Randomized comparative effectiveness study of 1-session vs 8-session online behavioral treatment for chronic pain: protocol for the national PROGRESS study

Abstract

Introduction

Evidence-based behavioral pain treatments are often inaccessible or infeasible for patients. A 1-session pain relief skills class (Empowered Relief; ER) has shown efficacy for multidimensional symptom reduction in various homogenous pain populations. Comparative effectiveness evidence in heterogeneous chronic pain is needed to inform decision-making and patient matching to treatment.

Objectives

Conduct a non-inferiority comparative effectiveness trial of 2 evidence-based online (telehealth) behavioral group treatments for chronic pain (1-session ER vs 8-session CBT) in a nationally representative sample derived from 6 study sites.

Methods

1200 adults ≥18 years of age with any type of pain ≥3 months with past-month intensity ≥3/10 will be recruited from 6 study sites. Participants are randomized 1:1 to either the online 1-session or the 8-session behavioral intervention. Patient-reported data are collected to 6 months post-treatment. Medical history, diagnoses, and healthcare utilization are obtained via medical records.

Main outcome measures

Multi-primary outcomes are changes in pain intensity and pain interference from baseline to 3 months post-treatment with hypothesized non-inferiority between the 2 study treatments. Statistical analyses include mixed models for repeated measures (MMRM). Priority secondary outcomes include sleep disturbance, pain catastrophizing, anxiety, and pain bothersomeness. Exploratory heterogeneity of treatment effects will be examined.

Ethics and dissemination

The study protocol was approved by all 6 site Institutional Review Boards including Stanford University School of Medicine (IRB 65439) as the coordinating site. Results will be published in peer-reviewed journals. PCORI (funder), participants, and advisors will receive summaries of the findings.

Trial registration

ClinicalTrials.gov Identifier: NCT 05612750.

Evidence-based behavioral pain treatments are often inaccessible or infeasible for patients. In addition, pain research samples are often homogenous. The PROGRESS study is a randomized comparative effectiveness non-inferiority study comparing the 'gold standard' 8-session cognitive behavioral therapy to 1-session Empowered Relief. The 6-site study includes 1,200 patients with diverse chronic pain, and emphasizes representative enrollment via deep engagement methods with patients and stakeholders. Findings will inform shared decision making and patient matching to treatment.

Introduction

Chronic pain impacts approximately 50 million Americans and is one of the most common reasons people seek medical care. It is a pervasive and often debilitating condition and costs the US >$560 billion each year.¹ Chronic pain rates are rising despite advances in medical and interventional treatments and pharmacologic approaches.² Notably, racial and ethnic subpopulations, women, older adults, and patients with lower socioeconomic status are disproportionately impacted by chronic pain and experience the greatest barriers to effective care.³ Expanding more equal access to low-risk pain treatment remains a critical national need.

Behavioral pain management approaches can effectively reduce pain and associated symptoms. Cognitive behavioral therapy (CBT) is the gold-standard behavioral treatment for chronic pain with 40 years of robust empirical support^4–7 including multiple high-quality systematic reviews.⁶ While meta-analyses indicate CBT is most effective for improving mood and pain catastrophizing,⁶ a notable limitation is the variability of study methods (eg, protocol used), populations studied (pain conditions), and rigor (active comparator).⁶ CBT is a skills-based intervention that helps individuals manage chronic pain through pain education, promoting adaptive thinking, emotion regulation, behavioral activation, goal setting, coping skills, better sleep, social support, and reduced pain catastrophizing. Historically, trained psychologists deliver group CBT in-clinic with 8 weekly sessions lasting 2 hours each (total treatment time of 16 hours). The shift to remote care during the COVID-19 pandemic demonstrated that CBT can be feasibly received online (ie, via telehealth), with meta-analyses showing equivalence in efficacy between in-person and remote formats for various health conditions, including chronic pain.⁸ Despite some advances with regard to remote options and insurance coverage, clinically CBT is often unavailable, infeasible, or unwanted due to a variety of factors, including treatment burden, shortage of trained mental health professionals, and stigma surrounding mental health treatment.^9,10

A key question arises as to whether behavioral pain treatments may be offered to patients more efficiently without comprising efficacy; or, with an acceptable level of reduced efficacy, given that the common alternative is no behavioral treatment. In multisession pain research, treatment “completer” thresholds of 5 of 8 sessions are often applied, reflecting that session missingness is common. Clinically, many patients who begin 8-session behavioral pain treatments do not receive the full course of treatment. Difficulty arises in ascertaining which sessions and content elements are essential for efficacy, given that each weekly session introduces new information and skills. Given that the treatment information is delivered in segments, receiving any one session of CBT, or any other effective behavioral treatment, represents a small fraction of the overall information and skills. In contrast, 1-session interventions are narrower in scope, and compressed in format and presentation to provide a full treatment rationale, content elements, and resources that allow for complete clinician delivery and patient receipt of information. Disadvantages of the 1-session approach include lack of clinician contact and peer support, and possibly decreased motivation to complete between-session skills application “homework.” Advantages of a 1-session approach include convenience, wider access to behavioral pain care, and potentially enhanced feasibility and patient engagement.^11,12

Empowered relief (ER) is a 1-session pain relief skills intervention that streamlines key evidence-based principles and skills from CBT and mindfulness interventions into a single 2-h didactic class that includes pain neuroscience education. ER is delivered by a licensed clinician and certified instructor using a standardized slide presentation and manual with instructor scripts. ER is highly didactic and not psychological therapy per se; its highly structured class content allows for interprofessional delivery and wider reach because it is not constrained to psychologist instructors. Class participants engage in experiential exercises, self-tailor the content, create a personalized plan for empowered relief, and receive a free MP3/App that contains a 20-min calming binaural audiofile for ongoing daily use. The 1-session format and branding were designed to address stigma and facilitate patient engagement, with anonymous participation possible in online formats.

ER has demonstrated efficacy for reducing pain intensity, pain interference, pain bothersomeness, sleep disturbance, depression, and anxiety in heterogeneous pain conditions.^4,13–16 ER was shown to impart clinically meaningful effects that were non-inferior to 8-session CBT for reducing pain catastrophizing, pain intensity, pain interference, depression, anxiety, fatigue, and pain bothersomeness at 3 months post-treatment in a single-site NIH-funded chronic low back pain treatment study.⁴ While ER improved physical function, on this one variable ER was inferior to CBT. Durability of ER effects at 6 months post-treatment was also shown. Ten completed clinical trials (5 randomized controlled trials; RCTs) have described multidimensional and longitudinal symptom improvement months after receipt of ER. Both in person and online formats have been tested in various patient populations chronic pain with comorbid prescription opioid misuse and use disorder,¹⁷ orthopedic trauma surgery,¹⁵ spine surgery,¹⁸ breast cancer surgery,¹⁴ mixed etiology chronic pain,^13,16 chronic low back pain,⁴ and in Marfan syndrome and related rare connective tissue disease.¹⁹ To date, the intervention is being delivered to patients with acute and chronic pain by certified instructors in 30 countries and in 8 languages. While adoption of ER has been rapid, the absence of comparative effectiveness evidence creates risks for false conclusions. For instance, ER may be inferior to longer course behavioral pain treatments, yet premature adoption of ER might impede access to more intensive and needed treatments. Larger and diverse samples are needed to explore heterogeneity of treatment effects. Moreover, multiple positive efficacy studies might unwittingly promote a message that “ER is sufficient for everybody” when the majority of research involved single-site study designs^4,13,20,21 (3 enrolled nationally)^16,19,22 and relatively homogenous samples do not support such a conclusion.

Multiple evidence gaps motivated the PROGRESS (Pain Relief with Online Groups that Empower Skills-based Symptom Reduction) study, including a lack of effectiveness evidence for online behavioral pain treatments in national clinical trials, particularly in heterogenous populations, and a lack of evidence on heterogeneity of treatment effects. Funded by the Patient Centered Outcomes Research Institute (PCORI),²³ PROGRESS is a national, 6-site, 2-arm, randomized, online, non-inferiority pragmatic comparative effectiveness trial for Americans with chronic pain. PROGRESS is comparing online 8-session CBT to 1-session ER in a nationally representative sample of patients with mixed etiology chronic pain (ie, broad inclusivity of chronic pain) to yield the pragmatic evidence needed to determine which treatment is best for whom. Our results will inform shared decision-making, pain treatment “dosing” (1 vs 8 session), and precision matching to treatment to either a 1-session treatment or fast-tracking to more intensive 8-session CBT for specific subgroups.

Methods

Study setting and recruitment

We are enrolling a heterogeneous sample of 1,200 individuals with chronic pain (N = 200 from each of 6 study sites; see Table 1). Note that we are also conducting 3 additional randomized trials involving N = 150 patients in each of 3 specific chronic pain populations: Lupus, Pelvic Pain, and Young Adults. While the total PROGRESS study sample is 1,650 participants, this protocol describes the methods for the primary study sample (N = 1,200) which is enrolling patients with any type of chronic pain from all 6 study sites. Table 1 details the study sites which include an academic medicine site (Stanford Medicine), a closed payor network (Intermountain Health), a VA site (Arizona Veterans Research and Education Foundation), a non-academic community health network (Lehigh Valley Health Network), a Medicare/Medicaid/dually eligible payor (SCAN Health Plan), and a large Medicare Advantage plan (Humana Healthcare Research). Figure 1 details the study recruitment methods and participant flow. Study flyers were constructed by the research team and advisors with lived chronic pain experiences. Advertisements include contact information for site study coordinators (email and phone) and the study website (progress.stanford.edu) which allowed for online eligibility screening. PROGRESS is conducted in compliance with the Stanford IRB, local site IRBs, and PCORI research regulatory standards.

Table 1.

Study sites, location, and setting.

Study site	Location	Setting
Stanford Medicine	Bay Area, Northern California	35 Pain, Primary Care Clinics
SCAN Health Plan	California, Arizona, New Mexico, Nevada, and Texas	Statewide Payor System
Lehigh Valley Health Network	Pennsylvania	Outpatient Clinics
Intermountain Health	Colorado, Idaho, Kansas, Montana, Nevada, and Utah	Outpatient Clinics
Phoenix VA Health Care System	Pheonix & greater Arizona	Pain & Outpatient Clinics
Humana	Southeastern United States & National	National Payor System

Figure 1.

Recruitment and participant flow.

Data collection and management

CHOIR²⁴ is the study informatics, data capture, and participant communications platform. Study staff call participants to remind them to complete surveys and review CHOIR data as it enters the system to ensure quality.

Eligibility criteria

Adults ≥ 18 years of age with chronic pain (≥3 months duration) of any cause with past-month average pain intensity ≥3/10 receiving medical care at any of our 4 medical study sites or being covered by our 2 payor sites. Exclusion criteria: Inability to provide informed consent, inability to participate meaningfully in group treatments (eg, evident cognitive impairment or lack of English fluency), or perform study activities (complete surveys online or by postal mail; receive treatment online or, in rare cases, by telephone) and received either study treatment in the prior 3 months. Active suicidality at screening is exclusionary.

Informed consent, randomization, and blinding

The study is being conducted through the informatics platform, CHOIR—an open-source learning health system.²⁴ From the study website, interested patients select a study site based on where they receive medical care, taking them to the site-specific CHOIR portal for a study screening form, verification of eligibility, and informed e-consent. Participants are e-randomized (1:1) to treatment type after the screening survey but treatment is not revealed until after participant completion of baseline survey. Treatment allocation is communicated to participants by email and during a treatment scheduling call with the study coordinator. Randomization is continuous, with new treatment sessions opened on a rolling basis. Participants are given numerous scheduling options for receiving their assigned treatment. A small proportion of participants (less than 5% based on early study enrollments) communicate scheduling conflicts that impede their ability to receive their assigned treatment. Participants who do not receive their assigned treatment are invited to continue in the study by completing survey assessments only. Early examination of attrition patterns revealed that the majority of attrition was occurring before initiation of assigned treatment. To ensure adequate power for planned comparisons, the study implements over-enrollment to ensure 600 participants initiated their assigned treatment. The principal investigator, statisticians, and all co-investigators not involved in delivery of study treatments are blinded to treatment group allocation. The project manager, study coordinators, and study therapists are unblinded.

Interventions

Cognitive behavioral therapy (CBT) is considered the gold standard behavioral treatment for chronic pain^4–7 equipping learners with self-regulatory skills and action plans that support pain self-management. PROGRESS uses Dr Beverly Thorn’s manualized low literacy CBT protocol (Learning About My Pain, “LAMP”),^25–27 that is freely available online, and used in prior PCORI-funded research. The original LAMP protocol involves 10 sessions; for this study, the intervention was adapted into 8 2-hour sessions, delivered weekly for 8 weeks by doctoral-level pain psychologists with post-doctoral specialized pain training. CBT therapists were either experts on the specific LAMP protocol or completed training prior to study launch. The LAMP CBT study therapists include Phoenix VA clinical psychologists, a LAMP expert psychologist from Southeastern United States, and Stanford pain psychology fellows who progress through a mentored learning pathway: Observing a master LAMP therapist teaching a class, then teaching a non-research CBT group, and finally teaching a PROGRESS CBT treatment group under the observation/supervision of the master LAMP trainer. Licensed and experienced LAMP therapists receive ongoing consultation and support as needed. For PROGRESS, the CBT therapists and ER instructors are distinct.

Because CBT relies on group interaction and the therapeutic value of multiple participant voices, session scheduling targeted 12 active attendees. Sessions with fewer than 7 participants at initiation were reviewed for rescheduling. Class capacity was increased from 20 to 25 participants to account for participants who did not initiate their scheduled first session of treatment. The therapist manual provides agendas for each session, focusing on cognitive and behavioral pain management tools and relaxation scripts. The content balances education about chronic pain with actionable self-management tools. The manual encourages group leaders to create a supportive space for participants to share their experiences, receive therapist and peer validation, and promote camaraderie and connection. The participant workbook presents CBT skills in a digestible format with weekly opportunities to practice these new tools. Participants have access to recorded relaxation exercises and key takeaways from each session to reinforce skills and guide home practice. If a participant misses the first class, they are rescheduled into a new class cohort; absences from later classes do not result in rescheduling or staff intervention once treatment is initiated participants remain in their treatment cohort. Participants are required to be visible via camera and cameras are provided to participants as required.

LAMP CBT Fidelity Rating Forms were developed by Dr Beverly Thorn and validated in prior PCORI-funded studies to ensure fidelity across sites and therapists.^28,29 For PROGRESS, across all sites, the study coordinators are trained to complete the CBT fidelity rating form in conjunction with the therapist to verify that the classes are delivered as intended in content and structure (see Online Supplement A). CBT fidelity forms are monitored by the CBT Director (HK) to ensure adherence to the treatment protocol. Any participant issues are reported to the CBT Director and principal investigator, whether those issues are noted by the instructor during the session, or by verbal or written participant communication, with discussion on whether actions are needed.

Empowered relief (ER) is a 1-session 2-hour pain relief skills intervention that includes pain neuroscience education and self-regulatory skills common to CBT and mindfulness interventions⁴ Taught at a lay level using narrative stories and large picture visuals, participants learn about the relationship between stress, pain, and perceived threat; automatic and protective physiologic and psychologic responses and how those impact pain. The session is didactic with no group introductions or open sharing of one’s history. Content includes completion of a self-completed and self-scored “pain survey” to help participants understand their unhelpful patterns of thoughts and emotions in relation to pain, and 3 skills to interrupt unhelpful and automatic pain responses (central nervous system calming via relaxation, altering unhelpful thoughts and helplessness, and supportive actions). Participants complete a personalized plan during the class for ongoing use of the skills in the following months; they receive a free binaural guided calming app that they incorporate into their personalized plan with daily use encouraged. ER is to be distinguished from CBT because it is not “therapy,” does not benefit from the ongoing therapeutic alliance and peer alliances, and includes greatly streamlined content, omitting many CBT topics and materials, such as activity/pacing, problem solving, pain and sleep, different relaxation audiofiles provided, review of prior homework assignment and positive reinforcement for engagement and completion. Because ER does not rely on group interaction, class cohorts have no maximum enrollment and a minimum of 3 registered attendees; class sizes are dependent on the enrollment rates and to date have ranged from 3 to 17 participants.

ER instructors are certified via a 2-day clinician certification workshop offered through the Stanford Center for Continuing Medical Education PROGRESS-certified ER instructors include pain psychology staff at Stanford, social work staff at Lehigh Valley Health Network, and clinician consultants; the Phoenix VA mental health staff incorporate PROGRESS study participants into their standard of care ER classes. Participants assigned to ER receive emailed and mailed hard copy class materials in advance of their online treatment session. Treatment fidelity checklists are completed at every class by trained study coordinators (see Online Supplement B); incomplete checklists are reported to the project manager and the principal investigator (none to date). Unlike CBT, ER participants may attend the class anonymously and with their camera off if desired, and a family member is welcome to attend with them.

For both interventions, study coordinators document class attendance and ensure participants receive all class materials. Recognizing that ER may be insufficient for some participants, we sought to offer ER treatment completers who we determined to be non-responders access to 8-session CBT as a higher level of care. As such, the first 50 participants at each site who complete the ER session and 6-month post-treatment survey will be evaluated on the co-primary outcomes. Those who fail to achieve at least a 20% improvement for pain intensity and pain interference (dually) will be invited to cross-over and receive the CBT intervention. This cross-over design (from online ER to online CBT) enables us to assess whether non-responders engage in longer course, more intensive treatment, and to derive signal for whether CBT confers additional benefit for ER non-responders.

Study aims and hypotheses

The study primary aims were selected based on patient preference surveys administered received from more than 500 patients. Secondary outcomes included those recommended by patients and informed by prior efficacy studies.

Primary aim 1

Compare treatment groups for multi-primary outcomes (pain intensity and pain interference) at 3 months post-treatment. We hypothesize ER will be non-inferior to 8-session CBT for reducing pain intensity or pain interference.

Primary aim 2

Compare treatment groups for 4 priority secondary outcomes (sleep disturbance, pain catastrophizing, pain bothersomeness, and anxiety), treatment completion/adherence, and treatment burden. We hypothesize ER will be non-inferior to 8-session CBT for improving priority secondary outcomes at 3 months post-treatment; ER will have superior treatment completion/adherence (n; % attending ER or at least 5 sessions of CBT) and lower burden ratings than 8-session CBT.

Primary aim 3

Compare treatment groups for other secondary outcomes (satisfaction with social roles and responsibilities, patient global impression of change, anger, fatigue, and depression). We hypothesize ER will be non-inferior to CBT on improving secondary outcomes at 3 months post-treatment.

Primary aim 4

Compare durability of treatment effects at 6 months post-treatment. We hypothesize ER will be non-inferior to 8-session CBT for multi-primary and priority secondary outcomes at post-treatment 6 months.

Exploratory aim 5

Compare healthcare utilization and opioid utilization using electronic health record (EHR) and CMS data 3 months prior to enrollment to last 3 study months. We will assess medication status (eg, stable dose, voluntarily tapering, involuntarily tapering); as well as medication access (eg, no access problems; difficulty with medication access). We hypothesize both groups will show reduced utilization during the final 3 study months.

Exploratory aim 6

Allow early ER participants who were non-responders on either primary outcome at 3 months to cross-over to online 8-week CBT (and characterize response). We hypothesize that ER non-responders, after 8-week CBT, will achieve success for a multi-primary outcome.

We will also investigate the trajectory of both treatments individually (within group treatment outcomes).

Measurement table

Table 2 describes the primary and secondary outcome measures and the timepoints at which they are collected. Baseline demographics include birth date, gender, sex at birth, race, ethnicity, family structure (eg, marital status, number of children), education level, veteran status, disability status, employment status, and household income level. Healthcare utilization includes number of medical visits, type of medical visits, diagnoses, medical care coverage, and prescription data.

Table 2.

Primary, priority secondary, secondary, and exploratory outcomes.

Primary or secondary	Name of outcome	Specific measure to be used	Timepointsa
Multi-primary	Pain intensity	Numeric Rating Scale (0–10)30–33	B; IPreT; M1, 2, 3, 6
Multi-primary	Pain interference	PROMIS Pain Interference 6-item31,34–38	B; IPreT; M1, 2, 3, 6
Priority secondary	Sleep disturbance	PROMIS Sleep Disturbance 6-item39	B; IPreT; M1, 2, 3, 6
Priority secondary	Pain bothersomeness	Numeric Rating Scale (0–10)40	B; IPreT; M1, 2, 3, 6
Priority secondary	Pain catastrophizing	Pain Catastrophizing Scale (13 items)41,42	B; IPreT; M1, 2, 3, 6
Priority secondary	Anxiety	PROMIS Anxiety 6-item43,44	B; IPreT; M1, 2, 3, 6
Secondary	Satisfaction with social roles and responsibilities	PROMIS Satisfaction with Social Roles and Responsibilities 6-item45	B; IPreT; M1, 2, 3, 6
Secondary	Patient global impression of change	PROMIS Global Impression of Change 1-item46–48	M1, 2, 3, 6
Secondary	Depression	PROMIS Depression 6-item49,50	B; IPreT; M1, 2, 3, 6
Secondary	Fatigue	PROMIS Fatigue 6-item45,51–53	B; IPreT; M1, 2, 3, 6
Secondary	Anger	PROMIS Anger 6-item17,54	B; IPreT; M1, 2, 3, 6
Exploratory	Health service utilization	Number of medical visits during final 3 months of the study vs 3 months prior to enrollment.	Chart review or CMS data (Humana/SCAN)

^aB, baseline; IpreT, immediately pre-treatment (administered if >2 weeks has elapsed since baseline); M, post-treatment months.

Baseline survey

In addition to all of the measures displayed in Table 2, the baseline survey includes demographic questions regarding race, ethnicity, sex at birth, gender identification, education, household income, number of household members (and dependents), insurance type, employment status, marital status, and disability status. Clinical items include pain history and prior receipt of behavioral treatment for chronic pain, current pain treatments, and barriers to pain care. Baseline substance use is assessed with the NIDA 4-item Quick Screen V1.0⁵⁵ and the number of pain body sites is assessed with the Stanford CHOIR Visual Body Map.⁵⁶ Treatment expectancies are assessed immediately after revealing treatment group allocation via the 6-item Stanford Expectation of Treatment Scale.⁵⁷

Pre-treatment survey

Participants’ baseline survey will be used in the analyses except in cases where more than 2 weeks have elapsed between baseline survey and treatment initiation. In these cases, to minimize regression to the mean, and overinflating treatment effects, participants who exceed the 2 weeks threshold are administered a second pre-treatment baseline to provide an accurate capture of their current symptom profile. This pre-treatment battery of measures omits demographics and retains symptom measures for a current assessment. Pre-treatment assessments are considered the analytic baseline for participants for whom treatment timing required administration of this measure.

Treatment appraisal and satisfaction survey

Immediately after completion of assigned treatment participants are asked to rate Treatment burden (0–100); 0 = none and 100 = extreme (also qualitative); (2) Treatment satisfaction—5 items assess satisfaction with materials and instructor, perceived utility⁴; (3) Online ease—3 items assess ease/difficulty receiving online treatment (also qualitative); and for ER only (4) Family attendance. Free text feedback is also encouraged and collected.

Post-treatment surveys

At months 1, 2, 3, and 6, participants receive the Table 2 measures as well as free text items relating to health status (new health/medical problems, emergency or urgent care visits, and free text fields).

Participant compensation

Participants may receive up to $195 in the form of Amazon gift card codes for study survey completion as follows: $25 for baseline and a second pre-treatment baseline if >2 weeks have elapsed since initial survey, treatment appraisal and satisfaction survey, and months 1 and 2 follow-up surveys. Participants received $35 for completing surveys at months 3 (primary) and 6.

Engagement approach

The design of the PROGRESS study was informed by 3 national online surveys of approximately 2,000 diverse adults with chronic pain and a separate survey of 905 clinicians¹⁰ to understand stakeholder priorities, needs, and preferences related to behavioral pain treatment. As the study moved toward implementation, the engagement framework evolved through co-development with consultant patient advisors and research staff.⁵⁸ This multi-level approach employed multiple advisory boards that were engaged at key decision points throughout the project.

The overarching goal of the PROGRESS engagement activities was to ensure that all stages of the study were understandable, accessible, and free from barriers, including stigma. During advisor recruitment, the core advisory group prioritized representation across racial/ethnic, sex, age, geographic, socioeconomic, and educational backgrounds reflective of the target participant population. These advisors co-developed and substantially refined major study components—including study design, outcome measures, patient-facing surveys, and recruitment materials—before these materials were further optimized by both local and national advisory boards. Patient-facing recruitment language was specifically tailored by local groups to fit each local sites culture and target population. Advisory boards meet at least 3 times annually to review progress toward diversity goals, assess the accessibility of treatment and study procedures, develop new surveys, and recommend modifications as needed.

Analytic plan

For both study treatment groups (ER and CBT), we will summarize data for treatment burden, satisfaction, and completion; online ease, and missing data. We will examine effects by site, therapist, insurance type, and copayment burden. We will also examine CBT attendance as a continuous variable and baseline predictors for attrition. For ER, we will report the rate of family attendance.

Aim 1: We will generate one-sided 95% simultaneous CIs for differences in pain intensity and interference at month 3 between CBT and ER using mixed models for repeated measure (MMRM) regression analysis. Unlike 2-sided intervals that assess both upper and lower bounds, a one-sided interval is used here since we are only concerned with the lower bound (for non-inferiority). We will conduct 2 separate MMRM analyses for pain intensity and pain interference; for addressing missing data, MMRM is considered superior over other options such as multiple imputation or least observation carried forward.⁵⁹ In one MMRM analysis, the dependent variables will be the primary outcome (pain intensity and interference) at baseline, month 1, 2, 3, and 6, and the independent variables will be time (baseline, month 1, 2, 3, and 6; categorical variables), treatment group (CBT vs ER; binary variable), and interactions between months 1, 2, 3, and 6 and treatment group. We will apply an unstructured variance covariance matrix among dependent variables. First, we will estimate the average change from baseline in pain intensity and pain interference by treatment group and by month 1–3 based on the fitted MMRM. The estimated within-group changes provide preliminary evidence on the effectiveness of the intervention (CBT and ER). We expect to observe clinically meaningful improvement in all endpoints after the interventions. More importantly, the within-group and between-group difference at month 3 will be summarized as the estimated coefficient for interaction between month 3 and treatment indicator. We will further estimate covariance between the estimated differences in the 2 primary outcomes from 2 separate MMRM analyses by bootstrapping individual patients. Accounting for the estimated covariance, we will generate one-sided 95% simultaneous CIs for the differences in pain intensity and interference in the form of ($-\infty , U_1]\times (-\infty , U_2]$ to ensure that the probability that the true between-group difference in pain intensity and pain interference is less than $U_1$ and $U_2,$ respectively, is 95%. We will determine if the upper end of the confidence interval is below the corresponding non-inferiority margin. The non-inferiority margin is derived from the literature which cites the minimal important difference (MID) on each specific outcome of interest (see Table 3). The non-inferiority in the primary endpoint will be established if the upper end is below the non-inferiority margin.

Table 3.

Non-inferiority margin and power (based on prior and published data).

Primary or secondary	Outcome	Time point	Non-inferiority margin (NM)	NM/baseline SD (baseline SD)	Power
Primary outcomes
Multi-primary (1)	Pain intensity	Month 3	1.560	0.88 (1.7)	>99%
		Month 6	1.560	0.88 (1.7)	>99%
Multi-primary (2)	Pain interference	Month 3	4.535	0.74 (6.1)	93.6%
		Month 6	4.535	0.74 (6.1)	92.5%
Priority secondary outcomes
Priority secondary (1)	Sleep disturbance	Month 3	4.061	0.53 (7.5)	>99%
		Month 6	4.061	0.53 (7.5)	>99%
Priority secondary (2)	Pain bothersomeness	Month 3	1.562	0.68 (2.2)	>99%
		Month 6	1.562	0.68 (2.2)	>99%
Priority secondary (3)	Pain catastrophizing	Month 3	6.4863	0.67 (9.7)	>99%
		Month 6	6.4863	0.67 (9.7)	>99%
Priority secondary (4)	Anxiety	Month 3	3.064	0.34 (8.8)	>99%
		Month 6	3.064	0.34 (8.8)	>99%
Secondary outcomes
Secondary (1)	Satisfaction with social roles	Month 3	2.061	0.25 (8.0)	91%
Secondary (2)	Patient global impression of change	Month 3	0.565	0.33 (1.5)	>99%
Secondary (3)	Depression	Month 3	3.050,64	0.35 (8.5)	>99%
Secondary (4)	Fatigue	Month 3	4.050	0.49 (8.1)	>99%
Secondary (5)	Anger	Month 3	5.035	0.43 (11.6)	>99%

For Aims 2–4, we will repeat the MMRM analysis and one-sided confidence intervals for within and between group differences, and will apply the corresponding non-inferiority margins based on a priori and published thresholds cited in Table 3. Aim 2 will test 4 priority secondary outcomes (sleep disturbance, pain catastrophizing, pain bothersomeness, and anxiety). Aim 3 will test other secondary outcomes satisfaction with social roles and responsibilities, patient global impression of change, anger, fatigue, depression. Aim 4 will repeat tests for Aims 1 and 2 at the 6-month follow up.

For Exploratory Aim 5, we will extract the number of health services during the final 3 months of the study period and the 3 months prior to study enrollment and calculate proportions, means, mean proportions, and corresponding 95% CIs. A mean proportion (ratio of total number of services during the final 3 study months to the pre-enrollment 3-month period) of < 0.5 would provide evidence that the intervention reduces utilization.

For Exploratory Aim 6, we will examine the treatment effect of CBT on multi-primary outcomes (pain intensity and interference) among non-responders to ER at month 6 (change in both after completing the CBT intervention based on a paired t-test). We will examine treatment effects on secondary variables at months 3 and 6.

Planned heterogeneous treatment effects (HTE) testing

The goal of HTE is to explore variability in treatment response among key subgroups of interest. We will conduct sensitivity analyses to test all primary and secondary outcomes in 6 subgroups: (1) primary pain (head, musculoskeletal, neuropathic, other); (2) number of pain conditions (>1 vs <1); (3) race and ethnicity (Black, White, Hispanic, and Asian); (4) socioeconomic status (above vs below median); (5) disability status; or (6) Medicare, Medicaid, or dual recipient. The primary pain types were selected based on prevalence; the number of pain conditions was selected as chronic overlapping pain conditions are common yet an underappreciated index of pain burden. The remaining 4 HTE categories were selected because they are commonly associated with degree of pain burden, access to pain care, and pain treatment outcomes. We will test whether the difference in the specified subgroup is the same as that in the overall population. We will use Tukey’s method for multiple adjustment.

Non-inferiority margins

The specific non-inferiority margins are derived from the literature as minimal important difference thresholds in chronic pain populations. For instance, for pain intensity, 1.5 is cited as a minimal important difference.⁶⁰ If the difference between ER and CBT is <1.5, ER will be considered non-inferior because it is within the acceptable margin. Based on our prior study,⁴ the cross-sectional SD for pain intensity and interference is 1.7 and 6.1, respectively. The SD of 3-month change in pain intensity and interference is 2.0 and 7.0, respectively. The minimal important difference (MID) thresholds were derived from established literature where MIDs were calculated for each variable. While patients selected the outcomes that were meaningful to them, they were not asked to select MIDs for each variable. Because a full study on patient MIDs was out of scope, the most rigorous approach is to apply the standards set by other researchers who performed specific MID determinations.

Power consideration

Table 3 displays the non-inferiority margin and power for each outcome. Based on engagement rates for our online study⁶⁶ and estimates for online CBT, we conservatively assume a 84% follow-up rate at 3 months and 72% at 6 months, ie, 1008 patients at month-3 visit and 864 patients at month-6 visit. We estimate the power of the non-inferiority test based on the one-sided confidence interval for the between-group difference for the endpoint of interest. For Aim 1, the non-inferiority margins are 1.5 for pain intensity⁶⁰ and 4.5 for pain interference,³⁸ respectively. In our prior study, the observed SDs of the 3-month change of pain intensity and interference were 2.0 and 7.0, respectively.⁴ Furthermore, the observed 3-month reduction in pain intensity was 1.02 in ER and 1.76 in CBT. The observed 3-month reduction in pain interference was 4.27 in ER and 7.72 in CBT.⁴ Assuming the 3-month changes in these outcomes are independent and the differences between groups are identical to those observed in prior studies, the probability that the upper ends of both one-sided 97.5% CIs are below their non-inferiority margin is 93.6%. Accounting for potential positive correlation between outcomes, the upper ends of the 97.5% simultaneous CIs will be even lower and the actual power will be even greater. For Aim 2, the non-inferiority margins are 4.0, 1.5, 6.8, and 3.0 for sleep disturbance,⁶¹ pain bothersomeness, pain catastrophizing, and anxiety,⁶⁴ respectively. In our prior work, the observed SDs of the 3-month change for sleep disturbance, pain bothersomeness, pain catastrophizing, and anxiety were 8.9, 2.6, 10.6, and 7.5, respectively.⁴ Assuming that 3-month changes in these outcomes are independent and there is no difference between groups, we have >99% power that the upper ends of all 4 one-sided 98.75% CIs are below their corresponding non-inferiority margins. For Aim 3, the non-inferiority margins are 2.0, 1.0, 3.0, 4.0, and 5.0 for satisfaction with social roles and responsibilities,⁶¹ patient global impression of change, depression,^50,64 fatigue,⁵⁰ and anger,³⁵ respectively. In our prior study, the observed SDs of 3-month change were 7.2, 9.8, and 10.9 for depression, fatigue, and anger, respectively.⁴ Reported SDs of social roles and responsibility and patient global impression of change are 8.0 and 1.5, respectively.^67,68 Assuming a correlation coefficient of 0.5 between baseline and month 3 in our pilot study, the SDs of 3-month change are 8.0 and 1.5 for social roles and responsibility and patient global impression of change, respectively. Assuming no difference at month 3, the power for non-inferiority is >90% for all endpoints. For Aim 4, the sample size is reduced to 432 per arm. Under the same assumptions at month 6, the power of the non-inferiority is >99% for pain intensity, 92.5% for pain interference, and >99% for all 4 priority secondary endpoints. In summary, a sample size of 1,200 provides adequate power for all aims.

Data monitoring, harms, and auditing

The primary risk to participants is breach of confidentiality, including potential exposure of personal health information or sensitive disclosures during group sessions. Protections include secure, password-protected electronic and paper data storage, HIPAA training for staff, and structured guidance on appropriate group disclosures. ER participants may remain anonymous during sessions, and confidentiality ground rules are reviewed at the outset of CBT. Emotional discomfort or distress during relaxation exercises is a minor risk; participants are advised to discontinue audio use if distress occurs. Internal data validity is supported by built-in database checks and periodic monitoring by study staff. A centralized Data Safety and Monitoring Board will perform ongoing evaluations of risk and evaluate all adverse events on an annual basis, and ensure data integrity.

Participant protections for worsening symptoms

All individuals who undergo eligibility screening are given 24-h national crisis telephone hotlines and resources for suicidality, regardless of eligibility. Participant-reported data on depression and substance use are continuously monitored through CHOIR. Responses that exceed predefined clinical thresholds are flagged as adverse events. CHOIR alerts study staff, the project manager, and the PI when depression scores increase more than a half SD or a PROMIS Depression score >68. For alcohol use, CHOIR flags consumption of 5 or more drinks in a single day (monthly or weekly) and an alert is routed to study staff for follow-up; for daily use at that level an alert is sent to the site PI and, if the participant is assigned to CBT, the CBT instructor. Each site has standard operating procedures in place for worsening clinical symptoms that includes escalation to the site PI to assess whether a phone call or intervention is warranted.

Ethics, reporting, and dissemination

This study protocol is approved by the Stanford University IRB and each study site’s local IRB or WIRB-Copernicus Group. The trial is conducted in accordance with the requirements of PCORI.

This study follows CONSORT guidelines for reporting randomized controlled trials including (http://www.equator-network.org/) and CONSORT extensions for non-pharmacologic treatments, pragmatic trials, and patient-reported outcomes.

Study timeline

Enrollment began January 2023 and will close in late 2026 with data collection ending mid-2027 and database lock scheduled for June 2027. Data analyses are expected to complete January 2028.

Funding

This work is supported through a Patient-Centered Outcomes Research Institute (PCORI) Project Program Award to Stanford University (OPD-2021C1-22347; PI, B.D.D.).

Disclaimer

All statements in this report are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors, or Methodology Committee.

Conflicts of interest

Empowered Relief is owned and copyrighted by Stanford University. Dr Darnall receives no personal monies from Empowered Relief or from the Stanford CME clinician certifications workshops. Dr Darnall is Chief Science Advisor at AppliedVR and she receives consulting fees for this role (unrelated to the current work). Dr Darnall receives royalties for 4 pain treatment books she has authored or coauthored. She is the principal investigator for 2 pain research awards from the PCORI and is principal investigator for 2 NIH grants, each of which investigates either 8-session CBT, Empowered Relief, or both. Dr Darnall is on the Board of Directors for the Institute for Brain Potential, and is on the Medical Advisory Board for the Facial Pain Association. Dr Darnall is a scientific member of the NIH Interagency Pain Research Coordinating Committee, a member of the ACOEM chronic pain guidelines committee, a former member of the Centers for Disease Control and Prevention Opioid Workgroup (2020–2021), and a current member of the Pain Advisory Group of the American Psychological Association. Dr Mackey receives research funding from the NIH, Food and Drug Administration, and Patient-Centered Outcomes Research Institute (administered through Stanford University). He is an unpaid advisor to both ACTTION (Analgesic, Anesthetic, and Addiction Clinical Trial Translations, Innovations, Opportunities, and Networks) on their over-sight committee, and the American Chronic Pain Association (ACPA) for their scientific oversight. All other authors have no conflicts of interest to disclose.