Integrated Meditation and Exercise Therapy: A Randomized Controlled Pilot of a Combined Nonpharmacological Intervention Focused on Reducing Disability and Pain in Patients with Chronic Low Back Pain

Abstract

Objective

This pilot trial examined the effects of a combined intervention of mindfulness meditation followed by aerobic walking exercise compared with a control condition in chronic low back pain patients. We hypothesized that meditation before exercise would reduce disability, pain, and anxiety by increasing mindfulness prior to physical activity compared with an audiobook control group.

Participants

Thirty-eight adults completed either meditation and exercise treatment (MedExT) (n=18) or an audiobook control condition (n=20).

Setting

Duquesne University Exercise Physiology Laboratory.

Design

A pilot, assessor-blinded, randomized controlled trial.

Methods

Over a 4-week period, participants in the MedExT group performed 12–17 minutes of guided meditation followed by 30 minutes of moderate-intensity walking exercise 5 days per week. Measures of disability, pain, mindfulness, and anxiety were taken at baseline and postintervention. Pain perception measurements were taken daily.

Results

Compared with the control group, we observed larger improvements in disability in the MedExT intervention, although the changes were modest and not statistically significant (mean between-group difference, –1.24; 95% confidence interval [CI], –3.1 to 0.6). For secondary outcome measures, MedExT increased mindfulness (within-group) from pre-intervention to postintervention (P=0.0141). Additionally, mean ratings of low back pain intensity and unpleasantness significantly improved with time for the MedExT group compared with that of the control group, respectively (intensity P=0.0008; unpleasantness P=0.0022).

Conclusion

. Overall, 4 weeks of MedExT produced suggestive between-group trends for disability, significant between-group differences for measures of pain, and significant within-group increases in mindfulness.

Introduction

Low back pain is the most commonly reported type of pain [1] and the second leading cause of physician visits and disability among US adults [2]. Globally, 25% of adults report having low back pain over any 1 month [3]. Often due to nonspecific causes and complicated by comorbid symptoms [2], low back pain remains difficult to treat. Current treatments include nonsteroidal anti-inflammatory drugs (NSAIDs), muscle relaxants, opioids, psychological therapy, physical therapy, chiropractic manipulation, injections, and surgery [2, 4–6]. Chronic low back pain (cLBP) is further complicated by the potential for comorbid anxiety disorders [7]. In particular, musculoskeletal pain is often associated with fear avoidance anxiety behavior and kinesiophobia [8, 9]. This kinesiophobia—or fear of movement—can further exacerbate pain and subsequent disability [10, 11]. Fear avoidance may also reduce the potential benefit of physical treatments in patients by increasing state anxiety before and during exercise or movement therapies. In treating chronic pain, a major gap exists in not only treating the physiological condition, but also in addressing the interplay with psychological etiologies.

There has been a significant push in the last 20 years to identify and understand complementary and integrative therapies to supplement or replace pharmacology. Nonpharmacological therapies include aerobic exercise, tai chi, yoga, mindfulness-based stress reduction (MBSR), progressive relaxation, electromyography biofeedback, operant therapy, cognitive behavioral therapy, multidisciplinary rehabilitation, acupuncture, spinal manipulation, and massage, with many of these showing significant positive effects [12–15]. There has been considerable interest in programs that combine elements of complex interventions to treat chronic pain [16–20]. One of the most well-established integrative programs that involves the elements of stress reduction, exercise, and meditation is the 8-week MBSR program of group-based classes, which has been found to improve pain, depression, and quality of life [21]. However, this program requires extensive training and may not be easily accessible to some persons with cLBP. In the current study, we examined a more simplified pilot program of introductory mindfulness meditation that novice meditators could easily put into practice prior to aerobic walking exercise. Both meditation and exercise have been independently investigated in the context of back pain therapy and have been found to be effective, providing a framework for the current pilot study [16, 22].

Exercise interventions have been proven to have beneficial outcomes on pain severity, physical disability, psychological function, and health-related quality of life for various chronic pain conditions [22–26]. Mechanistically, aerobic exercise at a level of at least 70% of the maximum aerobic capacity generates the production of endorphins and elicits other pain inhibitory mechanisms driven by the central nervous system [27, 28]. In addition, aerobic exercise has been shown to reduce fatigue and improve peak oxygen uptake and physical fitness [29, 30]. Walking exercise, specifically, is a feasible intervention that requires no training and that has been shown to improve pain perception measured by quantitative sensory testing (QST) [31].

Similar to exercise, studies incorporating mindfulness meditation have been largely shown to improve pain and depression symptoms, quality of life, and well-being and increase mobility and functioning [21, 32]. Mindfulness meditation refers to the sustained recognition of the knowing quality of awareness itself [33]. The practice of mindfulness focuses on orientation to the present moment with openness and acceptance [33]. Mechanistically, meditation with mindfulness has been associated with decreased levels of cortisol [34], increased signaling connections in the brain [35], improved pain processing and emotional control [36], and altered amygdalar response to emotional stimuli [37]. As these therapies (exercise and meditation) independently decrease disability and pain, we sought to test a combined practice of these interventions vs a time-matched audiobook control condition. The Meditation and Exercise to Treat Chronic Low Back Pilot Trial (MedExT) consisted of a 4-week intervention of a guided mindfulness meditation program combined with moderate-intensity walking exercise performed 5 days per week in cLBP patients. We hypothesized that participants randomly assigned to the MedExT would have improved disability (primary outcome) and pain compared with audiobook control participants. We also predicted that MedExT participants would experience increased mindfulness and less anxiety following the intervention. This specific therapy combination has not been previously examined in cLBP patients.

Methods

Study Design

This pilot study was designed as a randomized controlled trial, with repeated measures testing the effect of a combined treatment of mindfulness meditation and aerobic walking exercise (MedExT) compared with a control intervention. QST was conducted by an investigator (Benedict J. Kolber) who was blinded to treatment assignment throughout the trial. Matthew C. Kostek was responsible for generating the random allocation sequence, and Anna M. Polaski was responsible for enrolling and assigning participants to interventions. A power analysis indicated that a minimum of 21 participants per group would be sufficient to detect statistical differences in our primary dependent variable—disability measured with the Roland Morris Disability Questionnaire (RMDQ) [38]. Mindfulness meditation has been shown to produce a large effect size (Cohen’s d=1.06), which is not unusual for mindfulness meditation and pain [19, 39]. For our power calculations, we set a conservative effect size of 0.8. We used G*Power to calculate the sample size of 21 using an alpha of 0.05, an effect size of 0.8, and a power of 0.80. All procedures were approved by the Duquesne University Institutional Review Board (Protocol 2017–05-12), and written consent was obtained from each participant prior to testing. All methods were performed in accordance with the relevant international and local guidelines and regulations for human research. This study is registered with ClinialTrials.gov under ID NCT03324659 (October 30, 2017). Participants were compensated up to $200 for participation.

Participants

Participants were 52 adults with cLBP (more than 6 months) with no evidence of neuropathic pain, radicular pain (i.e., sciatica), or referred somatic pain. There were no specific criteria for the level of low back pain severity. Participants were recruited using in-clinic recruitment by Eric R. Helm through the University of Pittsburgh Department of Physical Medicine and Rehabilitation Research Registry (PMR3) and the University of Pittsburgh Clinical and Translational Science Institute (CTSI) patient registry, “Pitt+Me.” Initial prescreening was completed during recruitment to the Pitt+Me database with phone follow-up by Anna M. Polaski. Full inclusion criteria included the following:

1. A body mass index (BMI) within the normal to overweight range (18.5–29.9).

2. A resting heart rate (HR) between 60 and 100 bpm.

3. Resting blood pressure (BP) less than or equal to 140/90 mm Hg.

4. The ability to independently ambulate community distances without external support (e.g., walker, cane).

Exclusion criteria included the following:

1. Cardiovascular or respiratory disease.

2. Neurological disease unrelated to low back pain.

3. Diabetes mellitus types 1 and 2.

4. Diagnosis of chronic pain condition unrelated to low back pain.

5. Acute pain.

6. Regular participation in high-intensity athletic or sporting activities as determined by the International Physical Activity Questionnaire (IPAQ)—Short Form.

7. Sedentary lifestyle as determined by the IPAQ—Short Form.

8. Currently pregnant individuals.

9. Current cigarette smokers.

10. Individuals with ongoing litigation associated with back pain.

11. Regular participation in meditation techniques or training in MBSR.

A priori, we determined that we would evaluate all participants that completed 80% of each week’s sessions. This value was determined by whether participants completed the daily Qualtrics survey before and after their session. Once participants missed more than one session in a week, these participants were ineligible to continue. All 38 participants included in the results presented in this article met this threshold. As this was a pilot study, the aforementioned inclusion and exclusion criteria were chosen to reflect a healthier population of chronic low back pain patients (i.e., lower BMI, nonsmokers, lower resting HR, and lower BP). This was done in accordance with the American Heart Association (AHA)/American College of Sports Medicine (ACSM) Health/Fitness Facility Pre-Participation Screening Questionnaire to maintain participant safety [40].

Procedures

In-clinic sessions were conducted at Duquesne University’s Exercise Physiology Laboratory over the course of the 4-week intervention period between January 2018 and April 2019. For participants meeting phone-screening criteria, informed consent was obtained and participants were enrolled in the study. An initial clinical screening examination was performed by one of three clinicians (Eric R. Helm or two trained physician assistants). During this screening (approximately 15 minutes), patients were evaluated for strength, lumbar range of motion, reflexes, and sensation in relation to their low back pain. This screening was done to verify back pain inclusion (i.e., exclude radicular patients) and to determine safety of participation in the exercise portion of the intervention. Of 55 patients recruited, no patients were excluded during this screening. Following the clinical screening, patients were scheduled to start the actual intervention. The average time between consent and the start of the trial was 26 days.

At the start of the full pilot trial (after the clinical screening), participants came in for an intake session during which they completed a battery of questionnaires (see the Survey Instruments and Administration section) and were oriented to the general study protocol. See Figure 1 for assessment timing and intervention details. The intake session consisted of a sequence of quantitative sensory tests and baseline assessments of pain (see the Quantitative Sensory Testing section). Benedict J. Kolber performed all QST blinded to the treatment group of the participants and remained blinded to treatment until after the final pain assessments were completed. Participants were blinded to treatment assignments for baseline intake testing. Following baseline testing, treatment assignments were disclosed to the participants.

Figure 1.

MedExT vs control treatment design. MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial.

Within 1 week of performing baseline pain assessments (the average time between baseline and first intervention session was 5 days), participants completed their first in-clinic intervention session. At the start of this session, patients received approximately 35–45 minutes of meditation training or stress management training from a clinical psychologist (Thomas J. Smith). These sessions discussed either the potential of and use of mindfulness and meditation (MedExT group) or general stress management and well-being for chronic pain (control group). Sessions were standardized by using a script developed by Thomas J. Smith (see Supplementary Data  1). Following this session, participants completed their first intervention session—either combined meditation and exercise (MedExT) or the control condition. Participants had the option to complete intervention sessions at home or in the clinic. At-home sessions were performed using a home treadmill or a treadmill at a local gym. No participants chose to exercise daily in the clinic, and access to a treadmill was not reported as a barrier by any participant. Interventions were performed 5 days per week for 4 weeks. In-clinic intervention sessions were typically attended once per week depending on participant scheduling conflicts. During these sessions, two experimenters were present and checked in with the participants to ensure that they were not experiencing any difficulties completing the assigned intervention. Forty-eight hours after the end of the 4-week period, participants attended the exit session, where they again completed surveys and underwent QST.

Meditation and Exercise Protocol

For participants in the MedExT experimental group, guided meditation recordings with a focus on mindfulness performed by meditation teacher and psychologist Dr. Tara Brach were used [41, 42]. Five different recordings were used; each recording was listened to one time per week and lasted between 12 and 17 minutes (see Supplementary Data  2 for links to the recordings). Participants were discouraged from listening to the recordings outside the designated intervention time. Recordings were selected by Thomas J. Smith along with clinical psychologist Ian C. Edwards for their focus on mindfulness and their overall length. Participants were given an MP3 player to borrow (SanDisk) loaded with each of the five meditation recordings. During the weekly in-clinic session, participants practiced the meditation portion of the intervention session in our interdisciplinary meditation room, which was a quiet space with low lighting and comfortable seating options. For at-home intervention sessions, participants were encouraged to perform meditation in a quiet, comfortable setting. The protocol for home meditation was the same as that for the in-clinic session.

Immediately following meditation, participants performed 30 minutes of moderate-intensity walking exercise on a treadmill. Prior to the first exercise session, resting HR and age were used to calculate an HR that corresponded to a 50% heart rate reserve (HRR) for each participant [43]. We used the 50% HRR estimate as the target HR for moderately intense exercise with a range of 40–60% HRR calculated for each participant. HR monitors (Polar H1, Polar Electro Oy, Kempele, Finland) were worn for each in-clinic exercise session to monitor exertion levels. During the first in-clinic exercise session, trial coordinator Anna M. Polaski would manipulate the speed and grade of the treadmill to achieve the calculated HR for an individual participant. The average grade was 2.4%, and the speed range was 2.2–3.8 mph. Once a 50% HRR was reached, this speed and grade combination was used as the walking prescription for subsequent exercise sessions for that particular participant. Prior to and following exercise, each MedExT experimental intervention participant rated their perceived exertion levels using the Borg Rating of Perceived Exertion (RPE) scale to ensure that exercise intensity was within the moderate-intensity range (i.e., 12–14) [44]. Each exercise session began with a 2-minute warm-up at 2.5 mph and concluded with a 2-minute cooldown (total time of 30 minutes on the treadmill). For at-home sessions, participants followed this same protocol using their prescribed dose from the initial in-clinic session. This dose was continued for the duration of the trial.

Control Protocol

Participants in the control group listened to an audiobook for 12–17 minutes followed by a 30-minute rest period five times per week for 4 weeks. Each session was time matched to the experimental intervention group. Participants were given an MP3 player with 20 (one for each day) recordings of Natural History and Antiquities of Selborne [45], which has been previously used and validated as a neutral comparison for guided relaxation interventions [46–48]. During the resting period, participants were free to read, watch television, listen to music, or complete any other activity that was less than moderate physical effort and not stressful.

Survey Instruments and Administration

All surveys were administered using Qualtrics Core XM software [49] either via a tablet for in-clinic sessions or via email for at-home sessions. Participants were given a brief training on how to self-administer the daily surveys. Participants were also given the option to complete paper versions of the daily logs, if needed. All participants completed the following questionnaires at baseline and exit: the RMDQ [38], the State-Trait Anxiety Inventory (STAI) (Form Y) [50], and the Fear-Avoidance Beliefs Questionnaire (FABQ) [51]. The AHA/ACSM Health/Fitness Facility Pre-Participation Screening Questionnaire [40] and the IPAQ—Short Form [52] were also completed at baseline to assess eligibility for enrollment. The Freiburg Mindfulness Inventory (FMI) [53] was administered prior to the mindfulness training session and again at the exit session for MedExT experimental intervention participants only.

We used a two-prong approach to assess pain in the participants. First, we used quantitative measures of pain (QST) looking at mechanical, thermal, and deep tissue pressure pain. Second, we used visual analog scales (VAS) to measure daily subjective changes in back pain intensity and unpleasantness during the trial. Pain was assessed using QST methods, as well as self-report measures of pain using a VAS consisting of a 10-cm line with the numbers 0 and 10 at either end for intensity and unpleasantness ratings. On each day of the assigned intervention, participants received reminder emails with a link to the daily VAS survey, on which participants would rate pre-intervention and postintervention VAS pain intensity and unpleasantness. This survey was able to capture time stamps of survey progress, allowing protocol compliance to be monitored.

Throughout the 4-week trial period, participants in both groups wore ActiGraph GT9X Link devices (Acti- Graph, Pensacola, FL, USA) to monitor their physical activity (steps per day). To potentially account for noncompliance, we reevaluated the ActiGraph GT9X Link watch data. We were able to monitor activity of all participants in clinic and outside the clinic to estimate compliance with the exercise protocol. Using walking step data from in-clinic sessions as comparison, in addition to Qualtrics survey daily log input from participants (timestamps for starting and finishing the completed intervention), we were able to estimate participation in the walking exercise portion of the intervention for at-home sessions. In a subanalysis (data not shown), we reran analyses of our primary and secondary outcomes for participants who were deemed to be fully compliant and found results similar to those of our analysis on our full data set.

During the exit session, participants were also given an exit survey that was used to identify likelihood of continued adherence (for the MedExT group) and any barriers to this intervention.

Quantitative Sensory Testing

QST was done on the bare skin of the participant’s low back and forearms at specific testing sites. Protocols for the following tests were based off those standardized for testing in low back pain patients [54]. We have previously found strong interrater reliability in these manual testing methods (ICC, 0.86–0.98; P<0.01) [55]. For this evaluation of a pilot intervention, we chose to include QST that covered multiple pain modalities from mechanical sensation to thermal and deep pressure pain to identify the best modalities to measure in a future large-scale clinical trial. These assays assessed each participant’s cutaneous mechanical sensitivity (threshold for mechanical detection for Touch Test filaments (North Coast Medical Inc., Morgan Hill, CA, USA) of 0.008, 0.02, 0.04, 0.07, 0.16, 0.4, 0.6, and 1.0 g in 3 of 5 trials for filament); cutaneous mechanical pain (threshold for mechanical detection for Touch Test filaments of up to 300 g); constant heat pain (45°C heating block of 3 × 5 cm applied for 3 seconds followed by 10-cm VAS for intensity and unpleasantness of pain); pressure pain threshold (1-cm round probe (Wagner Instruments, Greenwich, CT, USA) applied at constant ramping pressure until participant defined cutoff in kilograms at pain threshold); and constant pressure pain sensitivity (2-second pressure stimulus at participant-defined threshold followed by VAS for intensity of pain and unpleasantness of pain) as previously described [55]. Ten-centimeter VAS were numbered at 0 and 10. The score was measured to the nearest millimeter. The intensity scale ranged from 0 (“no pain”) to 10 (“the worst pain imaginable”). The unpleasantness scale ranged from 0 (“not unpleasant”) to 10 (“most unpleasant sensation imaginable”). Testing was performed at baseline and at the end of the trial to measure the overall change in sensitivity across the entire study.

Statistical Analysis

Prior to analysis, an a priori data analysis plan was developed (ClinialTrials.gov under ID NCT03324659). In the following sections, we note changes from this initial plan along with justification. Descriptive statistics were calculated using IBM SPSS, version 25, or JMP and graphed with either SPSS or GraphPad Prism (version 6.0). Normality of the data was evaluated by ensuring that the distribution of each data set was Gaussian. Nonparametric inferential statistics were used for data that were not normally distributed, including mechanical sensitivity and mechanical pain QST assessments for the forearm and low back. We have provided the raw data values in a table (see Supplementary Data  3). Only participants who were determined to have completed 80% of the weekly sessions were included in the data analysis. Data from our exit survey that asked about compliance, prescription use, and preferences for the trial were only qualitatively analyzed.

Primary Outcome

The primary outcome was defined as a change in the RMDQ. This questionnaire was chosen as the primary outcome measure because we recently used it in both a pilot study and a large-scale assessment of MBSR in cLBP patients [16, 20]. Additionally, the RMDQ is most sensitive for patients with mild to moderate disability due to cLBP [38]. A priori, we decided to statistically compare raw RMDQ scores taken at the end of the trial between the MedExT and control groups. However, we recognized the need to assess baseline data between the groups as well. Post hoc to account for raw baseline and postintervention data, a two-way analysis of variance (ANOVA) was used to identify a significant difference between groups and time points using P<0.05, where Sidak’s test was used to identify significant multiple comparisons. To more easily compare these data with many of our secondary outcomes, we also completed a mean change data analysis of the RMDQ scores for the MedExT and control groups using P<0.05.

Secondary Outcomes

Four groups of secondary outcomes were measured and analyzed. P values were adjusted for each group of analyses using a general Bonferroni correction to control for familywise error. To do this, the critical value or alpha for individual tests was calculated by dividing the familywise error rate (0.05) by the number of tests [56]. For example, if performing three tests in one grouping, the critical value would be 0.05/3=0.0167. The first analysis tested whether the MedExT group would significantly increase mean scores on the FMI as determined by a one-sample t test (P<0.05). The second explored whether the MedExT treatment would significantly influence a mean change in responses on three psychological inventories administered: 1) the FABQ; 2) the STAI state anxiety inventory; and 3) the STAI trait anxiety inventory. These were analyzed using one-sample t tests, where P<0.0167 was considered significant. The third group of secondary outcomes measured mean response changes in the series of 14 QST taken at baseline and at the completion of the 4-week intervention period on the low back and nondominant forearm. Analyses were grouped separately for the tested body site. Significant mean pre-intervention and postintervention differences between groups were identified using two-sample t tests. Nonparametric Mann-Whitney rank-sum tests were used for mechanical sensitivity and mechanical pain at each site. Given the number of statistical tests (n=7 per body site) required for the QST secondary outcome measurements, a corrected P<0.0071 was used for each body site to determine statistical significance.

Fourth, we assessed current back pain using a VAS during each day of the trial. These measures were the VAS pain intensity score and the VAS pain unpleasantness score. In our original statistics plan, a repeated-measures multivariate analysis of variance (MANOVA) was to be used to determine if the vector of timed responses was significantly different between the two study groups. Post hoc due to missing data from some days, the MANOVA statistical plan was modified to use a mixed error-component model during analysis of data. JMP was used to perform this analysis. We looked at this in two ways: 1) the overall effect of the 4-week intervention on intensity and unpleasantness ratings (each day’s pre-intervention measurement minus baseline) and 2) the acute effect of each day’s session on VAS ratings (postintervention VAS minus pre-intervention VAS).

Demographic and Background Exercise Variables

The following demographic variables were collected and compared between groups to further check against potential bias: age, gender, handedness, BMI, baseline HR, baseline BP, baseline IPAQ—Short Form scores, and mean number of steps taken per day over the 4-week intervention period (for all participants). This was done using two-sample t tests. The difference in the proportion of gender and handedness was tested using Fisher’s exact test, where P<0.05 was considered significant. All other continuous variables were tested using two-sample t tests for significant differences between the two study groups (P<0.05). Post hoc, we also reevaluated the steps taken per day (accounting for steps in the intervention) on a day-by-day basis. We used a two-way ANOVA with Sidak’s post-test (P<0.05). Data for each day’s steps were not available for all participants, so we have included the data for the first 25 (of 28) days. A continued patient compliance survey was administered at the exit session. Qualitatively, we assessed the need for pain treatments during the study, the likelihood of continuing the intervention practice, barriers to continuing treatment, and the patient’s selection of the most beneficial aspect of the intervention (i.e., exercise, meditation, or both). This was done using a Likert-scale answer format.

Results

Participant Characteristics

Fifty-two adult volunteers (25 women and 13 men aged 18–60) with cLBP were enrolled in this trial, and 38 participated in its entirety. Fourteen participants dropped out of the study after enrollment. This included three due to newly discovered ineligibility (e.g., neurological disorder, respiratory disorder, acute pain) after receiving multiple treatment sessions. These participants dropped out on days 5, 12, and 13, respectively. Ten participants dropped out due to scheduling conflicts and received intake (session 2 with baseline surveys and QST) but never participated in any active intervention sessions. One participant dropped out after 13 scheduled intervention sessions (a total of 19 days) due to an inability to complete the minimum required 80% of sessions per week. See Figure 2 for a flowchart diagram of sessions. Recruitment of participants began in January 2018 and ended in February 2019. There were no reported adverse events for this trial. The demographic and baseline characteristics of the participants are presented in Table 1. Two-sample t tests revealed no significant group differences for any of the demographic variables. A Fisher’s exact test found no significant relationships comparing treatment group with gender and also with handedness (P>0.05). Using the ActiGraph watch data, we compared the average number of steps taken per day during the trial for participants in both groups. After subtracting steps taken by the MedExT group during their 30-minute exercise session, we found no statistically significant difference between the groups (P>0.05). A similar day-by-day analysis of steps taken revealed no significant effect of treatment (P=0.2355), time (P=0.3655), or time × treatment (P=0.5829) (see Supplementary Data  4).

Figure 2.

Consolidated standard of reporting trial (CONSORT) flow diagram.

Table 1.

Participant characteristics

Characteristic	MedExT (n=18)	Control (n=20)	All (N=38)	t Test, P Value
Gender, N
Male	5	7	12	—
Female	13	13	26	—
Baseline characteristics, mean (SD)
Age, y	36.3 (14.1)	38.7 (16.8)	37.6 (15.4)	0.6432
BMI	24.5 (2.9)	26.3 (2.7)	25.4 (2.9)	0.0603
Resting HR, bpm	71.3 (12.4)	72.4 (12.0)	71.9 (12.0)	0.7892
Resting BPs, mm Hg	116.2 (10.7)	115.5 (8.8)	115.8 (9.6)	0.8208
Resting BPd, mm Hg	75.2 (8.1)	77.5 (6.2)	76.4 (7.2)	0.3331
IPAQ, MET min/wk	2,731 (2,463)	2,906 (2,428)	2,821 (2,413)	0.8285
Disability, RMDQ	3.4 (2.7)	4.7 (3.6)	4.1 (3.2)	0.2515
Pain intensity, VAS	3.2 (1.7)	2.7 (2.2)	2.9 (2.0)	0.4454
Pain unpleasantness, VAS	3.3 (1.6)	2.9 (2.5)	3.1 (2.1)	0.6592
Mindfulness, FMI	37.4 (6.7)	—	—	—
Fear avoidance, FABQ	20.0 (10.6)	19.1 (14.5)	19.5 (12.7)	0.8301
State anxiety, STAI	30.4 (6.1)	34.6 (11.4)	32.6 (9.4)	0.1754
Trait anxiety, STAI	34.3 (8.0)	39.0 (12.0)	36.8 (10.4)	0.1666
Steps per day, mean (SD)	10,786 (2,508)	12,030 (3,514)	11,441 (3,103)	0.2220

MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial; SD = standard deviation; BMI = body mass index; HR = heart rate; BP_s = systolic blood pressure; BP_d = diastolic blood pressure; IPAQ = International Physical Activity Questionnaire; MET = metabolic equivalent of task; RMDQ = Roland Morris Disability Questionnaire; VAS = visual analog scale; FMI = Freiburg Mindfulness Inventory; FABQ = Fear-Avoidance Beliefs Questionnaire; STAI = State-Trait Anxiety Inventory.

Steps per day data reflect the trial period.

Primary Outcome: Intervention Effects on Disability

Our primary outcome was the effect of treatment on scores of disability as measured by the RMDQ. Results of a two-way ANOVA indicated no overall effect of treatment (P=0.0598), time (P=0.0904), or time × treatment (P=0.1699) but did indicate a significant effect of participant (P<0.0001). However, Sidak multiple comparisons showed a significant difference between control and MedExT postintervention scores (P=0.0421; Figure 3A). To complement this analysis of the raw data and as a comparison with the secondary analyses, we also calculated a mean change score for each group (postintervention minus baseline). A t test comparing the mean change score for the RMDQ revealed no significant difference between the control and MedExT groups (mean between-group difference, –1.24; 95% CI, –3.1 to 0.6; P=0.1699) (Figure 3B).

Figure 3.

Intervention effect of MedExT vs control treatment on disability as measured by the RMDQ. (A) Raw data for pre- and postintervention scores showed no significant effect of treatment; however, there was a significant difference between the control and MedExT disability scores postintervention. (B) Mean change score data showed no significant difference between groups. Mean between-group difference, –1.24; 95% CI, –3.1 to 0.6. Data shown as mean ± SEM. RMDQ minimum score=0; maximum score=24. MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial; RMDQ = Roland Morris Disability Questionnaire; CI = confidence interval; SEM = standard error of the mean. *P=0.0421.

Secondary Outcomes: Mindfulness, Fear Avoidance, Anxiety, and Pain

In this study of a pilot intervention, we evaluated a number of secondary outcomes directly related to the efficacy of the intervention as well as those related to pain in the participants and the associated symptoms. We used the FMI to determine if there were any changes in mindfulness that developed during the trial. A one-sample t test revealed a significant increase in mindfulness for the MedExT group from pre- to postintervention (P=0.0141; Table 2). One-sample t tests showed no significant differences between pre-intervention and postintervention measures for the MedExT group for the FABQ (P>0.0167), STAI state anxiety inventory (P>0.0167), or STAI trait anxiety inventory (P>0.0167) (Table 2). No differences were observed from pre- to postintervention for the control group for any of these measures (data not shown).

Table 2.

Secondary outcome results for control and MedExT groups

Outcome Measure	MedExT			Control			t Test/Mann-Whitney, P
	Pre-Intervention	Postintervention	Mean Change	Pre-Intervention	Postintervention	Mean Change
Mindfulness, FMI	37.44 (6.7)	41.11 (8.2)	3.67 (6.5)	NP	NP	NP	0.0141
Psychological inventories
Fear avoidance, FABQ	20.00 (10.6)	19.11 (10.1)	–0.89 (9.0)	19.10 (14.5)	21.45 (14.9)	2.35 (10.0)	0.6812
State anxiety, STAI	30.39 (6.1)	33.28 (10.1)	2.89 (10.5)	34.55 (11.4)	36.80 (10.7)	2.25 (8.1)	0.2577
Trait anxiety, STAI	34.28 (8.0)	33.94 (7.6)	–0.33 (5.6)	39.00 (12.0)	39.50 (13.5)	0.50 (5.7)	0.8049
QST—Low back
Mechanical sensitivity, g	0.49 (0.4)	0.34 (0.3)	–0.15 (0.4)	0.52 (0.2)	0.45 (0.3)	–0.08 (0.3)	0.6107
Mechanical pain, g	110.45 (132.8)	48.48 (78.9)	–61.98 (129.4)	56.43 (98.9)	52.86 (93.9)	–4.57 (75.5)	0.1078
Constant heat VAS intensity	3.84 (2.2)	2.88 (1.8)	–0.97 (1.9)	3.43 (2.2)	3.26 (2.4)	–0.17 (2.3)	0.2519
Constant heat VAS unpleasantness	4.57 (2.6)	3.37 (2.3)	–1.21 (2.1)	3.49 (2.8)	3.61 (2.8)	0.13 (2.2)	0.0635
Pressure pain threshold, kg	5.58 (1.8)	5.70 (1.6)	0.12 (1.9)	4.71 (2.0)	4.78 (1.8)	0.07 (1.4)	0.9236
Constant pressure VAS intensity	3.48 (1.6)	2.29 (1.4)	–1.19 (1.5)	2.83 (1.7)	2.51 (1.7)	–0.27 (1.6)	0.0746
Constant pressure VAS unpleasantness	3.92 (2.1)	2.45 (1.5)	–1.47 (1.6)	2.90 (1.7)	2.73 (2.1)	–0.17 (2.0)	0.0338
QST—Forearm
Sensitivity, g	0.28 (0.2)	0.27 (0.2)	–0.01 (0.2)	0.42 (0.2)	0.29 (0.2)	–0.13 (0.3)	0.1361
Mechanical pain, g	70.17 (101.3)	57.20 (99.4)	–12.97 (87.0)	65.66 (105.6)	62.66 (92.9)	–3.00 (67.9)	0.4259
Constant heat VAS intensity	1.76 (1.5)	1.56 (1.1)	–0.20 (1.2)	1.66 (1.2)	1.35 (1.1)	–0.32 (0.7)	0.7210
Constant heat VAS unpleasantness	1.98 (1.7)	1.75 (1.4)	–0.23 (1.6)	1.74 (1.9)	1.22 (1.3)	–0.52 (0.8)	0.4789
Pressure pain threshold, kg	3.63 (0.9)	3.93 (1.0)	0.30 (0.6)	3.54 (1.4)	3.53 (1.3)	–0.01 (0.8)	0.2104
Constant pressure VAS intensity	2.55 (1.6)	2.04 (1.6)	–0.51 (1.7)	2.03 (1.4)	1.57 (1.2)	–0.46 (1.1)	0.9141
Constant pressure VAS unpleasantness	3.11 (1.5)	2.56 (1.9)	–0.55 (2.1)	2.58 (2.1)	1.87 (1.6)	–0.71 (1.5)	0.7889

MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial; FMI = Freiburg Mindfulness Inventory (range, 14–56); NP = test not performed; FABQ = Fear-Avoidance Beliefs Questionnaire (range, 0–96); STAI = State-Trait Anxiety Inventory (range, 20–80); QST = quantitative sensory testing; VAS = visual analog scale.

All data are shown as mean (SD). P values for mindfulness and psychological inventories are a result of t tests comparing MedExT pre-intervention and postintervention scores. P values for QST outcomes are a result of comparisons between MedExT and control mean change scores. ∗P < 0.05 statistically significant difference for FMI. For all psychological inventories, P>0.0167, which was the threshold for a statistically significant difference. For all QST, P>0.0071, which was the threshold for a statistically significant difference.

Body site–specific QST pain data for each test are provided in Table 2. For the low back and forearm, two-sample t tests found no significant effects of treatment for constant heat pain intensity, constant heat pain unpleasantness, pressure pain threshold, constant pressure pain intensity, or constant pressure pain unpleasantness (P>0.0071). Additionally, nonparametric Mann-Whitney tests showed no significant differences between treatment for mechanical sensitivity or mechanical pain for either the low back or forearms (P>0.0071).

For VAS repeated measures of ongoing back pain, we found analgesic effects of the intervention that appeared to accumulate over time (Figures 4A and 4B). A mixed-effects model revealed a significant effect of time (P=0.0008) and time × treatment (P=0.0012) for intensity ratings on each day before undergoing the intervention session compared with baseline (Figure 4A). For unpleasantness ratings before the intervention, a mixed-effects model showed significant effects of time (P=0.0022) and time × treatment (P<0.0001) compared with baseline (Figure 4B).

Figure 4.

Intervention and acute effects of MedExT intervention compared with control. Data shown as mean ± SEM, with “analgesic” responses being values lower than 0 on y-axes. Intervention effects are shown in A and B comparing the VAS measurement taken immediately prior to each day’s session vs the baseline VAS measurement taken on intake day. Statistically significant analgesic effects were seen in the MedExT group for (A) VAS intensity (time × treatment [∗∗P=0.0012]; time [***P=0.0008]; and (B) VAS unpleasantness (time [**P=0.0022]; time × treatment [****P<0.0001]). Acute intervention effects shown in C and D comparing the VAS measurement taken after each day’s intervention to the VAS measurement taken immediately before the intervention. No significant differences were found for (C) VAS intensity (n.s.) nor (D) VAS unpleasantness (n.s.). MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial; SEM = standard error of the mean; VAS = visual analog scale; n.s. = not significant.

We also analyzed the acute effect of each day’s session by comparing the difference in VAS measurements between the pre-session measurement and a measurement made immediately following each day’s session. We predicted that an acute analgesic effect would show a reduction in VAS measurements each day in the MedExT group. Analysis of acute day-to-day effects of the intervention indicated no significant effects for intensity (Figure 4C)—i.e., the intensity the VAS measured immediately after the approximately 45-minute session was not significantly different from the VAS measured immediately before that day’s session. When we evaluated acute day-to-day effects on VAS unpleasantness, we similarly found no significant effects (Figure 4D).

Exit Survey Data

At the exit session, all patients were asked to complete a continued patient compliance survey. The results of these outcomes are shown in Figure 5 and Table 3. This survey sought to identify the need for pain treatments during the study (Figure 5); the likelihood of continued compliance after the study (Figure 5); any barriers to continuing the combined treatment (Table 3); and the most beneficial aspect of the intervention among meditation, exercise, or the combination (Table 3). Our qualitative data show that MedExT participants reported a greater decrease in pain medication use and seemed fairly likely to continue the intervention. When MedExT experimental participants were asked to identify the most beneficial aspects of the intervention (i.e., meditation, exercise, or both), over half of the participants stated that both components of the intervention were the most important.

Figure 5.

Continued patient compliance survey data—qualitative results of each outcome. Box plots are shown to represent data. Box represents IQR (bottom = Q1, middle = median, top = Q3). Whiskers represent range of data (minimum and maximum) (median = Q1 for MedExT and control for question 1). IQR = interquartile range; MedExT = Meditation and Exercise to Treat Chronic Low Back Pilot Trial.

Table 3.

Continued patient compliance survey data

Outcome Measure	No. of Responses
Barriers to continuing treatment
Time to complete both INTs	8
Motivation to complete both INTs	10
Pain with both INTs	0
No reduction in symptoms while doing both INTs	1
No access to a treadmill to complete the exercise	1
No access to guided meditations	3
Other	3
Most beneficial aspect
Meditation only	1
Exercise only	7
Both	10

INT = intervention.

Self-reported barriers to continued treatment are shown for the Meditation and Exercise to Treat Chronic Low Back Pilot Trial (MedExT) participants. MedExT participants identified what the most beneficial aspect of the intervention was between three choices: meditation only, exercise only, or both combined.

Discussion

In the current pilot study, we assessed the effect of a combined intervention of mindfulness meditation followed by aerobic walking exercise in patients with cLBP. The main findings of this study indicate that meditation and exercise together were able to increase mindfulness and decrease self-reported ratings of low back pain from baseline. Although the MedExT was able to reduce disability compared with the control, it is difficult to interpret this finding in light of the lack of other significant results for disability. To our knowledge, this specific therapy combination has not yet been tested in cLBP patients.

Although the present study took a unique approach to combined mindfulness and aerobic exercise, there is robust literature suggesting that such an approach could work. First and foremost, previous studies have tested MBSR [16–20, 57] and mindfulness meditation [58] alone, as well as aerobic walking exercise programs [24, 25, 59–64] in low back pain patients. Overall, these studies found improved disability, sleep quality, psychological function, depression, affective pain perception, fitness, and pain severity and reduced need for pain medications. One important aspect of the mindfulness used in the present study was the accessibility of the mindfulness. Beyond the introductory 45-minute session with a clinical psychologist, our participants were naive meditators. Nonetheless, using only five short recordings repeatedly, their mindfulness increased, as assessed by the FMI. Gains seen in this study via the FMI can be compared with more intensive training exercises [65]. Although the recordings used were curated for their emphasis on mindfulness, they were not specifically recorded for this intervention. We would hypothesize that the development of a mindfulness recording that specifically prepared participants for the subsequent exercise could be even more beneficial. The benefits seen with a brief meditation program are consistent with more recent data showing that only 4 days of mindfulness-based mental training can reduce pain [48, 66, 67]. These previous studies were done in healthy participants with models of acute nociception, whereas in this study we are showing gains in mindfulness in a cLBP patient population.

For our primary outcome (disability), we found some evidence for positive trends for disability in the MedExT group. Compared with postintervention values in the control group, MedExT participants had lower disability. The trends that we saw were likely minimized by the lower starting values of participants in both groups. Many studies include a minimum disability score during inclusion. Morone et al. [16] recruited patients who had functional limitations related to their cLBP. This was defined as a score of 11 or higher on the RMDQ. For the purposes of this study, we chose not to have a defined cutoff for disability in order to recruit a wider sample of patients, recognizing the potential barriers to the demanding physical nature of our exercise intervention. Perhaps not surprisingly, six of the participants reported scores of 0 on the RMDQ pre-intervention (Figure 3A), and only two participants reported a score higher than 11.

Surprisingly, although MedExT participants rated lower disability along with lower ongoing pain and lower average pain compared with their ratings at the start of the trial, they failed to show any changes in fear avoidance behavior or anxiety. This result is in contrast to data generated from a similar study that implemented combined mindfulness and exercise in the context of major depression [68]. That study found that 8 weeks of 60-minute mental and physical training twice a week significantly reduced depressive symptoms and ruminative thoughts. It is possible that the lack of a significant effect on fear and anxiety in the present study was driven by the lower starting anxiety and fear levels in our cohort of participants. We anticipate that the anxiolytic effects of the combined intervention may only present themselves in the context of higher baseline anxiety and fear avoidance behavior or with longer-duration studies (e.g., 8 vs 4 weeks).

During the trial, participants rated pain before and after each session. Thus, we were able to track their ongoing pain before each intervention compared with baseline and evaluate the potential for acute analgesic effects of the intervention itself. Analysis of day-to-day pre-intervention ratings of ongoing low back pain revealed significant effects of time and time × treatment for intensity and unpleasantness ratings. These data are indicative of a time-dependent effect of the intervention, with beneficial outcomes resulting from cumulative repeated treatment sessions. Interestingly, these data do not begin to show separation across time between the groups until approximately day 8 for pain intensity ratings and day 10 for pain unpleasantness ratings. This suggests that for this specific intervention, a sustained analgesic benefit can only be achieved after 8–10 sessions or approximately 2 weeks of effort.

Acutely, however, there were no major changes comparing the pre-exercise and rest session to the postexercise and rest session on single days. The lack of an observed acute effect of the exercise intervention on pain intensity and unpleasantness argues against a pure expectancy effect in this trial. One would anticipate that expectancy related to a placebo effect would occur each day with the meditation and exercise intervention.

Our data show significant differences in qualitative ratings of low back pain but no significant effects with QST. Interestingly, we do see trends for decreased ratings of induced pain unpleasantness (P=0.0338) in response to a noxious constant pressure stimulus applied to the low back. This trend was observed even though there were no trends for a change in participants’ pressure pain threshold (i.e., thresholds did not change, whereas the perception of the stimulus decreased slightly). These paradoxical effects are consistent with those of our previous study that tested aerobic treadmill walking exercise on painful QST in healthy participants [31]. Similar effects have also been shown in sustained aerobic exercise, where training induced increases in pain tolerance but did not alter the pressure pain threshold [69].

The results of our continued compliance survey given during the exit session show favorable outcomes for the combined meditation and exercise intervention. In particular, the finding that a majority of MedExT group participants found that the most beneficial component of the intervention was the combination of exercise and meditation suggests the potential for synergistic effects in this study. The need for pharmacological pain treatments decreased for the MedExT group compared with the control participants, which suggests an analgesic benefit of a solely nonpharmacological source. MedExT participants also indicated that they would be likely to continue the combined intervention on their own to manage their back pain symptoms.

Strengths and Limitations

Notable strengths of this study include the accessibility of the intervention, a low risk of detection bias through blinding of the outcome assessor for QST, and a lack of confounding demographic variables. Treadmills were easily accessible to patients both in the clinic and at home or at a local gym. The average age of study participants was 37.6 years, which accurately represents the range of our inclusion criteria [18–60]. Additionally, after subtracting intervention walking steps, there was no statistical difference in average steps per day between the control and MedExT participants. The patients in this trial were mostly women (n=26) compared with men (n=12); however, cLBP is reported to be more prevalent among women [3].

The most significant limitation to this pilot study is the lack of all possible study arms. Without exercise-only and meditation-only groups, the possibility of synergy between the individual therapies cannot be specifically assessed. A future larger-scale trial with all study arms would be necessary to address these questions. Another limitation is the small sample size associated with a pilot study. This study had a higher risk of performance bias due to lack of blinding of participants during the intervention, which is very difficult due to the nature of the active intervention. Although the exclusion criteria were selected to maintain the safety of study participants in completing at-home exercise sessions, this was a significant limitation to the generalizability of our findings, particularly in reference to obese individuals with cLBP. To speak further to the generalizability of our study, it is worth noting that our sample was younger in age and had readily accessible treadmills, which may not be the case with the totality of the cLBP population. Notably, a total of 14 participants (27% of the 52 randomized) were lost to follow-up, contributing to reduced statistical power and a possible threat to randomization. Selective data collection in our study could also be seen as a limitation. However, only one participant was excluded from our reported results for not completing 80% of the intervention sessions.

Another limitation of the study was the potential of a placebo effect in the outcome. Placebo effects are critical to consider in any intervention but are difficult to account for in the context of physical and behavioral interventions [70]. Placebo is a complex biopsychosocial phenomenon that includes factors such as time with the experimenter or clinician, selection bias, expectation of efficacy, context, and many other factors. Our neutral audiobook control condition was selected from a previously published work indicating that this condition has distinctly different neural mechanisms compared with mindfulness meditation in the context of providing acute pain relief in healthy participants [48]. We chose this control condition to control for the total amount of time with the clinical experimenter (at least one session per week just like the MedExT group), the time of intervention trial per day, the effect of listening to quiet spoken audio (audiobook vs mindfulness recordings), natural history (total 4-week study for all participants), and experience with our study psychologist (the control condition received a stress management session with our psychologist as a comparison treatment to the mindfulness training session that our MedExT participants received). Of course, there are limitations to this control group. Although we did not disclose the possible effects of the MedExT treatment, participants were informed of the possible study groups that they could be randomly assigned to during the consent process, leading to the potential of an expectation bias. However, similar studies have found the expectation of programs to reduce back pain to be equal between the control and intervention groups [16]. Other types of control groups will be considered in future iterations, including placebo cream controls and/or sham mindfulness controls. Another limitation related to the control group was the fact that the FMI survey was not administered to this group. Although we expect that the control group’s mindfulness scores would be unchanged from pre-intervention to postintervention, the absence of this information makes the within-group change in the MedExT group more difficult to interpret.

According to our continued compliance survey, the most prominently identified barriers to continuing this treatment after conclusion of the trial included time and motivation to complete both interventions. No patients reported increased pain with both meditation and exercise. Thus, although we cannot state conclusively that there was synergy between the treatments, we are confident that the therapies are not overtly antagonist. Other reported barriers to continuing the therapy included lack of access to a treadmill or guided meditations; however, access to additional meditations was provided to interested patients following their completion of the trial.

Conclusions and Future Implications

In conclusion, a 4-week pilot intervention of mindfulness meditation followed by moderate-intensity treadmill walking for patients with cLBP demonstrated trends of reduced disability compared with the control group and increased mindfulness and reduced perceptions of low back pain in MedExT participants compared with baseline. According to qualitative reports, patients in this treatment group report less need for pain medications and greater favorability for the combined approach as opposed to meditation or exercise therapy alone. To our knowledge, this pilot study is the first study testing this treatment combination in cLBP patients. Moreover, this pilot trial has identified a number of variables that will be adjusted in a large-scale trial, including restricting QST testing (pressure pain only), highlighting the importance of specific minimum disability cutoffs for low back pain, and demonstrating the need for more active and control comparison groups. Because synergistic benefits could not be definitively determined from this trial, future studies must be done to ascertain the most efficacious combination of this treatment regimen. Our laboratories are currently investigating this question. Nonetheless, we believe that the potential for this combined approach to improve outcomes for patients with cLBP is high. As exercise and meditation (as practiced in this study) are low cost, easy to implement, and carry few negative side effects, we are optimistic about the use of this or similar integrative therapy in clinic settings.

 

Funding sources: Funding for this work was supported by the National Institutes of Health through Grant UL1TR001857 and through a Pain Research Challenge Grant supported by the Clinical and Translational Science Institute at the University of Pittsburgh and the Virginia Kaufmann Foundation.

Conflicts of interest: The authors have no conflicts of interest to disclose.