Study design
This study was a mixed-methods sequential explanatory [26] pilot RCT with 6-month follow-up. The complete protocol, including description of the study design and methods was published prospectively [23] and registered retrospectively (27/07/2021) at ClinicalTrials.gov, #NCT04978961. All procedures were approved by the Institutional Review Board at The University of Texas Health Science Center at San Antonio (HSC20160512H) in accordance with federal codes for the conduct and protection of human subjects.
Setting and participants
Our study site was a primary care clinic affiliated with The University of Texas Health Science Center at San Antonio using the Primary Care Behavioral Health model [7] of integrated care; the census in the clinic at the time was approximately 6500 unique patients. The study site had one clinical half-time BHC and one study-funded BHC (approximately 0.1 FTE) who conducted all of the clinical procedures detailed below. Patient inclusion criteria required (a) age 18 and older; (b) at least one non-cancer pain condition persisting for 12 weeks or more; (c) a current primary care clinician at the study clinic; and (d) ongoing treatment for a non-cancer chronic pain condition. Exclusion criteria were minimized for generalizability: (a) social anxiety or unwillingness to participate in a class setting; (b) presence of symptoms of psychosis and/or delirium; (c) a medical condition or life circumstance that would contraindicate or prevent participating (e.g. upcoming surgery); and (d) inability to comprehend the informed consent process or study instructions.
Recruitment
Patients were recruited via: advertisements about the study posted in the clinic; referral by clinic personnel; and direct invitation based on pre-screening eligibility identified in the electronic health record. Recruitment and treatment were rolling, such that participants were enrolled as they completed pre-screening and consent procedures. Potential participants who indicated interest were pre-screened for eligibility by study staff then scheduled for their first visit where they completed informed consent procedures and baseline assessments. Participants were compensated for attending assessment visits. We also held a raffle for $10 gift card at each FACT-CP group visit to enhance retention. FACT-CP treatment was provided at no cost. This study started enrollment in October, 2017, and 6-month follow-up assessments concluded in December, 2018; exit interviews concluded in April, 2019.
Randomization
Following prescreening, consent, and baseline assessment, participants were randomized to receive either FACT-CP or Enhanced Treatment as Usual (ETAU). Because pain severity is a factor that could affect outcomes and/or responsiveness to intervention [27], we stratified to avoid misattributing effects of the intervention (i.e., type 1 error), which is especially important in small trials [28]. At baseline, participants rated their level of pain severity using the Numeric Rating Scale (NRS) for Pain, a frequently-used single-item measure consisting of a horizontal line anchored with numeric labels (0 = no pain and 10 = worst possible pain). We developed 3 levels of stratification based on clinical consensus in our team and previously-recommended cut-points [29, 30]: 1–3 (mild), 4–7 (moderate), and 8–10 (severe). Using SAS software (Cary, NC) we created randomized block sizes ranging from 4 to 12. Varying sizes were used to balance groups while blinding study staff/PI to assignment, with NRS score category entered into a custom web-based application to facilitate and mask the randomization process (see Fig. 1 for our CONSORT diagram) [31].
Data collection
Self-report measures described below were administered at 6 assessment visits: baseline, weeks 2, 3, 4, 12 (booster) and 24 (6-month follow-up; see Fig. 2). The study-appointed BHC was not involved in these visits and was blind to assessment results throughout. The study BHC did administer screening measures at each FACT-CP visit, including the NRS, in keeping with usual practice and standards of care in the clinic. Additional data were collected after the 6-month assessment visit by the PI (KEK), who contacted participants via phone and conducted semi-structured “exit interviews.”
Interventions
Focused Acceptance and Commitment Therapy treatment
Approximately 1 week after baseline quantitative assessment, participants in the FACT-CP arm attended a 30-min individual visit with the BHC consisting of typical initial visit activities [7], including role clarification of the BHC; contextual interview for functional analysis of the pain problem; biopsychosocial case conceptualization; the brief FACT-CP intervention; between-visit exercise recommendations; and collaboratively-developed goals.
Over the next 3 weeks, participants attended weekly 1-h group visits focused on increasing acceptance of chronic pain towards greater psychological flexibility in responding to and coping with chronic pain, followed by a “booster” visit approximately 2 months later [23]. The BHC communicated with participants’ primary care clinicians throughout the course of FACT-CP care (individual BHC visit through booster visit) and kept clinical notes in their electronic health records, consistent with usual practice.
Enhanced treatment as usual
Participants assigned to ETAU attended assessment visits where they completed study measures and then received a 1-page double-sided handout from the study research assistant that was based on CBT principles for coping with stress and pain (relaxation, pacing, sleep, goal-setting). Participants in both arms continued to access usual primary and specialty care treatment throughout the study.
Sample size
Recruitment was stopped at 13 per arm, in keeping with guidance on conducting pilot and feasibility studies [24, 25] (e.g., CONSORT [31]). We initially sought a larger sample size, however, at this stage, our team determined that a smaller sample was sufficient to pilot our goals examining feasibility and acceptability of the interventions, as well as study procedures, and to obtain preliminary data on the effectiveness of the FACT-CP intervention.
Primary outcome measures (quantitative data)
Feasibility
Feasibility was evaluated using a priori established benchmarks: (a) < 25% participant attrition; (b) at least 80% of participants rating the FACT-CP program as satisfactory as measured by response of at least 5 on a 7-point Likert-scale (1 = Not Satisfied at All, 7 = Very Satisfied). The satisfaction question was asked in the context of an “exit interview”: all participants who completed the 6-month follow-up assessment were contacted by the PI to provide feedback on their experiences in the study. The semi-structured phone interview lasted approximately 10 min and the quantitative data portion included 5 Likert-scale questions to assess patient experiences with study participation, ease or difficulty of learning pain management skills, amount of information learned, and satisfaction with treatment.
Feasibility measures also included fidelity checks of the study BHC. Fidelity to FACT-CP was independently assessed by the study’s external consultant (PR). All treatment visits were audiotaped. Our consultant randomly selected and listened to 22% of these visits, assessing fidelity using a standardized checklist based on the FACT-CP treatment manual. Fidelity was evidenced by greater than 95% adherence to treatment.
Acceptability
Acceptability of the FACT-CP intervention from the participants’ perspective was measured via 3 Likert-scale questions gathered during the semi-structured interview: perceived benefit, ease of learning about pain management, and whether the participant would recommend the FACT-CP treatment to a friend or family member.
Effectiveness
The primary outcome of effectiveness was self-reported physical disability, assessed using the modified and psychometrically sound Oswestry Disability Index (ODI) [32]. The ODI is a 10-item self-report measure using 6-point Likert scales, originally developed as a measure of back pain. We used an established modified version that asked about “pain” rather than “back pain” [33, 34]. Scores are summed to create a total score (maximum 50) that is then divided by the highest possible score based on items completed, then doubled to provide a percentage of disability. Reliability in our study was high (Cronbach’s alpha, α = .85).
Secondary outcome measures
Pain acceptance was examined using the Chronic Pain Acceptance Questionnaire-Revised (CPAQ) [35]. The 20-item CPAQ assesses the degree to which chronic pain and related experiences influence behaviors and the degree of effort put in to controlling pain. Items are responded to on a 0 to 6 Likert scale. Higher scores indicate greater acceptance; scores range from 0 to 120. We measured engagement in values-based activity with the Chronic Pain Values Inventory (CPVI) [36, 18], an inventory that identifies which values are important to a patient with chronic pain, and assesses the degree of success they are having in following their values. The valued domains are family, intimate relations, friends, work, health, and growth or learning. The 12-item CPVI uses 6-point Likert scale questions to measure the discrepancy between ratings of importance of valued life areas and success in engaging in those life areas; lower scores reflect greater alignment (i.e., less discrepancy) between values and actions in one’s life. The stand-alone success scale includes ratings on the engagement items only. We chose to employ the discrepancy scale rather than the success scale because we wanted to measure success in engagement in valued activities in the context of their perceived importance. CPVI scores range from 0 to 6. Reliability was high for both the CPAQ (α = .85) and the CPVI (α = .82).
Measures of participants’ experiences (qualitative data)
Qualitative data were also gathered during the exit interviews, which included open-ended questions assessing the following domains: what participants liked most and least about their participation, and any changes in pain management or quality of life due to participation (e.g., “In what ways has your participation in our study changed the way you think about or manage pain?”). Participants were given time to discuss anything else they wanted to share with the PI. The PI took near-verbatim contemporaneous notes during the interviews.
Analyses
Aim 1 analyses
We examined acceptability and feasibility of FACT-CP and study procedures by calculating percentages and frequencies. Physical disability (primary outcome) was examined using a general linear mixed (within and between groups) regression model with repeated measures, controlling for baseline pain severity, with the primary focus on comparing pre-post change in the 2 study arms. Fixed effects in these statistical models were treatment arm, time, and the treatment-by-time interaction. Although the analysis does produce conventional ANOVA-type tests, those are non-specific. Instead, the hypothesis tests were done using planned, a priori contrasts that compare the regression-based least-square means to estimate change in a group using all subjects (intent-to-treat analysis), including those with missing data. Baseline pain severity was included as a covariate because it was used to stratify randomization [37].
Aim 2 analyses
We examined secondary outcomes, acceptance of chronic pain (CPAQ) and engagement in values-based activity (CPVI) between and within groups using general linear mixed regression models with repeated measures, again with the primary focus on comparing pre-post change in the 2 study arms. Baseline pain severity was again included as a covariate.
Missing data analyses.
Across the 6 weeks of assessments, between 0 and 15.4% of data were missing from each measure. Missing data were handled using maximum likelihood estimation. This yields valid parameters given the usual assumption that data are missing at random. Additionally, Little’s MCAR test was non-significant for all variables, [ODI: X2 (15, N = 26) = 8.34, p = .909; CPAQ: X2 (11, N = 26) = 8.13, p = .702; CPVI: X2 (15, N = 26) = 14.08, p = .520]. All data were analyzed using SPSS 26.0 [38] and/or SAS v9.4 (Cary, NC).
Aim 3 qualitative analyses
We analyzed qualitative interview data using rapid qualitative analysis [39, 40]. This approach, compared with in-depth analyses, is particularly useful in studies with resource constraints (i.e., a pilot study) [41] and in research conducted in clinical settings, to aid in timely dissemination of patient feedback [39,40,41]. Rapid qualitative analyses has been compared directly with more traditional thematic analysis and found to produce closely aligned results, and is thus considered to have comparable rigor [40]. Participant responses were organized by question (domains for both groups: Best/Most Likeable Features; Worst/Most Disliked Features; Changes in Perception of Pain; Changes in Quality of Life; and Other Feedback). Analyses were structured to identify similarities and differences within and between groups. Interview notes were consolidated into a matrix, with rows for each domain and columns for individual respondents. Themes emerging within each domain and exemplar quotes were then identified by 2 co-authors who met in person to reach consensus (KEK and EPF); 3 other independent raters (BH, LSK, CM) reviewed the table of consolidated themes and quotes via email and/or in-person review and provided iterative feedback until consensus was achieved [40].