Autonomic inertia as a proximal risk marker for moments of perseverative cognition in everyday life in remitted depression

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2024 Sarah L. Zapetis
AUTONOMIC INERTIA AS A PROXIMAL RISK MARKER FOR MOMENTS OF
PERSEVERATIVE COGNITION IN EVERYDAY LIFE IN REMITTED DEPRESSION
by
Sarah L. Zapetis,B.S.
A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS, ARTS AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(PSYCHOLOGY)
December 2024

Copyright 2024 ii Sarah L. Zapetis
Dedication
To my partner, parents, friends, and cats—thank you for your unwavering emotional support.
Your encouragement, patience, and love have made this journey possible.

Copyright 2024 iii Sarah L. Zapetis
Acknowledgements
I would like to express my gratitude to my mentor, Jonathan P. Stange, and my lab mates for
their invaluable technical support throughout this project. Their guidance and expertise have
been essential to the completion of this work.

Copyright 2024 iv Sarah L. Zapetis
Table of Contents
Dedication .......................................................................................................................................ii
Acknowledgements .......................................................................................................................iii
List of Tables .................................................................................................................................vi
List of Figures ...............................................................................................................................vii
Abbreviations ...............................................................................................................................viii
Abstract ..........................................................................................................................................ix
Chapter 1: Introduction ...................................................................................................................1
Chapter 2: Materials and Methods ..................................................................................................9
Participants ..........................................................................................................................9
Procedures .........................................................................................................................10
Ecological Momentary Assessment ......................................................................10
Ambulatory Psychophysiology .............................................................................12
Statistical Analyses ...........................................................................................................14
Group Comparisons ..............................................................................................15
Within-Person Analyses ........................................................................................15
Between-Person Analyses .....................................................................................16
Comparisons of Inertia Metrics with Different Time Lags ..................................16
Chapter 3: Results .........................................................................................................................17
Descriptive Statistics about Ambulatory Data ..................................................................17
Group Comparisons ..........................................................................................................17
Within-Person Associations between Autonomic Complexity and PC.............................18
Between-Person Associations between Autonomic Complexity and PC..........................21
Chapter 4: Discussion ...................................................................................................................24

Copyright 2024 v Sarah L. Zapetis
Conclusions .......................................................................................................................32
References .....................................................................................................................................33
Supplementary Methods ...............................................................................................................42
Supplementary Figures .................................................................................................................46
Supplementary Table ....................................................................................................................48

Copyright 2024 vi Sarah L. Zapetis
List of Tables
Table 1. Group Differences in Demographic and Clinical Variables............................................11
Table 2. Group Differences in Ecological Momentary Assessment and Ambulatory
Psychophysiology Variables..........................................................................................................18
Table 3. Momentary Autonomic Complexity Predicting Subsequent Perseverative Cognition
(Within-Person)..............................................................................................................................19
Table 4. Person-Level Autonomic Complexity Predicting Perseverative Cognition (BetweenPerson)...........................................................................................................................................21
Supplementary Table S1. Inertia Time-Lag Analyses...................................................................48

Copyright 2024 vii Sarah L. Zapetis
List of Figures
Figure 1. Daily Ambulatory Assessment Procedure......................................................................14
Figure 2. Depression History as a Moderator of the Relationship between Within-Person
Fluctuations in the Inertia of Sample Entropy and Perseverative Cognition.................................20
Figure 3. Between-Person Association between the Mean of RMSSD and Perseverative
Cognition.......................................................................................................................................22
Figure 4. Depression History as a Moderator of the Relationship between the Inertia of Sample
Entropy and Perseverative Cognition at the Between-Person Level.............................................23
Supplementary Figure S1. Inertia of Autonomic Complexity Metrics..........................................46
Supplementary Figure S2. Sample Entropy Calculation from Heart Rate Interval Data..............47

Copyright 2024 viii Sarah L. Zapetis
Abbreviations
AC, autonomic complexity
CON, never-depressed control group
EMA, ecological momentary assessment
PC, perseverative cognition
rMDD, remitted major depressive disorder group
RMSSD, root mean square of successive differences between R-R intervals

Copyright 2024 ix Sarah L. Zapetis
Abstract
Background: Trait perseverative cognition (PC) is associated with inflexible autonomic activity
and risk for depressive recurrence. However, the identification of dynamic psychophysiological
markers of PC that fluctuate within individuals over time could facilitate the passive detection of
moments when PC occurs in daily life.
Methods: Using intensively sampled data across one week (3x/day) in adults with remitted
major depressive disorder (rMDD) and never-depressed controls, we investigated the utility of
monitoring ambulatory autonomic complexity to predict moments of PC engagement in
everyday life. Autonomic complexity metrics, including the root mean square of successive
differences (RMSSD), indexing vagal control, and sample entropy, indexing signal complexity,
were calculated in the 30 minutes before each measurement of PC to enable time-lagged
analyses. Multilevel models examined proximal fluctuations in the mean level and inertia of
complexity metrics as predictors of subsequent PC engagement.
Results: Momentary increases in the inertia of sample entropy, but not other metrics, predicted
higher levels of subsequent PC in the rMDD group, but not among never-depressed controls.
Conclusions: The inertia of sample entropy could index autonomic rigidity and serve as a
dynamic risk marker for real-world PC in individuals with a history of depression. This could
inform the development of technologies to passively detect fluctuations in risk for PC,
facilitating real-time interventions to prevent PC and reduce risk for depressive recurrence.
Keywords: ambulatory psychophysiology, autonomic complexity, ecological momentary
assessment, perseverative cognition, depression

Copyright 2024 1 Sarah L. Zapetis
Chapter 1: Introduction
Depression is a pervasive psychiatric disorder and a leading cause of disability
worldwide (Ferrari et al., 2013). The high rate of recurrence following remission from a
depressive episode is an important contributor to this public health challenge (Eaton et al.,
2008; Mueller et al., 1999). Accordingly, there is a pressing need for research to enhance our
understanding of factors that contribute to the onset and recurrence of depression. One key
risk factor is perseverative cognition (PC), characterized by the persistent and repetitive
activation of negative thoughts (Brosschot et al., 2006; Spinhoven et al., 2018). Engagement
in PC, which has been conceptualized as a way of responding to negative affect (NolenHoeksema, 1991; Papageorgiou & Wells, 1999), fluctuates throughout the day, reflecting the
dynamic interplay between emotions and cognitive processes. As such, just-in-time adaptive
interventions (JITAIs) that can identify and promptly intervene during moments of
heightened risk for PC may be particularly well-suited to target this cognitive risk factor.
Several interventions, like mindfulness-based cognitive behavioral therapy (CBT) and
rumination-focused CBT, have been developed to target PC and have shown some promise in
reducing rates of depressive recurrence (Piet & Hougaard, 2011; Watkins et al., 2011).
However, the real-world implementation of these intervention techniques is currently limited
by our inability to detect moments when PC occurs, precluding intervention deployment in
moments when it could be most beneficial. Moreover, identifying moments of heightened
risk for PC engagement in daily life could enable the implementation of intervention
strategies during critical moments to prevent PC. Utilized in combination with existing
interventions, this approach has the potential to enhance their efficacy. Thus, the goal of our

Copyright 2024 2 Sarah L. Zapetis
current study was to determine whether continuously monitoring psychophysiology in daily
life could facilitate the detection of moments of increased PC engagement.
PC can be divided into two key components: rumination, which involves engaging with
negative material from the past, and worry, which involves a focus on negative events that might
occur in the future (Brosschot et al., 2006; Ehring & Watkins, 2008; Nolen-Hoeksema, 1991;
Papageorgiou & Wells, 1999). Empirical evidence consistently indicates that trait rumination
predicts the onset, duration, and severity of major depressive episodes (Abela & Hankin, 2011;
Just & Alloy, 1997; Nolen-Hoeksema et al., 2008; Spasojević & Alloy, 2001; Stange et al.,
2016). In addition, recent research utilizing ecological momentary assessment (EMA) suggests
that ruminative responses to negative affect may be more automatic, and thus more difficult to
control, in individuals with a history of depression compared to never-depressed controls
(Hjartarson et al., 2022). This finding implies a habitual component of rumination that may
contribute to its persistence, thus prolonging or worsening negative affect (Moberly & Watkins,
2008) and increasing risk for depressive recurrence. Moreover, research indicates that worry
serves as an important risk factor for depression as well (Segerstrom et al., 2000; Starcevic,
1995). Specifically, trait levels of worry have been shown to predict the recurrence and
persistence of depressive disorders as well as the severity of depression symptoms (Spinhoven et
al., 2018). When examining specific features of trait worry, the lack of perceived control over
worrisome thoughts is uniquely related to depression above and beyond other features (Gorday et
al., 2018). Importantly, both rumination and worry are thought to prolong cognitive
representations of stressors, either retrospectively or in anticipation, along with stress-related
affect and physiological activation that can contribute to dysregulation and disease (Brosschot et
al., 2006). Furthermore, high trait PC has been linked to impairments in inhibitory control and

Copyright 2024 3 Sarah L. Zapetis
cognitive flexibility, reflected in difficulties shifting cognitive resources to changing situational
demands (Davis & Nolen-Hoeksema, 2000; Ottaviani et al., 2013), which may contribute to
challenges inhibiting or disengaging from these putatively maladaptive cognitive responses to
stress. Additionally, both components of PC have been conceptualized as serving an avoidance
function that interferes with emotional processing and hinders the deployment of adaptive
regulation strategies (Fresco et al., 2002; Mennin & Fresco, 2013; Stange et al., 2013; Stange,
Hamilton, et al., 2017b). Thus, despite differences in their temporal content, rumination and
worry are both types of PC that are difficult to control and are highly correlated with one another
(Fresco et al., 2002; Segerstrom et al., 2000; Watkins et al., 2005).
Whereas a substantial body of literature has examined trait-like PC, typically assessed
with one-time, self-report measures in which participants rate their usual engagement in PC
(Meyer et al., 1990; Treynor et al., 2003), limited research has used intensive sampling methods
to examine how PC fluctuates within individuals over time. Given the dynamic nature of
emotional experiences and cognitive responses to emotions in our daily lives, as well as the
susceptibility of retrospective measures to recall biases (Bradburn et al., 1987; Clark et al.,
1982), research designs that utilize in-the-moment data to capture within-person variability are
needed to aid in the development of more ecologically valid models of affect dysregulation
(Stange et al., 2019). Ambulatory assessment technologies offer a way to intensively sample an
individual’s momentary state within their everyday life to investigate fluctuations in constructs
of interest. Whereas measures like EMA require persistent effort from participants, ambulatory
psychophysiological measures can be passively and unobtrusively collected from wearable
devices, and therefore may be more feasible for long-term risk monitoring. To the extent that
people’s physiology changes in the moments before they engage in PC, ambulatory

Copyright 2024 4 Sarah L. Zapetis
psychophysiology has the potential to allow for the identification of moments when individuals
are at acute risk for PC in everyday life. In turn, this detection could facilitate the implementation
of in-the-moment intervention strategies to prevent real-world PC engagement.
Autonomic complexity is a promising psychophysiological construct that could serve
as a proximal marker for moments of risk for real-world PC. Autonomic complexity can be
derived from measures of variation in the time between consecutive heartbeats (e.g., heart
rate variability), and may reflect the capacity of the autonomic nervous system to respond
flexibly to changing demands (Stange et al., 2023). Neurovisceral integration theory posits
that, during rest, the prefrontal cortex exerts parasympathetic inhibitory control over thalamic
regions associated with sympathetic defensive responses via the vagus nerve (Smith et al.,
2017; Thayer et al., 2012; Thayer & Lane, 2002, 2009). This enables the autonomic nervous
system to be responsive to input from a variety of internal and external stimuli, which
manifests as slowed heart rate and greater variability in the intervals between heartbeats
(greater autonomic complexity). Conversely, PC may correspond to a failure of these
inhibitory processes resulting in prolonged physiological arousal, as evidenced by lower
resting autonomic complexity in individuals with high trait PC (Carnevali et al., 2018;
Ottaviani et al., 2016).
Having higher levels of autonomic complexity at rest is thought to signify the integration
of neural inhibitory circuits that support adaptive regulation of affect and physiology. In contrast,
reductions in autonomic complexity during stress or emotional challenges may represent a
contextually appropriate withdrawal of this inhibitory control that enables activation of
sympathetic defensive responses. The ability of the autonomic nervous system to flexibly
modulate responses to changing situational demands – as reflected in fluctuations in autonomic

Copyright 2024 5 Sarah L. Zapetis
complexity across contexts – is critical for effective emotion regulation (Balzarotti et al., 2017;
Stange, Alloy, et al., 2017; Thayer & Lane, 2009; Yaroslavsky et al., 2016). For example, in a
prior investigation, autonomic inflexibility in response to a laboratory sadness induction was
associated with higher levels of trait PC, as well as greater engagement in real-world PC
regardless of stress exposure (Stange et al., 2020). As such, inflexible autonomic activity may
reflect vulnerability to context-insensitive engagement in PC and could serve as a marker of
individual differences in risk for depression (Stange, Hamilton, et al., 2017b). However, at the
within-person level, little research has examined whether moments of risk for PC in daily life are
characterized by short-term decreases in autonomic flexibility. Moreover, recent evidence has
highlighted the importance of investigating intra-individual variability in autonomic complexity
metrics over time in real-world contexts (Gruber et al., 2015; Stange et al., 2023). As such, a key
aim of this study was to investigate fluctuations not only in the level of complexity but also the
inertia – a novel dynamic metric of autonomic sluggishness – as predictors of proximal
fluctuations in real-world PC.
Inertia represents the temporal dependency of a signal or how strongly two
consecutive measurements are associated with each other (Wenzel et al., 2023; Suls et al.,
1998). Our novel application of inertia to autonomic complexity builds upon an existing
literature on emotional inertia, which indicates that the inertia of negative emotions is
elevated in depression (Houben et al., 2015; Kuppens et al., 2010) and is positively
associated with rumination across individuals (Koval et al., 2012), suggesting that negative
emotions are more resistant to change in depression. At the physiological level, a recent
ambulatory study found that across individuals, reduced variability of autonomic complexity
across a week of measurement was associated with higher levels of negative affect and

Copyright 2024 6 Sarah L. Zapetis
ruminative brooding (Stange et al., 2023). However, to our knowledge, the inertia of
autonomic complexity metrics as a momentary indicator of proximal fluctuations in PC
within individuals has yet to be investigated. This investigation would more precisely address
our clinical interest in detecting moments when PC is likely to occur. In our study, greater
inertia would reflect more similarity or sluggishness in autonomic complexity metrics within
the intervals just prior to the measurement of PC in everyday life (See Supplementary
Methods and Supplementary Figure S1).
A commonly used index of autonomic complexity is the root mean square of successive
differences (RMSSD) between heartbeats. RMSSD is a linear time-domain measure that
approximates vagally mediated influences on the heart (Shaffer & Ginsberg, 2017). Therefore,
higher resting RMSSD reflects greater integration of parasympathetic inhibitory circuits in the
regulation of heart rate. Whereas linear measures like RMSSD are often used to estimate
physiological processes, the autonomic nervous system is influenced by a variety of regulatory
mechanisms, and thus its output is nonlinear in nature (Young & Benton, 2015). As such,
nonlinear measures like sample entropy may be better suited to capture autonomic complexity,
and may reflect shifts in the balance between the sympathetic and parasympathetic branches of
the autonomic nervous system. Sample entropy is a computation of signal irregularity that
captures the degree to which short-term patterns in a signal are predictable over time (See
Supplementary Methods and Supplementary Figure S2). While low levels of sample entropy
indicate that a system is more repetitive, showing consistent patterns across time, higher levels
may indicate that a system has more sources of input and is more flexible to changing demands
(Bakhchina et al., 2018).

Copyright 2024 7 Sarah L. Zapetis
While much of the extant literature has related autonomic complexity metrics and PC at
the between-person level of analysis, this approach has precluded our understanding of whether
state fluctuations in complexity have utility as a marker of moments when PC will increase. The
present study is the first to examine the within-person variation in autonomic complexity that
occurs in real-world moments just prior to PC engagement, using novel metrics of autonomic
complexity in everyday life. Specifically, this study investigated within-person fluctuations in
both the level and inertia of RMSSD and sample entropy as predictors of fluctuations in PC.
Furthermore, the time-lagged design of the current study allowed us to examine the ability of
fluctuations in autonomic complexity metrics to preemptively detect increases in PC. This will
provide insight into whether continuously monitoring psychophysiology in daily life could allow
for the identification of moments of increased likelihood of PC engagement. If so, these results
could enhance our understanding of the mechanisms of PC engagement in everyday life by
suggesting that certain autonomic states may confer vulnerability for PC. In addition, these
findings could inform the development of technologies, such as JITAIs, in that passive detection
of imminent fluctuations in autonomic complexity could potentially enable real-time
interventions to prevent PC and reduce risk for depressive recurrence.
Given trait-level associations between PC and reduced autonomic flexibility in both
laboratory paradigms (Stange et al., 2020) and daily life (Stange et al., 2023), we hypothesized
that periods of decreased level and periods of increased inertia of autonomic complexity metrics
would predict subsequent increases in EMA-reported PC. In our sample of adults with a history
of remitted major depressive disorder (rMDD) and never-depressed control individuals (CON),
we also examined depression history as a moderator of the within-person relationships between
fluctuations in autonomic complexity and PC hypothesizing that, relative to CONs, individuals

Copyright 2024 8 Sarah L. Zapetis
with rMDD would show a stronger relationship between maladaptive autonomic complexity
metrics (i.e., decreased mean and increased inertia) and subsequent engagement in PC. This
hypothesis was based on research showing that individuals with depression exhibit habitual
engagement in PC, characterized by a higher degree of automaticity in response to negative
affect (Hjartarson et al., 2022), as well as impairments in cognitive flexibility that may influence
the ability to inhibit or disengage from maladaptive cognitive responses to physiological arousal
or rigidity in everyday life. Finally, although our primary aims were to investigate within-person
associations between complexity and PC, we also explored whether participants’ mean level of
complexity metrics across the study period would predict their overall degree of engagement in
PC at the between-person level, and whether depression history would moderate this
relationship. We hypothesized that individuals with lower levels and higher inertia of autonomic
complexity metrics would engage in more PC, and that these associations would be stronger for
individuals with a history of depression compared to never-depressed controls.

Copyright 2024 9 Sarah L. Zapetis
Chapter 2: Materials and Methods
Participants
Study procedures were approved by the University of Illinois at Chicago (UIC) and
the University of Southern California (USC) Institutional Review Boards. All research
protocols were carried out in accordance with the provisions of the World Medical
Association Declaration of Helsinki. This study involved data collected from 36 young adults
with rMDD and 30 never-depressed CONs. The inclusion of individuals with rMDD allowed
for the investigation of relationships of interest in a population that is at an increased risk for
depressive recurrence while minimizing the influence of factors associated with the state of
current depression (e.g., symptoms). In addition, it allowed for the examination of how
depression history influences relationships between autonomic complexity metrics and PC.
Participants were 18-30 years old which served to minimize the cumulative effects of
recurrent depressive episodes, which would presumably be more prevalent in older samples.
Participants were 36.4% male and 63.6% female, with a mean age of 26.7 years (SD = 3.8).
Participants were recruited from the surrounding community using online advertisements. All
participants were required to have fluency in English and normal or corrected-to-normal
vision and were excluded if they had known cardiac arrhythmias. After providing informed
consent, participants were screened with the Diagnostic Interview for Genetic Studies
(Nurnberger et al., 1994). Participants in the rMDD group met DSM-5 criteria for lifetime
MDD (American Psychiatric Association, 2013), met criteria for full remission for at least 8
weeks, and scored a 7 or below on the 17-item Hamilton Depression Rating Scale (HDRS;
Hamilton, 1960). CON participants did not meet current or past criteria for MDD or any
other psychiatric disorder and had no first-degree family members with known psychiatric

Copyright 2024 10 Sarah L. Zapetis
disorders. Groups did not differ on demographic characteristics, although as expected,
participants in the rMDD group had higher rates of some comorbidities and psychotropic
medication use (Table 1).
Procedures
Ecological Momentary Assessment (EMA)
Following the screening visit, eligible participants were trained on the EMA
procedures. Specifically, a study staff member reviewed the EMA schedule, each EMA item,
and its associated response options with participants to ensure sufficient understanding.
Then, over the course of seven days, EMA survey links were sent to participants via text or
email six times a day. Surveys were presented and collected in REDCap (Harris et al., 2009).
Participants chose to receive either an “early” schedule (i.e., surveys were sent between the
hours of 8:00 am and 7:00 pm) or a “late” schedule that was offset by two hours (i.e., surveys
sent between 10:00 am and 9:00 pm) depending on their schedule preference. Within the
“early” and “late” groups, the EMA surveys were sent in a fixed schedule across participants.
To facilitate the appearance of a semi-random schedule, participants were not told of the
specific survey schedule beyond that they would receive surveys in the morning, afternoon,
and evening. The six surveys were sent in pairs with three “pre” and three “post” surveys per
day. For each pair, the post-survey was sent 30 minutes after completion of the pre-survey. If
the pre-survey was not completed, the participant was not sent a post-survey for that time
point. Participants were given up to one hour to complete each survey and were reminded
every 20 minutes until each survey was completed, resulting in a 30- to 90-minute period
between pre- and post-surveys. The mean between-person duration of time between pre- and

Copyright 2024 11 Sarah L. Zapetis
post-surveys was 46.5 min
(SD = 7.1 min), indicating
that on average people
completed the post-survey
within 15 minutes of
receiving the initial
prompt.
In all surveys,
participants answered
prompts about their affect
based on how they felt just
before the survey. In each
post-survey, participants
also reported their
engagement in ruminative
brooding and worry “since
the pre-survey about 30 -
90 minutes ago” on a scale
from 1 (not at all) to 10
(very much) using items
adapted for EMA from
the Spontaneous Affect
Regulation Scale (SARS;
Table 1. Group Differences in Demographic and Clinical
Variables. Educational attainment is reported in years where 12
years represents a high school diploma, 16 years represents a
bachelor’s degree, 18 years represents a master’s degree, and 20
years represents a doctoral degree. Psychotropic medication load
was calculated for each person following a coding system
(Phillips et al., 2008) which assigns a score for each medication
based on its dosage and frequency and sums scores for a total
medication load. * indicates significance (p < .05).
Measure CON rMDD
Demographics
Age -- Mean (SD) 26.2 (3.8) 27.1 (3.8)
BMI -- Mean (SD) 24.0 (3.3) 25.5 (5.4)
Gender Identity
Male 36.7% 36.1%
Female 63.3% 63.9%
Nonbinary 0.0% 0.0%
Race
American Indian or Alaska Native 0.0% 0.0%
Asian 26.7% 8.3%
White 50.0% 63.9%
Black or African American 20.0% 22.2%
Native Hawaiian or Other Pacific Islander 0.0% 2.8%
Other 0.0% 2.8%
Hispanic or Latino 16.7% 8.3%
Educational Attainment -- Mean (SD) 16.6 (2.3) 16.2 (1.9)
Clinical Features
Lifetime Comorbid Diagnoses
Dysthymia* 0.0% 19.4%
Panic Disorder 0.0% 11.1%
Agoraphobia 0.0% 2.8%
Specific Phobia 0.0% 5.6%
Social Phobia* 0.0% 25.0%
Generalized Anxiety Disorder 0.0% 2.8%
Obsessive-Compulsive Disorder 0.0% 2.8%
Post-Traumatic Stress Disorder 0.0% 11.1%
Other Specified/Unspecified Anxiety Disorder 0.0% 2.8%
Binge Eating Disorder 0.0% 8.3%
Substance Use Disorder (Alcohol) 0.0% 11.1%
Substance Use Disorder (Cannabis) 0.0% 2.8%
Attention-Deficit Hyperactivity Disorder 0.0% 5.6%

Current Psychotropic Medications
Antidepressant* 0.0% 27.8%
Mood Stabilizer 0.0% 2.8%
Antipsychotic 0.0% 2.8%
Anxiolytic 0.0% 5.6%
Benzodiazepine 0.0% 2.8%
Stimulant 3.3% 0.0%
Opioid 3.3% 0.0%
Total Medication Load* -- Mean (SD) 0.08 (0.37) 0.42 (0.74)

Copyright 2024 12 Sarah L. Zapetis
Egloff et al., 2006; Gruber et al., 2012; Stange et al., 2023; Stange, Hamilton, et al., 2017a).
Specifically, ruminative brooding was measured with the item “I thought about a recent
situation, wishing it had gone better” and worry was measured with the item “I worried about
something.” Responses to ruminative brooding and worry items were averaged at each time
point to provide a composite PC score. This PC score, while novel, demonstrated moderate
reliability in our sample (ωbetween = .701, ωwithin = .613; Lai, 2021).
Ambulatory Psychophysiology
During the seven-day EMA period, participants wore a lightweight biometric Hexoskin smart
shirt (Carré Technologies, Inc.) under their usual clothing during waking hours to allow for
the continuous collection of electrocardiogram data in their daily lives. The shirts contain a
single-lead ECG with three electrode sensors sampled at 256 Hz. The shirts also continuously
collected 3-axis accelerometer data sampled at 64 Hz which was used to control for activity
level in models using autonomic complexity metrics as predictors.
Sequential interbeat intervals were extracted from the Hexoskin shirts and entered
into HRVanalysis software (Pichot et al., 2016). Signals from each recording were visually
inspected and quality checked, such that any gaps of visibly noisy portions at the beginning
or end of the recording were removed before processing the signal. For segments of missing
or exceptionally messy signals that were noticeable, the R-R series was split into sections and
analyzed individually. The software detected and corrected anomalies due to heartbeat
rhythm disturbances such as ectopic beats.
The data were then analyzed in 5-minute epochs (Malik et al., 1996), which are
thought to reflect short-term cardiac dynamics (Shaffer & Ginsberg, 2017) and may be better
suited to detect periods of proximal risk for PC compared to longer-term measurements.

Copyright 2024 13 Sarah L. Zapetis
Epochs were excluded if they contained excessive noise or missing data, as detected by the
following criteria: (1) the percentage of artifacts in the R-R time series exceeded 10%; (2)
more than five seconds of data were missing from the beginning or end of the 5-min
segment; or (3) more than one-third of data were missing from the middle of the 5-min
segment (Stange et al., 2023). The first autonomic complexity metric, RMSSD, was obtained
by first calculating the successive time difference between each pair of heartbeats, squaring
each value, averaging them, and then squaring the result (Shaffer & Ginsberg, 2017). The
second metric calculated was sample entropy, which is defined as the logarithmic likelihood
that two sequences of RR intervals that are similar for m time points differ at time m+1
(Lanata et al., 2015; Richman & Moorman, 2000). Two sequences are determined to be
matched if their distance is within a certain threshold r. In our study, m was set to 2 and r to
0.2 times the standard deviation of the time series, which are standards commonly reported in
the literature (Fiskum et al., 2018). Time stamps were used to match the autonomic
complexity epoch values with the corresponding EMA survey data. For the analysis of
psychophysiological data, data in the 30 minutes before each EMA period was utilized to
enable time-lagged analysis (Figure 1). For each 30-minute autonomic complexity period,
the mean and inertia of RMSSD and sample entropy were calculated using the six 5-minute
epochs within that period. Inertia was defined as the Pearson autocorrelation between the six
consecutive epochs within a given autonomic complexity period and their lagged
counterparts (See Supplementary Methods and Supplementary Figure S1).
Physical activity levels served as a covariate in analyses that included autonomic
complexity metrics as a predictor. This was done to account for metabolically driven changes
in autonomic complexity (Verkuil et al., 2016), since psychologically driven changes were of

Copyright 2024 14 Sarah L. Zapetis
interest in this study. Activity levels were calculated as the vector of acceleration (square root
of the sum of squares of each of the three accelerometer axes) per second and averaged
across each 5-minute epoch. Then, the mean level of activity across each 30-minute
autonomic complexity period was calculated.
Statistical Analysis
All analyses were conducted in Mplus version 7 (Muthén & Muthén, 2010) and
utilized full information maximum likelihood (FIML) estimation to account for missing data.
Sensitivity analyses were conducted to examine the effects of relevant covariates including
age, gender, and body mass index (BMI) on the relationships of interest. In within-person
analyses, including those with group as a moderator, additional sensitivity analyses were
Figure 1. Daily Ambulatory Assessment Procedure. EMA post-surveys captured
engagement in PC “since the pre-survey about 30–90 min ago.” ECG data in the 30 min
before each EMA period was utilized to enable a time-lagged design. The RMSSD and
sample entropy of each 5-min epoch were calculated. Then, the mean level and inertia
across the six consecutive 5-min epochs in each autonomic complexity period were
calculated.
5 min
Autonomic
Complexity
Period
30 min
5 min 5 min 5 min 5 min 5 min
EMA
EMA
period
AC
period
Morning Afternoon Evening
30 min 30 min
EMA
EMA
period
AC
period
30 min 30 min
EMA
EMA
period
AC
period
30 min 30 min

Copyright 2024 15 Sarah L. Zapetis
conducted to examine the effect of including average activity level and the average of the
corresponding autonomic complexity metric across the study period.
Group Comparisons
Groups (rMDD and CON) were compared to determine differences in repeatedmeasures ambulatory variables (autonomic complexity metrics and PC). Multilevel models
included group as a between-person (Level 2) predictor of random intercepts (average levels)
of autonomic complexity metrics and PC across observation windows.
Within-Person Analyses
Multilevel models were estimated to examine relationships between within-person
fluctuations in autonomic complexity metrics (in the 30 minutes prior to the EMA period)
and the degree of engagement in PC during the subsequent 30-minute EMA period.
Autonomic complexity metrics and activity level were centered around each person’s mean.
Separate models were conducted for each of the four autonomic complexity metrics (mean of
RMSSD, inertia of RMSSD, mean of sample entropy, inertia of sample entropy). If models
including the inertia of autonomic complexity metrics indicated that inertia significantly
predicted PC, sensitivity analyses were then conducted that included both the inertia and
mean level of the respective autonomic complexity metric to determine the incremental effect
of inertia above and beyond the effect of mean level.
To investigate how depression history impacted the strength of the within-person
relationships between autonomic complexity metrics and subsequent PC, group (rMDD or
CON) was examined as a between-person (Level 2) moderator of the within-person
relationships between fluctuations in autonomic complexity metrics and subsequent PC,
while also including the main effect of group as a predictor of PC. Significant interactions

Copyright 2024 16 Sarah L. Zapetis
between group and autonomic complexity metrics were probed by testing simple slopes in
each group (Aiken & West, 1991).
Between-Person Analyses
To determine whether participants with lower (and more inert) autonomic complexity
would report more PC across the week of measurement, each autonomic complexity metric
and physical activity level were averaged across the week of observation for each participant
(approximately 21 time points, linked to the 30-min windows before EMA periods),
providing a single value for each person. These autonomic complexity metrics and activity
level were then z-scored across participants to aid in interpretation and entered as Level 2
predictors of intercepts of PC. To investigate how depression history impacted the strength of
the between-person relationships between autonomic complexity metrics and PC, group
(rMDD or CON) was examined as a moderator.
Comparisons of Inertia Metrics with Different Time-Lags
Our primary analysis plan was to calculate the inertia of autonomic complexity
metrics in successive (non-overlapping) 5-min epochs, as is standard when calculating
emotional inertia using EMA data (Suls et al., 1998). However, we also conducted sensitivity
analyses to determine how using overlapping epochs, with varying degrees of overlap (e.g.,
50% overlap with prior 5-min epoch), would alter the autocorrelations and the predictive
ability of the inertia variables. This analytical approach is described in the Supplementary
Methods, and the results of these analyses are shown in Supplementary Table S1.

Copyright 2024 17 Sarah L. Zapetis
Chapter 3: Results
Descriptive Statistics about Ambulatory Data
A total of 1,268 pre/post-survey pairs were completed (M = 19.21 per participant, SD
= 3.06, min = 10, max = 28). On average, participants completed 87.95% of possible “pre”
survey prompts (SD = 14.39%, min = 25%, max = 100%). Among the instances where presurveys were completed, 91.75% of possible “post” surveys (SD = 9.47%, min = 47.62%,
max = 100%) were completed. Of the post-surveys completed, 26.97% were missing the
corresponding psychophysiological data (i.e., the mean of autonomic complexity metrics
could not be calculated) due to no data being recorded (e.g., the shirt was not worn, or device
was not charged). For an additional 5.21% of completed post-surveys, the inertia of
autonomic complexity metrics could not be calculated due to an insufficient amount of data
(i.e., less than three consecutive epochs) within the autonomic complexity period. As a result,
a total of 32.18% of completed post-surveys were missing the inertia of autonomic
complexity metrics. Intra-class correlation coefficients (ICCs; Table 2) indicate the presence
of both within and between-person variability in EMA and mean level autonomic complexity
variables. However, the ICCs of the inertia of autonomic complexity variables were notably
low, indicating that a majority of the variance in these variables exists within individuals and
highlighting the importance of measuring these variables repeatedly over time.
Group Comparisons
Individuals with rMDD had higher PC than never-depressed controls (B = 0.706, SE=
0.296, 95% CI = [0.127,1.286], p = 0.017). However, the groups did not significantly differ
on any of the autonomic complexity metrics (Table 2).

Copyright 2024 18 Sarah L. Zapetis
Within-Person Associations between Autonomic Complexity and PC
Moments when people had higher entropy inertia (inertia of sample entropy) than
usual were associated with higher levels of PC in the next 30 minutes (Table 3). Conversely,
fluctuations in (1) the level of sample entropy, (2) the level of RMSSD, and (3) the inertia of
RMSSD did not predict subsequent levels of PC (Table 3). These results did not differ when
depression history and relevant covariates (age, gender identity, body mass index, average
activity level, average corresponding autonomic complexity metric) were included in the
models. In addition, although we did not control for PC at the previous time point in our
main models (Falkenström et al., 2017), the results were also consistent when we did include
this variable as a covariate (results available upon request). When both the mean and inertia
of sample entropy were included as independent variables in the same model, the inertia, but
CON rMDD Effect Size
Mean (SD) Mean (SD) Cohen's d
Ecological Momentary Assessment
Perseverative Cognition* 2.48 (1.20) 3.19 (1.25) -0.58 .314
Ambulatory Psychophysiology
Mean
RMSSD 41.99 (14.69) 36.95 (17.16) 0.31 .520
Sample Entropy 1.19 (0.15) 1.14 (0.15) 0.39 .179
Heart Rate* 79.42 (9.40) 84.75 (9.59) 0.56 .431
Activity 0.04 ( 0.02) 0.04 (0.02) 0.09 .148
Inertia
RMSSD 0.03 (0.16) 0.03 (0.18) 0.03 .014
Sample Entropy -0.11 (0.15) -0.09 (0.15) -0.10 .025
Measure ICC
Table 2. Group Differences in Ecological Momentary Assessment and Ambulatory
Psychophysiology Variables. ICC stand for intraclass correlation coefficient and indicates the
proportion of variance in within-person variables that occurred at the between-person level. *
indicates significance (p < .05).

Copyright 2024 19 Sarah L. Zapetis
not the mean, of sample entropy significantly predicted subsequent PC, and this effect
remained significant when covariates were included.
In the moderation analyses, we found that the within-person relationship between
entropy inertia and PC differed significantly based on depression history (Table 3), with
entropy inertia predicting subsequent PC in individuals with remitted depression (B = 0.576,
Table 3. Momentary Autonomic Complexity Predicting Subsequent Perseverative
Cognition (Within-Person). In these models, rMDD represents a group membership variable,
where never-depressed controls were coded as 0 and individuals with remitted major depressive
disorders were coded as 1. R2 for the full models represents the pseudo R2
, or the proportion of
variance explained by the model relative to an unrestricted (intercept-only) model containing no
predictors, separately at the within- and between-person levels (Rights & Sterba, 2020). R2 for
each predictor represents the ΔR2 of given predictor when it is excluded from the full model.
When interactions amongst variables were included in the full model (i.e., Sample Entropy
(Inertia) x rMDD), the computation of ΔR2 for each individual predictor (i.e., Sample Entropy
(Inertia) and rMDD) involved comparing a model excluding both the interaction and the
individual predictor to a model with the individual predictor added. Bolded lines indicate
significant main effects in dimensional models and significant interactions in models including
group as a moderator (p < .05).
Model Focal Predictor B SE 95% CI p _within ___between Model Focal Predictor B SE 95% CI p _within ___between
RMSSD (Mean) 0.004 0.001 RMSSD (Mean) 0.010 0.094
Perseveration (Intercept) 2.868 0.154 2.567, 3.169 < 0.001 Perseveration (Intercept) 2.479 0.200 2.087, 2.870 < 0.001
Fixed Effects Fixed Effects
Within-Subject (Level 1) Within-Subject (Level 1)
RMSSD (Mean) -0.006 0.004 -0.015, 0.002 0.143 0.003 < 0.001 Activity (Mean) -2.402 1.824 -5.976, 1.173 0.188 0.002 0.002
Activity (Mean) -2.490 1.815 -6.047, 1.068 0.170 0.004 0.001 RMSSD (Mean) -0.010 0.006 -0.021, 0.002 0.111 0.003 < 0.001
Random Effects Between-Subject (Level 2)
Perseveration (Random Intercept) 1.397 0.241 0.923, 1.870 < 0.001 rMDD 0.713 0.284 0.157, 1.270 0.012 0.002 0.086
RMSSD (Mean) x rMDD 0.006 0.008 -0.010, 0.002 0.462 < 0.001 < 0.001
Random Effects
Perseveration (Random Intercept) 1.267 0.213 0.849, 1.685 < 0.001
RMSSD (Mean) (Random Slope) < .001 0.006 -0.013, 0.013 0.987 0.009 0.005
RMSSD (Inertia) 0.001 0.002 RMSSD (Inertia) 0.004 0.094
Perseveration (Intercept) 2.868 0.154 2.566, 3.169 < 0.001 Perseveration (Intercept) 2.465 0.214 1.958, 2.972 < 0.001
Fixed Effects Fixed Effects
Within-Subject (Level 1) Within-Subject (Level 1)
RMSSD (Inertia) -0.009 0.103 -0.210, 0.193 0.932 < 0.001 < 0.001 Activity (Mean) -1.560 1.625 -4.745, 1.626 0.337 < 0.001 0.006
Activity (Mean) -1.568 1.600 -4.704, 1.569 0.327 0.001 0.001 RMSSD (Inertia) 0.147 0.166 -0.179, 0.473 0.376 < 0.001 < 0.001
Random Effects Between-Subject (Level 2)
Perseveration (Random Intercept) 1.396 0.241 0.923, 1.869 < 0.001 rMDD 0.729 0.328 0.087, 1.371 0.026 < 0.001 0.084
RMSSD (Inertia) x rMDD -0.259 0.210 -0.670, 0.153 0.218 0.002 < 0.001
Random Effects
Perseveration (Random Intercept) 1.267 0.214 0.847, 1.686 < 0.001
RMSSD (Inertia) (Random Slope) 0.017 0.004 0.009, 0.024 < 0.001 < 0.001 0.005
Sample Entropy (Mean) 0.005 0.001 Sample Entropy (Mean) 0.016 0.096
Perseveration (Intercept) 2.868 0.154 2.567, 3.170 < 0.001 Perseveration (Intercept) 2.511 0.207 2.105, 2.916 < 0.001
Fixed Effects Fixed Effects
Within-Subject (Level 1) Within-Subject (Level 1)
Sample Entropy (Mean) -0.374 0.317 -0.997, 0.248 0.238 0.004 < 0.001 Activity (Mean) -2.797 2.173 -7.056, 1.462 0.198 0.004 0.003
Activity (Mean) -2.929 2.216 -7.273, 1.415 0.186 0.005 0.001 Sample Entropy (Mean) -0.556 0.417 -1.373, 0.262 0.183 0.004 < 0.001
Random Effects Between-Subject (Level 2)
Perseveration (Random Intercept) 1.397 0.242 0.923, 1.871 < 0.001 rMDD 0.679 0.288 0.114, 1.244 0.019 < 0.001 0.082
Sample Entropy (Mean) x rMDD 0.389 0.436 -0.465, 1.243 0.372 < 0.001 < 0.001
Random Effects
Perseveration (Random Intercept) 1.264 0.212 0.849, 1.679 < 0.001
Sample Entropy (Mean) (Random Slope) 0.420 0.432 -0.428, 1.267 0.332 0.014 0.007
Sample Entropy (Inertia) 0.012 <0.001 Sample Entropy (Inertia) 0.075 0.097
Perseveration (Intercept) 2.867 0.154 2.566, 3.169 < 0.001 Perseveration (Intercept) 2.643 0.210 2.231, 3.055 < 0.001
Fixed Effects Fixed Effects
Within-Subject (Level 1) Within-Subject (Level 1)
Sample Entropy (Inertia) 0.366 0.146 0.080, 0.653 0.012 0.010 < 0.001 Activity (Mean) -1.757 1.549 -4.793, 1.279 0.257 0.015 0.007
Activity (Mean) -1.561 1.583 -4.665, 1.542 0.324 0.001 0.001 Sample Entropy (Inertia) -0.008 0.231 -0.460, 0.443 0.971 0.011 < 0.001
Random Effects Between-Subject (Level 2)
Perseveration (Random Intercept) 1.399 0.242 0.925, 1.873 < 0.001 rMDD 0.547 0.282 -0.006, 1.100 0.053 < 0.001 0.067
Sample Entropy (Inertia) x rMDD 0.584 0.286 0.025, 1.144 0.041 0.002 0.001
Random Effects
Perseveration (Random Intercept) 1.263 0.205 0.862, 1.664 < 0.001
Sample Entropy (Inertia) (Random Slope) 0.554 0.166 0.228, 0.879 0.001 0.072 0.006
Predicting Perseverative Cognition
Dimensional Models Moderation by Group Models
R2 R2 R2 R2

Copyright 2024 20 Sarah L. Zapetis
SE= 0.167, 95% CI = [0.249, 0.902], p = 0.001, ΔR² within of the random slope of entropy
inertia = 0.060, ΔR² between of the random slope of entropy inertia = .029), but not among
never-depressed controls (B = -0.008, SE= 0.231, 95% CI = [-0.460, 0.443], p = 0.971;
Figure 2)
1
. The relationships between PC and (1) the mean of RMSSD, (2) the inertia of
RMSSD, and (3) the mean of sample entropy did not differ based on depression history.
1 This moderation was attenuated when age (B = 0.558, SE= 0.285, 95% CI = [0, 1.117], p = 0.050), gender (B = 0.565, SE=
0.294, 95% CI = [-0.011, 1.141], p = 0.055), and BMI (B = 0.557, SE= 0.286, 95% CI = [-0.003, 1.118], p = 0.051) were
included as covariates in the model. However, the simple slopes of the rMDD group were still significantly positive (age: B =
0.560, SE= 0.160, 95% CI = [0.246, 0.874], p < 0.001; gender identity: B = 0.571, SE= 0.170, 95% CI = [0.238, 0.903], p =
0.001; BMI: B = 0.574, SE= 0.164, 95% CI = [0.252, 0.895], p < 0.001), whereas slopes of the CON group were non-significant
(age: B = -0.003, SE= 0.228, 95% CI = [-0.451, 0.445], p = 0.990; gender identity: B = 0.001, SE= 0.234, 95% CI = [-0.457,
0.459], p = 0.997; BMI: B = -0.003, SE= 0.228, 95% CI = [-0.449, 0.444], p = 0.991) when these covariates were included.
Figure 2. Depression History as a Moderator of the Relationship between WithinPerson Fluctuations in the Inertia of Sample Entropy and Perseverative Cognition.
Group membership (rMDD or CON) moderated the relationship between the inertia of
sample entropy and perseverative cognition at the within-person level such that
individuals with rMDD exhibited a significantly positive association and CON
individuals exhibited a nonsignificant association.
2
2.5
3
3.5
4
Low High
Perseverative Cognition
Inertia of Sample Entropy (Within-Person Fluctuation)
CON rMDD

Copyright 2024 21 Sarah L. Zapetis
Between-Person Associations between Autonomic Complexity and PC
At the between-person level, the mean of RMSSD was negatively associated with PC
(ΔR² = 0.081, Table 4 and Figure 3) such that individuals with a lower mean RMSSD
engaged in more PC overall. This association remained significant when covariates were
included in the model. Conversely, (1) the inertia of RMSSD and (2) the mean and (3) inertia
of sample entropy were not dimensionally associated with PC at the between-person level.
Table 4. Person-Level Autonomic Complexity Predicting Perseverative Cognition
(Between-Person). In these models, rMDD represents a group membership variable, where
never-depressed controls were coded as 0 and individuals with remitted major depressive
disorders were coded as 1. R2 for the full models represents the pseudo R2
, or the proportion
of variance explained at the between-person level by the model relative to an unrestricted
(intercept-only) model containing no predictors (Rights & Sterba, 2020). R2 for each predictor
represents the ΔR2 of given predictor when it is excluded from the full model. When
interactions amongst variables were included in the full model (i.e., Sample Entropy (Inertia)
x rMDD), the computation of ΔR2 for each individual predictor (i.e., Sample Entropy (Inertia)
and rMDD) involved comparing a model excluding both the interaction and the individual
predictor to a model with the individual predictor added. Bolded lines indicate significant
main effects in dimensional models and significant interactions in models including group as
a moderator (p < .05).
Model Focal Predictor B SE 95% CI p R2 between Model Focal Predictor B SE 95% CI p R2 between
RMSSD (Mean) 0.103 RMSSD (Mean) 0.172
Perseveration (Intercept) 2.853 0.144 2.582, 3.111 < 0.001 Perseveration (Intercept) 2.504 0.222 2.068, 2.940 < 0.001
Fixed Effects Fixed Effects
Between-Subject (Level 2) Between-Subject (Level 2)
RMSSD (Mean) -0.351 0.150 -0.039, -0.003 0.019 0.081 RMSSD (Mean) -0.208 0.242 -0.683, 0.267 0.390 0.056
Activity (Mean) 0.106 0.142 -7.620, 17.107 0.452 0.009 Activity (Mean) 0.135 0.143 -0.145, 0.414 0.346 0.013
Random Effects rMDD 0.626 0.285 0.066, 1.185 0.028 0.066
Perseveration (Random Intercept) 1.255 0.208 0.847, 1.663 < 0.001 RMSSD (Mean) x rMDD -0.141 0.295 -0.719, 0.437 0.633 0.003
Random Effects
Perseveration (Random Intercept) 1.159 0.189 0.788, 1.529 < 0.001
RMSSD (Inertia) 0.023 RMSSD (Inertia) 0.115
Perseveration (Intercept) 2.868 0.153 2.568, 3.167 < 0.001 Perseveration (Intercept) 2.474 0.213 2.057, 2.892 < 0.001
Fixed Effects Fixed Effects
Between-Subject (Level 2) Between-Subject (Level 2)
RMSSD (Inertia) 0.032 0.175 -0.311, 0.374 0.856 0.001 RMSSD (Inertia) -0.020 0.267 -0.544, 0.503 0.939 <0.001
Activity (Mean) 0.173 0.135 -0.092, 0.437 0.200 0.023 Activity (Mean) 0.192 0.138 -0.079, 0.463 0.164 0.026
Random Effects rMDD 0.724 0.293 0.150, 1.297 0.013 0.091
Perseveration (Random Intercept) 1.367 0.235 0.906, 1.829 < 0.001 RMSSD (Inertia) x rMDD 0.092 0.353 -0.599, 0.784 0.794 0.001
Random Effects
Perseveration (Random Intercept) 1.238 0.207 0.832, 1.643 < 0.001
Sample Entropy (Mean) 0.024 Sample Entropy (Mean) 0.119
Perseveration (Intercept) 2.866 0.152 2.567, 3.164 < 0.001 Perseveration (Intercept) 2.445 0.218 2.018, 2.872 < 0.001
Fixed Effects Fixed Effects
Between-Subject (Level 2) Between-Subject (Level 2)
Sample Entropy (Mean) -0.046 0.149 -0.337, 0.246 0.759 0.001 Sample Entropy (Mean) 0.149 0.200 -0.244, 0.542 0.458 0.001
Activity (Mean) 0.155 0.153 -0.144, 0.454 0.310 0.015 Activity (Mean) 0.227 0.173 -0.112, 0.566 0.190 0.029
Random Effects rMDD 0.749 0.305 0.150, 1.348 0.014 0.091
Perseveration (Random Intercept) 1.366 0.235 0.905, 1.827 < 0.001 Sample Entropy (Mean) x rMDD -0.161 0.250 -0.651, 0.329 0.519 0.004
Random Effects
Perseveration (Random Intercept) 1.233 0.209 0.824, 1.641 < 0.001
Sample Entropy (Inertia) 0.059 Sample Entropy (Inertia) 0.204
Perseveration (Intercept) 2.868 0.150 2.574, 3.162 < 0.001 Perseveration (Intercept) 2.471 0.215 2.049, 2.892 < 0.001
Fixed Effects Fixed Effects
Between-Subject (Level 2) Between-Subject (Level 2)
Sample Entropy (Inertia) 0.231 0.163 -0.090, 0.551 0.158 0.036 Sample Entropy (Inertia) -0.105 0.203 -0.503, 0.292 0.603 0.030
Activity (Mean) 0.148 0.134 -0.114, 0.410 0.268 0.016 Activity (Mean) 0.141 0.119 -0.093, 0.375 0.237 0.014
Random Effects rMDD 0.701 0.282 0.148, 1.254 0.013 0.085
Perseveration (Random Intercept) 1.317 0.231 0.864, 1.770 < 0.001 Sample Entropy (Inertia) x rMDD 0.590 0.282 0.037, 1.143 0.037 0.060
Random Effects
Perseveration (Random Intercept) 1.114 0.208 0.705, 1.522 < 0.001
Predicting Perseverative Cognition
Dimensional Models Moderation by Group Models
R2 R2

Copyright 2024 22 Sarah L. Zapetis
However, the relationship between entropy inertia and PC differed significantly by group
(Table 4 and Figure 4), indicating that people with more inertia tended to perseverate more,
but only if they had remitted MDD (rMDD group: B = 0.484, SE = 0.204, 95% CI = [0.085,
0.884], p = .017; CON group: B = -0.105, SE = 0.203, 95% CI = [-0.503, 0.292], p = .603)2.
The associations between (1) the mean and (2) inertia of RMSSD and (3) the mean of sample
entropy did not significantly differ based on depression history.
2 This moderation was no longer significant when age was included as a covariate in the model (B = 0.426, SE= 0.287, 95% CI =
[-0.137, 0.988], p = 0.138). However, the simple slope of the rMDD group was still significantly positive (B = 0.518, SE= 0.191,
95% CI = [0.144, 0.892], p = 0.007), and the simple slope of the CON group was non-significant (B = 0.092, SE= 0.211, 95% CI
= [-0.321, 0.505], p = 0.662) when this covariate was included.
Figure 3. Between-Person Association between the Mean of RMSSD and
Perseverative Cognition. The mean of RMSSD was negatively associated with
perseverative cognition in the full sample such that individuals with a lower
mean RMSSD engaged in more perseverative cognition. Group did not significantly
moderate this relationship, indicating that the slope of this association did not
significantly differ between groups.
0
1
2
3
4
5
6
7
0 20 40 60 80 100
Perseverative Cognition
RMSSD (mean)
CON rMDD CON rMDD

Copyright 2024 23 Sarah L. Zapetis
Figure 4. Depression History as a Moderator of the Relationship between the Inertia of
Sample Entropy and Perseverative Cognition at the Between-Person Level. The slope
of the between-person relationship between the inertia of sample entropy and perseverative
cognition differed significantly by group with rMDD individuals exhibiting a significantly
positive association and CON individuals exhibiting a nonsignificant association.
2
2.5
3
3.5
4
Low High
Perseverative Cognition
Inertia of Sample Entropy (Individual Differences)
CON rMDD

Copyright 2024 24 Sarah L. Zapetis
Chapter 4: Discussion
This study explored links between ambulatory autonomic activity and subsequent
engagement in perseverative cognition (PC) during daily life in individuals with and without
a history of depression. Notably, this study is the first to examine nonlinear autonomic
complexity metrics in daily life as they relate to proximal fluctuations in EMA-reported PC.
By investigating temporally lagged relationships within individuals – rather than between
individuals, as has been typically examined in the past – we aimed to gain a better
understanding of psychophysiological factors temporally linked with PC. The identification
of a proximal marker for real-world engagement in PC that can be passively assessed with
wearable devices could facilitate risk monitoring and inform the development of preventative
tools. We found that periods characterized by higher entropy inertia were more likely to be
followed by greater engagement in PC in individuals with a history of depression. We found
similar relationships at the between-person level, such that individuals with remitted
depression who exhibited higher inertia of sample entropy across the measurement period
engaged in more PC on average. This was not the case for never-depressed control
individuals. Taken together, these results suggest that entropy inertia may serve as a novel
ambulatory metric with utility for identifying both who may be at an increased risk for
engagement in PC, and perhaps more importantly, moments in daily life when risk for
engagement in PC is elevated, in individuals with a history of depression.
The observed within-person coupling between higher entropy inertia and increased
engagement in PC partially supports our initial hypotheses. While we predicted that this
coupling would be present to some degree in both groups and would be stronger in
individuals with a history of depression, we only observed a significant association in the

Copyright 2024 25 Sarah L. Zapetis
clinical group. Given the lack of group differences in entropy inertia, this finding suggests
that individuals with a history of depression may be more likely than never-depressed
individuals to respond to increases in entropy inertia by engaging in PC. However, it is also
possible that the reduced range of PC engagement in the CON group, compared to the rMDD
group, may have obscured certain associations that could be evident had a broader range of
PC been sampled in both groups.
The entropy inertia metric in this study is a measure of sluggishness in autonomic
signal complexity and may represent real-world autonomic inflexibility. Whereas higher
levels of signal complexity in the absence of stressors is thought to reflect the integration of
sympathetic and parasympathetic influences and the potential for the system to respond
flexibly (Bakhchina et al., 2018), changes in signal complexity across time in everyday life
may represent adaptive responses to shifting environmental or internal demands (Schubert et
al., 2009; Visnovcova et al., 2014). Conversely, sluggishness in signal complexity may
represent inflexibility in the balance between branches of the autonomic nervous system and
reduced capacity for affect regulation (Thayer et al., 2012). This autonomic inflexibility in
everyday life may align with existing literature on emotion context insensitivity in depression
(Bylsma et al., 2008; Rottenberg et al., 2005). Specifically, individuals with depression tend
to exhibit a general disengagement from the environment, characterized by attenuated
emotional reactivity to both positive and negative external stimuli and blunted physiological
reactivity in laboratory studies (Bylsma, 2021). Our findings may extend these results to
daily life contexts, as increases in entropy inertia may represent periods of heightened
disengagement from external stimuli and a shift of attentional resources to internal processes
such as spontaneous thought (Marchetti et al., 2016). Consequently, the increased

Copyright 2024 26 Sarah L. Zapetis
automaticity of PC engagement observed in individuals with a history of depression
(Hjartarson et al., 2022) may increase the likelihood that spontaneous thoughts transition to
PC in this population. Thus, while an increase in entropy inertia may indicate a universal
shift to internally focused attention, this hypothesis offers insight into why such increases
may result in increased PC engagement only among individuals with a history of depression.
Future research should investigate these hypotheses to clarify whether increases in entropy
inertia may represent periods of disengagement from external stimuli in everyday life and
how this may contribute to heightened risk for PC in certain individuals.
Fluctuations in the other autonomic complexity metrics assessed at the within-person
level, including the proximal mean and inertia of RMSSD and the mean of sample entropy,
were not associated with fluctuations in PC. Given the novelty of our analyses, we did not
make specific predictions about differences between autonomic complexity metrics in their
relationship to PC. However, variations in the timescale of these metrics may help to account
for these disparate findings. While RMSSD provides information about short-term variation
by comparing only successive heartbeat intervals, sample entropy measures the likelihood
that pairs of successive heartbeat intervals will exhibit similarity when an additional interval
is included across the sampling window (Richman & Moorman, 2000). Compared to
RMSSD, which is thought to mainly represent parasympathetic activity, sample entropy
takes into account sequences in the heart rate signal that are both closer together and farther
apart than RMSSD and is thus thought to reflect more complex information about the balance
of parasympathetic and sympathetic influences. Moreover, while limited studies have
explored fluctuations in linear versus nonlinear autonomic complexity in daily life, there is
some evidence from trait-level, laboratory findings that measures of signal complexity may

Copyright 2024 27 Sarah L. Zapetis
provide clinically relevant information, beyond linear measures. Specifically, lower resting
signal complexity, but not RMSSD, has been associated with increased subclinical symptoms
of depression in a sample of young adult females (Greco et al., 2018). In addition, a recent
study found that individuals with a history of depression who exhibited lower levels of
resting signal complexity while tapering off of antidepressants had a higher chance of
recurrence (George et al., 2023). Further, lower resting signal complexity, but not linear
measures, have been associated with worse cognitive inhibition performance (Young &
Benton, 2015), suggesting that nonlinear metrics may be particularly sensitive to fluctuations
in cognitive processes like PC. In sum, our results broadly align with existing research while
identifying a novel marker of proximal shifts in PC in everyday life. This work suggests that
nonlinear autonomic metrics, particularly signal complexity, may hold utility for predicting
clinically relevant outcomes such as risk for depressive recurrence, which should be a focus
of future research.
At the between-person level, individuals with a history of depression who had higher
levels of entropy inertia exhibited a higher level of PC engagement, paralleling our findings
about fluctuations at the within-person level. In some ways, this association between trait
entropy inertia and PC may help to clarify discrepancies in lab-based findings examining
autonomic reactivity in individuals with remitted depression. Specifically, some studies have
found blunted autonomic reactivity to stress or emotional challenges in individuals with
remitted depression compared to never-depressed controls (Yaroslavsky et al., 2013, 2014),
while others have found no difference (Ahrens et al., 2008; Bylsma et al., 2014; Salomon et
al., 2013). However, limited research has investigated the role of trait-level PC in autonomic
reactivity among individuals with rMDD. Our results suggest that, among individuals with a

Copyright 2024 28 Sarah L. Zapetis
history of depression, only those who have higher levels of trait PC may exhibit entropy
inertia, which may result in diluted group-level effects when combining all depressed
individuals in a single group regardless of PC tendencies. While a direct parallel cannot be
drawn between autonomic reactivity in laboratory paradigms and autonomic sluggishness in
everyday life, future research could investigate whether individuals with rMDD that have
higher trait PC are more likely to exhibit blunted autonomic reactivity than those with lower
trait PC to potentially clarify the noted discrepancy. Given the heterogeneity that exists in
depression (Buch & Liston, 2021; Monroe & Anderson, 2015), the identification of factors –
such as entropy inertia – that could confer risk of recurrence in certain subgroups of
individuals with remitted depression will be important to improving prevention and
treatment.
We also found that individuals with a lower level of RMSSD exhibited greater PC
engagement across the measurement period. This association was not moderated by
depression history, suggesting a dimensional relationship. This dimensional finding extends
trait-level observations examined in laboratory research (Brosschot et al., 2007; Carnevali et
al., 2018; Ottaviani et al., 2016; Williams et al., 2017) in which individuals with lower
RMSSD at rest tend to show greater trait PC. Establishing parallels between laboratory and
ambulatory studies may serve a dual purpose. Firstly, it bolsters confidence in the ecological
validity of laboratory findings by suggesting their real-world relevance. Further, it indicates
that, in cases where ambulatory measurements are impractical, laboratory-based assessments
may still provide valuable insights into trait levels of autonomic complexity as they relate to
real-world PC engagement. Secondly, these parallels provide evidence that ambulatory tools
can effectively capture relevant information about relationships that have been well-

Copyright 2024 29 Sarah L. Zapetis
characterized in laboratory research. Overall, our finding that the trait-level association
between RMSSD and PC extends to measurement in everyday life supports the robust nature
of this relationship.
The results of this study may have several important clinical implications. Firstly, the
within-person temporal association between fluctuations in entropy inertia and PC highlights
the potential for ambulatory psychophysiology to inform JITAI tools that can detect periods
of increased risk for engagement in PC in daily life. Specifically, using real-time data and
algorithms to compute proximal increases in inertia, JITAIs could provide individuals with a
history of depression support during moments when their physiology suggests they may need
it most, to help them prevent or inhibit PC and reduce risk for depressive recurrence. One
approach to intervention could involve directly targeting autonomic processes with
techniques like mobile biofeedback or biocueing (ter Harmsel et al., 2021), which aims to
increase awareness and self-regulation of physiological functions. While biofeedback has
shown some promise in treating depression symptoms (Lehrer et al., 2020), further research
is needed to determine the potential of biofeedback to modify entropy inertia, and whether
doing so could reduce episodes of PC. Another approach to a JITAI could involve utilizing
ambulatory psychophysiology to trigger behavioral interventions known to be effective in
preventing and reducing PC, such as cognitive restructuring and mindfulness strategies
(Querstret & Cropley, 2013), during moments when inertia fluctuates above an individual’s
own usual threshold. Secondly, the trait-level association between entropy inertia and PC in
individuals with a history of depression suggests the possibility of passively identifying
individuals who tend to engage in higher levels of PC based solely on their entropy inertia.
As wearable devices become more affordable and ubiquitous (Chandrasekaran et al., 2020),

Copyright 2024 30 Sarah L. Zapetis
this passive identification approach could be integrated into broader screening initiatives to
identify individuals at higher risk for depressive recurrence and connect them with resources
to aid in prevention.
This study has several strengths, including the use of a novel, dynamic metric to
index autonomic complexity in daily life. In addition, by investigating temporally lagged
associations between fluctuations in autonomic activity and PC, this study may provide
insight into factors that are temporally, and potentially even causally, linked to PC, offering
valuable information for preventative measures. Another significant strength is the use of
passive measurement methods, which may allow for unobtrusive monitoring of risk factors
over extended periods of time. In addition to the strengths, there are several important
limitations that should be considered. Firstly, since contextual information during the
measurement of autonomic complexity was not accounted for in the analyses, we cannot
definitively conclude that increases in entropy inertia represent a maladaptive lack of
flexibility to changing environmental demands. However, given that the autonomic
complexity metrics were measured across 30-minute time intervals, it seems probable that
some degree of fluctuation, even in response to one’s own thoughts and emotions, would be
expected and adaptive during this period. Secondly, the implications of this study may be
limited by the effect size of the primary within-person finding, as fluctuations in entropy
inertia explained 6% of the variance in subsequent fluctuations in PC in the rMDD group.
While this effect is modest, it is important to consider the real-world nature of this study and
thus the introduction of many contextual factors that could contribute to momentary
fluctuations in PC. Further, the identification of an association across modalities (i.e.,
between psychophysiology and EMA) minimizes the impact of shared method variance, and

Copyright 2024 31 Sarah L. Zapetis
thus may contribute to a smaller overall effect. Moreover, given that the goal of the current
study was to identify proximal markers of PC, to the extent that this association can be
reliably replicated, entropy inertia may have some utility for passively detecting periods of
increased risk for PC engagement, despite the small effect size. To enhance the predictive
capacity for PC, future research should incorporate additional passively collected measures
that may further increase the variance explained. Another limitation of this study involves the
restricted number of items used to measure PC. While this measure demonstrated moderate
reliability in our sample, subsequent investigations should consider including additional
items to capture other dimensions of PC (i.e., controllability, repetitiveness) and determine
how they may be uniquely related to entropy inertia. Additionally, while we used nonoverlapping 5-minute epochs to compute autonomic complexity metrics, which is standard in
the field (Malik et al., 1996; Shaffer & Ginsberg, 2017), future research should explore how
different epoch overlaps (as we examined in supplementary analyses) and epochs lengths
(e.g., 1 minute) impact these metrics and their associations with real-world PC. Moreover,
examining these associations in active MDD could help clarify state- and trait-level factors
influencing the relationship between entropy inertia and PC, potentially offering a broader
range of both variables and revealing stronger or more nuanced effects with additional
implications for the treatment of depression. Furthermore, considering evidence that PC
engagement is elevated across many mental health conditions (Kaplan et al., 2018; McEvoy
et al., 2013), further research should investigate the specificity of the relationship between
fluctuations in entropy inertia and PC to depression versus to a broader risk for
psychopathology transdiagnostically. Lastly, to provide additional support for establishing
whether the entropy inertia-PC relationship is potentially causal in nature, subsequent

Copyright 2024 32 Sarah L. Zapetis
investigations should examine whether and how entropy inertia can be manipulated and
whether altering this metric results in changes in PC engagement.
Conclusions
In conclusion, ambulatory monitoring of entropy inertia could have utility for
detecting moments of increased risk for engagement in PC. Given a wealth of evidence
implicating PC engagement as a key risk factor in depression, the ability to prevent or reduce
engagement in PC in individuals with a history of depression is essential to reducing the high
rate of recurrence. The identification of a passive physiological risk marker of PC may allow
for unobtrusive, long-term monitoring and in-the-moment interventions that could
complement conventional treatments and reduce the prevalence of depressive recurrence.
Overall, this study offers insights toward understanding the complex relationship between
autonomic activity and PC in daily life, offering promising avenues for future research on
physiological mechanisms of PC and for intervention development.

Copyright 2024 33 Sarah L. Zapetis
References
Abela, J. R. Z., & Hankin, B. L. (2011). Rumination as a vulnerability factor to depression
during the transition from early to middle adolescence: A multiwave longitudinal study.
Journal of Abnormal Psychology, 120(2), 259–271. https://doi.org/10.1037/a0022796
Ahrens, T., Deuschle, M., Krumm, B., van der Pompe, G., den Boer, J. A., & Lederbogen, F.
(2008). Pituitary-adrenal and sympathetic nervous system responses to stress in women
remitted from recurrent major depression. Psychosomatic Medicine, 70(4), 461–467.
https://doi.org/10.1097/PSY.0b013e31816b1aaa
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions.
Sage Publications, Inc.
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental
disorders: DSM-5 (Fifth edition). American Psychiatric Publishing.
Bakhchina, A., Arutyunova, K., Sozinov, A., Demidovsky, A., & Alexandrov, Y. (2018). Sample
entropy of the heart rate reflects properties of the system organization of behaviour.
Entropy, 20(6), 449. https://doi.org/10.3390/e20060449
Balzarotti, S., Biassoni, F., Colombo, B., & Ciceri, M. R. (2017). Cardiac vagal control as a
marker of emotion regulation in healthy adults: A review. Biological Psychology, 130,
54–66. https://doi.org/10.1016/j.biopsycho.2017.10.008
Bradburn, N. M., Rips, L. J., & Shevell, S. K. (1987). Answering autobiographical questions:
The impact of memory and inference on surveys. Science, 236(4798), 157–161.
https://doi.org/10.1126/science.3563494
Brosschot, J. F., Gerin, W., & Thayer, J. F. (2006). The perseverative cognition hypothesis: A
review of worry, prolonged stress-related physiological activation, and health. Journal of
Psychosomatic Research, 60(2), 113–124.
https://doi.org/10.1016/j.jpsychores.2005.06.074
Brosschot, J. F., Van Dijk, E., & Thayer, J. F. (2007). Daily worry is related to low heart rate
variability during waking and the subsequent nocturnal sleep period. International
Journal of Psychophysiology: Official Journal of the International Organization of
Psychophysiology, 63(1), 39–47. https://doi.org/10.1016/j.ijpsycho.2006.07.016
Buch, A. M., & Liston, C. (2021). Dissecting diagnostic heterogeneity in depression by
integrating neuroimaging and genetics. Neuropsychopharmacology, 46(1), 156–175.
https://doi.org/10.1038/s41386-020-00789-3
Bylsma, L. M. (2021). Emotion context insensitivity in depression: Toward an integrated and
contextualized approach. Psychophysiology, 58(2), e13715.
https://doi.org/10.1111/psyp.13715
Bylsma, L. M., Morris, B. H., & Rottenberg, J. (2008). A meta-analysis of emotional reactivity

Copyright 2024 34 Sarah L. Zapetis
in major depressive disorder. Clinical Psychology Review, 28(4), 676–691.
https://doi.org/10.1016/j.cpr.2007.10.001
Bylsma, L. M., Salomon, K., Taylor-Clift, A., Morris, B. H., & Rottenberg, J. (2014).
Respiratory sinus arrhythmia reactivity in current and remitted major depressive disorder.
Psychosomatic Medicine, 76(1), 66–73. https://doi.org/10.1097/PSY.0000000000000019
Carnevali, L., Thayer, J. F., Brosschot, J. F., & Ottaviani, C. (2018). Heart rate variability
mediates the link between rumination and depressive symptoms: A longitudinal study.
International Journal of Psychophysiology, 131, 131–138.
https://doi.org/10.1016/j.ijpsycho.2017.11.002
Chandrasekaran, R., Katthula, V., & Moustakas, E. (2020). Patterns of use and key predictors for
the use of wearable health care devices by US adults: Insights from a national survey.
Journal of Medical Internet Research, 22(10), e22443. https://doi.org/10.2196/22443
Clark, D. M., & Teasdale, J. D. (1982). Diurnal variation in clinical depression and accessibility
of memories of positive and negative experiences. Journal of Abnormal Psychology,
91(2), 87–95. https://doi.org/10.1037/0021-843X.91.2.87
Davis, R. N., & Nolen-Hoeksema, S. (2000). Cognitive inflexibility among ruminators and
nonruminators. Cognitive Therapy and Research, 24(6), 699–711.
https://doi.org/10.1023/A:1005591412406
Eaton, W. W., Shao, H., Nestadt, G., Lee, B. H., Bienvenu, O. J., & Zandi, P. (2008).
Population-based study of first onset and chronicity in major depressive disorder.
Archives of General Psychiatry, 65(5), 513–520.
https://doi.org/10.1001/archpsyc.65.5.513
Egloff, B., Schmukle, S. C., Burns, L. R., & Schwerdtfeger, A. (2006). Spontaneous emotion
regulation during evaluated speaking tasks: Associations with negative affect, anxiety
expression, memory, and physiological responding. Emotion, 6(3), 356–366.
https://doi.org/10.1037/1528-3542.6.3.356
Ehring, T., & Watkins, E. R. (2008). Repetitive negative thinking as a transdiagnostic process.
International Journal of Cognitive Therapy, 1, 192–205.
https://doi.org/10.1680/ijct.2008.1.3.192
Falkenström, F., Finkel, S., Sandell, R., Rubel, J. A., & Holmqvist, R. (2017). Dynamic models
of individual change in psychotherapy process research. Journal of Consulting and
Clinical Psychology, 85(6), 537–549. https://doi.org/10.1037/ccp0000203
Ferrari, A. J., Charlson, F. J., Norman, R. E., Patten, S. B., Freedman, G., Murray, C. J. L., Vos,
T., & Whiteford, H. A. (2013). Burden of depressive disorders by country, sex, age, and
year: Findings from the global burden of disease study 2010. PLoS Medicine, 10(11),
e1001547. https://doi.org/10.1371/journal.pmed.1001547
Fiskum, C., Andersen, T. G., Bornas, X., Aslaksen, P. M., Flaten, M. A., & Jacobsen, K. (2018).

Copyright 2024 35 Sarah L. Zapetis
Non-linear heart rate variability as a discriminator of internalizing psychopathology and
negative affect in children with internalizing problems and healthy controls. Frontiers in
Physiology, 9. https://www.frontiersin.org/articles/10.3389/fphys.2018.00561
Fresco, D. M., Frankel, A. N., Mennin, D. S., Turk, C. L., & Heimberg, R. G. (2002). Distinct
and overlapping features of rumination and worry: The relationship of cognitive
production to negative affective states. Cognitive Therapy and Research. 26(2), 179–
188. https://doi.org/10.1023/A:1014517718949
George, S. V., Kunkels, Y. K., Smit, A., Wichers, M., Snippe, E., van Roon, A. M., & Riese, H.
(2023). Predicting recurrence of depression using cardiac complexity in individuals
tapering antidepressants. Translational Psychiatry, 13(1), Article 1.
https://doi.org/10.1038/s41398-023-02474-7
Gorday, J. Y., Rogers, M. L., & Joiner, T. E. (2018). Examining characteristics of worry in
relation to depression, anxiety, and suicidal ideation and attempts. Journal of Psychiatric
Research, 107, 97–103. https://doi.org/10.1016/j.jpsychires.2018.10.004
Greco, A., Messerotti Benvenuti, S., Gentili, C., Palomba, D., Scilingo, E. P., & Valenza, G.
(2018). Assessment of linear and nonlinear/complex heartbeat dynamics in subclinical
depression (dysphoria). Physiological Measurement, 39(3), 034004.
https://doi.org/10.1088/1361-6579/aaaeac
Gruber, J., Harvey, A. G., & Gross, J. J. (2012). When trying is not enough: Emotion regulation
and the effort–success gap in bipolar disorder. Emotion, 12(5), 997–1003.
https://doi.org/10.1037/a0026822
Gruber, J., Mennin, D. S., Fields, A., Purcell, A., & Murray, G. (2015). Heart rate variability as a
potential indicator of positive valence system disturbance: A proof of concept
investigation. International Journal of Psychophysiology, 98(2), 240–248.
https://doi.org/10.1016/j.ijpsycho.2015.08.005
Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery and
Psychiatry, 23(1), 56–62. https://doi.org/10.1136/jnnp.23.1.56
Harris, P. A., Taylor, R., Thielke, R., Payne, J., Gonzalez, N., & Conde, J. G. (2009). Research
electronic data capture (REDCap)—A metadata-driven methodology and workflow
process for providing translational research informatics support. Journal of Biomedical
Informatics, 42(2), 377–381. https://doi.org/10.1016/j.jbi.2008.08.010
Hjartarson, K. H., Snorrason, I., Bringmann, L. F., & Ólafsson, R. P. (2022). Automaticity and
depression: Daily mood-reactive rumination in people with and without depression
history. Journal of Psychopathology and Clinical Science, 131(4), 327–340.
https://doi.org/10.1037/abn0000752
Houben, M., Van Den Noortgate, W., & Kuppens, P. (2015). The relation between short-term
emotion dynamics and psychological well-being: A meta-analysis. Psychological
Bulletin, 141(4), 901–930. https://doi.org/10.1037/a0038822

Copyright 2024 36 Sarah L. Zapetis
Just, N., & Alloy, L. B. (1997). The response styles theory of depression: Tests and an extension
of the theory. Journal of Abnormal Psychology, 106(2), 221–229.
https://doi.org/10.1037/0021-843X.106.2.221
Kaplan, D. M., Palitsky, R., Carey, A. L., Crane, T. E., Havens, C. M., Medrano, M. R., Reznik,
S. J., Sbarra, D. A., & O’Connor, M.-F. (2018). Maladaptive repetitive thought as a
transdiagnostic phenomenon and treatment target: An integrative review. Journal of
Clinical Psychology, 74(7), 1126–1136. https://doi.org/10.1002/jclp.22585
Koval, P., Kuppens, P., Allen, N. B., & Sheeber, L. (2012). Getting stuck in depression: The
roles of rumination and emotional inertia. Cognition & Emotion, 26(8), 1412–1427.
https://doi.org/10.1080/02699931.2012.667392
Kuppens, P., Allen, N. B., & Sheeber, L. B. (2010). Emotional inertia and psychological
maladjustment. Psychological Science, 21(7), 984–991.
https://doi.org/10.1177/0956797610372634
Lai, M. H. C., & Steinley, D. (2021). Composite Reliability of Multilevel Data: It’s About
Observed Scores and Construct Meanings. Psychological Methods, 26(1), 90–102.
Lanata, A., Valenza, G., Nardelli, M., Gentili, C., & Scilingo, E. P. (2015). Complexity index
from a personalized wearable monitoring system for assessing remission in mental health.
IEEE Journal of Biomedical and Health Informatics, 19(1), 132–139.
https://doi.org/10.1109/JBHI.2014.2360711
Lehrer, P., Kaur, K., Sharma, A., Shah, K., Huseby, R., Bhavsar, J., Sgobba, P., & Zhang, Y.
(2020). Heart rate variability biofeedback improves emotional and physical health and
performance: A systematic review and meta analysis. Applied Psychophysiology and
Biofeedback, 45(3), 109–129. https://doi.org/10.1007/s10484-020-09466-z
Malik, M., Camm, A. J., Bigger, J. T., Breithardt, G., Cerutti, S., Cohen, R. J., Coumel, P.,
Fallen, E. L., Kennedy, H. L., Kleiger, R. E., Lombardi, F., Malliani, A., Moss, A. J.,
Rottman, J. N., Schmidt, G., Schwartz, P. J., & Singer, D. H. (1996). Heart rate
variability. Standards of measurement, physiological interpretation, and clinical use.
European Heart Journal, 17(3), 354–381.
https://doi.org/10.1093/oxfordjournals.eurheartj.a014868
Marchetti, I., Koster, E. H. W., Klinger, E., & Alloy, L. B. (2016). Spontaneous thought and
vulnerability to mood disorders: The dark side of the wandering mind. Clinical
Psychological Science, 4(5), 835–857. https://doi.org/10.1177/2167702615622383
McEvoy, P. M., Watson, H., Watkins, E. R., & Nathan, P. (2013). The relationship between
worry, rumination, and comorbidity: Evidence for repetitive negative thinking as a
transdiagnostic construct. Journal of Affective Disorders, 151(1), 313–320.
https://doi.org/10.1016/j.jad.2013.06.014
Mennin, D. S., & Fresco, D. M. (2013). What, me worry and ruminate about DSM‐5 and RDoC?
The importance of targeting negative self‐referential processing. Clinical Psychology:

Copyright 2024 37 Sarah L. Zapetis
Science and Practice, 20(3), 258–267. https://doi.org/10.1111/cpsp.12038
Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and
validation of the Penn State Worry Questionnaire. Behaviour research and
therapy, 28(6), 487–495. https://doi.org/10.1016/0005-7967(90)90135-6
Moberly, N. J., & Watkins, E. R. (2008). Ruminative Self-Focus and Negative Affect. Journal of
Abnormal Psychology, 117(2), 314–323. https://doi.org/10.1037/0021-843X.117.2.314
Monroe, S. M., & Anderson, S. F. (2015). Depression: The shroud of heterogeneity. Current
Directions in Psychological Science, 24(3), 227–231.
https://doi.org/10.1177/0963721414568342
Mueller, T. I., Leon, A. C., Keller, M. B., Solomon, D. A., Endicott, J., Coryell, W., Warshaw,
M., & Maser, J. D. (1999). Recurrence after recovery from major depressive disorder
during 15 years of observational follow-up. American Journal of Psychiatry, 156(7),
1000–1006. https://doi.org/10.1176/ajp.156.7.1000
Muthén, L. K., & Muthén, B. (2010). Mplus statistical analysis with latent variables: User’s
guide. http://www.statmodel.com
Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of
depressive episodes. Journal of Abnormal Psychology, 100(Nov 91), 569–582.
Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination.
Perspectives on Psychological Science, 3(5), 400–424. https://doi.org/10.1111/j.1745-
6924.2008.00088.x
Nurnberger, J. I., Jr, Blehar, M. C., Kaufmann, C. A., York-Cooler, C., Simpson, S. G., HarkavyFriedman, J., Severe, J. B., Malaspina, D., & Reich, T. (1994). Diagnostic Interview for
Genetic Studies: Rationale, unique features, and training. Archives of General Psychiatry,
51(11), 849–859. https://doi.org/10.1001/archpsyc.1994.03950110009002
Ottaviani, C., Shapiro, D., & Couyoumdjian, A. (2013). Flexibility as the key for somatic health:
From mind wandering to perseverative cognition. Biological Psychology, 94(1), 38–43.
https://doi.org/10.1016/j.biopsycho.2013.05.003
Ottaviani, C., Thayer, J. F., Verkuil, B., Lonigro, A., Medea, B., Couyoumdjian, A., &
Brosschot, J. F. (2016). Physiological concomitants of perseverative cognition: A
systematic review and meta-analysis. Psychological Bulletin, 142(3), 231–259.
https://doi.org/10.1037/bul0000036
Papageorgiou, C., & Wells, A. (1999). Process and meta-cognitive dimensions of depressive and
anxious thoughts and relationships with emotional intensity. Clinical Psychology &
Psychotherapy, 6(2), 156–162. https://doi.org/10.1002/(SICI)1099-
0879(199905)6:2<156::AID-CPP196>3.0.CO;2-A
Phillips, M. L., Travis, M. J., Fagiolini, A., & Kupfer, D. J. (2008). Medication effects in

Copyright 2024 38 Sarah L. Zapetis
neuroimaging studies of bipolar disorder. The American Journal of Psychiatry, 165(3),
313–320. https://doi.org/10.1176/appi.ajp.2007.07071066
Pichot, V., Roche, F., Celle, S., Barthélémy, J.-C., & Chouchou, F. (2016). HRVanalysis: A free
software for analyzing cardiac autonomic activity. Frontiers in Physiology, 7.
https://www.frontiersin.org/articles/10.3389/fphys.2016.00557
Piet, J., & Hougaard, E. (2011). The effect of mindfulness-based cognitive therapy for
prevention of relapse in recurrent major depressive disorder: A systematic review and
meta-analysis. Clinical Psychology Review, 31(6), 1032–1040.
https://doi.org/10.1016/j.cpr.2011.05.002
Querstret, D., & Cropley, M. (2013). Assessing treatments used to reduce rumination and/or
worry: A systematic review. Clinical Psychology Review, 33(8), 996–1009.
https://doi.org/10.1016/j.cpr.2013.08.004
Richman, J. S., Lake, D. E., & Moorman, J. R. (2004). Sample entropy. Methods in enzymology,
384, 172–184. https://doi.org/10.1016/S0076-6879(04)84011-4
Richman, J. S., & Moorman, J. R. (2000). Physiological time-series analysis using approximate
entropy and sample entropy. American Journal of Physiology-Heart and Circulatory
Physiology, 278(6), H2039–H2049. https://doi.org/10.1152/ajpheart.2000.278.6.H2039
Rights, J. D., & Sterba, S. K. (2020). New Recommendations on the Use of R-Squared
Differences in Multilevel Model Comparisons. Multivariate behavioral research, 55(4),
568–599. https://doi.org/10.1080/00273171.2019.1660605
Rottenberg, J., Gross, J. J., & Gotlib, I. H. (2005). Emotion context insensitivity in major
depressive disorder. Journal of Abnormal Psychology, 114(4), 627–639.
https://doi.org/10.1037/0021-843X.114.4.627
Salomon, K., Bylsma, L. M., White, K. E., Panaite, V., & Rottenberg, J. (2013). Is blunted
cardiovascular reactivity in depression mood-state dependent? A comparison of major
depressive disorder remitted depression and healthy controls. International Journal of
Psychophysiology : Official Journal of the International Organization of
Psychophysiology, 90(1), 50–57. https://doi.org/10.1016/j.ijpsycho.2013.05.018
Schubert, C., Lambertz, M., Nelesen, R. A., Bardwell, W., Choi, J.-B., & Dimsdale, J. E. (2009).
Effects of stress on heart rate complexity—A comparison between short-term and chronic
stress. Biological Psychology, 80(3), 325–332.
https://doi.org/10.1016/j.biopsycho.2008.11.005
Segerstrom, S. C., Tsao, J. C. I., Alden, L. E., & Craske, M. G. (2000). Worry and rumination:
Repetitive thought as a concomitant and predictor of negative mood. Cognitive Therapy
and Research, 24(6), 671–688. https://doi.org/10.1023/A:1005587311498
Shaffer, F., & Ginsberg, J. P. (2017). An overview of heart rate variability metrics and norms.
Frontiers in Public Health, 5, 258. https://doi.org/10.3389/fpubh.2017.00258

Copyright 2024 39 Sarah L. Zapetis
Smith, R., Thayer, J. F., Khalsa, S. S., & Lane, R. D. (2017). The hierarchical basis of
neurovisceral integration. Neuroscience & Biobehavioral Reviews, 75, 274–296.
https://doi.org/10.1016/j.neubiorev.2017.02.003
Spasojević, J., & Alloy, L. B. (2001). Rumination as a common mechanism relating depressive
risk factors to depression. Emotion, 1(1), 25–37. https://doi.org/10.1037/1528-
3542.1.1.25
Spinhoven, P., van Hemert, A. M., & Penninx, B. W. (2018). Repetitive negative thinking as a
predictor of depression and anxiety: A longitudinal cohort study. Journal of Affective
Disorders, 241, 216–225. https://doi.org/10.1016/j.jad.2018.08.037
Stange, J. P., Alloy, L. B., & Fresco, D. M. (2017). Inflexibility as a vulnerability to depression:
A systematic qualitative review. Clinical Psychology: Science and Practice, 24(3), 245–
276. https://doi.org/10.1111/cpsp.12201
Stange, J. P., Boccia, A. S., Shapero, B. G., Molz, A. R., Flynn, M., Matt, L. M., Abramson, L.
Y., & Alloy, L. B. (2013). Emotion regulation characteristics and cognitive
vulnerabilities interact to predict depressive symptoms in individuals at risk for bipolar
disorder: A prospective behavioural high-risk study. Cognition & Emotion, 27(1), 63–84.
https://doi.org/10.1080/02699931.2012.689758
Stange, J. P., Connolly, S. L., Burke, T. A., Hamilton, J. L., Hamlat, E. J., Abramson, L. Y., &
Alloy, L. B. (2016). Inflexible cognition predicts first onset of major depressive episodes
in adolescence. Depression and Anxiety, 33(11), 1005–1012.
https://doi.org/10.1002/da.22513
Stange, J. P., Hamilton, J. L., Fresco, D. M., & Alloy, L. B. (2017a). Flexible parasympathetic
responses to sadness facilitate spontaneous affect regulation. Psychophysiology, 54(7),
1054–1069. https://doi.org/10.1111/psyp.12856
Stange, J. P., Hamilton, J. L., Fresco, D. M., & Alloy, L. B. (2017b). Perseverate or decenter?
Differential effects of metacognition on the relationship between parasympathetic
inflexibility and symptoms of depression in a multi-wave study. Behaviour Research and
Therapy, 97, 123–133. https://doi.org/10.1016/j.brat.2017.07.007
Stange, J. P., Hamilton, J. L., Shepard, R., Wu, J., Fresco, D. M., & Alloy, L. B. (2020).
Inflexible autonomic responses to sadness predict habitual and real-world rumination: A
multi-level, multi-wave study. Biological Psychology, 153, 107886.
https://doi.org/10.1016/j.biopsycho.2020.107886
Stange, J. P., Kleiman, E. M., Mermelstein, R. J., & Trull, T. J. (2019). Using ambulatory
assessment to measure dynamic risk processes in affective disorders. Journal of Affective
Disorders, 259, 325–336. https://doi.org/10.1016/j.jad.2019.08.060
Stange, J. P., Li, J., Xu, E. P., Ye, Z., Zapetis, S. L., Phanord, C. S., Wu, J., Sellery, P., Keefe,
K., Forbes, E., Mermelstein, R. J., Trull, T. J., & Langenecker, S. A. (2023). Autonomic
complexity dynamically indexes affect regulation in everyday life. Journal of

Copyright 2024 40 Sarah L. Zapetis
Psychopathology and Clinical Science. https://doi.org/10.1037/abn0000849
Starcevic, V. (1995). Pathological worry in major depression: A preliminary report. Behaviour
Research and Therapy, 33(1), 55–56. https://doi.org/10.1016/0005-7967(93)E0028-4
Suls, J., Green, P., & Hillis, S. (1998). Emotional reactivity to everyday problems, affective
inertia, and neuroticism. Personality and Social Psychology Bulletin, 24(2), 127–136.
https://doi.org/10.1177/0146167298242002
Taylor D. (1990). Time-series analysis. Use of autocorrelation as an analytic strategy for
describing pattern and change. Western Journal of Nursing Research, 12(2), 254–261.
https://doi.org/10.1177/019394599001200210
ter Harmsel, J. F., Noordzij, M. L., Goudriaan, A. E., Dekker, J. J. M., Swinkels, L. T. A., van
der Pol, T. M., & Popma, A. (2021). Biocueing and ambulatory biofeedback to enhance
emotion regulation: A review of studies investigating non-psychiatric and psychiatric
populations. International Journal of Psychophysiology, 159, 94–106.
https://doi.org/10.1016/j.ijpsycho.2020.11.009
Thayer, J. F., Åhs, F., Fredrikson, M., Sollers, J. J., & Wager, T. D. (2012). A meta-analysis of
heart rate variability and neuroimaging studies: Implications for heart rate variability as a
marker of stress and health. Neuroscience & Biobehavioral Reviews, 36(2), 747–756.
https://doi.org/10.1016/j.neubiorev.2011.11.009
Thayer, J. F., & Lane, R. D. (2002). Perseverative thinking and health: Neurovisceral
concomitants. Psychology & Health, 17(5), 685–695.
https://doi.org/10.1080/08870440290025867
Thayer, J. F., & Lane, R. D. (2009). Claude Bernard and the heart–brain connection: Further
elaboration of a model of neurovisceral integration. Neuroscience & Biobehavioral
Reviews, 33(2), 81–88. https://doi.org/10.1016/j.neubiorev.2008.08.004
Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S. (2003). Rumination reconsidered: A
psychometric analysis. Cognitive Therapy and Research, 27(3), 247–259.
https://doi.org/10.1023/A:1023910315561
Verkuil, B., Brosschot, J. F., Tollenaar, M. S., Lane, R. D., & Thayer, J. F. (2016). Prolonged
non-metabolic heart rate variability reduction as a physiological marker of psychological
stress in daily life. Annals of Behavioral Medicine, 50(5), 704–714.
https://doi.org/10.1007/s12160-016-9795-7
Visnovcova, Z., Mestanik, M., Javorka, M., Mokra, D., Gala, M., Jurko, A., Calkovska, A., &
Tonhajzerova, I. (2014). Complexity and time asymmetry of heart rate variability are
altered in acute mental stress. Physiological Measurement, 35(7), 1319–1334.
https://doi.org/10.1088/0967-3334/35/7/1319
Watkins, E., Moulds, M., & Mackintosh, B. (2005). Comparisons between rumination and worry
in a non-clinical population. Behaviour Research and Therapy, 43(12), 1577–1585.

Copyright 2024 41 Sarah L. Zapetis
https://doi.org/10.1016/j.brat.2004.11.008
Watkins, E. R., Mullan, E., Wingrove, J., Rimes, K., Steiner, H., Bathurst, N., Eastman, R., &
Scott, J. (2011). Rumination-focused cognitive–behavioural therapy for residual
depression: Phase II randomised controlled trial. British Journal of Psychiatry, 199(4),
317–322. https://doi.org/10.1192/bjp.bp.110.090282
Wenzel, M., Bürgler, S., Rowland, Z., & Hennecke, M. (2023). Self-control dynamics in daily
life: The importance of variability between self-regulatory strategies and strategy
differentiation. European Journal of Personality, 37(1), 33–56.
https://doi.org/10.1177/08902070211043023
Williams, D. P., Feeling, N. R., Hill, L. K., Spangler, D. P., Koenig, J., & Thayer, J. F. (2017).
Resting heart rate variability, facets of rumination and trait anxiety: Implications for the
perseverative cognition hypothesis. Frontiers in Human Neuroscience, 11, 520.
https://doi.org/10.3389/fnhum.2017.00520
Yaroslavsky, I., Bylsma, L. M., Rottenberg, J., & Kovacs, M. (2013). Combinations of resting
RSA and RSA reactivity impact maladaptive mood repair and depression symptoms.
Biological Psychology, 94(2), 272–281. https://doi.org/10.1016/j.biopsycho.2013.06.008
Yaroslavsky, I., Rottenberg, J., Bylsma, L. M., Jennings, J. R., George, C., Baji, I., Benák, I.,
Dochnal, R., Halas, K., Kapornai, K., Kiss, E., Makai, A., Varga, H., Vetró, Á., &
Kovacs, M. (2016). Parasympathetic nervous system activity predicts mood repair use
and its effectiveness among adolescents with and without histories of major depression.
Journal of Abnormal Psychology, 125(3), 323–336. https://doi.org/10.1037/abn0000149
Yaroslavsky, I., Rottenberg, J., & Kovacs, M. (2014). Atypical patterns of respiratory sinus
arrhythmia index an endophenotype for depression. Development and Psychopathology,
26(4pt2), 1337–1352. https://doi.org/10.1017/S0954579414001060
Young, H., & Benton, D. (2015). We should be using nonlinear indices when relating heart-rate
dynamics to cognition and mood. Scientific Reports, 5(1), 16619.
https://doi.org/10.1038/srep16619

Copyright 2024 42 Sarah L. Zapetis
Supplementary Methods
Inertia Computation
The inertia of autonomic complexity variables in this study was calculated as the lag 1
autocorrelation across the six epochs within each autonomic complexity period (See
Supplementary Figure S1). See the formula below, adapted from Taylor (1990).
�(�) = ∑ (�! − �̅)(�!"# − �̅) $%#
!%&
∑ (�! − �̅) $ '
!%&
�! = observation at time t
� = degree of lag (k =1 in this study)
� + � = serial dependency of one observation point with a specified lag
� = number of observations in the time series where k < N
� = mean of the total time series
Sample Entropy Computation
Sample entropy (SampEn) measures the likelihood that two sequences of R-R intervals
that are similar over m timepoints differ at the next time point (m+1; Lanata et al., 2015;
Richman & Moorman, 2000; See Supplementary Figure S2). For example, when m = 2, the
first pair of R-R intervals is compared against all other subsequent pairs, then the second pair is
compared all subsequent pairs, and so on until every duplet of R-R intervals has been compared
with the rest. Then, the same procedure would be repeated for sequences of three R-R intervals
(triplets). Two sequences are considered a match if the distance between them is within a
threshold r. The proportions of matched sequences of length m and m+1 are calculated across all
comparisons, and sample entropy is derived from the negative logarithm of the ratio between
these two proportions. Thus, sample entropy requires three parameters to be defined: m, which is
the embedding dimension (i.e., length) of patterns; r, which is the tolerance threshold for

Copyright 2024 43 Sarah L. Zapetis
distance below which two patterns are considered matched; and N, which is the sample size. In
our study, m was set to 2 and r to 0.2 times the standard deviation of the time series, which are
standards commonly reported in the literature (Fiskum et al., 2018).
For a 5-min epoch of R-R intervals �(�) = �(1), �(2), … , �(�), SampEn is calculated
as follows (Richman & Moorman, 2000):
1. Form N-m-1 vectors (“templates”) of length m. Each pair of templates is defined
as:
�)(�) = [�(�), �(� + 1), … , �(� + � − 1)], 1 ≤ � ≤ � − �
�)(�) = [�(�), �(� + 1), … , �(� + � − 1)], 1 ≤ �� ≤ � − �,� ≠ �
2. Let �<�)(�), �)(�)= be the Euclidean distance between �)(�) and �)(�).
3. Let �* denote the number of �)(�)’s that are within r unit distance from �)(�),
such that �<�)(�), �)(�)= ≤ �.
4. Define �*
)(�) = +!
$%)%&
, which is the probability of finding a match for the
template �)(�).
5. Define �)(�) = ∑"#$ !%& +!
$(.)
$%) , which is the average probability of finding a match
for templates of length m.
6. Repeat the above steps with templates of length m+1. Define �* as the number of
�)"&(�)’s that are within r unit distance from �)"&(�), such that
�<�)"&(�), �)"&(�)= ≤ �.
7. Define �*
)"&(�) = 0!
$%)%&
, which is the probability of finding a match for the
template �)"&(�).
8. Define �)"&(�) = ∑"#$ !%& 0!
$'&(.)
$%) , which is the average probability of finding a
match for templates of length m+1.
9. ��(�, �, �) = −�� F
0$'&(.)
+$(.) G, where 0$'&(.)
+$(.) is the conditional probability
that two sequences match for m+1 data points given that they are matched for m
data points.
If there are many length-m matches but few length-(m+1) matches, this means that most
length-m sequences do not continue to match at the next timepoint. In this case, similarity
of patterns does not help predict the next measurements, which suggests high irregularity
of the time series. This will result in a small value of 0$'&(.)
+$(.) ,and consequently a large
value of −�� F
0$'&(.)
+$(.) G (i.e., SampEn).

Copyright 2024 44 Sarah L. Zapetis
Inertia Time-Lag Analyses
Sensitivity analyses explored the impact of increasing epoch overlap on autocorrelations and the
predictive ability of inertia variables. In contrast with EMA data, continuous sampling of
psychophysiological data allows for data to be segmented in windows that are either adjacent or
overlapping. Upon initial examination of our data, we noticed that the grand average of sample
entropy inertia was negative and significantly different from zero. This would suggest that, on
average, within an individual, having higher sample entropy during one 5-minute epoch was
associated with a lower sample entropy during the next 5-minute epoch. In addition, we found
that the average RMSSD value was slightly positive but did not significantly differ from zero.
Given that the inertia of autonomic complexity metrics, to our knowledge, has not been
previously investigated, we decided to conduct additional analyses to investigate how lagging the
inertia of the autonomic complexity metrics in different ways may impact their average value
and their relationship with the outcome of interest (PC). Specifically, as a sensitivity analysis, we
computed the inertia of autonomic complexity metrics and activity as the autocorrelation
between 5-minute epochs within each 30-minute period that were 50% overlapping (where each
2.5 minutes within each 5-minute epoch overlapped with the previous 5-minute epoch) and 90%
overlapping (where 4.5 of each 5-minute epoch overlapped with the previous epoch). We then
investigated whether the average of these values across people differed significantly from zero
and in which direction (positive or negative). Next, we calculated the average within-person
correlation between inertia variables with different lags to determine whether they were
providing distinct information. Lastly, we repeated the primary dimensional analyses using the
overlapping inertia metrics as predictors of PC.
We found that inertia of autonomic complexity variables with larger lags were associated

Copyright 2024 45 Sarah L. Zapetis
with larger positive average values that were significantly different from zero (Supplementary
Table S1A). In addition, the non-overlapping inertia variables used in primary analyses were
moderately correlated with inertia variables that used a 50% overlap (RMSSD: r = .670; Sample
Entropy: r = .525) and were less correlated with the inertia variables with a 90% overlap
(RMSSD: r = .436; Sample Entropy: r = .288; Supplementary Table S1B). None of the inertia
variables with overlapping epochs significantly predicted subsequent PC (Supplementary Table
S1C). Future research should continue to investigate the timescale on which fluctuations in the
inertia of autonomic complexity metrics is optimal for predicting PC.

Copyright 2024 46 Sarah L. Zapetis
Supplementary Figures
Supplementary Figure S1. Inertia of Autonomic Complexity Metrics. To calculate the inertia
of autonomic complexity metrics, the values of RMSSD and Sample Entropy (shown here) for
the six epochs within each autonomic complexity period (points connected by solid line) were
lagged by 1 (points connected by dotted line). Then, the Pearson correlation between the
unlagged and lagged data was computed for each autonomic complexity period.

Copyright 2024 47 Sarah L. Zapetis
Supplementary Figure S2. Sample Entropy Calculation from Heart Rate Interval Data.
The two empty circles on the left represent a template of length m (where m = 2) to which all
other consecutive pairs of points in the signal are compared. The three empty shapes on the right
represent a template of length m + 1 to which all other m + 1 sets of points are compared. The
threshold for matching (r) was set to 0.2 times the standard deviation of the signal. The number
of matches for each template length are determined (Ai and Bi ) and then entered into the
computation for sample entropy shown at the bottom. Figure adapted from Richman et al.
(2004).

Copyright 2024 48 Sarah L. Zapetis
Supplementary Table
Supplementary Table S1. Inertia Time-Lag Analyses
A)

B)
C)

Abstract (if available)

Abstract Background: Trait perseverative cognition (PC) is associated with inflexible autonomic activity and risk for depressive recurrence. However, the identification of dynamic psychophysiological markers of PC that fluctuate within individuals over time could facilitate the passive detection of moments when PC occurs in daily life.
Methods: Using intensively sampled data across one week (3x/day) in adults with remitted major depressive disorder (rMDD) and never-depressed controls, we investigated the utility of monitoring ambulatory autonomic complexity to predict moments of PC engagement in everyday life. Autonomic complexity metrics, including the root mean square of successive differences (RMSSD), indexing vagal control, and sample entropy, indexing signal complexity, were calculated in the 30 minutes before each measurement of PC to enable time-lagged analyses. Multilevel models examined proximal fluctuations in the mean level and inertia of complexity metrics as predictors of subsequent PC engagement.
Results: Momentary increases in the inertia of sample entropy, but not other metrics, predicted higher levels of subsequent PC in the rMDD group, but not among never-depressed controls.
Conclusions: The inertia of sample entropy could index autonomic rigidity and serve as a dynamic risk marker for real-world PC in individuals with a history of depression. This could inform the development of technologies to passively detect fluctuations in risk for PC, facilitating real-time interventions to prevent PC and reduce risk for depressive recurrence.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Impulsivity interferes with emotion regulation strategy prioritization in everyday life

Asset Metadata

Creator Zapetis, Sarah L. (author)

Core Title Autonomic inertia as a proximal risk marker for moments of perseverative cognition in everyday life in remitted depression

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Psychology

Degree Conferral Date 2024-12

Publication Date 11/22/2024

Defense Date 03/27/2023

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

Tag ambulatory psychophysiology,autonomic complexity,depression,ecological momentary assessment,perseverative cognition

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Stange, Jonathan P. (committee chair), Martins-Klein, Bruna (committee member), Schneider, Stefan (committee member)

Creator Email sarahzapetis@gmail.com,zapetis@usc.edu

Unique identifier UC11399DUCT

Identifier etd-ZapetisSar-13651.pdf (filename)

Legacy Identifier etd-ZapetisSar-13651

Document Type Thesis

Format theses (aat)

Rights Zapetis, Sarah L.

Internet Media Type application/pdf

Type texts

Source 20241122-usctheses-batch-1224 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

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

Repository Email cisadmin@lib.usc.edu