University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 1

Addressing Federal Pain Research Priorities:
Drug Policy, Pain Mechanisms, and Integrative Treatment

by
Chun Nok Lam

____________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
PREVENTIVE MEDICINE: HEALTH BEHAVIOR RESEARCH

December 2019
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 2

Acknowledgements
I would like to extend my gratitude to my committee members, Dr. David Black, Dr.
Chih-Ping Chou, Dr. Joel Milam, Dr. Jennifer Unger, and Dr. Michael Menchine for their
guidance and support throughout my doctoral program and the dissertation process. I am
thankful for Dr. Black being an extremely dedicated mentor and friend. He has committed
enormous amount of time helping me become a better writer and building my critical thinking
skills to understand complex concepts through theoretical constructs. I’m thankful for his time to
chat news, family, life and career, and, most memorably, being my basketball buddy playing
outdoor at Soto under the 95°F sun. I am thankful for Dr. Chou for his father-like mentor
character, walking through each and every concern I have from how to impute missing data to
how to raise a child. Our many conversations gave me reassurance that I was doing just fine. I
would like to thank Dr. Milam and Dr. Unger for accepting me into this program. Because of that
decision, I became very busy the past 5 years but also gained so much knowledge. Last, but
certainly not least, I would like to thank Dr. Menchine for being a remarkable boss allowing me
to power through this program while working full time at the Department of Emergency
Medicine. I was fortunate to have a unique opportunity where I experienced research firsthand
through a clinical perspective, and learned how to bring impact directly to patient care. All of
these experiences created a unique experience helping me to grow personally and professionally.
I would like to acknowledge the Health Behavior Research program for providing the
dedicated support, warm friendship and a nurturing environment during the doctoral training. I
am grateful to have Marny Barovich, one of the kindest persons I have met, who always
addressed our needs, regardless if big or small, with ease and a smile. Thank you to the friends I
have made at this program, especially Sydney O’Connor, Karen Ra, Jessica Tobin, Nicolas
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 3

Goldenson, and Patricia Escobedo for being in my cohort. We all moved through the challenges
together by providing each other support.
Finally, I am deeply grateful to my family for their support. I am thankful for my parents,
Sam and Catherine, who showed me, through example, that learning is life-long journey by
themselves enrolling in an undergraduate program after many years of their professional careers.
I am grateful to my in-laws, who continuously came to the US from China for extended periods
of time to help out with kids, meals and all family needs. I am thankful for my two kids, Jeriah
and Jasmine, for missing some play time from daddy while going over the literature for night
time reading. I own my greatest thanks to my wife Lena, who took on the greatest challenge of
becoming a full-time mom when our second child was born. Her love and dedication for the
family has given me the courage to move on and gain my success in the doctoral program.
Above all, I would like to thank God for giving me this opportunity, allowing me to continue my
Christian faith and use what I have from Him to serve the people in need.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 4

Table of Contents
List of Tables .................................................................................................................................. 7
List of Figures ................................................................................................................................. 8
Abstract……………………………………………………………………………………………9
Chapter 1: Introduction ................................................................................................................. 12
Background & Significance ...................................................................................................... 12
Federal Pain Research Priorities ........................................................................................... 15
Introduction to the Dissertation Studies.................................................................................... 16
Specific Aims and Hypothesis .............................................................................................. 18
Chapter 2: Impact of the prescription drug monitoring program on state prescription opioid
demand: A piecewise growth curve modeling approach ............................................ 21
Background ............................................................................................................................... 21
Introduction ............................................................................................................................... 23
Research Questions ................................................................................................................... 26
Specific Aims and Hypothesis .............................................................................................. 26
Methods .................................................................................................................................... 27
Study Design ......................................................................................................................... 27
Measures ............................................................................................................................... 28
Statistical Analyses ............................................................................................................... 30
Equations ............................................................................................................................... 32
Results ....................................................................................................................................... 34
Discussion ................................................................................................................................. 40
Study Limitations .................................................................................................................. 43
Conclusions ........................................................................................................................... 44
Chapter 3: Psychological mechanisms linking mindfulness disposition and pain: a multiple
mediation model of individuals with orthopedic problems ........................................ 45
Introduction ............................................................................................................................... 45
Conceptual Model ..................................................................................................................... 51
Research Questions ................................................................................................................... 51
Specific Aims and Hypothesis .............................................................................................. 51
Methods .................................................................................................................................... 52
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 5

Study Design and Study Sample ........................................................................................... 52
Measures ............................................................................................................................... 52
Statistical Analyses ............................................................................................................... 55
Equations ............................................................................................................................... 57
Results ....................................................................................................................................... 58
Discussion ................................................................................................................................. 65
Study Limitations .................................................................................................................. 67
Conclusions ........................................................................................................................... 68
Chapter 4: Effect of telephone call versus text message reminders on patient return to
acupuncture follow-up treatment: a randomized controlled trial ................................ 69
Background ............................................................................................................................... 69
Introduction ............................................................................................................................... 71
Research Questions ................................................................................................................... 74
Specific Aims and Hypothesis .............................................................................................. 74
Methods .................................................................................................................................... 75
Trial Design and Randomization .......................................................................................... 75
Conceptual Framework ......................................................................................................... 75
Setting and Participants ......................................................................................................... 76
Procedures ............................................................................................................................. 77
Interventions ......................................................................................................................... 78
Outcome and Assessments .................................................................................................... 78
Covariates ............................................................................................................................. 79
Statistical Analysis ................................................................................................................ 83
Equations ............................................................................................................................... 85
Results ....................................................................................................................................... 88
Discussion ............................................................................................................................... 101
Study Limitations ................................................................................................................ 104
Conclusions ......................................................................................................................... 104
Chapter 5: Summary and Conclusions ........................................................................................ 105
Summary of Aims and Findings ............................................................................................. 105
Implications and future Research ........................................................................................... 110
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 6

Limitations .............................................................................................................................. 113
Contribution to the Literature ................................................................................................. 114
Alphabetic Bibliography ............................................................................................................. 116
Appendix A. Comparison of MMEPC growth profiles, full dataset (1997-2014) and early period
(1997-2009)............................................................................................................... 128
Appendix B. MMEPC growth profiles in the pre-PDMP period by the three-level strength of
PDMP mandates, 1997-2014 .................................................................................... 129
Appendix C. Acupuncture Study: Baseline Questionnaire ......................................................... 130
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 7

List of Tables
Table 2-1. Prescription opioids and morphine milligram equivalent conversion factors ............. 29
Table 2-2. Summary table for PDMP and Pill Mill laws .............................................................. 35
Table 2-3. Comparison of pre- vs. post-PDMP MMEPC growth profiles and effects of state-
level covariates from the piecewise growth curve modeling, 1997-2014 .................. 37
Table 2-4. Comparison of pre- vs. post-PDMP MMEPC growth profiles at specific dispensaries
(hospital and pharmacy) and type of opioids (fentanyl and hydrocodone) ................ 38
Table 3-1. Participant characteristics (N=525) ............................................................................. 59
Table 3-2. Pairwise correlation matrix from observed data .......................................................... 61
Table 3-3. A cross-examination of indirect effects of five FFMQ facets through three DASS
domains on UPAT in multiple mediation analysis ..................................................... 64
Table 4-1: Baseline characteristics of participants by randomized group assignment ................. 90
Table 4-2: Exploratory analysis of factors predicting higher number of follow-up treatments in
30 days ........................................................................................................................ 94
Table 4-3: Secondary analysis predicting attendance to follow-up treatments (number of visits) in
30 days, focusing on the brief illness perception scale and acupuncture expectancy
scale domains .............................................................................................................. 96
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 8

List of Figures
Figure 2-1. Raw data plot and locally weighted scatterplot smoothing of morphine milligram
equivalent per capita (MMEPC) by Year across 45 States, 1997-2014 .................... 31
Figure 3-1. A mediation model examining the effect of mindfulness disposition on pain through
the psychological distress pathway ........................................................................... 51
Figure 3-2. Universal Pain Assessment Tool ................................................................................ 53
Figure 3-3. Mediation effect of psychological distress (DASS) between mindfulness disposition
(FFMQ) and pain (UPAT) ......................................................................................... 62
Figure 3-4. Proportion mediated by different distress domains between mindfulness disposition
and pain in a multiple mediation model .................................................................... 63
Figure 4-1. Study overview / conceptual framework of attendance to acupuncture follow-up
treatment .................................................................................................................... 76
Figure 4-2. Flow of randomized participants ................................................................................ 88
Figure 4-3. Adjusted follow-up treatment attendance rates by group assignment ....................... 92
Figure 4-4. Adjusted means of participant intent to attend follow-up treatment among the as-
treated sample over 30 days, by intervention assignments (N=94) ........................... 97
Figure 4-5. Adjusted mean levels of pain among the pain cohort over 30 days, by attendance to
follow-up treatment (n=73) ....................................................................................... 99
Figure 4-6. Adjusted mean levels of pain disability among the pain cohort over 30 days, by
attendance to follow-up treatments (n=73) ............................................................. 100

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 9

Abstract
Pain is a multidimensional problem and is widespread in the United States. In 2017, the
National Institute of Health released the Federal Pain Research Strategy to guide research on
better understanding pain and improving pain care. The document highlights research priorities
to examine pain from various perspectives, including policies, mechanisms, and integrative
treatments. These priorities target the forefront issues of the current pain epidemic: lacking a
comprehensive pain management system, having a poor understanding of the biological and
psychological mechanisms involving pain, as well as having limited treatment options for pain
conditions. It is of prominent significance to address these issues through research to help relieve
the burden of pain in America. The objectives of this dissertation were to: 1) evaluate the policy
impact of the Prescription Drug Monitoring Programs on reducing the demand of prescription
opioids, 2) examine the mechanisms linking mindfulness disposition and pain through the impact
on psychological distress, and 3) testing the effect of telephone call and text message reminders
on follow-up treatment attendance rates among acupuncture patients using a randomized
controlled trial.
Findings from Study 1 showed that the rates of prescription opioid shipments were
significantly reduced after states implemented a Prescription Drug Monitoring Program (PDMP).
We used the random effects piecewise growth curve modeling approach to compare the pre- and
post-PDMP trends, while controlling for the presence of Pill Mill law and timing of PDMP
implementation. States that had a Pill Mill law and implemented PDMP after 2010 demonstrated
a greater reduction in opioid shipments at the post-PDMP period (p<.001), adding to the effect of
the PDMP. Contrary to the hypothesis, the strength of PDMP mandate did not moderate the
trends of the post-PDMP period. This result suggested that states that adopted a stronger policy
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 10

did not result in greater opioid reduction. The findings support current literature showing PDMP
as an effective policy to address the opioid epidemic by reducing the overall demand of
prescription opioids. However, a more precise assessment of the policy mandate is needed,
including the use of actual provider registration and access data to the PDMP exchange.
Findings from Study 2 showed that psychological distress mediated the association
between mindfulness disposition and pain in orthopedic patients. Over 90% of the sample
experienced pain over the past two weeks, and 49% reported at least a mild level of depression,
anxiety or stress symptoms. We observed a significant correlation between overall mindfulness
disposition and pain (r=.-23, p<.01), but this direct effect disappeared in the mediation model
(beta=.01, p=.90) after psychological distress was included as the mediator (indirect effect
beta=.-22, p<.001). Results from the multiple mediation analysis showed that depression and
stress mediated 48.3% and 39.6% (p=.02 and p=.04, respectively) of the association between
mindfulness disposition and pain. The examination of specific mindfulness domains showed that
non-judging, non-reactivity and acting-with-awareness indirectly affected pain through their
influence on depression and stress. The overall finding supports a mindfulness-distress-pain (i.e.
attention-affect-pain) construct for a future study and specific pathways involved in pain
mechanisms.
The third study of this dissertation used a randomized controlled trial to test if adherence
strategies, including telephone call and text message reminders, could improve patient
attendance to acupuncture follow-up treatment. Contrary to our hypothesis, both one-time
interventions did not result in greater follow-up treatment attendance (i.e. measured by any
follow-up visit in 30 days after baseline) compared to no intervention control (56.3%, 57.3% and
57.0%, respectively). However, our exploratory analysis using the Poisson regression model
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 11

showed that being White, perceiving one’s illness to be more severe, having a higher expectation
of acupuncture treatment, and scheduling a follow-up appointment predicted higher number of
follow-up visits within 30 days after the initial treatment. For the longitudinal outcomes, pain
patients who returned for one or more follow-up visit demonstrated greater improvement in pain
disability over a 30-day period compared to patients with no follow-up, even though the
difference was marginal (p=.05). The findings provided evidence for the clinical values of
acupuncture follow-up treatment on pain-related outcomes, and the implication for designing
targeted interventions to improve follow-up attendance rate among acupuncture patients.
This dissertation makes important contributions to the body of research on pain from a
policy, mechanism and integrative treatment perspective. Overall, findings from the three studies
demonstrate that pain is a multidimensional issue and requires multilayer strategic approaches to
address the extent of concern. The study results show that on a societal level, public policy such
as the PDMP can have a significant impact on prescription opioid outcomes; on a mechanistic
view, psychological distress mediates the association between mindfulness disposition and pain;
and from a behavioral perspective, improving adherence to acupuncture treatment can result in
better pain outcomes. These findings address key priorities in the Federal Pain Research Strategy
to help advance the field of pain research.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 12

Chapter 1: Introduction
Background & Significance
Pain is prevalent in the United States and poses a significant burden to the American
economy. In 2012, 56% of U.S. adults experienced pain during the past three months; 10%
reported having more severe daily chronic pain symptoms (Nahin, 2015). The estimated
healthcare costs attributable to pain ranged from 560 to 635 billion U.S. dollars, and lost
productivity due to pain ranged from 299 to 335 billion dollars (Gaskin & Richard, 2012). These
combined costs was about 5% of the U.S. GDP in 2010 (World Bank, 2019). Due to the severe
magnitude of the pain epidemic, in 2011 the Institute of Medicine released a comprehensive
report to address pain from a broader, public health perspective (Institute of Medicine, 2011).
The report provides a wide range of recommendations that focus on four areas: pain prevention,
pain care, pain education and pain research. Six years following the report, the Interagency Pain
Research Coordinating Committee coordinated by the National Institute of Health consolidated
the findings and published the National Pain Strategy (NPS) in 2015 (Interagency Pain Research
Coordinating Committee, 2015), and the Federal Pain Research Strategy (FPRS) in 2017
(Interagency Pain Research Coordinating Committee, 2017). The two documents lay out the
long-term strategic plan and provide guidance to improve pain care and pain research in the U.S.
This dissertation uses specific FPRS research priorities to guide the development of three
independent studies focusing on drug policy, psychological mechanisms, and integrative
treatments targeting pain.
Pain is a multidimensional problem, affecting from a personal to a societal level (Gatchel,
Peng, Peters, Fuchs, & Turk, 2007). Of the problems addressed in the FPRS, a prominent and
immediate threat to the American public is the rapid increase in opioid misuse. Since the early
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 13

2000s, the U.S. has faced an unprecedented opioid crisis due to the surge in opioid addiction and
overdose deaths (Guy et al., 2017; Rudd, Aleshire, Zibbell, & Matthew Gladden, 2016). More
individuals with poorly controlled acute and chronic pain symptoms became dependent on
prescription opioids for pain relief. These individuals include patients with cancer in remission
and other non-malignant conditions, a growing population who continue to face malingering pain
despite treatment of their primary concerns. However, many were possibly undertreated for pain
due strict regulatory controls of opioids to prevent illegal use in the mid-19
th
century (Hill Jr,
1996). When opioid therapy was reestablished for the treatment of acute and cancer pain in late
1990s, it resulted in a mass increase in opioid prescribing (Frieden & Houry, 2016). The surge in
demand created opportunities for opioid diversion and overprescribing patterns. Although
policies such as the Prescription Drug Monitoring Programs (PDMP) exist to regulate
prescription drugs including the opioids (Finley et al., 2017; National Alliance for Model State
Drug Laws, 2017; Substance Abuse and Mental Health Services Administration, 2018), more
evidence is needed to support the effectiveness of these programs. With 49 states having an
operational PDMP in 2019, there is an opportunity to assess the impact of the PDMP policy on
the opioid crisis on a national level.
A key emphasis in the FPRS is to further understand the biopsychosocial mechanisms
revolving pain to help develop more specific treatment targets. Pain mechanisms are
complicated in nature, as the definition of pain has evolved over the past two decades (Gatchel et
al., 2007; Loeser & Melzack, 1999; van Hecke, Torrance, & Smith, 2013; Williams & Craig,
2016). Risk factors include biological (i.e. genetics, co-morbidities), psychological (i.e. emotion,
mental disorders), and social (i.e. demographics, lifestyle, behavior, treatment access). These risk
factors can uniquely or interact to affect the experience of acute pain, chronic pain and the
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 14

transition between the two phases, in particular in populations with specific health conditions.
For example, individuals with orthopedic problems often report musculoskeletal pain (Yelin,
Weinstein, & King, 2016). In addition to the biological causes of their pain symptoms, the
presence of psychological distress can alter the overall pain experience (C. S. Lin, Wu, & Yi,
2017; Ocanez, McHugh, & Otto, 2010; Theunissen, Peters, Bruce, Gramke, & Marcus, 2012).
The lack of proper management in stress, anxiety and depression can exacerbate pain
catastrophizing, and subsequently result in worse pain outcomes and impede long-term
rehabilitation success (Castillo, Archer, Newcomb, & Wegener, 2016; Castillo et al., 2013).
Mindfulness, a psychological disposition, is associated with lower levels of distress (Branstrom,
Duncan, & Moskowitz, 2011; K. W. Brown, Weinstein, & Creswell, 2012; de Frias & Whyne,
2015; Prakash, Hussain, & Schirda, 2015). As such, cultivating mindfulness may have a
protective role on pain, especially through reducing symptoms of depression, anxiety and stress
(Dorado et al., 2018; A. C. Lee et al., 2017; Poulin et al., 2016). Investigating the significance of
specific mindfulness-distress-pain pathways based on different mindfulness facets and distress
types will enhance the understanding of pain mechanisms and development of targeted mind-
body interventions.
The lack of effective pain treatment options has prompted the scientific community to
develop a more comprehensive approach to pain management (Qaseem, Wilt, McLean, Forciea,
& for the Clinical Guidelines Committee of the American College of, 2017). It includes
exploring and validating existing alternative therapeutic models, such as complementary health
approaches (CHA) (Nahin, Boineau, Khalsa, Stussman, & Weber, 2016). In 2011, the Joint
Commission revised its standards to recognize non-pharmacologic therapies including massage,
relaxation, chiropractic and acupuncture in the management of pain (Baker, 2017). There is
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 15

evidence showing chiropractic and acupuncture treatments are effective in managing back pain
and joint disorders (R. Chou et al., 2016; Y. Lin, Wan, & Jamison, 2017; Nahin, Boineau, et al.,
2016; Tice et al., 2017; Tick et al., 2017). However, access to CHA is a forefront issue, as less
than 20% of users have complete coverage for acupuncture, chiropractic or massage (Nahin,
Barnes, & Stussman, 2016b). With U.S. adults spending a total of $14.1 billion out-of-pocket
expenses on practitioner visits for CHA in 2012 (Nahin, Barnes, & Stussman, 2016a), there is a
clear demand of CHA for pain management, especially in a sustaining way that can allow users
to adhere to a long-term therapeutic plan for chronic pain conditions. While more private
insurances offer coverage for CHA (Tice et al., 2017), it is important to evaluate additional
adherence strategies for non-pharmacological treatments to improve treatment success rate. This
dissertation uses acupuncture as a model to examine whether a mobile reminder strategy can
affect patient adherence to acupuncture follow-up treatment and, subsequently, result in
improved pain outcomes.
Federal Pain Research Priorities
This dissertation uses specific FPRS priorities to develop a research agenda focusing on
pain (Interagency Pain Research Coordinating Committee, 2017). The first study examines the
role of the PDMP policy on the opioid crisis, in which responses to the FPRS call to “assess
effects of policy changes on pain care” (p.8). The second study assesses psychological
mechanisms linking mindfulness disposition and pain levels through the influence on
psychological distress, in which responses to the FPRS calls to “determine optimal approaches
for use of self-management strategies in chronic pain” (p.17), “determine the bidirectional
relationship between common comorbidities and chronic pain” (p.16), and “determine the
mechanisms that sustain or resolve chronic pain and which of these elements can be intrinsically
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 16

and extrinsically modulated” (p.16). The third study evaluates the impact of two mobile
intervention strategies, including text message and telephone call reminders, on patient
adherence to follow-up treatment and subsequent results on pain outcomes, in which responds to
the FPRS calls to “determine optimal safe and effective chronic pain management” (p.16) and
“evaluate efficacy, safety and interactions of new therapies” (p.4) , more specifically the
“determination of optimal dosing and adherence strategies for non-pharmacological treatments
should be included in this evaluation process” (p.5). Consolidating current evidence and new
findings on these diverse fields as addressed in the FPRS will offer the much needed multilateral
approach to tackle the pain epidemic in the U.S.
Introduction to the Dissertation Studies
This dissertation addresses pain research from three specific angles. First, we start from a
policy perspective to evaluate the impact of PDMP on prescription opioids. Specifically, we test
whether there has been a significant change on prescription opioid shipment trajectories across
45 states from 1997 to 2014 after the implementation of state PDMP. The years selected were in
the height of the rising opioid epidemic, with thousands of Americans suffering from opioid
addiction and overdose deaths. Second, we examine the role of distress, a common comorbidity
of pain from a mechanistic perspective. Specifically, we test whether the associations between
mindfulness facets and pain are mediated by specific distress types. The findings will help
identify pathways to modulate distress-related pain experience through the cultivation of
mindfulness facets. Third, we focus on acupuncture therapy and ways to improve patient
adherence to follow-up treatments. Specifically, we test the effect of telephone call and text
message reminders on follow-up return rates in a randomized controlled trial. In addition, we test
whether attendance to follow-up treatment will result in greater improvement in pain outcomes.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 17

In summary, the dissertation presents an overarching theme on pain, with a more specific
focus on three areas: impact of a drug policy on opioids, pain psychological mechanisms, and the
effect of an adherence strategy on non-pharmacologic pain treatment. The studies follow the
research priorities proposed in the Federal Pain Research Strategy as guiding principles. Findings
will add to the current evidence on pain research to build a more comprehensive pain
management system in the U.S. and across the globe.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 18

Specific Aims and Hypothesis
STUDY 1: Impact of the prescription drug monitoring program on state prescription
opioid demand: a two-piece growth curve modeling approach
Aim 1: To determine whether PDMP implementation decreases the rates of prescription opioid
shipments across states from 1997 to 2014.
Hypothesis 1: The implementation of PDMP will decrease the rates of prescription
opioid shipments across states from 1997 to 2014.
Aim 2: To test whether the strength of PDMP mandate at the state level moderates the effects of
PDMP on prescription opioid shipments.
Hypothesis 2: The strength of PDMP mandate at the state level will moderate the effect
of PDMP on prescription opioid shipments. States with weaker mandates will
demonstrate less effect of PDMP on opioid shipments.
STUDY 2: Psychological mechanisms linking mindfulness disposition and pain: a multiple
mediation model of individuals with orthopedic problems
Aim 1: To examine the association between mindfulness disposition and pain in orthopedic
patients.
Hypothesis 1: Higher mindfulness disposition is associated with lower levels of pain.
Aim 2: To examine if psychological distress mediates the effect of mindfulness disposition on
pain.
Hypothesis 2: Psychological distress mediates the effect of mindfulness disposition on
pain.
Aim 3: To examine whether the mediation in Aim 2 is driven by specific mindfulness domains
and distress types.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 19

Hypothesis 3: Mindfulness domains including non-judging and non-reactivity from the
Five Facet Mindfulness Questionnaire will affect pain by their influence on depression
and stress levels.
STUDY 3: Effect of telephone call versus text message reminders on patient return to
acupuncture follow-up treatment: a randomized controlled trial
Aim 1: To test whether telephone call or text message reminders to participants increases their
likelihood of attending acupuncture follow-up treatments within 30 days after their initial visit.
Hypothesis 1: Participants who receive a telephone call or text message reminder will
have a higher likelihood of attending at least one follow-up treatment visit than the no
intervention control group within 30 days after their initial visit.
Aim 2: To test whether participants who received the intervention report a higher level of intent
to attend follow-up treatment over a 30-day period.
Hypothesis 2: Participants who received the intervention will report a higher level of
intent to attend follow-up treatment over a 30-day period compared to no intervention
control group.
Aim 3: To examine clinical and psychosocial factors that are associated with more follow-up
visits.
Hypothesis 3: Treatment expectancy, severity of illness, and access to treatment will
predict a higher number of follow-up visits.
Aim 4: To evaluate whether attending follow-up visits is associated with reduction in pain levels
and pain disability among a subset of pain participants over a 30-day period.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 20

Hypothesis 4: Among participants who reported a pain condition, those who attended
follow-up visit(s) will demonstrate a higher reduction in pain levels and pain disability
over a 30-day period than those who did not attend follow-up.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 21

Chapter 2: Impact of the prescription drug monitoring program on state prescription
opioid demand: A piecewise growth curve modeling approach
Federal Pain Research Strategy Research Priorities:
 Develop, Evaluate, Improve Models of Pain Care
- Assess Effects of Policy Changes on Pain Care
Background
The recent opioid epidemic in the United States is like history repeating itself. Since the
early 16
th
century, there have been documentations about addiction to opium in different parts of
the world, including Europe, India, China and the U.S. After the active ingredient in opium,
Morphine was isolated in the early 1800s. The number of morphine addicts skyrocketed,
possibly linked to its common use for chronic pain relief. In response, alternative forms of
opiate, such as cocaine, was used as an experimental antidote to morphine addiction. It also
included the once-believed non-addictive opiate, heroin, which had become widespread as drug
industries started mass production. Adults and children with various ailments such as acute pain,
whooping cough and heart diseases were prescribed morphine and opium by medical
professionals without detailed knowledge regarding potential consequences of addiction. Until
1880-1990, the abuse of these additive forms of opioid were finally recognized as a major public
health problem, particularly in the U.S. (Markel, 2011)
Due to the rise of the abuse of opium and heroin, several U.S. states instituted legal
controls of narcotics in the early 1900s (Quinn & McLaughlin, 1972). The strict regulatory
controls restricted legal use of the drugs only as physician prescribed because of the concerns of
addiction (Ballantyne & Mao, 2003). Physicians became largely liable for these prescription
drugs, and could face prosecution if prescribed inappropriately. As a result, many physicians
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 22

became reluctant to prescribe opioid therapies to conditions such as acute pain and cancer pain,
and pain became largely undertreated (Rosenblum, Marsch, Joseph, & Portenoy, 2008). Until
1980, a letter written by Porter and Jick (1980) published at the New England Journal of
Medicine argued that opioid addictions were rare among those treated. Despite the lack of
evidence to support this claim, it prompted a substantive re-evaluation of using opioid treatments
for pain conditions (Leung, Macdonald, Stanbrook, Dhalla, & Juurlink, 2017). In 1984, Purdue
Pharma introduced MS Contin, a form of oxycodone and had misrepresented the risk of
addiction, resulting in serious abuse of the drug, including Oxycontin later in 1996 (Van Zee,
2009). The state laws to allow regular use of opioids for chronic pain, as well as the adoption of
pain as the fifth vital sign during that time period, further complicated the prescribing and usage
of opioids for pain, and potentially contributed to the current opioid crisis (Franklin et al., 2005;
Levy, Sturgess, & Mills, 2018).
Only starting from the early 2000s, the scale of the recent opioid epidemic has surfaced
as more individuals died from opioid overdosed. Due to the lack of funding for substance abuse
treatment and other biopsychosocial approaches to pain management, there was limited patient
access to properly control their pain. A National Center for Health Statistics report showed that
in 2008, death from drug overdoses, mainly from opiates, had surpassed motor vehicle traffic
deaths (Warner, Chen, Makuc, Anderson, & Minino, 2011). Deaths involving opioids was
estimated 15,000 in 2008, and had increased to 29,000 in 2014 (Rudd et al., 2016). To combat
the epidemic, the Trump Administration declared the opioid crisis a national Public Health
Emergency under federal law on October 26, 2017 (Hodge Jr, Wetter, Chronister, Hess, & Piatt,
2017). The declaration includes enhancing surveillance effort through the already existing state
Prescription Drug Monitoring Program to help prevent and address opioid abuse and overdose
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 23

(The White House, 2018). The following introduction will discuss current evidence on the
impact of this program on prescription opioids in the U.S in the past two decades.
Introduction
A 2017 Centers for Disease Control report highlighted a parallel increase in prescription
opioid-related deaths and opioid prescribing in the United States during 1999-2010 (Guy et al.,
2017). Opioid-related deaths continued to rise in the following years, reaching 42,000 overdose
deaths in 2016 (Scholl, Seth, Kariisa, Wilson, & Baldwin, 2019). However, opioid prescribing
rates began to decline after its peak in 2010, dropping from 782 morphine milligram equivalent
(MME) per capita to 640 MME per capita in 2015 (Guy et al., 2017). Some research has
attributed the decline to state policies and systems-level interventions that combat the opioid
epidemic (Clark, Eadie, Kreiner, & Strickler, 2012; Pezalla, Rosen, Erensen, Haddox, & Mayne,
2017). Although the evidence is mixed (Haegerich, Paulozzi, Manns, & Jones, 2014; Meara et
al., 2016; Paulozzi, Kilbourne, & Desai, 2011; Reisman, Shenoy, Atherly, & Flowers, 2009), one
of these policies is the Prescription Drug Monitoring Program (PDMP). This study examines the
effect of the implementation of PDMP laws on changes in opioid shipments from 1997-2014,
using a piecewise growth curve modeling approach that is novel to this field of research.
A PDMP is a state-run database that collects data on controlled substance prescriptions
dispensed by pharmacies and providers (Brandeis University; Centers for Disease Control and
Prevention, 2017; Clark et al., 2012). A PDMP‘s primary goal is to solve the information
asymmetry between patients and prescribers. It can help identify prescription drug misuse and
detect doctor shopping behaviors in patients, flag fraudulent or excessive prescription patterns by
pharmacies or providers, and prevent diversion of controlled substances. By 2017, 49 states have
an operational PDMP (Pezalla et al., 2017). State laws require pharmacists to report patient
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 24

prescription fills to the PDMP administrative agency. Pharmacists and prescribers, in addition to
other authorized entities on a state-by-state basis, can obtain patient prescription history of
controlled substances, and prescriber/dispenser information via the state PDMP exchange
(Substance Abuse and Mental Health Services Administration, 2018). The balance of research
shows that PDMPs reduced opioid prescribing, use and distribution. Findings included a 30%
decrease in Schedule II opioid prescribing by physicians (Bao et al., 2016), a 2.36kg reduction in
prescription opioid use on a month basis in Medicare beneficiaries (Moyo et al., 2017), and 8%
reduction of oxycodone shipments from manufacturers and distributors (Mallatt, 2017).
However, state laws that mandate registration and use of the database, and the presence of other
opioid-related policies vary across the U.S. (Haffajee, Jena, & Weiner, 2015; Pezalla et al.,
2017). Whether these factors affect the impact of PDMPs requires further examination.
In 2014, only 22 states had some level of mandate that required prescribers and
dispensers to register or use the PDMP database (National Alliance for Model State Drug Laws,
2014). A study showed that states with a mandatory statute demonstrated a significant reduction
in number of Schedule II opioid prescriptions by 9-10%, but not in states without the mandates
(Wen, Schackman, Aden, & Bao, 2017). In another study, the author ranked the state PDMP
program from a score of 1-5 based on statutory regulation and best practice (Pardo, 2017). The
study concluded that prescription opioid overdose deaths were reduced by 1% for every 1-point
increase in PDMP strength. For opioid policies, 8 states implemented the Pill Mill laws that
increase oversight of selected healthcare facilities, such as doctor’s offices, for inappropriately
prescribing and dispensing controlled substances (Mallatt, 2017; Rutkow et al., 2015). Rutkow et
al. (2015) evaluated the joint effects of PDMP and Pill Mill laws in Florida. They observed a
decreased in opioid prescribing by 1.4%, opioid volume by 2.5%, and MME per transaction by
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 25

5.6% in high level prescribers and patients. The effects of PDMP mandates and Pill Mill laws are
key factors to account for when evaluating the impact of PDMP on opioid trends.
Our study has two goals. First, we test whether prescription opioid shipment trajectories
were reduced after PDMP user access was implemented at a state-level during the 1997-2014
period. Second, we test whether the strength of PDMP mandates moderated the effect of PDMPs
on prescription opioid shipments. We hypothesize that PDMPs significantly reduced the
trajectories of opioid shipments, and a stronger PDMP mandate would result in greater shipment
reduction. We apply a novel analytic strategy in the field of policy research by using the random
effect piecewise growth curve modeling approach. As Mallatt (2017) suggested, current analyses
employing fixed effects modeling can be limited in model specifications, specifically violating
the parallel-trend assumption. Therefore, our study using a different statistical approach may
offer additional justification to confirm PDMPs’ effectiveness in controlling the opioid epidemic.
In addition, we compare the changes in opioid shipments to hospitals and pharmacies, and on
selected opioids (i.e. fentanyl and hydrocodone) to determine the policy’s specific effects.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 26

Research Questions
Question 1: Does the implementation of PDMP reduce the demand of prescription
opioids during 1997-2014?
Question 2: Does the strength of PDMP mandate moderate the effects of PDMP on
prescription opioid shipments?

Specific Aims and Hypothesis
Aim 1: To determine whether PDMP implementation decreases the rates of prescription
opioid shipments across states from 1997 to 2014.
Hypothesis 1: The implementation of PDMP will decrease the rates of prescription
opioid shipments across states from 1997 to 2014.
Aim 2: To test whether the strength of PDMP mandate at the state level moderates the
effects of PDMP on prescription opioid shipments.
Hypothesis 2: The strength of PDMP mandate at the state level will moderate the effect
of PDMP on prescription opioid shipments. States with weaker mandates will demonstrate less
effect of PDMP on opioid shipments.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 27

Methods
Study Design
Prescription opioid data were queried from the Automation of Reports & Consolidated
Orders Systems (ARCOS) during the 1997-2014 period. ARCOS is a federal surveillance system
of all controlled substance shipments within the U.S., and is maintained by the Drug
Enforcement Administration (DEA) Office of Diversion (Drug Enforcement Administration).
We use the time-series data from ARCOS to compare differences in opioid shipment trajectories
before and after PDMP implementation using the piecewise growth curve modeling. Given the
model specification requires at least 2 years of data to measure the trajectory at each period, we
included 45 states in our analysis, excluding 2 states that only either the pre- or post-PDMP
period due to the implementation year on 1997 or 2014 (New Hampshire, Maryland), and three
states with that year outside of the data range (i.e. Hawaii, Missouri, Pennsylvania). Specifically,
we operationalized PDMP year of implementation using the year that states granted prescriber /
user access to the electronic database, which will be used as the state PDMP implementation
reference point in this study and mentioned as the “year of access” throughout the manuscript
and tables (Mallatt, 2017).
Mandates on prescribers or delegates to register and access the PDMP database varied
across states. We summarized three levels of PDMP strength using documentation from Wen et
al. (2017). We included the presence of a Pill Mill law in each state in our model to account for
the effect of non-PDMP policies on change in the opioid trends. The literature and plotting of
raw data suggested that there is a possible national decline of prescription opioids starting in
2010. We included a state-level interaction term for states with PDMP implemented before and
after 2010 to account for this time effect.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 28

Measures
Outcome. The demand for prescription opioids is operationalized by the yearly per capita
state-level shipments of opioids from manufacturers and distribution warehouses to pharmacies
and hospitals in the U.S. during 1997-2014. The study included 18 prescription opioids
continuously available in the ARCOS dataset using the following morphine milligram equivalent
(MME) conversion factors (Table 2-1): alfentanil (15), codeine (0.15), dextropropoxyphene
(0.1), dihydrocodeine (0.25), fentanyl base (100), hydrocodone (1), hydromorphone (4),
levorphanol (11), meperidine/pethidine (0.1), methadone (3), morphine (1), opium (1),
oxycodone (1.5), oxymorphone (3), remifentanil (200), sufentanil (500), sufentanil base (500),
and tapentadol (0.4). The study outcome is MME per capita (MMEPC), normalized by state
population per the specific year. Our dataset contain 810 MMEPC observed values by the 45
states in the analysis.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 29

Table 2-1. Prescription opioids and morphine milligram equivalent conversion factors
Prescription Opioids (18 total) MME conversion factor
ALFENTANIL 15
CODEINE 0.15
DEXTROPROPOXYPHENE 0.1
DIHYDROCODEINE 0.25
FENTANYL BASE 100
HYDROCODONE 1
HYDROMORPHONE 4
LEVORPHANOL 11
MEPERIDINE (PETHIDINE) 0.1
METHADONE 3
MORPHINE 1
OPIUM 1
OXYCODONE 1.5
OXYMORPHONE 3
REMIFENTANIL 200
SUFENTANIL 500
SUFENTANIL BASE 500
TAPENTADOL 0.4
MME: Morphine milligram equivalent
Source:
1. https://www.cms.gov/Medicare/Prescription-Drug-
Coverage/PrescriptionDrugCovContra/Downloads/Opioid-Morphine-
EQ-Conversion-Factors-March-2015.pdf
2. http://olh.ie/wp-content/uploads/2014/09/The-Use-of-Alfentanil-in-
a-Syringe-Driver-in-Palliative-Medicine.pdf
3. http://palliative.org/NewPC/_pdfs/tools/INSTRUCTIONsMEDD.pdf

Predictors. The key predictors are PDMP status and year (Moyo et al., 2017). First, we
tested whether the rates of MMEPC increase significantly over time in the pre- and post-PDMP
periods. The pre-PDMP period (P1) was operationalized as the years leading up to and including
the PDMP year of access starting 1997. The post-PDMP period (P2) was operationalized as the
years following the PDMP year of access until 2014. After testing whether there was a
significant growth in MMEPC at each period, we tested whether the slopes for P1 and P2 were
different statistically using post-estimation.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 30

Moderator. We defined PDMP strength based on a function of whether there is 1)
mandated registration, 2) mandated access, or 3) requirement of both (Wen et al., 2017). A weak
PDMP mandate is when 1) and 2) were not met; an intermediate is when 1) or 2) was met, and a
strong mandate is when 1) and 2) were met. The three-level PDMP strengths for each state are
listed in Table 1.
Covariates. The model controlled for two state-level factors interacted with P2. The first
factor was the presence of the Pill Mill law (Mallatt, 2017). We included an interaction term
between P2 and states with an existing Pill Mill law before or during the post-PDMP period to
account for the Pill Mill law effects on MMEPC. The second factor was whether states granted
PDMP access before or after 2010 (early versus late adopter effect). We included an interaction
term between P2 and pre- vs post-2010 PDMP access to account for the timing of
implementation that may coincide with changes in MMEPC.
Statistical Analyses
We first performed a raw data time plot overlaying with the locally weighted scatterplot
smoothing (LOWESS) – a non-parametric least square fitting that fits a smooth curve to data
point – to show the observed trend of prescription opioids during 1997-2014 (Figure 2-1). We
observed three data point outliers from the scatter plot and they were removed from the analysis
(i.e. Delaware on 2011, South Dakota in 2011, South Carolina in 2013). For the primary
analysis, we used the piecewise growth curve modeling approach to test whether the rates of
prescription opioid shipments (MMEPC) change between the pre- and post-PDMP periods (C. P.
Chou, Yang, Pentz, & Hser, 2004). A piecewise growth curve model uses random effect
multilevel modeling to estimate within-state and between-state variances. The estimates of the
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 31

slopes of the two growth profiles were compared using post-estimation for significant difference.
For this two-growth profile application, it is considered a two-piece growth curve modeling.
Figure 2-1. Raw data plot and locally weighted scatterplot smoothing of morphine
milligram equivalent per capita (MMEPC) by Year across 45 States, 1997-2014

We included the two state-level covariates, pre- vs. post-2010 PDMP access and Pill Mill
law into the model separately to observe how each affects the coefficients. Then, we tested
whether the strength of PDMP moderated its impact on the MMEPC trends. We conducted
secondary analyses to examine the effects of PDMP on 1) prescription opioids dispensed by
hospital and pharmacy, and 2) fentanyl and hydrocodone. The LOWESS plot was computed in
Stata 13 (StataCorp, 2013), and two-tailed significant tests were computed in SAS 9.4 using
PROC MIXED with α set to 0.05 (SAS, 2014).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 32

Equations
The two-piece growth curve model uses the following multilevel modeling equations
(Level-1: within-state, Level-2: between-state) which then formulates the combined equation.
This equation represents Model 3 in Table 2-3:
Level-1: mmepcti = B0i Int1i+ B1iPeriod1ti + B2i Int2i + B3iPeriod2ti + eti
Level-2: B0i = a1 +  01 Z1i +  02 Z2i + U0i
B1i = b1 +  11 Z1i +  12 Z2i + U1i
B2i = a2 +  21 Z1i +  22 Z2i + U2i
B3i = b2 +  31 Z1i +  32 Z2i + U3i
Combined equation (Reduced form):
mmepcti = (a1 +  01 Z1i +  02 Z2i + U0i) Int1i + (b1 +  11 Z1i +  12 Z2i + U1i) Period1ti
+ (a2 +  21 Z1i +  22 Z2i + U2i) Int2i + (b2 +  31 Z1i +  32 Z2i + U3i) Period2ti + eti
Level-1: within-state
Level-2: between-state
t = time (time centered at PDMP year of operation)
i = state
B0i = (within-state) initial status of pre-period
B1i = (within-state) growth trajectory of pre-period
B2i = (within-state) initial status of post-period
B3i = (within-state) growth trajectory of post-period
eti = normally distributed with mean 0 and variance σ²e
a1 = (between-state) fixed initial status of pre-period
U0i = (between-state) initial status variance of pre-period
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 33

b1 = (between-state) fixed growth trajectory of pre-period
U1i = (between-state) growth trajectory variance of pre-period
a2 = (between-state) fixed initial status of post-period
U2i = (between-state) initial status variance of post-period
b2 = (between-state) fixed growth trajectory of post-period
U3i = (between-state) growth trajectory variance of post-period
Z1i = (between-state) Pill Mill law at state-level
Z2i = (between-state) PDMP implementation before or after 2010 at state-level
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 34

Results
Of the 45 states in the analysis, the first PDMP year of access was in 2004 (Table 2-2).
The average length of the pre-PDMP period (P1) was 13 years (range 8-17 years) and the post-
PDMP period (P2) was 5 years (range 1-10 years). In 1997, the first year of the data, the average
MMEPC crude rate started at 105.6 (SD: 43.0) (Intercept 1); there was a seven-fold increase
across states when P2 began (Intercept 2), reaching 736.1 (SD: 190.3). The average unadjusted
rate increase in P1 was 54.5 (SD: 18.1) MMEPC per year. Figure 2-1 shows a possible national
declining trend in MMEPC starting 2010. Raw data from Table 2-2 indicates that the P2 slopes
of 23 out of 25 pre-2010 access states were positive, with average rate at 19.2 (SD: 15.7)
MMEPC per year. In contrast, for the 20 states that had a post-2010 access, the average slope of
P2 was -26.9 (SD: 39.5). Although the overall unadjusted rate of the post-PDMP period was
negative (-1.3, SD: 36.7), further adjustment is needed to account for external effects other than
the PDMP policy. These include the pre-/post-2010 PDMP access time, presence of the Pill Mill
laws (8 states), and the strength of PDMP mandates to be controlled in the piecewise growth
curve models.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 35

Table 2-2. Summary table for PDMP and Pill Mill laws
State
PDMP
Year of
Access
PDMP
Strength
Slope
MMEPC
Pre-PDMP
Slope
MMEPC
Post-PDMP
Pill Mill
Law
Year
NEVADA 2004 Intermediate 95.7 32.0

WEST VIRGINIA 2004 Strong 77.8 43.5 2014
WYOMING 2004 Weak 50.2 28.1

KENTUCKY 2005 Strong 59.9 31.6 2011
MAINE 2005 Intermediate 70.0 24.5

NEW MEXICO 2005 Strong 42.2 29.8

OHIO 2006 Strong 68.1 0.4 2011
OKLAHOMA 2006 Intermediate 73.4 39.4

TENNESSEE 2006 Strong 101.8 44.3 2012
UTAH 2006 Intermediate 70.1 10.7

VIRGINIA 2006 Weak 40.2 23.2

ALABAMA 2007 Intermediate 71.1 26.3

NORTH DAKOTA 2007 Intermediate 37.8 16.3

ARIZONA 2008 Strong 63.7 15.6

COLORADO 2008 Strong 47.7 9.6

CONNECTICUT 2008 Intermediate 59.3 4.3

IDAHO 2008 Intermediate 47.0 33.4

ILLINOIS 2008 Weak 27.7 5.5

INDIANA 2008 Intermediate 56.4 14.5

NORTH
CAROLINA 2008 Intermediate 54.5 29.2

SOUTH
CAROLINA 2008 Weak 54.8 31.5

CALIFORNIA 2009 Intermediate 42.2 -8.7

IOWA 2009 Weak 34.1 2.3

LOUISIANA 2009 Intermediate 64.7 2.7 2005
VERMONT 2009 Strong 58.1 -11.2

MASSACHUSETTS 2010 Strong 40.4 -27.2

MINNESOTA 2010 Intermediate 30.7 -12.4

FLORIDA 2011 Weak 87.8 -158.2 2011
KANSAS 2011 Weak 53.5 0.5

MICHIGAN 2011 Weak 52.1 -10.0

MISSISSIPPI 2011 Strong 48.0 4.0 2011
NEBRASKA 2011 Weak 27.7 -6.3

OREGON 2011 Weak 77.7 -57.0

ALASKA 2012 Weak 38.2 -17.1

DELAWARE 2012 Strong 87.5 -77.7

MONTANA 2012 Weak 55.9 -40.6

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 36

NEW JERSEY 2012 Weak 49.8 -19.1

RHODE ISLAND 2012 Strong 44.2 -65.0

SOUTH DAKOTA 2012 Weak 40.2 18.7

TEXAS 2012 Weak 25.5 -16.3 2009
WASHINGTON 2012 Intermediate 51.6 -32.7

ARKANSAS 2013 Weak 53.8 12.3

GEORGIA 2013 Intermediate 39.5 -15.9

NEW YORK 2013 Intermediate 37.1 -8.3

WISCONSIN 2013 Weak 44.2 -10.5

PDMP Year of Access and Pill Mill laws from Mallatt 2017. PDMP Strength from Wen 2017
PDMP: Prescription Drug Monitoring Program
MMEPC: Morphine Milligram Equivalent Per Capita
PDMP Year of Access: year that states granted prescriber/user access to the PDMP electronic
database

Our findings from the two-piece growth curve models show that the rate of MMEPC
decreased significantly when comparing the pre- and post-PDMP period across states during
1997-2014 (Table 2-3). Model 1, the base model without covariates, shows that the rate increase
was at 54.2 (SE: 2.7) MMEPC per year during P1. Although the rate increase at P2 had
significantly declined to 9.9 (SE: 3.2) MMEPC, a 44.3 unit reduction from P1 (p<.001), the
growth remained positive. Model 2, adjusting for pre-/post-2010 access interacted with P2,
demonstrates that states with post-2010 PDMP access have a 23.9 units (SE: 6.1) further
reduction in MMEPC at P2 (p<.01). Model 3, which includes the Pill Mill law interaction with
P2, shows that states with Pill Mill law had a further decline in growth of MMEPC by 28.0 units
(SE: 5.2, p<.001). Model 4 tests our second hypothesis of the effect of the strength of PDMP
mandates on MMEPC trend at P2. However, the result shows that neither state with intermediate
nor strong PDMP mandates had further reduction in their MMEPC rate compared to states with a
weak PDMP policy.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 37

Table 2-3. Comparison of pre- vs. post-PDMP MMEPC growth profiles and effects of
state-level covariates from the piecewise growth curve modeling, 1997-2014
Model 1 Model 2 Model 3 Model 4
Intercept 1 107.8 (6.7) 94.8 (8.9) 94.2 (8.9) 89.9 (14.4)
Intercept 2 727.5 (26.5) 727.3 (27.4) 720.9 (26.4) 716.6 (28.6)
Slope Period 1 54.2 (2.7)*** 54.2 (2.7)*** 54.2 (2.7)*** 54.2 (2.7)***
Slope Period 2 9.9 (3.2)** 12.0 (3.0)*** 14.5 (3.6)*** 14.6 (6.5)*

Post-2010 access

28.9 (12.5)* 30.2 (12.4)* 32.6 (13.9)*
Pill Mill law

2.8 (25.8) 1.9 (25.8)

Post-2010 access x Period 2

-23.9 (8.4)** -18.6 (8.3)* -19.6 (8.7)*
Pill Mill law x Period 2

-28.0 (5.2)*** -28.7 (5.3)***

Strength of PDMP
Mandates

Intermediate vs Weak

6.9 (15.8)
Strong vs Weak

2.8 (16.8)

Intermediate x Period 2

-3.3 (7.4)
Strong x Period 2

5.9 (7.8)

Slope Period 2 - Period 1 -44.3 (3.7)*** -42.2 (3.8)*** -39.7 (3.8)*** -36.7 (6.3)***

N, States 807 (45 states) 807 (45 states) 807 (45 states) 807 (45 states)
Chi-Square 1475.6*** 1475.8*** 1516.3*** 1438.4***
Beta (S.E.) shown at each cell. *p<.05, **p<.01, ***p<.001
3 outliers data point excluded (from 810 state-year observation to 807)
PDMP: Prescription Drug Monitoring Program, MMEPC: Morphine Milligram Equivalent Per
Capita
Pre-PDMP: Intercept 1 and Slope Period 1; Post-PDMP: Intercept 2 and Slope Period 2
Model 1: base model with intercepts and slopes
Model 2: control for post-2010 PDMP access interaction with Period 2
Model 3: control for post-2010 PDMP access and Pill Mill laws interactions with Period 2
Model 4: test for effect of PDMP mandates on Period 2
Post-2010 access: state granted PDMP prescriber/user access after 2010
Chi-Square represents the null model likelihood ratio test

As secondary analyses, in Table 2-4 we show the PDMP effect on the type of dispensary
affected (pharmacy and hospital) and on distribution of specific opioids in terms of change in
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 38

MMEPC volume. Controlling for covariates, results from the two-piece growth curve models
show that for pharmacy-dispensed opioids, the MMEPC rate increase in P2 (M=15.1, SE=3.6)
was significantly slower than in P1 (M=50.6, SE=2.7) (p<.001). For hospital-dispensed opioids,
the P1-P2 rate difference was also different (p<.001). However, unlikely the continued growing
trend of P2 in the pharmacy-dispensed opioids, the P2 trend in the hospital-dispensed opioids
showed an overall downward slope (M=-0.8, SE=0.4, p=.06).
Table 2-4. Comparison of pre- vs. post-PDMP MMEPC growth profiles at specific
dispensaries (hospital and pharmacy) and type of opioids (fentanyl and hydrocodone)

Hospital
MMEPC
Pharmacy
MMEPC

Fentanyl
MMEPC
Hydrocodone
MMEPC
Intercept 1 22.0 (1.7) 71.5 (8.1)

18.0 (2.2) 25.7 (2.7)
Intercept 2 62.9 (4.2) 657.2 (25.9)

157.7 (6.2) 125.6 (10.7)
Slope Period 1 3.5 (0.2)*** 50.6 (2.7)***

12.7 (0.6)*** 8.1 (0.7)***
Slope Period 2 -0.8 (0.4) 15.1 (3.6)***

-1.2 (1.1) 1.5 (0.8)

Post-2010 access 8.9 (2.4)** 23.6 (11.1)

11.7 (2.7)*** 4.7 (3.1)
Pill Mill law -3.7 (5.1) 6.2 (24.3)

-8.7 (8.3) 9.0 (5.6)

Post-2010 access x
Period 2 -2.3 (1.4) -16.1 (7.9)

-0.7 (2.4) -0.8 (1.8)
Pill Mill law x Period 2 0.4 (0.9) -29.0 (5.0)***

-1.7 (1.8) -6.8 (1.2)***

Slope Period 2 - Period 1 -4.2 (0.5)*** -35.5 (3.8)***

-13.9 (1.3)*** -6.6 (0.9)***

N, States 807 (45 states) 807 (45 states)

807 (45 states) 807 (45 states)
Chi-Square 796.8*** 1574.5***

598.2*** 2026.3***
Beta (S.E.) shown at each cell. *p<.05, **p<.01, ***p<.001
PDMP: Prescription Drug Monitoring Program
MMEPC: Morphine Milligram Equivalent Per Capita
Pre-PDMP: Intercept 1 and Slope Period 1; Post-PDMP: Intercept 2 and Slope Period 2
Post-2010 access: state granted PDMP prescriber/user access after 2010
Chi-Square represents the null model likelihood ratio test

Among selected opioids, fentanyl (M=-13.9, SE=1.3) and hydrocodone (M=-6.6, SE=0.9)
both show a significant reduction in P2 compared to P1 (p<.001). The growth profiles of P2 in
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 39

both opioids were insignificant, meaning the growth in MMEPC has paused (fentanyl: M=-1.2,
SE=1.1; hydrocodone: M=1.5, SE=0.8, p=.27). Pill Mill law showed no significant effect on
hospital-dispensed opioids (M=0.4, SE=0.9, p=.69) but in pharmacy-dispensed opioids (M=-
29.0, SE=5.0, p<.001), consistent with their intended effects, as well as no effect on fentanyl in
P2 (M=-1.7, SE=1.8, p=.69) but significant effect in reducing hydrocodone (M=-6.8, SE=0.9,
p=.35). We also tested whether strength of PDMP affected these subgroups; however, the results
were insignificant and therefore not included in the models.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 40

Discussion
Findings from our piecewise growth curve models on a national dataset show that there
was an overall reduction in prescription opioid shipments across states after PDMP user access
became available during 1997-2014. However, the overall trend remained upward even though
PDMPs were in effect. The strength of PDMP policy mandates did not moderate a PDMP’s
effect on opioid shipments, such that the states implementing an intermediate or strong PDMP
offered no further reduction of MMEPC in the post-PDMP period. We observed a flat growth of
opioid shipments to hospitals and, specifically, in fentanyl and hydrocodone at the post-PDMP
period. States that granted PDMP access after 2010 showed a weaker growth of opioid shipments
in the post-PDMP period, which suggests an implementation timing effect that may coincide
with other changes related to prescription opioids nationwide. States that adopted the Pill Mill
law also showed a significant reduction in prescription opioid shipments, in particular those to
the pharmacies. These results provide evidence to support an overall association between PDMP
implementation and reduction in opioid shipments during the study period.
We observed a reduction in prescription opioid demand after PDMP implementation,
which is consistent with some literature. Bao et al. (2016) found that PDMP implementation was
associated with 30% reduction in the prescribing of Schedule II opioids, while Moyo et al.
(2017) showed similar PDMP effects in the use of prescription opioids among Medicare
beneficiaries. However, other studies also showed no effect on PDMPs (Meara et al., 2016).
Paulozzi et al. (2011) and Reisman et al. (2009) analyzed PDMPs’ effect on opioids during
1997-2005, a period with the fastest growth of opioids in the nation (Guy et al., 2017). An
unintended effect is that their early assessment on PDMPs were selective to those early adopter
states, in which the implementation strategies and actual provider registration and use of the
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 41

PDMP database exchange could be limited (Haffajee et al., 2015). To assess the early PDMP
effects from our modeling approach, we conducted a sensitivity analysis by restricting our data
to 1997-2009, a year prior to the peak of opioid shipments at 2010. We observed a smaller
difference in opioid shipment trajectories comparing pre- vs. post-PDMP periods relative to our
full data, but the difference remained statistically significant. This might suggest that the data
source or model specification might have affected the finding during this early period of PDMP
effect. (Appendix A)
Our result demonstrates contrary findings to the current literature with regards to the
effect of the strength of PDMPs on opioids. Wen’s study found a 9-10% reduction in opioid
prescriptions in states with policy mandates versus states without (Wen et al., 2017). As Wen et
al. (2017) noted, the lack of effect could likely be due to the states that adopted the mandates
were those facing a more serious opioid epidemic and needed a more robust policy to combat the
growth. In particular, the growth could continue into the post-PDMP period and could be harder
or take longer for the state PDMP to show effects. We further evaluated our data and observed
that states with a strong or intermediate mandate trended toward a greater increase in opioid
shipments prior to their PDMP implementation, although the interactions were not significant
(Appendix B). An advantage using the piecewise growth curve model is that while estimating
the growth profile of the post-PDMP period, the model adjusted for the growth profile of the pre-
PDMP period. To better assess the effect of the strength of PDMP, a longer post-PDMP data
period is needed to generalize the overall impact of the policy.
For the timing of the PDMP implementation, states that granted access after 2010 showed
a much slower growth in opioid shipments. This finding supports that there were additional
factors influencing the change, specifically during the later period of the study time frame. This
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 42

might include policies that target opioid prescribing and use (Meara et al., 2016), as well as an
increased awareness among the public and professionals of the opioid epidemic (Guy et al.,
2017). Future research may account for these factors to provide a more robust assessment of
PDMPs as the model specification allows. For the Pill Mill law, it’s impact on opioid shipments
distributed to pharmacies (vs. hospitals) has confirmed the law’s intent to address pain
management clinics that had inappropriately prescribed and dispensed opioids (Lyapustina et al.,
2016; Rutkow et al., 2015). Accounting for policy timing and presence of the Pill Mill law help
strengthen the evaluation of the actual PDMP effects.
We observed a flat growth in fentanyl and hydrocodone in the post-PDMP period, even
though the shipment volume of both drugs grew significantly in the pre-PDMP period. For
fentanyl, the change could be linked to the increased fentanyl-related overdose deaths
(Manchikanti et al., 2018). Although these deaths were mostly associated with illicit fentanyl
instead of prescription fentanyl (Seth, Rudd, Noonan, & Haegerich, 2018), it may result in an
overall chilling effect on provider’s prescribing patterns. For hydrocodone, it is among the most
common opioids abused by opioid addicts in the U.S. (Cicero, Ellis, Surratt, & Kurtz, 2013). A
tighter regulatory oversight of controlled substances through the PDMPs may discourage doctor
shopping behaviors of this drug, as well as having providers to reconsider additional pain
management approaches and more appropriate use of opioids (Cicero, Ellis, & Kasper, 2017).
We display the results of these two prescription opioids given they play a major role in the
opioid crisis in terms of opioid overdose deaths and opioid abuse. We also recognize the key
influence of oxycodone in the epidemic; however, due to its reformulation in 2010 and ongoing
law suits involving its drug producing company (Severtson et al., 2013; Webster, 2012), it would
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 43

be difficult to assess PDMP’s effect. Our results, therefore, are consistent with current findings
of PDMPs on fentanyl and hydrocodone specifically.
Study Limitations
Our study has several limitations. We used prescription opioid shipment data from
ARCOS to proxy for the demand of prescription opioids during 1997-2014 in the U.S. As
discussed in Piper, Shah, Simoyan, McCall, and Nichols (2018) and Reisman et al. (2009), the
ARCOS dataset does not account for medications from outside of the U.S. in several states,
excluded Schedule IV or V opioids, and captures only prescription opioids detected under the
surveillance system. These factors likely underestimate the amount of prescription opioid
consumed in each state during the study period. As Paulozzi et al. (2011) suggested, using state
aggregated MME retains its ecologic finding only on state but not individual level. Wen et al.
(2017) also noted that mandates based on state policy information do not accurately reflect actual
physician prescribing pattern unless using data that directly reflects physician PDMP registration
and use. In addition, some states adopted the Pill Mill law prior to their state PDMP
implementation. However, our modeling only accounted for the law’s impact at the post-PDMP
period, which could underestimate its effects prior to PDMP. This affected two states (i.e.
Louisiana and Texas), while the Pill Mill / PDMP effects were unable to parse out for states that
had both policy / law implemented in the same year (i.e. Florida and Mississippi). The results of
the Pill Mill laws, therefore, should be interpreted with caution to the limitations.
Despite these limitations, our study utilized the piecewise growth curve modeling
technique, a novel random effects modeling approach compared to existing studies in this field
primarily using fixed effects analysis. An advantage of this modeling technique is that it allows
for higher degree of freedom (Jones, Wright, & Bell, 2013). The conventional approach to
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 44

analyze time-series data is to include the within (i.e. year) and between (i.e. state) fixed effects
dummies, which would lead to a highly conservative estimate in a fully saturated model. Second,
a random effects model can account for state-level factors, in this case the strength of PDMP
mandates, given in a fixed effects model the state fixed effects has accounted for the state-level
variances. In addition, the model specification using a growth curve approach can allow for
comparison between the pre- vs post-PDMP trends instead of the averaged difference. As
described in (Reisman et al., 2009), other confounders such as state policies targeting opioids
could affect prescription opioid trends; however our model is limited with state-level units and
therefore we only accounted for the Pill Mill law. With more papers supporting the use of
random effects modeling strategies for time-series multilevel data, our application will set a good
example in policy research (Bell & Jones, 2015; Jones et al., 2013).
Conclusions
Our study found that there is an overall reduction in prescription opioid shipments after
states implemented a PDMP during 1997-2014. The result has potential implications for the
effectiveness of the PDMPs on lowering the demand of prescription opioids across the U.S.
Whether PDMPs have effectively decreased the use of opioids, curbed opioid diversion, limited
doctor shopping behaviors, or restricted excessive prescribing patterns will require further
research to determine. Although our result suggests that the strength of PDMP mandates has no
additional impact on opioid shipment volumes, it could be due to states implementing a stronger
policy were facing a more severe opioid epidemic. The application of the random effects
piecewise growth curve modeling approach provides an alternative analytic strategy in policy
research. The study findings build on the current evidence supporting the PDMPs as an effective
policy to tackle the opioid crisis.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 45

Chapter 3: Psychological mechanisms linking mindfulness disposition and pain: a multiple
mediation model of individuals with orthopedic problems
Federal Pain Research Strategy Research Priorities:
 “Determine optimal approaches for use of self-management strategies in chronic pain”
 “Determine the bidirectional relationship between common comorbidities and chronic pain”
 “Determine the mechanisms that sustain or resolve chronic pain and which of these elements
can be intrinsically and extrinsically modulated”
Introduction
Pain is a leading complaint in orthopedic patients (Picavet & Hazes, 2003; Woolf &
Pfleger, 2003; Yelin et al., 2016). The musculoskeletal pain patient experiences is often a
physiological response to tissue damage, but the nature of pain experience can also involve
sensory, affective and behavioral dimensions (Adams, Rose, & Ravey, 1997). In the affective
domain, psychological features of mood and attention may explain variance in levels of patient
pain (Castillo et al., 2016; Day, Jensen, Ehde, & Thorn, 2014). However, psychological distress,
an affective state which is often presented in orthopedic patients, is often underdetected in
clinical settings (Vincent, Horodyski, Vincent, Brisbane, & Sadasivan, 2015). Mindfulness
disposition, a mental state of present-focused attention, may affect the perception of pain by
offering a different interpretation to pain-related psychological responses (Day et al., 2014). This
study integrates pain and mindfulness theoretical constructs to test a proposed attention-
affective-pain mediation model in a clinical cohort of orthopedic patients. The goal is to
understand whether psychological distress mediate the association between mindfulness
disposition and pain intensity, and the specific pathways involved in the process.

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 46

The Role of Psychological Distress on Pain Experience
Psychological distress plays an important role in the experience of pain. The general
concept of psychological distress is often embedded within the context of strain, stress, and
distress (Ridner, 2004). Ridner (2004) refines the definition to “the unique discomforting,
emotional state experienced by an individual in response to a specific stressor or demand that
results in harm, either temporary or permanent, to the person” (p.539). General representation of
these emotional states includes symptoms of depression, anxiety and stress (Henry & Crawford,
2005). In the experience of pain, these emotions can be the stressor, or harm in Ridner (2004)’s
definition. A longitudinal study of patients who suffered from lower extremity injury showed a
bi-directional relation between anxiety and pain (Castillo et al., 2013). Anxiety persisted to affect
pain throughout the study and the correlations became more intense as time progressed
(standardized beta: .11 at 3-6 months, .13 at 6-12 months, and .18 at 12-24 months). A study by
Ullrich, Turner, Ciol, and Berger (2005) showed that perceived stress from baseline to 6 months
was associated with greater pain at 12 months in men who reported persistent pelvic pain
(p=.03). Keefe, Rumble, Scipio, Giordano, and Perri (2004) proposed that pain catastrophizing,
pain-related anxiety and fear and helplessness contributed to poor pain adjustment. These are
more refined psychological constructs associated with the experience of pain (Quartana,
Campbell, & Edwards, 2009). A recent cross-sectional survey showed that pain severity, pain
catastrophizing, and ruminative anxiety were significantly associated in individuals experiencing
chronic musculoskeletal pain (Curtin & Norris, 2017). Baer, Smith, Hopkins, Krietemeyer, and
Toney (2006) found that individuals with chronic low back, hip or knee pain with comorbid
anxiety and depression reported a higher level of pain severity than those with pain alone
(p<.001). These literature findings suggest a strong linkage between psychological distress and
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 47

the experience of pain, thus prompting the need to account for psychological factors including
emotion and affect when conducting pain evaluations.
Mindfulness Disposition and Its Relationship with Pain and Distress

Mindfulness is a mental state of present-focused awareness that one engages fully in the
present experience, moment by moment, without judgment (Grossman, 2015; Ludwig & Kabat-
Zinn, 2008). This attentional stance is relevant to orthopedic patients, as being mindful might
help them encounter pain experience without unnecessary elaboration, catastrophizing or
reactivity (Day et al., 2014; Keefe et al., 2004). Even in the case when psychological distress
further exacerbates pain reactive responses, mindfulness may offer stress-buffering effects to
mitigate pain perception by adjusting emotional regulation and reducing suffering (McCracken,
Gauntlett-Gilbert, & Vowles, 2007; Prakash et al., 2015). To characterize the type of
mindfulness disposition that is involved in these psychological processes, Baer et al. (2006)
proposed a five-domain mindfulness construct. The domains include observing, describing,
acting-with-awareness, non-judging, and non-reactivity. Specifically, the non-judging and non-
reactivity domains refer to the processing of inner experience, in particular, thoughts and feelings
with an openness and non-evaluative manner (Bohlmeijer, ten Klooster, Fledderus, Veehof, &
Baer, 2011). Given pain is often associated with attentional and affective responses, one’s
mindfulness state may influence how pain is perceived and reported under the circumstances of
musculoskeletal conditions.
More recent literature examines mindfulness disposition as a protective mechanism against
the experience of pain. In a cross-sectional survey of patients with upper extremity conditions,
results found that mindfulness disposition was correlated with lower pain intensity; however the
association was not significant after adjusting for pain interference (Beks et al., 2018). Curtin
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 48

and Norris (2017) showed that in a sample of middle age adults with chronic musculoskeletal
pain, individual domains of the Five Facet Mindfulness Questionnaire (FFMQ) were not
associated with pain intensity, but they overall improved the predicted variance (R
2
) of the
regression model. Similar results were found in Schutze, Rees, Preece, and Schutze (2010) and
Komandur, Martin, and Bandarian-Balooch (2018) that individuals with chronic pain or
headache condition showed no significant association between the overall mindfulness
disposition and pain, except that the FFMQ describing domain predicted headache pain intensity.
Although there is a lack of consistent evidence supporting a direct association between
mindfulness disposition and pain intensity, these studies showed an association between
mindfulness disposition and psychological processes involved in pain, including pain
catastrophizing (Komandur et al., 2018; Schutze et al., 2010) and anxiety (Curtin & Norris,
2017). These observations warrant further investigation into an indirect mechanism that
mindfulness disposition may influence pain through affecting psychological processes associated
with pain experience.
To establish a model linking mindfulness disposition and pain mediated through
psychological distress, it is crucial to provide evidence supporting the association between
mindfulness and distress. Elvery, Jensen, Ehde, and Day (2017) surveyed a sample of
undergraduate students who self-report chronic pain. The results showed that the FFMQ non-
judging domain was negatively associated with depression scores (standardized beta: -0.17).
Brooks et al. (2018) showed that in adults with chronic musculoskeletal pain, greater perceived
mindfulness was associated with lower perceived stress, depressive symptoms and pain
catastrophizing (standardized beta: -.40, -.14, and -.30, respectively). Day, Smitherman, Ward,
and Thorn (2015) observed a significant correlation between FFMQ domains and pain
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 49

catastrophizing (acting-with-awareness: r=-.16, non-judging r=-.23, non-reactivity r=-.23).
Results from their latent variable analysis confirmed that as a single latent factor, mindfulness
predicted pain catastrophizing, while only non-reactivity contributed significantly to the
prediction. Prakash et al. (2015), in a study comparing older (age 60-79) and younger (age 18-
30) adults, found that emotional regulation mediated the associations between mindfulness and
perceived stress in both groups (indirect effect: -.18 and -.16, respectively). In addition,
McCracken et al. (2007) examined a clinical cohort attending chronic pain assessment and found
mindfulness disposition was negatively associated with depression and pain-related anxiety
(beta: -.37, -.20, respectively). This cumulating evidence shows that greater mindfulness
disposition are manifested with lower psychological distress in various sample populations. The
results establish a foundation to test for mindfulness’ role in psychological distress and pain
experience.
The Role of Mindfulness Disposition in the Pain-Distress Relation
Current literature made important advances to account for the role of mindfulness
disposition when assessing the psychological aspects of pain experience. However, research on
specific mindfulness facets that influence distress and pain remains nascent in the literature. In a
knee osteoarthritis clinical cohort, the authors tested mindfulness disposition to moderate the
association between stress and pain in a cross-sectional sample (A. C. Lee et al., 2017). Findings
showed that the acting-with-awareness facet measured in the FFMQ moderated the pain-stress
association. Participants in the higher level group of acting-with-awareness reported consistently
lower levels of perceived stress disregard of intensifying pain levels. In a cross-sectional survey
studying chronic cancer neuropathic pain, survivors in the group with a higher mindfulness
disposition showed a weaker correlation between pain catastrophizing and pain intensity (Poulin
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 50

et al., 2016). The study also found that in a regression model, higher levels of observing and non-
judging facets predicted lower levels of pain intensity. In a cross-sectional survey of female
sample with fibromyalgia, those reporting a higher level of observing showed a weaker
relationship between pain catastrophizing and pain intensity (Dorado et al., 2018). Current
studies focus on testing how mindfulness disposition, as a moderator, influences the strength of
the relationship between pain and psychological aspects of pain. However, these models lack a
direction of association (i.e. pain-distress verses distress-pain) and fail to examine psychological
distress as an intermediate pathway. Testing psychological distress as a mediator can help
explain the relationship between mindfulness disposition and pain. Our proposed model, adapted
from Day et al. (2014), specifically look at how affect (i.e. psychological distress) mediate the
influence of mindfulness disposition on pain. Findings from our study will provide a more in-
depth mechanistic examination of this relationship and specific domains involved in the process.
The goal of this study is to gain knowledge about how mindfulness disposition and
psychological distress are associated with pain in an orthopedic population, and to test the
plausibility of a proposed mechanism. We hypothesized that the direct effect of mindfulness on
pain would be negative, and that this effect would be mediated by psychological distress. In a
multiple meditation model, we examined the three distress domains (depression, anxiety and
stress) in the indirect effect path, and compared the proportion mediated for each distress
domain. We then expanded the multiple mediation model to include the five domains of
mindfulness disposition as predictors.

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 51

Conceptual Model
Figure 3-1. A mediation model examining the effect of mindfulness disposition on pain
through the psychological distress pathway

a = regression weight of mindfulness on distress, b = regression weight of distress on pain, c’ =
direct effect of mindfulness on pain. FFMQ: Five Facet Mindfulness Questionnaire. DASS:
Depression, Anxiety, Stress Scale, UPAT: Universal Pain Assessment Tool

Research Questions
Question 1: Is there an association between mindfulness disposition and pain in orthopedic
patients?
Question 2: Is the effect of mindfulness disposition on pain mediated by psychological distress?
Question 3: Is the mediation in Question 2 driven by specific mindfulness domains and distress
types?

Specific Aims and Hypothesis
Aim 1: To examine the association between mindfulness disposition and pain in orthopedic
patients.
Hypothesis 1: Higher mindfulness disposition is associated with lower levels of pain.
a
b
Mindfulness
(FFMQ)

Distress
(DASS)
Pain
(UPAT)
c’
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 52

Aim 2: To examine if psychological distress mediates the effect of mindfulness disposition on
pain.
Hypothesis 2: Psychological distress mediates the effect of mindfulness disposition on
pain.
Aim 3: To examine whether the mediation in Aim 2 is driven by specific mindfulness domains
and distress types.
Hypothesis 3: Mindfulness domains including non-judging and non-reactivity from the
Five Facet Mindfulness Questionnaire will affect pain by their influence on depression and
stress levels.

Methods
Study Design and Study Sample
We conducted a single-site, cross-sectional study from January to December 2015. Two
trained data collectors recruited adults from an orthopedic clinic in a not-for-profit teaching
hospital at Los Angeles. Inclusion criteria were patients age 18 years or older with upper
extremity joint complaints. Exclusion criteria included being non-fluent in English and with
apparent cognitive impairment. Patients who visited the orthopedic clinic were assessed for
eligibility prior to study enrollment. Individuals who were eligible and agreed to participate
provided signed informed consent at the clinic or by mail. University of Southern California
Institutional Review Board approved all procedures.
Measures
Pain (Outcome). The outcome measure was obtained using the Universal Pain
Assessment Tool (UPAT) shown in Figure 3-2. UPAT is a validated, single-item measure
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 53

routinely used in clinical practice and combines best elements of the validated Wong Baker
Faces Pain Scale, Visual Analogue Scale, and Verbal Rating Scale to ensure that users have both
visual and verbal information to support their accurate reporting of pain (Campbell & Lewis,
1990; Ohnhaus & Adler, 1975; Wong & Baker, 1988), in particular for those with limited
communication skills as well as intellectual disability (Dugashvili, Van den Berghe, Menabde,
Janelidze, & Marks, 2016). UPAT assesses the average pain experienced in the past two weeks
with response options ranging from 0-10. Higher scores indicate greater level of pain.

Figure 3-2. Universal Pain Assessment Tool

Mindfulness Disposition (Predictor). The 24-item Five Facet Mindfulness Questionnaire
(FFMQ) is a validated scale measuring five domains of mindfulness disposition: observing (i.e.
“I notice the smells and aromas of things”), describing (i.e. “I am good at finding the words to
describe my feelings”), non-reactivity (i.e. “I watch my feelings without getting carried away by
them”), acting-with-awareness (i.e. “I find it difficult to stay focused on what’s happening in the
present moment”), and non-judging (i.e. “I tell myself I should not be thinking the way I’m
thinking”) (Bohlmeijer et al., 2011). The initial 39-item FFMQ was developed as a consolidated
instrument by combining items from several mindfulness assessment tools (Baer et al., 2006),
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 54

and later modified to this 24-item short form which retained a similar psychometric properties
and content validity (Bohlmeijer et al., 2011). Each item ranges from 1 (never or very rarely
true) to 5 (very often or always true), and some items require reserved scoring. Higher scores
represent a better mindfulness state. We observed high internal consistency in the current sample
(Cronbach’s =.87). Recall for the FFMQ referred to the past two weeks.
Psychological Distress (Mediator). We used the 21-item Depression, Anxiety, Stress
Scale (DASS) as the mediator to measure three domains of psychological distress, including
depression, anxiety and stress (Henry & Crawford, 2005). DASS assesses emotional disturbance
while the users may not meet the cut-off for clinical diagnosis. Examples of items in each
construct include depression (i.e. “I felt that I have nothing to look forward to”), anxiety (i.e. “I
found it hard to wind down”), and stress (i.e. “I was close to panic”). The original instrument is a
42-item DASS scale developed by Lovibond and Lovibond (1995). Depression items intent to
measure low positive affectivity, anxiety items measure physiological hyperarousal, and stress
items measure a distress construct related to negative affectivity. Studies have shown that DASS
items have excellent psychometric properties for both clinical and non-clinical samples (T. A.
Brown, Chorpita, Korotitsch, & Barlow, 1997; Henry & Crawford, 2005). Item scoring ranges
from 0 to 3, with higher scores representing being more distressed. Recall of DASS referred to
the past two weeks. We observed high internal consistency in the current sample (Cronbach’s 
= .89).
Covariates. We controlled for participant sex and age in the mediation models, to account
for their correlations with pain and distress factors. Studies have shown that pain sensitivity
reduces as age increases, and women demonstrate lower pain threshold and tolerance than men
(Bartley & Fillingim, 2013; Edwards, Haythornthwaite, Sullivan, & Fillingim, 2004; Fillingim,
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 55

King, Ribeiro-Dasilva, Rahim-Williams, & Riley, 2009; Lautenbacher, Peters, Heesen, Scheel,
& Kunz, 2017). In a clinical sample of chronic pain patients, men demonstrate a stronger
association between pre-treatment pain severity and anxiety than women (Edwards, Augustson,
& Fillingim, 2003). In addition, Riley III et al. (2014) shows that older non-Hispanic black
reported greater decrease in pain tolerance than non-Hispanic White as well as compared to their
younger age group, which suggests a compound effect of age and race on pain. However, given
our sample consisted of mainly White participants and the rest with small proportions of other
race and ethnic categories, to maintain a parsimonious model and to simplify the multiple
imputation, we did not include race and ethnicity, as well as injury location as covariates in the
mediation models. Nevertheless, we evaluated their associations with mindfulness, pain and
distress in the correlation analyses and the result yielded weak and non-significant with most
factors.
Statistical Analyses
We first conducted pairwise correlations using available data to examine basis of the
mediation analysis (Rucker, Preacher, Tormala, & Petty, 2011). We then conducted the Little’s
test and showed that our data were not missing completely at random (Li, 2013). To handle
missing data, we performed multiple imputation, and ran all mediation analyses using the
imputed datasets. Our first model assessed the indirect effect between FFMQ sum score and
UPAT through a single mediator DASS sum score. To compare the mediation effect of different
distress domains, we conducted a multiple mediation model with the three DASS domains as
separate mediation pathways and assessed the proportion mediated by each domain. We
expanded the analysis to include all five FFMQ facets as predictors on each DASS domain to
observe any significant indirect effect from the five-predictor-by-three-mediator combinations.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 56

Finally, we conducted sensitivity analysis using complete cases and compared results with the
imputed data. All analyses were performed in Stata 15 with α set to .05 (StataCorp, 2017).
Missing data. We used fully conditional specification multiple imputation to handle
missing data (Lee 2010). We included the dependent variable pain in the imputation model, and
we imputed all data fields disregard of the amount of missingness. As discussed in (Graham,
2009), omitting the dependent variable in the imputation procedure would suppress its
correlation with other independent variables to zero and biases the sample. For the amount of
missingness, Schlomer, Bauman, and Card (2010) showed that when data is missing at random,
regression weights using multiple imputation strategy for 10, 20 or 50 percent of missingness did
not vary substantially. For both FFMQ and DASS, we imputed at the domain level. Plumpton,
Morris, Hughes, and White (2016) described that imputing at the subscale total (i.e. domain
level) can be a remedy when the number of individual items from multiple-item scales became
too large for imputation to process. As such prior to imputation, we first calculated raw domain
scores using available raw items, while treating the domain score as missing if any of its item
was missing. Then, we performed multiple imputation using the -mi impute chained- in Stata 15
(Multiple imputation in Stata. UCLA: Statistical Consulting Group). To impute, we included all
domains in FFMQ (five domains), DASS (three domains), as well as UPAT, sex and age. We
performed 50 imputations. We summed the imputed domain scores to generate a scale sum
score, and then transformed all imputed scores into z-score within each imputation for mediation
analysis. To perform mediation analysis with multiple imputation in Stata, we used the -cmdok-
option on -mi estimate- to accommodate the -sureg- analytic command.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 57

Equations
The equations below are two path models to assess the indirect effect of mindfulness
disposition (FFMQ) on pain (UPAT) through the mediator psychological distress (DASS).
Model (1) assesses the regression weight of the predictor FFMQ (a) on the mediator DASS, and
Model 2 assesses the regression weight of the mediator DASS (b) on the outcome UPAT, while
adjusting for the predictor’s direct effect (c’) on UPAT. Both models control for the effects of
participant sex and age as covariates (BnCov).

DASSi = B1 + aFFMQi + BnCovi + e1i (1)
UPATi = B2 + bDASSi + c’FFMQi + BnCovi + e2i (2)

B = intercept coefficient
a = FFMQ’s regression weight on DASS (path a)
b = DASS’s regression weight on UPAT (path b)
c’ = FFMQ’s regression weight / direct effect on UPAT (path c’)
Bn = covariates’ regression weight on UPAT
Covi = two covariates including participant’s sex and age
e = error term
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 58

Results
Our study included 525 adult orthopedic patients in the analysis. Participant’s average
age was 54.1 years (range 18-98), with 61.4% male and 63.7% White (Table 3-1). The primary
injury location was the shoulder (72.4%), followed by elbow (20.8%) and clavicle/wrist (6.9%).
91% of respondents reported at least mild pain in the past 2 weeks with UPAT scoring 1 or
greater (UPAT: M=4.2 ± 2.5). Respondents also reported presence of distress (DASS sum score:
M=13.0 ± 11.5), with 49% with at least mild level of either depression, anxiety or stress (30.5%,
34.5% and 30.0%, respectively). For sample characteristics including age, male, race/ethnicity,
and injury location, the amount of missing was 2.9% to 6.7%. For scale measurements, the
amount of missing was larger, ranging from 12.4% to 21.7% for domain scores, and 30.3% for
UPAT. As Schlomer (2010) suggested, handling missing data with multiple imputation would
produce minimal bias toward the regression estimates. Therefore, we conducted multiple
imputation and used the imputed datasets for the mediation analysis.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 59

Table 3-1. Participant characteristics (N=525)
Characteristics n % Missing
Age (Mean ± SD) 54.1 ± 15.6 35 (6.7%)
Male 310 61.4% 20 (3.8%)
Race/ethnicity

35 (6.7%)
White 312 63.7%

Hispanic/Latino 75 15.3%

Asian 41 8.4%

African American 22 4.5%

Other 40 8.2%

Primary injury location

15 (2.9%)
Shoulder 369 72.4%

Elbow 106 20.8%

Other (i.e. clavicle, wrist, other) 35 6.7%

Measurements (domain score range) Mean ± SD Missing*
Pain (UPAT) 4.2 ± 2.5 159 (30.3%)
Distress (DASS) 13.0 ± 11.5

Depression (0-21) 3.9 ± 4.6 83 (15.8%)
Anxiety (0-21) 3.3 ± 3.7 81 (15.4%)
Stress (0-21) 5.8 ± 4.4 88 (16.8%)
Mindfulness Disposition (FFMQ) 84.7 ± 10.9

Observing (4-20) 13.8 ± 3.8 65 (12.4%)
Describing (5-25) 19.2 ± 3.8 80 (15.2%)
Non-reactivity (5-25) 16.0 ± 3.9 99 (18.9%)
Acting-with-awareness (5-25) 18.4 ± 4.5 93 (17.7%)
Non-judging (5-25) 17.4 ± 4.1 114 (21.7%)
UPAT: Universal Pain Assessment Tool (single item, range 0-10)
DASS: Depression, Anxiety, Stress Scale sum score (21 items, scale sum range 0-63)
FFMQ: Five Facet Mindfulness Questionnaire sum score (24 items, scale sum range
24-120)
*Any missing at item level was considered missing at domain and scale level

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 60

Table 3-2 displays results of the pairwise correlations from the observed data. The
predictor, FFMQ sum score, was negatively correlated with the outcome, UPAT (r=.-23,
p<.001). FFMQ sum score was also negatively correlated with the mediator DASS sum score
(r=-.53, p<.001). DASS sum score was positively correlated with UPAT (r=.42, p<.001).
Correlations with the scale domains shows that UPAT was moderately correlated with all
domains in DASS (depression: r=.39, anxiety: r=.35, stress: r=.38, p=<.001), while weakly and
negatively correlated with FFMQ facets describing (r=-.14, p=.01) and non-reactivity (-.13,
p=.02), and moderately with non-judging (r=-.21, p<.001). For other covariates, male sex was
corrected with lower pain levels (r=-.15, p<.01), while age was positively correlated with FFMQ
sum score (r=.15, p<.01), FFMQ act-with-awareness (r=.11, p=.03) and negatively with male sex
(r=-.15, p<.001).

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 61

Table 3-2. Pairwise correlation matrix from observed data
UPAT FFMQ DASS FFMQ
OBS
FFMQ
DES
FFMQ
NOR
FFMQ
ACT
FFMQ
NOJ
DASS
DEP
DASS
ANX
DASS
STS
Male Age
UPAT 1

FFMQ -0.23 1

DASS 0.42 -0.53 1

FFMQ_OBS 0.02 0.43 0.07 1

FFMQ_DES -0.14 0.77 -0.37 0.28 1

FFMQ_NOR -0.13 0.64 -0.22 0.36 0.46 1

FFMQ_ACT -0.11 0.61 -0.50 -0.03 0.37 0.09 1

FFMQ_NOJ -0.21 0.49 -0.49 -0.26 0.19 -0.03 0.45 1

DASS_DEP 0.39 -0.52 0.93 0.07 -0.34 -0.22 -0.47 -0.49 1

DASS_ANX 0.35 -0.44 0.89 0.03 -0.37 -0.20 -0.38 -0.38 0.75 1

DASS_STS 0.38 -0.48 0.91 0.05 -0.30 -0.21 -0.50 -0.45 0.76 0.71 1

Male -0.15 0.02 -0.06 -0.15 -0.05 0.06 0.14 0.12 -0.04 -0.11 -0.03 1

Age 0.02 0.15 -0.04 0.08 0.05 0.03 0.11 0.07 -0.03 0.00 -0.04 -0.15 1
Italic-Bold: Correlations are significant at the p<.05 level
UPAT: Universal Pain Assessment Tool
FFMQ: Five Facet Mindfulness Questionnaire sum score
OBS: Observing, DES: Describing, NOR; Non-reactivity, ACT: Acting-with-awareness, NOJ: Non-judging
DASS: Depression, Anxiety, Stress Scale sum score, DEP: Depression, ANX: Anxiety, STS: Stress

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 62

For mediation analysis, we computed pooled estimates using the imputed datasets. In
Figure 3-3, we tested the indirect effect of FFMQ sum score on UPAT through the DASS sum
score. We observed a significant indirect effect (beta=.-22, p<.001), while the original (r=-.21,
p<.001) direct effect between FFMQ and UPAT became non-significant (beta=.01, p=.90). Then,
we conducted a multiple mediation analysis to test the different indirect effects and proportion
mediated through each DASS domain as separate indirect paths (Figure 3-4). Depression scores
mediated 48.3% of the total effect between FFMQ and UPAT (indirect effect beta=-.10, p=.02),
and stress mediated 39.6% (indirect effect beta=-.08, p=.04). The anxiety domain was non-
significant (p=.33). The sum mediated proportion was 104.2% of the total effect. We will discuss
possible reasons to our observed sum exceeding 100% in the discussion section.

Figure 3-3. Mediation effect of psychological distress between mindfulness disposition and
pain

Mindfulness
(FFMQ)
Pain
(UPAT)
a = -.52*** b = .42***
c’ = .01
Distress
(DASS)
*p<.05, **p<.01, ***p<.001, standardized coefficients reported, model used imputed data
FFMQ’s effect on UPAT (c) in the correlation matrix (raw data) is -0.23
UPAT: Universal Pain Assessment Tool, DASS: Depression, Anxiety, Stress Scale sum score
FFMQ: Five Facet Mindfulness Questionnaire sum score. Model controlled for sex and age
Indirect effect (a*b) -.22*** Total effect (a*b + c’) -.21***
Direct effect (c’) .01 Proportion mediated 100%

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 63

Figure 3-4. Proportion mediated by different distress domains between mindfulness
disposition and pain in a multiple mediation model

We further explored how each FFMQ domain indirectly affected pain through the three
unique distress domains. In Table 3-3, results from the multiple mediation analysis show that
through depression, two FFMQ facets including acting-with-awareness (indirect effect beta=-.05,
95% CI: -.10, -.00) and non-judging (indirect effect beta=-.07, 95% CI: -.14, -.01) indirectly
affected UPAT. In the same model, through stress, three FFMQ facets acting-with-awareness
(indirect effect beta=-.07, 95% CI: -.13, -.01), non-reactivity (indirect effect beta=-.04, 95% CI:
0 10 20 30 40 50 60 70
Proportion (%) Mediated by DASS Domain
DASS Domain
16.3% mediated
Indirect effect: -.03
39.6% mediated
Indirect effect: -.08*
All three DASS domains were modeled as separate mediators in one multiple mediation model
Total effect (-.21). Total mediated proportion was 104.2%
* p<.05. P-value refers to significance of indirect effect from each mediation pathways

48.3% mediated
Indirect effect: -.10*

Depression
Anxiety
Stress
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 64

-.07, -.00) and non-judging (indirect effect beta=-.06, 95% CI: -.11, -.01) also indirectly affected
UPAT. However, all five indirect effect pathways through anxiety were not significant.

Table 3-3. A cross-examination of indirect effects of five FFMQ facets through three DASS
domains on UPAT in multiple mediation analysis
FFMQ
DASS-Depression DASS-Anxiety DASS-Stress
Beta (95% C.I.) Beta (95% C.I.) Beta (95% C.I.)
Observing .01 (-.01, .03) .00 (-.01, .01) .01 (-.01, .03)
Describing -.03 (-.06, .00) -.02 (-.05, .02) -.01 (-.03, .01)
Non-reactivity -.03 (-.07, .00) -.01 (-.02, .01) -.04 (-.07, -.00)
Acting-with-awareness -.05 (-.10, -.00) -.01 (-.04, .02) -.07 (-.13, -.01)
Non-judging -.07 (-.14, -.01) -.02 (-.05, .02) -.06 (-.11, -.01)
Italic-Bold: Correlations are significant at the p<.05 level
UPAT: Universal Pain Assessment Tool, DASS: Depression, Anxiety, Stress Scale
FFMQ: Five Facet Mindfulness Questionnaire
Mediation analysis used the multiple imputation dataset
A total of 15 mediation paths between 5 FFMQ facets and 3 DASS domains were set-up in 1 multiple
mediation model to compare the standardized indirect effects of FFMQ on UPAT through DASS

We reran the mediation models with complete cases to reaffirm our findings from the
imputed datasets. We observed similar results in terms of association, directionality and
statistical significance for the indirect, direct and total effect coefficients. Again, repeating the
same mediation analysis in Figure 3-3 using complete cases (n=245) showed a significant
indirect effect (beta=.-20, p<.001) between mindfulness disposition and pain, while the direct
effect was non-significant (beta=-.03, p=.70).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 65

Discussion
Our finding shows that among orthopedic patients, higher mindfulness disposition is
associated with lower pain levels in the correlation analysis. The mediation models show that
psychological distress mediates the association between mindfulness disposition and pain, while
mindfulness’ direct effect on pain becomes insignificant. Results from the cross-examination
between FFMQ and DASS domains show that the non-judging and non-reactivity domains, as
well as acting-with-awareness, which is additional to our hypothesis, have a significant indirect
effect on pain. For the mediator, depression and stress are the main source of stressors that these
mindfulness domains affect upon. Our study results provide empirical evidence to support the
protective role of mindfulness on pain, as well as the importance to account for the presence of
psychological symptoms in orthopedic patients.
Our findings are consistent with current literature. Results from moderation analyses
show that participants in the higher mindfulness disposition category demonstrated a weaker
association between psychological distress and pain intensity (Dorado et al., 2018; A. C. Lee et
al., 2017; Poulin et al., 2016). Mindfulness disposition may change the psychological processes
of pain perception. This is, in particular, relevant to non-judging and non-reactivity domains as
our initial hypothesis focused on. As Bohlmeijer et al. (2011) summarized Baer et al. (2006)
finding on the five mindfulness facets, the non-judging domain is defined in terms of “taking a
non-evaluative stance toward thoughts and feelings”, and the non-reactivity is defined in terms of
“allowing thoughts and feelings to come and go”, both in which referred to one’s processing of
inner experience (p.309). Being non-judging and non-reacting to the pain-associated depressive
and stressful feelings could mitigate these emotions’ negative impact on pain. As such, our
findings is in accordance with the theory that being mindful may help one to encounter pain
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 66

experience without unnecessary elaboration, catastrophizing or reactivity (Day et al., 2014;
Keefe et al., 2004; Veehof, Trompetter, Bohlmeijer, & Schreurs, 2016). The role of mindfulness
is to buffer against the negative emotions and affects (i.e. distress or catastrophizing) associated
with pain, which could provide a positive psychological adjustment to patients experiencing
these types of pain-associated psychological symptoms (Dorado et al., 2018; Poulin et al., 2016).
Our result from the multiple mediation model shows that the sum mediated proportion
from depression, anxiety and stress exceeded 100%. VanderWeele and Vansteelandt (2014)
describes that this observation is possible and may due to three reasons: 1) mediators affecting
one another, 2) interaction between mediators affecting the outcome, and 3) other mediators with
a “negative” proportion being unaccounted in the model. Considering the sizable intercorrelation
(phi=.48) between measurements of the stress and depression constructs using DASS in a clinical
sample in T. A. Brown et al. (1997), as well as those observed in our sample (r=.71 to .76), it is
possible that these mediators affected one another, or interacted to affect the outcome pain,
which were not specifically assessed in our model. While there are other possible pathways
involved between mindfulness disposition and pain, we chose to retain a parsimonious model to
examine our proposed mindfulness-distress-pain construct. This finding prompts future studies to
consider additional mediators, as well as be caution to scale measurement errors and
intercorrelations between different distress domains that may affect the results.
Measuring levels of psychological distress and mindfulness states in a pain clinical cohort
have significant implication to the approach of pain care. The presence of psychological
symptoms such as depression is well-established with patients with musculoskeletal pain
(Duffield et al., 2017; Yelin et al., 2016). As Duffield et al. (2017) noted, a negative cycle can
develop between distress and pain, and further deteriorate one’s health condition and ability to
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 67

seek proper treatment. Our results confirm this co-morbidity, and identify that varying levels of
mindfulness might modify the distress-pain relationship. Individuals with higher mindful
awareness and being less judgmental might be able to shift their thoughts and observe their
emotions toward pain differently, a notion of reperceiving suggested by Day et al. (2014). This
process could disrupt the distress-pain cycle, or compete against rumination from pain
catastrophizing (Poulin et al., 2016). As more mind-body therapies target on psychological
symptoms for pain management (Rajguru et al., 2014), our results offer additional mechanistic
details to the design of specific mindfulness-facet-by-distress-type interventions for orthopedic
populations.
Study Limitations
This study has several limitations. The study recruited from a single non-profit teaching
hospital. The selective patient cohort limits the external validity and generalizability of the
findings. Convenience sampling may introduce self-selection bias. The analytic sample may be
of better health or more trust in research. In addition, some participants may experience
questionnaire fatigue and affected data validity. To address some of these internal threats, we
recruited a large study sample to increase the variance of the subject pool and diversify the
responses. For data validity, we encountered issue with missing data. To remedy, we performed
multiple imputation so that we can produce an estimation of the full sample. Although imputing
FFMQ and DASS scales at the domain level did not fully utilize the information that were
available at the item level (i.e. any item missing was treated as domain missing), our initial
attempt using individual item failed to converge due to a large volume of scale items (FFMQ: 24
items, DASS: 21 items). While we were fully aware that item level imputation is often the
preferred method (Gottschall, West, & Enders, 2012; Plumpton et al., 2016), Plumpton et al.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 68

(2016) suggests that a simplification approach is to impute using the subscale total, for which we
did. This allows us to reduce the imputation model size, while the potential trade-offs are losing
the ability to further explore individual items. We took caution of our imputed data findings by
rerunning our analyses with complete cases and confirmed that our results were consistent
compared to the imputed datasets. In addition, we tested the mediation analysis using cross-
sectional data. Although some may find this analytic approach arguable, Disabato (2016)
suggests that the goal is to offer theoretical contributions. Our study aims to provide insights
from an associational standpoint to inspire longitudinal study design for future reference
(Disabato, 2016; O'Laughlin, Martin, & Ferrer, 2018).
Conclusions
Mindfulness disposition may function to remediate the negative impacts of psychological
distress on pain. Our findings support mindfulness as a positive psychology disposition,
specifically the non-judging and non-reactivity domains in orthopedic patients to address their
pain-associated distress. Examining the presence of psychological distress as a co-morbid
condition of pain, and its associated mechanisms with pain symptoms are research priorities in
the Federal Pain Research Strategy (IPRCC 2017). Future research can use our findings to design
longitudinal studies to test specific mechanistic targets of mind-body therapies as self-
management strategies for the orthopedic pain population.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 69

Chapter 4: Effect of telephone call versus text message reminders on patient return to
acupuncture follow-up treatment: a randomized controlled trial
Federal Pain Research Strategy Research Priorities:
 “Determine optimal safe and effective chronic pain management”
 “Determination of optimal dosing and adherence strategies for non-pharmacological
treatments should be included in this evaluation process.”
Background
In 2012, an estimated 25 million U.S. adults suffer from daily chronic pain (Nahin,
2015). Amid the height of the opioid epidemic, both public and health professionals urgently call
for a multifaceted, integrative approach to manage pain symptoms. Complementary health
approaches (CHA) has a long history of use for the management of unremitting physical ailments
and chronic illness, including pain. CHA are non-pharmacological therapies including
acupuncture, chiropractic manipulation, meditation, massage therapy, natural product
supplements and homeopathy (Nahin, Boineau, et al., 2016). A national survey in 2012 shows
that over 30% of the U.S. adults had used some forms of CHA (Clarke, Black, Stussman, Barnes,
& Nahin, 2015). $14.1 billion was spent on CHA practitioner visits from out-of-pocket expenses
in that same year alone (Nahin, Barnes, et al., 2016a). Among CHA, acupuncture is one with a
consistent growth of usage even though coverage is limited. An estimated 75% of acupuncture
visits were not covered by health insurance (Nahin, Barnes, et al., 2016b); however, use had
increased from 1.1% in 2002 to 1.5% in 2012 (Clarke et al., 2015). Pain is the most common
reason people use acupuncture treatment (Wang et al., 2018). This study will focus on
acupuncture’s role in the management of pain conditions and strategy to improve adherence to
acupuncture treatment.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 70

Acupuncture is a Traditional Chinese Medicine (TCM) technique using needle
stimulation upon the body’s meridian points to remediate physical and psychological complaints.
The needling technique stimulates the meridians by unblocking chi stagnation. Given chi
stagnation causes pain in TCM theory, acupuncture can target and treat pain symptoms. In
biological research, studies shows that acupuncture changes the level of endogenous opiates and
serotonin (J. G. Lin & Chen, 2008). Both endogenous opiates and serotoninergic descending
inhibitory pathways are critical biological mechanisms for pain control, thus supporting the
scientific basis for acupuncture on pain. In 1998, the National Institute of Health published a
rigorous report outlining scientific assessments of acupuncture (National Institutes of Health,
1998). The report concluded that acupuncture shows efficacy on the management of
postoperative pain and reduces side effects from chemotherapy. In the following two decades,
evidence continues to emerge and supports acupuncture’s effectiveness on pain conditions such
as chronic headache, non-specific musculoskeletal pain of back and neck, osteoarthritis, and
shoulder pain (Cherkin & Herman, 2018; R. Chou et al., 2016; Furlan et al., 2010; M. S. Lee &
Ernst, 2011; Y. Lin et al., 2017; Liu, Skinner, McDonough, Mabire, & Baxter, 2015; Nahin,
Boineau, et al., 2016; Qaseem et al., 2017; Tice et al., 2017; Tick et al., 2017; Yuan, Purepong,
et al., 2008). With the pressing demand to identify nonpharmacological pain treatment during the
opioid crisis, acupuncture is recognized as one of the effective CHA to address the pain epidemic
in the U.S.
Acupuncture is a practice wherein patient adherence is important, specifically in the
context of attendance to follow-up treatments. A collaborative acupuncture trial showed that
patients improve in pain symptom by 0.1 standard deviation by every five session they received
(MacPherson et al., 2013). This finding suggests that acupuncture may offer a cumulative
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 71

therapeutic benefit when adhere over a continuous regimen. The recommended regimen for low
back pain treatment consists of 10 acupuncture sessions (Molsberger, Zhou, Arndt, & Teske,
2008; Yuan, Kerr, Park, Liu, & McDonough, 2008). Clinical trials use a median of 8-10 sessions
as a benchmark for their treatment design (Yuan, Kerr, et al., 2008), and trial patients in general
adhere to the recommendation (Barlow et al., 2011). However, in observational studies using real
clinical data, patient adherence rates to acupuncture pain treatment were modest (Bishop,
Yardley, Cooper, Little, & Lewith, 2017; Moroz, Spivack, & Lee, 2004). A study evaluated
follow-up treatment rates at an acupuncture clinic showed that 60% of their new patients did not
have a fifth follow-up visit (Marx, Rubin, Milley, Hammerschlag, & Ackerman, 2013). Another
clinic showed that 49% of their patients only had less than or equal to 21 days of treatment
duration (3 weeks of treatment) (Hsu, Dunn, Bradshaw, & Conboy, 2014). These findings
demonstrate a significant gap between the recommended acupuncture regimen and the actual
patient follow-up attendance behaviors. Improving attendance rates to follow-up visits, therefore,
will be an important research topic to warrant the clinical effects of acupuncture as intended.
Introduction
Acupuncture patients often receive practitioner recommendation for follow-up treatment
to manage their clinical symptoms. Follow-up treatment in the context of a treatment course has
been linked to reduction in pain levels, improvement in functional disability, and persistence of
treatment effects (Bishop et al., 2015; Li et al., 2014; MacPherson et al., 2013; MacPherson et
al., 2017). Each year, an estimated 3.4 million U.S. adults use acupuncture for different health
problems (Clarke et al., 2015). Acupuncture patients often complain of musculoskeletal pain
(Marx et al., 2013; Wang et al., 2018). These conditions often require multiple treatment sessions
in order for the therapeutic benefits to be experienced over time (Molsberger et al., 2008; Yuan,
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 72

Kerr, et al., 2008). However, follow-up treatment attendance rates are modest in acupuncture
clinics. 40 to 51% of patients complete all recommended treatment prescribed at the initial clinic
visit (Bishop et al., 2017; Moroz et al., 2004). Studies with different follow-up benchmarks
reveal that 60% of new patients had less than five follow-up visits (Marx et al., 2013), 49% had
21 days or less of a consecutive treatment duration (Hsu et al., 2014), and 50% had no follow-up
visit at all in acupuncture teaching clinics (Lam et al., 2019). Low follow-up treatment rates can
limit the health benefits that patients may receive from continued care.
Despite the potential health benefits of attending follow-up treatment, only a handful of
studies have addressed the factors related to follow-up treatment return in acupuncture patients.
Bishop et al. (2017)’s study found that severe symptoms and challenge to access treatment were
associated with lower treatment completion rate. Noncompliance to a treatment course could be
attributed to worsening symptoms and perceiving acupuncture treatment as ineffective (Moroz et
al., 2004). Qualitative findings suggest that patients stopped treatment if they lost coverage or
did not expect to receive any additional benefits (Barlow et al., 2011; Bishop, Barlow, Coghlan,
Lee, & Lewith, 2011; Rugg, Paterson, Britten, Bridges, & Griffiths, 2011). Of the variables that
predicted return to follow-up treatment, receiving a treatment plan and attending the first follow-
up treatment within 7 days resulted in a higher number of additional follow-up visits (Lam et al.,
2019). Beyond these preliminary findings, we have not identified any literature that uses a
systematic approach to affect an acupuncture patient’s treatment attendance behavior, in
particular to return for follow-up visits. Utilizing cues to action from the Health Belief Model is
a promising strategy to encourage patients to return for acupuncture treatment (Janz & Becker,
1984).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 73

Mobile phone reminder systems are widely used in conventional medicine. Applications
of mobile phone reminders have been used to encourage attendance to follow-up appointments,
improve medication adherence, and support recommended lifestyle changes (Chow et al., 2015;
Schwebel & Larimer, 2018; Thakkar et al., 2016). Two mobile interventions, telephone call and
text message, are shown to improve attendance to healthcare appointments (Gurol-Urganci, de
Jongh, Vodopivec-Jamsek, Atun, & Car, 2013; Kannisto, Koivunen, & Välimäki, 2014; Parikh et
al., 2010). Arora et al. (2015) showed that the follow-up visit rate were increased by 18% in
patients who received a multi-cue text message intervention compared to no reminder control in
the emergency department setting (73% vs 62%, p=.05). The text message reminders can serve
as external cues of action to improve appointment attendance. However, the strategies have not
been empirically evaluated in the integrative health arena. Specifically, it is not known if using
telephone call or text message reminder will promote return visits in acupuncture patients.
The overall goal of this study is to learn if mobile phone reminders will improve patient
return rates to acupuncture follow-up treatment. We first test whether the use of a telephone call
or text message reminder increases participant attendance to follow-up treatment within a 30-day
period after the initial visit. We predict that follow-up treatment rates will be higher in telephone
call and text message groups compared to usual-care at the site (i.e. no follow-up contact). Then,
we test the active interventions’ ability to sustain participant intent to attend follow-up treatment.
Additionally, we conduct an exploratory analysis to examine the association between clinical and
psychosocial factors and the number of follow-up visits in the 30-day period. Lastly, we test the
effect of attending follow-up treatment on pain levels and pain disability across a 30-day survey
period following the initial visit. We hypothesize that attendance to any follow-up treatment will
result in greater improvement in pain symptoms over time.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 74

Research Questions
Question 1: Do telephone call and text message reminders increase follow-up treatment
attendance rate among acupuncture patients within a 30-day period after their initial visit?
Question 2: Does the reminder system increase participant intent to attend follow-up treatment?
Question 3: What are the clinical and psychosocial factors that predict more follow-up visits?
Question 4: Does attending follow-up visits (versus no follow-up) increase pain reduction and
reduce pain disability over a 30-day period?

Specific Aims and Hypothesis
Aim 1: To test whether telephone call or text message reminders to participants increases their
likelihood of attending acupuncture follow-up treatments within 30 days after their initial visit.
Hypothesis 1: Participants who receive a telephone call or text message reminder will
have a higher likelihood of attending at least one follow-up treatment visits than the no
intervention control group within 30 days after their initial visit.
Aim 2: To test whether participants who received the intervention report a higher level of intent
to attend follow-up treatment over a 30-day period.
Hypothesis 2: Participants who received the intervention will report a higher level of
intent to attend follow-up treatment over a 30-day period compared to no intervention
control group.
Aim 3: To examine clinical and psychosocial factors that predict more follow-up visits.
Hypothesis 3: Treatment expectancy, severity of illness, and access to treatment will
predict a higher number of follow-up visits.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 75

Aim 4: To evaluate whether attending follow-up visits is associated with reduction in pain levels
and pain disability among a subset of pain participants over a 30-day period.
Hypothesis 4: Among participants who reported a pain condition, those who attended
follow-up visit(s) will demonstrate a higher reduction in pain levels and pain disability
over a 30-day period than those who did not attend follow-up.

Methods
Trial Design and Randomization
This was a single-site, three-group randomized controlled trial. We randomized
participants using age-stratification (≥50 years) in blocks of 3 conditions on the date of
enrollment using a random generator (1:1:1 telephone call to text message to no intervention
control). Individual participants were the unit of randomization. All participants provided written
informed consent. The sample size for each group was set to 40, which resulted in a total of 120
participants for this preliminary trial. The University of Southern California Health Science
Institutional Review Board approved all procedures. (Trial registration: clinicaltrial.gov,
Identifier: NCT03645083)
Conceptual Framework
The study proposed a conceptual framework to study the utilization of acupuncture
treatment. The framework synthesizes relevant constructs from two major health models that
guide the field of health behavior research. These models include the Health Belief Model
(Becker, 1974) and Behavioral Model of Health Services Use (Andersen, 1995). The design of
the trial, selection of data elements and analytic plans were guided by the conceptual framework
and presented with study elements as a study overview in Figure 4-1. Hypothesis 1 (H1) tests
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 76

the association between Cues to Action (i.e. mobile interventions) and the behavioral outcome
(i.e. attending follow-up visit) derived from the Health Belief Model. Hypothesis 2 (H2) tests if
the interventions change participant intent to attend follow-up treatment over a 30-day period.
Hypothesis 3 (H3) test if any baseline characteristics would predict a higher number of follow-up
visits over the 30-day period. Hypothesis 4 (H4) tests whether attending follow-up visit(s)
(versus not) will result in greater improvement in treatment responses, including pain levels and
pain disability outcomes.
Figure 4-1. Study overview / conceptual framework of attendance to acupuncture follow-up
treatment

Setting and Participants
The study was conducted in Los Angeles from February 1 to September 30, 2017. Adult
patients (≥18 years) visiting the acupuncture teaching clinic were recruited by a research
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 77

assistant at the front desk. Inclusion criteria were age of at least 18 years, agreement to
randomization, patient seeking a new treatment consultation, owner of a mobile phone with text
message capability, and treatment provided by an intern (i.e. acupuncture trainee under a
licensed acupuncturist) in the teaching clinic. Exclusion criteria were treatment in the auricular
acupuncture clinic or treatment in the specialty clinic. The acupuncture clinic’s own staff and
students were also excluded from the study.
Procedures
Participants who showed interest in the study completed three screening items to
determine eligibility. The items included age of at least 18 years, new treatment consultation, and
text-capable phone ownership. Both screening and enrollment took place between clinic arrival
and the start of the treatment session. All participants completed a self-report baseline
questionnaire in the clinic waiting area following their treatment session. Prior to leaving the
clinic, participants received a reminder card with a study telephone number that would be used to
notify them if they were assigned to a study intervention group. Study enrollment and baseline
questionnaire were completed on Day 0. Participants were blinded to the randomization until
they were contacted by a research assistant based on their group assignment on Day 3.
Participants were asked to complete an online follow-up questionnaire on Day 10 and Day 30
about their experience with the overall treatment. We conducted chart reviews to determine if
participants had any attended follow-up visits within a 30-day period after their initial visit. We
compensated all participants with a $5 gift card for completing the baseline questionnaire. Those
who completed a follow-up questionnaire were eligible to win a voucher for a one-time
acupuncture treatment at the same clinic for free (vouchers were mailed out after study was
complete).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 78

Interventions
Telephone Call Reminders
The telephone call intervention was a one-time, one-to-one contact between the research
assistant and the participant scheduled to receive the intervention. The research assistant used the
telephone number provided by the participant in the baseline questionnaire, and attempted to call
up to three times between Day 3 and Day 4. When connected, the research assistant recited a
scripted greeting and conveyed the key intervention message. The message stated: “we
encourage all patients to schedule a follow-up visit to help with their symptom management.” If
the call was not connected after three attempts, in the final attempt, the research assistant left a
voicemail with the same intervention message. The intervention was considered delivered
successfully when participants answered the telephone call within three attempts.
Text Message Reminders
The text message intervention was intended to be a one-time delivery of the intervention
message to the participant’s mobile phone. We expanded the “one-time” definition given that
some participants received the text message up to three times if they had not text-replied that the
intervention message was received. All messages were sent by the research assistant to each
participant’s preferred phone number between Day 3 and Day 4. The text message stated: “we
encourage all patients to schedule a follow-up visit to help with their symptom management”
after a scripted greeting text. The text intervention was considered delivered successfully when
participants text-replied their receipt of the text message as promoted at the end of the message.
Outcome and Assessments
Participants completed self-assessments before, during and after the 30-day study period.
The primary outcome was attendance to follow-up treatment within 30 days after initial visit
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 79

(y/n). This was obtained through chart review by the research assistants. Secondary outcomes
included the number of follow-up visits, as well as participant intent to attend follow-up
treatment at Day 0, Day 10 and Day 30 (0: strongly disagree to 4: strongly agree). Other
secondary outcomes included changes in pain scores and pain disability scale in participants
reporting a pain condition at baseline, and whether the change differed over time by attending
follow-up treatment (or not) throughout the 30-day study period. Pain scores were measured
using the visual analog scale (VAS: 0 – 100)

at 4 time points: Day 0 (before / after the initial
treatment), Day 10, and Day 30 (Campbell & Lewis, 1990). For pain disability, we used the
validated Pain Disability Index (PDI), a 7-item scale scoring from 0 (no disability) to 10 (worst
disability) (Tait, Chibnall, & Krause, 1990). Sum score of the 7-item PDI was assessed at three
time points: Day 0 (after treatment), Day 10, and Day 30, with Cronbach α = 0.92 at Day 0.
Covariates
We measured participant and clinical characteristics at baseline. Participant
characteristics included age, sex, race and ethnicity, years of education, employment, and health
insurance status. Clinical characteristics included pain-related condition (y/n), chronicity (acute
vs chronic, defined by three or more months), immediate post-treatment symptom improvement
(1: no improvement to 4: cure), prior use of acupuncture (y/n), out-of-pocket payment for
acupuncture visit (y/n), concurrent treatments, barriers to receive follow-up treatment, treatment
satisfaction (1-10), being recommended for follow-up treatment (y/n), and the scheduling of
follow-up appointment (y/n). Detailed description of the items are listed the following.
Age. Age was a self-repot item in the baseline questionnaire and was a continuous
variable in the model. A study using national data shows that middle age and older users (40
year+) are more likely to have used acupuncture. (Zhang 2012) Another study shows that age is
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 80

associated with an increased likelihood of attendance to acupuncture follow-up treatment. (Lam
2019)
Sex. Participants provided their birth sex in the baseline questionnaire (male vs female).
A study using national data shows that female are more likely to be acupuncture users than male
(Zhang, Lao, Chen, & Ceballos, 2012).
Sociodemographic and treatment access: Participants provided race and ethnicity
(American Indian/Alaska native, Asian, Black/African American, Native Hawaiian/Pacific
Islander, Hispanic/Latino, White/Caucasian, other), employment status (Y/N), years of
completed education as representation of their sociodemographic status at baseline. Participants
reported the type of insurance coverage (private, Medicare, Medi-CAL, other government
programs, no health insurance), and whether they paid out-of-pocket for the acupuncture
treatment as proxies for their ability to access healthcare treatment, and specifically for
acupuncture. Meghani and Liou (2019) argue that race/ethnicity and treatment coverage play a
role in patients accessing non-pharmacological treatments.
Pain and chronic condition. These were two separate questions at baseline asking
participants if their primary health concern for the visit was related to physical pain (Y/N), and
whether their condition was acute or chronic (chronic: three months and more) (Nahin, 2015).
Acupuncture is often used for musculoskeletal pain disorders in the U.S. (Clarke, Nahin, Barnes,
& Stussman, 2016), thus patients with related conditions may have greater tendency for further
treatment utilization.
Symptom levels and symptom improvement. The baseline questionnaire participants
completed after their initial visit included two items to evaluate their pre- (recall) and post-
(current) treatment overall symptom levels using the visual analog scale (0-100, using “Not at all
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 81

severe” to “Extremely severe” as anchors)(Campbell & Lewis, 1990; Gould, Kelly, Goldstone, &
Gammon, 2001). Immediate symptom improvement was a subjective report of possible
improvement immediately after the treatment from one of the four categories (cure, clear
improvement, slight improvement, no improvement/don’t know). Acupuncture treatment can
provide immediate pain relief and, thus, may affect patient’s decision to continue the
recommended regimen (Barlow et al., 2011; Xiang, Cheng, Shen, Xu, & Liu, 2017).
Treatment plan and Appointment Scheduled. Participants reported in the baseline
questionnaire whether they received a treatment plan (Y/N) from their acupuncturist
recommending follow-up visit, and if they scheduled an upcoming appointment (Y/N) prior to
their clinic departure. A study showed that when acupuncture patients received a treatment plan
from their practitioner, they were 2.5 times more likely to have a follow-up visit in 30 days (Lam
et al., 2019).
Acupuncture-related items. We asked participants if the current visit was their first time
using acupuncture treatment, and we coded that as repeated acupuncture user if they answered
no. We also asked participants to rate their overall satisfaction of the treatment they received,
from a scale of 0-10 (not satisfied at all to complete satisfied). Studies have shown prior
acupuncture usage and satisfaction with acupuncture treatment are possible reasons that lead to
further usage of acupuncture services (Barlow et al., 2011; Bishop et al., 2011).
Concurrent treatment. Participants reported if they were seeking other treatment or using
any self-care methods to address their health condition. The options included but not limited to
visiting a medical doctor, visiting a psychologist, using prescription medications, and/or other
CHA. The use of other care or therapy may offer health benefits to their primary condition, thus
could compete with their decision to continue care with acupuncture follow-up treatment.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 82

Follow-up treatment-related items. Participants reported whether they were
recommended by the acupuncturist to return for follow-up treatment at the baseline questionnaire
(Y/N). They reported if they anticipated any potential barrier to attend an acupuncture visit,
including time, cost and distance. Each barrier was an individual item and scoring from 0 (not at
all) to 4 (very much). For the scheduling of follow-up appointment, at Day 3, the research
assistant reviewed the patient chart to determine if the participant had scheduled a follow-up
appointment within the first three days after the initial visit.
Additionally, we measured participants’ perception of illness severity using the 8-item
Brief Illness Perception Questionnaire (Broadbent, Petrie, Main, & Weinman, 2006), levels of
physical disability using the 12-item World Health Organization Disability Assessment Schedule
2.0 (Andrews, Kemp, Sunderland, Von Korff, & Ustun, 2009), treatment attitudes using the 6-
item Attitudes toward Complementary and Alternative Medicine (Greco et al., 2016) and patient
expectancy of acupuncture treatment using the 4-item Acupuncture Expectancy Scale (Mao, Xie,
& Bowman, 2010). Summed scores were reported for all these measurements at baseline.
Detailed description of these items are listed as the following.
Perception of illness severity. Perception of illness severity was measured using the
validated 8-item Brief Illness Perception Questionnaire and the composite score was used
(Broadbent et al., 2006). All items score from 0 (lowest) to 10 (highest) with the anchors
correspond to the construct being measured for each item. A study found that acupuncture
patients with lower perceived pain were more likely to attend all recommended treatment visits
(Bishop et al., 2017). The scale has a moderate internal consistency (Cronbach’s α = 0.67, higher
score more severe).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 83

Level of physical disability. The measurement of physical disability was operationalized
by the 12-item World Health Organization Disability Assessment Schedule 2.0 (Andrews et al.,
2009). Each item scores from 0 (no difficulty) to 4 (extreme difficulty or cannot do). A study
found that patients required little physical effort to see their acupuncturist were more likely to
attend all recommended treatment visits. (Bishop 2017) The scale has a high internal consistency
in this study (Cronbach’s α = 0.88, higher score more severe).
Treatment attitude. We used the 6-item Attitudes toward Complementary and Alternative
Medicine (CAM) developed by Greco et al. (2016), using items derived from the PROMIS
(Patient-Report Outcomes Measure Information System) item bank. Items focused on patient’s
confidence, feeling, value toward the CAM treatment, as well as the treatment would be
successful and right for the patients. Each item scores from 0 (not at all) to 4 (very much). The
scale has a high internal consistency in this study (Cronbach α = 0.88, higher score more positive
attitude).
Expectancy of acupuncture treatment. The sum score of the 4-item Acupuncture
Expectancy Scale measured participants' expectancy of their acupuncture treatment (Mao et al.,
2010). Each item scores from 1 (not at all agreed) to 5 (completely agreed). Patient’s higher
expectation of their acupuncture treatment outcome is shown to be positively associated with
better improvement (Linde et al., 2007). The scale has a high internal consistency in this study
(Cronbach’s α = 0.89, higher score higher expectancy).
Statistical Analysis
The primary outcome of the study was the differing rates of attendance to acupuncture
follow-up treatment within 30 days after the initial visit comparing the telephone call, text
message, and no intervention control groups. Both the intention-to-treat and as-treated rates are
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 84

analyzed. The analyses accounted for factors with baseline differences across groups from the
randomization to compute an adjusted rate using multivariate logistic regression models.
Analyses were performed in Stata 13 (StataCorp, 2013). For secondary outcomes, we performed
an exploratory analysis to determine if any baseline factors would predict the total number of
follow-up treatments within 30 days. We first conducted bivariate analyses and used the forward-
selection approach with p-value <.10 set in advance to then analyze using the multivariate
Poisson regression model.
For analyses using the longitudinal data, between-group contrasts in the trajectories of
intent to attend follow-up treatment was tested by three intervention groups over the assessment
period (Day 0, Day 10 and Day 30). In addition to the intent measure, between-group contrast in
the trajectories of pain levels and pain disability were tested by attendance to follow-up treatment
over the assessment period. The analyses used the repeated measures mixed model (xtmixed
command using restricted maximum likelihood estimation) from Stata 13 (Multiple imputation
in Stata. UCLA: Statistical Consulting Group). We used the carried-forward imputation method
for missing data in Day 10 and Day 30. The intention model controlled for baseline acupuncture
expectancy, while the pain level and pain disability outcomes we controlled for group
intervention and the chronicity of the condition (i.e. acute vs chronic). We assessed whether our
data provided evidence of benefits from 1) study interventions on participant intention to attend
follow-up treatment, and 2) attendance to follow-up treatment on pain outcomes. Baseline levels
of participant intent and pain outcomes between participants with vs without missing data were
compared using t-tests.
Power calculation indicated that it would require more than 300 participants per group to
detect a 10% difference in our primary outcome (i.e. 60% intervention vs 50% control for return
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 85

rates); therefore, our sample size was set to a total of 120 (40 per group) as a preliminary RCT
based on our available study time and resources.
Equations
Aim 1 we conducted a logistic regression to test if the interventions (telephone call, text
message) increased the likelihood of participants attending follow-up visits from the RCT
design, while adjusting for covariates that are imbalance from the randomization. Aim 2 we
tested the difference in changes in participant intent to attend follow-up treatment over time by
comparing participants in each intervention group using the repeated measures mixed model. We
subset to only participants who received the interventions, and the group (telephone call, text
message and control)*time (Day 0, Day 10, Day 30) interactions are the main predictors. Aim 3
we conducted an exploratory analysis using a multivariate regression model to test if any
baseline factors predicted more follow-up visits. Aim 4 we subset on the pain cohort. We tested
difference in changes in pain levels and pain disability over time by comparing participants who
attended follow-up visits versus not using the repeated measures mixed model. The group
(follow-up attendance)*time (pre-/post-treatment, Day 10, Day 30) interaction was the predictor.
Given the RCT design in Aim 1 was designed to influence follow-up attendance but not intention
measured in Aim 2, we still controlled for the intervention group assignments as covariates in the
model in case there were unintended residual effects.
Aim 1: Attni = B0 + B1Calli + B2Texti + BnCovi + ei (1)
Attni is a binary indicator as to whether participant i attended any follow-up treatment
within 30 days after initial visit. Calli is a binary indicator of the participants who were assigned
to receive the telephone call intervention at Day 3, and Texti were participants assigned to
receive the text message intervention at Day 3. Both groups are compared to the no intervention
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 86

control group. Covi represents variables that were imbalance from the randomization, which
included pain condition and self-report symptom improvement.
Aim 2: Intentit = B0 + B1Calli + B2Texti + B3Time1t + B4Time2t + B5Calli*Time1t +
B6Texti*Time1t + B7 Calli*Time2t + B8Texti*Time2t + eti (2)
Intentit is a continuous indicator of participant intent to attend follow-up visits of
participant i at time t. Calli is a binary indicator of participants who received the telephone
intervention on Day 3, and Texti represents participants who received the text message
intervention on Day 3. Time1t and Time2t are the time indicators of three time measurement
points (Day 0, Day 10, Day 30) using two dummies. Calli*Timet and Texti*Timet are the
interactions of participant i who received telephone call or text message intervention versus no
intervention control at time t.
Aim 3: nFUVi = B0 + B1Calli + B2Texti + BnCovi + ei (3)
nFUVi is a count indicator representing the total number of follow-up visits within 30
days after the initial visit. Calli is a binary indicator of participants who received the telephone
intervention on Day 3, and Texti were those who received the text message intervention on Day
3. Covi represents vectors of multiple individual and clinical features, which enter the adjusted
Poisson model based on the forward-selection criteria.
Aim 4: Sympit = B0 + B1Attni + B2Time1t + B3Time2t + B4Attni*Time1t + B5Attni*Time2t +
B4Calli + B5Texti + B6Covi + et (4)
Sympit is a continuous indicator of the pain levels or pain disability of participant i at time
t. Attni is a binary indicator as to whether participant i attended any follow-up treatment within
30 days after initial visit. Time1t and Time2t are the time indicators of three measurement points
(Day 0, Day 10, Day 30) using two dummy variables (in the case for pain disability there were
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 87

four time points). Attni*Time1t and Attni*Time2t are the interactions of participant i attended
follow-up treatment versus not at time t. The term Calli is a binary indicator of participants who
received the telephone call intervention on Day 3, and Texti were those who received the text
message intervention on Day 3. Covi is a binary indicator of whether participant i reported an
acute vs chronic condition at baseline as a covariate.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 88

Results
Participant Flow and Characteristics
Participant flow through enrollment, randomization, Day 10 and Day 30 follow-ups, and
analysis is shown in Figure 4-2. After being screened for eligibility, 120 adults were randomized
to three groups. In the telephone call group, 23 of 40 (58%) participants received the phone
intervention by answering the call. In the text message group, 31 of 40 (78%) participants
received the text intervention by text-reply as prompted. All 120 participants were analyzed for
the primary outcome. Participants did not report any adverse events related to the study.
Figure 4-2. Flow of randomized participants

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 89

Table 4-1 lists descriptive statistics for the three study groups at baseline. The mean (SD)
age of participants was 42.6 (13.0) years, and 73% (87 of 120) were female. The mean (SD)
symptom score (before initial treatment) of 52.9 (25.0) indicated a sample with moderate levels
of symptom complaint. The mean (SD) score of intention to attend follow-up treatment was 3.5
(1.0), suggesting that most participants had a strong intent to return to the clinic for follow-up
treatment of their condition. Most participants (90.0%) received a recommendation from their
acupuncturist to return for follow-up visits, although only 62.5% of the recommendations were
documented on the patient chart, while 27.5% were communicated verbally as an as-needed
basis. Two baseline variables, pain condition and immediate symptom improvement differed
across group assignments by chance (p=.03 and p=.01, respectively).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 90

Table 4-1: Baseline characteristics of participants by randomized group assignment

Telephone Call
n = 40
mean ± SD / n (%)

Text Message
n = 40
mean ± SD / n (%)

Control
n = 40
mean ± SD / n (%)
p-
value
Female 28 (70.0)

28 (70.0)

31 (77.5) .69
Age (years) 44.1 ± 13.0

41.0 ± 11.9

42.7 ± 14.2 .58
White/Caucasian 26 (65.0)

32 (80.0)

29 (72.5) .32
≥16 years of school 32 (80.0)

27 (67.5)

26 (65.0) .29
Currently employed 35 (87.5)

35 (87.5)

33 (82.5) .76
Had health insurance 38 (95.0)

35 (87.5)

36 (90.0) .50

Chronic condition 28 (70.0)

24 (60.0)

23 (57.5) .47
Pain condition 20 (50.0)

31 (77.5)

22 (55.0) .03
Symptom VAS: before
baseline treatment 49.9 ± 26.3

53.1 ± 23.5

55.6 ± 25.6 .60
Symptom VAS: after
baseline treatment 34.2 ± 23.8

31.4 ± 16.9

39.4 ± 23.9 .25
Symptom improvement
(1 to 4) 2.8 ± 1.2

2.1 ± 1.1

2.1 ± 1.1 .01

Illness perception 42.8 ± 9.8

44.2 ± 8.4

40.4 ± 10.9 .22
Physical disability 20.4 ± 7.9

21.7 ± 7.1

19.7 ± 8.3 .50
Attitude toward CAM 20.1 ± 4.9

19.4 ± 4.9

20.3 ± 4.6 .67
Acupuncture expectancy 16.0 ± 3.5

16.2 ± 2.7

16.6 ± 3.0 .65

Repeated acupuncture
user 30 (75.0)

27 (67.5)

30 (75.0) .69
Acupuncture: paid out-
of-pocket 36 (90.0)

38 (95.0)

37 (92.5) .70
Concurrent treatment(s)

MD or psychologist
visits 15 (37.5)

9 (22.5)

12 (30.0) .34
Prescription
medications 10 (25.0)

9 (22.5)

7 (17.5) .71
Complementary health
approaches 26 (65.0)

28 (70.0)

31 (77.5) .47
Follow-up treatment
barriers

Time 10 (25.0)

6 (15.0)

7 (17.5) .50
Cost 7 (17.5)

7 (17.5)

7 (17.5) 1.00
Distance 10 (25.0)

7 (17.5)

8 (20.0) .70
Treatment satisfaction
(10 of 10) 19 (47.5)

17 (42.5)

24 (60.0) .27
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 91

Recommended for
follow-up treatment 25 (62.5)

27 (67.5)

23 (57.5) .65
Follow-up appointment
scheduled 27 (67.5)

23 (57.5)

17 (42.5) .08
Intention to receive
follow-up treatment
(0-4) 3.6 ± 1.0

3.5 ± 1.0

3.5 ± 0.9 .92
Symptom improvement (1: no improvement, 2: some improvement, 3: clear improvement, 4: cure)
Follow-up treatment barriers: participant reported 4 or 5 out of a 5-point scale (higher more barrier)
VAS: visual analog scale; CAM: complementary and alternative medicine

Primary Outcome
Primary outcome intention-to-treat analysis included all participants (N=120) based on
original group assignment. The adjusted rates of attendance to follow-up treatment in the 30 days
following baseline were 56.3% for the telephone call group, 57.3% for the text message group,
and 57.0% for the control group (Figure 4-3). As-treated analysis focused on participants who
received the intervention (23 via telephone call and 31 via text message). The adjusted rates of
attendance were 57.0% for the telephone call group and 66.0% for the text message group. There
was no significant difference in the attendance rates comparing the interventions to the control
group in both the intention-to-treat and as-treated analyses.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 92

Figure 4-3. Adjusted follow-up treatment attendance rates by group assignment

Note: Attendance rate is measured by the percentage of participants attended follow-up treatment
in 30 days after their initial visit for each group assignment. The number of participants included
in each group is shown at the bottom of the bar. Results show that there is no group difference in
the intention-to-treat analysis. Results of the as-treated analysis (participants who received the
intervention) show that attendance rates in the text message group increased but remains non-
significant comparing to the control group (p=.44). The same control group data is shown in both
sets of analysis. Standard error bars are presented. Models adjusted for pain condition and post-
treatment symptom improvement due to baseline data imbalance after randomization.
Secondary Outcomes
Predictors of Number of Follow-up Treatments
Table 4-2 posts the results from the Poisson regression models. The unadjusted models
showed that having at least 16 years of school education (IRR: 1.80, 95% CI: 1.04, 3.12, p=.04),
higher before-treatment symptom VAS (IRR: 1.01, 95% CI: 1.00, 1.02, p=.04), higher levels of
56.3 57.0
57.3
66.0
57.0
57.0
30
40
50
60
70
80
90
Intention-to-treat As-treated
% of Participants Attended Follow-up
Treatment
Telephone call Text message Control
p =.98
p =.99 p = .44
p = 1.00
40 40 40 40
31 23
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 93

illness perception (IRR: 1.03, 95% CI: 1.00, 1.05, p=.02), recommended for follow-up treatment
visit (IRR: 1.68, 95% CI: 1.01, 2.79, p=.05), a follow-up appointment scheduled (IRR: 2.85,
95% CI: 1.70, 4.78, p<.001), and stronger intent to attend follow-up treatment (IRR: 2.02, 95%
CI: 1.07, 4.78, p=.03) were significantly associated with a higher number of follow-up visits in
the 30 days following the initial visit. The forward-selection approach set in priori included five
additional covariates in the adjusted model with p<.10. Results from the adjusted Poisson model
showed that being a White/Caucasian (IRR: 1.90, 95% CI: 1.26, 2.88, p<.01), perceiving higher
levels of illness severity (BIP, IRR: 1.03, 95% CI: 1.01, 1.14, p<.01), having a higher expectancy
of acupuncture treatment (AES, IRR: 1.07, 95% CI: 1.00, 1.14, p=.04), and scheduling a follow-
up appointment (IRR: 1.95, 95% CI: 1.18, 3.20, p<.01) significantly predicted a greater number
of follow-up treatment visits within 30 days following baseline.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 94

Table 4-2: Exploratory analysis of factors predicting higher number of follow-up
treatments in 30 days

Unadjusted (N=120)

Adjusted (n=117)
IRR (95% CI) p-value

IRR (95% CI) p-value
Study Intervention

Telephone call 1.13 (0.67, 1.90) .65

0.84 (0.53, 1.33) .46
Text message 1.18 (0.71, 1.96) .53

0.84 (0.54, 1.31) .45

Female 0.88 (0.58, 1.32) .53

Age (years) 1.01 (0.99, 1.03) .16

White/Caucasian 1.66 (1.00, 2.76) .05

1.90 (1.26, 2.88) <.01
≥16 years of school 1.80 (1.04, 3.12) .04

1.60 (0.96, 2.64) .07
Currently employed 1.02 (0.51, 2.01) .96

Had health insurance 1.76 (0.69, 4.49) .24

Chronic condition 1.34 (0.85, 2.09) .21

Pain condition 0.89 (0.59, 1.36) .60

Symptom VAS: before treatment 1.01 (1.00, 1.02) .04

1.00 (0.99, 1.01) .70
Symptom VAS: after treatment 1.01 (0.99, 1.02) .19

Symptom improvement (1 to 4) 1.14 (0.97, 1.35) .12

Brief illness perception 1.03 (1.00, 1.05) .02

1.03 (1.01, 1.14) <.01
Physical disability (WHODAS
2.0) 1.01 (0.98, 1.04) .38

Attitude toward CAM 1.03 (0.98, 1.08) .23

Acupuncture expectancy scale 1.08 (1.00, 1.16) .05

1.07 (1.00, 1.14) .04

Repeated acupuncture user 1.25 (0.74, 2.12) .41

Acupuncture: paid out-of-pocket 0.67 (0.44, 1.01) .06

0.74 (0.50, 1.10) .14
Concurrent treatment(s)

MD or psychologist visits 1.33 (0.88, 2.03) .17

Prescription medications 1.05 (0.66, 1.66) .84

Complementary health
approaches 1.06 (0.67, 1.69) .79

Follow-up treatment barriers

Time 0.86 (0.72, 1.02) .09

0.83 (0.69, 1.01) .07
Cost 0.98 (0.84, 1.15) .83

Distance 0.86 (0.72, 1.03) .10

1.00 (0.82, 1.22) 1.00
Treatment satisfaction (1 to 10) 1.11 (0.96, 1.29) .16

Recommended for follow-up
visits 1.68 (1.01, 2.79) .05

1.14 (0.75, 1.77) .55
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 95

Follow-up appointment
scheduled 2.85 (1.70, 4.78) <.001

1.95 (1.18, 3.20) <.01
Intention to receive follow-up
treatment (0 to 4) 2.02 (1.07, 3.83) .03

1.53 (0.97, 2.42) .07
We performed Poisson regression for the unadjusted and adjusted model (for adjusted model forward
selection with p-value <.10)
Follow-up treatment barriers: report 4 or 5 out of a 5-point scale
IRR: incidence rate ratio, CI: confidence interval

Given we observed statistical significance for sum BIP and AES scores, we conducted
additional analyses using each subscale item as the only predictor for the scale in separate model
to understand which domains may be more influential to predicting attendance rate (Table 4-3).
For BIP, the perception at baseline that illness will continue (IRR: 1.12, 95%: 1.04, 1.21, p<.01),
having less control over illness (IRR: 1.09, 95% CI: 1.01, 1.17, p=.03), having more concern
about illness (IRR: 1.12, 95% CI: 1.04, 1.20, p<.01), and illness affecting emotion (IRR: 1.07,
95% CI: 1.00, 1.15, p=.05) predicted more follow-up visits in 30 days after initial visit. For AES,
all four domains at baseline predicted more follow-up visits, including illness will improve a lot
(IRR: 1.35, 95% CI: 1.07, 1.70, p=.01), able to cope better (IRR: 1.36, 95% CI: 1.08, 1.73,
p=.01), symptoms will disappear (IRR: 1.26, 95% CI: 1.03, 1.53, p=.03), and energy level will
increase (IRR: 1.33, 95% CI: 1.04, 1.71, p=.02) after receiving the acupuncture treatment.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 96

Table 4-3: Secondary analysis predicting attendance to follow-up treatments (number of
visits) in 30 days, focusing on the brief illness perception scale and acupuncture expectancy
scale domains
IRR (95% CI) p-value
Brief illness perception

Illness affects life 1.07 (0.99, 1.15) .10
Illness will continue 1.12 (1.04, 1.21) <.01
Control over illness* 1.09 (1.01, 1.17) .03
Treatment can help* 0.94 (0.79, 1.11) .46
Experience symptoms 1.02 (0.91, 1.15) .72
Concern about illness 1.12 (1.04, 1.20) <.01
Understand illness* 0.94 (0.86, 1.03) .17
Illness affects emotion 1.07 (1.00, 1.15) .05

Acupuncture expectancy scale

Illness will improve a lot 1.35 (1.07, 1.70) .01
Able to cope better 1.36 (1.08, 1.73) .01
Symptoms will disappear 1.26 (1.03, 1.53) .03
Energy level will increase 1.33 (1.04, 1.71) .02
*Reserved scoring
We performed separate multivariate Poisson regression models for
each subscale item. All models controlled for covariates listed in the
Table 4-2 adjusted model
IRR: Incidence rate ratio, CI: confidence interval

Intent to Attend Follow-up Treatment by Intervention Assignments
We examined the change in participant intent to attend follow-up treatment across the 30-
day period following baseline by group assignment in the as-treated sample (total n=94: 23
telephone call, 31 text message, 40 control). The levels of intention decreased significantly in the
text message (3.5 to 3.2, p=.03) and control (3.5 to 3.3, p=.03) groups across the 30-day period,
while the change in the telephone call group was not significant (3.7 to 3.5, p=.27). However,
after adjusting for the level of acupuncture expectancy at baseline, there was no difference in the
trajectories of intention by group over the 30-day period (p=.98) (Figure 4-4).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 97

Figure 4-4. Adjusted means of participant intent to attend follow-up treatment among the
as-treated sample over 30 days, by intervention assignments (N=94)

Note: The analysis included the subset of 94 participants who received the interventions
(telephone call: 23, text message: 31) and all in the control group (40). Baseline intent was
measured immediately after the initial treatment. Intent measurement ranges from 0-4. Group by
time difference is non-significant (p=.98). Text message and control groups (p=.03 and p=.03)'s
level of intent decreased significantly from Baseline to Day 30, but not telephone call group
(p=.46). The grey bar represents the intervention time period at Day 3 and 4. Model adjusted for
baseline acupuncture expectancy sum score.
Pain Level and Pain Disability by Attendance to Follow-up Treatment
Participants with a pain-related complaint at baseline (n=73) demonstrated moderate to
severe levels of pain prior to their initial treatment with an adjusted mean (SD) score at 58.8
(24.5) (Figure 4-5). Participants in both groups showed significant improvement in pain levels
immediately after the initial treatment (p<.001). However, after adjusting for study interventions
3.7
3.6
3.5
3.5
3.3
3.2
3.5
3.3
3.3
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
4.0
Baseline Day 10 Day 30
Intention to Follow-up
Telephone Call Text Message Control
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 98

and chronicity, the difference in the trajectories of pain levels by attendance to follow-up
treatment was not significant over a 30-day period following baseline (p=.42). For pain
disability, participants reported an adjusted mean (SD) of 25.7 (19.3) at baseline (Figure 4-6).
Participants who attended follow-up treatment demonstrated a significant improvement in pain
disability over a 30-day period (28.1 to 20.7, p<.001). In contrast, the non-attendees showed a
small initial improvement at Day 10 (22.2 to 19.7), but their pain disability scores bounced back
at Day 30 (19.7 to 21.0). After adjusting for study interventions and chronicity, the difference in
trajectories across the two groups achieved a marginal significance over the 30-day period
following baseline (p=.05).
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 99

Figure 4-5. Adjusted mean levels of pain among the pain cohort over 30 days, by
attendance to follow-up treatment (n=73)

Note: Participants’ pain levels were measured using the visual analog scale at four time points.
Day 0 had two measures: before and immediately after the initial treatment. The group adjusted
means are presented at each time point in the figure. Group by time trend difference is non-
significant (p=.42), meaning that participants who attended follow-up visit did not show a greater
improvement in pain levels over 30 days compared to those did not attend follow-up visit.
However, both groups present a significant improvement in pain levels immediately after the
initial treatment (p<.001). The model controlled for study interventions and participant’s self-
report of an acute vs. chronic condition at baseline.
62.0
34.0 33.9
33.1
54.2
34.7
32.8 32.0
0
10
20
30
40
50
60
70
Before treatment After treatment Day 10 Day 30
Pain Levels
Attended Not Attended
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 100

Figure 4-6. Adjusted mean levels of pain disability among the pain cohort over 30 days, by
attendance to follow-up treatments (n=73)

Note: Levels of pain disability were measured by the sum score of Pain Disability Index at three
time points. Participants reported the baseline pain disability immediately after the initial
treatment. We observe a marginal significance for group by time trend difference (p=.05),
meaning that participants who attended follow-up treatment had a greater improvement in pain
disability over 30 days compared to those who did not. There is significant improvement in pain
disability over 30 days in participants who attended follow-up treatment (p<.001), but not for
those who did not (p=.42). The model controlled for study interventions and participant’s self-
report of an acute vs. chronic condition at baseline.
28.1
21.9
20.7
22.2
19.7
21.0
10
15
20
25
30
35
Baseline Day 10 Day 30
Pain Disability
Attended Not Attended
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 101

Discussion
This randomized controlled trial compared the effect of a one-time telephone call
reminder, a one-time text message reminder, and treatment as usual (i.e. no phone contact) on
patient return to acupuncture follow-up treatment within a 30-day period. Our main findings
indicate that a telephone call and text message did not improve patient return rate or their intent
to attend follow-up treatment. Results from the exploratory analyses showed that being
White/Caucasian, perceiving a more severe illness, having a higher expectancy of acupuncture
treatment, and scheduling a follow-up appointment yielded a higher number of follow-up visits
during the 30 days following baseline. Participants who returned for follow-up treatment
reported greater improvement in pain disability across a 30-day assessment period. Although our
findings did not support that a one-time telephone call or text message reminder benefits patient
follow-up treatment rates, attending follow-ups may have a role in addressing pain disability.
This observation provides a key implication for further research into strategies that affect follow-
up treatment attendance in acupuncture patients.
Our telephone call and text message reminders failed to increase patient return to
acupuncture follow-up treatment. We adopted a minimal, one-time reminder approach based on
practical concerns in real clinic environments. Implementing a reminder system requires staffing
and monetary resources, which are often scarce in private practice. These concerns are applicable
to most acupuncture clinics in the U.S. because many practices are privately owned (Kaptchuk,
2002; Wang et al., 2018). On the other hand, a multi-cue intervention strategy may be more
effective (Schwebel & Larimer, 2018; Thakkar et al., 2016). Arora et al. (2015) sent participants
text message reminders 7 days, 3 days, and 1 day prior to their follow-up appointments. This
strategic difference could be linked to our lack of positive finding. A one-time reminder system
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 102

is suitable for private clinics because it is cost-effective and not labor intensive. The
disadvantages, however, include low intervention dosage and cannot overcome more important
psychosocial factors that dictate patient’s follow-up attendance behavior (i.e. uncertain of
treatment outcomes, lack of coverage) (Barlow et al., 2011; Bishop et al., 2011; Rugg et al.,
2011). Future studies can conduct a more robust evaluation by targeting their interventions on
patient cohorts with appointments scheduled and a treatment plan for follow-up visits (Lam et
al., 2019).
In our exploratory analyses, scheduling a follow-up appointment predicted more follow-
up visits in the 30-day after the initial treatment. Appointment scheduling forms a contract
between patients and their providers (Bosch-Capblanch, Abba, Prictor, & Garner, 2007). This
can reflect patient’s intent to continue care using acupuncture. Our previous study showed that
having multiple chief complaints was associated with more follow-up visits (Lam et al., 2019).
This is similar to our current finding, showing that having the perception of a more severe illness
may lead participants to seek additional treatments. However, this observation challenges
Bishop’s, who showed more severe illness could act as a barrier for patients to access follow-up
services (Bishop et al., 2017). Interestingly, being White results in more follow-up visits. This is
a potential artifact due to the clinic located in an urban, wealthy, White majority neighborhood.
Nevertheless, this is worth noting because race/ethnicity can play a role in utilization of CHA, in
particular acupuncture due to socioeconomic status and treatment coverage (Meghani & Liou,
2019). Our results showed that patients with higher expectancy on acupuncture had more follow-
up visits. This is a novel observation given that higher expectancy on acupuncture has only been
shown to associate with better treatment outcomes (Linde et al., 2007; Prady, Burch,
Vanderbloemen, Crouch, & MacPherson, 2015). Presumably, higher treatment expectancy
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 103

combining with positive treatment outcome may encourage patients to continue their prescribed
treatment plan (Barlow et al., 2011). However, both ours and Bishop et al. (2017)’s findings do
not show that symptom improvement increases follow-up attendance rates. A future study using
a prospective mediation analysis design may provide further explanation to these relationships.
A meta-analysis by MacPherson et al. (2013) found that patients pain symptoms
improved for every five acupuncture sessions they received. Our study supports this finding
given that participants who attended follow-up treatment demonstrated greater improvement in
pain disability over the 30-day period following baseline. For low back pain, international
experts (Molsberger et al., 2008) and acupuncture literature (Yuan, Kerr, et al., 2008) indicate
the use of an extended treatment regimen (i.e. median 10-session). Repeated acupuncture
stimulation can have a cumulative therapeutic effect to produce a habitual brain response (Li et
al., 2014). These findings support the role of having multiple acupuncture treatment sessions for
pain-related conditions. Our negative finding for group differences in pain levels could be due to
the fact that participants who continued treatment presented with more severe symptoms at
baseline and across time, which limited their capability to achieve better outcomes than the non-
attendees. However, we noticed that both groups improved significantly immediately after the
initial treatment, and that the treatment effect persisted over 30 days even among the non-
attendees. Our positive finding on pain disability may support acupuncture as a method for
addressing functional abilities (Shin et al., 2013) and psycho-social factors including pain
catastrophizing and fear-avoidance associated with pain symptoms (Bishop et al., 2015). More
studies are required to verify the longitudinal effect of acupuncture on treatment of pain
conditions.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 104

Study Limitations
Our study has several limitations. First, the small sample size underpowered our trial and
hindered our ability to detect significant group difference. However, we observed a trend in
which text message recipients yielded a higher rate for follow-up treatment in the 30 days
following baseline. Second, our study took place at a single teaching acupuncture clinic, thus
limiting the generalizability of our findings to other private practices with potentially different
patient populations. Third, we applied our single intervention approach to all participants, rather
than health condition specific. Targeting patients with chronic conditions requiring extended
treatment may result in higher follow-up return rates than patients presenting with an acute
symptom. Fourth, we used the carried-forward imputation method for missing data at Day 10 and
30 on the longitudinal outcomes. Results should be interpreted with caution even though we
adopted this conservative imputation approach. Nevertheless, our study provides the first
systematic evaluation of the effect of two reminder systems, telephone call and text message, to
improve patient return rates to acupuncture follow-up treatment. Our finding on a greater
improvement in pain disability in follow-up attendees provides further evidence to support the
need to improve follow-up attendance for acupuncture treatment.
Conclusions
In conclusion, a one-time reminder system using a telephone call or text message did not
result in higher patient return rates to acupuncture follow-up treatment. Participants returning for
follow-up treatment, however, demonstrated greater improvement in pain disability across a 30-
day assessment period. Targeted and multi-dimensional interventions designed to improve
treatment attendance hold promises to provide further health benefits to patients using
acupuncture treatment.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 105

Chapter 5: Summary and Conclusions
Summary of Aims and Findings
This dissertation uses research priorities from the Federal Pain Research Strategy from
the NIH to develop a research agenda focusing on pain policy, pain mechanisms, and integrative
pain treatments. This dissertation has three overarching aims: 1) to examine the effect of a drug
policy on the demand of prescription opioids, 2) to explore the mechanisms linking mindfulness
disposition and pain through the impact on psychological distress, and 3) to evaluate the impact
of two mobile interventions on improving attendance to follow-up acupuncture treatment. These
aims were fulfilled via three studies. The overall objective of Study 1 was to investigate the
effect of the Prescription Drug Monitoring Programs in reducing the overall amount of
prescription opioid shipments in 45 states during 1997-2014. The overall objective of Study 2
was to examine whether psychological distress including depression, anxiety and stress mediate
the association between mindfulness disposition and pain intensity among orthopedic patients.
The overall objective of Study 3 was to assess the effect of telephone call and text message
reminders in improving the follow-up treatment attendance rate among acupuncture patients, and
to explore whether attending follow-up visits is associated with greater improvement in pain
symptoms.
The results from Study 1 demonstrated that the growth rate of prescription opioid
shipments was significant reduced after the implementation of Prescription Drug Monitoring
Programs (PDMP). We used the random effects piecewise growth curve modeling approach to
compare the pre- and post-PDMP growth profiles of prescription opioid shipments across 45
states during 1997-2014. We controlled for whether a state implemented PDMP before or after
2010, had adopted the Pill Mill law during the study period, and their interaction with the post-
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 106

PDMP period. For states that implemented PDMP after 2010, we observed a further reduction in
the prescription opioid shipment trajectory in the post-PDMP period compared to states
implemented PDMP before 2010. For states that adopted the Pill Mill law, we also observed a
further reduction in prescription opioid shipments in the post-PDMP period. Contrary to our
second hypothesis, the strength of PDMP mandate did not moderate the growth trajectory of the
post-PDMP period. This observation suggested that states that adopted a stronger policy did not
result in a greater reduction in prescription opioid shipments. Our secondary analyses showed
that the Pill Mill laws affected prescription opioid shipments only to the pharmacies but not the
hospitals, while the pre-PDMP growth profiles of both fentanyl and hydrocodone had plateau at
the post-PDMP period. Overall, our study findings indicated that the PDMP policy was
associated with a reduction in prescription opioid shipments.
Findings from Study 2 showed that psychological distress mediated the association
between mindfulness disposition and pain in orthopedic patients. We conducted a multiple
mediation analysis using data from a cross-sectional survey to examine the potential mechanisms
linking domains of mindfulness disposition on pain via the influence on depression, anxiety and
stress. As hypothesized, we observed a significant correlation between the overall mindfulness
disposition measured by the Five Facet Mindfulness Questionnaire (FFMQ) sum score and pain
intensity, but this direct effect disappeared in the mediation model. Results from the single
mediation model demonstrated a significant indirect effect of mindfulness disposition on pain,
mediated through psychological distress measured by the Depression, Anxiety, Stress Scale
(DASS). Depression and stress were the significant mediators in the multiple mediation model,
each mediated at least 30% of the association between mindfulness disposition and pain. The
examination of specific FFMQ domains showed that non-judging, non-reactivity and acting-
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 107

with-awareness indirectly affected pain through their influence on depression and stress. The
overall findings support a mindfulness-distress-pain (i.e. attention-affect-pain) model for future
study design and the specific pathways involved in the mechanisms.
Study 3 findings provided insights into factors associated with attendance to acupuncture
follow-up treatment. Our primary hypothesis was to show that a one-time telephone call and text
message reminders would increase follow-up treatment attendance rate within 30 days after
patient’s initial visit in a randomized controlled trial. However, neither interventions improved
the follow-up visit rate compared to the no intervention control group. Our exploratory Poisson
regression analysis suggested that baseline characteristics including being White, perceiving
one’s illness to be more severe, having a higher expectancy on acupuncture treatment, and
scheduling a follow-up appointment predicted higher number of follow-up visits within 30 days
after the initial visit. For longitudinal data analysis involving three time points (i.e. Day 0, Day
10, Day 30), participant’s intent to attend follow-up visit deteriorated over the 30-day period and
the trajectories did not differ by intervention assignment. Contrary to our hypothesis, attending
follow-up visit(s) did not result in greater improvement in pain levels across the 30-day period
after the initial visit among patients reporting a pain condition. However, we observed a
significant improvement in pain disability among patients who attended follow-up visit(s), while
there was bounce back of pain disability scores at Day 30 among those who did not have a
follow-up. The findings provided evidence for the clinical values of acupuncture follow-up
treatment on pain-related outcomes, and the implications for designing targeted interventions to
improve follow-up attendance rate among acupuncture patients.
The findings from these three studies provide new insights into pain research from
multiple angles by addressing specific research priorities in the Federal Pain Research Strategies.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 108

Overall, findings from Study 1 support that a pain policy, such as the Prescription Drug
Monitoring Programs, can impact the overall demand of prescription opioids across the United
States. As shown in the data, nearly all 50 states had implemented the policy during 1997-2014,
and each state-specific PDMP had a differential effect on the observed reduction of prescription
opioid shipments. This positive result is consistent with studies focusing on a similar period,
while their outcomes focused on the opioid prescribing patterns among physicians and the use of
prescription opioids among Medicare beneficiaries (Bao et al., 2016; Moyo et al., 2017). Based
on findings from previous research (Pardo, 2017; Wen et al., 2017), we expected that there
would be a moderating effect on the level prescription opioid reduction for states with different
policy strengths. Specifically, we postulated that states with a stronger policy mandate would
result in greater opioid reduction. However, this hypothesis was not supported by the findings,
suggesting that there may be other factors such as the actual provider registration and access to
the PDMP database would affect the policy results (Haffajee et al., 2015).
Study 2 examined the mechanisms of how mindfulness disposition and psychological
distress may affect pain intensity. We anticipated that greater mindfulness disposition would be
associated with lower pain levels, and psychological distress would mediate this relationship.
Our results confirmed this cascade of effects, and showed that acting-with-awareness, non-
judging and non-reactivity mindfulness domains affected pain through their influence on
depression and stress symptoms. Our findings are consistent with current literature, by showing
that mindfulness disposition moderated the relationship between distress and pain (Dorado et al.,
2018; A. C. Lee et al., 2017; Poulin et al., 2016). The findings suite the proposed attention-
stress-pain model, and is in accordance with the theory that being mindful may help one to
encounter pain experience without unnecessary elaboration, catastrophizing or reactivity (Day et
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 109

al., 2014; Keefe et al., 2004; Veehof et al., 2016). The non-judging and non-reactivity domains
are particular relevant because they intent to measure and reflect the attention on and processing
of inner experience (Baer et al., 2006; Bohlmeijer et al., 2011), in this case possibly related to the
musculoskeletal pain that orthopedic patients often experience and encounter on their daily life.
Overall, our hypotheses of a mindfulness-distress-pain / attention-stress-pain mechanism were
supported by the study findings, and this mechanism could have important implication to future
studies aiming at expending the construct and applying to the design of pain management
interventions.
Study 3 addressed a broader context of the FPRS on the evaluation of non-
pharmacological pain treatment and adherence strategies. To our knowledge, Study 3 is the first
randomized controlled trial to test the effectiveness of a reminder system aiming to improve
patient return rate to acupuncture follow-up treatment. Only until recent years, the issue of low
treatment adherence in acupuncture setting has been addressed by research (Bishop et al., 2017;
Lam et al., 2019). Our findings did not support our one-time telephone call and text message
reminders to improve patient attendance rate to follow-up treatment. This observation, contrary
to previous finding about text message improving appointment attendance (Arora et al., 2015),
could due to strategic differences in reminder frequency and a tailored messaging approach. In
our exploratory analysis, we found that patient scheduling follow-up appointments and having a
higher treatment expectancy predicted more follow-up visits. Appointment scheduling may
reflect a patient’s intention to continue care using acupuncture, thus establishing a contract with
their provider (Bosch-Capblanch et al., 2007). Presumably, a higher treatment expectancy
combining with positive treatment outcome may encourage patients to continue their prescribed
treatment plan (Barlow et al., 2011). Findings from Study 3 point to the directions for future
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 110

research on strategies to improve treatment adherence to acupuncture that can be applied in real
clinical setting.
Implications and future Research
The Federal Pain Research Strategy (FPRS) specifically highlights research priorities on
pain policy, pain mechanisms, and integrative pain treatments to advance the science of pain and
improve pain care. Understanding how regulatory and legislative policies have impacted models
of pain care is important because it allows more rigorous evaluation on policy design and policy
adoption across states. Policy evaluations provide insights into whether the policy can achieve
the original goals and is effective as intended. The PDMP policy is designed to monitor and
prevent the diversion of controlled substances. However, PDMP’s functions to curb doctor
shopping behaviors and prescription drug misuse in patients, as well as to identify excessive
prescription patterns in providers or pharmacies have become the central focus amid the opioid
epidemic. Findings in Study 1 showed that there was a rate reduction in prescription opioid
shipments post PDMP implementation. This observation provides evidence on the effectiveness
of the policy to lower opioid demands in a broad picture, although not specific to the level that is
directly assessing the policy effect on opioid consumption, diversion, or prescribing pattern.
Nevertheless, the results will advocate for stronger government support to strengthen the policy
programs at states where deeply affected by opioid abuse and overdose deaths. Although our
result did not show that the level of opioid shipments varied upon on the strength of PDMP
mandate, it may suggest that a more precise and appropriate evaluation approach is to use the
actual PDMP registration and access data rather than basing on the state policy recommendation.
A direction for future study is to combine multiple data sources which include PDMP
registration and access data to address how these mandates are followed through and its impact
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 111

on specific opioid outcomes concerning the opioid crisis. Overall, our findings have the
implications for the federal government to further support this state-implemented policy, and to
ensure the intended effects of the PDMP on opioid reduction are delivered in a consistent
manner.
The FPRS emphasizes the study of pain mechanisms in the context of biopsychosocial
models because pain can have a differential effect across demographic and clinical subgroups.
Study 2 focused on orthopedic patients given pain is a leading complaint in this population. We
examined the psychological aspects of the pain mechanism, while focusing on how and attention
and affect might play a role in the overall pain experience. Our results showed that psychological
distress mediated the association between mindfulness disposition and pain. Mindfulness may
function to remediate pain through its impact on improving distress levels. Findings have
implications to support mindfulness as a positive psychology disposition that assists orthopedic
patients to better self-manage their distress-related pain problems. This is in particular relevant
because our study also found that 49% of our sample had at least mild level of either depression,
anxiety or stress. Examining the roles of common comorbidities and their associated mechanisms
with pain are key focuses in the FPRS. Results underscore the importance for providers to
identify and address problems related to psychological distress when they are treating pain
conditions. In addition, identifying the associations of mindfulness on distress and pain provide
evidence for non-pharmacologic approach such as mindfulness meditation as self-management
strategies for orthopedic pain patients. Our findings that mindfulness domains including non-
judging and non-reactivity indirectly affected pain through their influence on depression and
anxiety allow interventions to target these specific mechanistic pathways. A future study can
expand on including other comorbidities, including sleep disturbance in the mediation model,
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 112

and whether these pathways differ by sex. For example, females may benefit from mind body
interventions on pain primarily via the reduction of stress symptoms, as compared to improving
sleep quality in males. These findings will help future studies to design interventions that are
mechanism-specific and tailored to individual characteristics.
The FPRS calls for the identification and evaluation of non-pharmacological approaches
for pain management. Part of the evaluation process is to determine the adherence strategies for
therapies that require multiple treatments over a therapeutic course. Improving adherence, in the
context of patient attendance to follow-up treatment has a strong implication for acupuncture
treatment. Studies have shown that acupuncture patients benefit from a cumulative therapeutic
result over a continuous regimen, in particular for pain and chronic conditions. Study 3 used a
randomized controlled trial to test two evidence-based reminder systems, telephone call and text
message to determine their impact on the follow-up attendance rates in real acupuncture clinical
setting. Although both interventions did not improve follow-up attendance rate in the intention-
to-treat analysis, those responded to the text message intervention showed a higher but non-
significant rate of return. In addition, the exploratory analysis shed light on other factors
including illness perception and appointment scheduling on patients’ returning to follow-up
treatment. More importantly, results from the longitudinal analysis confirmed existing findings
that attending follow-up treatment would improve pain disability over the 30-day period after the
initial visit. This positive finding advocates for the need to boost treatment follow-up rates in
order for patients to receive the optimal therapeutic benefits from acupuncture treatment. Future
studies can continue to focus on text message with a more robust design, which include multiple
reminders and tailored messaging content, and directly focus on patient cohorts that have
received a recommendation for follow-up treatment. This will offer further insights to strengthen
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 113

the protocol and clinical operation of acupuncture treatment to better address the pain treatment
needs.
Limitations
This dissertation has several limitations and is study specific. Study 1 tested the effect of
the PDMP policy on prescription opioid demands and used prescription opioid shipments as the
proxy outcome measure. This approach is a major strength given the ARCOS dataset has
consistent reporting of specific prescription opioids by state across years; however, opioid
shipment data can only provide a broad picture, and cannot make a direct reference to opioid
consumption, diversion, or prescribing pattern. In addition, we compiled a three-level PDMP
strength based on state policy mandate recommendation. Whether how closely the providers
followed these mandates, in terms of registering and accessing the PDMP data exchange, would
have a direct impact on how the mandate could affect the policy outcomes. The unique
application of a random effects piecewise growth curve modeling approach on policy research
still needs further evaluation to maximize it utility for an accurate assessment. We focused on a
parsimonious model to only control for the presence of Pill Mill law, while we are fully aware
there are additional state-level policy that could affect the demand of prescription opioids during
the study period. Although there are cautions to be made when interpreting the results, the study
findings build onto the current evidence supporting the PDMPs as an effective policy to tackle
the opioid crisis.
For Study 2 and Study 3, both studies took place in a teaching hospital and a teaching
clinic. The single-site study design can limit the external validity and generalizability of the
findings due to the specific patient population at this location. Both studies had a majority White
sample; therefore, the results could not make cross-reference to other racial and ethnic groups.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 114

Given both study cohorts were recruited in a clinical setting, these participants could be a more
health-conscious subset of the population or had greater awareness toward their symptoms and
treatment options, thus affecting their response to the surveys. For Study 2, we used cross-
sectional data for a mediation analysis. This approach limits our ability to draw conclusion on a
casual interference due to a lack of timing factor in the study design, but is capable to analyze
from an associational perspective. For Study 3, we implemented the intervention to any clinic
patients agreed to participate in the acupuncture trial. The small sample of 120 lacked adequate
power to detect a statistical difference in follow-up attendance rates based on group assignment.
Over 90% of patients in this clinic paid out-of-pocket for the acupuncture service they received.
This could affect their follow-up treatment attendance behavior and we would expect to see a
different pattern for those who have insurance coverage. Nevertheless, the major strength of
Study 3 is that to our knowledge, this is the first study to test a behavioral intervention using the
cues-to-action construct from the Health Behavior Theory in an acupuncture treatment setting.
With the FPRS calling for the search of non-pharmacological therapies for pain, understanding
the mechanisms, effectiveness and finding the appropriate adherence strategies will contribute to
the overall scope of pain research.
Contribution to the Literature
Findings from this dissertation address the Federal Pain Research Strategies by making
important contributions to the body of pain research focusing on pain policy, pain mechanisms
and integrative pain treatment. Our pain policy study has used a random effects piecewise
growth curve modeling approach to study the impact of the Prescription Drug Monitoring
Programs on prescription opioid shipments. It introduced an alternative approach to traditional
fixed effects analysis in policy research which allows for different modeling assumptions and
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 115

analytic strategies. The null finding of policy strength on opioid reduction also confirms the need
to use actual PDMP registration and access data in future study. The pain mechanism study
explored psychological processes involved in the association between mindfulness disposition
and pain. The positive finding of a mediation through psychological distress suggests that
mindfulness disposition as an attention construct can influence pain through reducing the
negative impact of psychological affect associated with the experience of pain. Results from our
cross-sectional data can also inspire longitudinal study design to confirm findings with a
temporal sequence. Additionally, this dissertation includes the first randomized controlled trial
testing an adherence strategy to acupuncture treatment. The negative result of using a one-time
telephone call or text message reminders to improve patient follow-up treatment attendance rate
suggests that more robust intervention design and selective patient cohorts should be considered
in future research. The finding of further reduction in pain disability over time among patients
who attended follow-up treatment provides justification to invest in adherence strategies for
acupuncture treatment. In summary, the findings from this dissertation provide insights and
future directions to advancing research in pain policy, pain mechanisms and integrative pain
treatments.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 116

Alphabetic Bibliography
Adams, N., Rose, M. J., & Ravey, J. (1997). The psychophysiology of low back pain: WB
Saunders Company.
Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: does it
matter? Journal of health and social behavior, 1-10.
Andrews, G., Kemp, A., Sunderland, M., Von Korff, M., & Ustun, T. B. (2009). Normative Data
for the 12 Item WHO Disability Assessment Schedule 2.0. PLoS ONE, 4(12), e8343. doi:
10.1371/journal.pone.0008343
Arora, S., Burner, E., Terp, S., Lam, C. N., Nercisian, A., Bhatt, V., & Menchine, M. (2015).
Improving Attendance at Post –Emergency Department Follow ‐up Via Automated
Text Message Appointment Reminders: A Randomized Controlled Trial. Academic
emergency medicine, 22(1), 31-37.
Baer, R. A., Smith, G. T., Hopkins, J., Krietemeyer, J., & Toney, L. (2006). Using self-report
assessment methods to explore facets of mindfulness. Assessment, 13(1), 27-45. doi:
10.1177/1073191105283504
Baker, D. W. (2017). The Joint Commission's Pain Standards: Origins and Evolution. Oakbrook
Terrace, IL: The Joint Commission.
Ballantyne, J. C., & Mao, J. (2003). Opioid Therapy for Chronic Pain. New England Journal of
Medicine, 349(20), 1943-1953. doi: 10.1056/NEJMra025411
Bao, Y., Pan, Y., Taylor, A., Radakrishnan, S., Luo, F., Pincus, H. A., & Schackman, B. R.
(2016). Prescription drug monitoring programs are associated with sustained reductions
in opioid prescribing by physicians. Health Affairs, 35(6), 1045-1051.
Barlow, F., Scott, C., Coghlan, B., Lee, P., White, P., Lewith, G. T., & Bishop, F. L. (2011).
How the psychosocial context of clinical trials differs from usual care: a qualitative study
of acupuncture patients. BMC Med Res Methodol, 11, 79. doi: 10.1186/1471-2288-11-79
Bartley, E. J., & Fillingim, R. B. (2013). Sex differences in pain: a brief review of clinical and
experimental findings. British journal of anaesthesia, 111(1), 52-58. doi:
10.1093/bja/aet127
Becker, M. H. (1974). The health belief model and sick role behavior. Health education
monographs, 2(4), 409-419.
Beks, R. B., Mellema, J. J., Menendez, M. E., Chen, N. C., Ring, D., & Vranceanu, A. M.
(2018). Does mindfulness correlate with physical function and pain intensity in patients
with upper extremity illness? Hand, 13(2), 237-243.
Bell, A., & Jones, K. (2015). Explaining fixed effects: Random effects modeling of time-series
cross-sectional and panel data. Political Science Research and Methods, 3(1), 133-153.
Bishop, F. L., Barlow, F., Coghlan, B., Lee, P., & Lewith, G. T. (2011). Patients as healthcare
consumers in the public and private sectors: a qualitative study of acupuncture in the UK.
BMC Health Serv Res, 11, 129. doi: 10.1186/1472-6963-11-129
Bishop, F. L., Yardley, L., Cooper, C., Little, P., & Lewith, G. (2017). Predicting adherence to
acupuncture appointments for low back pain: a prospective observational study. BMC
Complement Altern Med, 17(1), 5. doi: 10.1186/s12906-016-1499-9
Bishop, F. L., Yardley, L., Prescott, P., Cooper, C., Little, P., & Lewith, G. T. (2015).
Psychological covariates of longitudinal changes in back-related disability in patients
undergoing acupuncture. Clin J Pain, 31(3), 254-264. doi:
10.1097/AJP.0000000000000108
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 117

Bohlmeijer, E., ten Klooster, P. M., Fledderus, M., Veehof, M., & Baer, R. (2011). Psychometric
properties of the five facet mindfulness questionnaire in depressed adults and
development of a short form. Assessment, 18(3), 308-320. doi:
10.1177/1073191111408231
Bosch-Capblanch, X., Abba, K., Prictor, M., & Garner, P. (2007). Contracts between patients
and healthcare practitioners for improving patients' adherence to treatment, prevention
and health promotion activities. Cochrane Database Syst Rev(2), Cd004808. doi:
10.1002/14651858.CD004808.pub3
Brandeis University. Prescription during monitoring program training and technical assistance
center. Retrieved June 10, 2018, from http://www.pdmpassist.org/
Branstrom, R., Duncan, L. G., & Moskowitz, J. T. (2011). The association between dispositional
mindfulness, psychological well-being, and perceived health in a Swedish population-
based sample. Br J Health Psychol, 16(Pt 2), 300-316. doi: 10.1348/135910710x501683
Broadbent, E., Petrie, K. J., Main, J., & Weinman, J. (2006). The brief illness perception
questionnaire. J Psychosom Res, 60(6), 631-637. doi: 10.1016/j.jpsychores.2005.10.020
Brooks, J. M., Iwanaga, K., Cotton, B. P., Deiches, J., Blake, J., Chiu, C., . . . Chan, F. (2018).
Perceived mindfulness and depressive symptoms among people with chronic pain.
Journal of rehabilitation, 84(2), 33.
Brown, K. W., Weinstein, N., & Creswell, J. D. (2012). Trait mindfulness modulates
neuroendocrine and affective responses to social evaluative threat.
Psychoneuroendocrinology, 37(12), 2037-2041. doi: 10.1016/j.psyneuen.2012.04.003
Brown, T. A., Chorpita, B. F., Korotitsch, W., & Barlow, D. H. (1997). Psychometric properties
of the Depression Anxiety Stress Scales (DASS) in clinical samples. Behav Res Ther,
35(1), 79-89.
Campbell, W. I., & Lewis, S. (1990). Visual analogue measurement of pain. The Ulster Medical
Journal, 59(2), 149-154.
Castillo, R. C., Archer, K. R., Newcomb, A. B., & Wegener, S. T. (2016). Pain and
psychological distress following orthopedic trauma: a call for collaborative models of
care. Techniques in Orthopaedics, 31(4), 228-234.
Castillo, R. C., Wegener, S. T., Heins, S. E., Haythornthwaite, J. A., Mackenzie, E. J., & Bosse,
M. J. (2013). Longitudinal relationships between anxiety, depression, and pain: results
from a two-year cohort study of lower extremity trauma patients. PAIN, 154(12), 2860-
2866. doi: 10.1016/j.pain.2013.08.025
Centers for Disease Control and Prevention. (2017). What states need to know about PDMPs.
Retrieved June 10, 2018, from https://www.cdc.gov/drugoverdose/pdmp/states.html
Cherkin, D. C., & Herman, P. M. (2018). Cognitive and Mind-Body Therapies for Chronic Low
Back Pain and Neck Pain: Effectiveness and Value. JAMA internal medicine.
Chou, C. P., Yang, D., Pentz, M. A., & Hser, Y. I. (2004). Piecewise growth curve modeling
approach for longitudinal prevention study. Computational Statistics & Data Analysis,
46(2), 213-225.
Chou, R., Deyo, R., Friedly, J., Skelly, A., Hashimoto, R., Weimer, M., . . . Griffin, J. (2016).
Noninvasive Treatments for Low Back Pain.
Chow, C. K., Redfern, J., Hillis, G. S., Thakkar, J., Santo, K., Hackett, M. L., . . . Thiagalingam,
A. (2015). Effect of lifestyle-focused text messaging on risk factor modification in
patients with coronary heart disease: A randomized clinical trial. JAMA, 314(12), 1255-
1263. doi: 10.1001/jama.2015.10945
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 118

Cicero, T. J., Ellis, M. S., & Kasper, Z. A. (2017). Increased use of heroin as an initiating opioid
of abuse. Addictive behaviors, 74, 63-66.
Cicero, T. J., Ellis, M. S., Surratt, H. L., & Kurtz, S. P. (2013). Factors influencing the selection
of hydrocodone and oxycodone as primary opioids in substance abusers seeking
treatment in the United States. PAIN®, 154(12), 2639-2648.
Clark, T., Eadie, J., Kreiner, P., & Strickler, G. (2012). Prescription drug monitoring programs:
an assessment of the evidence for best practices. The Prescription Drug Monitoring
Program Center of Excellence.
Clarke, T. C., Black, L. I., Stussman, B. J., Barnes, P. M., & Nahin, R. L. (2015). Trends in the
Use of Complementary Health Approaches Among Adults: United States, 2002–2012.
Natl Health Stat Report(79), 1-16.
Clarke, T. C., Nahin, R. L., Barnes, P. M., & Stussman, B. J. (2016). Use of Complementary
Health Approaches for Musculoskeletal Pain Disorders Among Adults: United States,
2012. Natl Health Stat Report(98), 1-12.
Curtin, K. B., & Norris, D. (2017). The relationship between chronic musculoskeletal pain,
anxiety and mindfulness: Adjustments to the Fear-Avoidance Model of Chronic Pain.
Scandinavian journal of pain, 17(1), 156-166.
Day, M. A., Jensen, M. P., Ehde, D. M., & Thorn, B. E. (2014). Toward a theoretical model for
mindfulness-based pain management. J Pain, 15(7), 691-703. doi:
10.1016/j.jpain.2014.03.003
Day, M. A., Smitherman, A., Ward, L. C., & Thorn, B. E. (2015). An investigation of the
associations between measures of mindfulness and pain catastrophizing. Clin J Pain,
31(3), 222-228. doi: 10.1097/ajp.0000000000000102
de Frias, C. M., & Whyne, E. (2015). Stress on health-related quality of life in older adults: the
protective nature of mindfulness. Aging Ment Health, 19(3), 201-206. doi:
10.1080/13607863.2014.924090
Disabato, D. (2016). The double standard against cross-sectional mediation. Applied Statistics
Blog. Retrieved April 1, 2019, from http://www.daviddisabato.com/blog/2016/5/22/the-
double-standard-against-cross-sectional-mediation
Dorado, K., Schreiber, K. L., Koulouris, A., Edwards, R. R., Napadow, V., & Lazaridou, A.
(2018). Interactive effects of pain catastrophizing and mindfulness on pain intensity in
women with fibromyalgia. Health Psychol Open, 5(2), 2055102918807406. doi:
10.1177/2055102918807406
Drug Enforcement Administration. Automation of Reports and Consolidated Orders System
(ARCOS). Retrieved 2018, June 16, from
https://www.deadiversion.usdoj.gov/arcos/index.html
Duffield, S. J., Ellis, B. M., Goodson, N., Walker-Bone, K., Conaghan, P. G., Margham, T., &
Loftis, T. (2017). The contribution of musculoskeletal disorders in multimorbidity:
Implications for practice and policy. Best Practice & Research Clinical Rheumatology,
31(2), 129-144. doi: https://doi.org/10.1016/j.berh.2017.09.004
Dugashvili, G., Van den Berghe, L., Menabde, G., Janelidze, M., & Marks, L. (2016). Use of the
universal pain assessment tool for evaluating pain associated with TMD in youngsters
with an intellectual disability. Medicina oral, patologia oral y cirugia bucal, 22(1), e88-
e94. doi: 10.4317/medoral.21584
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 119

Edwards, R. R., Augustson, E., & Fillingim, R. (2003). Differential relationships between
anxiety and treatment-associated pain reduction among male and female chronic pain
patients. Clin J Pain, 19(4), 208-216.
Edwards, R. R., Haythornthwaite, J. A., Sullivan, M. J., & Fillingim, R. B. (2004).
Catastrophizing as a mediator of sex differences in pain: differential effects for daily pain
versus laboratory-induced pain. PAIN, 111(3), 335-341.
Elvery, N., Jensen, M. P., Ehde, D. M., & Day, M. A. (2017). Pain Catastrophizing, Mindfulness,
and Pain Acceptance: What's the Difference? Clin J Pain, 33(6), 485-495. doi:
10.1097/ajp.0000000000000430
Fillingim, R. B., King, C. D., Ribeiro-Dasilva, M. C., Rahim-Williams, B., & Riley, J. L. (2009).
Sex, gender, and pain: a review of recent clinical and experimental findings. The Journal
of Pain, 10(5), 447-485. doi: 10.1016/j.jpain.2008.12.001
Finley, E. P., Garcia, A., Rosen, K., McGeary, D., Pugh, M. J., & Potter, J. S. (2017). Evaluating
the impact of prescription drug monitoring program implementation: a scoping review.
BMC Health Services Research, 17, 420. doi: 10.1186/s12913-017-2354-5
Franklin, G. M., Mai, J., Wickizer, T., Turner, J. A., Fulton ‐Kehoe, D., & Grant, L. (2005).
Opioid dosing trends and mortality in Washington State workers' compensation, 1996 –
2002. American journal of industrial medicine, 48(2), 91-99.
Frieden, T. R., & Houry, D. (2016). Reducing the risks of relief—the CDC opioid-prescribing
guideline. New England Journal of Medicine, 374(16), 1501-1504.
Furlan, A. D., Yazdi, F., Tsertsvadze, A., Gross, A., Van Tulder, M., Santaguida, L., . . .
Galipeau, J. (2010). Complementary and alternative therapies for back pain II. Evid Rep
Technol Assess (Full Rep)(194), 1-764.
Gaskin, D. J., & Richard, P. (2012). The economic costs of pain in the United States. J Pain,
13(8), 715-724. doi: 10.1016/j.jpain.2012.03.009
Gatchel, R. J., Peng, Y. B., Peters, M. L., Fuchs, P. N., & Turk, D. C. (2007). The
biopsychosocial approach to chronic pain: scientific advances and future directions.
Psychol Bull, 133(4), 581-624. doi: 10.1037/0033-2909.133.4.581
Gottschall, A. C., West, S. G., & Enders, C. K. (2012). A comparison of item-level and scale-
level multiple imputation for questionnaire batteries. Multivariate Behavioral Research,
47(1), 1-25.
Gould, D., Kelly, D., Goldstone, L., & Gammon, J. (2001). Examining the validity of pressure
ulcer risk assessment scales: developing and using illustrated patient simulations to
collect the data INFORMATION POINT: Visual Analogue Scale. Journal of Clinical
Nursing, 10(5), 697-706. doi: 10.1046/j.1365-2702.2001.00525.x
Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual review of
psychology, 60, 549-576.
Greco, C. M., Yu, L., Johnston, K. L., Dodds, N. E., Morone, N. E., Glick, R. M., . . . Pilkonis,
P. A. (2016). Measuring nonspecific factors in treatment: item banks that assess the
healthcare experience and attitudes from the patient's perspective. Qual Life Res, 25(7),
1625-1634. doi: 10.1007/s11136-015-1178-1
Grossman, P. (2015). Mindfulness: awareness informed by an embodied ethic. Mindfulness, 6(1),
17-22.
Gurol-Urganci, I., de Jongh, T., Vodopivec-Jamsek, V., Atun, R., & Car, J. (2013). Mobile
phone messaging reminders for attendance at healthcare appointments. Cochrane
Database of Systematic Reviews(12). doi: 10.1002/14651858.CD007458.pub3
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 120

Guy, J. G., Zhang, K., Bohm, M. K., Losby, J., Lewis, B., Young, R., . . . Dowell, D. (2017).
Vital Signs: Changes in Opioid Prescribing in the United States, 2006-2015. MMWR.
Morbidity and mortality weekly report, 66(26), 697-704.
Haegerich, T. M., Paulozzi, L. J., Manns, B. J., & Jones, C. M. (2014). What we know, and don’t
know, about the impact of state policy and systems-level interventions on prescription
drug overdose. Drug & Alcohol Dependence, 145, 34-47.
Haffajee, R. L., Jena, A. B., & Weiner, S. G. (2015). Mandatory Use of Prescription Drug
Monitoring Programs. JAMA, 313(9), 891-892. doi: 10.1001/jama.2014.18514
Henry, J. D., & Crawford, J. R. (2005). The short-form version of the Depression Anxiety Stress
Scales (DASS-21): construct validity and normative data in a large non-clinical sample.
Br J Clin Psychol, 44(Pt 2), 227-239. doi: 10.1348/014466505x29657
Hill Jr, C. S. (1996). Government regulatory influences on opioid prescribing and their impact on
the treatment of pain of nonmalignant origin. Journal of pain and symptom management,
11(5), 287-298.
Hodge Jr, J. G., Wetter, S. A., Chronister, D., Hess, A., & Piatt, J. (2017). Redefining Public
Health Emergencies: The Opioid Epidemic. Jurimetrics, 58, 1, 3.
Hsu, K. Y., Dunn, J. E., Bradshaw, Y. S., & Conboy, L. (2014). Lessons from an acupuncture
teaching clinic: patient characteristics and pain management effectiveness. Explore (NY),
10(5), 284-293. doi: 10.1016/j.explore.2014.06.004
Institute of Medicine. (2011). Relieving Pain in America: A Blueprint for Transforming
Prevention, Care, Education, and Research. Washington, DC: The National Academies
Press.
Interagency Pain Research Coordinating Committee. (2015). National Pain Strategy: a
comprehensive population health-level strategy for pain. Washington, DC: Department of
Health and Human Services.
Interagency Pain Research Coordinating Committee. (2017). Federal Pain Research Strategy.
Washington, DC: Department of Health and Human Services.
Janz, N. K., & Becker, M. H. (1984). The Health Belief Model: a decade later. Health Educ Q,
11(1), 1-47. doi: 10.1177/109019818401100101
Jones, K., Wright, C., & Bell, A. (2013). Do Multilevel Models Ever Give Different Results? Or
Why Bother With Multilevel Models.
Kannisto, K. A., Koivunen, M. H., & Välimäki, M. A. (2014). Use of Mobile Phone Text
Message Reminders in Health Care Services: A Narrative Literature Review. Journal of
Medical Internet Research, 16(10), e222. doi: 10.2196/jmir.3442
Kaptchuk, T. J. (2002). Acupuncture: theory, efficacy, and practice. Ann Intern Med, 136(5),
374-383.
Keefe, F. J., Rumble, M. E., Scipio, C. D., Giordano, L. A., & Perri, L. M. (2004). Psychological
aspects of persistent pain: current state of the science. J Pain, 5(4), 195-211. doi:
10.1016/j.jpain.2004.02.576
Komandur, B., Martin, P. R., & Bandarian-Balooch, S. (2018). Mindfulness and Chronic
Headache/Migraine: Mechanisms Explored Through the Fear-Avoidance Model of
Chronic Pain. Clin J Pain, 34(7), 638-649. doi: 10.1097/ajp.0000000000000580
Lam, C. N., Murnane, D., Ruth, C., Kim, Y., Chou, C. P., & Black, D. S. (2019). Demographic
and Clinical Factors Associated with Patient Adherence to Acupuncture Treatment. J
Altern Complement Med. doi: 10.1089/acm.2018.0267
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 121

Lautenbacher, S., Peters, J. H., Heesen, M., Scheel, J., & Kunz, M. (2017). Age changes in pain
perception: A systematic-review and meta-analysis of age effects on pain and tolerance
thresholds. Neurosci Biobehav Rev, 75, 104-113. doi: 10.1016/j.neubiorev.2017.01.039
Lee, A. C., Harvey, W. F., Price, L. L., Morgan, L. P. K., Morgan, N. L., & Wang, C. (2017).
Mindfulness is associated with psychological health and moderates pain in knee
osteoarthritis. Osteoarthritis Cartilage, 25(6), 824-831. doi: 10.1016/j.joca.2016.06.017
Lee, M. S., & Ernst, E. (2011). Acupuncture for pain: an overview of Cochrane reviews. Chin J
Integr Med, 17(3), 187-189. doi: 10.1007/s11655-011-0665-7
Leung, P. T. M., Macdonald, E. M., Stanbrook, M. B., Dhalla, I. A., & Juurlink, D. N. (2017). A
1980 Letter on the Risk of Opioid Addiction. N Engl J Med, 376(22), 2194-2195. doi:
10.1056/NEJMc1700150
Levy, N., Sturgess, J., & Mills, P. (2018). “Pain as the fifth vital sign” and dependence on the
“numerical pain scale” is being abandoned in the US: Why? British journal of
anaesthesia, 120(3), 435-438.
Li, C. (2013). Little's test of missing completely at random. The Stata Journal, 13(4), 795-809.
Li, C., Yang, J., Park, K., Wu, H., Hu, S., Zhang, W., . . . Zhang, X. (2014). Prolonged Repeated
Acupuncture Stimulation Induces Habituation Effects in Pain-Related Brain Areas: An
fMRI Study. PLoS ONE, 9(5), e97502. doi: 10.1371/journal.pone.0097502
Lin, C. S., Wu, S. Y., & Yi, C. A. (2017). Association between Anxiety and Pain in Dental
Treatment: A Systematic Review and Meta-analysis. J Dent Res, 96(2), 153-162. doi:
10.1177/0022034516678168
Lin, J. G., & Chen, W. L. (2008). Acupuncture analgesia: a review of its mechanisms of actions.
Am J Chin Med, 36(4), 635-645. doi: 10.1142/s0192415x08006107
Lin, Y., Wan, L., & Jamison, R. N. (2017). Using Integrative Medicine in Pain Management: An
Evaluation of Current Evidence. Anesthesia & Analgesia, 125(6), 2081-2093. doi:
10.1213/ane.0000000000002579
Linde, K., Witt, C. M., Streng, A., Weidenhammer, W., Wagenpfeil, S., Brinkhaus, B., . . .
Melchart, D. (2007). The impact of patient expectations on outcomes in four randomized
controlled trials of acupuncture in patients with chronic pain. PAIN, 128(3), 264-271. doi:
10.1016/j.pain.2006.12.006
Liu, L., Skinner, M., McDonough, S., Mabire, L., & Baxter, G. D. (2015). Acupuncture for low
back pain: an overview of systematic reviews. Evid Based Complement Alternat Med,
2015, 328196. doi: 10.1155/2015/328196
Loeser, J. D., & Melzack, R. (1999). Pain: an overview. The Lancet, 353(9164), 1607-1609. doi:
10.1016/S0140-6736(99)01311-2
Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the depression anxiety stress scales:
Psychology Foundation of Australia.
Ludwig, D. S., & Kabat-Zinn, J. (2008). Mindfulness in medicine. JAMA, 300(11), 1350-1352.
doi: 10.1001/jama.300.11.1350
Lyapustina, T., Rutkow, L., Chang, H. Y., Daubresse, M., Ramji, A. F., Faul, M., . . . Alexander,
G. C. (2016). Effect of a “pill mill” law on opioid prescribing and utilization: the case of
Texas. Drug and alcohol dependence, 159, 190-197.
MacPherson, H., Maschino, A. C., Lewith, G., Foster, N. E., Witt, C. M., & Vickers, A. J.
(2013). Characteristics of acupuncture treatment associated with outcome: an individual
patient meta-analysis of 17,922 patients with chronic pain in randomised controlled trials.
PLoS ONE, 8(10), e77438. doi: 10.1371/journal.pone.0077438
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 122

MacPherson, H., Vertosick, E. A., Foster, N. E., Lewith, G., Linde, K., Sherman, K. J., . . .
Vickers, A. J. (2017). The persistence of the effects of acupuncture after a course of
treatment: a meta-analysis of patients with chronic pain. PAIN, 158(5), 784-793. doi:
10.1097/j.pain.0000000000000747
Mallatt, J. (2017). The Effect of Prescription Drug Monitoring Programs on Opioid Prescriptions
and Heroin Crime Rates.
Manchikanti, L., Sanapati, J., Benyamin, R. M., Atluri, S., Kaye, A. D., & Hirsch, J. A. (2018).
Reframing the prevention strategies of the opioid crisis: focusing on prescription opioids,
fentanyl, and heroin epidemic. Pain physician, 21(4), 309-326.
Mao, J. J., Xie, S. X., & Bowman, M. A. (2010). Uncovering the expectancy effect: the
validation of the acupuncture expectancy scale. Altern Ther Health Med, 16(6), 22-27.
Markel, H. (2011). An anatomy of addiction: Sigmund Freud, William Halsted, and the miracle
drug cocaine: Vintage.
Marx, B. L., Rubin, L. H., Milley, R., Hammerschlag, R., & Ackerman, D. L. (2013). A
prospective patient-centered data collection program at an acupuncture and oriental
medicine teaching clinic. J Altern Complement Med, 19(5), 410-415. doi:
10.1089/acm.2011.0774
McCracken, L. M., Gauntlett-Gilbert, J., & Vowles, K. E. (2007). The role of mindfulness in a
contextual cognitive-behavioral analysis of chronic pain-related suffering and disability.
PAIN, 131(1-2), 63-69. doi: 10.1016/j.pain.2006.12.013
Meara, E., Horwitz, J. R., Powell, W., McClelland, L., Zhou, W., O'malley, A. J., & Morden, N.
E. (2016). State legal restrictions and prescription-opioid use among disabled adults. New
England Journal of Medicine, 375(1), 44-53.
Meghani, S. H., & Liou, K. T. (2019). Re: Demographic and Clinical Factors Associated with
Patient Adherence to Acupuncture Treatment by Lam, et al. (J Altern Complement Med
2019;25:522-525). J Altern Complement Med, 25(6), 659-660. doi:
10.1089/acm.2019.0049
Molsberger, A. F., Zhou, J., Arndt, D., & Teske, W. (2008). Chinese acupuncture for chronic low
back pain: an international expert survey. J Altern Complement Med, 14(9), 1089-1095.
doi: 10.1089/acm.2008.0138.A
Moroz, A., Spivack, S., & Lee, M. H. M. (2004). Adherence to acupuncture treatment for
chronic pain. Journal of Alternative & Complementary Medicine, 10(5), 739-740.
Moyo, P., Simoni ‐Wastila, L., Griffin, B. A., Onukwugha, E., Harrington, D., Alexander, G.
C., & Palumbo, F. (2017). Impact of prescription drug monitoring programs (PDMPs) on
opioid utilization among Medicare beneficiaries in 10 US states. Addiction.
Multiple imputation in Stata. UCLA: Statistical Consulting Group. Retrieved June 8, 2018,
from https://stats.idre.ucla.edu/stata/seminars/mi_in_stata_pt1_new/
Nahin, R. L. (2015). Estimates of pain prevalence and severity in adults: United States, 2012. J
Pain, 16(8), 769-780. doi: 10.1016/j.jpain.2015.05.002
Nahin, R. L., Barnes, P. M., & Stussman, B. J. (2016a). Expenditures on Complementary Health
Approaches: United States, 2012. Natl Health Stat Report(95), 1-11.
Nahin, R. L., Barnes, P. M., & Stussman, B. J. (2016b). Insurance Coverage for Complementary
Health Approaches Among Adult Users: United States, 2002 and 2012. NCHS Data
Brief(235), 1-8.
Nahin, R. L., Boineau, R., Khalsa, P. S., Stussman, B. J., & Weber, W. J. (2016). Evidence-
Based Evaluation of Complementary Health Approaches for Pain Management in the
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 123

United States. Mayo Clinic Proceedings, 91(9), 1292-1306. doi:
10.1016/j.mayocp.2016.06.007
National Alliance for Model State Drug Laws. (2014). States that require prescribers and/or
dispensers to access PMP database in certain circumstances.
National Alliance for Model State Drug Laws. (2017). MANDATED USE OF STATE
PRESCRIPTION DRUG MONITORING PROGRAMS (PMPS) - HIGHLIGHTS OF
KEY STATE REQUIREMENTS.
National Institutes of Health. (1998). NIH Consensus Conference. Acupuncture. JAMA, 280(17),
1518-1524.
O'Laughlin, K. D., Martin, M. J., & Ferrer, E. (2018). Cross-Sectional Analysis of Longitudinal
Mediation Processes. Multivariate Behav Res, 53(3), 375-402. doi:
10.1080/00273171.2018.1454822
Ocanez, K. L., McHugh, R. K., & Otto, M. W. (2010). A meta-analytic review of the association
between anxiety sensitivity and pain. Depress Anxiety, 27(8), 760-767. doi:
10.1002/da.20681
Ohnhaus, E. E., & Adler, R. (1975). Methodological problems in the measurement of pain: a
comparison between the verbal rating scale and the visual analogue scale. PAIN, 1(4),
379-384.
Pardo, B. (2017). Do more robust prescription drug monitoring programs reduce prescription
opioid overdose? Addiction, 112(10), 1773-1783.
Parikh, A., Gupta, K., Wilson, A. C., Fields, K., Cosgrove, N. M., & Kostis, J. B. (2010). The
effectiveness of outpatient appointment reminder systems in reducing no-show rates. Am
J Med, 123(6), 542-548. doi: 10.1016/j.amjmed.2009.11.022
Paulozzi, L. J., Kilbourne, E. M., & Desai, H. A. (2011). Prescription drug monitoring programs
and death rates from drug overdose. Pain medicine, 12(5), 747-754.
Pezalla, E. J., Rosen, D., Erensen, J. G., Haddox, J. D., & Mayne, T. J. (2017). Secular trends in
opioid prescribing in the USA. Journal of pain research, 10, 383.
Picavet, H. S. J., & Hazes, J. M. W. (2003). Prevalence of self reported musculoskeletal diseases
is high. Annals of the rheumatic diseases, 62(7), 644-650. doi: 10.1136/ard.62.7.644
Piper, B. J., Shah, D. T., Simoyan, O. M., McCall, K. L., & Nichols, S. D. (2018). Trends in
medical use of opioids in the US, 2006–2016. American journal of preventive medicine,
54(5), 652-660.
Plumpton, C. O., Morris, T., Hughes, D. A., & White, I. R. (2016). Multiple imputation of
multiple multi-item scales when a full imputation model is infeasible. BMC research
notes, 9(1), 45.
Porter, J., & Jick, H. (1980). Addiction rare in patients treated with narcotics. The New England
journal of medicine, 302(2), 123.
Poulin, P. A., Romanow, H. C., Rahbari, N., Small, R., Smyth, C. E., Hatchard, T., . . . Wilson,
K. G. (2016). The relationship between mindfulness, pain intensity, pain catastrophizing,
depression, and quality of life among cancer survivors living with chronic neuropathic
pain. Support Care Cancer, 24(10), 4167-4175. doi: 10.1007/s00520-016-3243-x
Prady, S. L., Burch, J., Vanderbloemen, L., Crouch, S., & MacPherson, H. (2015). Measuring
expectations of benefit from treatment in acupuncture trials: a systematic review.
Complement Ther Med, 23(2), 185-199. doi: 10.1016/j.ctim.2015.01.007
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 124

Prakash, R. S., Hussain, M. A., & Schirda, B. (2015). The role of emotion regulation and
cognitive control in the association between mindfulness disposition and stress. Psychol
Aging, 30(1), 160-171. doi: 10.1037/a0038544
Qaseem, A., Wilt, T. J., McLean, R. M., Forciea, M., & for the Clinical Guidelines Committee of
the American College of, P. (2017). Noninvasive treatments for acute, subacute, and
chronic low back pain: A clinical practice guideline from the american college of
physicians. Ann Intern Med, 166(7), 514-530. doi: 10.7326/M16-2367
Quartana, P. J., Campbell, C. M., & Edwards, R. R. (2009). Pain catastrophizing: a critical
review. Expert review of neurotherapeutics, 9(5), 745-758. doi: 10.1586/ern.09.34
Quinn, T. M., & McLaughlin, G. T. (1972). The evolution of Federal drug control legislation.
Cath. UL Rev., 22, 586.
Rajguru, P., Kolber, M. J., Garcia, A. N., Smith, M. T., Patel, C. K., & Hanney, W. J. (2014).
Use of Mindfulness Meditation in the Management of Chronic Pain A Systematic Review
of Randomized Controlled Trials. American Journal of Lifestyle Medicine,
1559827614522580.
Reisman, R. M., Shenoy, P. J., Atherly, A. J., & Flowers, C. R. (2009). Prescription opioid usage
and abuse relationships: an evaluation of state prescription drug monitoring program
efficacy. Substance abuse: research and treatment, 3, SART. S2345.
Ridner, S. H. (2004). Psychological distress: concept analysis. J Adv Nurs, 45(5), 536-545.
Riley III, J. L., Cruz-Almeida, Y., Glover, T. L., King, C. D., Goodin, B. R., Sibille, K. T., . . .
Fessler, B. J. (2014). Age and race effects on pain sensitivity and modulation among
middle-aged and older adults. The Journal of Pain, 15(3), 272-282.
Rosenblum, A., Marsch, L. A., Joseph, H., & Portenoy, R. K. (2008). Opioids and the treatment
of chronic pain: controversies, current status, and future directions. Experimental and
clinical psychopharmacology, 16(5), 405-416. doi: 10.1037/a0013628
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011). Mediation analysis in
social psychology: Current practices and new recommendations. Social and Personality
Psychology Compass, 5(6), 359-371.
Rudd, R. A., Aleshire, N., Zibbell, J. E., & Matthew Gladden, R. (2016). Increases in drug and
opioid overdose deaths—United States, 2000–2014. American Journal of
Transplantation, 16(4), 1323-1327.
Rugg, S., Paterson, C., Britten, N., Bridges, J., & Griffiths, P. (2011). Traditional acupuncture
for people with medically unexplained symptoms: a longitudinal qualitative study of
patients’ experiences. The British Journal of General Practice, 61(587), e306-e315. doi:
10.3399/bjgp11X577972
Rutkow, L., Chang, H., Daubresse, M., Webster, D. W., Stuart, E. A., & Alexander, G. C.
(2015). Effect of Florida’s prescription drug monitoring program and pill mill laws on
opioid prescribing and use. JAMA internal medicine, 175(10), 1642-1649.
SAS. (2014). Version 9.4. Cary, NC: SAS Institute Inc.
Schlomer, G. L., Bauman, S., & Card, N. A. (2010). Best practices for missing data management
in counseling psychology. Journal of Counseling psychology, 57(1), 1.
Scholl, L., Seth, P., Kariisa, M., Wilson, N., & Baldwin, G. (2019). Drug and opioid-involved
overdose deaths—United States, 2013–2017. Morbidity and Mortality Weekly Report,
67(5152), 1419.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 125

Schutze, R., Rees, C., Preece, M., & Schutze, M. (2010). Low mindfulness predicts pain
catastrophizing in a fear-avoidance model of chronic pain. PAIN, 148(1), 120-127. doi:
10.1016/j.pain.2009.10.030
Schwebel, F. J., & Larimer, M. E. (2018). Using text message reminders in health care services:
A narrative literature review. Internet Interventions, 13, 82-104. doi:
10.1016/j.invent.2018.06.002
Seth, P., Rudd, R. A., Noonan, R. K., & Haegerich, T. M. (2018). Quantifying the Epidemic of
Prescription Opioid Overdose Deaths. Am J Public Health, 108(4), 500-502. doi:
10.2105/AJPH.2017.304265
Severtson, S. G., Bartelson, B. B., Davis, J. M., Muñoz, A., Schneider, M. F., Chilcoat, H., . . .
Dart, R. C. (2013). Reduced abuse, therapeutic errors, and diversion following
reformulation of extended-release oxycodone in 2010. The Journal of Pain, 14(10), 1122-
1130.
Shin, J. S., Ha, I. H., Lee, J., Choi, Y., Kim, M. R., Park, B. Y., . . . Lee, M. S. (2013). Effects of
motion style acupuncture treatment in acute low back pain patients with severe disability:
a multicenter, randomized, controlled, comparative effectiveness trial. PAIN, 154(7),
1030-1037. doi: 10.1016/j.pain.2013.03.013
StataCorp. (2013). Stata Statistical Software: Release 13. College Station, Tx: StataCorp LP.
StataCorp. (2017). Stata Statistical Software: Release 15. College Station, Tx: StataCorp LLC.
Substance Abuse and Mental Health Services Administration. (2018). Using prescription drug
monitoring program data to support prevention planning. SAMHSA's Center for the
Application of Prevention Technologies: SAMHSA.
Tait, R. C., Chibnall, J. T., & Krause, S. (1990). The Pain Disability Index: psychometric
properties. PAIN, 40(2), 171-182.
Thakkar, J., Kurup, R., Laba, T., Santo, K., Thiagalingam, A., Rodgers, A., . . . Chow, C. K.
(2016). Mobile telephone text messaging for medication adherence in chronic disease: A
meta-analysis. JAMA internal medicine, 176(3), 340-349. doi:
10.1001/jamainternmed.2015.7667
The White House. (2018). How we will win the war on opioids. Retrieved June 8, 2018, from
https://www.whitehouse.gov/articles/will-win-war-opioids/
Theunissen, M., Peters, M. L., Bruce, J., Gramke, H. F., & Marcus, M. A. (2012). Preoperative
anxiety and catastrophizing: a systematic review and meta-analysis of the association
with chronic postsurgical pain. Clin J Pain, 28(9), 819-841. doi:
10.1097/AJP.0b013e31824549d6
Tice, J. A., Kumar, V., Webb, M., Seidner, M., Rind, D., Chapman, R., . . . Pearson, S. D.
(2017). Cognitive and Mind-Body Therapies for Chronic Low Back and Neck Pain:
Effectiveness and Value: Institute for Clinical and Economic Review.
Tick, H., Nielsen, A., Pelletier, K., Bonakdar, R., Simmons, S., Glick, R., . . . Zador, V. (2017).
Evidence-based Nonpharmacologic Strategies for Comprehensive Pain Care. A
Consortium Pain Task Force White Paper The Pain Task Force of the Academic
Consortium for Integrative Medicine and Health.
Ullrich, P. M., Turner, J. A., Ciol, M., & Berger, R. (2005). Stress is associated with subsequent
pain and disability among men with nonbacterial prostatitis/pelvic pain. Ann Behav Med,
30(2), 112-118. doi: 10.1207/s15324796abm3002_3
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 126

van Hecke, O., Torrance, N., & Smith, B. H. (2013). Chronic pain epidemiology – where do
lifestyle factors fit in? British Journal of Pain, 7(4), 209-217. doi:
10.1177/2049463713493264
Van Zee, A. (2009). The promotion and marketing of oxycontin: commercial triumph, public
health tragedy. American journal of public health, 99(2), 221-227. doi:
10.2105/AJPH.2007.131714
VanderWeele, T. J., & Vansteelandt, S. (2014). Mediation Analysis with Multiple Mediators.
Epidemiologic methods, 2(1), 95-115. doi: 10.1515/em-2012-0010
Veehof, M. M., Trompetter, H. R., Bohlmeijer, E. T., & Schreurs, K. M. (2016). Acceptance-
and mindfulness-based interventions for the treatment of chronic pain: a meta-analytic
review. Cogn Behav Ther, 45(1), 5-31. doi: 10.1080/16506073.2015.1098724
Vincent, H. K., Horodyski, M., Vincent, K. R., Brisbane, S. T., & Sadasivan, K. K. (2015).
Psychological Distress After Orthopedic Trauma: Prevalence in Patients and Implications
for Rehabilitation. Pm r, 7(9), 978-989. doi: 10.1016/j.pmrj.2015.03.007
Wang, H., Yang, G., Wang, S., Zheng, X., Zhang, W., & Li, Y. (2018). The Most Commonly
Treated Acupuncture Indications in the United States: A Cross-Sectional Study. Am J
Chin Med, 1-33. doi: 10.1142/s0192415x18500738
Warner, M., Chen, L. H., Makuc, D. M., Anderson, R. N., & Minino, A. M. (2011). Drug
poisoning deaths in the United States, 1980-2008. NCHS Data Brief(81), 1-8.
Webster, P. C. (2012). Oxycodone class action lawsuit filed: Can Med Assoc.
Wen, H., Schackman, B. R., Aden, B., & Bao, Y. (2017). States With Prescription Drug
Monitoring Mandates Saw A Reduction In Opioids Prescribed To Medicaid Enrollees.
Health Aff (Millwood), 36(4), 733-741. doi: 10.1377/hlthaff.2016.1141
Williams, A. C. d. C., & Craig, K. D. (2016). Updating the definition of pain. PAIN, 157(11),
2420-2423. doi: 10.1097/j.pain.0000000000000613
Wong, D. L., & Baker, C. M. (1988). Pain in children: comparison of assessment scales. Pediatr
Nurs, 14(1), 9-17.
Woolf, A. D., & Pfleger, B. (2003). Burden of major musculoskeletal conditions. Bulletin of the
World Health Organization, 81(9), 646-656.
World Bank. (2019). GDP (constant 2010 US$). World Development Indicators. Retrieved
June, 8, 2019, from The World Bank Group, 2019,
https://data.worldbank.org/indicator/NY.GDP.MKTP.KD?end=2010&locations=US&star
t=1960
Xiang, A., Cheng, K., Shen, X., Xu, P., & Liu, S. (2017). The Immediate Analgesic Effect of
Acupuncture for Pain: A Systematic Review and Meta-Analysis. Evidence-Based
Complementary and Alternative Medicine, 2017, 13. doi: 10.1155/2017/3837194
Yelin, E., Weinstein, S., & King, T. (2016). The burden of musculoskeletal diseases in the
United States. Semin Arthritis Rheum, 46(3), 259-260. doi:
10.1016/j.semarthrit.2016.07.013
Yuan, J., Kerr, D., Park, J., Liu, X. H., & McDonough, S. (2008). Treatment regimens of
acupuncture for low back pain--a systematic review. Complement Ther Med, 16(5), 295-
304.
Yuan, J., Purepong, N., Kerr, D. P., Park, J., Bradbury, I., & McDonough, S. (2008).
Effectiveness of acupuncture for low back pain: a systematic review. Spine (Phila Pa
1976), 33(23), E887-900. doi: 10.1097/BRS.0b013e318186b276
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 127

Zhang, Y., Lao, L., Chen, H., & Ceballos, R. (2012). Acupuncture Use among American Adults:
What Acupuncture Practitioners Can Learn from National Health Interview Survey 2007?
Evidence-based Complementary and Alternative Medicine : eCAM, 2012, 710750. doi:
10.1155/2012/710750

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 128

Appendix A. Comparison of MMEPC growth profiles, full dataset (1997-2014) and early
period (1997-2009)
1997-2014 1997-2009
Intercept 1 107.8 (6.7) 100.4 (5.4)
Intercept 2 727.5 (26.5) 629.4 (28.8)
Slope Period 1 54.2 (2.7)*** 55.8 (2.6)***
Slope Period 2 9.9 (3.2)** 46.2 (3.9)***

Slope Period 2 - Period 1 -44.3 (3.7)*** -9.6 (4.0)*

N, States 807 (45 states) 585 (21 states)
Chi-Square 1475.6*** 1158.04***
Beta (S.E.) shown at each cell. *p<.05, **p<.01, ***p<.001
PDMP: Prescription Drug Monitoring Program, MMEPC:
Morphine Milligram Equivalent Per Capita
Pre-PDMP: Intercept 1 and Slope Period 1; Post-PDMP:
Intercept 2 and Slope Period 2
Pill Mill law not controlled given only 1 state had the law
implemented during the early period (Louisiana in 2005)
Chi-Square represents the null model likelihood ratio test

Note. Appendix A includes a subset data model with 21 states that state PDMP implemented in
between 1997 and 2009 to show how the timing of PDMP implementation might affect the effect
of PDMP. The table also includes the full data model with 45 states during 1997-2014, the same
model as Model 1 in Table 2-2 as contrast. The 1997-2009 subset model shows that the growth
of MMEPC reduced after the implementation of PDMP (P2: 46.2 vs P1: 55.8, difference: -9.6);
however, the slope difference in the subset model is much smaller than the full model (-44.3).
This result suggests that there is a timing effect of PDMP implementation, which support our
approach to control for PDMP implementation timing in our analysis.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 129

Appendix B. MMEPC growth profiles in the pre-PDMP period by the three-level strength
of PDMP mandates, 1997-2014
Beta (S.E)
Intercept 1 108.3 (10.6)
Intercept 2 727.6 (27.9)
Slope Period 1 52.7 (3.0)***
Slope Period 2 10.2 (3.2)**

Strength of PDMP Mandates

Intermediate vs Weak -1.4 (15.0)
Strong vs Weak -0.5 (1.0)

Intermediate x Period 1 2.9 (2.6)
Strong x Period 1 1.6 (2.8)

N, States 807 (45 states)
Chi-Square 1420.7***
*p<.05, **p<.01, ***p<.001
PDMP: Prescription Drug Monitoring Program
MMEPC: Morphine Milligram Equivalent Per Capita
Chi-Square represents the null model likelihood ratio test
Pre-PDMP: Intercept 1 and Slope Period 1; Post-PDMP:
Intercept 2 and Slope Period 2

Note. The analysis examines whether states implemented different strength of PDMP mandate
might have a different trajectory of opioid growth prior to implementing the state PDMP. This
analysis is tested by taking the interaction between two variables representing the three-level
PDMP strength and the pre-PDMP period (Period 1). The interactions show that states with a
strong or intermediate PDMP had higher MMEPC growths compared to states with a weak
PDMP prior to implementation (2.9 and 1.6, respectively); however the interactions were not
significant.
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 130

Appendix C. Acupuncture Study: Baseline Questionnaire
DEMOGRAPHICS
1. Your sex:
a. Male
b. Female

2. Your age: _____________

3. Years of completed education: (Graduating high school means 12 years of education):
__________

4. Your Race/Ethnicity:
Mark one of the following that most accurately describes your race/ethnic background.
a. Hispanic/Latino
b. American Indian/Alaska Native
c. Asian
d. Black/African American
e. Native Hawaiian/Pacific Islander
f. White
g. Other, describe _____________

5. Your current marital status:
a. Never been married
b. Married
c. Separated
d. Widowed
e. Divorced
f. Other, describe _____________

6. Are you presently employed?
a. Yes
b. No

7. In which of the categories did your total family income fall last year? This includes income from
work plus other sources such as interest, social security, etc.
a. $0 - $34,999
b. $35,000 - $49,999
c. $50,000 - $74,999
d. $75,000 - $99,999
e. $100,000 and over

8. Do you have any of the following healthcare coverage?
a. Private insurance
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 131

b. Medicare
c. Medi-Cal
d. Other government sponsored programs
e. I do not have healthcare coverage

9. Attitude toward CAM
INSTRUCTIONS: For each of the following statement, circle the one best response.
CAM (Complementary and Alternative
Medicine) is a non-conventional,
holistic, or natural approach to health
care. Common CAM treatment may
include acupuncture, massage therapy,
meditation, or herbal remedies.
Not
at all
A little
bit
Some-
what
Quite
a bit
Very
much
CAM is effective 0 1 2 3 4
I prefer CAM over conventional medicine 0 1 2 3 4
It is important to be open to CAM 0 1 2 3 4
CAM can be used to treat serious illness 0 1 2 3 4
CAM can prevent health problems 0 1 2 3 4
I prefer natural remedies 0 1 2 3 4

10. Is today your first time using acupuncture?
a. Yes (go to item 10 and 11)
b. No (skip to item 12)

11. How would you rate your overall past experience with acupuncture?
0 1 2 3 4 5 6 7 8 9 10
Not satisfied at all………………………………..………………….……………Completely satisfied

12. On average, how many acupuncture follow-up visit(s) did you have for your previous health
issues when you used acupuncture? Follow-up visit refers to acupuncture visit(s) made as a
follow-up to an initial visit to treat the same health concern.
a. No follow-up visit (only initial treatment)
b. 1-2 follow-up visits
c. 2-4 follow-up visits
d. 4-8 follow-up visits
e. More than 8 follow-up visits

13. How effective do you consider acupuncture in general?
a. Very effective
b. Effective
c. Slightly effective
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 132

d. Not effective
e. Don’t know

14. What do you expect personally from the acupuncture treatment you have received?
a. Cure
b. Clear improvement
c. Slight improvement
d. No improvement
e. Don’t know

15. Treatment Expectancy
INSTRUCTIONS: For each of the following statement, circle the one best response.
Please think of the acupuncture
treatment you just received.
Not
at all
A little
bit
Some-
what
Quite
a bit
Very
much
I am confident in this treatment 0 1 2 3 4
This treatment will be successful 0 1 2 3 4
I feel good about this treatment 0 1 2 3 4
I expect good outcomes from this
treatment
0 1 2 3 4
This treatment is right for me 0 1 2 3 4
I value this treatment 0 1 2 3 4

16. Acupuncture Expectancy Scale
INSTRUCTIONS: For each of the following statement, circle the one best response.
Every individual may have different
expectation for the effects of acupuncture. If
we use the following sentences to describe
your expectation of acupuncture’s effect on
your illness/symptom after the entire course
of acupuncture therapy, how much do you
agree? For each statement, please choose
the closest answer.
Not
at all
agree
A little
agree
Moder
ately
agree
Mostl
y
agree
Compl
etely
agree
My illness will improves a lot 1 2 3 4 5
I will be able to cope with my illness 1 2 3 4 5
The symptoms of my illness will disappear 1 2 3 4 5
My energy level will increase 1 2 3 4 5

17. What is your main reason for today’s acupuncture visit?
a. A health problem or concern
b. For prevention purposes, health maintenance
c. Just curious about acupuncture, no real health issue

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 133

18. What is your PRIMARY health concern for today’s acupuncture visit? (select one apply)
a. Back pain or problem
b. Neck pain or problem
c. Joint pain or stiffness, or other joint condition
d. Other musculoskeletal pain or problem
e. Arthritis
f. Fibromyalgia
g. Sprain or strain
h. Severe headache or migraine
i. Regular headaches
j. Insomnia or trouble sleeping
k. Stress
l. Depression
m. Anxiety
n. Head or chest cold
o. Stomach or intestinal illness
p. Hypertension
q. Coronary heart disease
r. Cholesterol
s. Diabetes
t. Other, please describe: ______________________________
Acute vs. Chronic condition
19. Is your primary health concern:
a. A chronic health problem (persisting more than three months)
b. An acute health problem (less than three months)

20. Brief Illness Perception Questionnaire
INSTRUCTIONS: For the following questions, please circle the number that best corresponds
to your views:
How much does your illness affect your life?
0 1 2 3 4 5 6 7 8 9 10
No affect at all………………….……………………………………Severely affects my life

How long do you think your illness will continue?
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 134

0 1 2 3 4 5 6 7 8 9 10
A very short time……………………………………………………………………..Forever

How much control do you feel you have over your illness?
0 1 2 3 4 5 6 7 8 9 10
Absolutely no control……………………………………………Extreme amount of control

How much do you think your treatment can help your illness?
0 1 2 3 4 5 6 7 8 9 10
Not at all…………………….…………………………………………..…Extremely helpful

How much do you experience symptoms from your illness?
0 1 2 3 4 5 6 7 8 9 10
No symptom at all……………………………………………………Many severe symptoms

How concerned are you about your illness?
0 1 2 3 4 5 6 7 8 9 10
Not at all concerned……………………………………………………Extremely concerned

How well do you feel you understand your illness?
0 1 2 3 4 5 6 7 8 9 10
Don’t understand at all………………………………………………Understand very clearly

How much does your illness affect you emotionally? (e.g. does it make you angry, scared,
upset or depressed?)
0 1 2 3 4 5 6 7 8 9 10
Not at all affected emotionally..……………… ……………Extremely affected emotionally

Please list in rank-order the three most important factors that you believe caused your illness.
The most important causes for me:
_________________________________
_________________________________
_________________________________

21. Is your primary health concern related to physical PAIN?
a. Yes (answer 21, 22, 23, then skip to 27)
b. No (answer 24, 25, 26, then continue on 27)

22. Pain Disability Index
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 135

INSTRUCTIONS: The rating scales below are designed to measure the degree to which
aspects of your life are disrupted by chronic pain. In other words, we would like to know how
much pain is preventing you from doing what you would normally do or from doing it as well as
you normally would. Respond to each category indicating the overall impact of pain in your life,
not just when pain is at its worst.

For each of the 7 categories of life activity listed, please circle the number on the scale
that describes the level of disability you typically experience. A score of 0 means no disability at
all, and a score of 10 signifies that all of the activities in which you would normally be involved
have been totally disrupted or prevented by your pain.

Family/Home Responsibilities: This category refers to activities of the home or family. It
includes chores or duties performed around the house (e.g. yard work) and errands or favors
for other family members (e.g. driving the children to school).
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Recreation: This disability includes hobbies, sports, and other similar leisure time activities.
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Social Activity: This category refers to activities, which involve participation with friends and
acquaintances other than family members. It includes parties, theater, concerts, dining out, and
other social functions.
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Occupation: This category refers to activities that are part of or directly related to one’s job.
This includes non-paying jobs as well, such as that of a housewife or volunteer.
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Sexual Behavior: This category refers to the frequency and quality of one’s sex life.
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Self Care: This category includes activities, which involve personal maintenance and
independent daily living (e.g. taking a shower, driving, getting dressed, etc.)
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Life-Support Activities: This category refers to basic life supporting behaviors such as eating,
sleeping and breathing.
0 1 2 3 4 5 6 7 8 9 10
No Disability…………..……………………………………………………Worst Disability

Visual Analogue Scale – Please put a mark (X) on the line
23. (For pain-related condition) How severe is your pain BEFORE you received the acupuncture
treatment?
0 100
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 136

|------------------------------------------------------------------------------------------------|
No pain Pain as bad as it could possibly be
24. (For pain-related condition) How severe is your pain NOW after you received the acupuncture
treatment?
0 100
|------------------------------------------------------------------------------------------------|
No pain Pain as bad as it could possibly be
25. WHO Disability Assessment Schedule 2.0
INSTRUCTIONS: For each of the following statement, circle the one best response.
In the past 30 days, how much
difficulty did you have in…
No
difficult
y
Mild
difficult
y
Moderat
e
difficulty
Severe
difficult
y
Extreme
difficult
y or
cannot
do
Standing for long periods such as 30
minutes?
0 1 2 3 4
Taking care of your household
responsibilities?
0 1 2 3 4
Learning a new task, for example,
learning how to get to a new place?
0 1 2 3 4
How much of a problem did you
have in joining in community
activities (for example, festivities,
religious or other activities) in the
same way as anyone else can?
0 1 2 3 4
How much have you been
emotionally affected by your health
problems?
0 1 2 3 4
Concentrating on doing something
for ten minutes?
0 1 2 3 4
Walking a long distance such as a
kilometer [or 0.6 mile]?
0 1 2 3 4
Washing your whole body? 0 1 2 3 4
Getting dressed? 0 1 2 3 4
Dealing with people you do not
know?
0 1 2 3 4
Maintaining a friendship? 0 1 2 3 4
Your day-to-day work/school? 0 1 2 3 4

Visual Analogue Scale – Please put a mark (X) on the line
26. (For non-pain-related condition) How severe is your symptom today BEFORE you received the
acupuncture treatment?
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 137

0 100
|------------------------------------------------------------------------------------------------|
Not at all severe Extremely severe

27. (For non-pain-related condition) How severe is your symptom NOW after you received the
acupuncture treatment?
0 100
|------------------------------------------------------------------------------------------------|
Not at all severe Extremely severe

Treatment Outcome
28. Upon receiving today’s treatment, have you experienced any of the following from your
primary health concern?
a. Cure
b. Clear improvement
c. Slight improvement
d. No improvement
e. Don’t know
Concurrent Treatment
29. Are you currently seeking other treatment(s) (e.g. medications, physical therapy) or using any
self-care method(s) for your primary health concern? (select all apply)
a. Medical doctor/provider office visits or specialist consultations
b. Consultations with psychologists or social workers
c. Prescription medications (i.e. with a prescription written by medical providers)
d. Non-prescription medications (i.e. over-the-counter)
e. Natural products (i.e. herbs), dietary supplements, or vitamins and minerals
f. Chiropractic or osteopathic manipulation
g. Physical or occupational therapy
h. Massage therapy
i. Homeopathy or naturopathy
j. Yoga, Tai Chi, or Qi Gong
k. Meditation, deep breathing, progressive relaxation, or guided imagery
l. Others: __________________
m. No concurrent treatment

Treatment Cost Coverage
30. How do you normally pay for your acupuncture treatment (including today’s visit)?
a. Out-of-pocket expense
b. Insurance coverage/reimbursement
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 138

Access Barriers
INSTRUCTIONS: For each of the following, circle the one best response
Do you anticipate any of the following
factors to be potential barriers that will
prevent you from coming to an acupuncture
visit?
Not at
all
A little
bit
Some-
what
Quite a
bit
Very
much
Time 0 1 2 3 4
Cost 0 1 2 3 4
Distance 0 1 2 3 4
Transportation 0 1 2 3 4
Existing medical condition(s) 0 1 2 3 4
Other (please
describe:___________________)
0 1 2 3 4
Other (please
describe:___________________)
0 1 2 3 4

Treatment Referral
31. Has anyone referred you to seek care at Emperor’s College?
a. Self-referred
b. Referred by Emperor’s College student/supervisor/employee/alumni
c. Referred by outside medical provider/healthcare worker
d. Referred by friend/family member
Patient Satisfaction
32. Which of the following best describes how you feel about the overall care you have received
from today’s visit?
0 1 2 3 4 5 6 7 8 9 10
Not satisfied at all…………………………………………………….……………Completely satisfied

33. How satisfied are you with your decision to seek acupuncture for your health concern?
0 1 2 3 4 5 6 7 8 9 10
Not satisfied at all…………………………………………………….……………Completely satisfied

34. Patient-Provider Connection
INSTRUCTIONS: For each of the following statement, circle the one best response
Think of the healthcare provider who
provides your current treatment…
Not
at all
A little
bit
Some-
what
Quite
a bit
Very
much
I am satisfied with my healthcare provider 0 1 2 3 4
ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 139

I trust my healthcare provider 0 1 2 3 4
My healthcare provider pays attention to
my individual needs
0 1 2 3 4
My healthcare provider respects me 0 1 2 3 4
I feel my healthcare provider understands
me
0 1 2 3 4
My healthcare provider gives me support
and encouragement
0 1 2 3 4
Never Rarely Some-
times
Often Almost
always
My healthcare provider gives me enough
information
0 1 2 3 4

Treatment Plan
35. Did your acupuncturist provide you with a treatment plan for your health concern?
a. Yes, a treatment plan including the number of follow-up visits (go to 36)
b. No, acupuncturist only advised to return as needed (skip to 37)
c. No, no mention of treatment plan nor follow-up visit (skip to 37)

36. What is the recommended course of treatment provided by your acupuncturist?
a. Number of treatment(s) a week:______________
b. Estimated course of treatment: ___________ to ____________ week(s)
Intention
37. I INTEND to return for a follow-up visit for the same health concern.
a. Strongly agree
b. Somewhat agree
c. Neither agree nor disagree (answer 38)
d. Somewhat disagree (answer 38)
e. Strongly disagree (answer 38)

38. Please describe the reason(s) you might not return for a follow-up visit.

Please describe: _______________________________________________________

39. I INTEND to complete the recommended course of treatment.
a. Strongly agree
b. Somewhat agree
c. Neither agree nor disagree (answer 40)
d. Somewhat disagree (answer 40)
e. Strongly disagree (answer 40)

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 140

40. Please describe the reason(s) you might not complete the recommended course of treatment.

Please describe: _______________________________________________________

41. Have you already made an appointment for a follow-up visit?
a. Yes (skip to 44)
b. No (go to 42)

42. I PLAN to make an appointment for a follow-up visit.
a. Strongly agree
b. Somewhat agree
c. Neither agree nor disagree (answer 43)
d. Somewhat disagree (answer 43)
e. Strongly disagree (answer 43)

43. Please describe the reason(s) you might not consider making an appointment.

Please describe: _______________________________________________________

44. What led you to belief that acupuncture would be the appropriate treatment for your current
health concern?

Please describe: _______________________________________________________

45. Why did you choose Emperor's College Clinic to receive your acupuncture treatment?

Please describe: _______________________________________________________
Patient Preference on Follow-up Methods
46. We would like to follow-up with our patients after their initial treatment. Which of the
following method(s) you would prefer? (select all apply)
a. Telephone call
b. E-mail
c. Text-messaging
d. Postcard / letter via mail
e. Other (please describe:______________)
f. I prefer no follow-up
47. I think post-treatment follow-ups will make me more likely to return for follow-up visits.
a. Strongly agree
b. Somewhat agree
c. Neither agree nor disagree
d. Somewhat disagree
e. Strongly disagree

ADDRESSING FEDERAL PAIN RESEARCH PRIORITIES 141

48. If you are randomized to receive a follow-up telephone call in this study, what is the best time to
call you? It could be weekday or weekend based on your scheduled follow-up contact time.
(select all apply)
a. Morning (9am-11am)
b. Afternoon (12pm-5pm)
c. Evening (6pm-9pm)
Intervention and Follow-up Questionnaire
Instruction: We will collect some of your contact information for study purposes, including
mobile phone number and email address. We understand and care about your privacy, therefore,
your information will only be used for the study follow-up and will be deleted when the study is
complete.

49. Please provide the mobile phone number where we can best reach you (capable of send/receive
text-message):
a. Mobile number: _____________________________
b. Re-enter mobile number: ______________________

50. You will receive a follow-up questionnaire approximately 10 days from today. By completing
the follow-up questionnaire, you will enter into a drawing of 10 vouchers to receive a free
acupuncture session at the Emperor’s College clinic. Please provide the method you prefer to
complete the follow-up questionnaire:
a. Online (please provide your email address so that we can send you a link)
i. Email: _______________________________
ii. Re-enter email: ________________________
b. Telephone call (with same mobile number previously provided)
c. Paper survey via mail is also available, however we encourage selecting one of the above
methods

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Applications of mindfulness toward promoting psychosocial health in diverse communities

PDF

Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health

PDF

Opioid withdrawal symptoms, opioid use, and injection risk behaviors among people who inject drugs (PWID)

PDF

The evaluation of the long-term effectiveness of zero/low fluoroscopy workflow in ablation procedures for the treatment of paroxysmal and persistent atrial fibrillation

PDF

Trunk control during dynamic balance: effects of cognitive dual-task interference and a history of recurrent low back pain

Asset Metadata

Creator Lam, Chun Nok (author)

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

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior Research)

Publication Date 12/11/2019

Defense Date 08/13/2019

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

Tag Acupuncture,anxiety,Depression,Distress,follow-up,health behavior,integrative treatment,intervention,longitudinal,mechanism,mediation,Meditation,mindfulness,OAI-PMH Harvest,opioid,Pain,phone call,piecewise growth curve modeling,policy,Prescription Drug Monitoring Program,Psychology,Public Health,randomized controlled trial,research,statistics,Stress,text message,trend

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Black, David (committee chair), Chou, Chih-Ping (committee chair), Menchine, Michael (committee member), Milam, Joel (committee member), Unger, Jennifer (committee member)

Creator Email chunnok.lam@med.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-251196

Unique identifier UC11673893

Identifier etd-LamChunNok-8050.pdf (filename),usctheses-c89-251196 (legacy record id)

Legacy Identifier etd-LamChunNok-8050.pdf

Dmrecord 251196

Document Type Dissertation

Rights Lam, Chun Nok

Type texts

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

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

Repository Name University of Southern California Digital Library

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