Design and setting

This was a quasi-experimental pilot study using a one-group pretest-posttest design for feasibility testing. Recruitment was restricted to U.S. residents of Washington State due to practitioners’ licensing restrictions. A non-randomized convenience sample was selected of adults with a chronic pain diagnosis who responded to online advertisements or invitations from staff at partnering clinical sites between October 2020 and October 2021. Recruitment also occurred via e-mail messages, university distribution lists, posting flyers, and local news posts.

Inclusion/Exclusion criteria

Eligible participants were at or above 18 years of age with a chronic (non-cancer) pain diagnosis and reported experiencing at least a moderate level of pain intensity considered a minimum of a “5” on a zero to 10 visual analog scale. Participants were also required to have been prescribed an opioid-based medication to help manage their pain. Any individuals who reported having medical and/or psychiatric conditions that may require hospitalization or frequent medication adjustments were excluded, as were any females who were pregnant or expected to become pregnant during the study period. Finally, due to the digital nature of the project, participants were required to have access to a computer or device with internet access for study activities.

Intervention and procedures

After a screening process with a study team member via phone or digital video conference was completed, eligible participants were sent an electronic consent (eConsent) to be completed within the university’s Research Electronic Data Capture (REDCap) instance [17]. Once the completed eConsent was submitted by the participant, they were sent a link to complete a baseline survey via REDCap.

Participants were then scheduled to complete an initial pain assessment that was conducted via a cloud-based videoconferencing platform by a registered nurse using the McCaffery Pain Assessment Tool [18]. Participants were asked to provide a current list of medications including name, dose, and directions. This meeting was also used to set up participants’ videoconferencing accounts and work out any technology challenges. After completion of the assessment, participants were asked to select two of the four modalities for their six consultation sessions, choosing from yoga, massage, chiropractic, or physical therapy. Their preference was recognized to align with person-centered care. The results of the pain assessment were shared with the practitioners prior to their first session using a shared cloud-based project folder. Participants and practitioners worked with the research team to schedule three 45-minute videoconference sessions once weekly for three weeks at times that were convenient for both. After the day and time of the first appointment was agreed upon, researchers would create a private meeting within the university’s Health Insurance Portability and Accountability Act of 1996 (HIPAA)-compliant videoconference platform. During the third week, three 45-minute sessions were scheduled for the participant and practitioner for the second selected modality, allowing for a total of six sessions per participant.

All practitioners were certified or licensed in their specialty, met state regulations for practice, and held liability insurance as required by the university leading the study. Practitioners were given minimal guidance on how to conduct their pain-focused interventions, and sessions were unmonitored. Most had at least some experience delivering an online consultation and met with the principal investigator to review the research and privacy protocols prior to their first session. They were directed to use their clinical judgement during their allotted time with participants doing activities they would normally do in person to the best of their ability, such as asking participants to perform movements to evaluate mobility, posture and gait. Practitioners were encouraged to use their cameras to demonstrate interventions and observe participant responses. They were asked to keep brief clinical notes that could be accessed by the other practitioners before scheduled sessions and the research team to document progress. Fidelity testing was not conducted based on the pilot nature of the project and the expectation that variations could be necessary to adapt a novel intervention.

Instruments and data collection

At baseline, demographic data were collected using REDCap surveys along with valid and reliable measurements common to interventional pain research that were repeated post-intervention at three and six weeks [19]. Primary outcomes included the Numeric Pain Rating Scale as a well-tested 0 to 10 measurement of current pain intensity [20], and the Patient-Reported Outcomes Measurement Information System (PROMIS) pain intensity Short Form 3a [21] and pain interference Short Form 8a [22] to measure pain in the past seven days. Mood was measured with the eight-item Patient Health Questionnaire (PHQ-8) for depressive symptoms [23] and the Generalized Anxiety Disorder-7 (GAD-7) for anxiety symptoms [24]. For all scales, higher values indicate higher symptom burdens.

Secondary exploratory outcomes were captured at baseline and after six weeks in order to detect whether any improvements could be expected. Measures included: PROMIS self-efficacy for managing symptoms and emotions (Short Forms 4a) to measure participants’ self-rated confidence [25]; the nine thwarted belongingness items from the Interpersonal Needs Questionnaire (INQ) used to evaluate relationships and belonging [26]; the Four-Item Spiritual Well-Being Index (4-ISWBI) to capture spiritual quality of life [27]; and the Current Opioid Misuse Measure (COMM) to capture self-reported misuse of opioids [28].

Finally, treatment satisfaction was measured with a post-test survey adapted from the IBM computer usability and satisfaction questionnaires that asked about how easy, convenient, and comfortable the treatment sessions were, with Likert-style items such as “Overall, I am satisfied with how easy it was to participate in this treatment” (1 = strongly disagree; 7 = strongly agree) [29].

After completion of the study protocol, participants were asked to complete a follow-up interview with a member of the study team via videoconferencing that focused on experiences throughout the active study period. Results of these interviews have been previously published [30]. Figure 1 provides a timeline of all study activities over the six-plus-week study. All data were collected between January 2021 and February 2022. Participants were provided a USD 75 gift card payment when all study tasks were completed to compensate for time completing study activities.

Display full size

Figure 1. Timeline of study activities.

Data analysis

De-identified exports of data collected via REDCap were used in the analysis process. Data spreadsheets were used in both Microsoft Excel and SPSS^® version 27 to complete the analysis. Counts and frequency statistics were used to describe categorical variables, while mean and standard deviation (SD) were used for numeric variables. All analyses were considered exploratory due to the pilot design with no hypothesis testing. Protocol feasibility was assessed with regard to participant recruitment and retention data, while acceptability was measured in session attendance and self-reported treatment satisfaction. PROMIS measures (pain intensity, pain interference, and self-efficacy for managing symptoms, and self-efficacy for managing emotions) were calculated by converting the sum of the ratings for each scale into a T-score, where a T-score of 50 is equivalent to the U.S. general population average and +/– 10 points is equivalent to the SD [22]. For the remaining measurements, total scores were summed. After confirmation of normality via skewness and kurtosis, to evaluate post-treatment effects, continuous outcomes were assessed using paired t-tests in a pre (baseline)/post (week 6)-analysis with significance level set at .05. Effect sizes were calculated with Cohen’s d and categorized [31]. Cronbach’s alpha was calculated as a measure of internal reliability. Categorical outcomes were assessed via contingency tables and Fisher’s exact test due to the small sample size. Missing data were accounted for using last observation carried forward approach. The Consolidated Standards of Reporting Trials (CONSORT) extension to pilot and feasibility trials method was followed for reporting [32].

Ethical considerations

A protocol for this pilot study was registered a priori through Clinical Trials Network (#NCT05098158). Washington State University Institutional Review Board (IRB) approval was secured from the principal investigator’s university (IRB #18111) on 7/10/2020. Good clinical and research practices, including those in the Declaration of Helsinki and the Belmont Report, were followed throughout the study period including all requirements within the HIPAA. All participants provided written informed consent via a secure online platform.

Sample characteristics

The average age of study participants (n = 16) was 60.4 years of age (SD: 10.4; low: 38; high: 75) and was 75% female. All participants, except one, described their race as White and ethnicity as non-Hispanic/Latino(a) (93.7%). Marital status was evenly split and half of the study’s participants reported having a graduate degree. The majority (87.5%) of participants were no longer working with federal health insurance (Medicare) being the primary insurance for over half of the participants (56.2%).

At baseline, all participants, except one (weekly pain), noted they had daily pain, with the average numeric pain scale score of 6.1 (SD: 1.7; range: 6; minimum: 4; maximum: 10). PROMIS pain intensity (mean: 68.3; SD: 6.3) and pain interference (mean: 69.9; SD: 5.5) T-Scores were well above the average adult. The average participant had 4.5 (SD: 1.8) painful medical conditions, with over half reporting five or more (56.2%). All participants (100%) reported back pain, with three-quarters (75%) also reporting arthritis and nerve pain and over half reporting neck pain (62.5%). Additional painful medical conditions included headaches and migraines (50%), post-surgical pain (31.3%), fibromyalgia (18.8%), and complex regional pain syndrome (CRPS)/reflex sympathetic dystrophy (RSD) syndrome (12.5%). One-quarter (25%) of participants also noted “Other” conditions.

Nearly half (43.8%) of the study sample reported having a concurrent mental health disorder, with one participant reporting three diagnoses, and the remaining six participants listing one each. The most common diagnosis was major depressive disorder (31.3%), followed by generalized anxiety disorder (18.8%) and bipolar (6.3%). Full participant characteristics can be reviewed via Table 1.

Pilot feasibility

Over a 12-month course of recruitment taking place during the COVID-19 (SARS-CoV-2) pandemic, 73 individuals were referred to the team, by themselves or study partners, for possible inclusion. The last participant completed the final interview in February 2022. A total of 36 individuals were excluded prior to screening for eligibility due to unsuccessful outreach (n = 22), not being interested (n = 7), disclosure of not being on an opioid-based medication before eligibility screening (n = 6) or living outside of state (n = 1). The remaining 37 individuals were directly assessed for eligibility (Figure 2) with six not meeting study requirements due to not being on an opioid-based medication. Of the 31 individuals sent the consent document, 21 submitted a completed form to the study team via REDCap. Finally, of the 16 participants beginning appointments, 15 completed all study requirements.

Display full size

Figure 2. Consolidated Standards of Reporting Trials (CONSORT) flowchart of study participants.

Treatment acceptability

Of the expected 96 total provider-led sessions, participants completed 92 (96% attended). One yoga therapy appointment and three physical therapy appointments were missed. The majority of responses to the post-treatment satisfaction survey showed some level of agreement with the ease of participation in the COMFORT study treatments with “strongly agree” being the most common response (57.1%) followed by “agree” (35.7%) and “somewhat agree” (7.1%). The same response frequency was found when participants were asked about the time commitment required to participate in the treatments. When asked about satisfaction with the support available, two-thirds of participants “strongly agree” followed by “agree” (26.7%) and “somewhat agree” (6.7%). Finally, when asked about their comfort while participating in the study, 80% strongly agreed, with 13.3% agreeing and 6.7% somewhat agreeing. When asked “Did you find anything about this treatment especially useful?” nearly all participants (86.7%) described learning at least one especially useful concept during the study period. Yoga therapy and physical therapy exercises, techniques, and providers were most commonly mentioned as useful. When asked “Is there anything you would change about this treatment if you could?”, six participants (40%) noted potential changes that could be made for future studies. The most commonly reported change was to hold all appointments in person, but other changes included allowing participants to cycle through all four modalities rather than selecting two, providing additional appointments, and improving the scheduling procedures.

Preferred non-pharmacological modalities

Overall, the most commonly selected modality was yoga therapy (34.4%) followed by physical therapy (28.1%), massage therapy (25%), and chiropractic therapy (12.5%). Table 2 describes the full breakdown of modality selections.

Adverse events

During the study, there was one adverse event that resolved without sequelae. Following completion of a yoga therapy session which included lateral head and neck tilts, the participant noted feeling “queasy.” The primary investigator contacted the participant, who noted they had previous issues with vertigo. The participant continued in the study with no additional problems noted.

Post-treatment effects

All of the primary outcome measurements’ mean values showed numeric improvement across the six-week study period (Table 3). The PROMIS pain intensity and pain interference showed significant reductions. Affective symptoms of depression, assessed by PHQ-8, also showed significant reductions. Also, participants’ feelings of thwarted belongingness significantly decreased over the study period demonstrating improvement. Multiple outcomes showed positive changes that did not reach the level of significance, as indicated by Cohen’s d, such as numeric pain scale score and PROMIS self-efficacy for managing symptoms.

Participants in this study had a low level of anxiety, indicated by the mean GAD-7 score in the “mild” range for both baseline (pre) and week 6 (post). Also, spiritual well-being scores were high on average, with insignificant change over the study. Mean COMM scores remained fairly constant from baseline to week 6, with an insignificant 0.3 change in scores. At study entry, six participants (37.5%) were above the cutoff score (≥ 9), signifying a “positive” screening for medication misuse behaviors; this reduced to five individuals (31.3%) at the end of the study.

Limitations

This was a brief six-week pilot study with a small number of participants completing study activities from a single state in the U.S. They may not have represented a general population prescribed opioids for pain, as internet service was required, and most were White, had a college degree, endorsed multiple pain conditions, and were not presently employed. The lack of a controlled design prohibits attribution of positive effects to the intervention versus other pain management interventions. Fidelity was not assured, and not all participants tested all modalities, and the order of exposures varied. A few sessions were completed through voice alone because video was unavailable. We had minimal missing data from only one participant; however, the last observation carried forward method could have had some impact on the accuracy of evaluative measures. Practitioners were unable to use manual therapy techniques or assessments that they would be able to use in person, which may have reduced the evaluation quality and potential benefits. Professional licensing restrictions disallowed telehealth to be conducted outside of any state other than the one where practitioners were licensed, reducing the geographic diversity of participants.

Selection bias and regression to the mean can be addressed in future studies with a control group; however, a double-blinded study design would not be possible given the nature of the intervention. Therefore, future studies may choose to compare telehealth with in-person sessions or against other well-studied non-pharmacologic pain interventions for comparative effectiveness. Studies will require a well-developed intervention with fidelity testing to examine outcomes between different therapists and modalities. Although we collected data at three weeks to capture effects after each modality, we chose to forgo analysis across modalities due to the small sample, unevenly dispersed data, and order effects that would need to be considered because participants had the option to select the sequence of their two chosen modalities; these types of analyses are recommended for future studies.

Implications for policy, research, and practice

Our findings indicate that for at least some people, online sessions may be useful and allow for positive change over time. Despite skepticism expressed by some practitioners prior to delivering sessions, they were able to adapt to the virtual environment and accommodate people who had travel or transportation restrictions due to pain. Although the modalities offered in this study are traditionally rooted in physical touch, the experience was able to transcend the limitations of the digital world. Technical struggles such as internet lagging, mastering camera angles, and video-microphone settings added complications to interventions that may not be present in traditional face-to-face settings. Yet, there was no evidence that these created significant hindrances to providing engaging telehealth sessions that were seen as beneficial. From the practitioners’ perspective, they reported that most participants were highly engaged and the telehealth process was convenient, free of distractions, and beneficial for patient education.

Due to the emergence of SARS-CoV-2 and the spread of COVID-19, the project methods were altered to be completely virtual, as detailed elsewhere [9]. In the future, the project team would like to return to the original goal of conducting in-person consultations and comparing them to virtual sessions to determine efficacy. While some studies have found telerehabilitation to be equally effective to in-person physiotherapy in a few limited studies [38], widespread experimentation has not occurred testing a variety of non-pharmacologic therapies among various populations with pain.

Although this pilot study was conducted during the COVID-19 pandemic, the intervention aligned with recommendations that were later reinforced in newer clinical and policy guidelines emphasizing non-pharmacologic, interprofessional and telehealth-based approaches to chronic pain management [12, 13]. At the time of implementation, empirical research examining structured interprofessional telehealth consultations for individuals with chronic pain, particularly those prescribed opioids, remained limited. To our knowledge, few studies have examined the feasibility of delivering synchronous telehealth consultations that integrate multiple complementary and integrative health disciplines for this population. Our team has taken lessons learned from this study into clinical implementation of telehealth pain assessments using a nurse navigator model to assist in patient goal setting [41], and plans are underway to expand virtual yoga therapies within rural communities. More study is needed to determine how to consistently and reliably deliver multimodal, multidisciplinary pain care in an equitable fashion, and examine how this can affect opioid use and harms. Some studies suggest that persons from minoritized racial/ethnic and/or rural communities, and those with lower income, may be less likely to use telehealth [42]. Thoughtful testing and implementation will be necessary to ensure that those with the greatest need for care can access and realize the benefits of technology-based interventions.

In conclusion, virtual sessions using videoconferencing technology served as a feasible option for adults with chronic pain to access complementary and integrative therapies during a global pandemic. Yoga therapy and physical therapy were most commonly selected out of four non-pharmacologic options. Improvements were detected on outcomes important to pain management after six weeks. Future research using randomized designs can control for placebo and confounding effects and further explore barriers to digital health. The findings of this study support the expansion of telehealth services to address inequities in pain care. Telehealth holds promise in widening access to effective non-pharmacologic interventions and improving health outcomes for people with chronic pain.