Cooperative pain education and self-management (COPES): study design and protocol of a randomized non-inferiority trial of an interactive voice response-based self-management intervention for chronic low back pain

Participants are 230 patients with CLBP receiving treatment at the VACHS. Specific eligibility criteria include: 1) presence of at least a moderate level of pain (i.e., pain intensity rating of ≥ 4 on the 0 to 10 NRS) for a period of ≥ 3 months; 2) diagnosis of a low back pain condition in the electronic health record (EHR); 3) absence of a medical or psychiatric condition that could impair participation (e.g., severe COPD, terminal cancer, active substance abuse, psychosis, suicidality or severe depression Beck Depression Inventory-II score >28 ); 4) self-reported ability to walk at least one block without chest pain; 5) absence of dementia defined by St. Louis University Mental Status examination (SLUMS); [16] 6) absence of surgical interventions for pain during participation in this study (participants undergoing surgery will be discharged from the study in order to maintain the integrity of the active treatments); 7) availability of a touch-tone landline or mobile telephone; and 8) absence of any sensory deficits that would impair participation (e.g., hearing loss to a degree that telephone usage is not possible).

Participants are recruited from the VACHS by mailing letters to those who are identified via EHR review as having a back-pain-related diagnosis including low back and spine conditions and nerve compression (ICD-9 codes 724.02, 724.03, 724.2, 724.3, and 724.4) and a pain intensity rating of ≥ 4 (indicating moderate pain) [17] on the 0–10 NRS during their most recent clinic visit. Potential participants are called to solicit enrollment unless an “opt out” response is received by study staff, either by returning a pre-stamped and addressed letter or by telephone call. Additional participants are obtained through provider referral and advertisements placed in clinical care areas throughout the medical center and outpatient clinic locations of VACHS. Potential participants who respond to the referral or advertisement are screened by the study research assistant regarding key eligibility criteria for the study (e.g., confirmation of presence of chronic back pain of at least moderate intensity and absence of medical and psychiatric comorbidities that preclude eligibility). If initial screening suggests potential eligibility, a face-to-face appointment is scheduled for obtaining written informed consent and a full eligibility screening.

After consent, the final assessment of study eligibility is determined through review of the EHR by the study psychiatric advance practice registered nurse and psychologists, and participant responses to validated measures. The study nurse verifies the diagnosis of CLBP and uses a structured chart abstraction and assessment tool based on a classification system recommended in clinical practice guidelines [18] to identify the diagnosis and treatments of CLBP and classify participants’ pain as either non-specific low back pain, low back pain with a radicular component, or pain associated with other specific spinal causes. A trained research assistant under supervision of a psychologist reviews each participant’s chart to identify disqualifying psychiatric diagnoses or active suicidality. The research assistant administers a brief semi-structured interview to collect demographic and pain-related information, the Mini-International Neuropsychiatric Interview for DSM-IV (MINI) [19] to identify the presence of any DSM-IV psychiatric disorders, the Beck Depression Inventory-II (BDI-II) [20] to screen for severe depressive symptoms, and the SLUMS examination to screen for cognitive deficits likely to interfere with treatment participation. Prescribed pain medications and their dosage, as noted in the EHR, are recorded. Individuals who meet eligibility criteria are enrolled in the IVR system and TrialDB, a secure, encrypted clinical trials management database that facilitates participant randomization and collection and management of questionnaire data.

Eligible participants are randomized to either in-person CBT or IVR-CBT immediately after completing the baseline assessments. The study research assistant enters the participant’s demographic information and stratification factors into TrialDB, which holds the randomization schedule and is masked from users thereby enabling concealed allocation. The treatment allocation ratio is 1:1 using a permuted stratified block design with variable block size. Randomization is done within strata defined by patients’ distance from the VACHS-West Haven campus (<10, 11–25, 26 or greater miles) and cause of low back pain as determined by the study nurse based on EHR review (non-specific low back pain, low back pain with a radicular component, or pain associated with other specific spinal causes) [18]. The research assistant informs a study therapist of the participant’s treatment assignment.

The research assistant enrolls participants in the IVR system and TrialDB, a separate, secure, encrypted web-interface used for completing the study outcome measures. Participants are provided with training in the use of TrialDB, the IVR system and the study pedometer by a brief demonstration supplemented by written instructions included in the patient handbook. Participants’ stride length is measured and they are provided with an Omron Go Smart Model HJ-112 pocket pedometer to use in the walking portion of the CBT and IVR-CBT interventions.

At the end of the baseline appointment participants are given either a CBT or IVR-CBT patient handbook.The patient handbooks were adapted from materials used in a prior trial of in-person CBT for CLBP by psychologists with experience in the delivery of CBT for chronic pain [21]. The materials are written at a 6^th grade reading level and pictures and simple figures are incorporated as often as possible to enhance patients’ understanding and engagement. As a check on comprehension, participants are asked five true/false questions about each treatment module in the week it is presented. The IVR-CBT patient handbook and IVR scripts were pilot tested with 17 individuals with chronic pain who reviewed the materials and an additional four individuals who received treatment using the new materials. Both groups provided feedback via semi-structured interviews and revisions were made to the materials based on feedback.

After enrollment, but prior to beginning treatment, participants receive seven daily IVR calls to assess their baseline pain intensity, pedometer-measured step counts and self-reported sleep duration in hours [22]. Starting on the first day of treatment and continuing for 70 days, participants receive daily IVR calls to answer seven daily questions that assess pain intensity, sleep quality and duration, pedometer-measured step count, catastrophizing (I worried my pain would never end”, “I felt my pain was so bad I could not stand it anymore”) and adherence to the current week’s skill practice goal. Once per week participants are asked to report: 1) any adverse events associated with the graduated walking portion of the treatment, 2) any increase or decrease in pain medication dose made on the advice of their physician or their own initiative, 3) how often they practiced their weekly, self-selected pleasant or meaningful activity goal and if it improved their happiness or satisfaction 4) if they continued to use any of the pain coping skills learned in prior weeks, and 5) their comprehension of the module material via five true/false questions about the week’s pain coping skill. All of the information reported during a call is automatically captured in a database and time and date stamped for later review by a therapist. Participants receive a call at their chosen time; if they do not complete the call, they are called again fifteen minutes after the designated time, and if they do not complete the second call, they are called again one hour later. If they do not answer any of these calls, the call day is considered missed and they are not called again until the following day. Participants have the option of calling into the system to make their daily report in circumstances when they are unavailable at the scheduled call time. The IVR system flags any instances of a participant missing the first call day or two consecutive call days, and the study research assistants contact the participant to assist them if there is a technical difficulty or to encourage adherence to the call schedule. When participants are unable to complete calls for an extended period of time such as during vacation, calls can be paused and resume when a participant returns. Participants are not paid for completing daily IVR calls.

Both CBT conditions include 10 treatment modules delivered over 10 consecutive weeks. Treatments are delivered by either a PhD-level psychologist or the study nurse trained and supervised by a clinical psychologist with specific competencies and experience in delivering CBT for chronic pain. Therapists follow treatment manuals and participants are provided with a treatment-specific participant handbook. The 10-week course of therapy consists of an introductory module, followed by 8 consecutive pain coping skills modules, and concludes with a 10th module emphasizing skill consolidation and relapse prevention. Both conditions include the same pain coping skills, which were selected from a wider collection of possible skills because they were rated most highly by participants in a prior funded trial on importance, interest and ability to engage in them [21] (see Table 1 for description of skills). Participants assigned to both conditions continue to receive routine care of their CLBP by their current healthcare providers. Study staff do not attempt to influence clinical care other than CBT in any way.

Each module includes: (1) the presentation of an explicit rationale for development and use of the specific coping skill being taught, (2) a description of each coping skill, skill modeling and practice of the skill during the session, (3) problem-solving about the practice and use of the specific skill, and (4) assignment of a specific goal for practice of the week’s skill and collaboration between the therapist and participants to develop a weekly goal for engaging in pleasant or meaningful activities.

Each week the therapist reviews participants’ IVR-reported data via a web interface that summarizes the daily reports (see Fig. 2). Therapists comment on participants’ average pain intensity, sleep, steps, and skill practice for the week and note the weekly average, any change from the prior week, either positive or negative, and note any patterns (e.g., more pain on weekends versus week days). Therapists also provide reinforcement for any goal accomplishment and note potential associations among pain, goal accomplishment, and pain coping skill practice. Therapists provide corrective feedback on any incorrect answers to the true/false questions. Finally, therapists assign a steps goal for the upcoming week.

In-person CBT and IVR-CBT Treatment Receipt/Comprehension: The measures of treatment receipt (i.e., the five- item true/false content questionnaire) have been described above. We will analyze and report these data regarding average level of skill comprehension and whether any treatment modules were more difficult for participants to understand. Treatment fidelity will be assessed by psychologists with expertise in CBT for chronic pain, who will rate audiotapes of 30 % of the in-person CBT sessions and the IVR-CBT weekly feedback recordings to assure that key components of the manuals are covered. Percentages of treatment integrity/violations will be calculated. The study’s psychologist principal investigator will provide corrective feedback to the psychologist whenever drift occurs.

Although walking is not considered risky for patients with chronic musculoskeletal pain and is often recommended, information about adverse events associated with participation in the walking portion of the treatment is solicited once per week during the IVR assessment calls. Participants may report adverse events by recording a brief audio message. These messages are reviewed each business day by study staff and any reports that indicate a cause for concern are followed-up with the participant, their primary care provider, or both.

Survey outcomes are assessed at 4 time points, baseline, post-treatment, and 6 and 9 months post baseline. Participants are given up to 14 weeks to complete treatment in order to accommodate rescheduled sessions due to vacations, illness or travel difficulties. Post-treatment data are collected when a participant completes treatment or at 15 weeks post-baseline regardless of whether the participant has completed treatment. Survey outcomes are assessed using the TrialDB system, which allows participants to complete the self-report questionnaires online via a secure web-based interface. Participants who do not have an Internet connection may come to the medical center to complete the measures via TrialDB or complete a paper version of the measures and return them by mail. IVR outcomes will be assessed daily during the baseline week and over the 10 treatment weeks.

Measures were selected based on the Initiative on Methods, Measurement and Pain Assessment in Clinical Trials (IMMPACT) recommendations [23], which called for the assessment of multiple domains of the pain experience in all pain treatment trials and recommended measures based on their psychometric properties and respondent burden. These measures are consistent with CONSORT guidelines for non-inferiority trials [24] that recommend that outcome measures be similar to those used in studies to establish efficacy of the reference treatment.

Primary Outcome - Pain intensity: We assess participants’ average pain intensity over the past week via survey, at the four outcome assessment points using the Numeric Rating Scale of pain intensity (NRS-I) [25] an 11-point numeric rating scale (0 = no pain, 10 = worst pain imaginable). The primary outcome is the average weekly pain intensity at post-treatment.

Secondary Outcomes – Pain intensity will also be assessed daily during the trial via the IVR system using the NRS scale. Subjects will have a maximum of 77 daily scores: 7 daily calls for the baseline assessment week and 7 daily calls for each of the 10 weeks of treatment. We will use these scores to compute weekly means for each subject. The mean of seven daily assessments has been found to provide reliability in excess of 0.90 [26].

Physical functioning: The nine-item Interference subscale of the West Haven-Yale Multidimensional Pain Inventory (WHYMPI) is used to assess pain-related interference in social, work and household activities and has demonstrated good internal consistency (.86) and stability (.85 over 2 weeks) [27]. The 24 item Roland and Morris Disability Questionnaire is designed to assess physical functioning in patients with low back pain [28]. It has demonstrated good internal consistency (.84-.93) and stability (.83 over 3 weeks) [29].

Emotional functioning: Overall emotional functioning is assessed using the 65-item Profile of Mood States (POMS) [30], a multidimensional measure of emotional functioning designed to assess six dimensions of mood and to be used in non-psychiatric or physically ill populations. Internal consistency (0.84 to 0.95, depending on subscale) and test-retest reliability (0.65 to 0.74, depending on subscale) is good. Depression symptom severity is assessed using the 21-item Beck Depression Inventory-II [31] (BDI-II) a widely used self-report measure with excellent internal consistency (.94) and evidence of convergent and discriminant validity in primary care medical patients (.73–.96) [20].

Quality of Life: The Veterans SF-36 is used to assess health-related quality of life. This measure has demonstrated good internal consistency (.78–.93 across 8 subscales) and is strongly correlated with socioeconomic status and morbidities [32].

Call adherence is calculated as the total number of IVR calls made divided by the total number of expected calls (77).

Patient satisfaction is assessed by a modified version of the Client Satisfaction Questionnaire-8 [33], an 8-item satisfaction survey designed to assess global satisfaction with treatment. This measure has been used widely to assess treatment satisfaction across numerous types of interventions and demonstrates good internal consistency (α = 0.87–0.93 across three samples) and has shown correlation to treatment attendance and retention.

Session attendance is tracked and the number of participants who do not complete follow-up assessments is calculated. In order to understand reasons for treatment dropout and evaluate “missingness” we prospectively collect data regarding reasons for treatment dropout and missing data. The research assistant, who monitors IVR and questionnaire completion, collects these data.

The sample size was calculated based on a test of non-inferiority comparing IVR-CBT to CBT, with 80 % power, Type I error (one-sided) of 0.025, and assuming the true difference between means is 0. Based on preliminary data from similar participants in an ongoing study of the efficacy of CBT for chronic back pain [21], the estimated baseline NRS pain intensity score will be 7 ± 2.45 units. A 20 % reduction in the NRS pain score from baseline to post-treatment, that is, from 7 to 5.6 or 1.4 units, is considered to be clinically relevant. The non-inferiority margin was set at one NRS unit, so that if the mean pain rating for IVR-CBT is less than one unit higher than the mean in-person CBT score (mean IVR-CBT < mean in-person CBT + 1), IVR-CBT will be deemed non-inferior. Conducting a test of this hypothesis can be done by comparing the upper limit of the 95 % confidence interval for the mean pain difference between IVR-CBT and in-person CBT with the non-inferiority margin 1; if this upper limit falls below 1, we will be able to conclude IVR-CBT is non-inferior. Based on these assumptions and 15 % inflation for losses, the total required sample size is 230 participants (115 per group).

The adequacy of the randomization will be assessed by comparing baseline demographic and clinical characteristics between the two treatment groups. Additional covariates will include number and location of pain sites and difference in medication use at baseline in order to assess whether these variables are associated with a differential response to the treatment.

We present below the analysis of primary and secondary outcome measures only. Analysis of primary and secondary outcomes will employ linear mixed-effects models for longitudinal data, which will account for the clustering induced by repeated measures on individual patients. Mixed effects models make use of all available measurements on subjects and provide valid inferences even in the presence of missing data as long as the missingness is at random. Data from different subjects are assumed to be independent, while the correlation structure of the repeated measurements within subjects is modeled via parameterization of the covariance structure.

Between-group comparisons of the effectiveness of IVR-CBT and in-person CBT at each assessment time point will be conducted. Each of these comparisons can be tested separately within the same mixed-effect model using a treatment dummy variable, time dummy variables, and treatment-by-time interaction terms and appropriate contrasts. The outcome variable in each model will be the outcome at the follow-up time points, with the baseline value of the outcome and the stratification variables included as covariates in the model. Results will be summarized as LS- means (and their 95 % confidence intervals) within and between treatment groups at each time point. Non-inferiority will be demonstrated by the upper limit of the 95 % confidence interval for the LS-mean pain difference between IVR-CBT and in-person CBT at post-treatment being less than 1. Following CONSORT recommendations [24], non-inferiority analyses will be conducted both on a per protocol (e.g., having completed 3 weeks of treatment which is considered a dose of treatment) basis and an intent-to-treat basis. All other analyses will be conducted according to intent-to-treat basis, that is, by considering patient group status as randomized.

We will compare the distribution of lost patients by reason between the two study groups to evaluate any differences in the reasons for losses. We will compare the rate of loss to follow-up between the two groups using the chi-square test, and if this rate is different between the two groups, interpretation of the results will be made in view of this finding. If we find any baseline variables to be associated with the loss to follow-up, then we will include these baseline variables as covariates in the models evaluating the intervention effect. In case of problematic missingness, we will use multiple imputation based on sequential regression imputation to impute missing variables. Following Rubin's method [34] for multiple imputation inference, each of the simulated complete datasets will be analyzed by standard methods, and the results will be combined to produce estimates and confidence intervals that incorporate missing-data uncertainty.