Yoga for Pediatric Tension-Type Headache

Adolescents (ages 13 to 18) with TTH will be randomly assigned to either a Hatha yoga intervention (n = 20) or a wait-list control condition (n = 20). The yoga intervention will consist of 8 weekly classes lasting one hour each taught by an experienced yoga teacher. Each class will have a maximum of 10 participants. A baseline assessment including an experimental pain protocol, questionnaires, and hair samples (to determine cortisol concentrations in the past month) will be completed prior to the intervention. Biweekly questionnaires will be completed by both intervention and wait-list groups during the 8-week intervention period. The post-intervention assessment will include the experimental pain protocol, questionnaires, and hair samples (to determine cortisol concentrations in the past month). The wait-list control group will subsequently receive the yoga intervention and complete post-intervention and follow-up assessments (i.e., the wait-list group will complete one more experimental pain protocol and will provide one more hair sample than the intervention group); these data will then be merged with that of the yoga group to examine pre- and post-intervention effects. Both intervention and wait-list groups will complete questionnaires 3 months following completion of the yoga protocol.

Specific Aim 1: To determine the safety, feasibility, and acceptability of an 8-week Hatha Yoga Protocol (HYP) in adolescents (ages 13 to 18) with tension-type headache (CTTH). Hypothesis: At least 80% of adolescents will complete the HYP. Attrition rates will be compared between HYP and control groups to determine whether results may be biased given different durations of the study protocols between the two groups and interval between recruitment and intervention (i.e., control group will be in the protocol for two additional months and the intervention will be delayed by 8 weeks).

Specific Aim 2: To evaluate changes in primary outcomes (frequency and intensity of headaches, pain-related functional disability, health-related quality of life, medication use) from baseline to post-intervention in the HYP and usual care wait-list controls. Hypothesis: Participants in the HYP group will demonstrate greater reduction in headache frequency and intensity, decreases in pain-related functional disability, improvement in health-related quality of life, and reduction in medication use compared to wait-list controls from baseline to post-intervention. Secondary analyses will examine changes from pre- to post-intervention for the entire sample once the waitlist group has completed the HYP; in addition, sustained intervention effects will be determined by examining changes from post-intervention to 3-month follow-up for the entire sample.

Specific Aim 3: To assess changes in potential mediators of clinical outcomes following the HYP, including responses to experimental pain [conditioned pain modulation (CPM), temporal summation to second pain (TSSP)], alterations in stress response systems [hair cortisol concentrations, self-reported stress levels, stress hormone responses to experimental pain (cortisol, alpha-amylase, DHEA), heart rate variability responses to experimental pain] and pain-related cognitions (catastrophizing, self-efficacy, acceptance). Hypothesis: Participants in the HYP group will show increased CPM, decreased TSSP, reduced hair cortisol concentrations and self-reported stress levels, lower cortisol and alpha-amylase responses to experimental pain, higher DHEA responses to experimental pain, lower heart rate variability prior to experimental pain, reduced pain catastrophizing, increased pain self-efficacy, and increased pain acceptance compared to wait-list controls at post-intervention.

Although yoga has shown promise as a complementary health approach for chronic pain in adults, few studies have examined the potential efficacy of yoga for adolescents with chronic pain. Anticipated findings from the proposed study will help to determine whether - and how - an 8-week yoga intervention can improve pain-related outcomes in adolescents with TTH; these findings will be critical for refinement of the HYP for youth with TTH and for the design and implementation of a larger-scale, randomized controlled trial with an active control condition.

Background and Significance

TTH is common in youth and associated with significant functional disability. Complementary health approaches such as Hatha yoga, which emphasize physical postures, breathing exercises, and meditation, have been shown to reduce chronic pain-related disability. The only known pilot study of yoga for youth with headache showed moderate but non-significant improvement in daily functioning from baseline to post-intervention (Hainsworth et al., 2014). A randomized controlled trial of yoga for headache pain in adults found that the yoga group exhibited reduced migraine frequency and intensity, lower pain severity, and less medication use for headache compared to the control group after 12 weeks. Potential mechanisms underlying improvements in pain outcomes following yoga interventions include physiological, behavioral and psychological changes. Adults with TTH are characterized by impaired CPM and enhanced TSSP. Moreover, increased pain sensitivity in adults with TTH appears to be exacerbated by chronic stress, sustained physiological arousal, and negative cognitions. The proposed study addresses a critical gap in the literature by investigating the safety, feasibility, and efficacy of Hatha yoga as a complementary treatment approach for youth with TTH.

Preliminary Studies

Members of this research team have previously employed a similar experimental pain protocol to demonstrate elevated TSSP in female adolescents with functional abdominal pain, elevated TSSP in adolescents with both functional abdominal pain and a history of physical or sexual assault/abuse, and racial differences in both TSSP and CPM in healthy youth ages 10 to 17.

Sample Size Justification and Statistical Analysis Plan

Sample size estimation: Prior yoga interventions in youth with chronic pain suggest medium-to-large effect sizes (Cohen's d comparing mean changes from baseline to post-intervention) for indices of changes in pain-related functional disability. For power calculations, G*Power 3.1.9.2 was used to estimate the sample size required for estimated effect sizes ranging from .50 to .80. Assuming a Type I error rate of α = .05 and statistical power of 0.80, results indicated a range from 16 to 34 total participants. Assuming a 20% attrition rate and adopting the more conservative anticipated total n = 34, the investigators estimate that 40 total subjects will need to be recruited (20 per group).

Data analytic strategy: Group differences in primary outcomes (Specific Aim 2) will be tested using analysis of covariance, controlling for baseline scores on these outcome variables. An intent-to-treat analysis of primary outcome measures (including all enrolled participants and utilizing the last observation carry forward approach) will be conducted to confirm findings in participants who complete the HYP.

A series of linear regressions will test the direct effects of the intervention on outcome variables and the indirect effects of mediators on the relationship between mediator and outcome (Specific Aim 3). Mediation analyses will be run using the Preacher and Hayes (2008) SPSS Macro for Multiple Mediation. In order for statistical mediation to occur, two criteria are necessary: 1) there is an effect to be mediated (i.e., a significant association between intervention and the outcome), and 2) the indirect effect of intervention on outcome via mediator was significant. The indirect effect will be considered significant at the p < .05 level if the corresponding bootstrapped confidence interval does not contain zero. Bootstrapping procedures will test each mediation model by generating a 95% confidence interval (bias corrected) around the indirect effect test statistic in a series of 1000 Bootstrap samples repeatedly drawn from the full sample.

Potential limitations. The pilot study sample size is modest but will yield critical data regarding effect sizes for primary pain-related outcome variables as well as potential mediators of treatment effects.