University of Southern California Dissertations and Theses

The acute relationship between affective states and physiological stress response, and the moderating role of moderate-to-vigorous physical activity

(USC Thesis Other)

The acute relationship between affective states and physiological stress response, and the moderating role of moderate-to-vigorous physical activity

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

The Acute Relationship between Affective States and Physiological Stress Response,
and the Moderating Role of Moderate-to-Vigorous Physical Activity

by
Cheng Kun Wen

A Dissertation Presented to the
Faculty of the Graduate School of the
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
PREVENTIVE MEDICINE: HEALTH BEHAVIOR RESEARCH
December 2018

5 | P a g e

Acknowledgments
I would like to acknowledge my mentors and committee co-chairs Dr. Donna Spruijt-
Metz and Dr. Chih-Ping Chou for their guidance and support throughout the years in the health
behavior research doctoral program and during the dissertation process. I am extremely grateful
for their advice and encouragement on pushing me to think harder and deeper on not only on my
research topics but more importantly, to become a better scientist. I am very grateful for Dr.
Marc J. Weigensberg’s support through the Imagine Health Study, and his wisdom that has
helped to shape my interests in stress research. I am also very grateful for the support from Dr.
Britni R. Belcher and Dr. David S. Black. Their advice and feedback have helped to put my
research interests and study results in perspectives. Every member of this dissertation committee
has uniquely contributed to the development of this dissertation and my career as a scientist.
I would also like to acknowledge the Health Behavior Research Program for providing a
nurturing and supportive environment. I am especially grateful for the countless and tireless help
from Marny Barovich, as well as the support from my great cohort mates Robert Garcia, Jennifer
Tsai, Eldin Dzubur, and Angelical Delgado Rendon, as well as all the HBR friends, Gillian
O’Reilly Gentner, Lauren Cook Martinez, Kimberly Miller, Trevor Pickering, Daniel Chu, and
Em Arpawong, to name a few. There are also faculty members and staff that I would like to
thank for providing me with the opportunities to learn from their great strengths. Among these
are Dr. Genevieve Dunton, Dr. Jimi Huh, and Dr. Adam Leventhal from the Department of
Preventive Medicine, Dr. Stefan Schneider from the USC Center for Self-Report Science, and
Luz Antunez Castillo from Dr. Spruijt-Metz’s lab. These individuals all have offered great
advice that significantly impacted my career developments.
6 | P a g e

Finally, my deepest gratitude is to my loving family. Thank you to my parents, Shang-
Sen Wen and Chih-Yueh Hsu, and my siblings Cheng-Chueh and Cheng-Hsuan Wen for all your
support through my journey of becoming the first doctor in our family. I am also extremely
blessed to have my loving wife Yi-Chen Sylvene Tsai’s supports through the many ups and
downs during this journey. In addition to her tireless supports, she and my son Huey-Cheng Iber
Wen have provided me with an even stronger motivation to charge through the last mile of this
journal. This journey could have been much more challenging without all your support, patience,
and love. For this, I owe you my greatest gratitude.

7 | P a g e

Table of Contents
Abstract ......................................................................................................................................... 13
Chapter 1: Introduction ................................................................................................................. 15
Background and Significance: .................................................................................................. 15
Stress in the United States ......................................................................................................... 15
Stress and Stressors: Definitions ............................................................................................... 15
Stress Responses .................................................................................................................... 16
The Autonomous Nervous System and the Fight or Flight Reaction .................................... 16
The Hypothalamus-Pituitary-Adrenal Axis Activity and Cortisol ........................................ 17
Adapting to the Stressors: Allostasis, Allostatic Load, and the HPA Axis ........................... 18
Disease Implications of a Maladaptive HPA Axis ................................................................ 18
Indices of HPA Axis Activity ................................................................................................... 19
Cortisol Awakening Response (CAR) ................................................................................... 21
Diurnal Cortisol Slope (DCS)................................................................................................ 22
Correlates of the HPA Axis Activity......................................................................................... 23
Trait Correlates of the HPA Axis Activity: ........................................................................... 24
State Correlates of the HPA Axis Activity ............................................................................ 26
Focusing on the Affective States as Predictors of HPA Axis Activity ..................................... 29
Definition of Affective States ................................................................................................ 30
Quantifying the Affective States ........................................................................................... 31
The linkage between Affect and HPA Axis Activities .......................................................... 33
Summary on the Affective States as Predictors of HPA Axis activities ............................... 35
8 | P a g e

Physical Activity as a Possible Moderator of the Relationship Between Affect and HPA Axis
Activities ................................................................................................................................... 36
Gaps in the Literature ................................................................................................................ 37
Study Hypotheses, and Aims .................................................................................................... 39
Study 1 and 2: ........................................................................................................................ 39
Study 3: .................................................................................................................................. 40
Chapter 2: The Day-Level Association Between Affective States, Self-Reported Time in
Exercise, and Diurnal Cortisol Rhythm in Adults in The United States ...................................... 42
Background ............................................................................................................................... 42
Methods ..................................................................................................................................... 46
Study Sample ......................................................................................................................... 46
Measurements ........................................................................................................................ 48
Statistical Analysis ................................................................................................................ 53
Results: ...................................................................................................................................... 56
Descriptive Statistics ............................................................................................................. 56
Affective States and the Same Day Diurnal Cortisol Slope .................................................. 57
Affective States and the Subsequent-Day Cortisol Awakening Response ............................ 58
Time Spent in Exercising as a Potential Moderator for Diurnal Cortisol Rhythms .............. 58
Discussion ................................................................................................................................. 61
Chapter 3: The Daily Affective States and Diurnal Cortisol Rhythm in Urban Minority Youth
and the Moderator Role of Moderate-to-Vigorous Physical Activity .......................................... 74
Background ............................................................................................................................... 74
Methods ..................................................................................................................................... 76
9 | P a g e

Study Sample ......................................................................................................................... 76
Study Procedures ................................................................................................................... 77
Measurements ........................................................................................................................ 79
Statistical Analysis ................................................................................................................ 83
Results: ...................................................................................................................................... 85
Demographic and Descriptive Statistics ................................................................................ 85
Day-Level Association between the Affective States and Diurnal Cortisol Slope and Time-
Spent in MVPA as a Potential Effect Modifier ..................................................................... 86
Discussion: ................................................................................................................................ 86
Chapter 4: The Momentary Affective States and Diurnal Cortisol Rhythm in Minority Youth .. 93
Background ............................................................................................................................... 93
Methods ..................................................................................................................................... 95
Study Sample: ........................................................................................................................ 95
Study Procedure ..................................................................................................................... 96
Measurements ........................................................................................................................ 97
Statistical Analysis ................................................................................................................ 99
Results: .................................................................................................................................... 101
Demographic and Descriptive Statistics .............................................................................. 101
The Momentary Relationship between the Affective States and Subsequent Cortisol Levels
............................................................................................................................................. 102
Time Spent in MVPA as a Potential Effect Modifier .......................................................... 102
Discussion ............................................................................................................................... 103
Chapter 5: Discussion and Conclusion ....................................................................................... 114
10 | P a g e

Summary of Findings and Contributions to the Literature ..................................................... 114
Limitations .............................................................................................................................. 118
Implications of Study Findings ............................................................................................... 119
Future Research Directions ..................................................................................................... 121
References ................................................................................................................................... 123

11 | P a g e

List of Figures
Figure 1: Two main neuroendocrinological stress responses systems.......................................... 17
Figure 2: The range of diurnal cortisol levels (10th, 25th, 50th, 75th, and 90th percentiles)....... 20
Figure 3: Example of cortisol awakening response. The triangular area denoted “Cortisol
Awakening Response” represents the total amount of additional cortisol secreted in response to
the awakening (AUCI). ................................................................................................................. 22
Figure 4: Illustration of two diurnal cortisol slope definitions. Wake-to-bed slope (DCS Wake,
dashed line) and peak-to-bed slope (DCS Peak, dotted line). .......................................................... 23
Figure 5: Trait- and state-like correlates of HPA axis and potential adverse health consequences
....................................................................................................................................................... 29
Figure 6: The circumplex model of affect .................................................................................... 32
Figure 7: Conceptual model for study 1 ....................................................................................... 46
Figure 8: Timeline of MIDUS II Data Collection ........................................................................ 47
Figure 9: Timeline of Imagine Health 7-day observation data collection .................................... 79
Figure 10: Screenshot of affective states questionnaire on ZEMI ................................................ 81
Figure 11: Average Cortisol Levels by Time ............................................................................. 108
Figure 12: Average Cortisol Levels by Time by Meal type (LS v.s. HS breakfast condition) ... 109

12 | P a g e

List of Tables
Table 1: Participant Demographic Characteristics ....................................................................... 59
Table 2: Descriptive Statistics for Affective States ...................................................................... 60
Table 3: Descriptive Statistics for Raw and Log-Transformed Cortisol Values by Sampling Time
....................................................................................................................................................... 60
Table 4: Daily Affective States Predicting Diurnal Cortisol Slope of the Same Day .................. 65
Table 5: Daily Affective States Predicting Cortisol Awakening Response (AUCi) of the
Subsequent Day ............................................................................................................................ 70
Table 6: Demographic Characteristics of the Study Participants ................................................. 90
Table 7: Descriptive Statistics for Affective State Scores and Time Spent in MVPA ................. 90
Table 8: Salivary Cortisol Levels by Time and Diurnal Cortisol Rhythm Estimation ................. 91
Table 9: Daily Affective States Predicting Diurnal Cortisol Slope and the Moderator Effect of
Accelerometer-Measured Time-Spent in MVPA ......................................................................... 92
Table 10: Descriptive Statistics of the Study Participants .......................................................... 107
Table 11: Descriptive Statistics of the Affective States and Time-Spent in MVPA .................. 107
Table 12:Models of Negative Affect, Time Spent in MVPA, and Cortisol Levels Across Visit
and By Meal Type ....................................................................................................................... 110
Table 13: Supplemental Table S1-Models of Average Negative Affect, Time Spent in MVPA,
and Cortisol Levels ..................................................................................................................... 111
Table 14: Supplemental Table S2: Models of Feeling of Panicky, Time Spent in MVPA, and
Cortisol Levels Across Visit and By Meal Type ........................................................................ 113

13 | P a g e

Abstract
Mounting evidence has suggested the elevated disease morbidity associated with a
maladaptive diurnal cortisol rhythm. Identifying predictors of the diurnal cortisol rhythm and its
moderators could potentially bear significant health implications by offering viable targets for
interventions. Affective states, or the individuals’ experience toward an actual or perceived
stimulus, have shown to be associated with diurnal cortisol rhythm. However, the mixed results
suggest the need for studies that examine how affective states predict diurnal cortisol rhythm and
whether there are behavioral moderators to this relationship. This dissertation examined two
main research questions: 1) the relationship between affective states and HPA axis activities and
2) whether the relationship between affective states and HPA axis activities were different when
participants engaged in more MVPA both at the within- and between-person levels.
Study 1 found that mentally healthy adults exhibited a blunted cortisol awakening
response (CAR) on the day after they have reported the higher-than-usual level of high arousal
positive affective states, but not average positive affective states. At the between-person level,
larger CARs were observed among adults who on average reported higher low-arousal negative
affective states on the previous day. Self-reported time spent in the exercise was also not related
to the subsequent day CAR and did not moderate the relationship between affective states and
the subsequent day CAR. Contrary to the hypotheses, study 1 did not support the notion that
affective states are related to the diurnal cortisol slope (DCS) at the within-person level. At the
between-person level, DCSs were 1.2% flatter among adults who reported higher feelings of
calm/peaceful and satisfied (i.e., low arousal positive affect) than others.
Contrary to the findings from study 1, results of the study did not support the hypotheses
that affective states (i.e., positive or negative affective states) are related to the DCS in mentally
14 | P a g e

healthy youth, both at the within- and between-person levels. Additionally, accelerometer-
measured time spent in MVPA was neither related to the DCS nor moderated the relationship
between affective states and DCS.
Study 3 further examined the acute relationship between affective states and cortisol
levels at the subsequent 30 minutes in youth. Results of this study showed that average negative
affective states were related to a slightly heightened (1.91%) salivary cortisol level at the
subsequent 30 minutes at the within-person level only on the lab visit when young participants
were given a breakfast with high in sugar contents, but not on the visit when they were given a
low-sugar high-fiber breakfast. Accelerometer-measured time-spent in MVPA was associated
with heightened (1.7%) cortisol level at the subsequent 30 minutes at the within-person level at
both visits but did not moderate the relationship between affective states and cortisol activity.
This dissertation offers unique insights into the growing literature investigating how
affective states affect the human neuroendocrinological system. Overall, results of this
dissertation provide partial support to the assertion that affective states can be an important
correlate of HPA axis activities at both the day- and moment-level and that energy-balance
related behavior could alter how affective states are related to the hypothalamic-pituitary-adrenal
axis.

15 | P a g e

Chapter 1: Introduction
Background and Significance:
Stress in the United States
In the past decades, the average levels of perceived stress in the United States have been
steadily increased in both adults (American Psychological Association, 2017) and pediatric
(Association, 2014) populations. Stress is related to elevated risks for many adverse physical
(e.g., obesity (Moore & Cunningham, 2012), metabolic syndrome (Chrousos, 2000), slowed
wound healing (Bosch, Engeland, Cacioppo, & Marucha, 2007), cancer proliferation (Emma K
Adam et al., 2017)) and mental (e.g. major depression (Lopez-Duran, Kovacs, & George, 2009;
Pariante & Lightman, 2008)) health outcomes. Identifying correlates of stressful experiences and
understanding the underlying mechanisms between stress and disease morbidity may provide
valuable insights for reducing the negative impact of stress on health.
Stress and Stressors: Definitions
The human body is constantly challenged by internal (e.g., the perception of threats,
insecurity, and emotion) and external demands (e.g., the presence of a predator and high-
intensity physical activity), or stressor. The human body has a complex physiological system to
cope with these demands and to restore and maintain a steady physiological condition (e.g., body
temperature, oxygen concentration, and other vital conditions), or homeostasis(Cannon, 1932).
Under this framework, a stressor is defined as an internal or external demand and the experience
of stress is considered a state in which homeostasis is actually -or perceived to be- threatened
(Chrousos, 1995, 2009).
16 | P a g e

Stress Responses
When confronted by stressors, the amygdala, the emotion-processing center of the central
nervous system (Pessoa & Adolphs, 2010), signals the hypothalamus to invoke a cascade of
neuroendocrinological activities in the central nervous system and various peripheral organs and
tissues (Everly Jr & Lating, 2012). Hallmarks of this stress response are the activation of two
major neuroendocrine systems: the autonomous nervous system (ANS) and the hypothalamus-
pituitary-adrenal (HPA) axis. These systems, along with a repertoire of physiological and
behavioral adaptive responses, serve to mitigate the acute impact of the stressors and to restore
homeostasis.
The Autonomous Nervous System and the Fight or Flight Reaction
When a psychological or physiological demand is perceived as a stressor, the amygdala
triggers the hypothalamus to activate one part of the autonomous nervous system (ANS): the
sympathetic nervous system (SNS). Following the activation of the SNS is a cascade of events
that stimulate the production of two kinds of catecholamines: epinephrine and norepinephrine.
These catecholamines bind to adrenergic receptors in various organs, including smooth muscle
and pupils, therein triggering autonomic physiological responses in these organs (e.g., increased
heart rate and pupil dilation; for a review of other physiological responses (Everly Jr & Lating,
2012)). The subsequent acute physiological responses to this initial surge of these
catecholamines (e.g., increased heart rate, dilated pupils, etc.) are commonly termed the “fight or
flight” reaction, an important acute stress response to provide individuals with physiological
resources to cope with the stressors (See Figure 1 path A).

17 | P a g e

Figure 1: Two main neuroendocrinological stress responses systems

Path A: the autonomic nervous system; Path B: the hypothalamus-pituitary-adrenal axis. CRH:
corticotropin-releasing hormone. ACTH adrenocorticotropin
The Hypothalamus-Pituitary-Adrenal Axis Activity and Cortisol
When confronting a stressor, the hypothalamus also triggers a series of
neuroendocrinological events across the HPA axis - the hypothalamus, the pituitary gland, and
the adrenal gland. Briefly, when the HPA axis is activated, the hypothalamus releases
corticotropin-releasing hormone (CRH) to stimulate adrenocorticotropin (ACTH) production,
which, in turn, stimulates the production and release of glucocorticoids, or cortisol in humans,
from the adrenal glands (See Figure 1 path B). Cortisol is the final effector of the HPA axis
activity with two essential roles: 1) to increase glucose production in response to the acute
stressors after the initial surge of epinephrine (during the fight or flight reaction) subside
(Dickerson & Kemeny, 2004) and 2) to inhibit CRH releases by negative feedback to terminate
the stress response after the stress episode and to restore homeostasis (Chrousos, 2000; Gunnar
& Quevedo, 2007; Tsigos & Chrousos, 2002).
18 | P a g e

Adapting to the Stressors: Allostasis, Allostatic Load, and the HPA Axis
Beyond the episodic stressor, the HPA axis activity is adaptive in that a normally
functioning HPA axis reacts progressively less rigorously when exposed to the same stressors
(McEwen, 2007, 2008). This active and adaptive process through which the human body changes
responses to stressors to maintain homeostasis is termed “allostasis”(Sterling & Eyer, 1988).
While this adaptive process protects against the negative physiological impacts of stressors,
frequent activation of this adaptive process due to chronic or frequent exposure of stressors can
lead to wear-and-tear on the human body and the neuroendocrine system. This accumulation of
this wear-and-tear is termed allostatic load (Susan Folkman, 2013; McEwen, 2007; McEwen &
Gianaros, 2010) and is hypothesized to be a precursor of a maladaptive HPA axis.
Disease Implications of a Maladaptive HPA Axis
One of the pathways through which stress could impact health is through chronic
exposure to an elevated level of cortisol (Lupien, McEwen, Gunnar, & Heim, 2009). Individuals
with a maladaptive HPA axis are exposed to an increased level of cortisol (McEwen, 2007;
Pruessner et al., 1997), which can lead to higher risks of developing symptoms of metabolic
syndrome (e.g. higher visceral adiposity) and obesity (in adults(Emma K Adam et al., 2017; Epel
et al., 2000; Rosmond, Dallman, & Björntorp, 1998) and youth (Emma K Adam et al., 2017;
Weigensberg, Toledo-Corral, & Goran, 2008)), as well as a host of adverse health outcomes (e.g.
cancer diagnosis and disease progression (Emma K Adam et al., 2017) and development of
depression among mentally healthy individuals (Bhattacharyya, Molloy, & Steptoe, 2008; Huber,
Issa, Schik, & Wolf, 2006; Lopez-Duran et al., 2009; Pariante & Lightman, 2008)). However,
research on factors that could exacerbate the development of a maladaptive HPA axis remains
limited (Emma K Adam et al., 2017; Beckie, 2012). Therefore, this dissertation examines
19 | P a g e

whether and how psychological factors, such as affective states, activate the HPA axis to address
this gap in research.
Indices of HPA Axis Activity
A normally functioning HPA axis follows a diurnal pattern in which cortisol levels
increase rapidly by 50-156% within 30-45 minutes of awakening (cortisol awakening response,
or CAR(Clow, Thorn, Evans, & Hucklebridge, 2004)), decrease within a few hours after the
peak, and then trail off until reaching a nadir at bedtime (Figure 2 (R. Miller et al., 2016)). These
diurnal patterns vary substantially across individuals (Emma K Adam et al., 2017; Almeida,
Piazza, & Stawski, 2009; Chida & Steptoe, 2009). For example, chronically stressed, but
otherwise healthy, individuals exhibit stronger CARs (Chida & Steptoe, 2009). Compared to
mentally healthy individuals, patients of several mental health conditions (e.g. chronic fatigue
syndrome (CFS) (Chida & Steptoe, 2009; Kumari et al., 2009; Nater et al., 2008), post-traumatic
stress disorder (Chida & Steptoe, 2009; Suglia, Staudenmayer, Cohen, & Wright, 2010), and
major depressive disorder (MDD) (Chida & Steptoe, 2009; Herbert, 2013; Huber et al., 2006))
awaken with higher cortisol levels, exhibit smaller CARs, and have slower declines in cortisol
levels throughout the day. These patterns are reflective of the impaired HPA axis functions in
acutely reacting to and recovering from the exposure of stressors or a hypoactive HPA axis.

20 | P a g e

Figure 2: The range of diurnal cortisol levels (10th, 25th, 50th, 75th, and 90th percentiles)
(R. Miller et al., 2016)

Earlier evidence examining the acute impacts of stressors on cortisol levels among
healthy individuals focused on evaluating cortisol reactivity (rate of cortisol increase in response
to acute stressors) and recovery (rate of cortisol decline after the acute stressors are removed) in
controlled laboratory settings by comparing participants’ serum or saliva samples before and
multiple times after exposure to laboratory-induced stressors (e.g., public speaking tasks). While
evidence from these studies has established valuable foundations for mapping out the acute
neurophysiological effects of stressors, there is an emerging interest in examining HPA axis
activities in free-living environments. Main methodological advantages with data collected in the
free-living environment can provide evidence with an improved external, and ecological validity
for that individuals may react differently when faced with stressors arise from day-to-day
challenges compared to laboratory-induced ones. In free-living studies, the cortisol awakening
response (CAR) and diurnal cortisol slopes (DCS) are the two most commonly utilized indices of
the HPA axis for examining the impact of day-to-day experiences on HPA axis activity.
21 | P a g e

Cortisol Awakening Response (CAR):
Since first documented by Pruessner and colleagues in 1997 (Pruessner et al., 1997), the
CAR has been the most extensively used physiological indicators for HPA axis activity in free-
living settings. The CAR is part of the HPA axis’ response to awakening and is characterized by
a 50-156% increase in cortisol level within 30-45 minutes of awakening (Clow et al., 2004). The
size of CAR is conventionally defined as the total amount of cortisol produced during this period
and is quantified using the trapezoidal rules (for the mathematical formulas: (Rotenberg,
McGrath, Roy-Gagnon, & Tu, 2012)). Briefly, CAR is calculated by subtracting the area under
the curve with respect to the ground level (AUC AG) by the total ground cortisol level (see Figure
3). The resulting triangular area illustrated in Fig. 3 is the area under the curve with respect to
increase (AUCI), which represents the amount of cortisol secreted in addition to the level of
cortisol upon awakening. Besides the trapezoidal method, others have used the difference in
cortisol levels between the awakening and peak to quantify the CAR, herein denoted as the mean
increase (MnInc) method. While both methods are acceptable, the AUC I method accounts for the
timing of the wake and the peak samples, whereas the MnInc method assumes equal time
differences between the two samples. In accordance with the current recommendations by the
International Society of Psychoneuroendocrinology (Stalder et al., 2016), the CARs discussed
and operationalized in this dissertation are the dynamic post-awakening portion of the diurnal
cortisol rhythm, or AUC I, when wake time data is available.

22 | P a g e

Figure 3: Example of cortisol awakening response. The triangular area denoted “Cortisol
Awakening Response” represents the total amount of additional cortisol secreted in response to
the awakening (AUCI).

Diurnal Cortisol Slope (DCS):
The diurnal cortisol slope (DCS) is an increasingly utilized physiological indicator for
HPA axis diurnal rhythm. As mentioned previously, cortisol levels decline throughout the day
after reaching the peak level. A faster decline in cortisol levels throughout the day, or steeper
DCS, signifies a more efficiently regulated, or adaptive, HPA axis. Therefore, DCS is believed to
reflect the negative feedback control of the HPA axis and an index of the HPA axis that is
sensitive to acute stressors during the day (Emma K Adam et al., 2017; Doane & Adam, 2010).
The DCS is generally referred to as the slope of the decline in cortisol levels throughout the day.
However, it has been operationalized in various ways, depending on the time and frequency of
cortisol sampling procedures and whether wake time is assessed. In studies with wake time
assessed, the DCS can be calculated by two ways: the wake-to-bed slope (DCS Wake )(E. K. Adam
23 | P a g e

& Kumari, 2009) and the peak-to-bed slope (DCS Peak )(Emma K Adam et al., 2017) (see Figure 4
for illustration). The current recommendations for quantifying DCS is the rate of decline in
cortisol levels relative to the individual’s wake time (DCS Wake)(E. K. Adam & Kumari, 2009),
although some have argued that DCS Peak offers unique insights as the physiological mechanisms
underlying the CAR and the decline after peak may be different (E. K. Adam et al., 2015; Clow,
Hucklebridge, Stalder, Evans, & Thorn, 2010; Maina, Palmas, Bovenzi, & Filon, 2009). Despite
the on-going debates, as DCSWake has been the most commonly used method for quantifying
DCS (Emma K Adam et al., 2017), the DCS in this proposal will be quantified as DCS Wake.
Figure 4: Illustration of two diurnal cortisol slope definitions. Wake-to-bed slope (DCS Wake,
dashed line) and peak-to-bed slope (DCSPeak, dotted line).

Correlates of the HPA Axis Activity
Both indices of the diurnal cortisol rhythm (i.e., CAR and DCS) exhibit substantial inter-
and intra-personal variability (Emma K Adam et al., 2017; Almeida, Piazza, et al., 2009; Chida
& Steptoe, 2009; Fries, Dettenborn, & Kirschbaum, 2009; Ross, Murphy, Adam, Chen, & Miller,
2014). This evidence suggests that the HPA axis could be affective by both trait- and state-like
0
2
4
6
8
10
0 2 4 6 8 10 12 14 16 18
Cortisol level (nmol/L)
Time of Day (Hour since awakening)
Example of Cortisol Diurnal Rhythm
24 | P a g e

variables. Currently, evidence on the trait- and state-correlates of diurnal cortisol rhythm will be
discussed separately in the following section, as trait-like characteristics and state variables
related to the HPA axis activities differently. Due to the limited available evidence on correlates
of the HPA axis in youth, the following sections present evidence primarily from the adult
literature, followed by relevant evidence on the youth population.
Trait Correlates of the HPA Axis Activity:
Trait correlates of the HPA axis activity in the existing literature can be roughly
categorized into two categories: demographic characteristics and mental health conditions.
Demographic characteristics:
Demographic characteristics, including gender, age, ethnicity and socioeconomic status (SES),
are commonly included in analyses of the HPA axis activities as covariates. The current
literature suggests that women generally exhibiting larger and prolonged CARs (Hollanders, van
der Voorn, Rotteveel, & Finken, 2017; Oskis, Loveday, Hucklebridge, Thorn, & Clow, 2009;
Stalder et al., 2016; Steptoe, Kunz-Ebrecht, Brydon, & Wardle, 2004; Vreeburg et al., 2009) but
also a steeper decline in DCS (Shirtcliff et al., 2012; Šupe-Domić, Milas, Hofman, Rumora, &
Klarić, 2016)). Regarding the effect of age on the HPA axis, the human HPA axis activities
remain stable after adolescents transitioning into young adulthood (R. Miller et al., 2016).
Individuals in developmental states that involve substantial hormonal changes, however, exhibit
slightly different diurnal cortisol rhythm. For example, several have reported that, compared to
the adult's counterparts, the size of CAR is larger (Platje et al., 2013) and the decline of DCS is
slower during adolescence (Shirtcliff et al., 2012). Additionally, altered diurnal cortisol rhythms
have also been documented in other developmental stages that involve significant change in
hormonal activities, including infancy (R. Miller et al., 2016; Stalder et al., 2013), and with aging
25 | P a g e

populations (Evans et al., 2011; R. Miller et al., 2016)). Other demographic characteristics, such
as ethnicity and socioeconomic status (SES), have also been commonly examined. Results of
several national studies indicated that both Hispanic and African American exhibit a hypoactive
HPA axis activities (i.e., blunted CAR and flattened DCS (Cohen et al., 2006; Hajat et al., 2010))
compared to their non-Hispanic White counterparts. Similar hypoactive HPA axis activities have
also been documented among individuals with low SES (i.e. blunted CAR (Bennett, Merritt, &
Wolin, 2004; Hajat et al., 2010) and flattened DCS (Cohen et al., 2006; Willner, Morris, McCoy,
& Adam, 2014)) compared to the higher SES counterparts, independent of the ethnic
background. Additionally, maladaptive diurnal cortisol rhythms (i.e. higher CAR or blunted
DCS) have also been found associated with higher body mass index (BMI) (Champaneri et al.,
2013; Incollingo Rodriguez et al., 2015; Ranjit, Young, Raghunathan, & Kaplan, 2005; Ursache,
Wedin, Tirsi, & Convit, 2012) and waist-to-hip ratio (Steptoe et al., 2004; Therrien et al., 2007),
although conflicting (Therrien et al., 2007) and null findings have also been documented
(Kumari, Chandola, Brunner, & Kivimaki, 2010; Ranjit et al., 2005; Steptoe et al., 2004). Albeit
the effect of these demographic characteristics on indices HPA axis activities is generally small,
the inclusion of these demographic characteristics as trait covariates in the analyses of HPA axis
activities are recommended (Stalder et al., 2016).
Chronic Fatigue, Stress, and Mental Health Conditions:
Studies have shown that patients with affective disorder (e.g. major depressive disorder
(Burke, Davis, Otte, & Mohr, 2005; Chida & Steptoe, 2009; Herbert, 2013; Huber et al., 2006))
and post-traumatic stress disorder (Chida & Steptoe, 2009; Suglia et al., 2010) exhibit a
maladaptive HPA axis activity patterns. More specifically, these patients exhibit cortisol
activities that resemble a hypoactive HPA axis, including blunted reactivity and recovery from
26 | P a g e

exposure to acute stressors (Burke et al., 2005) and altered diurnal patterns (i.e., blunted CAR
and flattened DCS). Similar hypoactive HPA axis activity patterns have been documented among
in studies of mentally healthy individuals who are chronically burned out (Pruessner,
Hellhammer, & Kirschbaum, 1999; Pruessner et al., 1997), fatigue (Chida & Steptoe, 2009;
Papadopoulos & Cleare, 2011; Powell, Liossi, Moss-Morris, & Schlotz, 2013), and diagnosed of
chronic fatigue syndrome(Chida & Steptoe, 2009; Kumari et al., 2009; Nater et al., 2008). Taken
together, the evidence suggests that individuals with diagnosed of affective disorders and PTSD,
and mentally healthy individuals who are chronically fatigue or stressed exhibit an altered
diurnal cortisol rhythm that suggests a maladaptive and possibly hypoactive HPA axis.
State Correlates of the HPA Axis Activity
A relatively small amount of studies have examined various day-level or momentary
behavior and psychological states on the HPA axis activity patterns. The current evidence is
leaning toward the notion that the mechanism underlying CAR and DCS are different (E. K.
Adam et al., 2015; Clow et al., 2010; Edwards, Braunholtz, Lilford, & Stevens, 1999; Schmidt-
Reinwald et al., 1999). Therefore, state correlates for the CAR and DCS will be discussed
separately in the following section.
State Correlates of the Cortisol Awakening Response (CAR)
Most studies focused on examining behavioral and psychological factors as state
correlates of the CAR. Among behavioral correlates, sleep-related factors have been shown to be
consistent correlates of the CAR. More specifically, individuals who wake up late exhibit a
blunted CAR (Almeida, Piazza, et al., 2009; Stalder, Evans, Hucklebridge, & Clow, 2010)
(Kudielka & Kirschbaum, 2003), possibly as a result of higher cortisol levels upon awakening on
days with later wake times. Altered sleep patterns have also been shown to relate to variation in
27 | P a g e

magnitude of the CAR (Doane et al., 2010; Law, Hucklebridge, Thorn, Evans, & Clow, 2013),
while null associations between sleep and CAR have also been reported (Kunz-Ebrecht,
Kirschbaum, Marmot, & Steptoe, 2004; Pruessner et al., 1997). The inconsistency in between
sleep, wake time, and the magnitude of CAR could be attributed to participant non-compliance in
sample collection (Law et al., 2013). Participant non-compliance in sample collection and
sampling time assessment have both been shown to produce misleading CAR values (Broderick,
Arnold, Kudielka, & Kirschbaum, 2004; Rotenberg & McGrath, 2014).
In terms of the state psychological state correlates, the current evidence favors the
hypothesis that the magnitude of CAR is reflective of individuals’ anticipation of upcoming
demands (Fries et al., 2009). Supporting this notion is the evidence that showed the acute
relationship between lower perceived controllability (e.g., feeling lack of control (E. K. Adam,
Hawkley, Kudielka, & Cacioppo, 2006) or perceived threat (E. K. Adam et al., 2006)) and higher
the CAR on the subsequent day. Along the same line, adults who reported higher day-level
perceived stress also exhibited a larger CAR on the subsequent day (Stalder et al., 2010).
However, whether the same notion remains valid for the adolescent population remain to be
explored, as null results between prior day daily stressors did not predict CAR of the subsequent
day in adolescents (Lippold, Davis, McHale, Buxton, & Almeida, 2016). Related to the
perception of stress, several studies have also examined the relationship between affective states
and the CAR. Current evidence suggests that participants exhibit a larger CAR on the day after
they have reported higher levels of average negative affect (Piazza, Charles, Stawski, &
Almeida, 2013) or specific affective states of negative valence (i.e. worry (Arbel, Shapiro,
Timmons, Moss, & Margolin, 2017), loneliness (E. K. Adam et al., 2006; Doane & Adam,
2010), tension (Stalder et al., 2010), or sadness (E. K. Adam et al., 2006)). The effect of positive
28 | P a g e

affect on the magnitude of the CAR of the subsequent day remains limited and mixed (Hoyt,
Craske, Mineka, & Adam, 2015; Polk, Cohen, Doyle, Skoner, & Kirschbaum, 2005).
State Correlates of the Diurnal Cortisol Slopes
There is only a limited amount of evidence on the state correlates of DCS. Within this
body of literature, the focus is mainly on the acute relationship between affective states (both
positive and negative) and DCS. The evidence, however, is mixed at best. For example, some
researchers have reported a steeper decline in DCS on the days when participants reported higher
positive affect than usual (Hoyt et al., 2015; Human et al., 2015; Matias, Nicolson, & Freire,
2011) or experienced a positive event(Sin, Ong, Stawski, & Almeida, 2017), while others have
found that positive affective states were associated with a blunted DCS (Polk et al., 2005) or a
null relationship (Chen et al., 2017). Similarly, studies on the day-level association between
negative affective states and DCS found a blunted DCS (Hoyt et al., 2015) (Chen et al., 2017),
and null relationship (Polk et al., 2005). Besides affective states and experiences, some
contextual state correlates (e.g., being alone (Matias et al., 2011) or experienced high
work/family conflicts (Almeida et al., 2016)) are related to a blunted DCS.
In summary, the current literature suggests that individuals’ HPA axis activities are different by
several trait-like characteristics and can be altered by some state-like predictors. Figure 5
illustrates briefly how these trait- and state-like predictors could impact health. Prolonged
exposure to cortisol, a usual consequence of a maladaptive HPA axis, is implicated in the
manifestation of several adverse physical and mental health consequences(Emma K Adam et al.,
2017). Among the mixed findings, a better of how state-like correlates like affective states are
related to the diurnal cortisol rhythm could bear potential importance. Affective states are
individuals’ experiences toward an actual or perceived stimulus in their lives (Barrett, 2017;
29 | P a g e

Tracy & Randles, 2011). The preliminary evidence summarized above has suggested that
affective states could acutely activate the HPA axis and affect diurnal cortisol activities. As
repeated activation of the HPA axis could lead to the development of a maladaptive HPA axis, a
better understanding of whether and under what context affective states activate the HPA axis
relationship may potentially prevent a healthy HPA axis from becoming maladaptive.
Figure 5: Trait- and state-like correlates of HPA axis and potential adverse health consequences

Focusing on the Affective States as Predictors of HPA Axis Activity
Accumulating evidence has suggested the need for identifying day-level and acute
correlates of HPA axis activities (Emma K Adam et al., 2017; Almeida, Piazza, et al., 2009).
Affective states, as mentioned previously, are individual’s momentary feeling to actual or
perceived external stimulus (Barrett, 2017; Tracy & Randles, 2011). When these stimuli are
perceived as a demand or a stressor, affective states could activate the stress response pathway,
which in turn activate the HPA axis (Susan Folkman, 2013; S. Folkman, Lazarus, Dunkel-
30 | P a g e

Schetter, DeLongis, & Gruen, 1986; McEwen, 2007). Evidence supporting this notion has been
accumulating (E. K. Adam, 2006; Chida & Steptoe, 2009; Hoyt et al., 2015; Polk et al., 2005),
while null findings also continue to emerge (Chen et al., 2017; K. G. Miller et al., 2016; Polk et
al., 2005; Smyth et al., 1998). Therefore, the current dissertation sought to contribute to this
literature by further investigating the relationship between affective states and HPA axis
activities.
Definition of Affective States
Affective states and emotion are commonly used interchangeably to describe the
psychological states of an individual. The definition of emotion is a frequently debated issue
among contemporary psychologists, especially with respect to its etiology. For example, the
theory of constructed emotion asserts that emotions are shaped by various neurological sensory
inputs, both consciously and unconsciously, to help the brain perceive the stimulus and thereby
more efficiently regulate homeostasis (Barrett, 2017). On the other hand, the models for basic
emotions asserts that certain regions of the brain are responsible for a small number of
psychologically primitive emotions (e.g., anger, fear, and happiness) and these primitive
emotions interact with other cognitive processes to create more complex emotions (the models
for basic emotions are a collection of many models built on the similar theoretical basis. For
review (Ekman & Cordaro, 2011; Tracy & Randles, 2011). Despite the debate on the etiology,
most theorists agree that emotions are products of complex neurological activities in response to
external or internal stimuli and that these emotions form the basis of how humans perceive the
surroundings (Barrett, 2017; Ekman & Cordaro, 2011). Affective states are viewed by both
schools of thoughts as the momentary feeling state or experience occurs throughout the day.
These psychological states provide an important source of information for updating the prior
31 | P a g e

experience (Barrett, 2017) or helping shape more complex emotion by interacting with basic
emotions(Tracy & Randles, 2011). Therefore, affective states only represent the momentary
portion of emotion and are not synonyms for emotion. Accordingly, affective states, in this
dissertation, is defined as the feeling and experience states of an individual.
Quantifying the Affective States
The two commonly used approaches for quantifying affective state are the discrete
approach (e.g. (Ekman & Cordaro, 2011; Lazarus, 1991)) and the dimensional approach (e.g.
(Posner, Russell, & Peterson, 2005)). The differences between these two approaches mainly stem
from the different definitions of emotion from the functionalistic (Ekman & Cordaro, 2011) and
the constructionistic views (Posner et al., 2005). The discrete approach, based on models of basic
emotions (Tracy & Randles, 2011), views affective states as products of individuals appraisal of
experiences at the moment of inquiry (Lazarus, 2001). This view is also reflected in the affect-
HPA axis activity research as the integrated specificity model (Dickerson, Gruenewald, &
Kemeny, 2004). Under this model, each item of affective states are used as individual predictors
for specific neurophysiology manifestations. The dimensional approach, based off “the
circumplex model of affect,” asserts that each affective state overlaps with one another and
without a discrete border that clearly differentiate one from another(Posner et al., 2005). The
dimensional approach of affective states categorizes affective states, based on commonality, into
dimensions by valence (i.e., positive or negative) and by state of arousal (i.e., high arousal or low
arousal), resulting in a global value for 4 distinctively different types of affective states (see
figure 6 for example of the circumplex model of affect). This model has further been adopted
into the theory of constructed emotion, which hypothesizes that more than one brain regions are
responsible for generating the experience of emotions. Under the dimensional approach, the
32 | P a g e

valence and the state of arousal components of affective states are extracted based on the
commonality among items of the affective states questionnaire.
Figure 6: The circumplex model of affect

Although different in theoretical and methodological underpinnings, these two
approaches offer unique and possibly complementary insights into an individual’s experience.
Both have been utilized in the literature that examines the linkage between affective states and
HPA axis activities. The following section will review the current literature on the day-level and
momentary relationship between affective states and HPA axis activities. Following the similar
structure in sections reviewing correlates of CAR and DCS, the subsequent sections will review
the acute (i.e., day-level and momentary) relationship between affective states and either index of
HPA axis activities (i.e., CAR or DCS) separately. Due to the limited evidence available for the
youth population, the association between affective states and HPA axis activities for youth will
be separately discussed when available instead of as its own section.
33 | P a g e

The linkage between Affect and HPA Axis Activities
The affective States and the Subsequent-Day Cortisol Awakening Response (CAR)
In the adult literature, studies have shown that mentally health adult exhibits a larger
CAR on the days after they had reported a higher level of average negative affect (Polk et al.,
2005) and other individual affective states of negative valence (e.g., feeling of stress and tension)
(Stalder et al., 2010). The relationship between prior day positive affect and CAR, however, was
less consistent (Polk et al., 2005; Steptoe, Dockray, & Wardle, 2009). In the youth literature, the
evidence that examined the relationship between affective states and the HPA axis activity using
the dimensional approach (i.e., positive and negative affective states) is limited and mixed (Hoyt
et al., 2015). On the other hand, several studies have reported that young participants exhibited a
larger CAR on the days after reported a higher level of worry (Arbel et al., 2017), loneliness (E.
K. Adam et al., 2006; Doane & Adam, 2010), and sadness (E. K. Adam et al., 2006) compared to
days with lower levels of these prior day experiences. Nonetheless, null results that conflict with
findings from the adult literature (e.g. for negative affect (K. G. Miller et al., 2016; Polk et al.,
2005; Smyth et al., 1998) and for positive affect(Chen et al., 2017; Polk et al., 2005)) has also
been reported. For example, prior day feeling of stressed did not predict
the CAR of the next day in youth (Lippold et al., 2016)).
The affective States and Diurnal Cortisol Slopes of the Same Day
In the adult literature, while some have reported that higher positive affective states are
associated with a steeper decline in DCS (Matias et al., 2011), some reported that higher positive
affective states are linked to a blunted DCS (Polk et al., 2005). Among studies that examined
affective states of negative valence, null results have been reported in studies that operationalized
negative affective states by its valence (i.e. average negative affective states (Matias et al., 2011;
34 | P a g e

Polk et al., 2005)), while some reported a blunted DCS had been reported on days that
participants reported higher in specific aspect of negative affect (e.g. feeling of stressed (Sladek,
Doane, & Stroud, 2017)). Mixed results have also been also documented in the youth literature.
Among youth, positive affective states are found to associate with a steeper decline in DCS (the
healthier pattern) in some (Hoyt et al., 2015), but not others (Chen et al., 2017). On the other
hand, the relationship between day-level negative affective states and the DCS in children is
more consistent. Youth exhibited a blunted DCS on days they reported more negative affective
states (Chen et al., 2017; Doane & Zeiders, 2014; Hoyt et al., 2015).
Momentary Relationship Between Affective States and HPA Axis Activities
The literature examining the momentary-level effect of affective states on HPA axis
comprises mostly studies that took place in laboratory settings. In an early meta-analysis that
examined the acute effect of emotional induction on cortisol level in laboratory settings, Denson
et al reported that individuals had higher cortisol levels after induction of specific domains of
affective states (e.g., social threat, various cognitive appraisal ratings, and ruminations),
suggested the potential role of affective states on acute activation of HPA axis (Denson,
Spanovic, & Miller, 2009). These results, however, may be limited by its external validity, as
momentary experience from the in-lab stressors can be distinctively different from experiences
that occur in natural and free-living environments. Among the limited evidence that has
examined the affect-HPA axis activity relation at a within-day level in the free-living
environment, Human et al. found that adults with a higher within-day variability in positive
affect exhibited a blunted DCS while lower within-day variation in positive affect is associated
with steeper DCS (Human et al., 2015). Nonetheless, in a separate study, Hoyt et al., reported
that positive affect did not predict subsequent cortisol levels in an hour (Hoyt, Zeiders, Ehrlich,
35 | P a g e

& Adam, 2016). In terms of negative affective states, a momentary increase in negative affect is
associated with higher cortisol output (Doane & Zeiders, 2014). Sadness has also been found to
marginally but positively predict subsequent cortisol levels in an hour (Hoyt et al., 2016).
Summary on the Affective States as Predictors of HPA Axis activities
Although the HPA axis in mentally healthy individuals follows a strong diurnal rhythm,
studies have shown that affective states are acutely associated with indices of HPA axis
activities. However, findings on both the magnitude and the valence of the acute effect of
affective states on HPA axis activities at the day- and momentary-level are mixed and limited.
One of the possible explanation for the mixed finding, especially with studies using average
negative or positive affective states as a predictor, is that the valence the only approach
overlooks the effect of the level of emotional activation, or the state of arousal. Preliminary
evidence has suggested that high arousal, not average, positive affect was cross-sectionally
associated with adaptive HPA axis activity patterns (i.e., steeper declines in DCS) in youth in
natural environments (Hoyt et al., 2015). However, as this is among the only few existing studies
that examined whether the state of arousal is related to the HPA axis activities, further replication
in other populations is necessary. In addition to the ways affective states are quantified, the
mixed findings between affective states and HPA axis activities could also suggest that there are
unidentified effect modifiers (Campbell & Ehlert, 2012; DeSteno, Gross, & Kubzansky, 2013).
While the literature is limited, moderators of the relationship between affect and HPA axis
activities have been documented in several recent studies, with most focused on the moderating
role of trait psychological factors (e.g. perceived discrimination (Doane & Zeiders, 2014),
perseverative cognition(Brosschot, Gerin, & Thayer, 2006)), contextual information (e.g.
36 | P a g e

solitude (Matias et al., 2011), and the use of coping strategies or receiving support (Almeida et
al., 2016; Savla, Zarit, & Almeida, 2017).
Physical Activity as a Possible Moderator of the Relationship Between Affect and HPA
Axis Activities
There is a well-established literature on the beneficial effect of physical activity on
individuals mental (Biddle & Asare, 2011) and physical health (Kerr, Anderson, & Lippman,
2017)). Moderate-to-vigorous physical activity (MVPA) has also been shown to acutely improve
individual’s positive affective states (Dunton et al., 2014; Ekkekakis, Parfitt, & Petruzzello,
2011) while alleviating negative affective states (Liao, Shonkoff, & Dunton, 2015; C. K. F. Wen
et al., 2018). MVPA, however, has a paradoxical role in the stress response system. While there
is a widely documented remedial effect on mental health conditions associated with a
maladaptive HPA axis (e.g., depression(Warburton, Nicol, & Bredin, 2006)),MVPA is also a
physical stressor that could acutely activate the HPA axis, even though such an effect only exist
at a high intensity (i.e. above 60% maximal oxygen consumption (VO 2max) for 30 minutes for
non-athletes (Hill et al., 2008) and above 80% VO 2max for athletes (VanBruggen, Hackney,
McMurray, & Ondrak, 2011). The literature examining this paradox leans toward supporting the
cross-stressor adapter hypothesis(Sothmann et al., 1996). The cross-stressor adapter hypothesis
suggests that that frequent exposure to physical activity, a stressor, can lead to physiological
adaptation and in turn, reduces individual’s reactivity to stressors. In laboratory-based studies,
habitually active individuals (i.e., athletes) have shown to react less strongly to standardized
stress test (e.g., Trier Social Stress Test (TSST)) compare to non-athletes (Rimmele et al., 2009;
Rimmele et al., 2007). Among non-athletes, a small number of recent studies have reported
similar buffering effect of MVPA, where habitually active individuals (adults(Puterman et al.,
37 | P a g e

2011) and children (Martikainen et al., 2013)) reacted less rigorously and recover faster from
TSST than their less active counterparts(Strahler, Fuchs, Nater, & Klaperski, 2016). Preliminary
evidence has further shown that the individuals assigned to a 30-minute unstructured moderate-
intensity walk prior to the TSST test secreted less cortisol than those who were not assigned to
the 30-minute walk condition(Wood, Clow, Hucklebridge, Law, & Smyth, 2017). Together,
evidence from the literature suggests that acute physical activity engagements may attenuate the
neurophysiological effect of stressor exposure.
Gaps in the Literature
Converging evidence demonstrates that elevated exposure to cortisol from a maladaptive
HPA axis is linked to adverse physical (Emma K Adam et al., 2017) and mental health outcomes
(Pariante & Lightman, 2008). These converging evidence also provides support that the HPA
axis is an important mediator between individuals’ perception of stress and these disease
morbidity (Fries et al., 2009). Research focusing on identifying antecedents of HPA axis
activation and possible behavioral moderators could improve the current understanding of HPA
axis activities and provide opportunities for behavioral interventions. Recent developments in the
literature have provided preliminary evidence regarding the role of affective states on HPA axis
activities, the literature is nascent and limited at best.
There are three main gaps in the literature that this proposal seeks to fill. Firstly, the
literature examining the acute relationship between affective states and HPA axis activities have
relied predominantly on studies conducted in laboratory settings. As feelings and experiences
occurring in the free-living situation can substantially differ from the laboratory-induced ones,
studies in the free-living environment can provide new and relevant insights. Two studies in this
dissertation used intensive longitudinal data collected using a daily diary method (study 1) and
38 | P a g e

ecological momentary assessment (study 2) in the free-living environment. The results may
provide evidence with improved ecological validity. Secondly, although several studies provide
preliminary evidence on the day-level and momentary relationship between affective states and
HPA axis activities, most studies that took the dimensional approach of quantifying affective
states when examining the affect-HPA axis relationship have been less fruitful than studies that
took the discrete approach. A possible explanation may also be that studies only quantified
participants’ affective states by their valence (i.e., average positive or negative affect) may have
overlooked the possible effect of the state of arousal. Therefore, study 1 sought to improve upon
the current understanding of the affect-HPA axis relationship by additionally investigate the
effect of high arousal/low arousal affective states on indices of diurnal cortisol rhythm. Thirdly,
the current evidence on the affect-HPA axis relationship has been largely developed on a body of
literature that focused primarily on adult population and patients with mental health conditions.
However, youth are different from the adult, both psychologically and physiologically.
Adolescence is a transitional period when the HPA axis (R. Miller et al., 2016), along with the
rest of children’s physiological systems, are developing rapidly. At the same time, children are
exposed to an increased variety of experiences (e.g., peer interaction) that could help in shaping
emotional experiences (Barrett, 2017; Tracy & Randles, 2011). With these transitions, the type
and magnitude of affective states that activates HPA axis activities in youth may be different
from their adult counterparts. Study 2 and 3 of this dissertation, therefore, sought to contribute to
this literature by focusing on the affect-HPA axis relationship in youth. Lastly, MVPA has been
shown to mitigate the impact of a laboratory-induced stressor on the human
neuroendocrinological system both at the between- (Martikainen et al., 2013; Puterman et al.,
2011) and the within-person (Wood et al., 2017) level. Nonetheless, whether acute increase in
39 | P a g e

physical activity can exert a similar protective effect on stress reactivity and recovery in the free-
living environment is less understood, especially in the youth population. Therefore, one of the
major goals of this dissertation, in addition, to examine the linkage between affective states and
diurnal cortisol rhythm, is to examine whether MVPA acutely moderate the linkage between
affective states and the parts of the diurnal cortisol rhythm that are reactive to acute stressors.
Study Hypotheses, and Aims
Study 1 and 2:
Aim 1: To examine the relationship between positive affective states and indices of the HPA
axis.
- Hypotheses 1.1: The WP version of the positive affective state score is negatively
associated with diurnal cortisol slope of the same day, such that on days when
participants reported a higher positive affective state than the BP version of that positive
affective state, the diurnal cortisol slope of that day is steeper.
- Hypothesis 1.2: The WP version of the positive affective state score is negatively
associated with the cortisol awakening response of the subsequent day, such that the day
after participants reported a higher positive affective state than the BP version of that
positive affective state, the size of the cortisol awakening response is smaller.
Aim 2: To examine the relationship between negative affective states and indices of the HPA
axis.
- Hypotheses 2.1: The WP version of the negative affective state score is positively
associated with diurnal cortisol slope of the same day, such that on days when
participants reported higher negative affective states than the BP version of that affective
state, the diurnal cortisol slope of that day is less steep.
40 | P a g e

- Hypothesis 2.2: The WP version of the negative affective state score is negatively
associated with the cortisol awakening response of the subsequent day, such that the day
after participants reported higher negative affective states than the BP version of that
affective state, the size of the cortisol awakening response is larger.
Aim 3: To examine the moderator role of accelerometer-measured MVPA on the relationship
between affective states and HPA axis activities
- Hypothesis 3.1: Minutes of MVPA spent in a day moderates the association between
positive affective state and diurnal cortisol slope of the same day.
- Hypothesis 3.2: Minutes of MVPA spent in a day moderates the association between
negative affective states and diurnal cortisol slope of the same day.
- Hypothesis 3.3: Minutes of MVPA spent in a day moderates the association between
positive affect states and the size of cortisol awakening response of the subsequent day.
- Hypothesis 3.4: Minutes of MVPA spent in a day moderates the association between
negative affect states and the size of cortisol awakening response of the subsequent day.
Study 3:
Aim 1: To examines the acute effect of momentary affective states on the change in salivary
cortisol level of the subsequent 30 minutes.
- Hypothesis 1.1: Higher negative affect score at the measurement, compared to the
participants’ personal average, predicts a slower decline in cortisol level for the
subsequent 30 minutes.
- Hypothesis 1.2: Higher positive affect score at the measurement, compared to the
participants’ personal average, predict change in decline in cortisol level for the
subsequent 30 minutes.
41 | P a g e

Aim 2: To examine the effect of time spent in MVPA on the acute relationship between affective
state and change in salivary cortisol level during the subsequent 30 minutes.
- Hypothesis 2.1: Time spent in MVPA moderates the acute relationship between negative
affective state and decline in cortisol level at the subsequent 30 minutes, such that the
decline in cortisol is steeper when participants engaged in more MVPA during the 30-
minute interval.
- Hypothesis 2.2: Time spent in MVPA moderates the acute relationship between positive
affective state and decline in cortisol level at the subsequent 30 minutes, such that the
change in cortisol is different when participants engaged in more MVPA during the 30-
minute interval.
Exploratory Aim: To compare the acute relationship between affective state and change in
salivary cortisol level during the subsequent 30 minutes between a high sugar low fiber (HS)
breakfast condition and a low sugar high fiber (LS) breakfast condition.
- Hypothesis E1: The acute relationship between negative affective state and decline in
cortisol level at the subsequent 30 minutes is different on the day with HS condition.
- Hypothesis E2: The acute relationship between positive affective state and decline in
cortisol level at the subsequent 30 minutes is different on the day with HS condition.

42 | P a g e

Chapter 2: The Day-Level Association Between Affective States, Self-Reported Time in
Exercise, and Diurnal Cortisol Rhythm in Adults in The United States
Background
The hypothalamic-pituitary-adrenal (HPA) axis plays a key role in a person’s
physiological responses to stressors. A healthy HPA axis could effectively regulate the
production of cortisol by acutely increase the production of cortisol in response to the physical or
psychological stressors (McEwen, 2008), by promptly terminating the stress response when
stressors are removed (Chrousos, 2000; Gunnar & Quevedo, 2007; Tsigos & Chrousos, 2002),
and by adapting to the same stressors(McEwen & Gianaros, 2010). Chronic exposure to stressors
repeatedly activates the HPA axis and may lead to the development of maladaptive stress
responses and dysregulated HPA axis activities (Dickerson & Kemeny, 2004), both of which are
associated with elevated risks for various adverse health outcomes (Emma K Adam et al., 2017;
Epel et al., 2000; Rosmond et al., 1998). Identifying psychological and behavioral correlates of
maladaptive HPA axis activity patterns could potentially provide insights for interventions aim to
mitigate the impact of stressors on the human neuroendocrine system.
A healthy HPA axis exhibits a regular diurnal rhythm characterized a 50-156% increase
in cortisol levels during the first 30-45 minutes after awakening, or the cortisol awakening
response (CAR)(Clow et al., 2004), and then decline throughout the remainder of the day(R.
Miller et al., 2016). Both the CAR and the rate of decline in cortisol levels from awakening to
bedtime, or diurnal cortisol slope (DCS), are commonly used indices of the diurnal cortisol
rhythms. Although the cortisol levels exhibit a circadian rhythm, both indices of the HPA axis
also exhibit significant within-individual variations (Almeida, Piazza, et al., 2009), which
suggests that these indices could differ according to an individual’s daily experiences. Current
43 | P a g e

evidence has also demonstrated that both CAR and DCS are acutely affected by individuals’
daily psychological experiences. For example, on the day that individuals reported experienced a
stressful event, adults exhibit a flatter DCS on the same day and a stronger CAR on a subsequent
day. Among psychological factors, affective states - an individual’s momentary feelings and
experiences of actual or perceived stimuli (Barrett, 2017; Ekman & Cordaro, 2011)- have been
shown to acutely activate the HPA axis (Susan Folkman, 2013; S. Folkman et al., 1986;
McEwen, 2007). Studies have shown that higher day-level negative affect is associated with a
flatter DCS on the same day (E. K. Adam et al., 2006; Sladek et al., 2017) and a stronger CAR
on a subsequent day (E. K. Adam, 2006; E. K. Adam et al., 2006). On the other hand,
preliminary evidence has also shown that positive affect could be related to adaptive diurnal
cortisol patterns (e.g. steeper DCS on the same day (K. G. Miller et al., 2016; Polk et al., 2005)
and smaller CARs on the subsequent day(K. G. Miller et al., 2016)). However, as null findings
on the relationship between both negative (K. G. Miller et al., 2016; Polk et al., 2005; Smyth et
al., 1998) and positive affective states(Chen et al., 2017; Polk et al., 2005) and HPA axis
activities continue to emerge, more studies may be necessary to further elucidate the role of
affective states on the HPA axis and whether there are other moderators that conceal the affect-
HPA axis relationship (Emma K Adam et al., 2017; Beckie, 2012).
A potential explanation for the mixed findings could be that the current evidence in the
affect-HPA axis relationship focused primarily on how affective states of either valence (i.e.,
positive or negative) are related to cortisol levels without considering the other dimension of
affective states: the state of arousal. The idea of the state of arousal is first described in the
circumplex model of affect (Posner et al., 2005), suggesting that among positive and negative
individuals’ affective experiences, some of these experiences arouse an individual more than
44 | P a g e

others. Therefore, it is possible that affective states with high level of arousal could relate to
HPA axis activities differently than the lower arousal counterpart. However, only a limited
number of studies to date have examined the effects of the arousal components of positive and
negative affect on HPA axis activities. In a recent study, Hoyt et al. reported that high arousal
positive affect, independent of the level of low arousal positive affect, was cross-sectionally
associated with adaptive HPA axis activity patterns (i.e., steeper declines in DCS) in youth in
natural environments (Hoyt et al., 2015). This preliminary result suggests a possible role of
arousal components of the affective states on HPA axis activities. Whether similar findings can
be observed in the adult population remains unexplored. Therefore, the first aim of this study is
to examine whether affective states, as described by the circumplex model of affect, are
associated with diurnal cortisol rhythms.
Additionally, the mixed findings on the affect-HPA axis relationships suggest that there
may be unexplored moderators of this relationship (Campbell & Ehlert, 2012; DeSteno et al.,
2013). Moderate-to-vigorous physical activity (MVPA), in addition to the widely documented
mental (Wegner et al., 2014; C. K. F. Wen et al., 2018) and physical (Kerr et al., 2017) health
benefits, has also been shown to mitigate the negative impacts of laboratory-induced stressors on
the HPA axis (Gerber & Pühse, 2009). Recent studies have shown that habitually active (e.g.,
athletes) individuals react less strongly to standardized stress tests (e.g. Trier Social Stress Test
(TSST)) compare to their less habitually active counterparts (e.g. non-athletes) (Rimmele et al.,
2009; Rimmele et al., 2007) (Puterman et al., 2011) (Martikainen et al., 2013). However, the
evidence on the acute effect of MVPA on physiological stress responses is preliminary. In a
recent study, Wood et al. reported that individuals who took a 30-minute unstructured moderate-
intensity walk, compared to those who did not, reacted less strongly to lab-induced stressors
45 | P a g e

(Wood et al., 2017). This evidence gave support to the cross-stressor adaptation hypothesis,
which asserts that MVPA promotes biological adaptations that contribute to reductions in
reactions of the HPA axis to stressors (Sothmann et al., 1996). Nonetheless, whether the cross-
stressor adaptation hypothesis can be supported at the within-person level (i.e., less HPA axis
reactivity on days when individuals are more active than their usual level) and in the free-living
environment has not been explored. As laboratory-induced stressors could be substantively
different from the psychological stimuli that individuals experienced in free-living situations,
examining whether the affect-HPA axis relationship is moderated by time spent in exercising
using data from the free-living environment could provide results of improved ecological
validity. Therefore, studies that examine the moderating role of exercising on the affect-HPA
axis relationship in free-living environments can provide new and relevant insights.
In summary, this study sought to address two main gaps in the literature on the affect-
HPA axis relationship by examining: 1) the day-level relationship between affective states and
HPA axis activities in mentally healthy adults by quantifying affective states using the
circumplex model of affect and 2) the potential moderator role of self-reported time spent in
exercise on the affect-HPA axis relationship. The conceptual model for this study is presented in
figure 7.

46 | P a g e

Figure 7: Conceptual model for study 1

Methods
Study Sample
Data for this study is from the National Study of Daily Experience (NSDE), a sub-study
of the second wave of the Midlife Development in the United States (MIDUS-II) study. The
MIDUS-II study is a national survey study aimed to examine the role of behavioral,
psychological, and social factors on the physical and mental health of the aging population in the
United States. The sub-study, NSDE, focuses on examining the effect of behavioral and
psychosocial factors, and exposure to day-to-day life stressors on physical and emotional
reactivity by engaging NSDE participants in an 8-day observational study using the daily diary
method (Ryff & Almeida, 2010). From a nationally representative sample of adults ranging in
age from 35-84 recruited to the MIDUS-II study, a random sample of 3600 participants were
47 | P a g e

contacted for participating in the NSDE, of which 2022 respondents completed the NSDE study.
Details of the NSDE sampling procedure has been reported previously (Ryff & Almeida, 2010).
The NSDE study is an 8-day observational study using the daily diary method. For eight
consecutive evenings, NSDE participants were contacted by study interviewers for a daily 10-15
minutes semi-structured phone interview in the evening, inquired about their experience,
including affective states, of that day. Starting on day 2, participants were asked to collect four
saliva samples (i.e., upon waking, 30 minutes post-waking, lunchtime, and bedtime) per day for
four consecutive days (i.e., from Day 2 to Day 5). The timeline of when these data were
collected is illustrated in figure 8.
Of the 2022 participants who completed the 8-day daily diary study, there is a maximum
possible of 8088 days of saliva samples for modeling diurnal cortisol patterns. Of the 8088 days,
612 days were excluded because the second sample of the day (the 30-min post-awakening
sample) were provided either less than 15 min or more than 60 min after waking, 312 days were
excluded because participants were awake for more than 20 hours or less than 12 hours, 71 days
were excluded because the raw cortisol values were above 120 nmol/l, 396 days were excluded
because participants’ lunch scores were higher than their 30-min scores by 10 nmol/l, and 1317
days were excluded because participants woke up before 4 a.m. or after 12 noon. These
Figure 8: Timeline of MIDUS II Data Collection
48 | P a g e

exclusion criteria were in accordance with the current consensus (Stalder et al., 2016) and studies
previously published using this dataset (Piazza et al., 2013; Sin et al., 2017). The analytic sample
of this study included 3618 days of saliva samples with 4 samples each day, representing a total
of 1203 participants in the analytic sample. Among these participants, 71.06% completed at least
3 days of saliva samples, 16.70% completed 2 days of saliva samples, and 12.24% completed
one day of saliva samples.
Measurements
Salivary Cortisol
From day 2 through day 5 of the 8-day study period, the NSDE participants were asked to
collect saliva samples for four times per days (i.e. upon awakening before getting out of bed, 30
minutes after awakening, before lunch, and before bed), resulting in a maximum of 16 samples
throughout the 4-day period. Participants were instructed not to eat, drink, brush their teeth, or
consume any caffeinated items (e.g., coffee, tea, or chocolate) before collecting the saliva
samples. Participants were asked to record the sample collection time on a paper log sent along
with the saliva sample collection kit. The saliva sample collection time was also inquired by the
study interviewers during the end-of-the-day interview on the nights of sample collection.
During the daily interview on the last day of saliva collection, participants were also inquired
about whether they have consumed any prescription and over-the-counter medications on days of
saliva sample collection.
After completing the home saliva sample collection, participants were instructed to send
the samples using a pre-addressed, paid courier package for the return mailing. The return
package was shipped to the MIDUS Biological Core at the University of Wisconsin for freezer
storage at -60 degrees Celsius. Prior to immunoassay, the salivettes are thawed and centrifuged at
49 | P a g e

3000 rpm for 5 minutes. Luminescence immunoassay (IBL, Hamburg, Germany) was used to
determine cortisol concentrations of the samples. The detailed information regarding the saliva
sample procedures and the reliability of the saliva sample collected have been previously
reported (Almeida, McGonagle, & King, 2009)
For this study, the diurnal cortisol slope will be estimated using regression models (for an
overview of commonly used HPA axis indices calculation, please see (Rotenberg et al., 2012)).
The size of the cortisol awakening response of the subsequent day will be operationalized by the
computing dynamic increase in the amount of cortisol secreted in response to awakening, or the
area under the curve relative to increase or awakening cortisol value (AUCi), in accordance with
the current consensus(Stalder et al., 2016),
Affective States
The day-level affective states were measured each day during the end-of-day phone
interview using a modified version of the original Positive and Negative Affective Scale
(PANAS) (Watson, Clark, & Tellegen, 1988). The modified version of PANAS is a 27-item
scale that contains 14 items for negative affect (e.g., How much of the time today did you feel
“nervous?”, “worthless?”, “so sad nothing cheer you up?”, and etc) and 13 items for positive
affect (e.g., How much of the time today did you feel “enthusiastic?”, “attentive?”, “proud?” and
etc). Participants were asked to choose from a 5-point scale with possible response ranged from
“None of The Time,” “A Little of The Time,” “Some of The Time,” “Most of The Time,” to “All
of The Time.” Participants also have the option to respond with “I Do Not Know” or not to
provide an answer.
This study operationalized affective states using two approaches: 1) the valence only
approach (i.e., positive or negative affective states) and 2) the valence and state of arousal
50 | P a g e

approach, which further categorizes positive and negative affective states into a high or low state
of arousal. For the valence only approach, an average score of the 13 positive affective state
items and 14 negative affective state items were calculated for each day, resulted in one average
score each for positive and negative affective state for each day. The Cronbach’s alphas for the
positive and negative affective states scales in the analytic sample were 0.94 and 0.86,
respectively, suggested high internal consistencies within each scales. Similar psychometric
properties have been reported in previous studies using this dataset (Piazza et al., 2013). For the
valence and state of arousal approach, two subscales representing the state of arousal (e.g., high
and low arousal) within each emotional valence (i.e., positive or negative affective states) were
created based on the results of principal component analyses with orthogonal rotation. For both
positive and negative affective states across the 8-day study period, factors with an eigenvalue
greater than 1.0 and items with a factor loading of 0.65 or greater will be considered. Results of
factor analyses suggest that the four affect scales identified are aligned with the circumplex
model of affect (Posner et al., 2005) in that the four variable represent high arousal positive
affect (“Did you feel in good spirits?” and “Did you feel cheerful?”, standardized Chronbach’s
alpha = 0.87), low arousal positive affect (“Did you feel calm and peaceful?” and “Did you feel
satisfied?”, standardized Chronbach’s alpha = 0.84), high arousal negative affect (“Did you feel
irritable?”, “Did you feel upset?”, “Did you feel angry?”, “Did you feel frustrated?” standardized
Chronbach’s alpha = 0.82), and low arousal negative affect (“Did you feel worthless?”, “Did you
feel so sad nothing cheer up?”, ”Did you feel hopeless?”, standardized Chronbach’s alpha =
0.75). An average score for each of the four affective states (i.e., low arousal positive affect, high
arousal positive affect, low arousal negative affect, and high arousal positive affect) was
calculated for each day. Originally, this study proposed to also examine the relationship between
51 | P a g e

each individual item of the PANAS scale and HPA axis activities as the third approach. Briefly,
this approach involved treating each item of the modified PANAS as an individual predictor in
the models. Results of this approach, however, were later excluded from the current report due to
potential risks for type 1 error as a result of multiple comparisons.
For this study, two versions of affective states were calculated for modeling the effect of
trait-like and state-like affective states on diurnal cortisol activity patterns. For trait-like affective
states, the average scores for each affective state scales (i.e., positive/negative affect, high
arousal positive/negative affect, and low arousal positive/negative affect) were calculated for
each individual participant to create personal average scores of affective states. The personal
average scores for the affective states scales were then centered by an average of the
corresponding affective state scores across the analytic sample. The centered version of the
personal average score represents the trait-like affective states of each participant across the 8-
day period relative to other participants included in the analytic sample and herein denoted as the
between-person (BP) version of affective states. To examine the relationship between state-like
affective states and the diurnal cortisol rhythm, within-person versions of the affective state were
created by subtracting the affective state scores of each day by the personal average of the
corresponding affective state scores. The resulting scores represent participants’ levels of
affective states on any given day within the study period relative to the average level of the
corresponding affective state of that participant across the 8-day period. This version of the
affective state scores is herein denoted as the within-person (WP) versions of affective state.
Both the BP and the WP versions were date-linked to the days when participants collected saliva
samples and were included in the same model, which permits disaggregating the inter- and intra-
individual variabilities in outcomes of interest (Curran & Bauer, 2011).
52 | P a g e

Self-Reported Time Spent in Exercise
Daily time spent in exercising was assessed via participant self-report during the end-of-
the-day phone interview. During the daily phone interview, the NSDE participants were asked:
“since (this time/we spoke) yesterday, how much time did you spend engaged in vigorous
physical activity or exercise.” The participants were asked to estimate the time spent in physical
activity in the unit of hours and minutes. For this study, the total daily amount of time spent in
vigorous physical activity or exercise is expressed in the unit of minutes. As time-spent in
exercise may fluctuate across the 8-day study period, both the between- and within-person
versions of time spent in exercising were calculated and included in the same models. The
proposed moderation effects were evaluated by the inclusion of interaction terms constructed
with the main predictors of the models.
Covariates
The analytic models included variables that are considered correlates of diurnal cortisol
rhythm, including demographic characteristics and socioeconomic status (SES). Demographic
characteristics, including age, gender, and race/ethnicity, and socioeconomic status,
approximated by self-reported education level were self-reported by the participants prior to the
start of the 8-day daily diary study. In addition to the covariates described in the proposal, other
relevant variables, including self-rated well-being, self-rated physical health, the presence of
major depressive symptoms, and presence of daily stressors were also included in the final
analytic models. Self-rated well-being and physical health were two separate 1-item questions
where participants self-rated their general well-being and physical health as poor, fair, good, very
good, or excellent. Presence of major depressive symptoms was defined as presence of
depressive mood, anhedonia, or loss of interest for most of the day or nearly every day in the past
53 | P a g e

year and at least four of the related symptoms (i.e., loss of energy, loss of appetite, trouble falling
asleep, trouble concentrating, felt worthless, or suicidal ideas) during a 2-week period (Wang,
Berglund, & Kessler, 2000). Presence of daily stressors was assessed using the Daily Inventory
of Stressful Events (Almeida, Wethington, & Kessler, 2002). At the end of the day interview,
participants were inquired about whether the following stressors had occurred in the past 24
hours: “had an argument”, “avoided an argument”, “stressor at work or school”, “stressor at
home”, “discrimination that happened to you”, “stressor happened to friend that stressed you”, or
“other stressor”. A dichotomous variable was created to indicate the days in which participants
reported experiencing at least one of these stressors.
Statistical Analysis
The current study utilized the multi-level modeling approach to examine the relationships
between affective states and diurnal cortisol patterns. The multi-level model approach allows for
parameter estimations accounting for the non-independent nature of the outcome variables (i.e.,
salivary cortisol levels) and supports examination of the within- and between-person association
between affective states and the diurnal cortisol rhythm (E. K. Adam et al., 2006; Curran &
Bauer, 2011). Multi-level models were estimated using full-information maximum likelihood
estimation using PROC MIXED in SAS v.9.4 (Cary, NC, USA).
As cortisol data were nested within days and persons, three-level models were used to
estimate the relationships between day-level affective states on the diurnal cortisol slope of the
same day. In these models, the outcome variable was the log-transformed cortisol level at any
given sampling time. Therefore, parameter estimates resulted from this modeling approach
should be interpreted after the parameter estimates were back-transformed and should be
interpreted as percent changes in cortisol (E. K. Adam, 2006). In the three-level models, level 1
54 | P a g e

included only variables at the individual saliva sample level, including time since waking, a
quadratic term for time since waking, and a dummy variable for the cortisol awakening response.
Only the 30-minute post awakening sample of each day was coded as 1 for this dummy variable,
while other samples were coded as 0. Level 1 of the analytic model was specified as
𝐶𝑜𝑟𝑡𝑖𝑠𝑜𝑙 = 𝜋 + 𝜋 (𝐶𝐴𝑅 ) + 𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 )
+ 𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 ) + 𝑒
, where 𝜋 represented the log-transformed cortisol level at waking, 𝜋 represents the percent
change in cortisol level between waking and 30 minutes post-awakening, 𝜋 and 𝜋 represent
the linear and quadratic percent changes in cortisol per hour since awakening, respectively.
The level 2 model included variables that vary at a day level (e.g., within-person affective states
score) and was specified as:
Level 2: 𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑟
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
, where 𝛽 , 𝛽 , 𝛽 , and 𝛽 represent the percent changes in waking level, CAR, linear
slope, and quadratic slope associated with the day-level fluctuation in affective states. A random
intercept 𝑟 was also specified in the model to allow intercept to vary within MIDUS
participants across days. Level 3 of the analytic model included variables that vary across
individual participants (e.g., the between-person version of the affective state scores) and was
specified as followed:
𝛽 = 𝛾 + 𝛾 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑢
𝛽 = 𝛾 + 𝛾 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
55 | P a g e

𝛽 = 𝛾 + 𝛾 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑢
𝛽 = 𝛾 + 𝛾 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
, where 𝛾 , 𝛾 , 𝛾 , and𝛾 represent the change in waking level percent changes in
waking level, CAR, linear slope, and quadratic slope associated with the differences in affective
states between participants.
Two-level models were used to examine the effect of day-level affective states on CAR of
the subsequent day. CAR, in accordance with the current recommendations(Stalder et al., 2016),
was defined as the dynamic increase portion of the CAR (AUCi). In these models, both the
within-person and the between-person versions of each respective affective state were included
in the models as independent variables, with CAR of the subsequent day as the dependent
variable. Therefore, level 1 of the analytic model included variables that vary by day, such as
within-person affective states score. The level 1 model was specified as:
𝐶𝐴𝑅 ( ) = 𝜋 + 𝜋 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑒
, where 𝜋 represented the average CAR on the morning (d+1) subsequent to the affective state
data collection on the previous evening (d) and 𝜋 represented the slope of change associated
with within-person levels of affective states. Level 2 of the analytic model included variables
that vary across individual participants, such as between-person affective state scores. The level
2 model was specified as:
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 ) + 𝑟
𝜋 = 𝛽
, where 𝛽 represented the sample average CAR, 𝛽 represented the association between
between-person affective states and CAR, and 𝛽 represents the sample average slope. In both
2-level and 3-level models, the within-person time spent in exercise included in the day-level
56 | P a g e

(i.e. level 2 in the 3-level model and level 1 in the 2-level model) of the analytic models, whereas
the between-person time-spent in exercising were included in the person-level (i.e., level 3 in the
3-level model and level 2 in the 2-level model).
Interaction terms were constructed by multiplying the main predictor of the model with
time-spent in exercising at the same level (e.g., within-person affect* within-person time-spent in
exercising) to investigate the potential moderation effect of time spent in exercise. Additional
analyses that examined the relationship between days of extremely high variations in affective
states and diurnal cortisol rhythms were conducted on a post-hoc basis. In this post-hoc analysis,
quadratic terms of the main affective state predictor were included in the model to estimate the
effect of extreme affective states on the diurnal cortisol rhythm. Presence of major depressive
symptoms and presence of daily stressors were included at the day-level of the analytic models
as covariates. Age, gender, ethnicity, education level self-rated well-being, self-rated physical
health, were also included in the person-level of the analytic models as covariates.
Results:
Descriptive Statistics
The analytic sample for study 1 included 1203, predominantly white (92.25%) adults
with an average age of 55.93 ± 12.03 years old (range: 33.00-84.00). More than half of the
analytic sample were female (56.61%). Approximately half (49.33%) of the analytic sample
completed at least college education. Most rated their physical health (n=987, 89.00%) and
mental health (n=1035, 93.33%) as good, very good or excellent, and only a small fraction
(6.57%) of the participant's self-reported symptoms of the major depressive disorder (Table 1).
Descriptive statistics for the main predictors of interest, including average positive and negative
affect, high arousal positive and negative affect, and low arousal positive and negative affect, are
57 | P a g e

presented in Table 2. In this study, participants self-reported spending an average of 40.55 ±
82.97 minutes in physical activity or exercise per day during the study period.On the days of
cortisol collection, the average self-reported wake time was 6:37AM±1:17 hours. On average,
the time of first saliva sample was 6:43AM±1:21 hours. Descriptive statistics of the raw and
transformed cortisol value for the four sampling times (i.e., waking, 30-minute post-awakening,
lunch, and bedtime) are presented in Table 3.
Affective States and the Same Day Diurnal Cortisol Slope
Three-level models were used to examine whether average, high-arousal, or low-arousal
positive or negative affective states were associated with diurnal cortisol slope. After controlling
for demographic characteristics and other relevant covariates, none of the within-person affective
state predictors were associated with diurnal cortisol rhythm on the same day. At the between-
person level, nonetheless, average positive affective states score was associated with a slower
rate of deceleration in cortisol level (𝑒 . -1=-0.068%, p=0.0235, model 1.3 in table 4),
such that individuals with one standard deviation higher in average positive affective state score
exhibited a 0.07% slower rate of deceleration in cortisol levels throughout the day. Furthermore,
high arousal positive affective states were associated with a 0.07% slower rate of deceleration in
cortisol levels (𝑒 . -1=-0.068%, p=0.0278, model 2.3 in table 4), while low arousal
positive affective states were associated with both a 1.19% slower rate of decline in cortisol
levels (𝑒 . -1=1.19%, p=0.0297, model 3.3 in table 4) and a 0.09% slower rate of deceleration
in cortisol level (𝑒 . -1=-0.090%, p=0.0016, model 3.3 in table 4) throughout the day.
Results for the post-hoc models that included quadratic terms for the respective affective states
score were shown in table 4 (model 1.4, 2.4, 3.4). The parameter estimates were similar after
58 | P a g e

inclusion of the quadratic terms for affective states. None of the negative affective states
predictors were associated with DCS (data not shown).
Affective States and the Subsequent-Day Cortisol Awakening Response
Two-level models were used to examine the relationship between day-level affective
states scores predict CAR of the subsequent day. After controlling for demographic
characteristics and relevant covariates, high arousal positive affective state score was negatively
associated CAR of the subsequent day at the within-person level (Est=-0.0211, SE=0.01019,
p=0.0384, model 8 in table 5), such that individuals exhibited a smaller CAR after the days when
they reported a higher level of high arousal positive affect than their usual levels. Neither
average positive affect nor low arousal positive affect scores were associated with CAR of the
subsequent day. Other the other hand, low arousal negative affective state score was associated
with higher CAR of the subsequent day at the between-person level (Est: 0.09527, SE: 0.3857,
p=0.0135, model 6 in table 5). Neither average negative affect nor high arousal negative affect
scores were associated with CAR of the subsequent day.
Time Spent in Exercising as a Potential Moderator for Diurnal Cortisol Rhythms
The relationships between affective states and indices of diurnal cortisol rhythm on the
same day (i.e., the diurnal cortisol slopes and rate of cortisol deceleration) or the subsequent day
(i.e., the cortisol awakening response) were not moderated by participants’ self-reported time
spent in exercise. The interaction terms in each of the respective models were not statistically
significant, and the magnitude of the significant relationship between affective states and diurnal
59 | P a g e

Table 1: Participant Demographic Characteristics

Analytic Sample (N=1203) n (%)
Gender
Male 522 43.39
Female 681 56.61
Ethnicity
White 1023 92.25
African American 32 2.89
Native American or Alaska Native 19 1.71
Asian 3 0.27
Other 28 2.52
Don know 3 0.27
Refused 1 0.09
Education Level
Some High School 50 4.52
High School Degree 253 22.85
Some College 258 23.31
College Degree 316 28.55
Master Degree 176 15.90
Doctoral Degree 54 4.88
Self-Rated Physical Health
Good, Very Good, or Excellent 987 89.00
Fair 88 7.94
Poor 34 3.07
Self-Rated Mental/Emotional Health
Good, Very Good, or Excellent 1035 93.33
Fair 64 5.77
Poor 10 0.90
Symptoms of Major Depressive Disorder
Yes 79 6.57
No 1124 93.43
60 | P a g e

Table 2: Descriptive Statistics for Affective States

Table 3: Descriptive Statistics for Raw and Log-Transformed Cortisol Values by Sampling Time
Sampling
Time
Raw Values Log-Transformed Values
Mean
(nmol/L)
Standard
Deviation
(nmol/L) Skewness Kurtosis Mean
Standard
Deviation Skewness Kurtosis
Waking
Sample 15.91 10.24 38.76 1754.17
2.52 0.81 -1.51 13.22
30-M Post
Awakening 23.17 13.14 82.36 6785.33
2.87 0.77 -0.65 8.48
Lunch
Time 6.79 4.57 15.48 302.51
1.79 0.78 0.09 4.41
Bedtime 3.31 5.46 30.24 1106.23
0.75 1.12 0.68 3.11

Variable Labels Mean Std. Dev. Min Max
Average positive affective states 2.737 0.790 0 4
High arousal positive affective states 2.913 0.851 0 4
Low arousal positive affective states 2.844 0.882 0 4
Negative affect average 0.194 0.325 0 3.5
High arousal negative affective states 0.308 0.529 0 4
Low arousal negative affective states 0.066 0.301 0 4
61 | P a g e

cortisol rhythm identified in the previous sections were not altered after the inclusion of the
interaction terms.
Discussion
This study examined relationships between day-level affective states and diurnal cortisol
rhythms, as well as the potential moderating effect of self-reported time spent in exercise among
adults living in the United States. We found that high arousal positive affect was related to lower
CAR of the subsequent day at the within-person level. This study is among the first to examine
the relationship between the arousal components of affective states on CAR and diurnal cortisol
slopes. However, the results did not find a significant association between negative affective
states and diurnal cortisol slope, nor did the result not support the assertion that self-reported
time spent in exercise could moderate the relationship between affective states of either valence
and diurnal cortisol rhythms. Nonetheless, by investigating this relationship both at the between-
and within-person level and further examining the potential effect of state of arousal on HPA
axis activities, results of this study contributed to the understanding of whether day-to-day
fluctuation in affective states, specifically the ones that are potentially highly activating, is
associated with indices of diurnal cortisol rhythms.
The null day-level relationship between average positive affect and the CAR conforms
with results from other studies conducted on the adult population (Polk et al., 2005), but not
others(K. G. Miller et al., 2016). Despite this null relationship, results of this study indicate that
regardless of the level of low arousal affective states, adults exhibit a smaller CAR on the day
after they had reported a higher high arousal positive affective state than their personal average.
The discrepancies between results for average positive affect and for high arousal positive affect
offer preliminary evidence showing that only parts of the positive affect that are high in their
62 | P a g e

arousal components are associated with adaptive HPA axis activity patterns. Although the exact
mechanism underlying this within-person relationship is unclear, it is possible that high arousal
affective states alleviate the effect of other psychological burdens that are shown to relate with
higher CAR, including anticipation of demands on the upcoming days (Fries et al., 2009). On the
other hand, the results of this study did not support the relationship between any of the positive
affective states (i.e., average, high arousal, or low arousal) and DCS of the same day. Although
high arousal positive affective states may affect diurnal cortisol slope in adolescents (Hoyt et al.,
2015), neither high nor low arousal positive or negative affective states were found to be
associated with DCS of the same day among adults included in this analysis. We could only
speculate reasons for the divergent results documented in the adolescent population and in this
study. One such reason could be that the extents to which affective states are related to HPA axis
activities are different in adolescents compared to adults because affective states in adults are
less volatile than that of adolescents (Hoppmann & Riediger, 2009; Larson & Richards, 1994).
Therefore, it is possible that the effect of positive affective states as psychological stimuli does
not manifest in adults’ HPA axis immediately on the same day but the next morning. Altogether,
these results suggest that positive affective states that are on the high activation end in the
spectrum of affective states are acutely related to HPA axis activities at the within-person level.
Contrary to findings from other studies in adults(E. K. Adam et al., 2006), results of this
study did not support a day-level relationship between negative affect (i.e., average, high arousal,
or low arousal) and indices of the HPA axis activities. One of the explanations for the null
findings could be that the low variation and average in negative affect scores observed in this
item were not potent enough to activate the HPA axis. Although the within-person approach was
utilized to examine whether the diurnal cortisol rhythms are different on days when participants
63 | P a g e

reported higher or lower negative affect than usual, the low variation may have hindered the
abilities of this analysis to identify such relationship. In lights of other similar null findings, this
study also examined whether self-reported time spent in exercise could be a potential moderator
to this relationship.
While the null findings from the moderator analyses may suggest that physical activity
does not moderate the relationship between affective states and diurnal cortisol rhythm, we are
hesitant to draw this conclusion. In this study, both the time and intensity of exercise were
captured by self-report. Although self-reported physical activity is feasible for assessing activity
level in large-scale studies, it is subject to overestimation due to participant over-report, both in
the amount and in the intensity (Spruijt-Metz et al., 2013). Studies with physical activity
measured in other objective modalities (i.e. accelerometers, indirect calorimetry, or direct
observation) showed that physical activity level could attenuate the impact of affective states on
the HPA axis (Rimmele et al., 2009; Rimmele et al., 2007) (Puterman et al., 2011) (Martikainen
et al., 2013; Wood et al., 2017). Therefore, future studies with objectively assessed physical
activity could potentially help elucidate the relationship between affective states, physical
activity, and diurnal cortisol rhythms. Furthermore, as the intra-class coefficient in the final
models suggests a 59% unexplored variation at the within-person level, this result suggests that
future studies examining other moderators or mediators may help elucidate the relationship
between negative affective states and diurnal cortisol rhythms.
While identifying the day-level relationship between high arousal positive affect and the
CAR of the subsequent day offers novel contributions to the literature, some of the limitations
this study may preclude this study from generating other insights. One notable limitation of this
study is that affective states were only collected once at the end of the day. While affective states
64 | P a g e

assessed at the end-of-day could offer summative insights to an individual’s emotional
experiences of that particular day, they may not be able to capture the nuanced fluctuations that
an individual actually experiences as these affective states occur. Affective states are individuals’
experiences toward actual or perceived stimuli (Barrett, 2017). As these experiences could
activate the HPA axis on a momentary basis, studies that capture individuals’ experiences as they
unfold in ambulatory settings could potentially offer more insights to whether and how affective
states relate to the HPA axis. Secondly, the participants in this study were predominantly white.
Both the individuals’ emotional experiences and how HPA axis reacts to stressors (DeSantis,
Adam, Hawkley, Kudielka, & Cacioppo, 2015; Zeiders, Hoyt, & Adam, 2014) in ethnic minority
populations can be different from that of Caucasians. Therefore, despite the novel contribution,
the generalizability of this result to other ethnic minority population is yet to be established.
In summary, this study examined whether affective states are related to the diurnal cortisol
rhythm of adults. While the relationship between positive affect on the relationship between
positive affect and CAR has been previously documented (Dockray & Steptoe, 2010; K. G.
Miller et al., 2016), this study is among the first to examine whether the fluctuation in positive
affects with different arousal components relate to the CAR of the subsequent day in the adult
population. This result provides preliminary evidence on the differential effect of specific aspects
of affective states beyond their valence (i.e., positive or negative) on adults’ diurnal cortisol
rhythm. Furthermore, the acute effect of high arousal positive affect on CAR may also suggest
that feeling of cheerful and other related affective states are salient targets for future
interventions that aim to reduce the impact of cortisol exposure and health.

65 | P a g e

Table 4: Daily Affective States Predicting Diurnal Cortisol Slope of the Same Day

Cortisol Level At Waking
Estimate (SE)
Cortisol Awakening
Response
Estimate (SE)
Time Since Awakening
Estimate (SE)
Time Since Awakening
2

Estimate (SE)
Main Predictor: Average Positive Affect
Model 1.1: Minimally-adjusted model
Intercept 2.5572*** 0.01597 0.518*** 0.01179 -0.159*** 0.00333 0.00261*** 0.00018
Daily Positive Affect (BP) -0.0076 0.02278 0.0244 0.01699 0.00630 0.00472 -0.000420 0.00025
Daily Positive Affect (WP) 0.009 0.02824 -0.0488 0.034 -0.0133 0.00874 0.000829 0.0005
Model 1.2: Model including time-spent in exercise and interaction terms
Intercept 2.5558*** 0.01599 0.5185*** 0.01186 -0.158*** 0.00335 0.00257*** 0.00018
Daily Positive Affect (BP) -0.0117 0.02288 0.029 0.01714 0.00739 0.00478 -0.000490 0.00025
Daily Positive Affect (WP) 0.01069 0.02828 -0.0487 0.03406 -0.0137 0.00877 0.000821 0.00051
Daily Time-Spent in
Exercise (BP)
0.00041 0.00029 -0.0004 0.00021 -0.0000500 5.3E-05 0.0000022 2.54E-06
Daily Time-Spent in
Exercise (WP)
-0.0004* 0.00017 0.00025 0.00021 0.000133** 5.1E-05 -0.0000050 2.91E-06
Daily Positive Affect*
Daily Time-Spent in
Exercise (BP)
0.0002 0.00033 -5E-05 0.00024 -0.0000700 5.9E-05 0.0000027 2.77E-06
Daily Positive Affect*
Daily Time-Spent in
Exercise (WP)
-0.0002 0.00051 0.00054 0.00061 0.0000260 0.00015 -0.0000024 9.09E-06
Model 1.3: Fully-adjusted model
Intercept 2.5761*** 0.1037 0.4406 *** 0.08005 -0.139*** 0.02248 -0.000160 0.0012
Daily Positive Affect (BP) -0.0368 0.02598 0.01178 0.02004 0.00900 0.00562 -0.000680* 0.0003
Daily Positive Affect (WP) -0.0033 0.02986 -0.0344 0.036 -0.00887 0.00932 0.000651 0.00054
Daily Time-Spent in Exercise
(BP)
0.00021 0.0003 -0.0003 0.00023 -0.0000600 0.00006 0.0000033 3.10E-06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00019 0.00014 0.00022 0.000133* 5.7E-05 -0.0000056 3.22E-06
Daily Positive Affect*
Daily Time-Spent in Exercise
(BP)
-5E-05 0.00037 0.00019 0.00029 -0.0000300 7.3E-05 0.0000040 3.73E-06
Daily Positive Affect*
Daily Time-Spent in Exercise
(WP)
-0.0004 0.00055 0.0008 0.00065 0.0000850 0.00017 -0.0000058 0.00001
Gender (Male=1) -0.0693* 0.03274
0.08562
***
0.02513 -0.0305*** 0.00706 0.00190*** 0.00038
Age (per 10 years) 0.04028** 0.0133 0.01874 0.01021 0.00929** 0.00292 -0.000180 0.00016
Self-Rated Emotional Health -0.0707*** 0.01981 0.00348 0.01534 0.0103* 0.0043 -0.000190 0.00023
Self-Rated Physical Health 0.0233 0.0221 -0.01 0.01705 -0.00485 0.00479 0.000159 0.00026
Education 0.03393** 0.01347 -0.0381 *** 0.01035 -0.00857** 0.00293 0.000315* 0.00016
Depression (Yes=1) -0.124 0.06801 0.04034 0.05321 0.0237 0.01486 -0.00148 0.00081
Experience Daily Stressor
(Yes=1)
0.02247 0.02177 0.04096 0.02541 0.0170** 0.00651 -0.000910* 0.00037
Model 1.4: Fully-adjusted model plus modeling extreme emotion
Intercept 2.5764*** 0.1038 0.4397*** 0.08004 -0.139*** 0.02248 -0.00017 0.0012
66 | P a g e

Daily Positive Affect (BP) -0.0373 0.026 0.01201 0.02005 0.00884 0.00562 -0.00067* 0.0003
Daily Positive Affect (WP) -0.0078 0.0307 -0.0323 0.03697 -0.0113 0.00956 0.000854 0.00055
Daily Positive Affect*
Daily Positive Affect (WP)
-0.0238 0.04041 0.01056 0.04665 -0.0149 0.01207 0.00123 0.00069
Daily Time-Spent in Exercise
(BP)
0.0002 0.0003 -0.0003 0.00023 -0.00006 0.00006 0.00000327 3.10E-06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00019 0.00014 0.00022 0.000131* 5.7E-05 -0.00000555 3.22E-06
Daily Positive Affect*
Daily Time-Spent in Exercise
(BP)
-5E-05 0.00037 0.00019 0.00029 -0.00003 7.3E-05 0.00000394 3.73E-06
Daily Positive Affect*
Daily Time-Spent in Exercise
(WP)
-0.0004 0.00056 0.00077 0.00066 0.00009 0.00017 -0.00000647 0.00001
Gender (Male=1) -0.0683* 0.03278 0.08529*** 0.02516 -0.0301*** 0.00707 0.00187*** 0.00038
Age (per 10 years) 0.0405** 0.01331 0.01869 0.01021 0.00941** 0.00292 -0.00019 0.00016
Self-Rated Emotional Health -0.0705*** 0.01982 0.00341 0.01534 0.0105* 0.0043 -0.00021 0.00023
Self-Rated Physical Health 0.02361 0.02211 -0.01 0.01705 -0.00479 0.00479 0.000153 0.00026
Education 0.03368* 0.01348 -0.038*** 0.01036 -0.00859** 0.00293 0.000317* 0.00016
Depression (Yes=1) -0.1212 0.06812 0.03902 0.05339 0.0248 0.01489 -0.00157 0.00081
Experience Daily Stressor
(Yes=1)
0.02263 0.02177 0.04078 0.0254 0.0170** 0.00651 -0.00091* 0.00037
Main Predictor: High Arousal Positive Affect
Model 2.1: Minimally-adjusted model
Intercept 2.5572*** 0.01597 0.518*** 0.0118 -0.159*** 0.00333 0.00262*** 0.00018
Daily High Arousal Positive
Affect (BP)
-0.0028 0.02381 0.0151 0.01783 0.00490 0.00498 -0.000390 0.00027
Daily High Arousal Positive
Affect (WP)
0.00113 0.01915 -0.0202 0.02315 -0.00540 0.00593 0.000256 0.00034
Model 2.2: Model including time-spent in exercise and interaction terms
Intercept 2.556*** 0.01595
0.5187
***
0.01182 -0.159*** 0.00334 0.00259*** 0.00018
Daily High Arousal Positive
Affect (BP)
-0.0064 0.02384 0.01873 0.01792 0.00583 0.00502 -0.000440 0.00027
Daily High Arousal Positive
Affect (WP)
0.00094 0.01917 -0.0193 0.02316 -0.00513 0.00595 0.000231 0.00034
Daily Time-Spent in Exercise
(BP)
0.00041 0.00028 -0.0004 0.00021 -0.0000500 5.1E-05 0.0000025
2.43E-
06
Daily Time-Spent in Exercise
(WP)
-0.0004* 0.00017 0.00023 0.00021 0.000129* 5.2E-05 -0.0000049
2.92E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
0.00029 0.00034 -0.0001 0.00025 -0.0000700 6.2E-05 0.0000022
2.95E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00038 0.00051 0.00044 0.0000560 0.00012 -0.0000036
6.70E-
06
Model 2.3: Fully-adjusted model
Intercept 2.5715*** 0.1033 0.4446*** 0.07971 -0.139*** 0.02239 -0.000210 0.0012
Daily High Arousal Positive
Affect (BP)
-0.0443 0.0268 0.00484 0.0207 0.00940 0.00581 -0.000680* 0.00031
Daily High Arousal Positive
Affect (WP)
-0.0086 0.02019 -0.0118 0.02438 -0.000900 0.00631 0.0000280 0.00036
67 | P a g e

Daily Time-Spent in Exercise
(BP)
0.00018 0.00029 -0.0003 0.00023 -0.0000600 5.7E-05 0.0000039
2.89E-
06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00018 0.00014 0.00022 0.000136* 5.6E-05 -0.0000058
3.20E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
0.0001 0.00039 9.5E-05 0.00032 -0.0000400 0.00008 0.0000037
4.18E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0002 0.00038 0.0006 0.00045 0.0000390 0.00012 -0.0000030
6.95E-
06
Gender (Male=1) -0.0687* 0.03273
0.08623**
*
0.02514 -0.0307*** 0.00706 0.00190*** 0.00038
Age (per 10 years) 0.04066** 0.01328 0.01939 0.0102 0.00952** 0.00291 -0.000200 0.00016
Self-Rated Emotional Health -0.0706*** 0.01979 0.00338 0.01533 0.0102* 0.00429 -0.000190 0.00023
Self-Rated Physical Health 0.02364 0.02186 -0.0121 0.01688 -0.00523 0.00475 0.000194 0.00026
Education 0.03463* 0.01341
-
0.0387***
0.01032 -0.00882** 0.00292 0.000334* 0.00016
Depression (Yes=1) -0.1317 0.06827 0.04226 0.05341 0.0250 0.01492 -0.00156 0.00081
Experience Daily Stressor
(Yes=1)
0.02231 0.02174 0.04133 0.02538 0.0175** 0.0065 -0.000960** 0.00037
Model 2.4: Fully-adjusted model plus modeling extreme emotion
Intercept 2.5699*** 0.1033 0.4475
***
0.07972 -0.138*** 0.02239 -0.00026 0.0012
Daily High Arousal Positive
Affect (BP)
-0.0435 0.02687 0.00227 0.02079 0.00857 0.00583 -0.00063* 0.00031
Daily High Arousal Positive
Affect (WP)
-0.0047 0.0215 -0.0228 0.02586 -0.00464 0.0067 0.000271 0.00039
Daily High Arousal Positive
Affect*
Daily High Arousal Positive
Affect (WP)
0.00905 0.01842 -0.0267 0.02143 -0.00968 0.00563 0.000639 0.00033
Daily Time-Spent in Exercise
(BP)
0.00018 0.00029 -0.0003 0.00023 -0.00006 5.7E-05 0.00000384 2.89E-
06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00018 0.00015 0.00022 0.000139* 5.6E-05 -0.00000597 3.20E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
9.8E-05 0.00039 9.6E-05 0.00031 -0.00003 0.00008 0.00000363 4.18E-
06
Daily High Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0002 0.00039 0.00055 0.00045 0.000029 0.00012 -0.00000235 6.95E-
06
Gender (Male=1)
-0.0687* 0.03274 0.08633**
*
0.02513 -0.0306*** 0.00706 0.00190*** 0.00038
Age (per 10 years)
0.04067** 0.01329 0.01971 0.0102 0.00963*** 0.00291 -0.00021 0.00016
Self-Rated Emotional Health
-0.0706** 0.0198 0.00359 0.01532 0.0103* 0.00429 -0.00019 0.00023
Self-Rated Physical Health
0.02325 0.02187 -0.0112 0.01689 -0.00489 0.00475 0.000173 0.00026
Education
0.03463** 0.01342 -
0.0386***
0.01032 -0.00877** 0.00292 0.00033* 0.00016
Depression (Yes=1)
-0.1327 0.06832 0.04558 0.05346 0.0260 0.01493 -0.00163* 0.00081
Experience Daily Stressor
(Yes=1)
0.02212 0.02174 0.04143 0.02537 0.0176** 0.00649 -0.00096** 0.00037
Main Predictor: Low Arousal Positive Affect
68 | P a g e

Model 3.1: Minimally-adjusted model
Intercept 2.5567*** 0.01596 0.5181*** 0.0118 -0.159*** 0.00333 0.00262*** 0.00018
Daily Low Arousal Positive
Affect (BP)
0.00377 0.02248 0.0172 0.01679 0.00946* 0.00466 -0.000650** 0.00025
Daily Low Arousal Positive
Affect (WP)
-0.0183 0.01911 -0.0139 0.02319 -0.00469 0.00588 0.000415 0.00034
Model 3.2: Model including time-spent in exercise and interaction terms
Intercept 2.555*** 0.01593 0.519*** 0.01181 -0.158*** 0.00334 0.00258*** 0.00018
Daily Low Arousal Positive
Affect (BP)
0.00019 0.0225 0.01975 0.01687 0.0106* 0.0047 -0.000710** 0.00025
Daily Low Arousal Positive
Affect (WP)
-0.0191 0.01911 -0.0131 0.0232 -0.00415 0.00589 0.000378 0.00034
Daily Time-Spent in Exercise
(BP)
0.00041 0.00028 -0.0004 0.00021 -0.0000500 5.1E-05 0.0000026
2.44E-
06
Daily Time-Spent in Exercise
(WP)
-0.0004* 0.00017 0.00027 0.00021 0.000133** 0.00005 -0.0000051
2.87E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
0.00037 0.00033 -2E-05 0.00024 -0.0000800 0.00006 0.0000039
2.84E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0002 0.00036 0.0002 0.00043 -0.0000300 0.00011 0.0000025
6.45E-
06
Model 3.3: Fully-adjusted model
Intercept 2.5658*** 0.1031 0.4437*** 0.0796 -0.139*** 0.02235 -0.000170 0.00119
Daily Low Arousal Positive
Affect (BP)
-0.0268 0.02502 0.00609 0.0194 0.0118* 0.00541 -0.000910** 0.00029
Daily Low Arousal Positive
Affect (WP)
-0.0323 0.02029 -0.0029 0.02457 0.000785 0.00628 0.000179 0.00036
Daily Time-Spent in Exercise
(BP)
0.00016 0.00029 -0.0003 0.00023 -0.0000600 5.7E-05 0.0000034
2.86E-
06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00018 0.00019 0.00022 0.000135** 5.5E-05 -0.0000058
3.15E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
0.00014 0.00036 0.00024 0.00029 -0.0000600 7.3E-05 0.0000060
3.80E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0002 0.00038 0.00024 0.00045 0.0000110 0.00012 0.0000002
6.84E-
06
Gender (Male=1) -0.0681* 0.03269
0.08548**
*
0.0251 -0.0306*** 0.00705 0.00190*** 0.00038
Age (per 10 years) 0.03946** 0.01333 0.01918 0.01024 0.00890** 0.00292 -0.000150 0.00016
Self-Rated Emotional Health -0.0692*** 0.01973 0.00351 0.01528 0.0100* 0.00428 -0.000170 0.00023
Self-Rated Physical Health 0.02517 0.02194 -0.0115 0.01694 -0.00434 0.00477 0.000116 0.00026
Education 0.03548** 0.01342
-
0.0387***
0.01032 -0.00866** 0.00292 0.000317* 0.00016
Depression (Yes=1) -0.1252 0.068 0.04065 0.05322 0.0250 0.01485 -0.001580 0.00081
Experience Daily Stressor
(Yes=1)
0.01478 0.02192 0.04395 0.0256 0.0197** 0.00655 -0.00105** 0.00037
Model 3.4: Fully-adjusted model plus modeling extreme emotion
Intercept 2.5668*** 0.1032 0.4441*** 0.07966 -0.139*** 0.02236 -0.00018 0.0012
69 | P a g e

Daily Low Arousal Positive
Affect (BP)
-0.0277 0.02511 0.00573 0.01954 0.0117* 0.00544 -0.0009** 0.00029
Daily Low Arousal Positive
Affect (WP)
-0.0346 0.02095 -0.0041 0.02536 0.000443 0.00647 0.000213 0.00037
Daily Low Arousal Positive
Affect*
Daily Low Arousal Positive
Affect (WP)
-0.0086 0.02134 -0.0041 0.02494 -0.00134 0.00633 0.000132 0.00036
Daily Time-Spent in Exercise
(BP)
0.00016 0.00029 -0.0003 0.00023 -0.00006 5.7E-05 0.00000340
2.86E-
06
Daily Time-Spent in Exercise
(WP)
-0.0003 0.00018 0.00019 0.00022 0.000136* 5.5E-05 -0.00000583
3.15E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(BP)
0.00014 0.00036 0.00024 0.00029 -0.00006 7.3E-05 0.00000601
3.80E-
06
Daily Low Arousal Positive
Affect*
Daily Time-Spent in Exercise
(WP)
-0.0002 0.00038 0.00023 0.00045 0.00000839 0.00012 0.0000003319
6.85E-
06
Gender (Male=1) -0.0677* 0.03272
0.08568**
*
0.02512 -0.0305*** 0.00706 0.00189*** 0.00038
Age (per 10 years) 0.03959** 0.01334 0.01922 0.01024 0.00890** 0.00292 -0.00015 0.00016
Self-Rated Emotional Health -0.0689*** 0.01974 0.00361 0.01529 0.0101* 0.00428 -0.00017 0.00023
Self-Rated Physical Health 0.02505 0.02195 -0.0115 0.01694 -0.00434 0.00477 0.000117 0.00026
Education 0.03546** 0.01342
-
0.0387***
0.01032 -0.00866** 0.00292 0.000317* 0.00016
Depression (Yes=1) -0.1243 0.06808 0.04126 0.05334 0.0253 0.01488 -0.0016* 0.00081
Experience Daily Stressor
(Yes=1)
0.01486 0.02193 0.04405 0.0256 0.0198** 0.00655 -0.00105** 0.00037
70 | P a g e

Table 5: Daily Affective States Predicting Cortisol Awakening Response (AUCi) of the
Subsequent Day

Minimally-adjusted
model
Model with interaction
term
Fully adjusted model
Parameter Estimate
Standard
Error
Estimate
Standard
Error
Estimate
Standard
Error
Model 4: Daily Negative Affect
Intercept 0.1158*** 0.0067 0.1160*** 0.0068 0.1208** 0.0436
Daily Negative
Affect (BP)
0.0529 0.0284 0.0546 0.0286 0.0645 0.0336
Daily Negative
Affect (WP)
0.0129 0.0263 0.0109 0.0264 0.0295 0.0277
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0001 0.0001
Daily Negative
Affect*
Daily Time-Spent
in Exercise (BP)
- - 0.0003 0.0006 0.0004 0.0006
Daily Negative
Affect*
Daily Time-Spent
in Exercise (WP)
- - -0.0004 0.0004 -0.0002 0.0004
Gender (Male=1) - - - - 0.0111 0.0136
Age (per 10 years) - - - - 0.0153** 0.0054
Self-Rated
Emotional Health
- - - - 0.0052 0.0083
Self-Rated Physical
Health
- - - - -0.0130 0.0090
Education - - - - -0.0183*** 0.0056
Depression (Yes=1) - - - - -0.0153 0.0288
Experience Daily
Stressor (Yes=1)
- - - - 0.0091 0.0118
Model 5: Daily High Arousal Negative Affect
Intercept 0.1154*** 0.0067 0.1156*** 0.0068 0.1091* 0.0433
Daily High Arousal
Negative Affect
(BP)
0.0026 0.0198 0.0029 0.0198 0.0144 0.0224
Daily High Arousal
Negative Affect
(WP)
-0.0157 0.0135 -0.0161 0.0135 -0.0088 0.0140
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0001 0.0001
Daily High Arousal
Negative Affect*
Daily Time-Spent
in Exercise (BP)
- - 0.0002 0.0004 0.0002 0.0004
Daily High Arousal
Negative Affect*
Daily Time-Spent
in Exercise (WP)
- - -0.0001 0.0002 -0.0001 0.0002
71 | P a g e

Gender (Male=1) - - - - 0.0113 0.0136
Age (per 10 years) - - - - 0.0150** 0.0056
Self-Rated
Emotional Health
- - - - 0.0065 0.0083
Self-Rated Physical
Health
- - - - -0.0107 0.0089
Education - - - - -0.0186*** 0.0056
Depression (Yes=1) - - - - -0.0079 0.0284
Experience Daily
Stressor (Yes=1)
- - - - 0.0177 0.0120
Model 6: Daily Low Arousal Negative Affect
Intercept 0.1154*** 0.0067 0.1165*** 0.0067 0.1193** 0.0433
Daily Low Arousal
Negative Affect
(BP)
0.0629* 0.0317 0.0895* 0.0356 0.0953* 0.0385
Daily Low Arousal
Negative Affect
(WP)
0.0231 0.0285 0.0222 0.0286 0.0270 0.0296
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0001 0.0001
Daily Low Arousal
Negative Affect*
Daily Time-Spent
in Exercise (BP)
- - 0.0015 0.0009 0.0016 0.0009
Daily Low Arousal
Negative Affect*
Daily Time-Spent
in Exercise (WP)
- - -0.0005 0.0005 -0.0001 0.0005
Gender (Male=1) - - - - 0.0108 0.0136
Age (per 10 years) - - - - 0.0146** 0.0054
Self-Rated
Emotional Health
- - - - 0.0048 0.0083
Self-Rated Physical
Health
- - - - -0.0118 0.0089
Education - - - - -0.0181** 0.0056
Depression (Yes=1) - - - - -0.0178 0.0290
Experience Daily
Stressor (Yes=1)
- - - - 0.0143 0.0111
Model 7: Daily Positive Affect
Intercept 0.1147*** 0.0067 0.1153*** 0.0067 0.1015* 0.0434
Daily Positive
Affect (BP)
0.0139 0.0097 0.0140 0.0097 0.0126 0.0109
Daily Positive
Affect (WP)
-0.0236 0.0152 -0.0235 0.0152 -0.0191 0.0151
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0000 0.0001
Daily Positive
Affect*
Daily Time-Spent
in Exercise (BP)
- - -0.0001 0.0001 -0.0001 0.0002
72 | P a g e

Daily Positive
Affect*
Daily Time-Spent
in Exercise (WP)
- - 0.0000 0.0003 0.0001 0.0003
Gender (Male=1) - - - - 0.0110 0.0136
Age (per 10 years) - - - - 0.0128* 0.0055
Self-Rated
Emotional Health
- - - - 0.0073 0.0083
Self-Rated Physical
Health
- - - - -0.0067 0.0092
Education - - - - -0.0173** 0.0056
Depression (Yes=1) - - - - -0.0027 0.0283
Experience Daily
Stressor (Yes=1)
- - - - 0.0161 0.0113
Model 8: Daily High Arousal Positive Affect
Intercept 0.1148*** 0.0067 0.1149*** 0.0067 0.1046* 0.0433
Daily High Arousal
Positive Affect (BP)
0.0097 0.0101 0.0101 0.0101 0.0099 0.0113
Daily High Arousal
Positive Affect
(WP)
-0.0188 0.0104 -0.0184 0.0104 -0.0211* 0.0102
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0000 0.0001
Daily High Arousal
Positive Affect*
Daily Time-Spent
in Exercise (BP)
- - -0.0001 0.0002 -0.0001 0.0002
Daily High Arousal
Positive Affect*
Daily Time-Spent
in Exercise (WP)
- - 0.0001 0.0002 0.0002 0.0002
Gender (Male=1) - - - - 0.0111 0.0136
Age (per 10 years) - - - - 0.0130* 0.0055
Self-Rated
Emotional Health
- - - - 0.0072 0.0083
Self-Rated Physical
Health
- - - - -0.0077 0.0091
Education - - - - -0.0177** 0.0056
Depression (Yes=1) - - - - -0.0036 0.0283
Experience Daily
Stressor (Yes=1)
- - - - 0.0156 0.0112
Model 9: Daily Low Arousal Positive Affect
Intercept 0.1148*** 0.0067 0.1153*** 0.0067 0.1032* 0.0432
Daily Low Arousal
Positive Affect (BP)
0.0154 0.0095 0.0154 0.0095 0.0141 0.0105
Daily Low Arousal
Positive Affect
(WP)
-0.0125 0.0103 -0.0123 0.0103 -0.0127 0.0102
Daily Time-Spent
in Exercise (BP)
-0.0001 0.0001 -0.0001 0.0001 -0.0001 0.0001
Daily Time-Spent
in Exercise (WP)
0.0000 0.0001 0.0000 0.0001 0.0000 0.0001
73 | P a g e

Daily Low Arousal
Positive Affect*
Daily Time-Spent
in Exercise (BP)
- - -0.0002 0.0001 -0.0001 0.0002
Daily Low Arousal
Positive Affect*
Daily Time-Spent
in Exercise (WP)
- - 0.0001 0.0002 0.0001 0.0002
Gender (Male=1) - - - - 0.0109 0.0136
Age (per 10 years) - - - - 0.0123* 0.0055
Self-Rated
Emotional Health
- - - - 0.0070 0.0083
Self-Rated Physical
Health
- - - - -0.0061 0.0091
Education - - - - -0.0173** 0.0056
Depression (Yes=1) - - - - -0.0031 0.0283
Experience Daily
Stressor (Yes=1)
- - - - 0.0154 0.0113
74 | P a g e

Chapter 3: The Daily Affective States and Diurnal Cortisol Rhythm in Urban Minority
Youth and the Moderator Role of Moderate-to-Vigorous Physical Activity
Background
The hypothalamic-pituitary-adrenal (HPA) axis is an important neuroendocrine system
that provides physiological resources for the human body when faced with psychological or
physical stressors or demands that arise from daily life (Fries et al., 2009). When faced with
psychological or physical stressors, a normally-functioning HPA axis is acutely reactive to
stressors (i.e. promptly activated and deactivated in response to stressors(Chrousos, 2000;
Gunnar & Quevedo, 2007; McEwen, 2008; Tsigos & Chrousos, 2002)) and is adaptive (i.e., it
reacts progressively less vigorously when encountering the same stressor (McEwen & Gianaros,
2010)). A normally-functioning HPA axis also follows a strong diurnal rhythm, characterized by
a sharp increase in cortisol within 30-45 minutes post-awakening (i.e., the cortisol awakening
response (CAR)), followed by a steady decline in cortisol levels throughout the remainder of the
day. Chronic and prolonged exposure to stressors can lead to dysregulation of the HPA axis
(Dickerson & Kemeny, 2004). Among mentally healthy individuals, patterns that resemble a
dysregulated HPA axis include a larger CAR and a slower decline in cortisol level throughout
the day, or a blunted diurnal cortisol slope (DCS). These patterns expose individuals to an
elevated level of cortisol and are hypothesized to increase individuals’ risks for various adverse
health outcomes (Emma K Adam et al., 2017; Epel et al., 2000; Rosmond et al., 1998).
Therefore, identify correlates of dysregulated HPA axis activity patterns, including increased
CAR and blunted DCS, could help to understand the underlying mechanisms and identify
potential intervention targets. HPA axis activity varies widely by population groups and by day-
to-day experiences (Almeida, Piazza, et al., 2009) (Emma K Adam et al., 2017; Beckie, 2012).
75 | P a g e

Therefore, it is important to investigate both the between- and within-person correlates of the
HPA axis in order to further understand how, and under what contexts, are individual’s daily
experiences impact HPA axis activities
Affective states are individual’s momentary feelings about an actual or perceived
stimulus (Barrett, 2017; Tracy & Randles, 2011). In the adult population, affective states have
been shown to relate slower decline in cortisol levels on the same day and larger CAR on the
next day(E. K. Adam et al., 2006). These results suggest that some affective states constitute
psychological burdens that could activate the HPA axis. However, evidence concerning the
impact of affective states on the HPA axis in youth is inconclusive at best. During childhood and
adolescence, many psychological and physiological functions are developing and changing
(Heim & Nemeroff, 2002; Romeo, 2010), including HPA axis reactivity to stressors (Romeo,
2010). As youths encounter various new experiences and feelings as physiological functions are
developing, the affect-HPA axis relationship may not be the same as in adults. There is some
evidence in youth populations that higher-than-usual negative affect is associated with flatter
DCS on the same day (Doane & Adam, 2010; Doane & Zeiders, 2014) and a larger CAR on the
subsequent day(E. K. Adam et al., 2006; Arbel et al., 2017; Doane & Adam, 2010). Although
this research shows a correlation between affective states and HPA axis diurnal rhythms, other
studies have reported null relationships (K. G. Miller et al., 2016; Polk et al., 2005; Smyth et al.,
1998) or correlations in the opposite direction (Chen et al., 2017; Polk et al., 2005) in youth.
These mixed findings suggest the need for further investigation of the affect-HPA axis
relationship in youth. Therefore, the first aim of this study is to examine whether affective states
are related to the diurnal cortisol rhythm in the youth population.
76 | P a g e

A potential explanation for the mixed findings on the affect-HPA axis relationships
observed in youth could be that there are unexplored moderators of this relationship (Campbell
& Ehlert, 2012; DeSteno et al., 2013). Moderate-to-vigorous physical activity (MVPA) could be
a potential moderator of how HPA axis reacts to psychological and physical stressors. MVPA
presents a physical demand on the human body that activates the HPA axis. The cross-stressor
adaptation hypothesis asserts that physical demands, such as MVPA, promotes biological
adaptations and dampens magnitude of the HPA axis’ reaction to stressors (Sothmann et al.,
1996). At the between-person level, recent evidence has shown that habitually active (e.g.
athletes) individuals react less strongly to laboratory-induced psychological stressors compared
to their less habitually active counterparts (e.g., non-athletes) (Martikainen et al., 2013; Puterman
et al., 2011; Rimmele et al., 2009; Rimmele et al., 2007). At the within-person level, some
studies found that adults who took a 30-minute unstructured moderate-intensity walk, compared
to those who did not, reacted less strongly to psychological stressors (Wood et al., 2017).
However, whether MVPA also moderates the relationships between affect and HPA axis activity
in youth in the free-living environment has yet to be studied.Therefore, the second aim of this
study is to examine whether time-spent in MVPA moderates the relationship between affective
states and diurnal cortisol rhythm in youth.
Methods
Study Sample
This study used data from the baseline measurements of the Imagine Health Study 3
(IHS-3). The IHS-3 study is a block-randomized controlled intervention trial that examines the
effect of a 12-week guided imagery-based intervention program on reducing stress and
increasing a suite of healthy behavior in a population of urban minority adolescents
77 | P a g e

(Weigensberg et al., 2018). Study participants were recruited from four high schools in the
metropolitan Los Angeles area from 2014 to 2016. Participants were enrolled in the study after
they were screened by the study staff for inclusion and exclusion criteria. Study inclusion criteria
included: 1) age 14-17, in second or third year (i.e., sophomore or junior year) of high school at
time of consenting with a stated intention to complete high school to graduation, and 2)
agreement to attend up to 3 after-school classes per week for the 12 weeks of the program. Study
exclusion criteria included: 1) serious chronic illness or physical, cognitive, or behavioral
disability, 2) taking medication known to affect body composition (e.g., prednisone, stimulant
medications for attention deficit disorder), 3) prior diagnosis of clinical eating disorder or
psychiatric disorder, 4) lack of English fluency, 5) participation in similar school-based council-
based programs previously, 6) participation in formal weight-loss programs in preceding 3-
months, 7) pregnancy, 8) sibling or other household member enrolling for the intervention, and
9) participation in after-school sports or other extracurricular activities, including more than one
advanced placement (AP) class that would make it difficult for the participant to attend the after-
school intervention. The full description of this study has previously been
published(Weigensberg et al., 2018).
Study Procedures
The Imagine Health 3 Study (IHS-3) participants participated in a week-long assessment
at four-time points throughout the study: before the intervention (baseline), immediately after the
12-week intervention (post-test), six months after the intervention, and 12 months after the
intervention. The full study procedure is described elsewhere(Weigensberg et al., 2018). Briefly,
participants visited the Diabetes and Obesity Research Institute (DORI) laboratory at the
University of Southern California for measurements and participant training on the Saturday
78 | P a g e

before the start of the week-long ambulatory assessment. Anthropometric data were collected,
and participants completed a battery of questionnaires on a secured internal electronic platform
(i.e., REDCap) via tablets. After the measurement, participants received a package with an
individually-assigned study phone, an accelerometer, and pre-labeled package that contained 4-
day supply of saliva sample collection apparatus, as well as printed instructions for saliva sample
collection. Participants also received a group oral training session for wearing the accelerometer
on the wrist and collecting saliva samples. Starting the following Monday, participants were
asked to wear the accelerometer on their non-dominant wrist always except for water-related
activities. From Tuesday to Thursday, participants were asked to collect three saliva samples
three times per day (i.e., upon waking, 30 minutes post-waking, and bedtime) for at least three
days. Participants were also provided with an extra package with the saliva sample collection
apparatus for an additional day. Including the extra package, each participant could have a
maximum of 3 saliva samples for four days for each 7-day period. This study used the ZEMI
application to facilitate compliance with cortisol sample collection. The ZEMI application is an
open source mobile application designed at the University of Southern California. The ZEMI
application emits alarms based on the wake time that participants provided to the research staff
during the training. Once the participant has responded to the alarm, the ZEMI app prompts
participant through taking a saliva sample, taking a picture of the corresponding saliva sample,
uploading the picture to a secure internal server at the University of Southern California, and
lastly, providing ratings on a list of five-item affective state questions. Both the uploaded
pictures and the affective state responses are time-stamped. The time-stamp information on the
sample pictures has been previously shown a valid measurement for assessing saliva samples
collection time, as compared to the gold standard measurements obtained by cap-like devices
79 | P a g e

(i.e., MEMS caps, Medical Event Monitoring System; Aardex Ltd) in this population(Cheng K
Fred Wen, Weigensberg, Schneider, Weerman, & Spruijt-Metz, 2015). The timeline of when
each of these data was collected is illustrated in figure 9.
Figure 9: Timeline of Imagine Health 7-day observation data collection

Measurements
Salivary Cortisol
From day 2 through day 4 of the 7-day study period, participants were asked to collect
saliva samples for three times per days (i.e., upon awakening before getting out of bed, 30
minutes after awakening, and before bed). Although the participants were asked to provide saliva
samples for three days, they were given an extra one-day supply of sampling apparatus, resulting
in a maximum possible of 12 samples (3 samples a day for four days) throughout the 7-day
period. Participants were instructed not to eat, drink, brush their teeth, or consume any
caffeinated items (e.g., coffee, tea, or chocolate) 5 minutes before collecting the saliva samples.
Participants were instructed to collect the saliva samples upon hearing the alarm reminders that
were personalized to their own schedule and to upload a picture of the collected sample via the
80 | P a g e

ZEMI application. The sampling time of the saliva samples was objectively assessed using the
timestamp collected by the ZEMI application.
After completing the saliva sample, participants were instructed to store the collected
sample in a designated bag in their home freezer before samples retrieval by the study staff.
After the saliva samples were retrieved at the end of the 7-day ambulatory assessment period, the
saliva samples were transported immediately to the University of Southern California DORI
freezer for storage at -80 degrees Celsius. Before immunoassay, the salivettes are thawed and
centrifuged at 2710 RPM for 15 minutes at 4 degrees Celsius. Luminescence immunoassay was
used to determine cortisol concentrations of the samples.
This study used log-transformed cortisol data for modeling the diurnal cortisol slope. The
log-transformation was warranted, as the raw cortisol levels were highly skewed (skewness:
24.78, kurtosis: 779.43). The raw data screening procedure and potential data transformation
processes are in accordance with the most recent recommendations by the International Society
for Psychoneuroendocrinology (Stalder et al., 2016).
Affective States
Affective states were measured using 5-items from the Adolescent Profile of Mood States
scale (Terry, Lane, Lane, & Keohane, 1999) immediately after the cortisol collection. For each
salivary cortisol sample, the participants were asked to provide self-reported ratings on five
different types of affective states, including the feeling of stress, worried, panicked, anxious, and
happy, using visual analog scales on one single screen (Figure 10).

81 | P a g e

Figure 10: Screenshot of affective states questionnaire on ZEMI

Affective state scores for each item will be averaged across each day to create personal
average scores. The personal average affective state scores were then grand mean-centered to
create the between-person (BP) version of the affective states. The between-person version of the
affective state scores represented the personal average and trait-like affective states of each
participant across the study period. To examine the relationship between state-like affective
states and the diurnal cortisol rhythm, the within-person version of affective state scores were
created by subtracting the affective state scores of each day by the BP version of each respective
affective state scores. The resulting scores represented a participant’s level of affective states on
any given day within the study period relative to the average level of each respective affective
state of that participant, or the within-person (WP) versions of the affective state. Both the BP
82 | P a g e

and the WP versions were date-linked to the days when participants collected saliva samples and
included in the same model. The incorporation of both the BP and WP version of the same
variable allowed for disaggregating the inter- and intra-individual variabilities of predictors in
the outcomes of interest (Curran & Bauer, 2011).
Time Spent in MVPA:
The amount of time spent in MVPA each day is assessed by wrist-worn accelerometer-
measured across seven days of ambulatory assessment. At the assessment visit, participants were
provided with a triaxial accelerometer (Actigraph model GT3X+, Actigraph LLC, Pensacola,
Florida). Participants were trained on how to wear the accelerometer properly and were
instructed to wear it on the wrist of their non-dominant hand at all time, except for water-related
activities (e.g., showering, swimming, etc.). They were also instructed to record time when the
accelerometer was taken off and put back on in an accelerometer wear log. The accelerometer
was set to collect participants’ bodily movement at a 30-sec epoch. Upon retrieving the
accelerometers, the raw data was immediately downloaded to the ActiLife software version 6.0
for wear time validation. Classification of MVPA was based on the Chandler et al. (Chandler,
Brazendale, Beets, & Mealing, 2016) cut points, which were developed specifically for data
collected using accelerometers worn on non-dominant wrists. The total amount of time spent in
MVPA was aggregated for each day. Both the WP and BP versions of time-spent in MVPA were
calculated and included in the models.
Anthropometric Measurements
During the measurement visit, participants’ height was measured using a wall-mounted
stadiometer, and weight was measured by clinical medical balance by trained research staff and
was used to calculate the Z-score for the body mass index (BMI-Z). Participants self-reported
83 | P a g e

their gender, race, and ethnicity in the demographic information section of the survey battery.
Participants’ perceived stress level was assessed using the Perceived Stress Scale-17 (PSS-17), a
scale modified from the original PSS-14 (Cohen, Kamarck, & Mermelstein, 1983) that was
validated in an adolescent population of similar demographic characteristics (Nguyen-Rodriguez,
Chou, Unger, & Spruijt-Metz, 2008).
Statistical Analysis
Multi-level models were used to examine the relationships between affective states and
diurnal cortisol patterns. These models allow for parameter estimations to account for the
statistical non-independence inherent in the nested outcome data (i.e., salivary cortisol levels).
Additionally, this modeling approach supports the examination of the within- and between-
person association between affective states and the diurnal cortisol rhythm (E. K. Adam et al.,
2006; Curran & Bauer, 2011). Multi-level models (MLM) were estimated using full-information
maximum likelihood estimation using PROC MIXED in SAS v.9.4 (Cary, NC, USA).
Outcome variables for these models were the log-transformed cortisol levels. Therefore,
parameter estimates resulted from MLM were interpreted as percentages, after the parameter
estimates were back-transformed (E. K. Adam, 2006). Due to lack of variation in cortisol levels
at the day level, two-level multi-level models were used to examine the relationship between
affective states and diurnal cortisol slope. In the two-level models, level 1 included only
variables in intra-individual levels. Variables included in the intra-individual levels included time
since the first sample, a quadratic term for time since the first sample, a dummy variable for the
cortisol awakening response, the within-person affective states score, and the within-person time
spent in MVPA. For the dummy variable for CAR, only the 30-minute post awakening sample of
each day was coded as 1 for this dummy variable, while other samples were coded as 0. Level 1
84 | P a g e

of the analytic model was specified as 𝐶𝑜𝑟𝑡𝑖𝑠𝑜𝑙 = 𝜋 + 𝜋 (𝐶𝐴𝑅 ) +
𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 ) + 𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 )
+𝜋 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
+𝜋 (𝐶𝐴𝑅 ) ∗ (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 )
∗ (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝜋 (𝑇𝑖𝑚𝑒 𝑆𝑖𝑛𝑐𝑒 𝑊𝑎𝑘𝑖𝑛𝑔 )
∗ (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑎𝑡𝑒𝑠 − 𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑒
, where 𝜋 represented the log-transformed cortisol level at waking, 𝜋 represents the percent
change in cortisol level between waking and 30 minutes post-awakening, 𝜋 and 𝜋 represent
the linear and quadratic percent changes in cortisol per hour since awakening, respectively. In
this model, 𝜋 to 𝜋 represent the percent change in cortisol levels at intercept, linear change,
and quadratic change associated with the within-person affective states on the intercept
The level 2 model included variables that vary across individual participants (e.g., the between-
person version of the affective state scores) and was specified as followed:
Level 2: 𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑟
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑟
𝜋 = 𝛽 + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽
𝜋 = 𝛽
𝜋 = 𝛽
𝜋 = 𝛽

85 | P a g e

, where 𝛽 , 𝛽 , 𝛽 , and 𝛽 represent the change in waking level percent changes in
waking level, CAR, linear slope, and quadratic slope associated with the differences in affective
states between participants. These models were constructed to examine the effect of affective
states on the diurnal cortisol rhythm. The effect of affective states on the CAR of the subsequent
day was not recommended due to the lack of self-reported or objectively measured wake time, a
crucial variable for accurate estimation of the CAR (Stalder et al., 2016).
Interaction terms between affective states and time-spent in MVPA were also included in
the analytic models to investigate the potential moderation effect of time spent in MVPA. The
analytic models included age, gender, race, ethnicity, and level of perceived stress at the person-
level as covariates.
Results:
Demographic and Descriptive Statistics
Demographic characteristics of the study participants are presented in Table 6. Table 7
presents the descriptive statistics for average negative affect, individual items for the negative
affective state (higher number being more negative), and accelerometer-measured time spent in
MVPA. Table 8 presents the average cortisol levels and time of sampling for each sampling
occasion (i.e., upon awakening, 30 minutes post awakening, and bedtime). Across the 3-day
saliva sampling protocol, 53.0% (n=123) participants provided three saliva samples for at least
three days, and 75.4% (n=175) participants provided at least one day with three saliva samples.
Results from the two-level unconditional model indicated that, on average, the cortisol
levels rose by 102.44% at the second sample of the day, declined at an average rate of 17.36%
per hour, and the rate of decline decelerate at the rate of 0.41% per hour until bedtime (Table 8).
This model also showed that intercept and linear term of time explained 19.2% and 0.3% of the
86 | P a g e

variation, respectively, at the between-person level while 80.5% of variation exists at the within-
person level.
Day-Level Association between the Affective States and Diurnal Cortisol Slope and Time-
Spent in MVPA as a Potential Effect Modifier
Results from two-level models showed that affective states, both at the within- and the
between-person level, were not related to the rate of decline in cortisol on the same day (Table 9
model 1). Models with interaction terms between affective states and accelerometer-measured
time spent in MVPA indicate that time-spent in MVPA did not modify the relationship between
affective states and the rate of decline in cortisol levels throughout the day (Table 9, model 3).
Discussion:
This study examined the day-level relationship between affective states and diurnal
cortisol rhythm and whether accelerometer-measured time-spent in MVPA moderated this
relationship in adolescents. Our results did not support the hypothesis that affective states are
associated with the rate of decline in cortisol both at the between- and within-person level in
youth. The results also did not support the hypothesis that the affect-HPA relationship is
moderated by time-spent in MVPA in youth. This study adds to the mixed findings on the affect-
HPA axis relationship among youth, providing results that do not agree with some of the existing
literature.
Contrary to the hypotheses, neither the average negative affective state nor the individual
items of the affective state scale were related to the diurnal cortisol rhythm in this study
population. While this might mean that there are no relationships to be found, we can speculate
on other possible explanations for these null findings. First, affective states captured in this study
were captured at home when cortisol samples were collected -- upon awakening, 30-minute post-
87 | P a g e

awakening, and at bedtime. Affective states collected at these time points may not be
representative of an individual’s feelings and experiences throughout the day for that youth may
have not yet been exposed to (i.e., right upon waking up) or have already recovered from
exposure to (i.e., bedtime) psychological or physical stimuli at these time points. Measuring
affective states upon awakening and before bedtime may only provide a summative day-level
view of adolescent participants’ affective experience. However, this sampling frequency may not
provide data sensitive enough to capture to how youths’ HPA axis reacts to affective states as
psychological or physical stimuli unfold in other more relevant social of physical settings (e.g.,
at school) or at the momentary levels. The limited amount of affective states data per day also
precludes our abilities to investigate other relevant aspects of daily affective experiences, for
example, variability in affective states. Results from study 3 of this dissertation corroborate the
latter notion, as we found that fluctuation in affective states is acutely related to altered HPA axis
activities during the subsequent 30 minutes. Therefore, the null results observed in this study
may be due to that adolescents’ affective states are related to HPA axis activities at a momentary
level, instead of day-level. Another potential explanation for the null results may be that, instead
of average affective states, HPA axis activities are related to specific aspects of affective states,
for example, high arousal positive affective states. The idea of arousal states originates from the
circumplex model of affect (Posner et al., 2005), which suggests that affective states should be
categorized both by its valence (i.e., positive and negative) and its level of activation (i.e., high
or low activation). Examples of high arousal positive affect could include the feeling of active
and alert (Hoyt et al., 2015), while negative affect may include items such as feeling upset,
irritable, angry, and frustrated (Charles, Mogle, Leger, & Almeida, 2017). The current literature
only provides supports that high arousal positive affect are related to altered diurnal cortisol
88 | P a g e

rhythm ((Hoyt et al., 2015) and study 1 of this dissertation). Results of this study also
corroborate the null findings on the relationship between high arousal negative affect and the
diurnal cortisol rhythm, as none of the potential high arousal negative affective states items (e.g.,
the feelings of anxious or panic) were related to the DCS. Future studies that examine whether
diurnal cortisol rhythms are related to affective states of high or low arousal components may
provide further insights into how affective states relate to the HPA axis.
Methodological limitations related to the daily cortisol sampling frequency may also
render the dataset not sensitive enough to pick up the nuanced within-person variation in diurnal
cortisol rhythm in youth. Recent study has shown that saliva sampling procedures of moderate
intensity (i.e., 3-4 cortisol samples per day) could provide comparable results as data collected
with more intensive sampling protocols (i.e., 5-6 samples per day) when evaluating how diurnal
cortisol rhythm relates to disease morbidity at the between-person level(Emma K Adam et al.,
2017). However, it is also speculated that moderate sampling procedure (i.e., 3-4 saliva samples
per day) may not provide adequate data for investigating within-person HPA axis processes
(Saxbe, 2008). Compliance with the saliva sampling protocol can be another pertinent factor that
further limits our ability to detect within-person variation. Youth participants in this study were
asked to provide three saliva samples per day for up to four days. However, only 53% of the
participants provided three samples per day for three days, lowering statistical power and
hindering our ability to detect the nuanced within-person variation in diurnal cortisol rhythm.
Therefore, it is possible that the absence of the hypothesized within-person affect-HPA axis
relationship could be partially attributed to both the limitations in sampling intensity and
participant compliance. Besides limitations related to the frequency and data availability of daily
saliva sample, the absence of self-reported or objectively measured wake time hindered our
89 | P a g e

ability to estimate the cortisol awakening response. Wake time detection using accelerometers is
challenging. Algorithms developed using data from wrist-worn accelerometer (Sadeh, Sharkey,
& Carskadon, 1994) commonly used for wake time detection exhibited poor specificity (46%
compared to polysomnography (Slater et al., 2015)) in wake time detection. Therefore,
inaccurate wake time estimations resulted from using the wake-time detection using algorithms
currently available for wrist-worn accelerometer may lead to misleading CAR results, as
accurate estimation of the CAR relies heavily on accurate estimation of wake time(Stalder et al.,
2016). Reanalysis using accelerometer-derived wake time from a more accurate wake time
detection algorithm could potentially offer valuable opportunities to examine the day-level
relationship between affective states and subsequent day CAR.
The current study is one of the few to examine the relationship between affective states
and HPA axis activities in youth. Our results did not corroborate the findings documented in
some other studies (E. K. Adam et al., 2006; Arbel et al., 2017; Doane & Adam, 2010; Doane &
Zeiders, 2014). There were several methodological limitations. These limitations, taken along
with the continued contradictory findings in the literature, highlight the need for strong studies
that examine the affect-HPA axis relationship throughout adolescence, an important
developmental stage that involves changes in both psychological and physiological functions.
Future studies with higher daily saliva sampling frequencies and momentary measurements that
also capture the arousal components of affective states could potentially further elucidate the
within-person relationship between affect and HPA axis activities.

90 | P a g e

Table 6: Demographic Characteristics of the Study Participants
(N=232)
Gender N(%)
Female 154 (66%)
Race N(%)
Hispanic White 204 (87.9%)
Non-Hispanic White 4 (1.7%)
African American 2 (0.9%)
American Indian 1 (0.4%)
Asian 8 (3.4%)
Other Hispanic 5 (2.2%)
Other Non-Hispanic 7 (3.0%)
Not Reported 1 (0.4%)
Variable Mean SD
Age 16.4 ± 0.6
BMI 24.6±5.7
Perceived Stress Scale Score 29.7±8.7 (4-61)

Table 7: Descriptive Statistics for Affective State Scores and Time Spent in MVPA
Variable Mean ± SD (Range)
Feeling of Stressed 1.52±0.51(1-4)
Feeling of Worried 1.45±0.49(1-4)
Feeling of Panicked 1.34±0.43(1-4)
Feeling of Anxious 1.39±0.45(1-4)
Negative Affect 1.42±0.43(1-4)
Time Spent in MVPA 212.41 ± 197.87 (0-717) mins

91 | P a g e

Table 8: Salivary Cortisol Levels by Time and Diurnal Cortisol Rhythm Estimation
Descriptive Statistics for Salivary Cortisol
Sample Sampling Time
(Mean±SD)
Cortisol Level (mg/dL)
(Mean±SD)
Waking 06:27±01:14 0.332±0.196
Waking +30 minutes 07:06±01:44 0.588±1.075
Evening 20:58±03:03 0.084±0.236
Diurnal Cortisol Activity Estimated from Two-Level Unconditional Model
Diurnal cortisol rhythm Estimate (SE) Interpretation
Cortisol Level at the First Sample -1.2858 (0.03166)*** 0.2764 mg/dL
Morning Rise
+
0.7053 (0.03845)*** 102.44% from the first sample

Linear Timesince first sample -0.1907 (0.03072)*** -17.36% per hour since the first sample
Quadratic Timesince first sample 0.004414 (0.002011)* +0.41% per hour since the first sample
Covariance Parameter Estimate
Covariance Parameter Level Estimate (SE) P-value
UN(1,1) BP 0.06876(0.01355)*** <0.0001
UN(2,1) BP 0.00021(0.00129) 0.8702
UN(2,2) BP 0.00109 (0.00023)*** <0.0001
Residual 0.2883 (0.01319)*** <0.0001
Model Fit Indices
2 Log Likelihood 2371.8
AIC 2387.8
BIC 2413.3
***: p<0.001, **:p<0.01, *:p<0.05
+: The term “Morning Rise” was used in place of CAR for that CAR estimation requires reliable wake time data. In
the absence of reliable wake time data, the regression coefficient for the second sample of the day represents only
the additional amount of cortisol secreted after the first saliva sample, instead of the CAR.

92 | P a g e

Table 9: Daily Affective States Predicting Diurnal Cortisol Slope and the Moderator Effect of
Accelerometer-Measured Time-Spent in MVPA

Cortisol Level at the
First Sample
Estimate (SE)
Morning Rise
+

Estimate (SE)
Time Since First
Sample
Estimate (SE)
Time Since First
Sample
2

Estimate (SE)
Model 1: Affective States and Diurnal Cortisol Rhythm
Intercept -1.0949 (0.2321) *** 0.6298 (0.2997)* -0.4115
(0.2380)+
0.0175 (0.0158)
Negative Affect (BS) -0.0062 (0.0901) 0.0592 (0.1109) 0.0964 (0.0951) -0.0067 (0.0063)
Negative Affect (WS) -0.2065 (0.1455) 0.1934 (0.2053) 0.1543 (0.1067) -0.0086 (0.0070)
Model 2: Affective States and Diurnal Cortisol Rhythm, Adjusting for Accelerometer-Measured Time-Spent in
MVPA
Intercept -1.2402 (0.2850) *** 0.6480 (0.3766)+ -0.4099 (0.2741) 0.0187 (0.0188)
Negative Affect (BS) 0.0004 (0.0922) 0.0787 (0.1134) 0.0760 (0.0984) -0.0053 (0.0066)
Negative Affect (WS) -0.1788 (0.1495) 0.1809 (0.2094) 0.1647 (0.1098) -0.0095 (0.0073)
Time in MVPA (BS) 0.0000 (0.0004) 0.0000 (0.0006) 0.0007 (0.0006) 0.0000 (0.0000)
Time in MVPA (WS) 0.0001 (0.0003) 0.0005 (0.0004) 0.0002 (0.0004) 0.0000 (0.0000)
Model 3: Accelerometer-Measured Time-Spent in MVPA as a Moderator of the Affective States and Diurnal Cortisol
Rhythm
Intercept -1.2424 (0.2849)*** 0.6216 (0.3766)+ -0.3656 (0.2748) 0.0156 (0.0189)
Negative Affect (BS) 0.0175 (0.0995) 0.0893 (0.1289) 0.0826 (0.1217) -0.0055 (0.0081)
Negative Affect (WS) -0.3792 (0.3250) 0.5275 (0.4727) 0.0272 (0.2816) 0.0038 (0.0184)
Time in MVPA (BS) 0.0000 (0.0004) 0.0000 (0.0006) 0.0006 (0.0006) 0.0000 (0.0000)
Time in MVPA (WS) 0.0001 (0.0003) 0.0006 (0.0004) 0.0001 (0.0004) 0.0000 (0.0000)
Negative Affect (BS)*
Time in MVPA (BS)
-0.0004 (0.0010) -0.0003 (0.0014) -0.0001 (0.0013) 0.0000 (0.0001)
Negative Affect (WS)*
Time in MVPA (WS)
0.0009 (0.0013) -0.0016 (0.0019) 0.0008 (0.0013) -0.0001 (0.0001)
Participants’ gender, age, ethnicity, BMI percentile, and perceived stress score were adjusted in all models.
***: p<0.001; *:p<0.05; +:p<0.10
+: The term “Morning Rise” was used in place of CAR for that CAR estimation requires reliable wake time data. In
the absence of reliable wake time data, the regression coefficient for the second sample of the day represents only the
additional amount of cortisol secreted after the first saliva sample, instead of the CAR.
93 | P a g e

Chapter 4: The Momentary Affective States and Diurnal Cortisol Rhythm in Minority
Youth
Background
The hypothalamic-pituitary-adrenal (HPA) axis is a critical neuroendocrinological system
that provides physiological resources for the human body to cope with psychological and
physiological demands that could challenge homeostasis (Fries et al., 2009). A normally
functioning HPA axis is adaptive in that it reacts progressively less rigorously when exposed to
the same stimuli (McEwen, 2007, 2008). Repeated activation of the HPA axis due to prolonged
exposure to these stimuli can lead to the development of a dysregulated and maladaptive HPA
axis (McEwen, 2007, 2008) and increase risks of adverse mental and physical health conditions
(Emma K Adam et al., 2017; Epel et al., 2000). Identifying predictors of and mechanisms that
lead to HPA axis activation could provide future research with valuable insights on viable
intervention targets for preventing healthy individuals from developing HPA axis dysregulation
and related adverse health outcomes.
Affective states are possible predictors of HPA axis activation. Affective states are
momentary experiences or feelings that an individual has about an actual or perceived internal or
external event (Barrett, 2017; Tracy & Randles, 2011), which could present psychological
demands and activate the HPA axis. An early meta-analysis showed that, compared to the control
groups, adults exposed to experimental conditions designed to elicit various feelings (e.g., fear of
losing social approval, rumination, threats, etc.) had increased cortisol levels as early as 30
minutes after exposure to the experimental condition (Denson et al., 2009). Evidence on acute
HPA axis reactions to affective states in adolescents, however, is limited. During childhood and
adolescence, many psychological and physiological functions develop and change (Heim &
94 | P a g e

Nemeroff, 2002; Romeo, 2010), including HPA axis reactivity to stressors (Romeo, 2010).
During adolescence, youths encounter various new experiences and feelings. Investigating the
acute HPA axis reactivity to affective states during adolescence may provide valuable insights
into how the HPA axis develops during adolescence. The HPA axis is reactive to fluctuations in
naturally-occurring, or unstimulated, affective states at the day level. Several studies have
pointed that youths exhibit a lower rate of decline in cortisol levels throughout the day, the
diurnal cortisol rhythm that indicates maladaptive HPA axis activities (for review, see (Emma K
Adam et al., 2017)), on days when they reported higher in affective states of negative valence (E.
K. Adam, 2006; E. K. Adam et al., 2006; Arbel et al., 2017; Doane & Adam, 2010). While
providing foundations for understanding how affective states are related to the HPA axis, these
studies provide limited insights into how fluctuations in naturally-occurring affective states are
related to HPA axis activities on a momentary basis. Laboratory-based studies have shown that
emotion induction and lab-induced emotional stressors could lead to noticeable changes in
cortisol levels as early as 30 minutes after the exposure (Denson et al., 2009). Therefore it is
possible that naturally-occurring affective states could lead to altered cortisol level at a similar
window. The first aim of this study is to examine whether affective states are related to cortisol
levels in the subsequent 30 minutes.
There are substantial between- and within-person variations in the effects of affective
states on HPA axis activities (Almeida, Piazza, et al., 2009; Denson et al., 2009). A possible
explanation for the substantial unexplained variation is that there are other moderators in the
affect-HPA axis relationship. Moderate-to-vigorous physical activity (MVPA) and dietary
behavior could be possible behavioral moderators in this relationship. Both MVPA and dietary
behavior have shown preliminary efficacies in acutely changing affective states. MVPA has been
95 | P a g e

shown to acutely improve youth’s affective states in youth (Dunton et al., 2014; C. K. F. Wen et
al., 2018). Preliminary evidence has further shown that participation in physical activity of
moderate intensity reduced adults’ HPA reactivity to stressors (Wegner et al., 2014; Wood et al.,
2017). On the other hand, dietary behavior, such as breakfast consumption(Benton & Brock,
2010) and sugar consumption (Benton, 2002; van de Rest, van der Zwaluw, & de Groot, 2018),
has been shown to affect individuals’ affective states. While these studies have provided
preliminary evidence on the possible effect of energy-balance related behaviors on affective
states, whether these behaviors could acutely change the relationship between affective states
and HPA axis activities remains unexplored. Understanding whether these energy-balance
related behaviors could modify the affect-HPA axis relationship could provide valuable targets
for prevention and intervention efforts that aim to reduce youth’s stress neurophysiological
responses. The second aim of this study is to examine the potential moderating role of MVPA
and sugar consumption in the acute predictive relationship between affective states and HPA axis
activities.
Methods
Study Sample:
This study used data from the Food, Adolescence, Mood, and Exercise (FAME) study
(O'Reilly et al., 2015). The FAME study is a laboratory-based randomized cross-over trial that
aimed to examine the effect of meal contents on ethnic minority adolescents’ mood states,
physical activity, and metabolic indices. In this study, participants visited the University
observation laboratory for two separate days that were 2-4 weeks apart. For the first visit, each
participant was randomized into one of two meal conditions, high sugar and low fiber (HS) or
low sugar and high fiber (LS), followed by an 8-hour in-lab observational period. Each
96 | P a g e

participant returned to the laboratory for the second time and received the other meal condition
after at least two weeks of wash-out period, followed by another 8-hour in-lab observational
period. The HS breakfast contained 41.0 grams of sugar and 1.0 gram of fiber, whereas the LS
breakfast contained 7.0 grams of sugar and 16.0 grams of fiber(O'Reilly et al., 2015). The FAME
study participants were recruited from the Los Angeles area from 2007-2010. The inclusion
criteria were: adolescents aged between 14 and 18 years old, African American or Latino
ethnicity, and with a body mass index ≥ 85th percentile for age and sex. The exclusion criteria
included: diagnosis of diabetes, participation in a weight loss or exercise program, use of
medications that influenced body weight or insulin sensitivity, or diagnosis of a syndrome that
influences body composition.
Study Procedure
After signed parental consent and participant assent was provided, the study participants
were scheduled for two visits at the observational laboratory at the University of Southern
California, each after a 10-hour overnight fast. Each participant received experimental breakfast
and lunch meals (either HS or LS condition for the first visit and the other condition at the
second visit) and was otherwise instructed to engage freely in activities available in the
observation laboratory throughout their 8-hour stay. The naturalistic observational laboratory
contained various options for participants to choose from, including treadmill, small trampoline,
jump rope, hula-hoops, free weights, video games (Nintendo Wii, Rock Band, Dance Dance
Revolution), books, movies, arts and craft center, and music. At the beginning of the laboratory
stay, a small saline lock intravenous catheter was placed into the participant’s forearm for
subsequent blood samples, and a uniaxial accelerometer (Actigraph GT1M) was fitted on the
participant’s waist for activity behavior assessment. For the first 5 hours of each laboratory stay,
97 | P a g e

participants were interrupted every 30 minutes for blood samples, and saliva samples resulted in
10 blood and saliva samples for each laboratory visit. Throughout the entire 8-hour laboratory
stay, participants were asked to provide ratings for their affective states. The same in-lab
measurement procedures were implemented for the second visit.
Measurements
Salivary Cortisol
Before the first meal, and at every 30 minutes during the first 5-hour of each laboratory
stay, participants were asked to put a cotton plug in their mouth until fully soaked with saliva.
This data collection schedule resulted in 10 salivary cortisol samples for each visit. Each soaked
cotton plug was collected in a labeled vial and stored on dry ice until the end of the lab visit
when they were immediately transported to an -80 degrees Celsius freezer at the Diabetes and
Obesity Research Institute (DORI) laboratory at the University of Southern California prior to
assay. Salivary cortisol levels were determined by immunometric assay on Tosoh AIA 600II
analyzer.
The raw cortisol values were skewed (Mean 0.784, Median 0.680, Min: 0.020, Max:
41.000 Skewness: 33.010, Kurtosis: 1292.640) with one single extreme outlier of 41 nmol/L.
After removing the extreme outlier, the distribution properties of the log-transformed cortisol
values were less skewed (Mean: -0.406, Median: -0.386, Min: -3.912, Max:1.435, Skewness: -
0.305, Kurtosis: 1.018). Therefore, the current study used the log-transformed cortisol values
after removal of the extreme outlier as the outcome variable.
Affective States
Prior to the first meal and at every 30 minutes during the entire 8-hour of each laboratory
stay, participants were asked to rate their current affective states, at the same time with serum
98 | P a g e

and saliva sample collection. Five affective states were collected at each time using visual analog
scales. The five-item scale included 4 items from the tension subscale of the Profile of Mood and
States (POMS) scale(Terry et al., 1999):panic, worry, anxiousness, and nervous, and 1 item for
the feeling of calmness, with a possible range from 0 to 100 for each item. The average negative
affective state score was calculated by averaging the scores for feelings of panic, worry,
anxiousness, and nervousness at each moment. Affective state scores for each item and the
average negative affective state were averaged across both visits to create personal average
scores for each person. The personal average scores were then subtracted by the grand mean for
the respective item to create the grand mean-centered affective state scores and herein denoted as
the between-person (BP) version of the affective state scores. The BP affective state scores
calculated this way represents the personal average and trait-like affective states of each
participant across the study period. The within-person (WP) versions of affective states were
created by subtracting the affective state scores at each measurement (i.e., every 30 minutes) by
the respective average affective state scores for that person. The resulting scores represent the
participant’s level of affective states at any given moment within the study period relative to the
average level of each respective affective state of that participant. Both the BP and the WP
versions will be included in the same model to disaggregate the inter- and intra-individual
variabilities in outcomes of interest (Curran & Bauer, 2011).
Time Spent in MVPA
The amount of time spent in MVPA was assessed by the waist-worn uniaxial
accelerometer (ActiGraph GT1X) across the 8-hour stay for each laboratory stay. Before the
beginning of the 8-hour laboratory stay, the accelerometer was fitted on the participants’ right
hip under the cloth using an elastic belt. The accelerometers were set to collect time-stamped
99 | P a g e

data at a 60-sec epoch. Accelerometer data was processed using SAS code developed by the
National Cancer Institute for use with National Health and Nutrition Examination Survey
(NHANES, http://riskfactor.cancer.gov/tools/nhanes_pam). The thresholds for physical activity
of moderate to vigorous intensity were age-adjusted using the criteria by Freedson et al.
(Freedson, Pober, & Janz, 2005). The time-stamped accelerometer data were matched to the time
of the affective states measurement and blood draw. Time-spent in MVPA were summed for
each 30-minute interval. Both BP and WP versions for time spent in MVPA were calculated
using the method described in the previous section.
Statistical Analysis
The current study utilized multi-level modeling (MLM) to examine whether affective states
are related to salivary cortisol levels in the subsequent 30 minutes. MLM accommodates
parameter estimations that account for the non-independent nature of the outcome variables (i.e.,
salivary cortisol level at each moment was nested within a laboratory visit and within a person)
and allows for disaggregating within- and between-person association between affective states
and the diurnal cortisol rhythm (E. K. Adam et al., 2006; Curran & Bauer, 2011). Multi-level
models were estimated using full-information maximum likelihood estimation using PROC
MIXED in SAS v.9.4 (Cary, NC, USA).
Cortisol data in this dataset were nested within visit and participant. Due to lack of variation
in cortisol levels between the two laboratory visits, two-level mixed models, instead of three-
level models, were used to estimate the momentary relationships between affective states and
log-transformed cortisol level of the subsequent 30 minutes. Since the outcome variables were
log-transformed, parameter estimates resulted from these models were interpreted after the
parameter estimates were back-transformed and should be interpreted as percent changes in
100 | P a g e

cortisol (E. K. Adam, 2006). In these two-level models, level 1 included only variables at the
individual saliva sample level, including time since the first sample, a quadratic term for time
since the first sample, and the affective state score at the beginning of the 30-minute interval.
The level 1 models were specified as
𝐶𝑜𝑟𝑡𝑖𝑠𝑜𝑙 ( ) = 𝜋 + 𝜋 (𝑇𝑖𝑚𝑒
) + 𝜋 (𝑇𝑖𝑚𝑒
) ∗ (𝑇𝑖𝑚𝑒
) + 𝜋 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑆𝑡𝑎𝑡𝑒 −
𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑆𝑡𝑎𝑡𝑒 ) + 𝑒
, where 𝜋 represented the log-transformed cortisol level 30 minutes after the assessment of
affective states (t+30 mins), 𝜋 and 𝜋 represented the linear and quadratic percent changes in
cortisol per 30-minute since the beginning of the visit to account for the cortisol rhythm,
respectively, and 𝜋 represented the percent changes in cortisol at t+30 mins predicted by
affective state at the beginning of that 30-minute interval (t).
The level 2 models included variables that vary at the participant level and were specified as:
𝜋 = 𝛽 + 𝛽 (𝑀𝑒𝑎𝑙𝑡𝑦𝑝𝑒 ) + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 ) + 𝑟
𝜋 = 𝛽 + 𝛽 (𝑀𝑒𝑎𝑙𝑡𝑦𝑝𝑒 ) + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽 + 𝛽 (𝑀𝑒𝑎𝑙𝑡𝑦𝑝𝑒 ) + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
𝜋 = 𝛽 + 𝛽 (𝑀𝑒𝑎𝑙𝑡𝑦𝑝𝑒 ) + 𝛽 (𝐴𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑎𝑡𝑒𝑠 )
, where 𝛽 , 𝛽 , 𝛽 , and 𝛽 represent the percent changes in cortisol at t+30 mins, linear
slope, quadratic slope, and momentary level affective states that were associated with the
breakfast assignment. A random intercept 𝑟 was also specified in the model to allow intercept
to vary among FAME participants. 𝛽 , 𝛽 , 𝛽 , and 𝛽 represented changes in cortisol at
t+30 mins, linear slope, quadratic slope, and momentary level affective states that were
associated with the differences in affective states between participants.
101 | P a g e

Interaction terms were constructed, by multiplying the main predictor of the model with
time spent in MVPA at the same level (e.g., within-person negative affective states* within-
person time-spent in MVPA), to investigate the potential moderation effect of time spent in
MVPA on the affect-HPA axis relationship. Interaction terms were constructed by multiplying
the main predictors of the model with meal assignment, to investigate the potential moderation
effect of meal assignment on the relationship between affective states and subsequent cortisol
levels. The inclusion of interaction terms in the final models was made based on statistical
significance. Non-significant interaction terms were removed from the final models to maintain
model parsimony. Parameter estimates and model fit indices for the all model building steps are
summarized in table 13 (Supplemental Table S1). All the final models included age, gender, and
ethnicity as covariates.
Results:
Demographic and Descriptive Statistics
Table 10 presents descriptive information on the study participants. Table 11 presents
descriptive statistics of the main predictors (i.e., momentary affective state scores) and moderator
of interest (i.e., time spent in MVPA). Figure 10 illustrates the average salivary cortisol levels
across both visits. Salivary cortisol was the highest at the beginning of the laboratory stay
(Figure 10) and declined throughout the remainder of the morning. These patterns did not
significantly differ by meal assignment. The unstructured 2-level model with only the linear and
quadratic terms of time since the first saliva sample as predictors indicated that, on average,
cortisol levels declined at an average rate of 16.48% per 30 minutes, and the rate of decline
decelerated at a rate of 1.68% per 30 minutes throughout morning (Figure 10). Similar trends
were observed when salivary cortisol was separately analyzed for each meal type (Figure 11)
102 | P a g e

The Momentary Relationship between the Affective States and Subsequent Cortisol Levels
Across both the HS and LS conditions, the average negative affective state scores, both at
the between- and within-person levels, were not associated with levels of salivary cortisol at the
subsequent 30 minutes and the rate of decline in cortisol levels (table 2, model 1). After adjusting
for current affective states and time spent in MVPA, the significant meal type by time interaction
suggested that, compared to the LS visit, participants had a 17.85% lower average cortisol level
(est= -0.20, p=0.0155) and a slower rate of decline in cortisol level (est=0.08, p=0.0391) during
the HS visit. Additional analyses that separately examined the same models for each meal
assignment were conducted on a post-hoc basis. The results revealed that, at the within-person
level, average negative affective state score was associated with 1.91% (est: 0.02, p=0.0343)
higher cortisol level 30 minutes after they had reported one point higher in average affective
states compare to their personal average during the HS visit only (Table 2, Model 2). The
average negative affective state score was not related to subsequent cortisol level during the LS
visit (Table 2, Model 3). In models with the individual item of affective state (i.e., panic, worry,
nervousness, anxiousness, and calm) as the main predictors, only levels of feeling panic were
related to cortisol levels at the subsequent 30 minutes at the within-person level (est=0.02,
p=0.002) during the HS visit (Supplemental Table S2). None of the other individual affective
state items were associated with levels of salivary cortisol at the subsequent 30 minutes and the
rate of decline regardless of meal types.
Time Spent in MVPA as a Potential Effect Modifier
Across both breakfast conditions, accelerometer-measured time-spent in MVPA during
the 30 minutes after assessment of affective states was associated with levels of salivary cortisol
at the subsequent 30 minutes at the within-person level (est: 0.014, p=0.0128, Table 12, Model
103 | P a g e

1). This result indicated that participants exhibited a 1.44% increase in salivary cortisol level
after they had spent one minute more in accelerometer-measured MVPA than their personal
average level. Similar main effects with time spent in MVPA were observed in models with
individual items of affective states as the main predictors. However, the amount of time spent in
MVPA during the 30 minutes interval, both at the between- and the within-person level did not
moderate the relationship between affective states and cortisol levels at the subsequent 30
minutes did not moderate the relationship. Post-hoc analysis results that separately analyzed the
same model by meal condition indicated that the within-person association between time spent in
MVPA and subsequent cortisol levels were significant only during the HS visit (est=0.016,
p=0.495 Table 12 Model 2). Both within- and between- person time spent in MVPA were not
related to subsequent cortisol level during the LS visit (Table 12, Model 3). Time spent in
MVPA did not moderate the acute relationship between affective states and cortisol level in the
subsequent 30 minutes for either breakfast conditions.
Discussion
This study is one of the first to examine the relationship between fluctuations in affective
states and subsequent cortisol levels and the possible moderating effects of physical activity or
specific nutrient intake on this relationship in ethnic minority youths. Results of this study
showed that negative affect and MVPA were independently related to cortisol levels measured
30 minutes later, at the within-person level, and only during the high sugar condition laboratory
visit. During the HS visit, participants were provided with breakfast that contained 41.0 grams of
sugar (O'Reilly et al., 2015), a comparable level of sugar in breakfast observed in other studies
conducted among urban ethnic minority youth (Schembre et al., 2013). The identified association
104 | P a g e

in the HS condition, therefore, could represent how youths’ neuroendocrine system functions
after consuming breakfast on a regular day.
This study further builds on the existing literature by further showing that HPA axis is
acutely reactive to the fluctuation in affective states on a momentary basis during when
adolescents were provided with a breakfast high in sugar content. The identified within-person
process during the HS condition suggests that when adolescents experience higher-than-usual
negative affective states, especially the feeling of panic, the psychological state places demands
on the body, which in turn activate the HPA axis. Additionally, higher cortisol levels were
followed by higher within-person time spent in MVPA during the 30 minutes leading up to the
saliva sample collection. Although research suggests that time spent in MVPA can improve
affective states in adolescents (Dunton et al., 2014; C. K. F. Wen et al., 2018), we did not find
that MVPA alleviated the effects of negative affect on youths’ neuroendocrine system. As
suggested by the cross-stressor adapter hypothesis (Sothmann et al., 1996), frequent exposure to
physical activity, which is a physical stressor, could lead to physiological adaptation and in turn
reduce individual’s reactivity to stressors. However, obese adolescents are in general less
habitually active than their non-obese counterparts (Belcher et al., 2010). Therefore, it is possible
that the effect of MVPA in reducing stress reactivity would not be observed in this population.
Taken together, these observed within-person associations indicate that the HPA axis is acutely
reactive to both psychological and physical demands when these demands are greater their
respective usual levels.
The null relationship between affective states and subsequent salivary cortisol during the
LS condition is unexpected. We speculate that the possible decrease in hunger-related
metabolites due to increased consumption of fiber during the LS condition may explain the
105 | P a g e

absence of this acute relationship between affective states and subsequent cortisol level. Prior
research has shown that ghrelin levels, one of the metabolites associated with hunger, are
associated with higher HPA axis reactivity(Ulrich-Lai & Ryan, 2014). Consumption of soluble
fiber has also been shown to reduce biomarker and self-reported rating of hunger (Ye,
Arumugam, Haugabrooks, Williamson, & Hendrich, 2015). Therefore, it is possible that after
consuming 16.0 grams of the soluble fiber during the LS condition(O'Reilly et al., 2015),
participants’ HPA axes were less reactive to psychological and physical demands, compared to
the day when they were exposed to the HS condition. While this may explain the discrepancies
observed within participants between an HS and LS breakfast condition, we could not examine
this physiological mechanism due to the small subsample (n=23) of participants with processed
ghrelin assays available in the data. Nonetheless, the discrepant results between the HS and LS
visit may suggest a possible role of sugar consumption and HPA axis reactivity.
This study had several limitations. First, while this study took place in an observation
laboratory that was purposefully decorated like a regular living room and equipped with
entertainment and exercise equipment attuned to the specific participants, youths’ emotional
experience in the naturalistic environment can still be different from that of ambulatory and free-
living environment. An important way that fluctuations in affective states captured in this
naturalistic environment differ from the ambulatory environment is the absence of stressors that
occur in an individuals’ daily life. Therefore, future studies that examine whether the number or
kind of stressors moderates the relationship among affective states, MVPA, and subsequent
cortisol levels could further elucidate the intricate relationship among affective states, MVPA,
and the HPA axis. Second, this study compared how the HPA axis reacts to fluctuation in
affective states in two different meal-type conditions, HS and LS meals. Although this study
106 | P a g e

provided novel insights on the differential impact of nutrient consumption on the impact of
affective states on HPA axis activation, future studies conducted with multiple days of specific
meal conditions could yield stronger evidence for the effect of specific nutrients on the HPA axis
reactivity. Third, the results of this study might not be generalizable to youth who are not
overweight or not African American, or Latino. Studies that include youths of other ethnic
groups would generate more insights into how racial differences are related to HPA axis
reactivity and its associated mental (e.g., depression (Pariante & Lightman, 2008)) and physical
(e.g., adiposity(Epel et al., 2000)) consequences.
This study is among the first to examine the momentary relationship between naturally-
occurring affective states and HPA axis activities at the within-person level, along with
identifying potential lifestyle moderators for the affect-HPA relationship, i.e., dietary intake and
physical activity. The momentary within-person relationship between negative affect and
subsequent salivary cortisol level suggests that fluctuations in negative mood place specific
demands on adolescents’ neuroendocrine system. Additionally, the discrepancies in the affect-
HPA axis relationships found between the two meal conditions indicate that the sugar and fiber
content of meals may have a role in HPA axis reactivity to psychological demands in youth.
While results of this study provide novel evidence toward untangling the momentary relationship
between affective states and the HPA axis, future studies with more extended observational
periods and a more representative sample may yield valuable insights to how meal contents
affect the HPA axis functions.

107 | P a g e

Table 10: Descriptive Statistics of the Study Participants
N=88 Mean Standard
Deviation
Range
Age 16.3 1.2 14.2-
18.5
N=88 Percentage
Gender (%) 48.96% Male
Ethnicity
African
American
43.18%
Latino 56.82%
Weight Status
Overweight* 5.7%
Obese** 94.3%
*: BMI Percentile ≥ 85
th
and <95
th
**: BMI Percentile ≥ 95
th

Table 11: Descriptive Statistics of the Affective States and Time-Spent in MVPA

Affective States Mean Standard Deviation Range
Feeling of panicky 4.64 9.58 0-100
Feeling of anxiousness 8.79 17.48 0-100
Feeling of worry 5.48 11.48 0-95
Feeling of nervousness 6.90 14.26 0-100
Average negative affective states 6.46 11.02 0-87.75
Feeling of calmness 83.32 23.40 0-100
Time Spent in MVPA (mins) 0.77 2.18 0-28.17
108 | P a g e

Figure 11: Average Cortisol Levels by Time

Salivary Cortisol Levels Mean Standard
Deviation
Range
Beginning of the lab visit 1.22 0.59 0.24-4.20
+30 Mins 0.98 0.47 0.25-3.15
+60 Mins 0.85 0.38 0.21-2.30
+90 Mins 0.72 0.31 0.20-1.80
+120 Mins 0.64 0.28 0.19-1.71
+150 Mins 0.60 0.26 0.02-1.41
+180 Mins 0.65 0.34 0.16-2.25
+210 Mins 0.62 0.31 0.13-1.81
+240 Mins 0.61 0.27 0.08-1.36
+270 Mins 0.72 0.39 0.12-1.83
+300 Mins 0.75 0.34 0.15-1.72
Parameter Estimates for the Effect of Time (linear and quadratic) on Cortisol Levels
Effect Estimate Standard Error p-values
Intercept 1.17 0.02 <.0001
Time -0.19 0.01 <.0001
Time*Time 0.02 0.001 <.0001
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
0 30 60 90 120 150 180 210 240 270 300
Salivary Cortisol Levels
Minutes Since Observation Started
109 | P a g e

Figure 12: Average Cortisol Levels by Time by Meal type (LS v.s. HS breakfast condition)
LS Condition HS condition Pairwise
Comparison
Time Mean Standard
Deviation
Mean Standard
Deviation
p-values
0 Min 1.22 0.58 1.22 0.59 0.9184
+30 Mins 0.98 0.45 0.97 0.50 0.8325
+60 Mins 0.88 0.37 0.81 0.39 0.1955
+90 Mins 0.74 0.30 0.71 0.32 0.5335
+120 Mins 0.66 0.29 0.62 0.26 0.5092
+150 Mins 0.61 0.26 0.59 0.26 0.7451
+180 Mins 0.63 0.34 0.67 0.35 0.4916
+210 Mins 0.59 0.29 0.64 0.32 0.4050
+240 Mins 0.59 0.25 0.63 0.28 0.5741
+270 Mins 0.70 0.40 0.74 0.38 0.4967
+300 Mins 0.77 0.36 0.73 0.33 0.5144
Parameter Estimates for the Effect of Time (linear and quadratic)
and Meal Type on Cortisol Levels
Effect Estimate Standard
Error
p-values
Intercept 1.19 0.03 <.0001
Time -0.19 0.01 <.0001
Time*Time 0.015 0.001 <.0001
Meal type -0.03 0.04 0.4594
Time*Meal type 0.006 0.02 0.7515
Time*Time*Meal type -0.0002 0.002 0.9383
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
0 30 60 90 120 150 180 210 240 270 300
Salivary Cortisol Levels
Minutes Since Observation Started
LSHF Visit
HSLF Visit
110 | P a g e

Table 12:Models of Negative Affect, Time Spent in MVPA, and Cortisol Levels Across Visit and
By Meal Type

Model 1:
Across Both Visits
Model 2:
HS
Model 3:
LS
Effect Estimate (SE) Estimate (SE) Estimate (SE)
Intercept 0.088 (0.654) 0.713 (0.757) -0.850 (0.796)
Negative Affect
(WS)
0.003 (0.002) 0.008 (0.003)** -0.002 (0.003)
Negative Affect
(BS)
-0.004 (0.003) -0.005 (0.003) -0.003 (0.004)
MVPA (WS)
0.014 (0.006)* 0.016 (0.008)*

0.014 (0.008)
+

MVPA (BS)
0.059 (0.036) 0.073 (0.043) 0.047 (0.044)
Time -0.226 (0.027)*** -0.148 (0.025)*** -0.231 (0.025)***
Time*Time 0.020 (0.003)*** 0.015 (0.002)*** 0.021 (0.002)***
Meal Type -0.197 (0.081)* - -
Meal
Type*Time
0.077 (0.037)* - -
Meal Type*
Time*Time
-0.006 (0.004) - -
Model Fit Statistics
-2 Log
Likelihood
1643.1 810.9 774.2
AIC 1677.6 839.5 802.8
BIC 1718.9 873.3 836.4
Age, gender, ethnicity, BMI percentile, and a binary variable adjusting for the order
of randomization were included in all models. AIC: Akaike information criterion;
BIC: Bayesian information +: p<0.10; *:p<0.050; **:p<0.010; ***:p<0.001
111 | P a g e

Table 13: Supplemental Table S1-Models of Average Negative Affect, Time Spent in MVPA, and
Cortisol Levels
Model 1 Model 2 Model 3 Model 4
Effect
Estimate
(SE)
Estimate
(SE)
Estimate
(SE)
Estimate
(SE)
Intercept 0.080 (0.66) 0.090 (0.650) 0.100 (0.660) 0.110 (0.660)
Negative Affect (WS) 0.011 (0.01)
+
0.010 (0.010) 0.010 (0.010)
+
0.000 (0.010)
Negative Affect (BS) -0.007 (0.005) -0.010 (0.010) -0.010 (0.010) 0.000 (0.010)
MVPA (WS) 0.014 (0.01) * 0.010 (0.010)* 0.020 (0.010)* 0.020 (0.010)*
MVPA (BS) 0.06 (0.04) 0.060 (0.040) 0.060 (0.040) 0.060 (0.040)
Negative Affect (WS)*
MVPA (WS)
- -
0.000
(0.000)
0.010 (0.010)
Negative Affect (BS)*
MVPA (BS)
- - 0.000 (0.010) 0.000 (0.010)
Meal Type -0.21 (0.08)* -0.210 (0.080)* -0.210 (0.080)* -0.220 (0.080)**
Negative Affect (WS)*
Meal Type
- - - 0.020 (0.020)
Negative Affect (BS)*
Meal Type
- - - -0.010 (0.010)
Negative Affect (WS)*
MVPA (WS)*
Meal Type
- - - 0.000 (0.010)
Negative Affect (BS)*
MVPA (BS)*
Meal Type
- - - 0.000 (0.010)
Time -0.23 (0.03)***
-0.230 (0.030)
***
-0.230 (0.030)*** -0.230 (0.030)***
Negative Affect (WS)*Time -0.005 (0.003) 0.000 (0.000) -0.010 (0.000) 0.000 (0.010)
Negative Affect (BS)*Time 0.001 (0.002) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000)
Negative Affect (WS)*
MVPA (WS)*
Time
- - 0.000 (0.000) 0.000 (0.000)
Negative Affect (BS)*
MVPA (BS)*
Time
- - 0.000 (0.000) 0.000 (0.000)
Meal Type*Time 0.082 (0.038)* 0.080 (0.040)* 0.080 (0.040)* 0.080 (0.040)*
Negative Affect (WS)*
Meal Type*Time
- 0.000 (0.000) - -0.010 (0.010)
112 | P a g e

Negative Affect (BS)*
Meal Type*Time
- 0.000 (0.000) - 0.000 (0.000)
Negative Affect (WS)*
MVPA (WS)*Meal Type*Time
- - - 0.000 (0.000)
Negative Affect (BS)*
MVPA (BS)*Meal Type*Time
- - - 0.000 (0.000)
Time*Time 0.020 (0.003)*** 0.020 (0.000) *** 0.020 (0.000) *** 0.020 (0.000)***
Negative Affect (WS)*
Time*Time
0.000 (0.000) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000)
Negative Affect (BS)*
Time*Time
0.000 (0.000) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000)
Negative Affect (WS)*
MVPA (WS)*
Time*Time
- 0.000 (0.000) 0.000 (0.000)
Negative Affect (BS)*
MVPA (BS)*
Time*Time
- 0.000 (0.000) 0.000 (0.000)
Meal Type*Time*Time -0.006 (0.004)
+
-0.010 (0.000)
+
-0.010 (0.000)
+
-0.010 (0.000)
+

Negative Affect (WS)*
Meal Type*Time*Time
- 0.000 (0.000) - 0.000 (0.000)
Negative Affect (BS)*
Meal Type*Time*Time
- 0.000 (0.000) - 0.000 (0.000)
Negative Affect (WS)*
MVPA (WS)*
Meal Type *Time*Time
- - - 0.000 (0.000)
Negative Affect (BS)*
MVPA (BS)*
Meal Type *Time*Time
- - 0.000 (0.000)
Model Fit Statistics
2 Log Likelihood 1637.5 1638.9 1630.7 1630.7
AIC (smaller is better) 1687.5 1692.9 1708.7 1708.7
BIC (smaller is better) 1748.9 1759.1 1804.4 1804.4
Age, gender, ethnicity, BMI percentile, and a binary variable adjusting for the order of randomization were included in all
models. AIC: Akaike information criterion; BIC: Bayesian information; +: p<0.10; *:p<0.050; **:p<0.010; ***:p<0.001
113 | P a g e

Table 14: Supplemental Table S2: Models of Feeling of Panicky, Time Spent in MVPA, and
Cortisol Levels Across Visit and By Meal Type

Model 1:
Across Both Visits
Model 2:
HS
Model 3:
LS
Effect Estimate (SE) Estimate (SE) Estimate (SE)
Intercept 0.091 (0.656) 0.701 (0.762) -0.840 (0.798)
Panic (WS) 0.002 (0.002) 0.007 (0.002)** -0.003 (0.003)
Panic (BS) -0.005 (0.004) -0.005 (0.005) -0.002 (0.005)
MVPA (WS) 0.014 (0.006)* 0.017 (0.008)* 0.014 (0.008)*
MVPA (BS) 0.056 (0.036) 0.071 (0.043)* 0.047 (0.044)
Time -0.227 (0.027)*** -0.149 (0.025)*** -0.231 (0.025)***
Time*Time 0.020 (0.003)*** 0.015 (0.002)*** 0.021 (0.002)***
Meal Type -0.193 (0.081)*
-

-

Meal Type*Time 0.076 (0.037)*

-
-

Meal
Type*Time*Time
-0.006 (0.004)
-

-

Age, gender, ethnicity, BMI percentile, and a binary variable adjusting for order of randomization
were included in all models. +: p<0.10; *:p<0.050; **:p<0.010; ***:p<0.001

114 | P a g e

Chapter 5: Discussion and Conclusion
Summary of Findings and Contributions to the Literature
This dissertation described three separate, but related studies that examined two main
research questions: 1) the relationship between affective states and HPA axis activities and 2)
whether the relationship between affective states and HPA axis activities were different when
participants engaged in more MVPA. Study 1 examined whether affective states are related to
diurnal cortisol rhythms at a day-level in adults and whether these relationships were moderated
by self-reported time spent in exercise. Study 1 is one of the first studies to explore whether high
or low arousal affective states, as described in the circumplex model of affect (Posner et al.,
2005), are associated with diurnal cortisol rhythms, including the diurnal cortisol slope (DCS) of
the same day and the cortisol awakening response (CAR) of the subsequent day. Study 2
examines whether affective states are related to HPA axis activities in an adolescent population.
This study is one of the few on this topic conducted in a youth population and the only study that
examines whether accelerometer-measured time-spent in MVPA moderates the relationship
between affective states and the HPA axis. Lastly, study 3 of this dissertation examines the
affect-HPA axis relationship in adolescents at a momentary level and potential role of MVPA as
a moderator of this affect-HPA axis relationship, using data collected from a randomized cross-
over trial that involved two 8-hour observational periods. Study 3 is among the first to examine
whether natural (i.e., not experimentally manipulated) fluctuations in affective states are acutely
related to cortisol levels at the subsequent 30 minutes. In the following section, the findings of
each study will be summarized separately, followed by discussions on how the results of the
three studies in this dissertation contribute to the existing literature.
115 | P a g e

Study 1 of this dissertation examined three research questions: 1) whether positive
affective states would be related to adaptive diurnal cortisol rhythm (i.e., steeper same day DCS
and blunted subsequent day CAR), 2) whether negative affective states would be related to
maladaptive diurnal cortisol rhythm (i.e., flatter same-day DCS and larger subsequent-day CAR),
and 3) whether time spent in exercise moderate the affect-HPA axis relationship. Study 1 found
partial support to the first hypothesis in that mentally healthy adults exhibited a blunted cortisol
awakening response (CAR) on the day after they have reported the higher-than-usual level of
high arousal positive affective states, but not average positive affective states. At the between-
person level, larger CARs were observed among adults who on average reported higher low-
arousal negative affective states on the previous day. On the other hand, study 1 results did not
support the hypothesis that negative affective states are related to the diurnal cortisol rhythm.
Self-reported time spent in exercise was also not related to the subsequent day CAR and did not
moderate the relationship between affective states and the subsequent day CAR. Study 1 did not
support Hypothesis stating that affective states (i.e., high/low arousal positive/negative affects)
are related to the diurnal cortisol slope (DCS) at the within-person level. At the between-person
level, DCSs were 1.2% flatter among adults who reported higher feelings of calm/peaceful and
satisfied (i.e., low arousal positive affect) than others. Lastly, although the main effect of self-
reported time spent in exercise on the same day DCSs was observed at the within-person level,
self-reported time spent in exercise did not moderate the relationship between affective states
and DCS.
Study 2 of this dissertation sought to address the same set of hypotheses in high school
adolescents. The results, however, did not support the hypotheses that affective states (i.e.,
positive or negative affective states) are related to the DCS. Additionally, accelerometer-
116 | P a g e

measured time spent in MVPA was neither related to the DCS nor moderated the relationship
between affective states and DCS. Although the null relationship between affective states and
DCS is consistent with the adult population, the discrepancy in the main effect of MVPA on
DCS between adults (study 1) and adolescents (study 2) is unexpected.
Study 3 of this dissertation examined the affect-HPA axis relationship in adolescents at
the momentary level in a crossover design in-lab setting where participants were randomized to
either high sugar meals or low sugar meals for the entire visit. Study 3 hypothesized that
affective states are acutely associated with cortisol levels at the subsequent 30 minutes and that
the accelerometer-measured time spent in MVPA during these 30 minutes moderates the affect-
HPA axis relationship. As an exploratory aim, study 3 further examined whether the affect-HPA
relationships differ on the day when they received high sugar meals compared to low sugar
meals. Results of this study revealed that average negative affect was related to a slightly
heightened (1.91%) salivary cortisol level at the subsequent 30 minutes at the within-person level
only on the lab visit when they were given a breakfast with sugar contents comparable to a
regular breakfast, but not on the visit when they were given a low-sugar high-fiber breakfast. In
study 3, accelerometer-measured time-spent in MVPA was associated with heightened (1.7%)
cortisol level at the subsequent 30 minutes at the within-person level at both visits but did not
moderate the relationship between affective states and cortisol activity.
Results from the three studies described in this dissertation provide partial support to the
assertion that affective states can be an important correlate of HPA axis activities at both the day-
and moment-level. More specifically, on the day-level, study 1 showed that high arousal positive
affect, rather than average positive affect, was associated with altered subsequent day CAR and
that none of the affective state predictors were associated with DCS of the same day in adults.
117 | P a g e

For study 2, although our ability to examine the effect of high arousal affective states on HPA
axis activities was limited by the affective state items used in this study, the null results observed
between average negative affect and DCS in adolescents were coherent with what was observed
in study 1 and in some of the available literature. The null findings observed across the two age
groups under investigation in this dissertation and the significant findings regarding high arousal
affective states may suggest that only a subset of affective states acutely activate the HPA axis.
These findings partially corroborate with the notion that when examined merely by its valence
(i.e., positive or negative), average affective states are not specific enough for detecting its
nuanced effect on the human neuroendocrinological system (Dickerson & Kemeny, 2004).
Along with the work by Hoyt and colleagues(Hoyt et al., 2015), study 1 is among the few that
examined whether different arousal components of affective states are related to diurnal cortisol
activity. Study 1 was also among the first to examine this question in an adult population. These
results highlight the importance of considering the arousal components of affective states, as
described in the circumplex model of affect, and examining the affect-HPA axis relationship.
Results from study 1, however, differ somewhat from those presented by Hoyt and colleagues.
Hoyt et al. found that, in adolescents, high arousal positive affective states are related to steeper
decline in DCS but not the subsequent day CAR(Hoyt et al., 2015), while study 1 of this
dissertation only found a significant association between high arousal positive affect and
subsequent day CAR. These discrepant findings might suggest that the adult HPA axis reacts to
naturally-occurring affective states differently from the adolescent HPA axis. This might be
related to the rapid changes occurring in many psychological and physiological functions during
adolescence (Heim & Nemeroff, 2002; Romeo, 2010). Future studies that investigate the
relationships between affective states and HPA axis activities by age group will be able to
118 | P a g e

elucidate the complex relationship among affect, the HPA axis, and age. relationship between
affective states and the HPA axis on the moment-level were somewhat different from those at the
day level. In study 3, we observed a heightened cortisol level after adolescents reported higher
average negative affect and the feeling of panic than their usual level, even in an environment
that was designed not to elicit any stressors. While small in magnitude, the study 3 results
showed that negative affect could momentarily affect subsequent salivary cortisol level.
Limitations
There are some limitations to the findings of this dissertation that should be noted. Most
noticeably, the affective states assessed in both study 1 and 2 may not necessarily represent how
study participants felt toward stressors or other stimuli. Affective states were collected right
before bedtime in study 1 and 2 and right upon waking for study 1. While affective state
measurements at these moments provide insights to how participants felt on the day of the
inquiry, it is possible that affective states measured at these moments are not representative of
the momentary experiences immediately after encountering stressors during the day. It is,
therefore, speculated that affective states measured at this sampling frequency or at these time-
points could not detect the nuanced changes in neuroendocrinology activities when the
participants encountered stressors. Study 3 of this dissertation, as well as the literature(Smyth et
al., 1998), show that unstimulated affective states have an acute impact on the HPA axis in
observational laboratory settings. These results offer preliminary evidence that, although effect
sizes were small in magnitude, affective states can acutely affect the HPA axis. Therefore, it is
possible that affective states measured at a higher frequency during the day could provide deeper
insights into whether and how affective states are related to HPA axis. Although future studies in
the ambulatory setting are necessary to confirm whether these results are generalizable in the
119 | P a g e

ambulatory environment, results of study 3 offer unique insights into the literature on the affect-
HPA axis relationship.
The other important limitation to be noted of this dissertation is the generalizability
across different ethnic groups: there is a lack of representation of ethnic minority groups in the
study 1 dataset while study 2 and 3 datasets included predominantly Hispanic and African
American youth. The differences in HPA diurnal cortisol rhythm and acute reactivity between
non-Hispanic White and various racial minorities have been documented (Ranjit et al., 2009).
Individuals’ momentary experiences can differ depending on their cultural and educational
backgrounds. Therefore, it is possible that the extent to which the HPA axis could react to an
individuals’ momentary experience at different magnitudes across ethnic groups. While results
from this dissertation have provided some novel contributions to the literature, it is important to
acknowledge that the affect-HPA axis relationship may differ according to race/ethnicity.
Implications of Study Findings
Despite these noted limitations, the three studies described in this dissertation offer
unique contributions to the growing literature investigating how affective states affect the human
neuroendocrinological system. For example, study 1 results offer support to the notion that high
arousal affective states, instead of average positive or negative affect, could be a day-level
correlate of the diurnal cortisol rhythm. This study is among the few, besides the work from Hoyt
and colleagues (Hoyt et al., 2015), that examined the specific contribution of high arousal
affective states on HPA axis activities. An important implication for these findings is that it is
important to consider both the valence and levels of emotional activation when examining the
relationship between affective states and the HPA axis. Studies that could afford to quantify the
level of emotional activation component of affective states often utilize scales that could be
120 | P a g e

burdensome for participants when implemented repeatedly. For example, participants in study 1
were asked to complete the modified PANAS, which contained 27 items, for eight days. Recent
efforts have attempted at reducing the number of items needed to capture the full circumplex
model of affective states (Yik, Russell, & Steiger, 2011). However, the utility of using these
scales to investigate the affect-HPA axis relationship is yet to be established. Therefore, future
investigations on the level of emotional activation on HPA axis activities need to consider the
balance between creating burdens to the survey respondents and obtaining affective states data
with enough detail to quantify the level of emotional activation when examining human affect-
HPA axis relationship. Study 3 provides several interesting contributions to the literature. One is
the preliminary evidence that affective states could acutely affect salivary cortisol levels.
Another interesting contribution is that the impacts of affective states on the HPA axis may differ
depending on the breakfast sugar and fiber contents. This finding may have potentially important
health implications. Although more studies are necessary to elucidate the underlying
mechanisms, results of study 3 provide preliminary evidence showing that adolescents’ HPA
axis were reactive to negative affect only when they were provided with high sugar meals.
Therefore, it is possible that modifiable health behaviors, such as reduction in sugar
consumptions, can be an effective strategy for attenuating the physiological impact of negative
affective states. Furthermore, these preliminary findings also could serve as a foundation for
more studies that investigate how hunger-related macronutrients and metabolites impact the
affect-HPA axis relationships. Finally, although preliminary, study 3 found that changes in HPA
axis activities could be detected at 30 minutes following fluctuation in affective states, instead of
the one-hour window previously documented in the literature (Smyth et al., 1998). Study 3
further indicates that the HPA axis is reactive to average negative affective states and feelings of
121 | P a g e

panic in the condition where youth were provided with high sugar meals while we did not detect
a significant affect-HPA relationship in the condition where they were provided with low sugar
meals. This result may suggest that behavioral factors, such as dietary behavior, could acutely
impact youths’ HPA axis reactivity to affective states. Therefore, modifying dietary behavior
may represent potential intervention targets for reducing HPA axis reactivity to emotional
experiences.
Future Research Directions
Chronic and repeated activation of the HPA axis could lead to the development of a
maladaptive HPA axis, exposing individuals to an elevated amount of cortisol, and increasing the
risks for various adverse health outcomes. This dissertation seeks to contribute to the literature
by examining how affective states relate to the HPA axis and whether energy-balance related
behaviors could moderate the affect-HPA axis relationship. Building on the preliminary findings
presented in this dissertation, future research that further examines the effect of high/low arousal
affective states on the HPA axis seems warranted. More specifically, studies with more intensive
sampling protocols for affective states and cortisol samples during the day may be able to
provide valuable insights into how cortisol rhythm fluctuates in response to affective states.
Future studies that examine the physiological impact of momentary experiences could potentially
offer viable intervention targets aiming to alter individuals’ exposure to excessive cortisol.
Furthermore, while this dissertation provides evidence on affect-HPA axis relationships using
intensive longitudinal data (e.g., study 1 collected 8-day of data and study 2 collected 10 saliva
samples for 2 days), it is yet unclear whether and how the acute HPA axis reactivity exacerbates
the development of a maladaptive HPA axis. Future studies that examine affect-HPA axis
relationships using intensive longitudinal data collected in multiple waves could potentially
122 | P a g e

provide more insights into this intricate relationship. Secondly, studies that examine other viable
behavioral moderators could potentially contribute to elucidating the affect-HPA axis
relationship. As illustrated in this dissertation and other studies (Wood et al., 2017), the
magnitude of HPA axis reactivity differs from acute exposure to behavioral interventions (e.g.,
high sugar meals). However, this literature is relatively nascent. Some behaviors might attenuate
HPA axis reactivity to multiple stressors. Cross-sectional studies that identify other potential
moderators and randomized cross-over trials similar to study 3 of this dissertation could
elucidate the underlying mechanisms that drive the relationship between emotional experiences
and the human neuroendocrine system. More importantly, this evidence could inform future
studies that aim to utilize behavioral intervention to attenuate the linkage between emotional
experience and an elevated cortisol exposure and thereby reduce the risks for disease morbidity.

123 | P a g e

References
Adam, E. K. (2006). Transactions among adolescent trait and state emotion and diurnal and
momentary cortisol activity in naturalistic settings. Psychoneuroendocrinology, 31(5),
664-679.
Adam, E. K., Hawkley, L. C., Kudielka, B. M., & Cacioppo, J. T. (2006). Day-to-day dynamics
of experience--cortisol associations in a population-based sample of older adults. Proc
Natl Acad Sci U S A, 103(45), 17058-17063.
Adam, E. K., Heissel, J. A., Zeiders, K. H., Richeson, J. A., Ross, E. C., Ehrlich, K. B., et al.
(2015). Developmental histories of perceived racial discrimination and diurnal cortisol
profiles in adulthood: A 20-year prospective study. Psychoneuroendocrinology, 62, 279-
291.
Adam, E. K., & Kumari, M. (2009). Assessing salivary cortisol in large-scale, epidemiological
research. Psychoneuroendocrinology, 34(10), 1423-1436.
Adam, E. K., Quinn, M. E., Tavernier, R., McQuillan, M. T., Dahlke, K. A., & Gilbert, K. E.
(2017). Diurnal Cortisol Slopes and Mental and Physical Health Outcomes: A Systematic
Review and Meta-analysis. Psychoneuroendocrinology.
Almeida, D. M., Davis, K. D., Lee, S., Lawson, K. M., Walter, K., & Moen, P. (2016).
Supervisor Support Buffers Daily Psychological and Physiological Reactivity to Work-
to-Family Conflict. J Marriage Fam, 78(1), 165-179.
Almeida, D. M., McGonagle, K., & King, H. (2009). Assessing daily stress processes in social
surveys by combining stressor exposure and salivary cortisol. Biodemography Soc Biol,
55(2), 219-237.
Almeida, D. M., Piazza, J. R., & Stawski, R. S. (2009). Interindividual differences and
intraindividual variability in the cortisol awakening response: an examination of age and
gender. Psychol Aging, 24(4), 819-827.
Almeida, D. M., Wethington, E., & Kessler, R. C. (2002). The daily inventory of stressful
events: an interview-based approach for measuring daily stressors. Assessment, 9(1), 41-
55.
American Psychological Association. (2017). Stress in America: Coping with Change. Stress in
America™ Survey. .
Arbel, R., Shapiro, L. S., Timmons, A. C., Moss, I. K., & Margolin, G. (2017). Adolescents'
Daily Worry, Morning Cortisol, and Health Symptoms. J Adolesc Health, 60(6), 667-
673.
Association, A. P. (2014). Stress in America: Are teens adopting adults’ stress habits. Stress in
America Surveys (〈 http://www. apa. org/news/press/releases/stress/2013/stress-report.
pdf〉).
Barrett, L. F. (2017). The theory of constructed emotion: an active inference account of
interoception and categorization. Soc Cogn Affect Neurosci.
Beckie, T. M. (2012). A systematic review of allostatic load, health, and health disparities. Biol
Res Nurs, 14(4), 311-346.
Belcher, B. R., Berrigan, D., Dodd, K. W., Emken, B. A., Chou, C.-P., & Spuijt-Metz, D. (2010).
Physical activity in US youth: Impact of race/ethnicity, age, gender, & weight status.
Medicine and science in sports and exercise, 42(12), 2211.
Bennett, G. G., Merritt, M. M., & Wolin, K. Y. (2004). Ethnicity, education, and the cortisol
response to awakening: a preliminary investigation. Ethn Health, 9(4), 337-347.
124 | P a g e

Benton, D. (2002). Carbohydrate ingestion, blood glucose and mood. Neurosci Biobehav Rev,
26(3), 293-308.
Benton, D., & Brock, H. (2010). Mood and the macro-nutrient composition of breakfast and the
mid-day meal. Appetite, 55(3), 436-440.
Bhattacharyya, M. R., Molloy, G. J., & Steptoe, A. (2008). Depression is associated with flatter
cortisol rhythms in patients with coronary artery disease. J Psychosom Res, 65(2), 107-
113.
Biddle, S. J., & Asare, M. (2011). Physical activity and mental health in children and
adolescents: a review of reviews. Br J Sports Med, 45(11), 886-895.
Bosch, J. A., Engeland, C. G., Cacioppo, J. T., & Marucha, P. T. (2007). Depressive symptoms
predict mucosal wound healing. Psychosom Med, 69(7), 597-605.
Broderick, J. E., Arnold, D., Kudielka, B. M., & Kirschbaum, C. (2004). Salivary cortisol
sampling compliance: comparison of patients and healthy volunteers.
Psychoneuroendocrinology, 29(5), 636-650.
Brosschot, J. F., Gerin, W., & Thayer, J. F. (2006). The perseverative cognition hypothesis: a
review of worry, prolonged stress-related physiological activation, and health. J
Psychosom Res, 60(2), 113-124.
Burke, H. M., Davis, M. C., Otte, C., & Mohr, D. C. (2005). Depression and cortisol responses to
psychological stress: a meta-analysis. Psychoneuroendocrinology, 30(9), 846-856.
Campbell, J., & Ehlert, U. (2012). Acute psychosocial stress: does the emotional stress response
correspond with physiological responses? Psychoneuroendocrinology, 37(8), 1111-1134.
Cannon, W. B. (1932). The wisdom of the body.
Champaneri, S., Xu, X., Carnethon, M. R., Bertoni, A. G., Seeman, T., DeSantis, A. S., et al.
(2013). Diurnal salivary cortisol is associated with body mass index and waist
circumference: the Multiethnic Study of Atherosclerosis. Obesity (Silver Spring), 21(1),
E56-63.
Chandler, J. L., Brazendale, K., Beets, M. W., & Mealing, B. A. (2016). Classification of
physical activity intensities using a wrist-worn accelerometer in 8-12-year-old children.
Pediatr Obes, 11(2), 120-127.
Charles, S. T., Mogle, J., Leger, K. A., & Almeida, D. M. (2017). Age and the Factor Structure
of Emotional Experience in Adulthood. J Gerontol B Psychol Sci Soc Sci.
Chen, L., Chi, P., Li, X., Zilioli, S., Zhao, J., Zhao, G., et al. (2017). The effects of trait and state
affect on diurnal cortisol slope among children affected by parental HIV/AIDS in rural
China. AIDS Care, 29(8), 1034-1040.
Chida, Y., & Steptoe, A. (2009). Cortisol awakening response and psychosocial factors: a
systematic review and meta-analysis. Biol Psychol, 80(3), 265-278.
Chrousos, G. P. (1995). The hypothalamic-pituitary-adrenal axis and immune-mediated
inflammation. N Engl J Med, 332(20), 1351-1362.
Chrousos, G. P. (2000). The role of stress and the hypothalamic-pituitary-adrenal axis in the
pathogenesis of the metabolic syndrome: neuro-endocrine and target tissue-related
causes. Int J Obes Relat Metab Disord, 24 Suppl 2, S50-55.
Chrousos, G. P. (2009). Stress and disorders of the stress system. Nat Rev Endocrinol, 5(7), 374-
381.
Clow, A., Hucklebridge, F., Stalder, T., Evans, P., & Thorn, L. (2010). The cortisol awakening
response: more than a measure of HPA axis function. Neurosci Biobehav Rev, 35(1), 97-
103.
125 | P a g e

Clow, A., Thorn, L., Evans, P., & Hucklebridge, F. (2004). The awakening cortisol response:
methodological issues and significance. Stress, 7(1), 29-37.
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. J
Health Soc Behav, 24(4), 385-396.
Cohen, S., Schwartz, J. E., Epel, E., Kirschbaum, C., Sidney, S., & Seeman, T. (2006).
Socioeconomic status, race, and diurnal cortisol decline in the Coronary Artery Risk
Development in Young Adults (CARDIA) Study. Psychosom Med, 68(1), 41-50.
Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person
effects in longitudinal models of change. Annu Rev Psychol, 62, 583-619.
Denson, T. F., Spanovic, M., & Miller, N. (2009). Cognitive appraisals and emotions predict
cortisol and immune responses: a meta-analysis of acute laboratory social stressors and
emotion inductions. Psychol Bull, 135(6), 823-853.
DeSantis, A. S., Adam, E. K., Hawkley, L. C., Kudielka, B. M., & Cacioppo, J. T. (2015). Racial
and ethnic differences in diurnal cortisol rhythms: are they consistent over time?
Psychosom Med, 77(1), 6-15.
DeSteno, D., Gross, J. J., & Kubzansky, L. (2013). Affective science and health: the importance
of emotion and emotion regulation. Health Psychol, 32(5), 474-486.
Dickerson, S. S., Gruenewald, T. L., & Kemeny, M. E. (2004). When the social self is
threatened: shame, physiology, and health. J Pers, 72(6), 1191-1216.
Dickerson, S. S., & Kemeny, M. E. (2004). Acute stressors and cortisol responses: a theoretical
integration and synthesis of laboratory research. Psychol Bull, 130(3), 355-391.
Doane, L. D., & Adam, E. K. (2010). Loneliness and cortisol: momentary, day-to-day, and trait
associations. Psychoneuroendocrinology, 35(3), 430-441.
Doane, L. D., Kremen, W. S., Eaves, L. J., Eisen, S. A., Hauger, R., Hellhammer, D., et al.
(2010). Associations between jet lag and cortisol diurnal rhythms after domestic travel.
Health Psychol, 29(2), 117-123.
Doane, L. D., & Zeiders, K. H. (2014). Contextual moderators of momentary cortisol and
negative affect in adolescents' daily lives. J Adolesc Health, 54(5), 536-542.
Dockray, S., & Steptoe, A. (2010). Positive affect and psychobiological processes. Neurosci
Biobehav Rev, 35(1), 69-75.
Dunton, G. F., Huh, J., Leventhal, A. M., Riggs, N., Hedeker, D., Spruijt-Metz, D., et al. (2014).
Momentary assessment of affect, physical feeling states, and physical activity in children.
Health Psychology, 33(3), 255-263.
Edwards, S. J., Braunholtz, D. A., Lilford, R. J., & Stevens, A. J. (1999). Ethical issues in the
design and conduct of cluster randomised controlled trials. [Review]. Bmj, 318(7195),
1407-1409.
Ekkekakis, P., Parfitt, G., & Petruzzello, S. J. (2011). The pleasure and displeasure people feel
when they exercise at different intensities: decennial update and progress towards a
tripartite rationale for exercise intensity prescription. Sports Med, 41(8), 641-671.
Ekman, P., & Cordaro, D. (2011). What is meant by calling emotions basic. Emotion Review,
3(4), 364-370.
Epel, E. S., McEwen, B., Seeman, T., Matthews, K., Castellazzo, G., Brownell, K. D., et al.
(2000). Stress and body shape: stress-induced cortisol secretion is consistently greater
among women with central fat. Psychosom Med, 62(5), 623-632.
126 | P a g e

Evans, P. D., Fredhoi, C., Loveday, C., Hucklebridge, F., Aitchison, E., Forte, D., et al. (2011).
The diurnal cortisol cycle and cognitive performance in the healthy old. Int J
Psychophysiol, 79(3), 371-377.
Everly Jr, G. S., & Lating, J. M. (2012). A clinical guide to the treatment of the human stress
response: Springer Science & Business Media.
Folkman, S. (2013). Stress: appraisal and coping: Springer.
Folkman, S., Lazarus, R. S., Dunkel-Schetter, C., DeLongis, A., & Gruen, R. J. (1986).
Dynamics of a stressful encounter: cognitive appraisal, coping, and encounter outcomes.
J Pers Soc Psychol, 50(5), 992-1003.
Freedson, P., Pober, D., & Janz, K. F. (2005). Calibration of accelerometer output for children.
Med Sci Sports Exerc, 37(11 Suppl), S523-530.
Fries, E., Dettenborn, L., & Kirschbaum, C. (2009). The cortisol awakening response (CAR):
facts and future directions. Int J Psychophysiol, 72(1), 67-73.
Gerber, M., & Pühse, U. (2009). Review article: do exercise and fitness protect against stress-
induced health complaints? A review of the literature. Scand J Public Health, 37(8), 801-
819.
Gunnar, M., & Quevedo, K. (2007). The neurobiology of stress and development. Annu Rev
Psychol, 58, 145-173.
Hajat, A., Diez-Roux, A., Franklin, T. G., Seeman, T., Shrager, S., Ranjit, N., et al. (2010).
Socioeconomic and race/ethnic differences in daily salivary cortisol profiles: the multi-
ethnic study of atherosclerosis. Psychoneuroendocrinology, 35(6), 932-943.
Heim, C., & Nemeroff, C. B. (2002). Neurobiology of early life stress: clinical studies. Semin
Clin Neuropsychiatry, 7(2), 147-159.
Herbert, J. (2013). Cortisol and depression: three questions for psychiatry. Psychol Med, 43(3),
449-469.
Hill, E. E., Zack, E., Battaglini, C., Viru, M., Viru, A., & Hackney, A. C. (2008). Exercise and
circulating cortisol levels: the intensity threshold effect. J Endocrinol Invest, 31(7), 587-
591.
Hollanders, J. J., van der Voorn, B., Rotteveel, J., & Finken, M. J. J. (2017). Is HPA axis
reactivity in childhood gender-specific? A systematic review. Biol Sex Differ, 8(1), 23.
Hoppmann, C. A., & Riediger, M. (2009). Ambulatory assessment in lifespan psychology: An
overview of current status and new trends. European Psychologist, 14(2), 98-108.
Hoyt, L. T., Craske, M. G., Mineka, S., & Adam, E. K. (2015). Positive and negative affect and
arousal: cross-sectional and longitudinal associations with adolescent cortisol diurnal
rhythms. Psychosom Med, 77(4), 392-401.
Hoyt, L. T., Zeiders, K. H., Ehrlich, K. B., & Adam, E. K. (2016). Positive upshots of cortisol in
everyday life. Emotion, 16(4), 431-435.
Huber, T. J., Issa, K., Schik, G., & Wolf, O. T. (2006). The cortisol awakening response is
blunted in psychotherapy inpatients suffering from depression.
Psychoneuroendocrinology, 31(7), 900-904.
Human, L. J., Whillans, A. V., Hoppmann, C. A., Klumb, P., Dickerson, S. S., & Dunn, E. W.
(2015). Finding the middle ground: Curvilinear associations between positive affect
variability and daily cortisol profiles. Emotion, 15(6), 705-720.
Incollingo Rodriguez, A. C., Epel, E. S., White, M. L., Standen, E. C., Seckl, J. R., & Tomiyama,
A. J. (2015). Hypothalamic-pituitary-adrenal axis dysregulation and cortisol activity in
obesity: A systematic review. Psychoneuroendocrinology, 62, 301-318.
127 | P a g e

Kerr, J., Anderson, C., & Lippman, S. M. (2017). Physical activity, sedentary behaviour, diet,
and cancer: an update and emerging new evidence. Lancet Oncol, 18(8), e457-e471.
Kudielka, B., & Kirschbaum, C. (2003). Awakening cortisol responses are influenced by health
status and awakening time but not by menstrual cycle phase. Psychoneuroendocrinology,
28(1), 35-47.
Kumari, M., Badrick, E., Chandola, T., Adam, E. K., Stafford, M., Marmot, M. G., et al. (2009).
Cortisol secretion and fatigue: associations in a community based cohort.
Psychoneuroendocrinology, 34(10), 1476-1485.
Kumari, M., Chandola, T., Brunner, E., & Kivimaki, M. (2010). A nonlinear relationship of
generalized and central obesity with diurnal cortisol secretion in the Whitehall II study. J
Clin Endocrinol Metab, 95(9), 4415-4423.
Kunz-Ebrecht, S. R., Kirschbaum, C., Marmot, M., & Steptoe, A. (2004). Differences in cortisol
awakening response on work days and weekends in women and men from the Whitehall
II cohort. Psychoneuroendocrinology, 29(4), 516-528.
Larson, R. W., & Richards, M. H. (1994). Family emotions: Do young adolescents and their
parents experience the same states? Journal of Research on Adolescence, 4(4), 567-583.
Law, R., Hucklebridge, F., Thorn, L., Evans, P., & Clow, A. (2013). State variation in the
cortisol awakening response. Stress, 16(5), 483-492.
Lazarus, R. S. (1991). Emotion and adaptation: Oxford University Press on Demand.
Lazarus, R. S. (2001). Relational meaning and discrete emotions.
Liao, Y., Shonkoff, E. T., & Dunton, G. F. (2015). The Acute Relationships Between Affect,
Physical Feeling States, and Physical Activity in Daily Life: A Review of Current
Evidence. Front Psychol, 6, 1975.
Lippold, M. A., Davis, K. D., McHale, S. M., Buxton, O. M., & Almeida, D. M. (2016). Daily
stressor reactivity during adolescence: The buffering role of parental warmth. Health
Psychol, 35(9), 1027-1035.
Lopez-Duran, N. L., Kovacs, M., & George, C. J. (2009). Hypothalamic-pituitary-adrenal axis
dysregulation in depressed children and adolescents: a meta-analysis.
Psychoneuroendocrinology, 34(9), 1272-1283.
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.
Maina, G., Palmas, A., Bovenzi, M., & Filon, F. L. (2009). Salivary cortisol and psychosocial
hazards at work. Am J Ind Med, 52(3), 251-260.
Martikainen, S., Pesonen, A. K., Lahti, J., Heinonen, K., Feldt, K., Pyhälä, R., et al. (2013).
Higher levels of physical activity are associated with lower hypothalamic-pituitary-
adrenocortical axis reactivity to psychosocial stress in children. J Clin Endocrinol Metab,
98(4), E619-627.
Matias, G. P., Nicolson, N. A., & Freire, T. (2011). Solitude and cortisol: associations with state
and trait affect in daily life. Biol Psychol, 86(3), 314-319.
McEwen, B. S. (2007). Physiology and neurobiology of stress and adaptation: central role of the
brain. Physiol Rev, 87(3), 873-904.
McEwen, B. S. (2008). Central effects of stress hormones in health and disease: Understanding
the protective and damaging effects of stress and stress mediators. Eur J Pharmacol,
583(2-3), 174-185.
McEwen, B. S., & Gianaros, P. J. (2010). Central role of the brain in stress and adaptation: links
to socioeconomic status, health, and disease. Ann N Y Acad Sci, 1186, 190-222.
128 | P a g e

Miller, K. G., Wright, A. G., Peterson, L. M., Kamarck, T. W., Anderson, B. A., Kirschbaum, C.,
et al. (2016). Trait positive and negative emotionality differentially associate with diurnal
cortisol activity. Psychoneuroendocrinology, 68, 177-185.
Miller, R., Stalder, T., Jarczok, M., Almeida, D. M., Badrick, E., Bartels, M., et al. (2016). The
CIRCORT database: Reference ranges and seasonal changes in diurnal salivary cortisol
derived from a meta-dataset comprised of 15 field studies. Psychoneuroendocrinology,
73, 16-23.
Moore, C. J., & Cunningham, S. A. (2012). Social position, psychological stress, and obesity: a
systematic review. J Acad Nutr Diet, 112(4), 518-526.
Nater, U. M., Youngblood, L. S., Jones, J. F., Unger, E. R., Miller, A. H., Reeves, W. C., et al.
(2008). Alterations in diurnal salivary cortisol rhythm in a population-based sample of
cases with chronic fatigue syndrome. Psychosom Med, 70(3), 298-305.
Nguyen-Rodriguez, S. T., Chou, C. P., Unger, J. B., & Spruijt-Metz, D. (2008). BMI as a
moderator of perceived stress and emotional eating in adolescents. Eat Behav, 9(2), 238-
246.
O'Reilly, G. A., Belcher, B. R., Davis, J. N., Martinez, L. T., Huh, J., Antunez-Castillo, L., et al.
(2015). Effects of high-sugar and high-fiber meals on physical activity behaviors in
Latino and African American adolescents. Obesity (Silver Spring), 23(9), 1886-1894.
Oskis, A., Loveday, C., Hucklebridge, F., Thorn, L., & Clow, A. (2009). Diurnal patterns of
salivary cortisol across the adolescent period in healthy females.
Psychoneuroendocrinology, 34(3), 307-316.
Papadopoulos, A. S., & Cleare, A. J. (2011). Hypothalamic-pituitary-adrenal axis dysfunction in
chronic fatigue syndrome. Nat Rev Endocrinol, 8(1), 22-32.
Pariante, C. M., & Lightman, S. L. (2008). The HPA axis in major depression: classical theories
and new developments. Trends Neurosci, 31(9), 464-468.
Pessoa, L., & Adolphs, R. (2010). Emotion processing and the amygdala: from a 'low road' to
'many roads' of evaluating biological significance. Nat Rev Neurosci, 11(11), 773-783.
Piazza, J. R., Charles, S. T., Stawski, R. S., & Almeida, D. M. (2013). Age and the association
between negative affective states and diurnal cortisol. Psychol Aging, 28(1), 47-56.
Platje, E., Vermeiren, R. R., Branje, S. J., Doreleijers, T. A., Meeus, W. H., Koot, H. M., et al.
(2013). Long-term stability of the cortisol awakening response over adolescence.
Psychoneuroendocrinology, 38(2), 271-280.
Polk, D. E., Cohen, S., Doyle, W. J., Skoner, D. P., & Kirschbaum, C. (2005). State and trait
affect as predictors of salivary cortisol in healthy adults. Psychoneuroendocrinology,
30(3), 261-272.
Posner, J., Russell, J. A., & Peterson, B. S. (2005). The circumplex model of affect: an
integrative approach to affective neuroscience, cognitive development, and
psychopathology. Dev Psychopathol, 17(3), 715-734.
Powell, D. J., Liossi, C., Moss-Morris, R., & Schlotz, W. (2013). Unstimulated cortisol secretory
activity in everyday life and its relationship with fatigue and chronic fatigue syndrome: a
systematic review and subset meta-analysis. Psychoneuroendocrinology, 38(11), 2405-
2422.
Pruessner, J. C., Hellhammer, D. H., & Kirschbaum, C. (1999). Burnout, perceived stress, and
cortisol responses to awakening. Psychosom Med, 61(2), 197-204.
129 | P a g e

Pruessner, J. C., Wolf, O. T., Hellhammer, D. H., Buske-Kirschbaum, A., von Auer, K., Jobst,
S., et al. (1997). Free cortisol levels after awakening: a reliable biological marker for the
assessment of adrenocortical activity. Life Sci, 61(26), 2539-2549.
Puterman, E., O'Donovan, A., Adler, N. E., Tomiyama, A. J., Kemeny, M., Wolkowitz, O. M., et
al. (2011). Physical activity moderates effects of stressor-induced rumination on cortisol
reactivity. Psychosom Med, 73(7), 604-611.
Ranjit, N., Diez-Roux, A. V., Sanchez, B., Seeman, T., Shea, S., Shrager, S., et al. (2009).
Association of salivary cortisol circadian pattern with cynical hostility: multi-ethnic study
of atherosclerosis. Psychosom Med, 71(7), 748-755.
Ranjit, N., Young, E. A., Raghunathan, T. E., & Kaplan, G. A. (2005). Modeling cortisol
rhythms in a population-based study. Psychoneuroendocrinology, 30(7), 615-624.
Rimmele, U., Seiler, R., Marti, B., Wirtz, P. H., Ehlert, U., & Heinrichs, M. (2009). The level of
physical activity affects adrenal and cardiovascular reactivity to psychosocial stress.
Psychoneuroendocrinology, 34(2), 190-198.
Rimmele, U., Zellweger, B. C., Marti, B., Seiler, R., Mohiyeddini, C., Ehlert, U., et al. (2007).
Trained men show lower cortisol, heart rate and psychological responses to psychosocial
stress compared with untrained men. Psychoneuroendocrinology, 32(6), 627-635.
Romeo, R. D. (2010). Adolescence: a central event in shaping stress reactivity. Dev Psychobiol,
52(3), 244-253.
Rosmond, R., Dallman, M. F., & Björntorp, P. (1998). Stress-related cortisol secretion in men:
relationships with abdominal obesity and endocrine, metabolic and hemodynamic
abnormalities. J Clin Endocrinol Metab, 83(6), 1853-1859.
Ross, K. M., Murphy, M. L., Adam, E. K., Chen, E., & Miller, G. E. (2014). How stable are
diurnal cortisol activity indices in healthy individuals? Evidence from three multi-wave
studies. Psychoneuroendocrinology, 39, 184-193.
Rotenberg, S., & McGrath, J. J. (2014). Sampling compliance for cortisol upon awakening in
children and adolescents. Psychoneuroendocrinology, 40, 69-75.
Rotenberg, S., McGrath, J. J., Roy-Gagnon, M. H., & Tu, M. T. (2012). Stability of the diurnal
cortisol profile in children and adolescents. Psychoneuroendocrinology, 37(12), 1981-
1989.
Ryff, C. D., & Almeida, D. M. (2010). Midlife in the United States (MIDUS 2): Daily Stress
Project, 2004-2009. ICPSR26841-v1. Ann Arbor, MI: Inter-university Consortium for
Political and Social Research [distributor], 02-26.
Sadeh, A., Sharkey, K. M., & Carskadon, M. A. (1994). Activity-Based Sleep—Wake
Identification: An Empirical Test of Methodological Issues. Sleep, 17(3), 201-207.
Savla, J., Zarit, S. H., & Almeida, D. M. (2017). Routine Support to Parents and Stressors in
Everyday Domains: Associations With Negative Affect and Cortisol. J Gerontol B
Psychol Sci Soc Sci.
Saxbe, D. E. (2008). A field (researcher's) guide to cortisol: tracking HPA axis functioning in
everyday life. Health Psychology Review, 2(2), 163-190.
Schembre, S. M., Wen, C. K., Davis, J. N., Shen, E., Nguyen-Rodriguez, S. T., Belcher, B. R., et
al. (2013). Eating breakfast more frequently is cross-sectionally associated with greater
physical activity and lower levels of adiposity in overweight Latina and African
American girls. Am J Clin Nutr, 98(2), 275-281.
130 | P a g e

Schmidt-Reinwald, A., Pruessner, J. C., Hellhammer, D. H., Federenko, I., Rohleder, N.,
Schürmeyer, T. H., et al. (1999). The cortisol response to awakening in relation to
different challenge tests and a 12-hour cortisol rhythm. Life Sci, 64(18), 1653-1660.
Shirtcliff, E. A., Allison, A. L., Armstrong, J. M., Slattery, M. J., Kalin, N. H., & Essex, M. J.
(2012). Longitudinal stability and developmental properties of salivary cortisol levels and
circadian rhythms from childhood to adolescence. Dev Psychobiol, 54(5), 493-502.
Sin, N. L., Ong, A. D., Stawski, R. S., & Almeida, D. M. (2017). Daily positive events and
diurnal cortisol rhythms: Examination of between-person differences and within-person
variation. Psychoneuroendocrinology, 83, 91-100.
Sladek, M. R., Doane, L. D., & Stroud, C. B. (2017). Individual and Day-to-Day Differences in
Active Coping Predict Diurnal Cortisol Patterns among Early Adolescent Girls. J Youth
Adolesc, 46(1), 121-135.
Slater, J. A., Botsis, T., Walsh, J., King, S., Straker, L. M., & Eastwood, P. R. (2015). Assessing
sleep using hip and wrist actigraphy. Sleep and Biological Rhythms, 13(2), 172-180.
Smyth, J., Ockenfels, M. C., Porter, L., Kirschbaum, C., Hellhammer, D. H., & Stone, A. A.
(1998). Stressors and mood measured on a momentary basis are associated with salivary
cortisol secretion. Psychoneuroendocrinology, 23(4), 353-370.
Sothmann, M. S., Buckworth, J., Claytor, R. P., Cox, R. H., White-Welkley, J. E., & Dishman,
R. K. (1996). Exercise training and the cross-stressor adaptation hypothesis. Exerc Sport
Sci Rev, 24, 267-287.
Spruijt-Metz, D., Belcher, B. R., Hsu, Y. W., McClain, A. D., Chou, C. P., Nguyen-Rodriguez,
S., et al. (2013). Temporal relationship between insulin sensitivity and the pubertal
decline in physical activity in peripubertal Hispanic and African American females.
Diabetes Care, 36(11), 3739-3745.
Stalder, T., Bäumler, D., Miller, R., Alexander, N., Kliegel, M., & Kirschbaum, C. (2013). The
cortisol awakening response in infants: ontogeny and associations with development-
related variables. Psychoneuroendocrinology, 38(4), 552-559.
Stalder, T., Evans, P., Hucklebridge, F., & Clow, A. (2010). State associations with the cortisol
awakening response in healthy females. Psychoneuroendocrinology, 35(8), 1245-1252.
Stalder, T., Kirschbaum, C., Kudielka, B. M., Adam, E. K., Pruessner, J. C., Wüst, S., et al.
(2016). Assessment of the cortisol awakening response: Expert consensus guidelines.
Psychoneuroendocrinology, 63, 414-432.
Steptoe, A., Dockray, S., & Wardle, J. (2009). Positive affect and psychobiological processes
relevant to health. J Pers, 77(6), 1747-1776.
Steptoe, A., Kunz-Ebrecht, S. R., Brydon, L., & Wardle, J. (2004). Central adiposity and cortisol
responses to waking in middle-aged men and women. Int J Obes Relat Metab Disord,
28(9), 1168-1173.
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 (pp. 629-649).
New York: John Wiley and Sons.
Strahler, J., Fuchs, R., Nater, U. M., & Klaperski, S. (2016). Impact of physical fitness on
salivary stress markers in sedentary to low-active young to middle-aged men.
Psychoneuroendocrinology, 68, 14-19.
Suglia, S. F., Staudenmayer, J., Cohen, S., & Wright, R. J. (2010). Posttraumatic stress
symptoms related to community violence and children’s diurnal cortisol response in an
131 | P a g e

urban community-dwelling sample. International journal of behavioral medicine, 17(1),
43-50.
Terry, P. C., Lane, A. M., Lane, H. J., & Keohane, L. (1999). Development and validation of a
mood measure for adolescents. J Sports Sci, 17(11), 861-872.
Therrien, F., Drapeau, V., Lalonde, J., Lupien, S. J., Beaulieu, S., Tremblay, A., et al. (2007).
Awakening cortisol response in lean, obese, and reduced obese individuals: effect of
gender and fat distribution. Obesity, 15(2), 377-385.
Tracy, J. L., & Randles, D. (2011). Four models of basic emotions: a review of Ekman and
Cordaro, Izard, Levenson, and Panksepp and Watt. Emotion Review, 3(4), 397-405.
Tsigos, C., & Chrousos, G. P. (2002). Hypothalamic-pituitary-adrenal axis, neuroendocrine
factors and stress. J Psychosom Res, 53(4), 865-871.
Ulrich-Lai, Y. M., & Ryan, K. K. (2014). Neuroendocrine circuits governing energy balance and
stress regulation: functional overlap and therapeutic implications. Cell Metab, 19(6), 910-
925.
Ursache, A., Wedin, W., Tirsi, A., & Convit, A. (2012). Preliminary evidence for obesity and
elevations in fasting insulin mediating associations between cortisol awakening response
and hippocampal volumes and frontal atrophy. Psychoneuroendocrinology, 37(8), 1270-
1276.
van de Rest, O., van der Zwaluw, N. L., & de Groot, L. C. P. G. (2018). Effects of glucose and
sucrose on mood: a systematic review of interventional studies. Nutr Rev, 76(2), 108-116.
VanBruggen, M. D., Hackney, A. C., McMurray, R. G., & Ondrak, K. S. (2011). The
relationship between serum and salivary cortisol levels in response to different intensities
of exercise. Int J Sports Physiol Perform, 6(3), 396-407.
Vreeburg, S. A., Kruijtzer, B. P., van Pelt, J., van Dyck, R., DeRijk, R. H., Hoogendijk, W. J., et
al. (2009). Associations between sociodemographic, sampling and health factors and
various salivary cortisol indicators in a large sample without psychopathology.
Psychoneuroendocrinology, 34(8), 1109-1120.
Wang, P. S., Berglund, P., & Kessler, R. C. (2000). Recent care of common mental disorders in
the United States : prevalence and conformance with evidence-based recommendations. J
Gen Intern Med, 15(5), 284-292.
Warburton, D. E., Nicol, C. W., & Bredin, S. S. (2006). Health benefits of physical activity: the
evidence. CMAJ, 174(6), 801-809.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures
of positive and negative affect: the PANAS scales. J Pers Soc Psychol, 54(6), 1063-1070.
Wegner, M., Helmich, I., Machado, S., Nardi, A. E., Arias-Carrion, O., & Budde, H. (2014).
Effects of exercise on anxiety and depression disorders: review of meta- analyses and
neurobiological mechanisms. CNS Neurol Disord Drug Targets, 13(6), 1002-1014.
Weigensberg, M. J., Spruijt-Metz, D., Wen, C. K. F., Davis, J. N., Ávila, Q., Juarez, M., et al.
(2018). Protocol for the Imagine HEALTH Study: Guided imagery lifestyle intervention
to improve obesity-related behaviors and salivary cortisol patterns in predominantly
Latino adolescents. Contemp Clin Trials, 72, 103-116.
Weigensberg, M. J., Toledo-Corral, C. M., & Goran, M. I. (2008). Association between the
metabolic syndrome and serum cortisol in overweight Latino youth. J Clin Endocrinol
Metab, 93(4), 1372-1378.
132 | P a g e

Wen, C. K. F., Liao, Y., Maher, J. P., Huh, J., Belcher, B. R., Dzubur, E., et al. (2018).
Relationships among affective states, physical activity, and sedentary behavior in
children: Moderation by perceived stress. Health Psychol, 37(10), 904-914.
Wen, C. K. F., Weigensberg, M., Schneider, S., Weerman, B., & Spruijt-Metz, D. (2015).
Validity of ZEMI on Ambulatory Salivary Cortisol Assessment in an Adolescent
Population. Paper presented at the Wireless Health 2015.
Willner, C. J., Morris, P. A., McCoy, D. C., & Adam, E. K. (2014). Diurnal cortisol rhythms in
youth from risky families: effects of cumulative risk exposure and variation in the
serotonin transporter gene-linked polymorphic region (5-HTTLPR) [corrected]. Dev
Psychopathol, 26(4 Pt 1), 999-1019.
Wood, C. J., Clow, A., Hucklebridge, F., Law, R., & Smyth, N. (2017). Physical fitness and prior
physical activity are both associated with less cortisol secretion during psychosocial
stress. Anxiety Stress Coping, 1-11.
Ye, Z., Arumugam, V., Haugabrooks, E., Williamson, P., & Hendrich, S. (2015). Soluble dietary
fiber (Fibersol-2) decreased hunger and increased satiety hormones in humans when
ingested with a meal. Nutr Res, 35(5), 393-400.
Yik, M., Russell, J. A., & Steiger, J. H. (2011). A 12-Point Circumplex Structure of Core Affect.
Emotion, 11(4), 705-731.
Zeiders, K. H., Hoyt, L. T., & Adam, E. K. (2014). Associations between self-reported
discrimination and diurnal cortisol rhythms among young adults: The moderating role of
racial–ethnic minority status. Psychoneuroendocrinology, 50, 280-288.
Šupe-Domić, D., Milas, G., Hofman, I. D., Rumora, L., & Klarić, I. M. (2016). Daily salivary
cortisol profile: Insights from the Croatian Late Adolescence Stress Study (CLASS).
Biochem Med (Zagreb), 26(3), 408-420.

Abstract (if available)

Abstract Mounting evidence has suggested the elevated disease morbidity associated with a maladaptive diurnal cortisol rhythm. Identifying predictors of the diurnal cortisol rhythm and its moderators could potentially bear significant health implications by offering viable targets for interventions. Affective states, or the individuals’ experience toward an actual or perceived stimulus, have shown to be associated with diurnal cortisol rhythm. However, the mixed results suggest the need for studies that examine how affective states predict diurnal cortisol rhythm and whether there are behavioral moderators to this relationship. This dissertation examined two main research questions: 1) the relationship between affective states and HPA axis activities and 2) whether the relationship between affective states and HPA axis activities were different when participants engaged in more MVPA both at the within- and between-person levels. ❧ Study 1 found that mentally healthy adults exhibited a blunted cortisol awakening response (CAR) on the day after they have reported the higher-than-usual level of high arousal positive affective states, but not average positive affective states. At the between-person level, larger CARs were observed among adults who on average reported higher low-arousal negative affective states on the previous day. Self-reported time spent in the exercise was also not related to the subsequent day CAR and did not moderate the relationship between affective states and the subsequent day CAR. Contrary to the hypotheses, study 1 did not support the notion that affective states are related to the diurnal cortisol slope (DCS) at the within-person level. At the between-person level, DCSs were 1.2% flatter among adults who reported higher feelings of calm/peaceful and satisfied (i.e., low arousal positive affect) than others. ❧ Contrary to the findings from study 1, results of the study did not support the hypotheses that affective states (i.e., positive or negative affective states) are related to the DCS in mentally healthy youth, both at the within- and between-person levels. Additionally, accelerometer-measured time spent in MVPA was neither related to the DCS nor moderated the relationship between affective states and DCS. ❧ Study 3 further examined the acute relationship between affective states and cortisol levels at the subsequent 30 minutes in youth. Results of this study showed that average negative affective states were related to a slightly heightened (1.91%) salivary cortisol level at the subsequent 30 minutes at the within-person level only on the lab visit when young participants were given a breakfast with high in sugar contents, but not on the visit when they were given a low-sugar high-fiber breakfast. Accelerometer-measured time-spent in MVPA was associated with heightened (1.7%) cortisol level at the subsequent 30 minutes at the within-person level at both visits but did not moderate the relationship between affective states and cortisol activity. ❧ This dissertation offers unique insights into the growing literature investigating how affective states affect the human neuroendocrinological system. Overall, results of this dissertation provide partial support to the assertion that affective states can be an important correlate of HPA axis activities at both the day- and moment-level and that energy-balance related behavior could alter how affective states are related to the hypothalamic-pituitary-adrenal axis.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

PDF

Investigating a physiological pathway for the effect of guided imagery on insulin resistance

PDF

Motivation and the meanings of health behavior as factors associated with eating behavior in Latino youth

PDF

The acute and longitudinal associations between sedentary behaviors, affective states, and emotional disorder symptoms among youth

PDF

Marginalization in acculturation is related to objectively measured physical activity in Latina adolecents

PDF

Influences of specific environmental domains on childhood obesity and related behaviors

PDF

Mixed methods investigation of user engagement with a smoking cessation app

PDF

The vicious cycle of inactivity, obesity, and metabolic health consequences in at-risk pediatric populations

PDF

Prenatal sleep health, cortisol, and gestational weight gain

PDF

Adolescent life stress and the cortisol awakening response: the moderating roles of emotion regulation, attachment, and gender

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

Effects of sugar and fiber consumption in minority adolescents and self-tracking as a potential dietary intervention tool

PDF

Addressing federal pain research priorities: drug policy, pain mechanisms, and integrative treatment

PDF

The role of social support in the relationship between adverse childhood experiences and addictive behaviors across adolescence and young adulthood

PDF

Effects of the perceived and objectively assessed environment on physical activity in adults and children

PDF

Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology

Asset Metadata

Creator Wen, Cheng Kun (author)

Core Title The acute relationship between affective states and physiological stress response, and the moderating role of moderate-to-vigorous physical activity

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior Research)

Publication Date 11/08/2020

Defense Date 10/18/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag affective states,diurnal cortisol rhythm,Exercise,moderate-to-vigorous physical activity,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Chou, Chih-Ping (committee chair), Spruijt-Metz, Donna (committee chair), Belcher, Britni R. (committee member), Black, David S. (committee member), Weigensberg, Marc J. (committee member)

Creator Email chengkuw@usc.edu,darse1112@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-102582

Unique identifier UC11675786

Identifier etd-WenChengKu-6948.pdf (filename),usctheses-c89-102582 (legacy record id)

Legacy Identifier etd-WenChengKu-6948-0.pdf

Dmrecord 102582

Document Type Dissertation

Format application/pdf (imt)

Rights Wen, Cheng Kun

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA