2001). Fetal growth and programming of the hypothalamic-pituitary- adrenal axis. Clin Exp Pharmacol Physiol, 28(11), 967-970.

Phillips, D. I. (2007). Programming of the stress response: a fundamental mechanism underlying the long-term effects of the fetal environment? J Intern Med, 261(5), 453-460. doi: 10.1111/j.1365-2796.2007.01801.x

Phillips, D. I., & Jones, A. (2006). Fetal programming of autonomic and HPA function: do people who were small babies have enhanced stress responses? J Physiol, 572(Pt 1), 45-50. doi: 10.1113/jphysiol.2005.104695

Phillips, D. I., & Matthews, S. G. (2011). Is perinatal neuroendocrine programming involved in the developmental origins of metabolic disorders? World J Diabetes, 2(12), 211-216. doi: 10.4239/wjd.v2.i12.211

Porter, F. L., Wolf, C. M., Gold, J., Lotsoff, D., & Miller, J. P. (1997). Pain and pain management in newborn infants: a survey of physicians and nurses. Pediatrics, 100(4), 626-632.

Porter, F. L., Wolf, C. M., & Miller, J. P. (1999). Procedural pain in newborn infants: the influence of intensity and development. Pediatrics, 104(1), e13.

Price, D. A., Close, G. C., & Fielding, B. A. (1983). Age of appearance of circadian rhythm in salivary cortisol values in infancy. Arch Dis Child, 58(6), 454-456.

173

Quinn, M. W., Otoo, F., Rushforth, J. A., Dean, H. G., Puntis, J. W., Wild, J., & Levene, M. I. (1992). Effect of morphine and pancuronium on the stress response in ventilated preterm infants. Early Hum Dev, 30(3), 241-248.

Rohrmeister, K., Kretzer, V., Berger, A., Haiden, N., Kohlhauser, C., & Pollak, A. (2003). Pain and stress management in the Neonatal Intensive Care Unit--a national survey in Austria. Wien Klin Wochenschr, 115(19-20), 715-719.

Ryckman, K. K., Cook, D. E., Berberich, S. L., Shchelochkov, O. A., Berends, S. K., Busch, T., . . . Murray, J. C. (2012). Replication of clinical associations with 17- hydroxyprogesterone in preterm newborns. J Pediatr Endocrinol Metab, 25(3-4), 301-305.

Saigal, S., Rosenbaum, P., Stoskopf, B., Hoult, L., Furlong, W., Feeny, D., . . . Torrance, G. (1994). Comprehensive assessment of the health status of extremely low birth weight children at eight years of age: comparison with a reference group. J Pediatr, 125(3), 411-417.

Saigal, S., Rosenbaum, P., Szatmari, P., & Campbell, D. (1991). Learning disabilities and school problems in a regional cohort of extremely low birth weight (less than 1000 G) children: a comparison with term controls. J Dev Behav Pediatr, 12(5), 294-300.

Saigal, S., Szatmari, P., Rosenbaum, P., Campbell, D., & King, S. (1991). Cognitive abilities and school performance of extremely low birth weight children and matched term control children at age 8 years: a regional study. J Pediatr, 118(5), 751-760.

Sari, F. N., Dizdar, E. A., Oguz, S. S., Andiran, N., Erdeve, O., Uras, N., . . . Dilmen, U. (2012). Baseline and stimulated cortisol levels in preterm infants: is there any clinical relevance? Horm Res Paediatr, 77(1), 12-18. doi: 10.1159/000332157

Scafidi, F. A., Field, T. M., Wheeden, A., Schanberg, S., Kuhn, C., Symanski, R., . . . Bandstra, E. S. (1996). Cocaine-exposed preterm neonates show behavioral and hormonal differences. Pediatrics, 97(6 Pt 1), 851-855.

Scott, S. M., & Watterberg, K. L. (1995). Effect of gestational age, postnatal age, and illness on plasma cortisol concentrations in premature infants. Pediatr Res, 37(1), 112-116.

174

Seckl, J. R., & Holmes, M. C. (2007). Mechanisms of disease: glucocorticoids, their placental metabolism and fetal 'programming' of adult pathophysiology. Nat Clin Pract Endocrinol Metab, 3(6), 479-488. doi: 10.1038/ncpendmet0515

Selye, H. (1976). Forty years of stress research: principal remaining problems and misconceptions. Can Med Assoc J, 115(1), 53-56.

Selye, H., & Horava, A. (1953). Stress Research. Science, 117(3045), 509. doi: 10.1126/science.117.3045.509

Simons, S., & Tibboel, D. (2006). Pain perception development and maturation. Semin Fetal Neonatal Med, 11(4), 227-231. doi: 10.1016/j.siny.2006.02.010

Simons, S., van Dijk, M., Anand, K., Roofthooft, D., van Lingen, R., & Tibboel, D. (2003). Do we still hurt newborn babies? A prospective study of procedural pain and analgesia in neonates. Arch Pediatr Adolesc Med, 157(11), 1058-1064.

Soldin, S. J., Brugnara, C., & Wong, E. C. (2003). Pediatric reference ranges (4th ed.). Washington, DC: AACC Press.

Stang, H. J., Gunnar, M. R., Snellman, L., Condon, L. M., & Kestenbaum, R. (1988). Local anesthesia for neonatal circumcision. Effects on distress and cortisol response. JAMA, 259(10), 1507-1511.

Stansbury, K., & Gunnar, M. R. (1994). Adrenocortical activity and emotion regulation. Monogr Soc Res Child Dev, 59(2-3), 108-134.

Stephens, B. E., Walden, R. V., Gargus, R. A., Tucker, R., McKinley, L., Mance, M., . . . Vohr, B. R. (2009). First-week protein and energy intakes are associated with 18- month developmental outcomes in extremely low birth weight infants. Pediatrics, 123(5), 1337-1343. doi: 10.1542/peds.2008-0211

Sterling, P., & Eyer, J. (1988). Allostasis: A new paradigm to explain arousal pathology. In S. Fisher & J. Reason (Eds.), Handbook of life stress, cognition and health (Vol. xxxiii, pp. 629-649). Oxford, England: John Wiley & Sons.

175

Stevens, B., McGrath, P., Gibbins, S., Beyene, J., Breau, L., Camfield, C., . . . Yamada, J. (2003). Procedural pain in newborns at risk for neurologic impairment. Pain, 105(1-2), 27-35.

Sullivan, M. C., & Hawes, K. (2007). A decade comparison of preterm motor performance at age 4. Res Nurs Health, 30(6), 641-654. doi: 10.1002/nur.20220

Sullivan, M. C., Hawes, K., Winchester, S. B., & Miller, R. J. (2008). Developmental origins theory from prematurity to adult disease. J Obstet Gynecol Neonatal Nurs, 37(2), 158-164. doi: 10.1111/j.1552-6909.2008.00216.x

Sullivan, M. C., McGrath, M. M., Hawes, K., & Lester, B. M. (2008). Growth trajectories of preterm infants: birth to 12 years. J Pediatr Health Care, 22(2), 83-93. doi: 10.1016/j.pedhc.2007.02.008

Szatmari, P., Saigal, S., Rosenbaum, P., Campbell, D., & King, S. (1990). Psychiatric disorders at five years among children with birthweights less than 1000g: a regional perspective. Dev Med Child Neurol, 32(11), 954-962.

Tegethoff, M., Pryce, C., & Meinlschmidt, G. (2009). Effects of intrauterine exposure to synthetic glucocorticoids on fetal, newborn, and infant hypothalamic-pituitary- adrenal axis function in humans: a systematic review. Endocr Rev, 30(7), 753- 789. doi: 10.1210/er.2008-0014

Thompson, J. A., & Regnault, T. R. (2011). In utero origins of adult insulin resistance and vascular dysfunction. Semin Reprod Med, 29(3), 211-224. doi: 10.1055/s- 0031-1275522

Tracy, M. F., & Chlan, L. (2011). Nonpharmacological interventions to manage common symptoms in patients receiving mechanical ventilation. Crit Care Nurse, 31(3), 19-28. doi: 10.4037/ccn2011653

Tyson, J. E., Parikh, N. A., Langer, J., Green, C., & Higgins, R. D. (2008). Intensive care for extreme prematurity--moving beyond gestational age. N Engl J Med, 358(16), 1672-1681. doi: 10.1056/NEJMoa073059

Unruh, A. M. (1992). Voices from the past: ancient views of pain in childhood. Clin J Pain, 8(3), 247-254.

176 van Baar, A. L., Vermaas, J., Knots, E., de Kleine, M. J., & Soons, P. (2009). Functioning at school age of moderately preterm children born at 32 to 36 weeks' gestational age. Pediatrics, 124(1), 251-257. doi: 10.1542/peds.2008-2315 van den Akker, C. H., te Braake, F. W., Schierbeek, H., Rietveld, T., Wattimena, D. J., Bunt, J. E., & van Goudoever, J. B. (2007). Albumin synthesis in premature neonates is stimulated by parenterally administered amino acids during the first days of life. Am J Clin Nutr, 86(4), 1003-1008. van Dijk, M., Roofthooft, D., Anand, K., Guldemond, F., de Graaf, J., Simons, S., . . . Tibboel, D. (2009). Taking up the challenge of measuring prolonged pain in (premature) neonates: the COMFORTneo scale seems promising. Clin J Pain, 25(7), 607-616.

Vanhatalo, S., & van Nieuwenhuizen, O. (2000). Fetal pain? Brain Dev, 22(3), 145-150.

Vermes, I., Dohanics, J., Toth, G., & Pongracz, J. (1980). Maturation of the circadian rhythm of the adrenocortical functions in human neonates and infants. Horm Res, 12(5), 237-244.

Wallace, A. M., Beesley, J., Thomson, M., Giles, C. A., Ross, A. M., & Taylor, N. F. (1987). Adrenal status during the first month of life in mature and premature human infants. J Endocrinol, 112(3), 473-480.

Ward, A. M., Moore, V. M., Steptoe, A., Cockington, R. A., Robinson, J. S., & Phillips, D. I. (2004). Size at birth and cardiovascular responses to psychological stressors: evidence for prenatal programming in women. J Hypertens, 22(12), 2295-2301.

Ward, A. M., Syddall, H. E., Wood, P. J., Chrousos, G. P., & Phillips, D. I. (2004). Fetal programming of the hypothalamic-pituitary-adrenal (HPA) axis: low birth weight and central HPA regulation. J Clin Endocrinol Metab, 89(3), 1227-1233.

Watterberg, K. L., Gerdes, J. S., & Cook, K. L. (2001). Impaired glucocorticoid synthesis in premature infants developing chronic lung disease. Pediatr Res, 50(2), 190- 195.

Weller, A., Rozin, A., Goldstein, A., Charpak, N., Ruiz-Pelaez, J. G., Figueroa de Calume, Z., . . . Sack, J. (2002). Longitudinal assessment of pituitary-thyroid axis

177

and adrenal function in preterm infants raised by 'kangaroo mother care'. Horm Res, 57(1-2), 22-26.

Whitfield, M. F., & Grunau, R. E. (2000). Behavior, pain perception, and the extremely low-birth weight survivor. Clin Perinatol, 27(2), 363-379.

Whitfield, M. F., Grunau, R. V., & Holsti, L. (1997). Extremely premature (< or = 800 g) schoolchildren: multiple areas of hidden disability. Arch Dis Child Fetal Neonatal Ed, 77(2), F85-90.

Wilson-Costello, D., Friedman, H., Minich, N., Fanaroff, A. A., & Hack, M. (2005). Improved survival rates with increased neurodevelopmental disability for extremely low birh weight infants inthe 1990s. Pediatrics, 115(4), 997-1003.

Winchester, S. B., Sullivan, M. C., Marks, A. K., Doyle, T., DePalma, J., & McGrath, M. M. (2009). Academic, social, and behavioral outcomes at age 12 of infants born preterm. West J Nurs Res, 31(7), 853-871. doi: 10.1177/0193945909339321

Wittekind, C. A., Arnold, J. D., Leslie, G. I., Luttrell, B., & Jones, M. P. (1993). Longitudinal study of plasma ACTH and cortisol in very low birth weight infants in the first 8 weeks of life. Early Hum Dev, 33(3), 191-200.

Wong, C. M., McIntosh, N., Menon, G., & Frank, L. S. (2003). The Pain (and Stress) in Infants in a Neonatal Intensive Care Unit. In N. l. Schechter, C. B. Berde & M. Yaster (Eds.), Pain in Infants, Children, and Adolescents (2nd ed., pp. 669-693). Philadelphia: Lipppincott Williams & Wilkins.

World Health Organization. (2012). Born too soon: The global action report on preterm birth. In C. Howson, M. Kinney & J. Lawn (Eds.). Geneva, Switzerland.

Xita, N., & Tsatsoulis, A. (2010). Fetal origins of the metabolic syndrome. Ann N Y Acad Sci, 1205, 148-155. doi: 10.1111/j.1749-6632.2010.05658.x

Yamada, J., Stevens, B., de Silva, N., Gibbins, S., Beyene, J., Taddio, A., . . . Koren, G. (2007). Hair cortisol as a potential biologic marker of chronic stress in hospitalized neonates. Neonatology, 92(1), 42-49. doi: 10.1159/000100085

178

Zhang, L. (2005). Prenatal hypoxia and cardiac programming. J Soc Gynecol Investig, 12(1), 2-13. doi: 10.1016/j.jsgi.2004.09.004

179

Appendix A: Data Collection Form

180

181

182

Appendix B: Nursing Standards of Care Document

183

184

Appendix C: Columbia University IRB Approval

185

Appendix D: Stony Brook University IRB Approval

186

187

Appendix E: Research Permission Form

188

189