Evaluation of Intervention Components to Maximize Yoga Practice among People with Chronic Pain Taking Opioid Agonist Therapy: A Factorial Experiment Using the Multiphase Optimization Strategy Framework

Abstract

Background:

Chronic pain affects up to half of individuals taking opioid agonist therapy (OAT; i.e., methadone and buprenorphine) for opioid use disorder (OUD), and yoga-based interventions may be useful for decreasing pain-related disability. Whereas more yoga practice (i.e., higher “dosage”) may improve pain-related outcomes, it can be challenging for people with chronic pain taking OAT to attend class regularly and sustain a regular personal yoga practice. Therefore, we plan to optimize a yoga-based intervention (YBI) package in order to support class attendance and personal practice, thus maximizing the yoga dose received.

Study Design:

Using the Multiphase Optimization Strategy (MOST) framework, we will conduct a factorial experiment to examine four intervention components that may be added to a weekly yoga class as part of a YBI. Components include: 1) personal practice videos featuring study yoga teachers, 2) two private sessions with a yoga teacher, 3) daily text messages to inspire personal practice, and 4) monetary incentives for class attendance. The primary outcome will be minutes per week engaged in yoga (including class attendance and personal practice). We plan to enroll 192 adults with chronic pain who are taking OAT for OUD in this 2×2×2×2 factorial experiment.

Conclusion:

Results of the study will guide development of an optimized yoga-based intervention package that maximizes dosage of yoga received. The final treatment package can be tested in a multisite efficacy trial of yoga to reduce pain interference in daily functioning in people with chronic pain who are taking OAT.

Chronic pain is a significant problem for 40–50% of all persons taking opioid agonist therapy (OAT) for opioid use disorder (OUD), including buprenorphine or methadone maintenance treatment [1–3]. In people with OUD taking OAT, chronic pain is associated with disability [1], health problems and psychiatric disorders [1,4], including depression and anxiety [2,5], increased misuse of opioids or other non-prescribed or illicit drugs [6,7], and increased suicide ideation [8]. Further, there are many challenges to treating chronic pain in people with OUD including the possibility of opioid-induced hyperalgesia [9]; increased tolerance of opioids; use of benzodiazepines, alcohol or marijuana to mitigate pain; and patients’ fears about medications and addiction. Providers may also have difficulty resolving the seeming inconsistencies between a pain management approach and an addiction management approach [10]. Additional systematic barriers to pain management include lack of comprehensive pain management programs [11], low reimbursement from Medicaid for non-pharmacological pain treatments [12], and lack of attention to pain management in opioid use disorder treatment programs [13]. As such, there is an urgent need for innovative approaches to manage pain this population so that people with OUD have more options.

Hatha yoga is an ancient Indian system of philosophy and practice [14]. Modern variants often practiced in the U.S. (in this paper, termed “yoga”) commonly involve physical postures (āsanas), breath control (prāṇāyāma), and meditative practices. A hatha yoga-based intervention (YBI) may be a useful adjunctive approach for decreasing pain-related interference in daily functioning and pain severity and preventing opioid misuse in people taking OAT. Yoga is an effective treatment for chronic pain conditions [15–17], and may target cravings [18–20] and other risk factors for misuse of opioids [21–23]. A YBI could also help overcome some of the barriers to chronic pain treatment among people with OAT. YBI providers can be trained to work with people who have chronic pain conditions and be sensitive to the intersecting chronic pain, OUD, OAT, and other stigmas this population faces. Yoga may also be a more accessible pain management approach when surgical treatment is inaccessible – and a YBI may also be a cost-effective treatment option. Indeed, growing research shows yoga to be cost-effective for chronic pain and musculoskeletal concerns [24–26], with recent research suggesting that yoga relative to physical therapy may produce similar results at greatly reduced cost [27]. Widespread access to in-person yoga is currently limited; yoga delivered virtually can help overcome this barrier.

Although there is increasing interest in the use of YBI in substance use disorder treatment [28,29], there are few studies of YBIs for people with substance use disorders [30–32] and fewer still conducted with people receiving OAT in any setting [33–35]. To our knowledge, we conducted the only study of a YBI for people with chronic pain and OUD taking OAT [35]. In this pilot study (n=40), we demonstrated feasibility, acceptability, instructor fidelity, and participant retention in a 12-week manualized YBI (vs. a 12-week health education program). The primary challenge was in class attendance; this is a common concern with behavioral interventions for people taking OAT who may face numerous barriers to attending group trainings regularly [36–38]. However, previous research has shown that more yoga practice is associated with better pain-related outcomes [35,39]. As such, future studies should maximize the “dosage” of yoga that people taking OAT receive; using an optimized intervention will allow a robust test of the hypothesis that a YBI is effective (i.e., reduces pain interference, improves other pain and substance use outcomes).

This paper provides an overview of the development of a YBI, Yoga MAT (Medication for Addiction Treatment), aimed at optimizing yoga “dose” (i.e., minutes per week of yoga practice, including time in class + personal practice time). The study design is the Multiphase Optimization Strategy (MOST). MOST [40] is an engineering-inspired framework for intervention optimization, i.e., identifying a multicomponent intervention that produces the best expected outcome obtainable within key constraints. We plan to identify components of a YBI package that, when added to a weekly yoga class, maximize yoga dosage received. Because adding components to a standard yoga class necessitates time and financial resources, it is essential to develop the most efficient, economical, and scalable intervention package for maximizing dosage received.

The MOST framework comprises three phases. In the Preparation Phase, investigators lay groundwork for optimization by developing a conceptual model, identifying intervention components thought to have an impact on the outcome of interest (in this case, yoga dosage received), conducting pilot studies, and choosing an optimization criterion. In this study, our criterion will include, in the ultimate package, all active components that produce a positive effect on increasing yoga dosage received. In the Optimization Phase, investigators conduct an optimization trial, using an efficient experimental design, such as a factorial experiment, to identify the best combination of intervention components that meets their optimization criterion. In this phase, investigators also test potential mediators by which intervention components have an impact on the outcome of interest. Finally, In the Evaluation Phase, investigators use a randomized controlled trial (RCT) design to study efficacy of the optimized intervention on a primary outcome such as pain-related disability.

This paper describes the design of an Optimization Phase study. In a 2×2×2×2 factorial design, we plan to test four experimental intervention components as potential adjuncts to a weekly synchronous, online yoga class. Each participant will be randomly assigned to receive or not receive each of four intervention components, including: a) access to study teacher-led personal practice videos, b) two private yoga sessions with the yoga teacher, c) text messages designed to cue and increase intrinsic motivation for personal practice, d) monetary incentives for class attendance. Results of this study will be used to design an optimized YBI that can be evaluated for efficacy in decreasing pain-related interference in daily functioning and improving other pain-related and substance use-related outcomes in a future RCT.

Methods

Study Design

Consistent with the MOST Optimization Phase, we will conduct a fully-powered (n=192) factorial experiment to elucidate whether any of the identified intervention components result in increased yoga dosage. Participants will be patients recruited from substance use disorder clinics (including in primary care) who are taking methadone or buprenorphine and who experience chronic pain. All participants will receive the core yoga intervention (12 weeks of weekly manualized yoga classes), with random assignment to receive or not receive each of the four experimental intervention components, (yielding 16 possible combinations).

We aim to recruit participants in three regions, two in the Northeastern U.S and one in Ohio. To determine whether the experimental intervention components have an impact on theorized mechanisms of action and in turn, yoga dosage, we also plan mediation analyses.

Interventions

Core Yoga-Based Intervention

All participants will receive weekly group yoga classes conducted via Zoom videoconference and will be asked to engage in personal practice. See Table 1 for further detail.

Table 1.

Core yoga-based intervention, described per to the CLARIFY (CheckList stAndardizing the Reporting of Interventions for Yoga) guidelines [41].

Activities	Each class consists of greeting and discussion of personal practice; centering body and breath awareness; warm up movements; pranayama (breathing practices); an asana sequence; a final seated meditation, breathing practice, and discussion of personal practice. Instructors will acknowledge the validity of participants’ pain and gently challenge them to try a new way of coping with it; emphasize breathing in every part of the class; and offer gentle postures.
Expertise	All instructors will have met criteria to be Registered Yoga Teachers at the Yoga Alliance, a professional organization that sets standards for yoga teacher training, and at least 2 years of prior teaching experience.
Delivery and Training	Although the study was originally designed to be in-person, procedures were changed to synchronous remote classes due to the COVID-19 pandemic. All participants will be provided a cell phone equipped with unlimited data plans for the 12 weeks that they are taking classes. In the event of lost or malfunctioning phones, replacement phones will be provided. Class size will range between 1 and 8 participants. There are two situations when a 1-person class might occur: 1) at study start-up, when it takes time to consent people and have them go through the research procedures necessary before the start of class; and 2) during the trial, when a number of participants plan to attend but only actually attends. We expect that the vast majority of classes will have more than one participant, and if a given class time persistently has only one person attending, we would cancel or change that class time. Yoga teachers will have study-specific training on research methods, OAT, chronic pain, adverse events, and the yoga manual comprising approximately 5–10 hours of didactics, observing at least one class, and having their first class reviewed by a fidelity rater. More guidance is available if needed. They will meet monthly for group supervision.
Dose and Personal Practice	Participants will be asked to attend one 60-minute class per week for 12 weeks. They will also be asked to engage in personal practice.
Duration/Frequency of Personal Practice	Participants are encouraged to practice up to 30 minutes per day, 5 days per week, with self-compassion encouraged when they do not engage in personal practice. We will give all participants a yoga mat, a curated list of YouTube videos, and written instructions for personal yoga practices.
Participant Adherence	Total number of minutes per week attending a study yoga class (documented by study staff) and in personal practice or outside yoga classes (using weekly yoga journals) will be used to assess participant yoga dosage. To encourage class adherence, all participants will receive a weekly reminder to attend class via phone call or text. This reminder is distinct from the intervention component comprised of daily inspirational text-based cues for personal practice.
Instructor Fidelity	Audio or video recordings of the first class will be reviewed plus an additional random 10% of all other classes to assess instructor intervention adherence. Results will be reviewed with teachers.

Rationale and Description of Intervention Components

Participants will be randomized to either receive or not receive the following four intervention components. Please see Figure 1 for a list of intervention components and hypothesized mechanisms by which they impact yoga dosage received.

Figure 1.

Conceptual model

Note. Although we will assess distal outcomes in the current project in order to document feasibility of capturing these outcomes, the primary analyses are focused on the proximal outcome of yoga dosage received.

Intervention Component 1: Personal practice Videos Featuring Study Yoga Teachers.

Social Cognitive Theory posits that one important influence on behavior (e.g., class attendance or personal practice) is self-efficacy (belief that one can do a behavior) [42]. Higher self-efficacy predicts initiating and maintaining types of physical activity [43–45]. The addition of study-specific yoga videos of varying lengths featuring study teachers and focusing on yoga practices taught in class may increase self-efficacy for personal practice more than the non-specific (YouTube) videos. Study-specific videos could also increase self-efficacy for practice in class (potentially increasing class attendance), as participants can “rehearse” outside of class and build confidence.

Intervention Component 2: Two private sessions with a yoga teacher.

Research with people with chronic pain [46,47] and associated literature [48] suggest that the interrelated fears of movement, injury, and pain may be a barrier to engaging in yoga or other physical activity. Personalized attention from a yoga teacher may potentially decrease this fear [46] while offering other benefits, reflected in the parent tradition of yoga that emphasized the student-teacher relationship [14]. In private sessions, the yoga teacher can address the individual’s specific physical concerns and provide individualized advice. Content will focus on a participant’s specific physical concerns and needs and will primarily involve discussion rather than an extended guided yoga practice (as in yoga classes). We hypothesize that this intervention component will lead to decreased fear of movement or injury.

Intervention Component 3: Text messages cuing personal practice.

According to Self Determination Theory, in order for people to engage in a behavior such as yoga, motivation needs to be present [49]. Motivation occurs on a continuum of increasing “self-determination” that is in turn associated with more long-term engagement in physical activity [49–51]. Text messages may be useful to cue participants to engage in yoga personal practice. Text message content will focus on self-determined motivation for personal practice by emphasizing personal choice and encouraging engagement in practices that are enjoyable or create a sense of competence.

Intervention Component 4: Monetary incentives for class attendance.

Based on operant conditioning, financial incentives can increase the likelihood of a person engaging in a behavior. They have been widely used as part of contingency management strategies in the substance use treatment literature [52,53]. A burgeoning literature suggests that they can also increase physical activity [54,55]. Financial incentives can be effective in promoting habit formation for people who are new to a behavior such as attending yoga classes [56], with the hope that internal motivation increases as the behavior is repeated. Participants assigned to this intervention component will receive $15 per class attended. We chose to provide a flat amount per class attended for the sake of simplicity, as we want the ultimate intervention package, if effective, to be easy to administrate in community settings. We hypothesize that this direct reinforcement will increase the number of classes attended without decreasing amount of personal practice, thus increasing amount of yoga dosage.

Recruitment and Informed Consent

Adults with chronic pain taking OAT will be recruited from three regions in the U.S. via screening of medical records by research staff at some sites, referrals from clinicians, and advertisements and flyers at substance use disorder clinics and office-based addiction treatment clinics. Participants will be asked screening questions, and, if they appear eligible, scheduled for a baseline visit. After informed consent, confirmation of study eligibility, and a baseline assessment, a study staff member who is not blinded to intervention component assignment will randomize the participation.

Eligibility Criteria

Participants must meet all eligibility criteria at initial screening and baseline to participate in the study. Inclusion criteria will be: ≥3 month continuous methadone or buprenorphine treatment plan to continue treatment for next 6 months; chronic pain that persists more than half of the last 90 days [57]; average pain interference score ≥4 on Brief Pain Inventory (BPI, 0–10) [58]; average pain severity of ≥4 on a Visual Analog Scale (VAS, 0–10) indicating “worst pain in the last week” [59,60]; age ≥18; proficiency in English sufficient to engage in informed consent in English, understand classes taught in English, and read short sentences; and available at least one of the three times study classes are offered.

Exclusion criteria will be: currently taking yoga classes or practicing yoga on their own; medical conditions that would make participation in yoga unsafe or not possible, including active malignancy treatment, fracture, recent joint surgery, or use of assistive ambulatory devices other than a cane; severe or progressive neurologic deficits; surgery requiring overnight hospitalization planned in the next 3 months; pregnancy; and a plan to move out of the area within 6 months (because of the high potential for disruption in OUD care). Finally, we will exclude people with other severe, disabling, chronic medical and/or psychiatric comorbidities deemed by the site Investigator on a case-by-case basis to prevent safe or adequate participation in the study. To make this determination, the site investigator will consider whether the person can participate in class safely and is unlikely to be disruptive to others in class. They will also consider whether the yoga teachers feel that they have the necessary skill to teach someone with particular limitations.

Note that we will include patients taking OAT who are and are not participating in a licensed opioid treatment program. Although this may introduce heterogeneity, it also increases generalizability. However, we will not include patients on XR-naltrexone which is not an opioid agonist and should have no effect on pain levels.

Once randomized, participants will be allowed to continue in the study even if they discontinue OAT, unless participation is deemed unsafe in some way.

Randomization

At the randomization visit, participants will be randomly assigned to one of 16 study conditions by study staff. The study statistician will create randomization tables and upload to the data collection system (REDCap) prior to the start of recruitment. There will be no stratification variables. Block size is effectively 12 (i.e., number of participants per cell). Study staff will not have access to randomization tables and allocation will be concealed prior to randomization. After randomization, participants will be considered enrolled.

Community Engaged Research Component

When we began this line of research, we conducted focus groups with people taking OAT to gain their feedback for intervention development. Subsequently, when piloting the yoga-based intervention, we conducted qualitative interviews to solicit feedback and used that feedback to further adapt the YBI (e.g., see Table 4 in [35] for some of the participant feedback). For the proposed project, we have also solicited feedback from patients on the content of text messages and used that feedback to modify planned messaging. Finally, for this study, we plan to recruit two participant-stakeholders per year to serve as consultants and provide feedback on study methods and results.

Measures

Assessment of personal yoga practice (i.e., weekly yoga journals) will be collected weekly during the 12-week intervention. All other assessments will be collected at baseline with follow-ups administered at 1, 2, and 3 months (end of intervention) as well as 6, 9 (yoga journals only), and 12 months. Staff who conduct follow-up assessments will be blinded to intervention component assignment. Participants may receive a total of up to $240 for assessments. Those randomized to receive payments for attending yoga classes will receive an additional $15 for each yoga class attended, for a possible total of $180 for 12 classes.

Primary Outcome.

Yoga dosage received will be calculated by adding total number of minutes per week in the study yoga class and total number of minutes per week in personal yoga practice (or outside yoga classes following study enrollment). Research staff will take attendance at study yoga classes. Personal practice will be measured using a weekly yoga journal, which has been shown to have good correspondence to a daily yoga log [61]. Staff will administer this measure via telephone or videoconference interview. Although they will attempt to administer this weekly, if one or more weeks are missed, using a Time-Line Follow-Back method [62,63], staff will be able to collect data for missing weeks. Prior research has shown a correlation between amount of yoga practice assessed by self-report, and improvement in pain-related outcomes [35]. Amount of practice will be assessed weekly in Months 0–3, and then again at M6, M9, and M12.

Mediators of intervention components on yoga dosage.

The Tampa Scale of Kinesiophobia (TSK), 11-item version, will be used to assess fear of movement or injury [64–66]. Self-efficacy for yoga practice will be assessed with a 5-item scale for exercise self-efficacy, adapted to refer to yoga [67]. The Behavioral Regulation in Exercise Questionnaire (BREQ-2), adapted to refer specifically to yoga, will assess self-determined motivation [68]. These scales will be assessed at baseline, and at months 1, 2, and 3.

Distal outcomes.

Distal outcome data will be collected to demonstrate feasibility of data collection for future research. All distal outcomes will be assessed at baseline and at months 1, 2, 3, 6, and 12. The Brief Pain Inventory - Pain Interference Subscale will be used to measure pain-related interference in daily functioning [69]. The pain interference subscale includes items assessing psychosocial/affective functioning (relations with others, enjoyment of life, mood) and physical functioning (walking, general activity, work). A 0–10 Numerical Rating Scale (NRS) will be used to assess average and worst pain severity in the previous week [59,70]. Participants will also complete the WHO Quality of Life (WHOQOL BREF), a valid and reliable 26-item self-report measure of quality of life [71]. These domains are all recommended as outcomes for chronic pain treatment trials [70].

Opioid use and use of illicit drugs will be assessed multiple ways. The Addiction Severity Index (ASI) will be used to assess days of methadone or buprenorphine use, alcohol use, and illicit drug use [72]. Retention in OAT will be measured via OAT site chart review (when available to researchers) and self-report. Finally, the McHugh Opioid Craving Scale will be used to assess cravings [73].

Strategies for retention for assessments.

We plan to use multiple strategies to retain participants for assessments, including provision of study cell-phones, payment for follow-up assessments, multiple outreaches to a participant via multiple methods, consultation with the referring clinic whenever possible to identify new contact information, and request for contact information of a person who can always help to locate the participant in the future.

Analytic Plan, Sample Size, and Power Estimates

Primary outcome.

Collins provides detailed recommendations for conducting a factorial experiment in the context of MOST [40]. The primary goal of this analysis is to determine which components will be included in the final, optimized intervention package. We will assess 4 candidate intervention components for possible inclusion in an optimized intervention. To do so, we will use a balanced factorial ANOVA. Each component will have two conditions producing a 2⁴ = 16 cell design with 4 main effects, 6 possible 2-way interactions, 3 possible 3-way interactions, and 1 possible 4-way interaction (see Table 1). Note that specific cells are not compared against other specific cells.

We will use a general linear model with each candidate component coded using effect (1 vs −1) coding. Recruitment site, sex, age, and OAT type will be covariates. If changes in OAT use are common, we will include a time-varying indicator of change in OAT use between assessments as an additional covariate in the analytical models. The pragmatic exigencies of field research will likely result in imperfect balance (e.g., individual cells may have slightly varying n). However, we anticipate only very small variations in cell sizes and thus estimated model coefficients will be nearly, but not perfectly, uncorrelated.

Missing data.

Number of minutes in a study yoga class will be assessed via observation (and is perfectly measured – i.e., without any missing data). Via weekly assessments [61], we will ask about any community (including online) yoga classes attended following study enrollment and yoga personal practice. Research staff will make every effort to ensure this primary outcome is assessed even if other outcomes are not assessed. To estimate missing data, we will use multiple imputation by chained equations [74] to generate 50 fully populated data sets. Variables in the imputation model will include effect coded indicators for the four evaluated intervention components and all observed data regarding yoga practice. It is anticipated that imputation will be feasible even in the event of higher attrition rates (e.g., 30–50% attrition), although clearly less missing data is always preferable and study staff will make every effort to retain participants in the study longer-term even if they stop taking OAT. We will also conduct sensitivity analyses post-hoc where we assume that if data are missing, the participant did not engage in any yoga practice.

Statistical power.

In RCTs a conclusion is made regarding the effect of an intervention based on rejecting the null hypothesis with Type I error < some predefined α. In contrast, in an optimization trial, a decision is made to include or exclude specific components in an intervention that will be tested in a future RCT. In this scenario, Type II error may be as pernicious as Type I error. Therefore, based on recommendations of Collins [40] this study will be powered with α/2 = .10. Because intervention effects are uncorrelated (or nearly uncorrelated), the alpha will not be corrected for multiple comparisons and all effects have equal statistical power. Assuming 17% attrition for our primary outcome of dosage (i.e., almost double the attrition seen in our pilot work) [35], we estimated power for n= 159 (192 recruited - 33 lost to attrition). To estimate the minimum detectable effect size we assumed a small correlation (r = .15) between recruitment site and yoga attendance and used the FactorialPowerPlan [75] macro in R. The proposed design has sufficient power (1 – β > .80) to detect a standardized difference in means of .40 or larger. Power to detect a medium standardized effect (d = .50) is > .90.

Optimization criteria.

We established optimization criteria for the project based on the recommendations of Collins [40]. In this study, our optimization criterion is to include, in the ultimate package, all active components that produce a positive effect on increasing yoga dosage received. Given that the four components we are testing are relatively inexpensive, we chose not to include any constraints based on cost or other factors. When deciding to include or exclude components in an optimized intervention, primacy is given to the main effects of components but we will also consider higher order interactions. Interactions may be synergistic wherein which the combination of 2 or more components is more beneficial than would be expected by an examination of the main effects only. In this case, one might include a component even though the main effect is relatively small. Or, interactions may be antagonistic where the combination of two or more components is less effective than would be expected based on the main effects alone. In this case, one might exclude a component from the optimized intervention even though the main effect appears strong.

Mediation analysis.

We will use the bias corrected and accelerated bootstrap (10,000 replications) as implemented in Mplus statistical software [76] to estimate the indirect effects of the intervention components on yoga dosage received via hypothesized mediators (see Figure 1). Additional exploratory analyses will estimate the indirect effects of intervention components via other intermediate variables (i.e., those variables hypothesized to mediate the effect of other components). This will provide valuable insight into the mechanisms through which intervention components impact total yoga practice time.

Discussion

Up to half of all persons taking OAT for opioid use disorder experience chronic pain [1,2,77]. Despite promising evidence implicating yoga as effective in other populations of people with chronic pain [15–17,78] as well as for the reduction of cravings [18] and other opioid relapse risk factors [21–23], attendance rates have been suboptimal, a serious concern as yoga “dosage” has been associated with improved pain [35,39]. Optimization of a yoga-based intervention to maximize yoga “dosage” received will thus ensure an adequate test of the hypothesis that yoga is efficacious, i.e., reduces pain-related interference in daily functioning, and improves other pain and substance use outcomes with significant potential for public health.

The Yoga MAT study is one of the first trials to use the MOST framework to optimize the development of a yoga-based intervention. Factorial experiments bypass some of the drawbacks of RCTs, which involve substantially greater resources and sample sizes to execute. It is also not possible with a traditional RCT to test a multicomponent treatment to elucidate the specific individual and synergistic effects of intervention components. The present study will address these limitations and result in an optimized yoga treatment package that best supports participant engagement in yoga practice – and can then be tested in a future traditional RCT (i.e., Phase 3 of the MOST framework).

In addition, the current study will allow us to assess potential mechanisms that may underlie the effects of intervention components on yoga dosage. This will provide information regarding not only whether the selected intervention components act on the outcome (i.e., the direct effect of each intervention component), but also whether this action occurs via the theorized mechanisms of action (i.e., whether the component acts on the hypothesized mediator, and whether the hypothesized mediator is associated with the outcome). For intervention components that have no direct effect on the outcome, mediation analysis may provide insight into why this was the case.

Limitations

One limitation of this project is that data is collected by self-report by participants, captured via weekly phone interviews or online surveys. Participants are given a daily calendar to record their practice, but not everyone will use this accurately and often. A related limitation may be the ability to consistently get participants on the phone on a regular basis to record the information. There are many psychosocial stressors and structural factors that make follow up challenging for some people with chronic pain and OUD, and treatment attrition rates can be high. The more time that passes between assessments of yoga practice, the higher the likelihood that the participant is inaccurately reporting their yoga practice time. A final limitation is our reliance on the use of self-report for OAT adherence. However, this is not the primary outcome, and confirmation options (e.g., urine drug screens) are not feasible given the lack of in-person contact.

Conclusions

This study addresses an urgent need for timely and accurate information on optimal approaches to reduce pain-related interference in daily functioning and other pain and substance use disorder outcomes in people taking OAT with chronic pain. Results of the study will guide development of an optimized yoga-based intervention package that yields the maximized dose of yoga with the fewest intervention components. The final treatment package can be tested in a multisite efficacy trial. Although yoga classes are widely available in many parts of the U.S., gentle yoga classes for people with chronic pain and opioid use disorder, and who are potentially low-income and naïve to yoga, are not widely available. Because this intervention can be delivered remotely (i.e., via synchronous yoga classes and other components), this could allow for widespread reach to people across the United States and worldwide.

Table 2.

Balanced Factorial Design Assessing Four Candidate Intervention Components

	COMPONENT
Cell	1	2	3	4	n
1	No	No	No	No	11
2	No	No	No	Yes	11
3	No	No	Yes	No	11
4	No	No	Yes	Yes	11
5	No	Yes	No	No	11
6	No	Yes	No	Yes	11
7	No	Yes	Yes	No	11
8	No	Yes	Yes	Yes	11
9	Yes	No	No	No	11
10	Yes	No	No	Yes	11
11	Yes	No	Yes	No	11
12	Yes	No	Yes	Yes	11
13	Yes	Yes	No	No	11
14	Yes	Yes	No	Yes	11
15	Yes	Yes	Yes	No	11
16	Yes	Yes	Yes	Yes	11