Changes in Perceived Stress After Yoga, Physical Therapy, and Education Interventions for Chronic Low Back Pain: A Secondary Analysis of a Randomized Controlled Trial

Abstract

Objective

Perceived stress and musculoskeletal pain are common, especially in low-income populations. Studies evaluating treatments to reduce stress in patients with chronic pain are lacking. We aimed to quantify the effect of two evidence-based interventions for chronic low back pain (cLBP), yoga and physical therapy (PT), on perceived stress in adults with cLBP.

Methods

We used data from an assessor-blinded, parallel-group randomized controlled trial, which recruited predominantly low-income and racially diverse adults with cLBP. Participants (N = 320) were randomly assigned to 12 weeks of yoga, PT, or back pain education. We compared changes in the 10-item Perceived Stress Scale (PSS-10) from baseline to 12- and 52-week follow-up among yoga and PT participants with those receiving education. Subanalyses were conducted for participants with elevated pre-intervention perceived stress (PSS-10 score ≥17). We conducted sensitivity analyses using various imputation methods to account for potential biases in our estimates due to missing data.

Results

Among 248 participants (mean age = 46.4 years, 80% nonwhite) completing all three surveys, yoga and PT showed greater reductions in PSS-10 scores compared with education at 12 weeks (mean between-group difference = −2.6, 95% confidence interval [CI] = −4.5 to −0.66, and mean between-group difference = −2.4, 95% CI = −4.4 to −0.48, respectively). This effect was stronger among participants with elevated pre-intervention perceived stress. Between-group effects had attenuated by 52 weeks. Results were similar in sensitivity analyses.

Conclusions

Yoga and PT were more effective than back pain education for reducing perceived stress among low-income adults with cLBP.

Introduction

Perceived stress occurs when individuals experience environmental demands that exceed their adaptive capacity [1]. Low-income and racially diverse populations are a high-risk group for perceived stress due to intersecting risk factors such as traumatic life experiences (e.g., violence, racism), work-related factors (e.g., high job demand, occupational injuries), and financial insecurity [2–5]. Perceived stress is associated with the onset and progression of many chronic conditions [6, 7], and its significant relationship with the onset and prevalence of persistent musculoskeletal pain has been described in various populations [8–10]. Low back pain is a common source of persistent musculoskeletal pain. Nearly all adults will experience at least one episode in their lifetime, and rates of chronic low back pain (cLBP) are higher in low-income populations [11, 12]. While perceived stress can increase the likelihood of acute low back pain becoming chronic [10], chronic pain also can generate ongoing stress [13]. Additionally, among adults with low-level or stable cLBP, a stressful life event may precipitate “flares” of back pain symptoms, which impact usual activities [14, 15]. While mind and body therapies contain components (e.g., meditation, aerobic exercise) that may alleviate both pain and perceived stress [16], few studies have evaluated treatments to reduce stress in patients with cLBP. Our study examined whether yoga and physical therapy (PT), two evidence-based cLBP treatments [17, 18], may reduce perceived stress in low-income, racially diverse adults with cLBP.

Yoga typically consists of physical poses, breathing exercises, and meditation. Randomized controlled trials (RCTs) suggest that yoga reduces perceived stress in healthy individuals [19, 20], workers with either back pain or stress [21], and among populations known to experience chronic pain, including cancer patients [22], sedentary adults [23], and workplace employees [24]. Clinical studies of yoga on perceived stress are limited by small sample sizes, lack of randomization, nonstandardized interventions, short follow-up periods, and variations in outcome measurement [20]. Few studies have examined yoga for perceived stress among low-income populations or adults living with cLBP. One large trial of yoga for cLBP found that reductions in perceived stress midintervention did not independently mediate improvements in physical function [25], and end-of-intervention effects on stress were not published [26]. Thus, whether patients receiving yoga for cLBP also experience changes in perceived stress remains unknown.

PT is a common treatment for cLBP and typically involves stretching, strengthening, and aerobic exercise in a one-on-one setting with a physical therapist, who supports and encourages the patient [17]. Physical therapists commonly manage painful musculoskeletal conditions, but little attention has been dedicated to understanding non-pain-related stress in the pain experience [13]. While aerobic exercise, a fundamental component of PT, can reduce patients’ sensitivity to stress [27] and improve stress overall [28, 29], we are unaware of any RCTs that have examined the effect of PT on perceived stress. Given the biological links between chronic stress and pain, physical therapists have been encouraged to screen for perceived stress and to educate patients on the role of stress in pain to reduce fear-based responses [13].

To address these gaps in knowledge, we used data from the Back to Health Study, a recent large RCT demonstrating that yoga was noninferior to PT for improved pain and back-related function in low-income and racially diverse adults with cLBP [30]. We evaluated the effect of yoga and PT on perceived stress at both 12 weeks and 52 weeks. We hypothesized that yoga and PT participants would experience greater improvements in perceived stress postintervention compared with an education control group and that this improvement would be sustained over the follow-up period.

Methods

Study Design

The Back to Health Study protocol [31] and primary results [30] have been described in detail elsewhere. The original trial was registered (NCT01343927), and all study procedures were approved by the Boston University Medical Campus (BUMC) Institutional Review Board. Briefly, the current study is a secondary analysis of an assessor-blinded, parallel-group RCT of 320 participants recruited from a large academic safety-net hospital and seven affiliated federally qualified community health centers between June 2012 and November 2013. Follow-up ended in November 2014. Participants were predominantly low-income, racially diverse, English-speaking adults aged 18 to 65. Inclusion criteria included low back pain lasting ≥12 weeks and pain intensity rated ≥4 on a 0–10 numeric pain rating scale. Persons with specific causes of cLBP (e.g., spinal canal stenosis, sciatica where leg pain was greater than back pain) were excluded. All participants provided informed consent.

Participants were randomized using a computer-generated sequence in a 2:2:1 ratio to group yoga, one-on-one PT, or a back pain education book, respectively [31]. After the 12-week intervention phase, participants in the yoga and PT groups were further randomized for a 40-week maintenance phase. Participants in the yoga arm were randomized to either a weekly drop-in yoga class or home practice only; PT participants were randomized to either five “booster sessions” or home practice only. Because the maintenance phase groups were not significantly different in pain and function at 52 weeks [30], we collapsed maintenance groups for the purpose of these analyses.

Interventions

A manualized hatha yoga program [31] based on previous studies [32, 33] consisted of weekly 75-minute classes over the 12-week treatment phase. Classes focused on yoga postures complemented by relaxation and breathing exercises. Low participant–instructor ratios (<5:1) allowed instructors to provide prespecified modifications to accommodate different abilities.

A manualized, high-dose PT intervention [31] consisted of 15 hour-long sessions over the treatment phase delivered according to the treatment-based classification system [34]. Each session consisted of one-on-one work with a physical therapist and supervised aerobic exercise. Additionally, participants who scored high on the Fear-Avoidance Beliefs Questionnaire were given a copy of The Back Book [35], a brief resource that addresses psychosocial comorbidities that can negatively affect prognosis.

The education group received The Back Pain Helpbook [36], which has been used as a credible control arm in previous studies of back pain treatments [37, 38]. The Back Pain Helpbook comprehensively discusses cLBP and self-management strategies such as stretching, exercise, and lifestyle changes. Participants were provided a reading schedule and received check-in calls from study staff every three weeks. Newsletters were also sent to summarize each section.

Data Collection

Participants were administered the 10-item Perceived Stress Scale (PSS-10) [1] at baseline, 12 weeks, and 52 weeks. The PSS-10 is an abbreviated version of the 14-item Perceived Stress Scale and has high validity and reliability. Participants rate the frequency with which they experienced 10 stress symptoms over the preceding 30 days on a five-point scale (0 = never, 4 = very often). Total scores range from 0 to 40, with higher scores indicating greater perceived stress. A cutoff of 17 for elevated perceived stress was selected because it is above all four estimates for the true population mean (13.0–16.1) derived from large studies (N = 1,143–2,387) [39–41]. This cutoff coincided with the median baseline PSS-10 score in our data set.

Sociodemographic characteristics included age, sex, race (white, nonwhite), ethnicity (Hispanic, non-Hispanic), employment status (working, nonworking), and annual income (≤$30,000, >$30,000). Clinical characteristics included body mass index (BMI), baseline back-related pain intensity measured by an 11-point numerical scale [42], and function as assessed by the Roland-Morris Disability Questionnaire (RMDQ) [43]. The Generalized Anxiety Disorder (GAD-7) and Patient Health Questionnaire (PHQ-8), measures of anxiety and depressive symptoms, respectively, were also reported.

Data Analysis

Baseline characteristics of the participants in each intervention group were compared using analyses of variance (ANOVA) for continuous variables and chi-square tests for categorical variables. If baseline differences existed (P < 0.05), these variables were adjusted for in our analyses. Our analytic sample included participants with complete data at 12 and 52 weeks. For participants who only answered nine of the PSS-10 questions, a total score was imputed using mean item substitution [44]. Adherence was defined a priori as attending at least nine yoga classes, 11 PT appointments (≥75% or ≥73% of sessions, respectively), and self-report of reading three-fourths of the assigned self-care book.

We calculated within-group PSS-10 differences from baseline to 12- and 52-week follow-up. Thus, negative values represent improvements in perceived stress. We calculated between-group differences in PSS-10 change scores at each follow-up using the education group as a reference. We performed an analysis of covariance (ANCOVA) to test between-group differences in 12- and 52-week changes in perceived stress after adjusting for age, gender, and baseline RMDQ score. Results were presented as mean change scores with 95% confidence intervals. We also computed Pearson correlation coefficients between baseline perceived stress and both back pain intensity and RMDQ scores. These calculations were repeated between change scores at 12 and 52 weeks.

Sensitivity analyses allowed us to estimate the effect of yoga and PT in two larger samples: participants who completed at least one survey at either 12 or 52 weeks and all 320 participants using multiple imputation. Estimates from these analyses were compared with results of the primary analysis (complete case) to assess potential bias.

Our sample included participants with low levels of stress who may have been excluded if the original study’s primary aim had been to evaluate changes in perceived stress. Thus, we repeated the above analyses separately in adult participants defined a priori as having elevated (PSS-10 ≥ 17) pre-intervention levels of perceived stress. We used published national median PSS-10 scores [39–41] to a priori establish the cutoff point.

All analyses were conducted using SAS Studio, version 3.7.

Results

Among 248 participants with complete data, most were female (71%), middle-aged (mean age = 46.4 ± 10.4 years), nonwhite (80%), and low-income (59%) (Table 1). There were significant between-group differences in baseline RMDQ (P = 0.03) and gender (P = 0.02). Baseline PSS-10 scores (mean = 16.9 ± 6.7) did not vary significantly by treatment group. The 72 excluded participants were similar in most baseline characteristics, except that they were less likely to be white (11%) and female (40%). For the subset analysis, the entire sample (N = 320, mean PSS-10 = 17.3 ± 7.0) was split into elevated (N = 174, mean PSS-10 = 22.5 ± 4.7) and lower (N = 146, mean PSS-10 = 11.2 ± 3.7) pre-intervention perceived stress groups (Supplementary Data). The numbers of participants included in the complete case analysis and each of the sensitivity analyses are shown in Figure 1.

Adherence in the treatment and maintenance phases has been described previously [30]. Of the participants with complete data, the median treatment phase yoga and PT attendance (interquartile range) was eight (4–11) classes and 10.5 (6–13) appointments, respectively. Based on an a priori definition of adherence, the yoga and PT groups both had 49% adherence, while education was at 44%.

At 12 weeks, participants in the yoga and PT treatment groups, but not the education group, reported significant reductions in perceived stress (Figure 2, Table 2). After adjusting for age, gender, and baseline RMDQ score, participants in these groups improved significantly relative to those receiving education (yoga: mean between-group difference = −2.6, 95% confidence interval [CI] = −4.5 to −0.66; PT: mean between-group difference = −2.4, 95% CI = −4.4 to −0.48). Among participants with elevated pre-intervention stress, we observed larger between-group differences for both yoga and PT compared with education (yoga: mean between-group difference = −3.4, 95% CI = −6.0 to −0.77; PT: mean between-group difference = −3.2, 95% CI = −6.0 to −0.54).

At 52 weeks, we observed within-group improvements among all three groups, with statistically significant improvements in perceived stress for yoga and PT only (Figure 2, Table 2). The effects of yoga and PT relative to education at 52 weeks were attenuated and no longer statistically significant (mean between-group difference = −2.1, 95% CI = −4.45 to 0.35, and mean between-group difference = −0.93, 95% CI = −3.4 to 1.5, respectively). Similar 52-week outcomes were observed in those with elevated pre-intervention PSS-10 scores.

Sensitivity analyses showed similar results at 12 and 52 weeks (Figure 1; Supplementary Data). The magnitude of the effect of yoga and PT compared with education at 12 weeks was partially attenuated by half of a point in analyses using multiple imputations for missing data but remained significant.

Perceived stress exhibited a nonsignificant positive relationship with reported back pain intensity (r = 0.12, P = 0.06) and a weak negative relationship with function (r = −0.17, P = 0.007) at baseline. Slightly stronger relationships were also observed between changes in stress and changes in pain and function at 12 weeks (r = 0.32 and −0.29, respectively, P < 0.001) and 52 weeks (r = 0.36 and –0.27, respectively, P < 0.001).

Discussion

In a large sample of predominantly low-income adults with cLBP, we observed a larger reduction in perceived stress among those randomized to yoga or PT interventions, compared with a back pain education control group at 12 weeks. The effect of yoga and PT on perceived stress was larger among adults with elevated pre-intervention perceived stress. While improvements in stress were maintained over the 52-week period, the effect of yoga and PT compared with education had attenuated by 52 weeks and was no longer statistically significant. Sensitivity analyses using several imputation methods for missing data observed similar effects of yoga and PT at 12 and 52 weeks.

Our findings are consistent with clinical trials indicating modest to moderate improvements in perceived stress from yoga. Of eight small RCTs (N = 59–100) evaluating the effect of similar yoga interventions using the PSS-10, seven detected significant improvements ranging from −2.7 to −7.8 points in populations including workers with either back pain or stress [21], pregnant women [45, 46], cancer patients [47], health profession students [48], and others [23, 49]. The nonsignificant RCT found only a small reduction (–1.5 points) in perceived stress among 100 nurses [50]. One larger study (N = 239) of yoga in a high-stress workplace population found a large reduction in perceived stress after 12 weeks (−8.2 points) [24]. The smaller effect observed in our study may be in part due to our lower mean baseline PSS-10 score (16.9 vs 24.9). Indeed, when we restricted analyses to 56 participants with elevated perceived stress (mean baseline PSS-10 of 22.7), we observed a larger reduction in PSS-10 at 12 weeks and 52 weeks of −3.7 and −5.5, respectively. Low adherence rates in yoga and PT interventions may have also contributed to modest treatment effects observed in both groups.

Although no studies, to our knowledge, have assessed PT’s effect on perceived stress, the effect of aerobic exercise, a major component of our PT intervention, is associated with lower stress levels [29, 51]. However, fewer studies have evaluated aerobic exercise as a stress reducer in a chronic pain population. Individuals with pain may avoid exercise because it exacerbates pain or they fear re-injury [52]. Stress itself may prevent people from being physically active [53]. Structured supportive interventions are needed to overcome stress- or pain-related barriers for individuals to start or progress through an exercise routine. Among studies with smaller samples (N = 50–151), the effectiveness of aerobic exercise interventions for reducing perceived stress has not been demonstrated [54–56].

One potential mechanism by which yoga or PT may improve perceived stress in adults with cLBP is through reducing pain and improving function. However, we found that improvements in pain and function were only weakly correlated with improvements in perceived stress, suggesting other aspects of these treatments may contribute to reducing perceived stress. Yoga has several components that are thought to reduce stress, such as meditation, rhythmic breathing, and stretching techniques, as well as social support [57]. Our PT intervention included screening for fear-avoidance beliefs, a treatment-based classification algorithm, and one-on-one education and support from the physical therapist. Studies have shown that a strong patient–therapist relationship, defined by a sense of collaboration, warmth, and support, is associated with reduced pain and disability and higher patient satisfaction [58, 59]. These aspects of PT may also help account for the improvements in perceived stress at 12 weeks, as the provider–patient relationship may provide unique opportunities to tailor care and reassure the participant, for example, addressing fear-avoidance beliefs that can perpetuate stress [52].

This study had several limitations. The complete case analysis did not maintain randomization. Given that this was a secondary analysis, the yoga and PT interventions tested in this trial were designed to target cLBP, not perceived stress. Modifying these interventions may enhance their effect on perceived stress, for example, a greater focus on mindfulness, meditation, or coping skills. As minimum pre-intervention levels of perceived stress were not required at study entry, participants with very low pre-intervention PSS-10 may have had less opportunity to improve due to floor effects. The PSS-10 does not ask participants to identify the source(s) of their stress (e.g., a traumatic life experience, work-related stress, or stress from their chronic pain experience). Therefore, we were unable to assess whether distinct sources of stress modify the effects of yoga or PT on stress outcomes. Finally, this trial did not include objective assessments of chronic stress biomarkers (e.g., salivary cortisol) that might elucidate potential psychoneuroendocrine mechanisms underlying yoga- and PT-related health benefits.

Study strengths include use of a validated measure of perceived stress, multiple time points of measurement over 52 weeks of follow-up, and a socioeconomically diverse sample that is particularly vulnerable to stress and cLBP [11, 60]. We also conducted sensitivity analyses using various imputation methods due to missing data, which obtained similar results.

Our findings may have public health and clinical implications. Back pain and mental health conditions are leading causes of disability worldwide [61]. Recent findings from the National Health Interview Survey show an exponential increase in the use of yoga in the United States from 2002 to 2017, with many Americans reporting yoga to be helpful for pain and mental health conditions [62]. As yoga can be done in a group with fairly minimal equipment, it would conceivably be a fairly low-cost option for patients with co-occurring stress and cLBP in low-income areas. However, further research is needed to determine whether our results are generalizable to community yoga, which is nonmanualized and often has higher participant–instructor ratios. Additionally, primary care providers, physical therapists, and other clinicians who manage cLBP may also want to ask their patients about perceived stress. Patients may be more comfortable talking about perceived stress than often-stigmatized mental health conditions such as anxiety and depression [63].

In conclusion, yoga and PT were effective for reducing perceived stress compared with an education control in a predominantly low-income, racially diverse cLBP population, especially among participants with high perceived stress pre-intervention. Yoga and PT are often offered in the community. Improving access to these treatment options in underserved communities may improve the management of stress in these settings.

Author Contributions

Mr. Berlowitz and Drs. Saper and Roseen had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Berlowitz, Sherman, Saper, Roseen. Acquisition, analysis, or interpretation of data: Berlowitz, Hall, Joyce, Fredman, Sherman, Saper, Roseen. Drafting of the manuscript: Berlowitz, Roseen. Critical revision of the manuscript for important intellectual content: Berlowitz, Hall, Joyce, Fredman, Sherman, Saper, Roseen. Statistical analysis: Berlowitz, Roseen. Obtained funding: N/A. Administrative, technical, or material support: Berlowitz, Saper, Roseen. Study supervision: Saper, Roseen.

Funding sources: The Back to Health Study (5R01-AT005956) was funded by the National Center for Complementary and Integrative Health (NCCIH). Funding from the NCCIH also supported the work of Drs. Hall (K23AT010157-01) and Roseen (1F32AT009272, K23AT010487-01). Mr. Berlowitz received funding through the Boston University Undergraduate Research Opportunities Program to support his work on this project.

Conflicts of interest: The authors have no financial or other relationships that would constitute a conflict of interest.

Trial registration: ClinicalTrials.gov Identifier NCT01343927.

References