PreprintPDF Available

REVISING THE SCRIPT: MIXED-METHOD STUDY OF TRAUMA DRAMA FOR COMPLEX-TRAUMA-EXPOSED YOUTH IN RESIDENTIAL TREATMENT A Dissertation

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Abstract Creative-arts interventions have historically been used to treat people exposed to complex trauma. In the United States, however, such interventions have failed to achieve widespread use, perhaps because of obstacles to systematic study, which have prevented development of a sufficient evidence base. This may be attributable in part to the challenges of measuring drama’s differential impact on the unique psychophysiological system of the individual, as well as the complications related to having no legitimized complex trauma diagnosis around which to organize a research agenda until publication of the complex post-traumatic stress (CPTSD) diagnostic classification of 2018 (ICD 11; World Health Organization). In order to simplify future field research and data aggregation across settings, the current study of Trauma Drama (TD; Spinazzola, 2019) made a first attempt to identify a core matrix of variables that might imply an increased flexibility in the underlying constituents that bind heterogeneous symptoms (Insel, 2014). The TD study, as described in this dissertation, is an embedded, combined, concurrent, and sequential mixed-method investigation in a complex-trauma-exposed population of 47 youth in residential treatment in the northeastern United States (treatment condition [n = 27] and a semi-matched comparison condition [n = 20]). Longitudinal pattern analysis of the subtracted distance between resting and maximum capacity heart rate variability (HRV) was used to calculate the novel high frequency HRV (HF HRV) variable, an index of parasympathetic nervous system balance at pre- vs. posttest. My analysis suggested the presence of three physiological subgroups across the treatment and comparison groups. The treatment condition of one of the subgroups appeared to account for improved depression symptoms in the overall treatment group as contrasted with the comparison group. Therefore, it appears that in this study, response to the TD intervention may be, at least partially, moderated by electrophysiological subtype. I used Interpretive Phenomenological Analysis (IPA; Smith & Osborn, 2008) to analyze interviews of treatment group participants (n = 29) and intervention-facilitators (n = 12). The superordinate theme of the qualitative arm of the study was the shift that occurs from a state of Absence (a survival state of fight, flight, and/or freeze that dissociates people from being in the now) to a state of greater Presence (being rooted in the now) via the Modification of Associations Process (MAP). The MAP reorganizes and expands associations undergirding subjective experience of self, others, and the world. Taken together, triangulated data imply that the treatment group experienced a decrease in dissociation (a central symptom in the network of the Post-Traumatic Stress Disorder [PTSD] cluster of the CPTSD diagnosis [Knefel & Lueger-Schuster, 2013]), as well as a decrease in depression (a central symptom in the network of the Disturbances of Self Organization [DSO] cluster of the diagnosis [Haselgruber et al., 2020]). These findings suggest that TD may work to diminish complex trauma pathology in this population, and as a result, further study of TD is indicated. The data also suggest that the stabilization phase of TD is essential to engagement in the intervention and for the consequent reduction of DSO symptoms in particular; therefore further study of TD may offer an opportunity to use the intervention as an exemplar through which the general characteristics, effect, and value of the stabilization phase of component-based treatments for CPTSD may be examined. Keywords: Complex Posttraumatic Stress Disorder (CPTSD); Drama Therapy; Heart-Rate Variability (HRV); Interpretive Phenomenological Analysis (IPA); Stress Sequelae Assessment; HeartMath
Content may be subject to copyright.
REVISING THE SCRIPT: MIXED-METHOD STUDY OF TRAUMA DRAMA
FOR COMPLEX-TRAUMA-EXPOSED YOUTH IN RESIDENTIAL TREATMENT
A Dissertation
Presented to the Faculty of the
School of Human Service Professions
Widener University
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy
by
Mimi Sullivan
Center for Social Work Education
June, 2021
Widener University
Center for Social Work Education
Title of Dissertation: Revising the Script: Mixed-Method Study of Trauma Drama for Complex-
Trauma-Exposed Youth in Residential Treatment
Author: Mimi Sullivan
Candidates Name: Miriam Elizabeth Sullivan
Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy
Copyright by
Miriam Elizabeth Sullivan
2021
i
Dedication
This dissertation is dedicated to the people who made me: my darling dad and mom, Burk
and Miriam Sullivan; my grandmother, Miriam Elizabeth Loughran Fenerty; the acting/music
teachers of my youth, Dolly Beechman Schnall, Jack and Sue Shaw, Eileen and Larry Zerone,
and Mike Lemon; my partners in motherhood, Stephanie Brielle Gullo, Maria Carolla Gullo,
Dolores Landgraf, and Yildiz Dogukan; the two loves of my life, my husband Andy Hyde and
our daughter Isabella Stephanie Hyde; and the generations of theatre artists, healers, teachers,
and scientists upon whose shoulders I stand.
ii
Acknowledgments
Complex trauma disconnects people from sense-of-self, others, and the larger world. This
dissertation is about connection. It describes the experiences of trauma-exposed adolescent
participants as they emerge from isolation and develop healthful, pleasurable connections in the
present. The many people who were involved in bringing this dissertation to publication were
themselves masters of connection. I want to extend particular thanks to:
My steadfast committee: Barbara Gilin, Joseph Spinazzola, Linda Houser, and especially
my Chairperson, Tom Young, who invested themselves in this project and who generously and
tirelessly contributed earned wisdom, scholarly capital, inspiration, pure stamina, and innovative
critical analysis to the TD study (even though the process of investigating a complex naturalistic
intervention was never anticipated to yield quick or easy answers). Barbara contributed insights
from her deep well of clinical wisdom. Joseph tapped into his vast developmental trauma and
publication experience to provide crucial editorial direction. Linda lent quantitative-method
design expertise that allowed me to manage the presentation of a massive amount of data. Tom
stewarded the entire process—always displaying patience and belief in me and in the project—
never forcing a particular perspective onto the research but always pursuing the
reflective/creative/scientific process.
Simon Moon, who tutored, advised, and partnered with me every step of the way in the
design and execution of statistical analysis. All of my professors at the Center for Social Work
Education at Widener University from my MSW days forward, who embraced and nurtured the
development of this expat actor turned social worker. Bessel van der Kolk, who listened when I
told him I wanted to make a contribution toward advancement of drama interventions for
traumatized people and, consequently, introduced me to Joseph Spinazzola, the author of the
iii
Trauma Drama intervention. Joseph himself, who entrusted me with this work. Wendy
D’Andrea, who helped me to come up with the novel heart-rate-variability variable used in the
physiological portion of the study. Jackie Waterman, Rollin McCraty, and Mike Atkinson from
HeartMath, who helped me to adapt HeartMath’s clinical equipment for this research project and
who coached me through every nuance of reliably extracting and preparing the
electrophysiological data for analysis. Lia Martin, Hilary Hodgedon, Mara Renz Smith, and
Linley Frankel, who invested in and supported data collection and preparation. Jen Norton and
Sachi Ando, who got me started on learning SPSS. Lee Ann Pingel of Expert Eye Editing, who
edited and proofread this dense document in its final stages. John Giugliano, Colleen Fleming,
and Lauree Padgett, who were unofficial but crucial readers. Pat Fletcher and Margo Campbell,
who invested their time and expert talents as official readers. Their thorough, wise, and
constructive critique increased the strength, accuracy, and clarity of this dissertation.
The administrators, residential directors, and staff members of the participants’
residences, who welcomed me and supported this research. The far-flung Trauma Drama troupes
who invited me to participate in their trainings, including Douglas Beckstead from Calgary
Family Services in Alberta, Canada; Anne Brinkley, Alan Brinkley, the Center for Children and
Families, and the University of Oklahoma in Norman, OK; Vera Salzburn at Chatham County
Safety Net Planning Council, and Front Porch Improv in Savannah, GA. Merle Perkins and
Kevin Smith, who generously shared their prodigious professional acting/educator/healer talents,
wisdom, heart, soul, skill, and experience as troupe members and creators of the Urban Improv
troupe and as Trauma Drama intervention trainers and contributors.
Awilda Bravo, Ed Hamlin, Thomas Morgan, Jeffrey Holman, Pat Romano McGraw,
Jeanne Metz, David Samuels, Sandra Bloom, and my fellow education committee members at
iv
the Pennsylvania Society of Clinical Social Workers, including Leda Sportalari, Jane Abrams,
and Jacqueline Falkenheim, who provided mentorship on my clinical path. Office-mates Roberta
Morgan, Tori Lynn, and Jill Brinker, who helped me to maintain the spirit and stamina needed to
cross the finish line. Jill Borin at the Widener Library. The leaders of my cross-institutional
doctoral moms’ social-work support group, Kylie Evans and Megan Holmes, and to all my
fellow group members, who shared their institutional privilege, power, and wisdom with me, and
who were a lifeline in the last stages of the dissertation.
My wonderful, nurturing, ever-curious, encyclopedia-minded, ex-monk, philosophy-
professor father, Burk Sullivan, who imbued me with a sense of purpose and an understanding of
the importance of taking a multifaceted approach to the examination of almost any question. To
my darling mom, Miriam Fenerty Sullivan, who has proofread every bit of my graduate-school
writing over the years, even through the loss of my sweet dad to Covid this past pandemic year.
My sweetheart husband Andy Hyde, who I met under the clock at Haddonfield High
School, and our treasure of a daughter Isabella Stephanie, both of whom somehow withstood the
neglect that was the price of my devotion to and completion of this project. (I can never make up
for their patience, but I aim to try.) Stephanie Brielle Gullo, Maria Carolla Gullo, Dolores
Landgraf, Yildiz Dogukan, Dilara Dogukan, Asena Dogukan, Kailey Purnell, and Olivia Horan
for being my partners in motherhood and for helping our little Isabella grow into the beautiful
young woman she has become. Mina Newstadt Segal and Bob Segal, Andy and Isabella Hyde,
and Miriam F. and Burk Sullivan, who made personal financial contributions to this primarily
self-funded research project. My amazing clients, who have taught me much and who have been
positive and supportive during this dissertation project.
v
My mentors and actor partners, Eileen Zerone, Larry Zerone, Dolly Beechman Schall,
Jack Shaw, Mike Lemon, Richard Kislan, George DiCenzo, Martell Spagnolo, Katie Hyde
Lewars, Sheri Mann Stewart, Sharon Elliott Hetland, Donna Marrazzo, Lauree Padgett, Bill
Rauch, Karen Mulle, Chris Sullivan Miller, Liz Fielder Olson, Bill Massey, Greg McMillan,
Milly Massey, Nick Weuhrmann, John Adler, Ron Brown, Becky Rogers Bushong, Elaine
Winder, Jennifer O’Neill, Sue O’Malley-Sheehan, Jim Waring, Janet and Don Waters, Gilbert
Hartke, Ed Cashman, Teddy Handfield, Jeanne Fahey, Bill Graham, Alan Alda at the Alda
Center for Communicating Science and the Clear and Vivid Podcast, John Poulin, Stephen
Kauffman, Richard Cooper, Carolyn Walter, Robin Goldberg Glenn, Paula Silver, Ange Puig,
Diane Alessi, Richard Kislan, Rebecca Stevens, and Rene Pagan.
Influential cultural progenitors John Heminges, Henry Condell, Lauren Gunderson,
Elizabeth Haddon Estaugh, J. Fithian Tatem, the Haddonfield Friends Meeting, Haddonfield
Public Schools, Jane Addams, Neva Boyd, Viola Spolin, Konstantin Stanislavski, Uta Hagen,
Amy Sherman Palladino, Bessel van der Kolk, Julian Ford, Chris Courtois, Rachel Yehuda, Ruth
Lanius, Wendy D’Andrea, Judith Herman, Lenore Terr, Babette Rothschild, Bruce Perry, and
Francine Shapiro.
Most of all, I wish to thank the youth participants and troupe-member informants of this
Trauma Drama study. I am immensely grateful to them. I did my best to accurately present and
transfer the essential truths they conveyed, and I hope that—if they ever read this work—they
will feel that it honors their contribution and represents them well.
vi
Abstract
Creative-arts interventions have historically been used to treat people exposed to complex
trauma. In the United States, however, such interventions have failed to achieve widespread use,
perhaps because of obstacles to systematic study, which have prevented development of a
sufficient evidence base. This may be attributable in part to the challenges of measuring dramas
differential impact on the unique psychophysiological system of the individual, as well as the
complications related to having no legitimized complex trauma diagnosis around which to
organize a research agenda until publication of the complex post-traumatic stress (CPTSD)
diagnostic classification of 2018 (ICD 11; World Health Organization). In order to simplify
future field research and data aggregation across settings, the current study of Trauma Drama
(TD; Spinazzola, 2019) made a first attempt to identify a core matrix of variables that might
imply an increased flexibility in the underlying constituents that bind heterogeneous symptoms
(Insel, 2014). The TD study, as described in this dissertation, is an embedded, combined,
concurrent, and sequential mixed-method investigation in a complex-trauma-exposed population
of 47 youth in residential treatment in the northeastern United States (treatment condition
[n = 27] and a semi-matched comparison condition [n = 20]).
Longitudinal pattern analysis of the subtracted distance between resting and maximum
capacity heart rate variability (HRV) was used to calculate the novel high frequency HRV (HF
HRV) variable, an index of parasympathetic nervous system balance at pre- vs. posttest. My
analysis suggested the presence of three physiological subgroups across the treatment and
comparison groups. The treatment condition of one of the subgroups appeared to account for
improved depression symptoms in the overall treatment group as contrasted with the comparison
group. Therefore, it appears that in this study, response to the TD intervention may be, at least
vii
partially, moderated by electrophysiological subtype. I used Interpretive Phenomenological
Analysis (IPA; Smith & Osborn, 2008) to analyze interviews of treatment group participants
(n = 29) and intervention-facilitators (n = 12). The superordinate theme of the qualitative arm of
the study was the shift that occurs from a state of Absence (a survival state of fight, flight, and/or
freeze that dissociates people from being in the now) to a state of greater Presence (being rooted
in the now) via the Modification of Associations Process (MAP). The MAP reorganizes and
expands associations undergirding subjective experience of self, others, and the world.
Taken together, triangulated data imply that the treatment group experienced a decrease
in dissociation (a central symptom in the network of the Post-Traumatic Stress Disorder [PTSD]
cluster of the CPTSD diagnosis [Knefel & Lueger-Schuster, 2013]), as well as a decrease in
depression (a central symptom in the network of the Disturbances of Self Organization [DSO]
cluster of the diagnosis [Haselgruber et al., 2020]). These findings suggest that TD may work to
diminish complex trauma pathology in this population, and as a result, further study of TD is
indicated. The data also suggest that the stabilization phase of TD is essential to engagement in
the intervention and for the consequent reduction of DSO symptoms in particular; therefore
further study of TD may offer an opportunity to use the intervention as an exemplar through
which the general characteristics, effect, and value of the stabilization phase of component-based
treatments for CPTSD may be examined.
Keywords: Complex Posttraumatic Stress Disorder (CPTSD); Drama Therapy; Heart-Rate
Variability (HRV); Interpretive Phenomenological Analysis (IPA); Stress Sequelae Assessment
viii
Table of Contents
Dedication ........................................................................................................................................ i
Acknowledgments .......................................................................................................................... ii
Abstract .......................................................................................................................................... vi
List of Illustrations ...................................................................................................................... xvii
List of Appendixes ..................................................................................................................... xviii
Chapter I: Introduction .................................................................................................................... 1
Organization of the Mixed-Methods Dissertation .......................................................... 7
Chapter II: A Review of the Literature ......................................................................................... 10
Epidemiology of Complex Stress Exposure in Childhood ........................................... 10
Relevance to Social Work ............................................................................................. 15
Theoretical Background: A Selective History .............................................................. 16
Attribution and Framing of the Condition ................................................................ 16
Early Complex Trauma Practitioner-Theorists ......................................................... 16
War ............................................................................................................................ 18
Social Systems .......................................................................................................... 19
Attachment Theory ................................................................................................... 20
Physiology: The Nervous System ............................................................................. 21
Theoretical Framing of TD Study ................................................................................. 23
High Frequency Heart Rate Variability (HF HRV) .................................................. 25
ix
The Ventromedial Prefrontal Cortex (vmPFC)..................................................... 28
Research Doman Criteria (RDoC) ............................................................................ 29
CPTSD Classification ............................................................................................... 30
Dynamic-System Theory .......................................................................................... 31
Subtypes/Phenotype Studies ................................................................................. 32
TD Treatment ................................................................................................................ 33
Drama ........................................................................................................................ 33
TD: Description of Intervention................................................................................ 33
Trauma-Drama Relevant Outcomes ..................................................................... 36
What the TD Study Seeks to Contribute ....................................................................... 37
Chapter III: Quantitative Methods, Results, and Discussion ........................................................ 39
Quantitative Problem Statement ................................................................................... 39
Hypotheses .................................................................................................................... 39
Methods......................................................................................................................... 39
Participants ................................................................................................................ 39
Demographics ....................................................................................................... 39
Trauma History ..................................................................................................... 40
Background Information ....................................................................................... 41
Recruitment and Inclusion/Exclusion Criteria ...................................................... 43
Protection of Human Subjects .............................................................................. 45
x
Data Collection ......................................................................................................... 47
Procedure .............................................................................................................. 47
Questionnaires....................................................................................................... 47
Assessments Completed by the Participant ...................................................... 48
Assessments Completed by Clinician ............................................................... 51
Assessments Completed by a Teacher or Primary Caregiver ........................... 53
Data Cleaning for Questionnaires ..................................................................... 54
Electrophysiological Data ..................................................................................... 55
Results ........................................................................................................................... 58
Questionnaire Data.................................................................................................... 58
Hypothesis I-A .......................................................................................................... 58
Pattern Analysis in Hypothesis I-A....................................................................... 63
Pattern I. Decreasing Trend in Treatment-Group Symptoms ........................... 63
Pattern II. Increasing Trend in Treatment-Group Symptoms ........................... 64
Pattern III. Curvilinear Convex Treatment-Group Trend ................................. 65
Pattern IV. Curvilinear Concave Treatment-Group Trend ............................... 65
Summary of Findings for Hypothesis I-A ............................................................ 66
Hypothesis I-B .......................................................................................................... 66
Summary of Findings for Hypothesis I-B ............................................................. 67
Hypothesis II ............................................................................................................. 67
xi
Exploratory Analysis of HRV Data .......................................................................... 68
Demographic Differences in Subtypes ................................................................. 73
Exploratory Analysis of Questionnaire Data ............................................................ 74
CDI 2 ..................................................................................................................... 74
BRIEF T and ADI ................................................................................................. 75
Discussion ..................................................................................................................... 77
Results ....................................................................................................................... 77
Hypothesis I-A ...................................................................................................... 77
Hypothesis I-B ...................................................................................................... 79
Hypothesis II ......................................................................................................... 79
Exploratory Analysis ............................................................................................ 80
Limitations and Implications ........................................................................................ 88
Limitations ................................................................................................................ 88
Implications............................................................................................................... 90
Chapter IV: Qualitative Methodology, Findings, and Discussion ................................................ 92
Methodology: Interpretative Phenomenological Analysis............................................ 92
Phenomenology......................................................................................................... 93
Hermeneutics ............................................................................................................ 93
Idiography ................................................................................................................. 95
Differentiation of the IPA Phenomenological Method ............................................. 96
xii
Replicability, Flexibility, and Creativity as Related to this Study ............................ 96
Replicability .......................................................................................................... 96
Flexibility .............................................................................................................. 98
Creativity............................................................................................................... 98
Triangulation of Informants within Qualitative Data ............................................. 101
Research Problem ....................................................................................................... 101
Researcher Description ............................................................................................... 102
Participants .................................................................................................................. 103
Protection of Human Subjects ................................................................................ 103
Sample Characteristics ............................................................................................ 104
Parent-Organization-Employee Informants ............................................................ 104
Procedure .................................................................................................................... 105
Interview Questions ................................................................................................ 106
Analysis and Criteria for Selecting Observations ................................................... 107
Results ......................................................................................................................... 109
Superordinate Theme: Absence to Presence ........................................................... 110
Structure of the Intervention ................................................................................... 110
Three-Stage Pattern in the Flow of the Overall Twenty-two Sessions ............... 113
Stage I ............................................................................................................. 113
Stage II ............................................................................................................ 113
xiii
Stage III ........................................................................................................... 114
Three-Stage Pattern in the Flow of the Discrete Session .................................... 114
Green-Room Activities: Procedure and Purpose ................................................ 116
On Stage Activities: Procedure and Purpose ...................................................... 118
Dressing Room Activities: Procedure and Purpose ............................................ 120
Folkways & Mores .................................................................................................. 121
Equal Participation .............................................................................................. 124
Self-Determined Boundaries ............................................................................... 126
Behavioral Self Disclosure ................................................................................. 127
Identification via Deepened BSD ................................................................... 130
Navigation ........................................................................................................... 132
Fun/Play .............................................................................................................. 135
Energy ............................................................................................................. 137
Group-Sourced Problem Solving and Meaning Making .................................... 138
Pushing and Testing Limits ................................................................................ 140
MAP (The Primary Change Process) ...................................................................... 143
Presence and Its Five Domains ............................................................................... 145
MAP and IMAJN During an Improvisational Exercise ...................................... 147
MAP Stage I: Green Room ................................................................................. 147
MAP Stage II: On Stage ..................................................................................... 147
xiv
MAP Stage III: Dressing Room .......................................................................... 148
MAP and IMAJN During Scene Work ............................................................... 150
MAP Stage I: Green Room ................................................................................. 151
MAP Stage II: On Stage ..................................................................................... 151
MAP Stage III: Dressing Room .......................................................................... 152
Population of Associations x-Time ........................................................................ 153
Shift in Sense of Identity......................................................................................... 155
Advocacy ............................................................................................................ 156
Example Cases with Applied Terminology ............................................................ 158
Jill’s Sense of Identity Shift ................................................................................ 159
I
MAJN ........................................................................................................ 163
PAxT ............................................................................................................... 164
Pablo’s Sense of Identity Shift: I Am a Non-Violent Person ............................. 165
I
MAJN ........................................................................................................ 168
PAxT ............................................................................................................... 168
Bond’s Sense of Identity Shift ............................................................................ 169
I
MAJN ........................................................................................................ 172
PAxT ............................................................................................................... 174
Indigo’s Sense of Identity Shift .......................................................................... 174
Indigo’s Observed Behavior ........................................................................... 175
xv
I
MAJN ........................................................................................................ 177
PAxT ............................................................................................................... 178
Exceptions ............................................................................................................... 178
Discussion ................................................................................................................... 181
Introduction and Overview of Notable Findings .................................................... 181
Stage I: Green Room Stabilization ......................................................................... 182
First Injuries First ................................................................................................ 183
Impaired-Caregiving System Repair: The MAP of Identification .................. 184
Therapeutic Alliance ........................................................................................... 185
BSD and the Authentically Open Therapeutic Stance .................................... 185
The Role of Egalitarianism in BSD ................................................................ 186
MAP Process in Improvisational Warm-Ups and Games................................... 190
Loss of Self-Consciousness through Irresistible Play ..................................... 191
Blends and Compressions ............................................................................... 192
Relation of Emotional Regulation and Executive Function to DSO .............. 193
Experientially Described Essences as Related to Recognized Variables ....... 195
Is Phase I Really Necessary? The Surprising Importance of Green Room .... 196
Stage II, On Stage ................................................................................................... 200
That Old Familiar Feeling and Diversifying the Reactions That It Elicits ......... 204
Dissociation via Aggression and Checking-Out ............................................. 205
xvi
HF HRV and Executive Function ........................................................................... 207
Limitations, Recommendations, and Implications...................................................... 210
Suggested Instrumentation Based on Qualitative Results .......................................... 213
CPTSD Measure ..................................................................................................... 214
Attachment Schemas Measure ................................................................................ 215
Evidence-Based Relationship Measure ................................................................... 215
Chapter V: Mixed-Method SynthesisLimitations, Recommendations, and Implications ...... 216
Missing Data ............................................................................................................... 216
Trauma History and Demographic Data ................................................................. 216
Administration of Assessments................................................................................... 217
Cross-Disciplinary Collaboration ............................................................................... 218
Further Recommendations for Instrumentation .......................................................... 219
Index of Nervous System Balance: HF HRV ......................................................... 219
Executive Function, Dissociation, and Depression Measures ................................ 220
Conclusion .................................................................................................................. 222
References ................................................................................................................................... 224
xvii
List of Illustrations
Illustration A ..................................................................................................................................62
Illustration B ..................................................................................................................................71
Illustration C ..................................................................................................................................71
Illustration D ..................................................................................................................................76
Illustration E...................................................................................................................................87
Illustration F .................................................................................................................................113
Illustration G ................................................................................................................................115
Illustration H ................................................................................................................................159
xviii
List of Appendixes
Appendix A ..................................................................................................................................249
Appendix B ..................................................................................................................................256
Appendix C ..................................................................................................................................263
Appendix D ..................................................................................................................................289
1
Chapter I: Introduction
For thousands of years, drama has been used to integrate traumatized people naturally
into society and to heal stressful experiences (Boyd, 2018; Shay, 1995). In the 21st century,
however, the treatment of trauma has become professionalized and often focuses on cognitive
behavioral protocols for posttraumatic stress that directly target specific or time-limited incidents
(e.g., an automobile accident or combat exposure in adulthood), with the goal of rapidly restoring
affected people to a prior state of well-functioning (Friedman et al., 2007; Rauch & Foa, 2006).
Yet, for people who have experienced more pervasively stressful events in early childhood—
such as chronic abuse and neglectstressful experience most commonly does not have a
beginning and an ending (Ford & Courtois, 2020; van der Kolk, 2015). It is often not
characterized by discrete events that can later be used as treatment targets. In addition,
chronically stressed people cannot be returned to a baseline of well-functioning if it was never
established in the first place.
Early life stress that is pervasive, continual, and cumulative is referred to as complex
stress (Ford & Courtois, 2020), and when it contaminates a person’s experience of life it is
referred to as complex trauma (Cloitre & Beck, 2017; Herman, 1992b; Shonkoff et al., 2009;
Spinazzola et al., 2005; Spinazzola et al., 2013). Complex trauma most often occurs within the
context of impaired caregiving systems. It triggers persistent employment of survival adaptations
that deplete physiological resources; it alters the developmental course; and it interferes with
flexible and well-calibrated responses to challenges in the environment (Karatsoreos & McEwen,
2013; Teicher & Samson, 2016; van der Kolk, 2005). The survival-adapted body habitually
deploys crisis-calibrated fight/flight/freeze defenses (Karatsoreos & McEwen, 2013). Confused
evaluations of relative threat versus safety make it impossible for people to trust fully in their
2
own judgment (Matsakis, 1998). Trauma-infused perceptual impairments are viewed as akin to
“driving while drunk” (Rosenberg, 2021, personal communication, May 12, 2021). Affected
individuals learn to mistrust self, others, and the world. They develop a negative sense of self
and become isolated and marginalized from engagement in society (Cloitre, 2021; D’Andrea et
al., 2012; Ford & Courtois, 2020; Herman, 1992b; Janet, 1925/1976; Matsakis, 1998; Spinazzola
& Briere, 2020; van der Kolk, 2010).
Early-life stress has a dose-related association with over forty types of physiological,
behavioral, and psychological sequelae (e.g., cardiovascular disease, cancer, addiction, and
depression), which develop across the lifespan (Anda et al., 2006; Bellis et al., 2019; Briere et
al., 2008; Cloitre & Beck, 2017; Finkelhor et al., 2007; Follette et al., 1996; Fox et al., 2015;
Spinazzola & Briere, 2020). The highest doses of early adverse experiences are associated with
life being cut short by two decades (Brown et al., 2009). The Adverse Childhood Experiences
(ACE) studies (Anda et al., 2006; Bellis et al., 2019; Felitti et al., 1998; Giano et al., 2020) have
demonstrated that early-life stress has impacted over 60% of the U.S. population (Centers for
Disease Control and Prevention [CDC], 2019). ACEs exact a tremendous toll that is paid for in
human suffering, in lives cut short, and in dollars. The cost of ACEs, calculated thorough the
lens of childhood abuse and neglect in the U.S., is estimated to be $2 trillion dollars per year
(Peterson et al., 2018).
Over the past century and a half, expert clinicians who have treated people with complex
trauma have recommended phased treatment (sometimes called component-based treatment), the
first phase of which concentrates on stabilization (Cloitre et al., 2012; Ford & Courtois, 2020;
Herman, 1992b; Hopper et al., 2019; Janet, 1925/1976). The stabilization phase includes
establishing a trusting and safe alliance with the therapist and instilling emotional-regulation
3
skills and other emotional protections. For people who have developed within the context of
inadequate caregiving systems and who have never had a stable foundation, the stabilization
phase revisits and fills gaps in the developmental architecture needed to support the trauma
processing done in the second treatment phase (Perry, 2009).
Despite longstanding clinical recommendations in its favor, the need for phased
intervention in the treatment of complex trauma is the focus of heated debate within the trauma
field (see Brewin, 2020). Proponents of phased treatment argue that an initial stabilization phase
is critical (Cloitre et al., 2012; Cloitre et al., 2010; Courtois, 2021; Herman, 1992b; ISTSS,
2019); however, proponents of focal-trauma treatments—designed for post-traumatic stress
disorder (PTSD)argue that there is little systematically collected clinical trial data to support
the superiority of phased complex trauma treatment. They maintain that trauma-focused
cognitive behavioral treatments—which bypass stabilization and directly and rapidly expose
people to trauma memorieshave a strong evidence base for treatment of the time-limited
traumatic incidents characteristic of PTSD (de Jongh et al., 2016; Ehring et al., 2014; Resick et
al., 2012) and should therefore be used for complex trauma until there is further evidence to
support phased treatment. Focal treatment proponents argue that to withhold evidence-based
PTSD treatments from chronically traumatized people is potentially to prolong their suffering
through the weeks of a stabilization phase that may be ineffective and unnecessary. In contrast,
proponents of phase-based complex trauma treatments argue that complex trauma is not merely a
more severe form of PTSDit is different in character. Phase-based treatment proponents
maintain that focal treatments will at best remain inert and will at worst be destabilizing and re-
traumatizing, since focal treatment omits the stabilizing installation of the relational and
4
emotional-regulation scaffolding required to initiate, contain, and metabolize complex trauma
processing (Brewin, 2020; Cloitre et al., 2017).
It is indeed the case that there is scant systematic clinical trial research examining phased
treatment for complex trauma (Karatzias et al., 2019) because the complex post-traumatic stress
disorder (CPTSD) classification was not legitimized as an official diagnosis until the 2018
edition of the International Classification of Diseases for Mortality and Morbidity Statistics:
11th Revision (ICD 11; World Health Organization). Therefore, until three years ago, there was
no official complex trauma diagnosis around which to organize a shared and unified research
agenda. In contrast, the PTSD diagnosis (American Psychiatric Association [APA], 1980) was
formulated and legitimized four decades ago, in the wake of the Vietnam War. Forty years of
PTSD research have allowed for a body of evidence to accumulate in support of direct and focal
approaches to PTSD (Foa et al., 2009; Friedman et al., 2007). In short, although there is little
systematically acquired evidence that phased treatments are better than focal treatments for
treating complex trauma, there is a similar lack of evidence that they are not. There is a pressing
need for systematic studies on phased treatments for CPTSD (Karatzias et al., 2019).
Complicating the study of CPTSD treatment is the fact that people who are affected are
often differentially impacted depending upon a variety of factors, including individual
physiology and the timing and dose of exposure (including length of time exposed and number
of different types of exposures; Finkelhor et al., 2007; Spinazzola et al., 2014; Teicher &
Samson, 2016). There is some evidence to indicate that once symptoms develop, they compound,
intensify, and combine to produce additional symptoms (Haselgruber et al., 2021). Complex
trauma produces a host of heterogeneous diagnoses and biopsychosocial sequelae, depending
upon developmental trajectories and physical maturation from infancy through the phases of
5
adulthood (Briggs et al., 2013; Cross et al., 2017) and other factors (e.g. genetics; Adams et al.,
2016; Franklin et al., 2010; Hodgdon et al., 2019; Koenen et al., 2009; Rutter et al., 2006;
Shonkoff et al., 2009; van der Kolk, 2010).
The CPTSD classification emerged subsequent to the introduction of a new research
framework by the National Institute of Mental Health (NIMH; Insel et al., 2010; Insel, 2014),
namely, the Research Domain Criteria (RDoC). The RDoC expanded the prevailing research
paradigm by promoting analysis of complex relationships and of convergences and divergences
among dimensional biopsychosocial variables. The RDoC assumes that symptomatology occurs
along a continuum of wellness to pathology, that phenotypes (observable characteristics that
emerge as a result of gene/environment interactions may moderate symptom profiles and
treatment responses, and that common substrates underpin heterogeneous sequelae (for a fuller
description see Cuthbert, 2020). The goal behind the RDoC paradigm has been to support
identification and delivery of precision treatment to individuals (Gordon, 2020a, 2020b).
The current study of Trauma Drama (TD) was conducted in light of the RDoC framework
and during the time of debate over the necessity of phased interventions for complex-trauma. TD
is a phased and manualized group intervention (Spinazzola, 2019) that synthesizes a naturalistic,
improvisational theatre form with the expert-consensus treatment guidelines for CPTSD (first
published in 2012 by Cloitre et al.). TD is an evidence-informed and promising practice model
adapted from the Urban Improv theatre-based secondary youth violence prevention model
(Kisiel et al., 2006; Zucker et al., 2010) recognized by the Office of Juvenile Justice and
Delinquency Prevention in its Model Program Guide (U.S. Department of Justice [DOJ], 2006).
The potential strength of an improvisational drama intervention is (arguably) its ability to
interact differentially with the unique characteristics of the individual in a complex and dynamic
6
manner. In theory, TD has the potential to seek, find, and address traumatic memory
differentially and can address pre-verbal attachment wounds and non-verbal traumatic-memories
through safe projection of unconscious and conscious experience onto the metaphor of the
improvisation (Blair, 2009; Boyd, 2018; Cook, 2009; Johnson, 1982; Spolin, 1963; Stanislavski,
1934). The potential strength of drama therapy in conferring personalized effects is also its
weakness, and until now treatments that employ drama have defied the type of systematic study
necessary to encourage wider dissemination and implementation. In line with the RDoC
framework, the current study suggests that if TD is effective, there will be variables on different
dimensions that will show convergent trends—and therefore will reveal general effects for the
entire treatment groupand/or there will be other variables that show divergent trends as
moderated by phenotypes or subtypes that are present within the treatment group (in this
dissertation I use the words subtypes and phenotypes interchangeably).
It would be hubris to proceed on the assumption that the differential impact of a
naturalistic art form like drama could be comprehensively reduced, but through triangulated
examination of TD, a matrix of variables that is capable of confirming effectiveness may emerge.
Consequently, the goal of this mixed-method study is to triangulate psychological questionnaire
data, electrophysiological data, and phenomenological interview data in order to begin the work
of identifying a parsimonious matrix of variables capable of capturing evidence of dynamic-
system perturbations and shifts to a higher order (Garland et al., 2010; Pascual-Leone, 2009). It
is also the goal of this study to identify variables that may simplify future field research and
support the data aggregation across settings that may enable statistical modeling. Such modeling
may help to show whether, how, and for whom TD is effective.
7
The TD study is an embedded, combined, concurrent, and sequential mixed-method
investigation of a complex-trauma-exposed population of 47 youth in residential treatment in the
northeastern U.S. (treatment condition [n = 27] and of a semi-matched comparison condition
[n = 20]). Participant demographics were as follows: aged 14 to 22 years; 61.7% female, 38.3%
male; average of 5 trauma types; 79.4% experienced their first trauma before age 1; average of
2.51 diagnoses; average of 2.54 medications; and average of 3 years behind in school (see
Appendix C, Table C1). Participants were divided into two groups, the first group being a TD
treatment group of those who attended Trauma Drama adjunctive treatment groups over the
course of a school year, and the second group being a comparison group of semi-matched
participants who experienced treatment as usual and adjunctive treatments that were not TD.
Organization of the Mixed-Methods Dissertation
This dissertation has three data strands: (a) secondary analysis of quantitative data from
questionnaires; (b) primary analysis of electrophysiological heart rate variability data; and (c)
primary analysis of qualitative interview data. In the past, creative-arts interventions have
resisted methodical study. This triangulation of methods is intended to serve as a beginning step
in the identification of a simplified matrix of variables for future study of this complex arts
intervention in complex populations (Creswell & Zhang, 2009). Triangulation of methods and
informants is also meant to compensate for a small quantitative sample size and the roughness of
the secondary analysis of the parent-organization-provided clinical questionnaire-data (originally
used as continuous quality control data).
Triangulated exploration considers the following research questions:
What are the phenomena of TD as experienced by participants?
8
What are electrophysiological and/or survey instrument variables (if any) that appear
to indicate treatment effect across the treatment group?
Is there evidence of increased biopsychosocial flexibility?
If so, what are the indicator variables?
Is there any evidence to indicate that phenotypical subtypes are embedded within the
participant groups?
If so, are there any differences in treatment response by phenotypical subtypes
(as indicated by quantitative variables)?
Based on the results of this exploratory study, what are recommendations for future
study and for further honing of a matrix of variables for a cross-setting aggregation of
data and for data analysis?
Chapter I introduces the dissertation. Chapter II includes a general literature review to
contextualize the overall dissertation within relevant theory and research. Chapter III presents
quantitative methods, results, and discussion as well as limitations, implications, and
recommendations that are specific to the quantitative data analysis. Sample characteristics and
human-subject considerations (for the entire dissertation) are contained in the quantitative
methodology section. Chapter IV presents qualitative interview methods, results, and discussion
as well as limitations, implications, and recommendations that are specific to the qualitative data
analysis. As is typical of qualitative writing, additional contextualizing literature is interwoven
with the qualitative discussion (Smith, 2008). Because of the complexity of the methods and
discussions, instead of repeating material I have provided hyperlinks1 within this document to
1 To follow a hyperlink, click on the blue text. To return to the previous location, use Alt+LeftArrow (hold
down the Alt key and press the Left Arrow key).
9
provide ease of access to revisited concepts. (For further clarity, Appendix A contains a list of
terms.) Chapter V provides limitations, implications, recommendations, and conclusions
synthesized from the three data strands.
10
Chapter II: A Review of the Literature
Resilience is the ability of an organism to withstand threats to stability in the
environment. In a sense, resilience represents the ability to bend without breaking in the
face of environmental or psychological perturbations.
Karatsoreos & McEwen, 2013
Since 1980, when the diagnosis of post-traumatic stress disorder (PTSD) was first
introduced in the third edition of the Diagnostic and Statistical Manual of Mental Disorders
(APA, 1980), a vast, diverse, and multi-dimensional body of biological, psychological, and
social research has been generated. An exhaustive review of the trauma literature would consume
volumes. In the following review of the literature, I strive to include a selection of the literature
that is limited but sufficient to situate the Trauma Drama (TD) study in the current context of
what has become known as complex trauma (Cloitre et al., 2019; Herman, 1992b). Additional
relevant literature on complex trauma is further woven into the discussion chapter of the
qualitative portion of the study, which is consistent with the norms of qualitative research
(Smith, 2008).
Epidemiology of Complex Stress Exposure in Childhood
A dose-related association exists between stress exposure during childhood and the
breadth and severity of sequelae that develop across the lifespan (Anda et al., 2006; Briere et al.,
2008; Cloitre & Beck, 2017; Finkelhor et al., 2007; Follette et al., 1996; Spinazzola & Briere,
2020). Injurious stress that usually begins in childhood has become known as complex trauma
(Cloitre et al., 2019; Cook et al., 2003; Cook et al., 2007; Herman, 1992b; ISTSS, 2018, 2019;
Spinazzola et al., 2021). Complex trauma may be described as interpersonal injury that
compounds over time, is potentially irreparable, and may result in deformations of identity,
11
including the capacity to integrate identity and experience (Ford, 2017; Ford & Courtois, 2020).
Complex trauma has an omnipresent quality that erodes sense of safety with intimates and
interferes with the ability to form the trusting relationships that are requisite for thriving in
society (Ford, 2017; Ford & Courtois, 2020). Complex-trauma exposure provokes
psychophysiological adaptations that enhance chances of survival during periods of inescapable
threat (McEwen & Lasley, 2004), but those survival-calibrated adaptations become a liability
when using crisis-fixated fight, flight, or freeze defenses to navigate safer environments
(Wamser-Nanney & Vandenberg, 2013). Sustained activation of stress-response physiology
alters the functioning of maturing organ systems (including the brain) and the nervous system
(Shonkoff et al., 2009) and negatively affects long-term psychophysiological wellbeing,
morbidity, and mortality (Felitti et al., 1998; Ford & Courtois, 2020; Institute of Medicine &
National Research Council, 2000; Porges, 1995; Sapolsky, 2004a; Shonkoff et al., 2009; Teicher
& Samson, 2016).
Since 1995, the Adverse Childhood Experiences (ACE) studies have systematically
assessed large, general-population adult cohorts for level of exposure to common childhood
stressors in order to understand cumulative long-term impacts of early life stress (Felitti et al.,
1998). Although other collectives of researchers have utilized somewhat different frames, the
ACE studies provide a broad-based (although not entirely complete) and contextualizing
snapshot of the effects of early life stress. The first ACE study enrolled 17,337 health
maintenance organization members who had completed a standard medical evaluation (Brown et
al., 2009; Felitti et al., 1998). ACEs may be described as stressful and potentially traumatic
events occurring between birth and the age of 17 that may interfere with bonding to others, sense
of safety, and sense of stability (Felitti et al., 1998). ACEs encompass events such as
12
maltreatment and neglect, witnessing of violence at home, parental mental illness, parental
divorce, and/or substance abuse, etc. (CDC, 2019). ACEs are common across all demographics.
Emotional abuse is the most commonly reported ACE (Merrick et al., 2018, 2019). The feature
of disorganized attachment to caregivers, which frequently accompanies emotional abuse and/or
neglect in early childhood, has been robustly related to indicators of psychopathology and to
dissociative symptoms in adolescence and adulthood (Spinazzola et al., 2021; Sroufe et al.,
2010).
Over half of prior ACE-study participants have reported experiencing at least one ACE.
One quarter of participants have reported exposure to more than three categories of ACEs, and
nearly one in six participants has reported exposure to four or more ACEs (Felitti et al., 1998;
Giano et al., 2020; Merrick et al., 2018, 2019). An exposure-gradient was found to exist between
the number of categories of ACEs and the severity of psychopathology, physical illness, and
early death. Other studies that also examine chronic stress in early life (but do not employ an
ACE frame) have similarly shown that the types of trauma, number of traumas, combinations of
traumas, and developmental timing of traumas compound, which means that that higher levels of
trauma exposure are associated with more complex symptom profiles and more severe outcomes
over the lifespan (Adams et al., 2016; Andersen et al., 2008; Briere et al., 2008; Finkelhor et al.,
2007; Teicher, 2018; Teicher & Samson, 2016). A recent network-analysis study by Haselgruber
et al. (2021) posits that it is the existence of chronic trauma symptoms themselves (rather than
total trauma load) that causes trauma symptoms to worsen and branch into additional mental
health syndromes over time.
A systematic review and meta-analysis of quantitative ACE studies (inspired by the
original ACE studies) from North America and Europe between 1990 and 2019 concluded that
13
30% of cases of anxiety and 45% of cases of depression in North America and 25% of anxiety
and depression cases in Europe were attributable to ACEs (Bellis et al., 2019; Merrick et al.,
2019). As compared with people who had experienced no ACEs, people who had experienced
four or more ACEs had “4- to 12-fold increased health risks for alcoholism, drug abuse, and
suicide attempts” (Felitti et al., 1998, p. 45), and people who had experienced six or more
categories of ACEs died an average of 20 years before people who had experienced no ACEs
(Brown et al., 2009). More than 40 outcomes have been associated with ACEs, including
ischemic heart disease, cancer, chronic lung disease, immune system problems (e.g., asthma and
arthritis), inflammation, metabolic abnormalities, and liver disease (CDC, 2019; Danese et al.,
2009; Felitti et al., 1998; Waehrer et al., 2020). ACEs are also associated with increased criminal
and violent behavior (Duke et al., 2010; Fox et al., 2015), as well as higher rates of alcohol and
drug abuse and risky sexual behavior (Dube, Anda, et al., 2002; Dube, Felitti et al., 2003; Hillis
et al., 2001; Pilowsky et al., 2009).
Certain groups are at higher risk for childhood stress than others, including people in
marginalized groups (e.g., people of color and LGBTQIA+) and those who lack a high-school
diploma or are under- or unemployed (Giano et al., 2020; Merrick et al., 2018). Lower socio-
economic status confers a high stress burden (Marmot et al., 1984; Sapolsky, 2004b, Syme &
Berkman, 1976; Wilkinson, 2000). Ancestral stress (also known as intergenerationally
transmitted trauma) is associated with mental health and physiological sequelae (Jovanovic et al.,
2011; Sotero, 2006; Yehuda & Lehrner, 2018). Certain combinations of stressors have been
associated with particular impacts upon children; for instance, exposure to violence in the family
and/or community, combined with impaired caregiving, confers a risk for compromised self and
relational functioning and for developmental issues (Spinazzola et al., 2018). Males and females
14
display distinctly different profiles of childhood adversity, and female gender is associated with a
more complex and varied history of exposure to adverse experiences (Haahr-Pedersen et al.,
2020). The type, timing, and compounding of stress during childhood development differentially
impacts the physical development of the brain and consequent sequelae. (For a review, see
Teicher & Samson, 2016.)
The estimated total economic burden of complex stress in the United States, as calculated
through the frame of childhood abuse and neglect, is $2 trillion dollars per year (Peterson et al.,
2018). An estimated $105 billion could be saved annually from a 10% reduction in ACE
prevalence (Bellis et al., 2019; Merrick et al., 2019). In summary, in the United States and
throughout the global community, complex stress in childhood appears to be dose-related, is
common, and leads to a wide array of negative psychological and physical outcomes that exact
severe tolls on individuals, society, and the economy.
Although I have used the term “ACE” to capture recent essential research on negative
stress during development and its sequelae, there is no one common term used in the extensive
literature of recent decades. Other terms (with associated variations in meaning) include (but are
not limited to) Early Life Stress (Lanius, Vermetten, et al., 2010), Developmental Trauma
Disorder (van der Kolk et al., 2009), and Complex Trauma (Herman, 1992a). As mentioned
previously, Complex PTSD (CPTSD) was legitimized as an official diagnostic classification in
2018 in the ICD 11. Prior to 2018, there was no official complex trauma diagnosis around which
to organize a shared research agenda (Karatzias & Cloitre, 2019). As of this writing, the CPTSD
formulation is emerging as the centerpiece of a coordinated research agenda. Previously, there
have been numerous separate streams of research occurring across multiple domains. The
streams of particular relevance to this paper began to emerge in the second half of the nineteenth
15
century, at which time scientist-researchers and clinical theorists began to wrestle with
empirically derived definitions of the constituents of aversive stress and with identification of
which people were affected, why they were affected, and when and how they were affected.
Relevance to Social Work
The current study of TD synthesizes concepts that derive from different but convergent
streams of the multi-disciplinary literature on stress that provide the theoretical dimensionality
necessary for a dynamic-system model. The circular-causality/dynamic-system model that is
employed is in line with the literature and with the complexity recognized by the Social Work
profession (Garland et al., 2010; Panksepp & Biven, 2012; Stiglbauer et al., 2013). Social Work
is a dimensional discipline dedicated to drawing together elements that intersect and interact to
enhance the complex biopsychosocial system. The practice of social work recognizes complex
systems and processes (International Federation of Social Workers [IFSW], 2014) and the need
for complex analysis. Social Work employs a strengths-based approach to intervention (IFSW,
2014). The Trauma Drama (TD) intervention itself is a holistic, nonpathologizing, strengths-
based approach to working with people in dynamic interaction with their environments.
The incidence and frequency of adverse childhood experiences have remained unchanged
since at least the early 1900s (Dube, Felitti, et al., 2003, p. 267). The mental health field has a
century-and-a-half-long history (at least) of episodic attuning to and then abandoning of the issue
of adverse childhood stress (van der Kolk, 2007). As an explicitly tenacious change-agent
profession, Social Work is a good fit for holding fast and long to a complex and difficult research
agenda that is aimed toward liberating the vulnerable and neglected with the intended result of
leveling the path to enhanced wellbeing and social inclusion (IFSW, 2014). As a practical
16
discipline, Social Work is concerned with facilitating changes that matter. This study looks for
intervention-facilitated changes that matter for the human condition.
Theoretical Background: A Selective History
Attribution and Framing of the Condition
Depending upon the ascendant theorists of a particular historical epoch, traumatized
people might be either blamed or held blameless for their symptoms (or both, in differing
proportions). They might be viewed as constitutionally weak, e.g., as cowardly (Moran, 1945,
cited in van der Kolk, 2007), as innocently injured by events that would have adversely affected
almost anyone (Tedeschi & Calhoun, 2004), and/or as having a predisposition or an inherited
genetic defect that rendered them particularly vulnerable to stress. In addition, attribution of
traumatization has cycled from the perception of having been caused by a single event or a
cluster of events during a discrete period in adulthood (like war, automobile accident exposure,
or rape) to having been caused by the cumulative effect of chronic stress, potentially of multiple
types, usually beginning in childhood (such as ongoing child abuse and neglect). In short, the
lens of particular historical epochs has framed perception of the injured person as responsible or
not responsible for injury and impacts, and symptoms as attributable to a simple cause requiring
a simple treatment or to a complex cause requiring a complex treatment. (For a thorough
summary, see Herman, 1992b and van der Kolk, 2007.)
Early Complex Trauma Practitioner-Theorists
French psychiatrist Jean-Martin Charcot (1825–1893) proposed that exposure to
adversity produced the psycho/physiological symptoms of hysteria (Charcot, 1887, cited in van
der Kolk, 2007). As influenced by fellow countryman Charcot, Pierre Janet (1859–1947), also a
psychiatrist who treated hysterics, theorized that stressful events generated vehement emotions
17
that fractured normal memory in such a way that original traumas were blocked from
consciousness. Janet observed that forgotten memories haunted sufferers and were manifested
through their repeated re-experiencing of intrusive and unresolvable fragments of the traumatic
past. Without the ability to integrate traumatic memory, traumatized people remained fixated in
an ever-present past that effectively halted emotional development and conferred an inability to
assimilate new experiences (described in van der Hart et al., 1989).
In contrast to Janet, who believed that hysteria developed through adverse events as
combined with an inherited pre-disposition, Sigmund Freud (1856–1939) suggested that
intrusive emotions and dissociation (i.e., aspects of hysteria) were instead solely the consequence
of traumatic experience (Sanfelippo & Dagfal, 2020). Yet, like Janet, Freud believed that
physical and emotional manifestations of forgotten trauma continued in repetition because the
traumatic experience itself had never been resolved (Sanfelippo & Dagfal, 2020; van der Kolk,
2007). As influenced by Charles Darwin’s (1808–1882) work on the universality of emotional
expression across species (Darwin, 1872; Ekman, 2009), Freud proposed that traumatic
emotional re-enactments provided traceable clues (via the universal language of the physical
expressions conveyed) that could lead to the actual origins of the unresolved trauma (Breuer &
Freud, 1895/1957). Freud imagined that the processing of trauma was much like the mourning
process, because the trauma sufferer, like the mourner, needed to work methodically through
fragmented memory and reconcile it with the present in order to conclude the traumatic
experience (Freud, 1917; Sanfelippo & Dagfal, 2020).
Now, in the 21st century, the field has circled back to Freud’s and Janets early
formulations. Janet, in concert with his clinical formulation of traumatization, recommended that
treatment proceed in three stages. His stages closely correspond to the three-phase treatment
18
model that is currently recommended for complex trauma (Cloitre et al., 2012; Herman, 1992b).
Van der Hart et al. (1989, p. 3) describe Janets (1925/1976) three-stage treatment model as
follows:
1. Stabilization, symptom-oriented treatment, and preparation for liquidation of
traumatic memories
2. Identification, exploration, and modification of traumatic memories
3. Relapse prevention, relief of residual symptomatology, personality reintegration, and
rehabilitation
In a well-known example of the way in which the trauma field has oscillated between
validating and then invalidating interpersonal injury as the cause of intractable traumatization,
Freud eventually abandoned the contention that his patients’ symptoms were caused by external
traumas and instead proposed that symptoms were generated by drives that originated within the
person themselves. Van der Kolk (2007, p. 34) quotes Freud’s rejection of interpersonal trauma
as cause and his eventual decision to disbelieve his patients’ stories:
I believed these stories (of childhood sexual trauma) and consequently supposed that I
had discovered the roots of the subsequent neurosis in these experiences of sexual
seduction in childhood. If the reader feels inclined to shake his head at my credulity, I
cannot altogether blame him. . . . I was at last obliged to recognize that these scenes of
seduction had never taken place, and that they were only fantasies which my patients had
made up.
Over the early and mid-twentieth century, Freud’s drive theory ascended in dominance, and
Charcots and Janet’s theories of interpersonal injuries-as-cause receded (van der Kolk, 2007).
War
In decades surrounding the turn of the 20th century, another relatively independent
stream of theory and clinical practice was developing in relation to the trauma of war. As with
trauma treatment in the private sector, trends of particular historical epochs influenced clinicians
and theorists examining the outcomes of battle experience. Alternately, post-combat stress
sequelae were attributed either to the overwhelming experience of war or to an inherent physical
19
and or mental weakness within the person (or sometimes a mixture of both). In the United States
after the Civil War, traumatic combat sequelae were attributed to irritable heart and nostalgia
(Hyams, 1996). During World War I, symptoms were attributed to shell shock and trench
neurosis. In World War II, symptoms were believed to arise from effort syndrome and battle
fatigue. In 1941, Abram Kardiner (1891–1981), who had previously treated World War I
soldiers, published The Traumatic Neurosis of War, in which he described a war-related stress
syndrome that included avoidance of memories (dissociation), intrusions and flashbacks, and
somatization. As such, Kardiner’s theory had much in common with Janet’s earlier theories and
with the framing of the post-traumatic stress disorder diagnosis (PTSD) as it exists today (van
der Kolk, 2007). After the Vietnam War, trauma symptoms were attributed to the new diagnosis
of PTSD (APA, 1980; Hyams, 1996), which, for the first time, legitimized PTSD as an officially
recognized categorical diagnosis. The PTSD diagnosis acknowledged that individuals had been
injured by events that would have overwhelmed most people’s capacity to cope, thereby no
longer “othering” those affected and thus rendering them deserving of treatment and support
(van der Kolk, 2007; Yehuda et al., 2015).
Social Systems
The discipline of Social Work arose in the United States at the end of the nineteenth
century. In contrast to approaches that emphasized single types of interpersonal traumas or
intrapersonal predispositions as etiological causes of symptomology, the progressive settlement
house movement attributed psychological disturbance to social forces, such as multiple points of
oppression and the socio-economic gradient (Addams, 1910; Gross, 2009; Hamington, 2019).
This included the impact of the society in synthesis with the biological, psychological, and social
systems of people. The first settlement house in the United States, Hull House, was opened in
20
1889 in Chicago by Jane Addams (1860–1935) and associates. Addams and the Hull House
workers embodied a humanistic approach centered on empathy, egalitarianism, and connection.
They collaborated with community members to enrich the experience of life by resolving
individual, community, and educational inequities, and to advance society through joint activism
and social justice. Hull House, while embedded in the community, provided practical supports,
including housing, education, child care, and participation in recreational and artistic
enrichments (Addams, 1910). The work at Hull House launched the Social Work profession in
the United States (Gross, 2009).
Attachment Theory
John Bowlby (1907–1990), an English psychiatrist educated in Freuds psychoanalytic
drive theory, was not able to reconcile the mismatch he found between that theory and the needs
and experiences of the children whom he observed and treated. Bowlby believed that
“[psycho]analysts, in their preoccupation with a child’s fantasy life, were paying too little
attention to actual events in the childs real life” (Ainsworth & Bowlby, 1991, p. 333). In the
aftermath of World War II, Bowlby published a World Health Organization (WHO) report on
children who had lost their families as a result of the war, entitled Maternal Care and Mental
Health (1951). Bowlby concluded that children, when physically and/or emotionally separated
from their caregivers, manifested mourning and grief. He attributed the grief reaction to
caregiver unavailability and inadequate caregiving systems, which he believed to be at the root
of children’s symptoms (rather than symptoms being attributable to internal drives). He proposed
that, in the service of building a person’s capacity for secure attachment, the therapist should
construct a trustworthy relationship with the “patient” and guide that person toward the
exploration of childhood relationships with caregivers, in order that the internalized model of
21
attachment could be revised to lessen attachment anxiety and to gradually enable improved
interpersonal relationships in the present. In the early 1950s, Bowlby and Canadian psychologist
and researcher Mary Ainsworth (1913–1999) joined forces to develop contemporary attachment
theory.
Physiology: The Nervous System
Another stream of traumatic stress theory developed around the physiology of stress.
Claude Bernard (1813–1878) and Walter Bradford Cannon (1871–1945) pioneered research on
the balance of the internal physiology, including the nervous system. Bernard considered a well-
regulated, flexible, and balanced internal milieu (glycemic function, blood flow and temperature
regulation, influence of the vagus nerve on heart rate, etc.) essential for living a free and
independent life (Bernard, 1879; Wehrwein et al., 2016). He theorized that the higher the order
of the living being, the more complicated the coordinated adjustment of various vital systems to
keep the body functioning within a stable zone during periods of external stress (e.g., adverse
traumatic experiences). Cannon coined the term “homeostasis” to describe the complex
coordination of the body’s internal material supplies (e.g., sodium, insulin, etc.) and
physiological processes (e.g., the action of the autonomic nervous system) in order to maintain a
stable balance and steady state in the face of the ever-changing external environment (Cannon,
1929). Cannon suggested that homeostasis could be maintained as a result of organized
government of the “open system such as our bodies represent, compounded of unstable material
and subjected continually to disturbing conditions” and that “constancy is in itself evidence that
agencies are acting, or ready to act, to maintain this constancy” (p. 424).
Hans Selye (1956) elaborated on the body’s necessary shift away from homeostasis
when under the stress of acute challenge. He observed the resetting of the body from the
22
regenerating (anabolic) functions that preserved the steady state to the systemically expensive
crisis-accommodating catabolic functions (a breakdown of metabolic compounds in the service
of producing rapid energy) that allowed the organism to respond rapidly to changing/emergent
situations. In addition to analyzing the acute-crisis response, Selye examined the systemic cost of
more prolonged stress that was of lesser magnitude but that was chronically present. He
theorized that chronic stress would require a sustained use of the same survival-calibrated
physiological-balance resources as had been episodically employed to manage acute challenges.
Like Seyle, Sterling and Eyer (1988) sought to understand the sustained uses of survival
resources to manage chronic stress. In the clinic they had observed a vast difference in cardiac
aging of patients whose systems had been taxed over time by stress. They expanded the concept
of homeostasis by introducing the term allostasis to describe the body’s fluctuating employment
of alternative set-points to balance and serve the level-of-demand and speed-of-response required
in more emergent circumstances. Sterling and Eyer noted that, in addition to contributing to
cardiac problems, chronic employment of allostasis caused people to misread even safe
situations through the lens of crisis, which exacted a toll on relationships and mental health. The
language and understanding of homeostasis and allostasis were further elaborated by McEwen &
Lasley (2004) with the term allostatic overload, which describes the perpetuation of the
expensive, resource-expending, crisis-oriented state over the long term. When experiencing
allostatic overload, the body is habitually engaged in crisis defenses. ACEs and other taxing
social constructs such as oppression are believed to be at the root of the depleting nature of
chronic stress that triggers allostatic overload. In more recent years, Sapolsky (2004a, 2004b) has
demonstrated in animal studies the ways in which lower social rank (e.g., social-order
oppression) acts as a form of chronic stress that negatively affects morbidity and mortality of
23
subordinate animals. The Whitehall Study, a long-term prospective study of British social
servants, also found that lower rank had a negative impact on morbidity and mortality (Marmot
et al., 1984; Marmot & Shipley, 1996; North et al., 1996).
Theoretical Framing of TD Study
Allostatic Overload as Connected to HRV. Allostatic overload is theorized to alter the
organism along epigenetic, chromosomal, cellular, structural, and psychological dimensions
(Herman et al., 2016; Karl et al., 2006; Koenen et al., 2009; Karatsoreos & McEwen, 2013;
Kaufman et al., 2004; Liston et al., 2009; Sapolsky, 1996; Shonkoff et al., 2009, Lasley, 2002).
The organism may become chronically depleted and consequently lose the flexibility necessary
to reset to restorative homeostasis. Progressive degeneration initiates a cascade of pathological
processes (Shonkoff et al., 2009). Allostatic overload has been connected to accelerated
reduction of telomere length, the DNA protein caps of chromosomes that are considered a proxy
for cellular aging (Price et al., 2013), and is theorized to cause changes in gene methylation
(epigenetic changes) that can reduce stress-response and immune-system resiliency across
generations, among other potential epigenetic alterations (Morrison et al., 2019). In theory, gene
methylation may affect not only the traumatized person but also pass the crisis-state settings
forward to offspring via altered genetic coding that is contributed by either parent. Recent
research suggests that effective treatment may be able to switch off epigenetic coding that was
switched on by chronic trauma (Vinkers et al., 2019)
On a functional level, allostatic overload may be reflected in pervasive use of fight,
flight, and/or freeze defense postures. Chronic employment of crisis defenses is believed to
deplete adrenaline and cortisol, two hormones that are required to support systemic flexibility
through regulation of the sympathetic (accelerator) and parasympathetic (braking) branches of
24
the autonomic nervous system and hypothalamic pituitary axis (Cannon, 1929; McEwen &
Lasley, 2004). The ongoing stress response interferes with the calming action of the
parasympathetic branch of the nervous system by disengaging the calming “brake” on the
sympathetic “accelerator” branch of the nervous system. The sympathetic nervous system
becomes mobilized at the expense of suppression of generative secretions such as insulin,
digestive hormones, and reproductive hormones (Sapolsky, 2004b). While sympathetic
activation during acute stress may sharpen quick cognitive assessment of a danger, “chronic
stress disrupts [cognition] and impairs synaptic plasticity as well as the birth of new neurons,
atrophies dendritic processes in neurons, and increases the incidence of neuron death” (Sapolsky,
2004, p. 395). According to Sterling & Eyer (1988), allostatic arousal affects metabolism and the
function of almost all regulatory systems and chemicals in the body and brain, and “rather than
simply local feedbacks, is a far more complex form of regulation than homeostasis” (p. 637).
The rapid speed with which the parasympathetic nervous system operates appears to
indicate that any rapid changes in heartbeat are more generally facilitated by the parasympathetic
nervous system rather than the sympathetic nervous system. A release of acetylcholine at the
neuro-muscular junction mediates parasympathetic activity. Acetycholinesterase quickly
degrades the acetylcholine in the extracellular compartment (Talman & Kelkar, 1993), so the
parasympathetic system is relatively rapid in its action. Alternatively, the sympathetic nervous
system releases adrenalin and slowly engages in its reuptake. Parasympathetic activity has seven
times more control over the timing of the heartbeat than does sympathetic activity (Berntson et
al., 1997). The parasympathetic versus sympathetic balance of the hypothalamic pituitary axis
can be reflected in cardiac vagal tone (Cannon, 1929; Porges, 2007; Selye, 1956). Vagal tone is
described as the inhibitory influence of the parasympathetic nervous system on the sympathetic
25
nervous system and the heart (Pickens & Field, 1995). Schore (2018, p. 349) states, “In other
words, the energy-expending sympathetic and energy-conserving parasympathetic components
of the autonomic nervous system (ANS) regulate the autonomic, somatic aspects of not only
stress responses but also emotional states.” Vagal tone may be assessed by measuring the power
of the high-frequency range of electrophysiological measures of heart rate variability (HF HRV),
which is the electrophysiological power-spectrum measure of amplitude/voltage that is used in
this study to index systemic flexibility vs. inflexibility over the course of the intervention.
High Frequency Heart Rate Variability (HF HRV)
In this study, the use of the physiological measure of HF HRV is viewed as an
operationalized index of pivotal aspects of the fear-emotion circuitry (Porges, 1995, 2007),
including flexible “regulation of physiological, affective, and cognitive processes” (Thayer &
Lane, 2009, p. 81). HF HRV is a continuous variable and is an indicator of systemic balance. The
HF HRV (.5 to .4 hertz) primarily reflects parasympathetic/vagal activity (Thurber et al., 2008).
The parasympathetic branch of the autonomic nervous system (PNS) modulates the arousal of
the sympathetic branch of the nervous system (SNS). Even though HF HRV was specified as the
electrophysiological variable of interest for this study, the HeartMath HRV equipment used for
the TD study (EmWave Pro Desktop, n.d.) captured all HRV frequency bands at once, so all
frequencies were examined for potential information of relevance (Thurber et al., 2008). Prior
studies (Aupperle et al., 2012; Beauchaine et al., 2007; Beauchaine & Thayer, 2015; Blase et al.,
2021; Blood, 2015; Choi & Jeon, 2020; Colzato et al., 2018; Forte et al., 2019; Hartmann et al.,
2019; Jandackova et al., 2016; Kemp et al., 2012; Koenig et al., 2016; Kogan et al., 2014; Kolacz
et al., 2020; Ottaviani et al., 2018; Porges, 2007; Pyne et al., 2016; Rasmusson & Abdallah,
2015; Shaffer & Ginsberg, 2017; Thayer et al., 2012) have linked HF HRV to constructs that
26
were evaluated via psychometric instrumentation in the TD study, including body awareness,
alexithymia, affect regulation, behavioral regulation, cognition, mood, socialization, and
executive function. Although perhaps sounding counter-intuitive, increased/higher levels of heart
rate variability (HRV) have been conceptualized as indicative of a healthier, more balanced,
more flexible, and more resilient system, and decreased/lower levels of HF HRV as indicative of
a less healthy, less balanced, less flexible, and less resilient system (McEwen & Lasley, 2004;
Porges, 2007).
Lower HF HRV has consistently been associated with depression (Choi & Jeon, 2020;
Hartmann et al., 2019; Kuang et al., 2017; Licht et al., 2008; Shinba, 2014), including in
adolescents (Blood, 2015; Koenig et al., 2016). Increases in HF HRV have been associated with
improvement in depression symptoms (Chambers & Allen, 2002; Koenig et al., 2018). HF HRV
has been used to describe the physiology associated with chronic stress exposure and sequelae
(Campbell et al., 2019; Herzog et al., 2018; Rabellino et al., 2017; Schneider & Schwerdtfeger,
2020), yet with the exception of small and pilot studies (e.g., D’Andrea & Pole, 2012), it has not
been used as an outcome measure for trauma-related disorders. At the conclusion of a recent
meta-analysis of potential physiological variables (Yang et al., 2021), the authors suggest that
electrophysiological measures be used to assess traumatization and recovery because such
measures are less limited than psychological questionnaires may be by cultural biases related to
preconceptions of emotional health versus pathology. However, Yang et al. (2021) also observe
that there is not a standard testing paradigm for electrophysiological measures (which is needed),
and that although increases in HF HRV have been associated with abatement of traumatic
sequelae in small or pilot studies, findings have not been so robust as to encourage HF HRV to
be broadly adopted as a measure of therapeutic outcome (see also Campbell et al., 2019; Cohen
27
et al., 1998; Forte et al., 2019; Grupe et al., 2020.; Karstoft et al., 2016; Liddell et al., 2016;
Schneider & Schwerdtfeger, 2020; Thayer et al., 2012).
The work of Lanius, Frewen, et al. (2010) and Lanius, Vermetten, et al. (2010) suggests
that the neurobiological system affected by complex stress may not only chronically under-
modulate but may alternate between under-modulation and over-modulation. A fixation to either
over-controlled modulation (e.g., shutting down or freezing) or under-controlled modulation
(e.g., aroused aggression)—or toggling back and forth between them—is not considered useful
in navigating safe environments. Karatzias et al. (2018) have found both hypo- and hyper-
activation to be characteristic of CPTSD. The findings of Lanius, Frewen, et al. (2010), Lanius,
Vermetten, et al. (2010), and Karatzias et al. (2018) would suggest that the theory related to HF
HRV requires elaboration, since present HRV theory considers increases in the power of HF
HRV to be indicative of treatment success. Yet, it would seem that for those people who are
over-modulated (frozen or disconnected from emotion and feeling), a decrease in
parasympathetic HF HRV dominance would be indicative of healing, because such a decrease
would be expected to allow them to awaken from the frozen state. Complex stress adaptation
appears to narrow the flexibility level of the fear-emotion circuitry, which affects the stress
response and influences the development of complex-stress symptoms/sequelae (Lanius,
Vermetten, et al., 2010; McEwen & Lasley, 2004; Rodrigues et al., 2009; Sherrington, 1906). It
would seem that a flexible balance between sympathetic and parasympathetic systems is optimal
for any HRV typological subgroups embedded in the TD study’s participant population. This
study proposes that successful treatment with TD will diminish reliance on over- and under-
modulating crisis settings and will begin restoring/instilling balance and flexibility to the fear-
emotion circuitry. In turn, increased flexibility would foster a broader range of behaviors that
28
would be optimally calibrated for and adapted to the particular circumstances at hand. This study
concerns itself with locating indicators of increased systemic flexibility and looks at HF HRV as
one of those potential indicators of balance.
This study acknowledges the possibility of heterogeneous physiological phenotypical
subtypes (as indexed by HF HRV) embedded within the population. It was anticipated that
participants could display different longitudinal physiological and symptomatic recovery paths as
moderated by physiological subtype and indexed by HF HRV (Alvares et al., 2016; Freed &
D’Andrea, 2015; Ge et al., 2020; Kemp et al., 2017; Kogan et al., 2014; Mann et al., 2015;
Shaffer & Ginsberg, 2017; Shinba, 2014; Sloan et al., 2017; Williams et al., 2019; Young-
Southward et al., 2020). The mixed methods of the TD study represent an attempt to triangulate
physiological, psychological-instrument, and subjective-participant perspectives in order to begin
the process of homing in on a refined matrix of variables for data aggregation in future study of
TD treatment effectiveness across treatment settings. HF HRV was proposed as a transdiagnostic
marker of pathology versus health and as an index of the flexible function of the organism’s
neurovisceral integration system (Thayer et al., 2021; Thayer & Lane, 2009).
The Ventromedial Prefrontal Cortex (vmPFC). In a meta-analysis of “studies relating
cerebral blood flow to HRV,” Thayer et al. (2012) deduced that there was support “for the idea
that HRV [might] index the degree to which a [medial prefrontal cortex] mPFC-guided core
integration system [would be] integrated with the brainstem nuclei that directly regulate the
heart” (p. 754). Thayer et al. concluded that HRV was important “as a potential marker of stress
and health and provide[d] evidence for the neural correlates that serve[d] to underpin the
relationship” (p. 754). Damasio (2007, p. 66) maintained that the vmPFC “provide[d] a cortical
platform for inputs and outputs that link the lower brain with the higher cortical regions and vice
29
versa.” According to Damasio (2007, p. 66), since the vmPFC acts as “a trigger to emotions or as
a producer of signals capable of influencing the activity of other brain regions, the vmPFC can
exert a wide influence. . . . Given its anatomofunctional position, the vmPFC sector is well suited
for monitoring the connection between decision options and action taken, in addition to the
outcome of the option/action, in factual as well as emotional terms.” In order to make reasoned,
optimal decisions that are oriented to the environment, Damasio (2000) contends that individuals
must feel the implications of decisions and also be able to reason cognitively about the
advantages or disadvantages of a particular course. Traumatized individuals have a great deal of
difficulty understanding emotions and undergo what Krystal (1978) conceptualized as a
dedifferentiation of affects. Studies of people with PTSD have shown functional and structural
abnormalities in the vmPFC (Grupe et al., 2020). In this study, movements toward increased
systemic flexibility, as indicated by balanced HF HRV, are also theoretically indicative of
movement toward improved functional interconnectivity of the vmPFC (Bluhm et al., 2012;
Kaplan et al., 2016; Lanius, Frewen et al., 2010; Shaffer & Ginsberg, 2017; Zelazo &
Cunningham, 2007) .
Research Doman Criteria (RDoC)
The National Institute of Mental Health initiated the Research Doman Criteria (RDoC) in
2009 to “integrate many levels of information (from genomics to self-report) to better understand
the basic dimensions of functioning underlying the full range of human behavior, from normal to
abnormal” within a research framework (National Institute of Mental Health, 2016, p. 2). The
RDoC guidelines recommend uncovering the transdiagnostic markers that underlie common
substrates of heterogeneous diagnoses through a triangulation of biology, behavior, and context.
The current mixed-method TD study was mindfully formulated with RDoC agenda in mind.
30
Creswell & Zhang (2009) observed that quantitative studies have dominated the trauma field,
and they have recommended that study design be diversified to include qualitative data within a
mixed-method model. They state that mixed-method studies are well suited to the field of
trauma, and should therefore be systematized, standardized, and pursued.
CPTSD Classification
At the time that this study was proposed and the data gathered (2013 to 2015), the
CPTSD diagnosis had not yet been legitimized. Since its appearance in the 2018 ICD 11, the
CPTSD construct has been tested in more than forty studies (including studies of adolescents and
children) conducted in over fifteen different countries (Cloitre, 2020, 2021), and CPTSD has
been shown to be distinct from PTSD. The CPTSD classification has provided the platform for a
coordinated research agenda across disciplines and allowed for the development of standardized
CPTSD assessments (International Trauma Questionnaire—Child and Adolescent (ITQ-CA,
n.d.) and treatment recommendations (Cloitre et al., 2012; Cloitre, 2021; Hopper et al., 2019).
The CPTSD classification has two factors: PTSD and Disturbances of Self Organization
(DSO). Each factor organizes three symptom clusters. The PTSD factor encompasses the clusters
of Re-experiencing, Avoidance, and Sense of Threat. The DSO item encompasses the three
symptom clusters of Emotion Regulation, Negative Self-Concept, and Interpersonal Problems
(Cloitre, 2020). Negative Self-Concept has been shown to be a central feature of CPTSD
(Cloitre, 2020). The results of this study were examined in the context of CPTSD, and
synthesized recommendations for future research are provided with the RDoC guidelines of
identifying transdiagnostic and common substrates of heterogeneous diagnoses through a
triangulation of biology, behavior, and the CPTSD classification.
31
Dynamic-System Theory
TDs impact on the condition of fear-emotion circuitry may be best understood if viewed
through the lens of a dynamic system model (Baines, 2008). In theory, if the dynamic systems of
study participants were effectively stimulated by TD, their neurovisceral fear-emotion circuitry
would self-organize to a more optimal level of flexibility. “Broadly defined, self-organization
refers to the emergence of novel patterns or structures, the appearance of new levels of
integration and organization in existing structures, and the spontaneous transition from states of
lower-order to states of higher-order” (Lewis, 2005, p. 173). Chaos/complexity theory asserts
that gentle, very small perturbations of a dynamic system may generate very large changes in
systemic organization: the butterfly effect (Bishop, 2008). A dynamic-system model was a good
theoretical fit for this study of TD (Lewis, 2011; Smith & Thelen, 1993), since there could be
countless interacting and compounding elements emerging from the intervention itself and from
the group dynamic. It was anticipated that elements might selectively and idiosyncratically
perturb the body/mind/psychology of the individual study participants, which would thereby
make it impossible to determine which element was the potent agent in instigating healing.
Different elements might have varying levels of potency for different people.
After a review of the literature on the neuroscience of emotional circuitry (affective
neuroscience), Garland et al. (2010) found scientific support for a dynamic-system “upward
spiral” of instilling/restoring/building positive emotional states and enduring positive traits. In
theory, positive emotions cause an upward spiral of ever-broadening attention that expands the
person’s “thought-action repertoire” (2010, p. 860) to a degree of flexibility and a flourishing
that utilizes all levels of the body and mind (the learning brain). Positive factors may “exert
actions on the brain by feeding back upon it to modify the expression of genes and thus the
32
function of nerve cells” (Kandel, 1998, p. 460). Because of Trauma Drama’s structure and
repetition of themes and activities, it was anticipated that the intervention might incorporate
reiteration in the service of producing upward spirals that could recontextualize life in a more
positive and empowering frame (Lakoff & Johnson, 1980). Complex circular systems might not
be reducible to component parts. They might not have a solution (Sreenath, 2008). That is one
reason the TD study sought to identify a matrix of variables with the potential for successful
indexing of pivotal aspects at the substrate level that initiate an upward spiral of flexibility in the
dynamic fear-emotion circuitry. For instance, heart rate variability might impact the fear-emotion
circuitry, which might impact the activity of the medial prefrontal cortex, which might impact
emotional regulation, which might impact executive function, which might impact depression
symptoms, which might impact negative sense of self, which might impact interpersonal
relationships.
Subtypes/Phenotype Studies. It would seem necessary to assume some heterogeneity
within samples of chronically trauma-exposed youth and to identify ways to filter for treatment
effects by phenotype or subgroup. Ford (2020) has been able to identify six trauma-profile
subtypes through latent class analysis (LCA) of a nationally representative group of adolescents.
Each trauma-profile subtype has distinct characteristics. (Two subgroups had primarily been
exposed to single-event trauma or had witnessed violence, and four subgroups had experienced
polyvictimization, that is, multiple personally victimizing events.) Adams et al. (2016) also used
LCA to identify trauma-exposure classes within a diverse clinical sample of 3,487 adolescents
within the National Child Traumatic Stress Network (NCTSN) and found five classes, four of
which were characterized by polyvictimized youth. In latent profile analysis (LPA) of Danish
treatment-seeking, combat-exposed veterans, Karstoft et al. (2016) identified four profiles related
33
to different levels of symptomatology, the least symptomatic of which exhibited unexpectedly
low vagal tone when compared to veterans who were more symptomatic. Although this TD
sample is small, it seemed prudent to look for evidence of phenotype that might moderate
longitudinal patterns related to treatment effect. Although this study was unable to draw
conclusions about phenotype, it begins the process of identifying candidate variables for
potential classification of phenotypes to be pursued when larger samples and/or aggregated data
sets become available.
TD Treatment
Drama
Since the evolution of modern humans, drama has been used to develop resiliency and to
heal the wounds of stressful experience (Boyd, 2001; McConachie, 2008, 2011; Shay, 1995).
Consequently, it is a natural vehicle with which to synthesize a trauma-informed intervention. In
Africa and Europe, research has been conducted on the use of school-based drama interventions
for early- and middle-school students. Drama interventions have been focused on increasing
children’s resiliency and on preventing and ameliorating the impacts of early life stress. Studies
of drama interventions have been found to decrease relational, verbal, and physical aggression, to
increase a sense of wellbeing (Ager et al., 2011), and to increase social and emotional skills that
include a better understanding of bullying and diversity (Joronen et al., 2011).
TD: Description of Intervention
Although it has been previously studied for effectiveness in violence-prevention, TD was
(and is) formulated as a “trauma-informed” intervention. The traumatic-stress field recognizes
trauma-informed treatment as explicit in its recognition of clientshistories of trauma, their
related symptoms, and the ways in which traumatic experiences and symptoms have affected
34
their lives (Bartlett et al., 2016; Hodgdon et al., 2013; Peterson et al., 2018). TD is built on a
naturalistic theatre model. It incorporates elements of the vast multi-disciplinary stress literature
into its drama format (Spinazzola, 2019). The eighteen- to twenty-four-session intervention
employs a phased trauma-treatment structure to (a) instill/create safety and build appropriate
trust, (b) learn and practice stress-coping skills and social skills, (c) work through
stressors/traumas previously experienced, (d) make personal meaning of experiences and connect
to the social environment, (e) obtain transcendent post-traumatic growth, (f) achieve self-
actualization and a hopeful outlook, and (g) instill resiliency to future stress. TD seeks to build
mind/body infrastructure and function by bolstering attachment, interoception, and executive
function. By focusing on core competencies that are often derailed by adverse childhood
experiences (Anda et al., 2010) that have resulted in stress sequelae, TD instills and/or reinforces
abilities in the areas of attunement, assessment of trust, safe attachment, interoception (somatic
awareness), self-regulation, identification of emotions, affect/emotion regulation, decision-
making, engagement of executive function under stress, conflict management and pro-social
behaviors, assertiveness, effective-interaction and bonding skills, self-awareness, self-
acceptance, and self-esteem (Kisiel et al., 2006; Spinazzola, 2019; Tishelman et al., 2010; van
der Kolk et al., 2009).
Detailed fictional-scenario outlines have been designed and written, which cover topics
that allow participants to rehearse alternative behaviors and feelings elicited by common stresses
and challenges faced by adolescents (for instance, peer pressure, prejudice, racism, homophobia,
and harassment). Troupe members then improvise dialogue and behaviors. The practice of new
coping mechanisms and alternative behaviors during improvisational scenarios that portray
domestic violence, dating violence, peer violence (e.g., gang violence, community violence),
35
and/or familial trauma (e.g., neglect and maltreatment) may help participants metabolize and
learn from difficult experience. It can also aid participants who experience depression, grief, and
loss by increasing interpersonal connection and by encouraging future orientation toward
individual goals and plans (Spinazzola, 2019). TD’s eighteen to twenty-four sessions are divided
into three major phases: (1) Safety and Skills Building, (2a) Skills Application: Youth Conflict,
(2b) Skills Application: Youth Trauma, and (3) Skills Generalization (Spinazzola, 2019). The
sessions themselves have common elements: an opening circle, therapeutic exercises, scenario-
driven theatre improvisations, post-scene discussion, and a closing circle. The “opening circle”
poses an introductory “question of the day” that assists participants in warming up to and
anticipating the day’s theme. Therapeutic exercises develop core competencies like
emotional/physiological regulation, self-expression, and social risk-taking. Cooperation is
emphasized over competition. One of the repeated exercises, “Simon Says” (not all exercises are
repeated), is meant to develop cortical reasoning under pressure (executive functioning), impulse
control, and attunement. Facilitators increase the difficulty and complexity of Simon Says as the
weeks progress and as the participants develop greater mastery (Spinazzola, 2019).
Scenario-scripted improvisational scenes become more trauma-focused toward the
middle phase of TD. For instance, during improvisations of the middle phase, troupe members
may act out a scenario in which a character is depressed or has experienced a familial trauma. At
pivotal points in the scene, the troupe members freeze the action and the participants brainstorm
potential actions that the main character and those around him/her might take. Troupe members
and participants alternate in taking over the roles of key characters and play out options that are
proposed by the group. At the close of the scenario, there is a post-scene discussion designed to
process the substance of the scene-work, to problem-solve, to consolidate learning, and to wrap
36
up the emotional work of the session. TD troupe members are trained to be mindful of
identifying any participants who require additional emotional support during or after the session
(Spinazzola, 2019). The scenes are not taken from the real lives of participants but are fictional
depictions of typical interactions experienced by youth who have experienced complex stress.
Scenarios are carefully designed to seem real but are written in a mindful and modulated manner
that helps participants to tolerate, manage, and metabolize the material.
Beginning sessions of the TD intervention are weighted toward tasks that relate to
development of the lower and mid-sections of the brain (involved in sensation, emotion, self-
regulation, attachment, etc.), and such tasks are integrated and reiterated throughout the weeks of
the intervention (even though in later weeks participants also work on the development of
higher-order cortical functions; Spinazzola, 2019). The process of acting within the safe holding
space of the fictional scene aims to lead participants to a conscious toleration of feeling and
knowing in the present. Participants are taught to turn toward, rather than away from,
emotionally evocative material contained in the safe field of play, and they are taught to
recognize and regulate differentiated emotions and affects (Spinazzola, 2019). Common rituals
and exercises of skill that are repeated in the session structure are designed to build upon one
another. Skills and themes are reiterated over the course of the school year. The
physical/emotional space provided by TD strives to be safe and predictable. The same eight
facilitators (four clinicians and four milieu staff) return to engage participants in the ritual of the
drama. TD joins self-agency and safety with the experience of social relationships.
Trauma-Drama Relevant Outcomes. In line with violence-prevention goals of the CDC
(David-Ferdon et. al., 2016) and DOJ (U.S. Department of Justice, 2006), two previous studies
of Urban Improv (UI; the predecessor of TD) examined behavioral outcomes related to youth
37
violence in urban school populations exposed to community violence. The prior UI studies found
a halting of aggression in the UI group versus controls and an increase in pro-social behaviors
that are identified as mitigators of youth aggression (Kisiel et al., 2006; Zucker et al., 2010). The
current study of TD in a complex-trauma exposed clinical population, while not unconcerned
with aggression abatement, is more concerned with substrates that indicate perturbation of the
dynamic system of the person, rebalancing toward greater flexibility, and resumption of
developmental and maturational processes (a transdiagnostic approach that, at least in the short
term, is diagnostically agnostic). Troupe members and participants involved in past TD
intervention groups reported that the intervention had a positive effect on the development of an
expanded pro-social-behavior repertoire and a sense of self, including self-esteem, increased
confidence, sense of self-efficacy, and personal discovery of talents, abilities, and strengths (D.
Beckstead, personal communication, March 11, 2011). As one teacher whose students
participated in TD put it, “Kids who would normally be aggressive or swear in confrontations
tried out new behaviors(K. Demasi, personal communication, August 14, 2009).
What the TD Study Seeks to Contribute
Focal-incident interventions (interventions meant to address a single type and/or time-
limited trauma) have been relatively well researched and disseminated, but interventions meant
to address the progressive cascade of sequelae that flow from early life stress (a far more
widespread problem) are not well researched because of barriers to the quantification of results
(D’Andrea et al., 2012; Ford & Courtois, 2009; Ford, 2011, van der Kolk et al., 2009). Recent
advances in neuropsychophysiological studies suggest that the cascade of heterogeneous
sequelae that emerge from chronic stress can be catalyzed by processes connected into the
dynamic fear-emotion circuitry (Shonkoff et al., 2009). In theory, an intervention that is capable
38
of selectively interacting with the dynamic fear-emotion circuitry of people as they
idiosyncratically develop over time cannot be reduced to component parts (because of its
circularity), so analysis is a challenge. It is likely that other effective interventions for treating
early life/developmental stress currently exist, but it is necessary to find a way to build an
evidence base sufficient to encourage their dissemination. This study seeks to uncover, to frame,
and to understand evidence of the helpfulness of a trauma-informed creative-arts intervention in
establishing a more flexible fear-emotion circuitry. Mixed methods (phenomenological,
electrophysiological, and psychological questionnaire data) are triangulated to begin the work of
identifying a manageable matrix of testable variables that may be used for future aggregation of
participant data across settings and subsequent assessment of intervention effectiveness in
particular populations and subpopulations.
39
Chapter III: Quantitative Methods, Results, and Discussion
Quantitative Problem Statement
Does TD intervention increase systemic flexibility in chronic-trauma exposed youth in
residential care? Are there clues to the identification of a core set of variables that may be used in
future research to investigate TDs impact on systemic flexibility?
Hypotheses
Based on a review of the literature, the following hypotheses were tested:
1. Hypothesis I predicted that participation in Trauma Drama would be associated with
an increase in the flexibility of the fear-emotion circuitry.
2. Hypothesis II predicted that participation in Trauma Drama would be associated with
a halting of aggression as indexed by a decrease in aggression and social-problem
symptoms.
Methods
Participants
Demographics. Participants were 47 complex-trauma exposed youth living in seven
residential treatment facilities operated by a large nonprofit organization dedicated to social
justice and delivery of services in the New England area of the United States. (I will refer to this
organization as the parent organization.) Demographic data related to the participants were
collected by the parent organization upon participants’ admission into their respective residences
and were entered into the parent organization’s Client Assessment Tracking System (CATS)
database, which housed continuous quality control and assessment data. Descriptive statistics on
demographics are presented in Table C1. Thirty-four participants (72.3%) resided in four
residential schools run by the parent organization. Thirteen participants (27.7%) resided in three
40
group homes run by the parent organization. Of the 47 participants, 27 participants (57.4%) were
members of the treatment group and 20 were members the comparison group (42.6%).
Participants ranged in age from 14 years to 22 years, with the average age being 17.8 years.
Twenty-nine (61.7%) participants were female, and 18 (38.3%) were male.
Trauma History. I was granted access to the de-identified and coded demographics,
medication records, and trauma histories for 38 of the 47 participants (although there were some
missing data in the available records). Descriptive statistics on trauma history are presented in
Table C1. The following trauma history characteristics apply to the available records of the total
participant sample. On average, each participant had experienced 4.9 types of trauma (f = 34)
over an average of 6.1 total years of life (f = 34). Most participants experienced more than one
type of trauma at a time.
There is no standardized way to quantify cumulative trauma load, and so I devised my
own rough calculation to describe cumulative trauma load. I subtotaled the years of occurrence
assigned to each type of trauma experienced by each participant and then added the subtotals to
get a total that represented cumulative trauma load. For instance, if a participant experienced
school violence from the age of 8 through the age of 12 (a subtotal of 4 years), and also
experienced sexual abuse from the age of 8 through the age of 12 (another subtotal of 4 years),
their cumulative trauma load would equal 8 years. This calculation yielded an average
cumulative trauma load of 14.6 years (f = 34) per participant. Twenty-six participants (79.4% of
the sample with accessible data) had experienced their first trauma before the age of 1; the mean
age for first exposure was 1.8 years of age (f = 34). I felt it was important to devise a rough
indicator of trauma load, since a vast literature has demonstrated the dose-dependent association
between severity of sequelae and the number of and chronicity of childhood adverse stress types:
41
higher numbers of stress type and higher levels of chronicity are associated in a dose-dependent
fashion with worse morbidity and mortality outcomes. In addition, higher levels of post-
traumatic stress and co-occurring depression lead to the poorest mental-health treatment
outcomes (Anda et al., 2006; Barawi, et al., 2020; Briere et al., 2008; Cloitre & Beck, 2017;
Finkelhor et al., 2007; Follette et al., 1996; Spinazzola & Briere, 2020). The cumulative trauma
load calculation may have particular merit if different levels of cumulative trauma load are
associated with any phenotypical subtypes identified by this study. In some cases, cumulative
trauma load is greater than a participant’s chronological age; however, the cumulative trauma
load is meant to provide a rough calculation for distinguishing trauma exposure among
participants in this study.
Of those participants with available data, 48.6% were suspected2 to have experienced
sexual trauma (f = 18). Physical abuse and/or physical maltreatment was suspected for 51.4% of
participants (f = 19). Emotional abuse was suspected to have been experienced by at least 69.4%
(f = 25), and neglect was suspected to have been experienced by at least 75.7% of participants
(f = 28). Fifty percent had been exposed to an impaired caregiver (f = 17). Of the 31 participants
who had been assigned a global assessment of functioning score (APA, 2000), the average score
was 46.2, implying “serious symptoms” (APA, 2000, p. 34).
Background Information. Study participants had academic difficulty. (Descriptive
statistics on background information are presented in Table C1.) The 23 participants with
available school data were, on average, 3 years behind the grade level corresponding to their
2 “Suspected” was used to describe an event that the clinician-evaluator judged as a likely occurrence, yet sources of
information might have disagreed. For instance, a caregiver might have reported sexual abuse of a child, but the
child might have denied the abuse had occurred (Hodgdon et al., 2019). Despite lack of corroboration between
reporters, the event was still presumed likely and, therefore, was “suspected.”
42
chronological age. Of those 23 participants, 63.9% had documented learning disabilities. Of the
participants with data, (a) at least eight (21.1%) were adopted, (b) 10 (26.3% ) were in custody of
child protective services, (c) 18 (48.6%) had experienced between one and 15 different foster
settings, (d) 14 (36.8%) had a history of substance abuse, (e) 21 (63.6%) had a history of suicidal
ideation, (f) nine (25.7%) had made between one and 10 suicide attempts, and (g) 17 (44.7%)
had a history of self-injury. Of the 29 participants for whom racial demographics were available,
one was African American (3.4%), one was Asian (3.4%), 20 were white (69%), and five were
bi- or multiracial (17.2%). Of the 29 participants for whom ethnicity was identified, six of the
participants (20.7%) identified as Latino/Hispanic and 23 (79.3%) as not Latino/Hispanic.
Of the 29 participants for whom medication records were available, 100% were
prescribed up to six psychotropic medications. The mean number of medications taken by each
participant was 2.54. Of the 29 participants with medication records, nine (32.2%) were on
antidepressants, 14 (50%) were on mood stabilizers, 14 (50%) were on antipsychotics, one
(3.6%) was on a benzodiazepine, 11 (39.3%) were on blood pressure medications, four (14.3%)
were on an antihistamine, three (10.7%) were on sleep medications, and 10 (35.7%) were on a
stimulant for ADHD.
Of the thirty-five participants for whom mental-health diagnosis was available (74.5%)
had a history of prior mental-health residential placement, with an average of 2.51 diagnoses.
Twenty-one participants (60%) had been diagnosed with mood disorder/depressive disorder, 13
(37.1%) with PTSD, 11 (31.4%) with reactive attachment disorder (RAD), two (5.7%) with
dissociative disorder, one (2.9%) with an eating disorder, five (14.3%) with bipolar disorder, two
(5.7%) with a psychotic disorder, two (5.7%) with being on the autism spectrum, eight (29.9%)
with generalized anxiety, and nine (25.7%) with attention deficit hyperactivity disorder (ADHD).
43
Data related to critical incidents (e.g., any reported violence) and other relevant data (e.g.,
changes to medication and/or diagnoses) were regularly updated while participants lived in the
parent organization’s residences. The treatment group and comparison groups were roughly
equivalent in terms of demographic information, trauma history background, and medication
records (see Table C1).
Recruitment and Inclusion/Exclusion Criteria. The research participants were not randomized
into this study; rather, participants comprised a convenience sample. Following review by both the
Widener University Institutional Review Board (IRB) and the parent organization IRB, the on-site
clinical supervisors at the seven residential facilities participating in TD were provided with the
inclusion/exclusion criteria for the study. Potential treatment-group participants were drawn from among
the resident youth who were already planning to attend the TD intervention. Potential participants’
clinicians determined whether they met the study criteria and asked them whether, in addition to
attending the TD treatment, they would like to join the TD study. (HRV-protocol-related preparation
material provided to participants may be found in Appendix B.) Clinicians explained the study to
participants as it was described on the consent form.
The non-randomized comparison-group was composed of youth who: (a) were the same
age and general description as treatment-group participants, (b) resided in the same schools
and/or group homes as the treatment-group participants, (c) were identified through the parent
organization’s CATS database, and (d) were willing to contribute de-identified existing
questionnaire data for secondary analysis by me (the CATS database held demographic and
treatment information and tracked client progress on an array of domains related to stress
sequelae). After identification through the CATS system, potential comparison-group youth
were provided with an invitation to be part of the study if their respective clinicians considered
44
them competent to provide consent/assent and if they expressed willingness to participate.
Participants 18 years-of-age and older provided consent; those younger than 18 gave assent with
their legal guardians providing consent on their behalf. Comparison group members were
receiving standard trauma-informed care at their residences and were also receiving adjunctive
treatments that were not TD. In utilizing a non-randomized, comparison group, I wanted to gain
a rough understanding of what TD might have contributed that was different from or additional
to outcomes of standard care in combination with adjunctive treatments that were not TD (as
experienced by the comparison-group participants). Was there something contributed by TD that
was recognizably unique from or additive to standard trauma-informed treatment (as combined
with other adjunctive treatments in the comparison group)? If so, did any related variables
emerge as promising candidates for use in future streamlined study of TD? (I did not measure or
keep record of the adjunctive treatments in which the comparison group participated.) In this
exploratory study, electrophysiological, quantitative questionnaire, and qualitative interview data
were triangulated in order to provide a starting point for identification of a matrix of streamlined
variables suitable for future data aggregation and analysis across treatment settings. In addition
to possibly identifying a useful matrix of variables, triangulation of primary analysis
electrophysiological and phenomenological data with secondary analysis of psychological-
instrument data may partially compensate for the small quantitative sample size and the
roughness of the secondary-analysis questionnaire data.
Criteria for non-randomized treatment-group participant inclusion, in addition to
consent/assent, were that the participant must be proficient in English, enrolled as a resident at
one of the seven included parent-organization facilities, and enrolled in the TD intervention.
Comparison group participants were required to meet these same criteria, except for participation
45
in the TD intervention. Criteria for study exclusion were (a) a lack of consent/assent, (b) a
participant’s decision to opt out of the study, (c) a legal guardian’s decision to remove their
underage child from the study, (d) a decision by the parent organization or the researcher that
participation in the study might cause harm to a participant, (e) lack of English proficiency, or (f)
discharge from the residential setting. Because the TD study was a naturalistic study of a
complex-trauma-exposed clinical population in residential treatment, most participants qualified
for a psychiatric diagnosis and were taking multiple medications; as such, there were no
exclusion criteria related to medication or for psychiatric diagnosis. Regular trauma-informed
treatment was similar for both the comparison and treatment groups. TD was an adjunctive
treatment for the treatment groups; the comparison group received other adjunctive treatments
that the treatment group did not. I did not separately measure or keep track of comparison-group
membersadjunctive treatments for the following reasons: the information was not readily
available to me; the adjunctive treatments in which the comparison participants engaged were
varied; the n involved in each adjunctive treatment would have been very small (e.g., one or two
people); this exploratory study is an attempt to obtain a rough idea of a matrix of relevant
variables that might help to differentially assess the phenomenon of TD in the future. The
study did not attempt to precisely measure TDs outcomes head-to-head against other
particular adjunctive treatments. All treatment group participants attended TD sessions
regularly and therefore received an equivalent dose of the treatment (with the exception of one
young woman who attended only three sessions because she said that she was not interested in
drama and did not identify sufficiently with other participants).
Protection of Human Subjects. In addition to study approvals by the IRBs of Widener
University and the parent organization, the parent organization provided me with necessary
46
letters of support from supervisors of each residence granting permission to recruit and to
conduct the proposed study at their residences. Each potential participant’s clinician supervised
the participant’s completion of the consent/assent form in order to ensure that participants fully
understood the nature of their potential participation and that consent was provided freely and
without coercion.
Potential study participants were supplied with pseudonyms by the parent organization in
order to protect their privacy. In addition to electrophysiological HRV data and
phenomenological interview qualitative data that I collected myself, this study utilizes secondary
analysis of the parent organization’s continuous quality control CATS data (psychological
questionnaires, background information, and demographic data), which had been de-identified
and matched to each participant’s pseudonym. As a condition of the Widener University IRB,
and as a further caution against my learning the real names of individual participants, all consent
forms across the seven residential sites were collected by a parent-organization employee and
shuffled before being sent to me in one combined package. I hired a research assistant who was
also an employee of the parent organization to (a) maintain a list of participant names attached to
pseudonyms (at no time did I have access to the linked identities); (b) acquire, track, receive, and
shuffle consent forms before forwarding them to me; (c) coordinate receipt of CATS data; and
(d) coordinate transfer of de-identified and pseudonym-tagged archival data to me.
As an incentive to complete the study, a $30 gift certificate to a retail store was offered to
participants in the comparison group upon study completion. Participants in the treatment group
were provided with a slightly higher gift certificate of $40 upon study completion, since their
participation required a brief interview that was not required of the comparison group. All
participants were offered a snack of their choosing (e.g., candy or pretzels) when completing
47
each HRV assessment and/or phenomenological interview that I conducted. I self-funded the
incentives.
Data Collection
Procedure. There were two cohorts in the TD group. The data collection with the first
cohort occurred from October of 2013 to May of 2014. Data collection for the second cohort
(different individuals from those in the first cohort) occurred from November of 2014 to June of
2015. CATS archival data were collected by the parent organization between August of 2013 and
December of 2015.
Questionnaires. The CATS comprehensive assessment battery was administered by the
parent organization to clients of the three residential treatment centers and three group homes as
part of their routine continuous quality control mechanism. Clients were evaluated via the web-
based, trauma-informed psychological CATS questionnaire battery at entry into the program,
every three months thereafter, and upon discharge. Questionnaires were presented (with express
permissions from the publishers, except in the case of free open-source measures) on scrolling
screens that showed identical items as the paper versions of those questionnaires and in the same
order, and they were scored by computer using the same raw score, t score, and percentile
algorithms as the paper versions.
Questionnaire data were maintained in the CATS database, which tracked client progress
on an array of domains related to stress sequelae. The battery of eight assessment instruments
evaluated for constructs and domains related to (a) posttraumatic stress symptoms, (b) self-
regulation, (c) behavior and social competencies, (d) depressive symptoms, (e) alexithymia, (f)
somatization, (g) executive function, and (h) types of trauma exposure. Taken together, the
assessments were meant to index the impact of complex developmental stress and recovery from
48
its sequelae (Briere & Spinazzola, 2005). Client demographic, diagnostic, and ongoing
medication information was also maintained in the database.
CATS data were transferred to me by the parent organization in an SPSS database
containing all participants’ CATS assessments, demographics, diagnostic, and medication
information (de-identified, as described above). Quarterly assessments occurred every three
months as counted from that person’s residential-program entry date; as such, none of these fell
at the same time (except by coincidence). I therefore selected the three assessment points for
each participant that were closest to the beginning, middle, and ending of the TD treatment. A
description of the eight psychological questionnaires utilized by the CATS database follows.
Assessments Completed by the Participant.
University of California Posttraumatic Stress Disorder Reaction Index. The
Posttraumatic Stress Disorder Reaction Index (PTSD RI; Elhai et al., 2013; Pynoos et
al., 1998; Steinberg et al., 2004) DSM-IV version is a widely used measurement that
assesses a young person’s reaction (fear, helplessness, horror) to exposure to 26
possible types of trauma during childhood and also assesses for DSM-IV PTSD
diagnostic criteria. Numerous studies assessing instrument validity have found a clear
dose-response relationship with regard to PTSD RI scores (Steinberg et al., 2013).
The PTSD RI is a 48-item, self-report measure in three parts: Part I assesses for
previous trauma exposures and forms, Part II asks for “yes” or “no” endorsement of
symptoms that have been experienced since the event, and Part III inquiries about the
frequency of symptoms as experienced in the prior month, as scored on a 5-point
Likert scale from 0 (none of the time) to 4 (most of the time). The instrument is
tabulated for a total score on a continuous scale. The PTSD RI has shown an internal
49
consistency estimate (Cronbachs α) of .90 and a test-retest reliability estimate of .84
(Roussos et al., 2005). The test has been used across a variety of trauma types, age
ranges, settings, and cultures. Although most of the assessments in the parent
organization’s CATS-database battery are administered at entry into the residential
program, then every three months, and then at discharge, the PTSD RI Part I (which
takes trauma history) is administered once, while Part II and Part III (symptoms and
symptom severity—from Question #14 onward) are administered quarterly and at
discharge.
The Abbreviated Dysregulation Inventory. The Abbreviated Dysregulation
Inventory (ADI; Mezzich et al., 2001) is a self-report measure used to assess
emotional, behavioral, and self-regulatory dysregulation in adolescents. The ADI
is meant to predict “delinquent behavior, aggression, and problems managing
frustration,” including prediction of substance use disorders (Mezzich et al.,
2001). The 30-item ADI is scored on a 4-point Likert scale from 0 (never true) to
3 (always true). There have been no norms established for the ADI. One study
found four factor-loadings on the instrument: Behavior, Emotion/Affect, and two
Cognitive Dysregulation factors (Cruz-Katz et al., 2010).
Marsee and Frick (2007) suggested that the estimated reliability and validity
that they established in their study were an encouraging indication that the ADI
would be a promising measure of adolescent dysregulation. The ADI
emotional/affective scale has been shown to display an internal consistency
estimate (Cronbachs α) of .75 and to be “associated with reactive aggression”
(Marsee, 2008, p. 524) in adolescent girls. Cronbach’s α for the behavioral
50
dysregulation subscale has been shown to be .80, and higher scores on that
subscale have been associated with impulse control and conduct problems
(Pardini et al., 2003).
The Children’s Depression Inventory 2. The Children’s Depression Inventory 2
(CDI 2; Kovacs, n.d.) is a broadly used, normed instrument used to measure the
extent and severity of depressive symptoms over the previous two weeks (state
rather than trait attributes). The instrument is a 28-item, self-report measure
scored on a 3-point Likert scale from 0 (none) to 2 (definite). There are two
higher-order factors on the instrument: Emotional Problems (with two subscales:
Negative Mood/Physical Symptoms and Negative Self-Esteem) and Functional
Problems (with two subscales: Ineffectiveness and Personal Problems). In internal
consistency reliability testing, the CDI 2 has shown coefficient alphas from .67 to
.91 on all scales and subscales for various ages and sex groups (Bae, 2012).
The Body Awareness Measure. The Body Awareness Measure (BAM; Stone,
2011) is a 20-item self-report questionnaire that utilizes a 5-point Likert scale
from 0 (never) to 4 (always) to index bottom-up somatic nervous system function.
The instrument consists of three factors: somatic sensory sensitivity, somatic
problems, and body awareness, with respective internal consistency coefficients
of .61,.87, and .86 (Stone, 2011). In the current study, the Cronbach’s α values
were, over a three-month interval at Time 1, Time 2, and Time 3, respectively,
.61, .70, and .63 for somatic sensory sensitivity, .67, .58, and .73 for somatic
problems, and .80, .81, and .73 for body awareness. The test-retest reliability
estimates from Time 1 to Time 2 and from Time 2 to Time 3 were, respectively,
51
.70 and .61 for somatic sensory sensitivity, .65 and .61 for somatic problems, and
.83 and .83 for body awareness.
Assessments Completed by Clinician.
Trauma History Profile. Unlike other instruments in the assessment battery (with
the exception of Section I of the PTSD RI), the Trauma History Profile (THP;
Briggs et al., 2012) interview is administered by the clinician upon admission to
the residential facility and is not readministered. The THP includes 20 common
types of traumatic experience, recording age-of-onset and duration of exposure.
The categories of trauma inventoried by the THP were drawn from the National
Traumatic Stress Network core data set of over 14,000 children and adolescents
who attended NCTSN treatment centers across the U.S. (Briggs et al., 2013).
Examples of two categories of traumatization are “impaired caregiver; history of
exposure to caretaker depression/mental health problems, other medical illness, or
alcohol/drug abuse” and “serious injury/accident: unintentional accident or
injury” (Briggs et al., 2013, p. 104). A previous study that utilized the data created
by the THP assessment instrument summarized interview results by converting
the number of trauma types from a possible score of 20 to variable categories: 1–
2, 3–4, or 5+ total types of trauma. The variable conceptualization was intended to
facilitate analysis of contrasts related to general dose response effects for multiple
trauma exposures (Briggs et al., 2013). The THP is described and was
administered as detailed in Hodgdon et al. (2019).
Child Behavior Checklist. The Child Behavior Checklist (CBCL; Achenbach,
2001; Achenbach & Rescorla, 2001, 2004) for ages 6 to 18 is a widely used
52
comprehensive assessment tool designed to assess child behavior problems and
social competencies. It seeks to “identify syndromes of co-occurring problems”
that contribute to psychopathology. The instrument has 118 questions related to
behavior issues and 20 questions related to social competencies. The instrument is
normed according to age, gender, cultural group, and informant. The
questionnaire is scored on a 3-point Likert scale from 0 (not true) to 2 (often true).
A total score of more than 65 indicates problems in the clinical range.
The parent-report form of the CBCL was filled out by each student
participant’s parent-organization assigned clinician. The CBCL includes eight
factors/syndromes: Anxious/Depressed, Withdrawn/Depressed, Somatic
Complaints, Social Problems, Thought Problems, Attention Problems, Rule-
Breaking Behavior, and Aggressive Behavior (Achenbach et al., 2008). Second-
order factors are “Internalizing” the Anxious/Depressed, Withdrawn/Depressed,
and Somatic Complaints syndromes and “Externalizing” the Rule-Breaking
Behavior and Aggressive Behavior syndromes (Achenbach et al., 2008, p. 257).
The CBCL also has “DSM-IV oriented scales” of Affective Problems, Anxiety
Problems, Somatic Problems, Attention Deficit/Hyperactivity Problems,
Oppositional Defiant Problems, and Conduct Problems (Achenbach et al., 2008,
p. 257). Achenbach et al. (2008) reported test-retest reliability estimates (r .80)
and interrater reliability estimates (r .57) across diverse samples of youth. In the
current study, the CBCL subscales showed test-retest reliability estimates within a
three-month interval ranging from .43 (Oppositional Defiant subscale) to .86
(Somatic Complaints subscale).
53
The Children’s Alexithymia Measure. The Children’s Alexithymia Measure
(CAM; Way et al., 2010) assesses limits in children’s ability to recognize and
name their own emotions (alexithymia). It is a 14-item self-report with a single-
factor structure utilizing a 4-point Likert scale from 0 (almost never) to 3 (almost
always). Way et al. (2010) reported an internal consistency estimate (Cronbachs
α) of .92 and a criterion-related validity of .73. When compared to the CBCL,
greater alexithymia was associated with “greater externalizing behavior problems
and greater withdrawn/depressed problems, which suggests preliminary
contrasted-groups validity” (Way et al., 2010, p. 313). In the current study, the
test-retest reliability estimates within a three-month interval were .84 (Time 1 to
Time 2) and .45 (Time 2 to Time 3).
Assessments Completed by a Teacher or Primary Caregiver.
Behavior Rating Inventory of Executive Function. The Behavior Rating Inventory
of Executive Function (BRIEF)—teacher (BRIEF T) and parent (BRIEF P)
versions—measures childrens “everyday behavior manifestations” of executive
function (Gioia et al., 2002). The assessment contains 86 items and yields a three-
factor structure: Behavior Regulation (including Inhibit and Self-Monitor scales),
Emotional Regulation (including Emotional Control and Shift scales), and Meta-
cognition (including Working Memory, Initiate, Plan/Organize, Organization of
Materials, and Task Monitor scales.)
Each item is rated using a Likert scale from 1 (never) to 3 (often) and is
computed by summing the weighted individual items on the scale. The parent
version has been normed, and the instrument has been standardized for parents
54
and teachers. The BRIEFs normative data were based on child ratings by 1,419
parents and 720 teachers from rural, suburban, and urban areas, reflecting 1999
U.S. Census estimates for socioeconomic status, ethnicity, and gender
distribution.
In the current study of TD, milieu staff workers of the parent organization
rated TD participants on the BRIEF P scale. Classroom teachers rated TD
participants on the BRIEF T scale. Executive Function is defined as “abilities that
direct and control goal-oriented Cognitive, Behavioral, and Emotional
Functioning” (Gioia et al., 2002, p. 250). In this study, BRIEF P and the BRIEF T
showed internal consistency estimates (Cronbachs α) ranging from .80 to .98 and
test-retest reliability estimates of .82 for BRIEF P and .88 for BRIEF T. There are
moderate correlations between teacher and parent ratings (rs = .32–.34) for both
parent and teacher forms. In the current study, from Time 1 to Time 2 (an interval
of three months), the BRIEF P showed test-retest reliability estimates ranging
from .38 to .47, and the BRIEF T showed test-retest reliability estimates within a
three-month interval ranging from .40 to .74. From Time 2 to Time 3 (an interval
of three months), the BRIEF P showed test-retest reliability estimates ranging
from .58 to .80, and the BRIEF T showed test-retest reliability estimates ranging
from .71 to .94.
Data Cleaning for Questionnaires. The de-identified CATS data were maintained in an
SPSS file (SPSS 25.0, 2017). All questionnaire data for the pre-, mid-, and posttest points for
each participant for each questionnaire and its subscales were moved from an SPSS file
containing every CATS quarterly assessment to a separate SPSS file containing only study data.
55
There were missing data (missed quarterly-administrations of questionnaires), which is not
unusual for data collected in naturalistic settings (Graham, 2009), but which did impact the final
sample sizes for ANOVA analysis and pattern mapping of each of the respective eight survey
questionnaires (sample size for each questionnaire is attached to its respective Table [see
Appendix C] and Figure [see Appendix D]).
Scale and subscale calculations were spot-checked for accuracy at the item level, when
item-level data were available. Parent-organization tabulation of scale and subscale items was
found to be correct on items that could be checked. Data were examined for inattentive
participant responses. Inattentive responses are those for which there is a suspicious lack of
variability in response to questions, or a suspicious lack of variability from one assessment to the
next assessment three months later (Meade & Craig, 2012). There were not many inattentive
responses; yet, if for instance a participant rated all items on an assessment with the same score,
or if a teacher gave a participant almost exactly the same item scores on every three-month
iteration of the BRIEF T, those assessments would be considered suspiciously lacking in
variability. There were a few responses judged to be inattentive; these were removed.
Electrophysiological Data. The raw electrophysiological HRV data, which had been
individually recorded for each participant (see Appendix B for HRV recording protocol), were
extracted from the HeartMath emWAVE Desktop HRV professional recording program and
equipment (emWave Pro Desktop, n.d.) via individual ASCII files of each recording for each
participant. All ASCII file data were manually inspected for artifacts. Artifacts were then
manually extracted, and the cleaned data were entered into an Excel spreadsheet.3 The
3 Artifacts are short periods of static or of obvious anomaly in the visual representation of the electrophysiological
signal. Artifacts interfere with accurate calculation of the amplitude of the various frequencies. Any static and/or
56
spreadsheet data were then entered into the Kubios HRV analysis program Version 2.1
(Niskanen et al., 2004; Tarvainen et al., 2014).4 For this study, the Kubios program imported
beat-to-beat interval (RR) data from the Excel spreadsheet and calculated frequency and time
domain variables for each individual HRV recording. Natural logarithm variables were also
created for frequency-domain data. The natural logarithm transformed frequency data corrected
for skewness, which rendered those data better suited for statistical testing. Natural logarithms
were calculated for power-spectrum variables of total power, very-low frequency (VLF), low
frequency (LF), high frequency (HF), and low frequency versus high frequency ratio (LF/HF).
Natural logarithms were also calculated for the time domain measure of root mean square of the
successive differences (RMSSD).
Individual PDF-document Kubios HRV Analysis reports were generated for each
participant, for each recording. Each participant’s Kubios RR time-series graph was visually
examined for RR time-series artifacts. Observed artifacts were manually removed, as
recommended by the 1996 report by the Task Force of the European Society of Cardiology and
the North American Society of Pacing and Electrophysiology (TFESC & NASPE). All artifact-
corrected reports were reprinted.
In addition to the five-minute electrophysiological measure of high-frequency heart rate
variability (HF5) and one-minute electrophysiological measure of high-frequency heart rate
variability (HF1) variables that were the focus for this study, Kubios analyzed all other variables
that are in standard use. Each time-domain variable, frequency-domain variable, and each natural
anomalies in the visual wave-tracing of the signal were manually removed so that HRV variables could be
accurately calculated.
4 Kubios is a freeware program that was developed as part of the academic research of its authors at the Department
of Applied Physics, University of Eastern Finland, Kuopio, Finland.
57
logarithm of the frequency-domain variable for each participant (for the first five-minute resting
and first one-minute paced breathing evaluation at pretest and for the first five-minute resting
and first one-minute paced breathing evaluation at posttest) were transferred from the Kubios
HRV Analysis PDF reports into the SPSS analysis system (see Tables C11 through C16 for a
listing of other HRV variables).
Box plots were created for each variable and outliers were identified. Outlying values
were removed in order to perform between-groups analysis of covariance (ANCOVA) and
within-subjects t tests, which were conducted for treatment and comparison groups and included
27 treatment group participants and 18 comparison group participants. Two treatment group
participants had no HRV data, and thus were not included. One treatment group participant
(Kate) did not have any HRV data because she entered the intervention immediately after the
HRV pretest evaluation of the first cohort. The HRV data of a second treatment group participant
(Ariel) were lost in data transfer to the ASCII file.
No single outlying values were excluded from HF-HRV pattern mapping analysis, but
five participants (three treatment group participants and two comparison group participants) were
not included in that analysis. Treatment group members Kate and Ariel were not included for
reasons stated directly above. A third treatment group participant (Pablo) and two comparison
group participants (Isabella and Margaret) produced a high number of artifacts in their HF-HRV
recordings, and thus their cases were entirely eliminated from HF pattern-mapping analysis. As a
result, 24 treatment group participants and 18 comparison group participants were included in
HF-HRV pattern mapping analysis.
58
Results
Questionnaire Data
Because repeated-measures analysis of variance (ANOVA; comparing the differences in
mean scores pre-, mid-, and posttest for each dependent variable) between the treatment and
comparison groups showed inconclusive patterns, I decided to run repeated measures ANOVA
trend analyses for only the treatment group on all scale, factor, and subscale scores of the eight
questionnaire instruments. For the analysis, the Type I error level was relaxed from .05 to a level
of .10 across all tests because of the small sample and the resulting low power of the analysis and
because of the exploratory nature of the study. The aim of the study was to identify variables of
interest for future streamlined assessment of the interventions effectiveness. Whenever the
sphericity assumption was violated, multivariate analysis results were reported. If both linear and
quadratic effects were significant for a particular variable, only the higher-order quadratic effects
have been reported.
Hypothesis I-A
Hypothesis I-A predicted that participation in TD would be associated with an increase in
the flexibility of the fear-emotion circuitry as indexed by a decrease in symptoms. A decrease in
symptoms would be indicated by a decrease in the total scores and any factor-scale or subscale
scores on each of eight psychological questionnaires. The results of the analysis are summarized
below.
Posttraumatic stress symptoms were measured by the Total Score of the PTSD RI Part II
and Part III. According to the hypothesis, there would be a decrease in the PTSD RI total score.
A repeated measures ANOVA trend analysis on the PTSD RI total score showed a decreasing
linear trend (p = .06). At the subscale level, Intrusion (p = .04), Cognition/Mood (p = .09), and
59
Arousal/Reactivity (p = .07) showed a decreasing linear trend (see Table C3, Figure D2, and
Figure D3).
Self-regulation symptoms were measured by the Abbreviated Dysregulation Inventory
(ADI) total score. According to the hypothesis, there would be a decrease in self-regulation
symptoms as assessed by the total score of the ADI. A repeated measures ANOVA trend analysis
on the ADI total score and on its subscales did not show any statistically significant trend (see
Table C4 and Figure D4).
Behavior and social competency symptoms were measured by the CBCL total score.
According to the hypothesis, there would be a decrease in behavioral and social competency
symptoms as assessed by the Total Syndrome Score of the Child Behavior Checklist for Ages 6–
18 (CBCL, 2001). A repeated-measures ANOVA trend analysis on the CBCL Total Syndrome
Score did not show any statistically significant trend (see Table C2 and Figure D1).
At the subscale level there were, however, significant curvilinear convex quadratic trends
(meaning that symptoms went down and came up again) for the Rule-Breaking Behavior
subscale (p = .07) and the Attention Problems subscale (p = .06; see Table C2 and Figure D5).
Depression was measured by the CDI 2. According to the hypothesis, there would be a
decrease in the total score of the CDI 2. A repeated-measures ANOVA trend analysis on the CDI
2 total score showed a decreasing linear trend (p = .08; see Table C5 and Figure D6). At the
factor level, Functional Problems showed a decreasing linear trend (p = .08; see Table C5 and
Figure D6). At the subscale level, Negative Mood (p = .07) and Interpersonal Problems (p = .02)
showed a decreasing linear trend (see Table C5 and Figure D7).
Alexithymia symptoms were measured by the total score of the CAM. According to the
hypothesis, there would be a decrease in alexithymia as assessed by the CAM. A repeated
60
measures ANOVA trend analysis on the CAM total showed an increasing linear trend (p = .09),
implying an increase in alexithymia, contradicting the hypothesis (see Table C6 and Figure D8).
Somatization symptoms were measured by the total score of the BAM. According to the
hypothesis, there would be a decrease in somatization as assessed by the BAM. A repeated-
measures ANOVA trend analysis on the BAM total score did not show any statistically
significant trend (see Table C7 and Figure D9). At the subscale level, Somatic Sensory
Sensitivity (p = .08) showed a curvilinear convex trend (see Table C7 and Figure D10).
Executive function symptoms were measured by the parent version of the BRIEF, the
BRIEF P. According to the hypothesis, there would be a decrease in executive function
symptoms as assessed by the Global Executive Composite (GEC; total score) of the BRIEF P.
Repeated-measures ANOVA trend analysis on the BRIEF P total score showed an increasing
linear trend in the GEC/total score (p = .04), implying an increase in symptoms, contradicting the
hypothesis (see Table C8 and Figure D11). At the factor level, there was a significant increasing
linear trend in the Metacognition factor (p = .01), implying an increase in symptoms,
contradicting the hypothesis (see Table C8, Figure D11, and Figure D12). At the subscale level,
there was an increasing linear trend on the subscales of Initiate (p = .01), Plan Organize
(p = .02), and Organization of Materials (p = .02), implying an increase in symptoms on those
subscales, contradicting the hypothesis. There was also a curvilinear concave quadratic trend on
the subscale of Inhibit (p = .06; see Table C8 and Figure D13).
Executive function problems were also measured by GEC (total score) of the teacher
version of the BRIEF, the BRIEF T. According to the hypothesis, there would be a decrease in
executive function symptoms as assessed by the GEC/total score of the BRIEF T. A repeated
measures ANOVA trend analysis on the BRIEF T showed an increasing linear trend in the
61
GEC/total score (p = .01), implying an increase in symptoms, contradicting the hypothesis (see
Table C9 and Figure D14). At the factor level there was a significant increasing linear trend (see
Table C9, Figure D15, and Figure D16) in both the Behavioral Regulation factor (p = .02) and
the Metacognition factor (p = .02), implying an increase in symptoms, contradicting the
hypothesis. There was also a curvilinear convex quadratic trend on the factor of Metacognition
(p = .07). At the subscale level (see Table C9 and Figure D17), there was an increasing linear
trend on the subscales of Shift (p = .01), Emotional Control (p = .06), Initiate (p = .003),
Working Memory (p = .01), and Plan Organize (p = .02), implying an increase in symptoms on
those scales, contradicting the hypothesis. There was also a curvilinear convex quadratic trend on
the subscale of Initiate (p = .06) and Plan Organize (p = .91; see Table C9 and Figure D17).
Illustration A on the next page summarizes the treatment-group trends described above.
62
Illustration A
Decreasing-Symptom Versus Increasing-Symptom Trends in the Treatment Group
Decreasing Symptoms
Increasing Symptoms
PTSD RI (measure of posttraumatic stress)
Total PTSD score
Intrusion subscale
Cognition/Mood subscale
Arousal Reactivity subscale
CDI 2 (measure of depression)
Total Depression score
Functional Problems factor
Negative Mood subscale
Interpersonal Problem subscale
BAM (measures somatization and
body awareness)
Body Awarenessa subscale
CAM (measure of alexithymia)
Total alexithymia score
Brief P (measure of executive function)
Total executive function
Metacognition factor
Initiate subscale
Organization of Materials subscale
Plan/Organize subscale
Brief T (measure of executive function)
Total executive function
Metacognition factor
Behavior Regulation factor
Shift subscale
Emotional Control subscale
Working Memory subscale
Plan/Organize subscale
Note. PTSD RI = Posttraumatic Stress Disorder Reaction Index. CAM = Children’s Alexithymia
Measure. CDI 2 = Children’s Depression Inventory-2. BRIEF P = Behavior Rating Inventory of
Executive Function – Parent Version. BAM = Body Awareness Measure. BRIEF T = Behavior Rating
Inventory of Executive Function – Teacher Version.
a For the Body Awareness subscale of the BAM, the comparison group (n = 13, (1, 12) = 9.25, p = .01;
η2 = .44) experienced an opposite linear trend to the treatment group.
63
Pattern Analysis in Hypothesis I-A. The goal of this dissertation was to identify
promising variables that might indicate treatment-related effect. In pursuit of that goal, I
attempted to separate signal from noise”; therefore, below, I more concisely present variables of
interest as culled from the section above. The culled variables may be grouped into four
statistically significant treatment-group trends or patterns. In this dissertation, treatment group
results are comprehensively presented.5
Pattern I. Decreasing Trend in Treatment-Group Symptoms From Pre- to Posttest.
Symptoms scales that showed a statistically significant treatment-group pattern consistent with
the hypothesis—namely, a decreasing linear trend in symptoms—were:
PTSD RI: For the self-rated PTSD RI posttraumatic stress questionnaire, there was a
significant decrease in the treatment group symptoms from pretest to posttest on the
following scales: total score (p = .06; see Table C3 and Figure D2) and subscales of
Intrusion (p = .04), Cognition/Mood (p = .09; see Table C3 and Figure D3), and Arousal
Reactivity (p =.07; see Table C3 and Figure D3).6
CDI 2: For the self-rated CDI 2 depression questionnaire, there was a significant decrease
in the treatment group symptoms from pretest to posttest on the following scales:
Depression total score (p = .08; see Table C5 and Figure D6), factor of Functional
Problems (p = .08; see Table C5 and Figure D6), and the subscales of Negative Mood
5 The presentation of comprehensive comparison-group results is beyond the scope of this dissertation; yet, while
probing the data, I did perform repeated measures ANOVA trend analyses for the comparison group. It may be of
interest to view comparison-group results and trends that contrasted with treatment-group results and trends, so I
have included those comparison-group results that contrast in footnotes to the corresponding treatment-group
results.
6 Comparison group: There was a significant curvilinear-concave quadratic trend in the comparison group for the
PTSD RI total score, F(1, 11) = 4.72, p = .05, η2 = .30); for the Intrusion subscale, F(1, 11) = 5.11, p =.05, η2 = .32);
and for the Cognition Mood subscale, F(1, 11) = 4.12, p = .07, η2 = .27, in which symptoms decreased from Time 1
to Time 2 but increased to Time 1 baseline scores from Time 2 to Time 3.
64
(p = .07; see Table C5 and Figure D7) and Interpersonal Problems (p = .02; see Table C5
and Figure D7).7
Pattern II. Increasing Trend in Treatment-Group Symptoms From Pre- to Posttest.
Symptoms scales that showed a statistically significant treatment-group pattern contradicting the
hypothesis—namely, an increasing linear trend in symptoms—were:
CAM: For the CAM clinician-rated alexithymia questionnaire total score, there was an
increasing linear trend in symptoms (p = .09; see Table C6 and Figure D8).8
BRIEF P: For the BRIEF P milieu staff-rated questionnaire of executive function, there
was a significant increase in the treatment group symptoms from pretest to posttest on the
following scales: Executive Function total score (p = .04; see Table C7 and Figure D11),
factor of Metacognition (p = .01; see Table C7 and Figure D11), subscales of Initiate
(p = .01), Plan/Organize (p = .02), and Organization of Materials (p = .02; see Table C7
and Figure D12).9
BRIEF T: For the BRIEF T questionnaire, there was a significant increase in total score
of Executive Function (p = .01; see Table C8 and Figure D14), the factor of Behavior
Regulation (p = .02; see Table C8 and Figure D14), and in the subscales of Shift
(p = .01), Emotional Control (p = .06), and Working Memory (p = .01; see Table C8 and
Figure D17).10
7 Comparison group: There were no significant trends.
8 Comparison group: There were no significant trends.
9 Comparison group: There were no significant trends.
10 Comparison group: There was a significant curvilinear-convex quadratic trend for the Working Memory subscale,
F(1, 8) = 4.60, p = .06, η2 =.37), in which symptoms increased from Time 1 to Time 2 but returned to Time 1
baseline scores from Time 2 to Time 3.
65
Pattern III. Curvilinear Convex Treatment-Group Trend. Two symptom scales showed
statistically significant curvilinear convex patterns. They were:
BAM: For the BAM (self-rated somatic symptom inventory) symptoms measured by the
somatization/body awareness subscale increased from Time 1 to Time 2 and decreased
from Time 2 to Time 3 (p = .09; see Table C7 and Figure D10).11
BRIEF T: For the BRIEF T (teacher-rated executive function) factor of Metacognition
(p = .07), symptoms increased from Time 1 to Time 2 and plateaued between Time 2 and
Time 3 (see Table C8 and Figure D14). Similarly, for the BRIEF T subscales (see Table
C8 and Figure D16) of Initiate (p = .03) and of Plan Organize (p = .91), symptoms
increased from Time 1 to Time 2 and plateaued between Time 2 and Time 3.12
Pattern IV. Curvilinear Concave Treatment-Group Trend. The symptom scales that
showed a statistically significant curvilinear concave pattern (symptoms decreased from Time 1
to Time 2, and increased from Time 2 to Time 3, ending slightly above Time 1’s score) were:
BRIEF P: For the BRIEF P (milieu staff-rated executive function), the subscale of Inhibit
(p = .06; see Table C7 and Figure D13) showed symptom decrease from Time 1 to Time
2 and increase from Time 2 to Time 3, ending slightly above the Time 1 score.
CBCL: For the CBCL (clinician-rated social and behavioral competency; see Table C2
and Figure D5), symptoms of Rule-Breaking Behavior (p = .07) and Attention Problems
11 For the comparison group: There was a positive linear trend in the comparison group, F(1, 12) = 9.25, p = .01,
η2 = .44.
12 Comparison group: There was also a significant Plan Organize subscale curvilinear-convex quadratic trend in the
comparison group, F(1, 8) = 5.64, p = .05, η2 =.41, in which symptoms increased from Time 1 to Time 2 but
returned to Time 1 baseline scores from Time 2 to Time 3.
66
(p = .06) decreased from Time 1 to Time 2 and increased from Time 2 to Time 3, ending
above Time 1’s score.13
Summary of Findings for Hypothesis I-A. Hypothesis I-A predicted that TD would be
associated with an increase in the flexibility of the fear-emotion circuitry as indexed by a
decrease in symptoms (a decrease in symptoms would be indicated by a decrease in the total
scores and any factor scale or subscale scores of each of the eight psychological questionnaires).
Hypothesis I-A was partially supported for the treatment group because some but not all
symptoms decreased over time. While depression, PTSD, and body awareness problems
decreased over the course of the intervention, executive function symptoms and alexithymia
symptoms increased. Rule Breaking Behavior and Attention Problems both showed a curvilinear
concave quadratic pattern but rebounded to approximate pretest levels by the end of treatment.
Hypothesis I-B
Hypothesis I-B predicted that participation in TD would be associated with an increase in
the flexibility of the fear-emotion circuitry as indexed by an increase in the HF power spectrum
measurement of HRV. In addition to HF HRV, all HRV time variables and HRV frequency
variables were probed for pre- to posttest changes for both the five-minute resting condition
(HF5) and the one-minute paced-breathing condition (HF1). Between-groups analysis of
covariance (ANCOVA) and within-subjects t tests were conducted for treatment and comparison
groups. For the between-groups ANCOVA, there were no significant differences between
treatment and comparison groups on any of the tested variables for either the five-minute
condition or the one-minute condition (see Table C11 and Table C12).
13 Comparison group: there were no significant trends.
67
For within-subjects t tests in the five-minute resting-condition analysis (HF5; see Table
C13 and Table C14), there was one statistically significant change: VLF power was significantly
higher in the posttest measurement of the comparison group. The finding lacks utility, however,
given that, in short-term recordings, VLF is an unstable variable/frequency that can fluctuate
with circadian rhythm and a variety of other factors and cannot be reliably interpreted (TFESC
& NASPE, 1996).
For the within-subjects t tests in the one-minute paced-breathing analysis (see Table C15
and Table C16), there was one significant change: posttest RR interval was significantly higher
in the comparison group (indicating lower heart rate). Lower heart rate is generally considered
more favorable but is not a strong indicator of increased systemic flexibility (TFESC & NASPE,
1996).
Summary of Findings for Hypothesis I-B. TD did not increase flexibility of the fear
emotion circuitry as indexed by an increase in HF. Overall, there were no meaningful differences
between treatment and comparison groups on any HRV variables. Therefore, Hypothesis I-B was
not supported.
Hypothesis II
Hypothesis II predicted that participation in TD would be associated with a decrease in
youth violence as indexed by a decrease in aggression and social-problem symptoms. Aggression
symptoms were measured by the CBCL Externalizing factor score; social-problem symptoms
were measured by the CBCL Social Problems subscale and the CBCL Internalizing (indexing
hyperactivity and withdrawal) second-order factor score. According to the hypothesis, there
would be a decrease in all of these measures. Repeated measures ANOVA trend analysis on
these scores did not show any statistically significant trend (see Appendix C, Table C2 and
68
Appendix D, Figure D1). Therefore, while TD appeared to halt the progression of aggression, it
did not reduce aggression as indexed by the CBCL Externalizing factor score.
Exploratory Analysis of HRV Data
Because chronic developmental trauma influences the physiological system differentially
depending upon a variety of factors, I expected to find physiological phenotypical subtypes
within my sample. It seemed to me that longitudinal intrapersonal comparisons on HRV
variables could be meaningful for this complex population. Thus, I probed the data to explore
whether electrophysiological subtypes moderated patterns of response to the TD intervention.
The HF variables of the HRV power spectrum are believed to best reflect systemic flexibility; for
that reason, I explored the HF1 and HF5 variables for evidence of subtypes. My exploratory
analysis began with intrapersonal comparisons of the HF1 and HF5 variables from pre- to
posttest.
Spielbergers framing of the difference between trait versus state conditions (Spielberger,
1966, 2012) provided a useful tool with which to conceptualize and label the difference between
the HF1 and HF5 variables. In this study, the HF1 paced-breathing task measured maximum
induced parasympathetic nervous system (PNS) capacity achieved at will (via a one-minute EKG
recording of a paced-breathing exercise). Thus, HF1 reflected the maximum parasympathetic
state (meaning the transient voltage/amplitude/power of HF) that the participant could induce
and achieve at will. In contrast, HF5 measured the strength of the PNS at rest via a five-minute
resting EKG recording. In general, resting HF would remain relatively stable over time
meaning that voltage/amplitude/power would stay relatively consistentunless a person
experienced a major physiological change (e.g., a significant increase in athleticism). Thus, for
this study, the HF5 resting measure reflected participant trait (stable) PNS function.
69
I analyzed the way in which HF1 (state HF) and HF5 (trait HF) changed in relation to
each other from pretest to posttest. If an individual’s HF1 state/maximum-induced
parasympathetic tone and their HF5 resting/trait parasympathetic tone moved closer together
from pretest to posttest, I reasoned that the nervous system would be more balanced,
consequently indicating greater nervous-system flexibility. Greater flexibility of the nervous
system would manifest in increased flexibility of the fear/emotion circuitry, enabling the
individual to more effectively respond to the changing environment. Such flexibility might be
important, because it would counteract the inflexible and habitual defenses that would have been
a disruptive consequence of complex traumatization.
In order to quantify the longitudinal changes in the ratio of state versus trait dominance
from pretest to posttest, I developed the following formulas: HF1pretestHF5 pretest = State-Trait
HFpretest (S-T HFpretest) And HF1posttestHF5posttest = State-Trait HFposttest (S-T HFposttest). The new
S-T HFpretest and S-T HFposttest variables were single variable snapshots of systemic balance at
pretest and at posttest (D’Andrea, personal conversation, June 24, 2016).14 With the new
variables, participants could be compared to themselves from pretest to posttest. For the
treatment group, I plotted each participant separately, mapping the participant’s directional
movement/pattern on S-T HF from pretest to posttest. Participants’ directional movement over
time on S-T HF appeared to separate them into three phenotypical subtypes (see Figure D18).
Following the same procedure, the comparison group could then be separated into the same three
phenotypical-subtype groups (see Figure D19).
14 Associate Professor Wendy DAndrea, PhD., of the New School helped to conceptualize and operationalize the
novel formula of S-T HF as a way to capture longitudinal shifts in state/trait PNS balance (personal communication,
June 24, 2016).
70
For the three subtypes (i.e. Calming Down, Sowing Seeds, and Waking Up subtypes)
from pretest to posttest, if the S-T HF variable produced was a negative number by posttest, that
meant that HF5 and trait PNS had become more dominant for the participant; if the S-T HF
number was positive, that meant that state PNS had become more dominant for the participant. If
the S-T HF variable was near zero, it meant that HF1 and HF5 had gained balance (neither state
nor trait was dominant over the other). See the next page for plottings of the three subtypes for
the treatment group in Illustration B and for the comparison group in Illustration C.
All other HRV variables were mapped by phenotypical subtypes for treatment and
comparison conditions (see Figures D24 through D63). For the Sowing Seeds and Waking Up
subtypes, other HRV variables supported and elaborated the systemic balance information that
could be gleaned from each subtype via the longitudinal movement of the S/T HF variables from
pre- to posttest (see Footnotes 16 and 17 for a description of those variables).
Regardless of treatment status (treatment or comparison group), the longitudinal pattern
of state versus trait balance (S-T HF) from pre- to posttest exhibited the following trends in the
three phenotypes (see Figures D18 and D19).15 Group 1 (Calming Down subtype) exhibited
positive S-T HFpretest that moved from above zero to near zero by S-T HFposttest, achieving trait
versus state balance. This meant that at pretest the Calming Down 1 subtype’s ability to achieve
state PNS activity at will was stronger than the ability to experience trait PNS activity at rest. At
posttest, trait PNS activity appeared to strengthen and balance with state PNS activity, imparting
maximal systemic flexibility of response to varying in-the-moment challenges.
15 See Figures D20 and D21 for the typological subtypes from the combined data as compared to norms.
71
Illustration B
State/Trait High Frequency (S/T HF) Balance in the Treatment Condition by Three Phenotypical
Subtypes Pre- Versus Posttest for the Treatment Group
Note. n = 24 (Subtype 1 = 10, Subtype 2 = 5, Subtype 3 = 9). This illustration is also included as Figure
D18 in Appendix D.
Illustration C
State/Trait High Frequency (S/T HF) Balance in the Comparison Condition by Three
Phenotypical Subtypes Pre- Versus Posttest for the Comparison Group
Note. n = 18 (Subtype 1 = 4, Subtype 2 = 4, Subtype 3 = 10). This illustration is also included as Figure
D19 in Appendix D.
72
Group 2 (Sowing Seeds subtype) exhibited balanced S-T HFpretest but S-T HFposttest had
fallen far below zero, which indicated that trait PNS activity had strengthened, grown dominant,
and had perhaps become excessive relative to maximum achievable state parasympathetic
activity. Excessive trait PNS activity might have caused participants to end the study “checked-
out” and/or defaulting to dissociation in the face of varying in-the-moment challenges. It is
possible that the Sowing Seeds subtype of the treatment group experienced a differential
treatment effect for TD, in that they became more markedly trait PNS dominant than did the
comparison group. The treatment condition of Sowing Seeds had a much sharper downward
slope from pre- to posttest on the S-H HF variable (see Figure D18) than did Sowing Seeds in the
comparison condition (see Figure D19).16
Group 3 (Waking Up subtype) had negative S-T HFpretest that moved from below zero to
above zero by S-T HFposttest. At pretest the Waking Up subtype appeared to experience excessive
trait PNS activity (perhaps being checked-out or dissociated at rest), which was stronger than the
ability to evoke state PNS activity under in-the-moment challenge. By posttest, it seemed that
state PNS activity became dominant, and trait PNS activity receded. The Waking Up subtype
may have ended the posttest feeling less calm at rest but more able to induce calm under
challenge.17
16 Other variables provided additional support for both state and trait posttest increases in PNS strength for the
Sowing Seeds subtype. On the longitudinal mapping of HF5 alone (higher levels of which indicate greater trait
parasympathetic strength), the treatment condition of the Sowing Seeds subtype showed a sharper and higher
upward slope (indicating greater trait parasympathetic activity, see Figure D50) than did the comparison group (see
Figure D51). Longitudinal movement of the Sowing Seeds group on the SDNN1 variable follows an upward slope
(see Figure D28), indicating an increase in state PNS strength (while there was a slight decrease in state PNS in the
comparison group [see Figure D29]).
17 The treatment condition of Waking Up #3 subtype appeared to become slightly more sympathetically aroused and
awake in the face of varying in-the-moment challenges. While sympathetic arousal cannot be explained by the S/T
HF graphs alone, it can be deduced when longitudinally mapping Log LF5 with the covariate Log HF 5 (see Figures
73
In summary, treatment and comparison phenotypical subtypes followed similar subtype
trends from pretest to posttest. The TD treatment condition itself did not appear to differentially
impact S-T HF by phenotypical subtype, except for sharper slopes that indicated greater trait
parasympathetic activity in the treatment condition of the Sowing Seeds subtype and slightly
greater sympathetic activation in the resting measure of the treatment condition of the Waking
Up subtype. Thus, in the total sample of complex-trauma-exposed residential youth, who all
received both trauma treatment as usual and adjunctive therapies over the course of the study
(with the treatment group alone receiving TD), the same three electrophysiological subtype
groupings (i.e., Calming Down, Sowing Seeds, and Waking Up) were longitudinally observed
regardless of treatment versus comparison condition. In other words, everyone’s physiology
longitudinally shifted in one of three subtype patterns, regardless of whether or not they had
received the TD treatment.
Demographic Differences in Subtypes. The possible differences in demographic
characteristics of the phenotypical subtypes were examined (see Table C17). The Calming Down
subtype (across treatment and comparison conditions) experienced a heavier cumulative trauma
load (17.5 years) as compared to both the Sowing Seeds subtype (10.14 cumulative years) and
the Waking Up subtype (11.62 years). It may also be worth noting that the sexual trauma burden
was higher for the Calming Down subtype. Sexual trauma was experienced by 72.7% of the
Calming Down subtype, in comparison to 44.4% of Sowing Seeds subtype and 30.8% of the
Waking Up subtype. The Sowing Seeds subtype had a much lower number of participants
D62 and D63). Greater sympathetic activation may be observed at rest during posttest (compared to pretest) for the
treatment condition of the Waking Up subtype. In the comparison group, longitudinal mapping at-rest sympathetic
activation, as indicated by Log LF5 with the covariate Log HF5 (see Figures D62 and D63), showed an opposite
decreasing trend.
74
reporting physical trauma (11.1%), as opposed to the Calming Down subtype (63.3%) and
Waking Up subtypes (53.8%).
Exploratory Analysis of Questionnaire Data
I also examined the data from the questionnaires in terms of the HRV subtypes.
CDI 2. I had hypothesized that a complex intervention such as TD might interact with
phenotypical subtypes by keying into the next sequentially needed emotional/developmental
building block required by a particular phenotypical subtype. Because longitudinal-path patterns
of S-T HF appeared to separate all participants into three subtypes (regardless of treatment vs.
comparison condition), I wondered whether one subtype might display more treatment-related
symptom changes than other subtypes and whether there would be longitudinal symptom
differences between treatment- and comparison-groups by subtype. I therefore decided to explore
how S-T HF HRV subtypes (Calming Down, Sowing Seeds, or Waking Up) might relate to
response to the CDI 2 (the questionnaire that had showed the greatest reduction of symptoms in
the overall treatment group vs. the comparison group [see Table C15, Figure D16, and Figure
D17]). For the three subtypes, I ran the trend analysis in repeated measures ANOVA for the CDI
2’s total, factor, and subscale scores, and I longitudinally mapped each scale by subtype (see
Figure D6 through Figure D77).
When divided by treatment- versus comparison-group conditions, the three subtypes had
small sample sizes, yet longitudinal movement on scales by subtype provided interesting hints of
potential targets for further exploration. When contrasting treatment to comparison conditions
(see Figure D6 through Figure D77), two subtypes (the Calming Down S-T HF subtype and
Sowing Seeds S-T HF subtype) showed opposite patterns (treatment vs. comparison) on most
scales and subscales of the CDI 2: the Calming Down subtype of the treatment group
75
consistently showed a decreasing trend in depression symptoms over time, while the Calming
Down subtype of the comparison group showed either a flat trend or an increasing trend in
depression symptoms over time. It seemed that most of the overall effect in treatment-group
depression symptom reduction could be accounted for by the scores of those whose heart rate
variability placed them in the Calming Down subtype.
The scope of this dissertation is limited, so for the remainder of the psychological
instruments, I explored total scores only for phenotypical subtype movement as divided by
treatment versus comparison conditions (see Figures D78 through D93 for longitudinal mapping
of Total Scores). For succinct presentation, only S-T HF subtype movement on the total scores of
remaining instruments is described in the next section.
BRIEF T and ADI. The differential patterns observed on the CDI 2 in the Calming
Down subtype can again be observed in the ADI total score (see Figures D78 and D79) and the
BRIEF T Executive Function total score (see Figures D82 and D83). For those in the Calming
Down subtype, as assessed by the CDI 2 and the ADI, treatment group symptoms decreased over
time, while comparison group symptoms stayed relatively the same. However, for those in the
Calming Down subtype, as assessed by the BRIEF T, symptoms increased for the treatment
group but decreased for the comparison group. Given the above, it seems that the Calming Down
subtype may account for most of the differential symptom changes captured by psychological
instruments in the overall treatment group. The total scores on the remainder of the
questionnaires did not show markedly consistent or notable patterns in either the treatment or
comparison group (see Figure D84 through D93).
It appears that the Calming Down subtype, which had the greatest cumulative trauma
load and a greater incidence of sexual trauma, had, paradoxically, the most marked differential
76
treatment versus comparison response on the CDI 2. The Calming Down phenotype in the
treatment condition (differentiated from the comparison condition of the Calming Down
subtype) exhibited a decreasing trend for depression, total PTSD, affect dysregulation, and total
body awareness symptoms, and an increase in teacher-rated executive function symptoms. In
contrast, the Sowing Seeds subtype and the Waking Up subtype showed no notable patterns in
the treatment condition relative to the comparison condition. See Illustration D.
Illustration D
Decreasing Versus Increasing Symptom Trends in the Calming Down Phenotypical Subtype
Treatment Condition
Decreasing Trend in Symptoms
Increasing Trend in Symptoms
CDI 2 (measure of depression)
Total Depression score
Negative Mood
Ineffectiveness
Interpersonal Problems
Emotional Problems
Functional Problems
ADI (measures affect dysregulation)
Total Score
BRIEF T (measure of executive function)
Total executive function
Note. CDI 2 = Children’s Depression Inventory-2. BRIEF T = Behavior Rating Inventory of Executive
Function – Teacher Version. ADI = Abbreviated Dysregulation Inventory. Because of the small n of the
subgroupings, participants were pattern mapped by subgroupings in treatment versus comparison
subgroupings (rather than probed for statistical significance). The comparison condition of the Calming
Down subtype showed opposite trends to those summarized in the table above for the Calming Down
subtype. All scales and subscales of the CDI were pattern mapped and examined. All scales and subscales
showed opposite patterns for the Calming Down comparison group. Only the total-score was pattern
mapped and examined for the ADI and the BRIEF T.
77
Discussion
Results
My initial plan was to test TD’s effectiveness by comparing complex-trauma-exposed
residential youth who had experienced the TD intervention (the treatment-condition) to a
comparison group of youth from the same population who had not experienced the TD
intervention. Initial analysis of group differences in trends was inconclusive. I therefore focused
primarily on trend analysis of the treatment group.
Hypothesis I-A. Hypothesis IA was partially supported. The supported findings indicated
that systemic flexibility did increase in the treatment group. Analysis of standardized-
questionnaire data revealed statistically significant decreases in depression and PTSD symptoms
among the TD treatment group. The treatment group also showed statistically significant
increases of alexithymia and executive function symptoms. Thus, the Hypothesis IA
questionnaire findings may possibly indicate systemic flexibility shifts in the treatment group. It
could be that patterns of improvement in certain symptoms, as well as an exacerbation (or
transient exacerbation) in other symptoms, may be indicative of a growth/recovery trajectory.
Findings related to Hypothesis IA appear to indicate that participantsfear-emotion
circuitry became more flexible by the end of the intervention. Yet, the equating of only
significant symptom decreases with increased systemic flexibility may not be the best
conceptualization of flexibility. I considered the possibility that decreases in some symptoms and
increases in others may have combined to present a more natural and accurate reflection of
emotional flexibility and emotional maturation as opposed to a simple decrease in all symptoms.
For instance, in this study, it is possible to speculate that increased flexibility of the fear-emotion
circuitry was captured by the significant decrease in PTSD and depression, as well as by the
78
significant increase in alexithymia and exacerbation in symptoms of executive function. Again,
patterns of lesser as well as greater symptoms (detected by the psychological questionnaires)
may have signaled the natural fluctuation of dynamic recovery from complex traumatization.
And thus, those patterns may have indicated increased flexibility of the fear-emotion circuitry.
When the flatness of depression and the negative memory intrusions of PTSD decrease,
those changes may be accompanied by the emergence of new/unfamiliar feelings and cognitions.
People who have not experienced themselves fully in the past and who are confronted with an
assortment of unfamiliar feelings and thoughts may not know how to synthesize that new
information, perhaps causing an increase in alexithymia (not knowing what to make of body
sensations) and a worsening of executive function symptoms. Considering these possibilities, if
we have mistakenly been using decreases in symptom scales across the board to indicate success
of an intervention—without taking into account that in some cases increasing symptoms may
indicate treatment progress—we may have previously discounted effective interventions.
Other possibilities arising from these findings are as follow.
1. Treatment may be effective in reducing some but not all symptoms.
2. Treatment may be effective in immediately reducing some symptoms, while other
symptom decreases may lag.
3. There could be an actor-observer bias: the questionnaires that reflected a
significant increase in symptoms for executive function and alexithymia measures
were rated by teachers and clinicians, respectively, while the questionnaires that
reflected a significant decrease in symptoms for depression and PTSD were self-
rated. It may be that subjects themselves internally experienced and assessed
improvement, while external observers (who may have been more concerned with
79
behavioral control and/or may not have been aware of internal shifts and changes)
perceived an increase in symptoms. Observation of complex-trauma-related
recovery phenomena may lead to opposite interpretations depending upon the
interpreter (self-reporter vs. other reporter). When the system becomes more
flexible, subjects may feel relieved; they may feel better and feel their symptoms
have decreased. However, others who are observing the same individuals may
interpret the expanding repertoire of emotional/behavioral/cognitive activity and
increased self-expression that comes with increased flexibility as a worsening of
symptoms (Berger et al., 2005; Kim et al., 2018; Verhulst & Ende, 1992).
4. Alternatively, different patterns and combinations of waxing and waning
symptoms over time may signal the presence of multiple specific phenotypes
(specific subtypes of behavioral and physical traits [Taylor & Lewotin, 2017]).
The first three possibilities should be explored through follow-up investigations. In the current
study, I explored the last possibility using the HRV data as presented in exploratory data
analysis.
Hypothesis I-B. In contrast to findings for Hypothesis I-A, no statistically significant
differences in trends were detected in the Hypothesis I-B HRV data. There were no meaningful
differences between treatment and comparison groups on any HRV variables. Therefore,
Hypothesis I-B was not supported.
Hypothesis II. Hypothesis II was not supported. Two previous studies of Urban Improv
(UI), a predecessor of TD (Kisiel et al., 2006; Zucker et al., 2010) demonstrated a halting of
aggressive behaviors in youth participating in this secondary violence prevention program versus
controls, and an improvement in pro-social behaviors identified as important potential mitigators
80
of youth aggression. In the current study, TD appeared to halt the progression of aggression, but
it did not reduce aggression as indexed by the CBCL, because neither aggression nor social-
problem symptoms significantly changed on the levels of total scores, factor scores, or subscale
scores. Therefore, the TD model when implemented with youth with histories of complex trauma
exposure and adaptation may not function primarily as a violence prevention model per se. This
observation is not necessarily contradictory to the designed intent of TD, which was developed
not as a secondary violence prevention program but a tertiary prevention and clinical
intervention model for violence and trauma-exposed youth exhibiting significant trauma-related
symptoms and difficulties. In addition, the UI studies and the current study did not use the same
outcome measures to index aggression. The UI studies used the Social Skills Rating System
(SSRS; Gresham & Elliott., 1990) to measure outcomes related to aggression, while the current
study used the second-order factor Externalizing scale of the CBCL (Achenbach et al., 2008) in
order to approximate the measure of aggression, because questionnaire data in the TD study were
confined to secondary analysis of already collected data available through the parent
organization’s quality-control database. (While Hypothesis II was not supported by secondary
analysis of the CBCL questionnaire results, reductions in aggression and increases in pro-
sociality were clearly supported by the qualitative portion of this dissertation, which indicates
that attitudes toward aggression and aggressive behaviors were reduced and pro-social attitudes
and behaviors increased.)
Exploratory Analysis. The possible moderation of the treatment effect by HRV
phenotypical subtypes was further explored. Three phenotypical subgroups were identified (see
Figure D18 and Figure D19) based on the participants’ directional movement over time on the S-
81
T HF variable (a snapshot of state/trait PNS balance). I have called those groups Calming Down,
Sowing Seeds, and Waking Up.
The Calming Down subtype had been state-PNS dominant at pretest, but by posttest had
developed an equal balance between state and trait PNS. At pretest, resting-state calm would not
have been as accessible as at posttest. At posttest, with both state and trait PNS accessible and
flexibly balanced, the Calming Down group may have been able to toggle fluidly between trait-
calm when safely at rest and at-will state calm as demanded by the circumstances at hand.
The Sowing Seeds subtype, which had exhibited balanced state and trait PNS at pretest,
became markedly trait-PNS dominant at posttest (although both state and trait PNS
strengthened), perhaps imparting a type of dissociative or checked-out/numb condition at
posttest. Although the treatment and comparison groups of the Sowing Seeds subtype followed
the same pattern, the slope of the treatment group was much sharper, indicating even greater trait
PNS dominance than demonstrated by the comparison group. This may have been a treatment
effect.
The Waking Up subtype had been excessively trait PNS dominant at pretest (perhaps
dissociative or checked-out at pretest). By posttest, the Waking Up subtype of both the treatment
and comparison group had developed state PNS dominance, suggesting that trait PNS had
receded (indicating that dissociation or separation from one’s own experience had diminished),
and state PNS (the ability to calm at will) had become dominant. For the Waking Up subtype,
state/trait longitudinal courses were similar for the treatment and comparison groups, but the
treatment group experienced slightly more sympathetic activation (as indicated by LF with
covariate HF variable pre- to posttest), which may have reflected a differential treatment effect.
82
In general, it seemed that the three distinctive phenotypical subtypes were present in both
the treatment and comparison groups. It appeared that the entire sample followed one of the three
subtype electrophysiological state/trait PNS patterns over time, regardless of treatment vs.
comparison assignment. The literature supports the notion that the characteristic represented by
the HF5 variable (i.e., trait resting high-frequency activity in a short-term recording) is stable
across time, except in the case of a major physiological shift in a person (e.g., losing or gaining
substantial physical fitness, or gain or loss of substantial body mass, etc.; Ge et al., 2020;
Goedhart et al., 2007; Spangler et al., 2018). It is perhaps of interest then that HF5 itself (see
Figures D50, D51, D54, and D55) longitudinally shifted by phenotypical subtypes across the
treatment and comparison conditions. This change in HF5 and trait PNS balance might possibly
indicate that all participants experienced electrophysiological effects as a result of the parent
organization’s treatment environment as usual.
Next, I explored whether questionnaires used in this study had captured significant
symptom changes. Because the CDI 2 had shown the most consistent treatment effect across
scales and subscales, I pattern mapped the treatment and comparison groups by subtypes and
compared the subtypes on all scales of the CDI 2. The Calming Down subtype of the treatment
condition appeared to account for most of the symptom decreases on the CDI 2, while the
Sowing-Seeds and Waking Up subtypes did not consistently show decreases in symptoms (see
Figures D64 through D77). Thus, it is possible that the electrophysiological subtypes had
confounded the treatment effect and masked the extent of the hypothesized effect on the
depressive symptoms.
I decided to briefly probe the other questionnaires for notable differential patterns by
longitudinally mapping questionnaire total scores by subtype for both treatment and comparison
83
groups (see Figures D78 thorough D93). The treatment condition of the Calming Down subtype
displayed decreasing trends in affect dysregulation symptoms on the ADI and an increasing trend
in teacher-rated executive function symptoms on the BRIEF T, while the comparison condition
showed the opposite patterns on these questionnaires. In contrast, the Sowing Seeds and Waking-
Up subtypes did not show consistent contrasting patterns of interest. Therefore, as in CDI 2, the
Calming Down subtype seems to be responsible for changes in affect dysregulation and
executive function symptoms.
In addition, as described before, there were some clear demographic differences among
the subtypes. Notably, individuals included in the Calming Down subtype had the highest mean
cumulative-trauma load and sexual-trauma burden (see Table C17), which might underlie the
greater pre- to posttest symptom changes and perhaps the greatest benefit from intervention. A
second possibility is that the questionnaires used in this study were not sensitive to any changes
experienced by the other two phenotypical groups or by the overall treatment group. A third
possibility is that missing data, or a low number of responders, might have masked subtype
pattern-mapping results. A fourth possibility is that there is a lag in symptom abatement for the
Sowing Seeds and Waking Up subtypes that might be revealed at a later time.
Although HRV norms are not particularly well-defined for adolescents and are not
necessarily relevant to this population (who vary widely because of physical development,
medication status, etc.), I did compare the three phenotypical subtypes on HF5 (vagal tone), both
pre- and posttest, to existing norms (Corrales et al., 2012) for this age group (see Figures D22 &
D23). At pretest the Calming Down subtype exhibited HF5 trait PNS vagal tone far below the
norm; at posttest, trait HF5 increased to come much closer to the norm. The Calming Down
subtype presented the phenotypical profile and responses that would be expected by existing
84
HRV-literature (as it relates to depression and trauma impacts), except that for executive
function the literature would have predicted a decrease in executive-function symptoms
(Beauchaine & Thayer, 2015) rather than the increase in executive-function symptoms displayed
by the Calming Down subtype on the teacher-rated BRIEF T. The unexpected increase in
executive-function symptoms may have been caused by observer bias on the part of the teacher-
observers.
It is also possible that improvement in executive-function symptoms lagged behind
improvement in depression and affect regulation symptoms, and that initial executive-function
disorganization was in fact a positive indicator of progression on a healing trajectory. As people
shake free from engrained neural patterns set to evoke instantaneous defense under survival
threat, a dynamic fluctuation in executive-function symptoms could precede a new ordering of
executive-function neural connections. For the Calming Down subtype, a new executive-function
order set to flexibly evaluate and navigate a safer environment might have been evident at a later
posttest assessment. In such a case, it would be beneficial to continue to follow participants for a
longer time to see whether executive function symptoms do, in fact, lag, before ultimately
showing the expected improvement. The cumulative effects of traumatization produce oscillation
between emotional under- and over-modulation (Lanius, Frewen, et al., 2010). As the
cumulative-trauma load and its impact increase, dysregulation of the biological system
progressively increases (Lanius, Frewen et al., 2010; Shonkoff et al., 2009; Yehuda et al., 2015).
It would seem logical that as the cumulative-trauma impact on a person’s system unwinds and
decreases via interventions, healing might be tracked by variables that index progressive
reorganization of better regulation and of greater stability of modulated affect.
85
The Calming-Down group, with its higher cumulative trauma load, accounted for most of
the significant symptom reductions in the CDI 2 depression scale and subscales. Prior studies
have associated high trauma load with a combination of depression and low HF HRV. Higher
trauma load and depression have been associated with poor treatment response (Barawi et al.,
2020), but in the current study of TD, the Calming Down phenotype, which displayed the highest
level of depression and highest cumulative trauma load, was the phenotypical subgroup that was
most responsive to the TD treatment as measured by the CDI 2 depression questionnaire. In a
study of 62 depressed individuals as compared to 65 healthy controls, Hartman et al. found that
major depression was associated with low HF HRV (2019) and that there was a progressive
improvement in both HF HRV/vagal tone and depression symptoms over the course of treatment.
Hartmans treatment was pharmacological, and he postulated that improved HRV parameters
directionally caused an improvement in arousal modulation (improved autonomic nervous
system function), which led to symptom improvement. However, it is possible that symptom
improvement directionally led vagal tone improvement. A Danish network analysis of the
symptoms of 2,782 adults seeking treatment (Fried et al., 2017) found that detachment and loss
of interest in activities were strongly connected to other PTSD symptoms. Therefore, the
depression-related symptoms of detachment (i.e., anhedonia) and loss of interest, as affected by
the TD intervention, might secondarily affect HF HRV. It is also possible that improvement in
depression and HF HRV were bi-directionally related (e.g., HF changes perturbed symptom
abatement, which in turn perturbed more HF changes, and so on).
The Calming Down subtype’s decrease in depression symptoms might also be
conceptualized as an increase in positivity (Garland et al., 2010; Kogan et al., 2014; Licht et al.,
2008; Rottenberg et al., 2007). Prior studies have supported the proposal that increases in HF
86
HRV functioning/vagal tone be used as a physiological indicator of positive affectivity, and
therefore a reflection of the positive-valence system of the NIMH Research Domain Criteria
(RDoC; National Advisory Mental Health Council Workgroup, 2018) and an indicator of a
strengthening medial-prefrontal-cortex core integration (Gruber et al., 2015; Thayer & Lane,
2009). In the current study, over the course of treatment, the Calming Down subtype experienced
an increase in trait HF HRV functioning, which would predict a decrease in depression
symptoms by posttest, unlike the other subtypes.
My findings led me to an idea that requires exploration at a later time: rather than being
discrete phenotypes, Calming Down, Sowing Seeds, and Waking Up might represent a cycling
trajectory of stepwise nervous-system adjustments, slowly and dynamically nudging the system
to a new norm of flexibility (as the impact of trauma unwinds). Illustration E on the next page
shows a theoretical model of cyclical progression. With each cyclic iteration, the system is
perturbed to operate in a more balanced, flexible, and stable fashion as it cycles through Calming
Down to Sowing Seeds to Waking Up, and then begins the next iteration of the cycle more
balanced than it was during the previous cycle.
87
Illustration E
Cycling Trajectory of Stepwise Nervous-System Adjustments
Note. Cycle A reflects data from the TD study. Cycles B, C, & Stable D are theoretical projections.
88
Limitations and Implications
Limitations
An important limitation of the current study is the small sample size available for
quantitative analyses of the questionnaire data (N = 47). Sample-size issues were further
aggravated when exploratory analyses were conducted using subtypes. The current study
proceeded with the small sample size because of the rare opportunity to access complex-trauma
exposed teenagers in residential care (for whom research is remarkably absent, yet critically
needed). Regardless, because of the small sample sizes, sampling errors are a threat to the
conclusions.
The detailed demographic information that the parent organization collected and provided
is a strength of this study, but a great deal of demographic data was missing or incomplete,
particularly data related to race and ethnicity. Because this data had not been collected for many
of the participants, it is possible that the participant group was more diverse than reflected and, if
so, minority participants may have experienced societal oppression that could have resulted in a
greater cumulative-trauma load that was not captured in the quantitative portion of the study.
Although I argue that, because of decreases in depression and affect dysregulation, the
Calming Down subtype benefited most from the TD treatment, it is also possible that treatment
effects for all subtypes were not sufficiently captured by the instruments. Examination of the
psychometric properties of the BAM uncovered some issues. The scoring instructions in the
manual were not correct for reverse-coded items. Based on item content and inter-item
correlations, items 5, 6, 7, and 8 must be reverse-coded and item 4 must not be reverse-coded,