Online yoga to reduce post traumatic stress in women who have experienced stillbirth: a randomized control feasibility trial

This was a three-group randomized feasibility trial. The Institutional Review Board at Arizona State University approved this study. All participants provided their consent to participate and granted the research team permission to collect medical records before beginning the study. This study adheres to CONSORT guidelines.

All data were collected online via Qualtrics.

All preliminary efficacy measures were collected at baseline, post-intervention (12-weeks), and follow-up (20-weeks).

Descriptive analyses were performed for demographic characteristics using means and standard deviation of continuous data and frequencies and proportions of discrete data for each intervention group. One-way analysis of variance were conducted to examine between-group differences on demographic characteristics. Frequency analyses were used to determine feasibility (acceptability and demand), compliance, and outcome measures using only participants that completed both baseline and post intervention surveys. To test the preliminary efficacy of the trial, repeated measures (collected at baseline, post-intervention, follow-up time points) analysis of covariance analyses (ANCOVA) were performed for each of the outcome measures of PTSD, anxiety (state and trait), depression, perinatal grief, self-compassion, emotional regulation and suppression, mindfulness attention awareness, self-rated health, and subjective sleep quality (PSQI). Lastly, to investigate the dose of yoga activity related to changes in the outcome measures, multiple regression analyses were conducted using the pooled sample (i.e. no group stratification) for each outcome for all participants with baseline and post-intervention data.

ANCOVA analyses were conducted independently on each outcome while controlling for age, race, household income, education level, marital status, BMI, and the level of peer support received (none, online, in-person, both online and in-person). Regression models were built using a hierarchical approach, adding demographic variables based on theoretical importance while consindering multicollinearity indices. All analyses were conducted while applying appropriate adjustments (i.e. Greenhouse-Gesser corrections when sphericity was not met, Bonferroni corrections for post hoc analyses) to account for non-parametric data and multiple comparisons and effect sizes were computed [43] using SPSS [44] with a statistical significance threshold set at p < .05. Post-hoc analysis to determine required sample size to achieve statistical significance (α < 0.05) with appropriate power (1- β = 0.80) for repeated measures within-between interaction effects was conducted using G*Power 3.1.9.4.

A total of 707 participants completed the eligibility questionnaire. Of those, 196 participants met the eligibility criteria and 94 enrolled in the study. Four enrollees did not start the study, yielding a total of 90 participants. Once eligibile but prior to randomization, 30% of women (n = 30) were asked to verify their stillbirth (i.e., doctor’s note) with 28 verified and two unverified and ineligible. Sixty participants were randomized to the intervention groups (LD n = 30; MD n = 30), and 30 participants were randomized to the control group (see Fig. 1).

Participants were predominantly White (n = 78; 86%). Most participants (n = 53; 59%) earned $61,000 per year or more, were college educated (baccalaureate n = 32; 36% and graduate n = 20; 22%), and married (n = 75; 83%). The average gestational age at the time of stillbirth was 30 weeks (SD = 6.99). See Table 3 for further demographic information.

Just under 46% of participants did not finish the study (i.e., dropped out or did not complete post), mainly because of time, mood (e.g., lack of energy and motivation), and stress. It was not surprising that time was a major reason for not completing the study as other physical activity interventions have also revealed time as a barrier, especially in pregnant and post-partum populations and women in the inter-conception period [45, 46]. These studies report that childcare and family obligations consume the time that they might have for physical activity/exercise [46]. Women also reported mood as a reason for not completing the study. Participants in our study were within 2 years of experiencing a baby’s death. Some literature suggests symptoms related to post traumatic stress are most prevalent and severe in the first year following the loss, and these symptoms interfere with normal life activities [19]. Mood-related symptoms associated with PTSD may have inhibited motivation to complete the prescribed videos each week. Although the prevalence and patterns of mood-related symptoms after stillbirth are not known, one study found that women diagnosed with PTSD after live-birth had higher levels of anxiety, depression, and parenting stress through 12 months postpartum [47]. The time since the baby’s death could play a key role in the ability of a woman to participate in inter-conception care (e.g., delivery of intervention based on time since the stillbirth) [48], and this information could inform the design of future interventions. Additionally, individuals experiencing symptoms of post-traumatic stress may perceive everyday stressors as more severe than prior to the traumatic event and may not have proper coping mechanisms to handle minor stressors [49]. If the intervention was perceived as an added stressor (e.g., another item on a to-do list), participants may have not been motivated to participate. That is, they may be more inclined to avoid these stressors until they have developed coping mechanisms to manage stressors.

Interestingly, participants who dropped out of the study reported they enjoyed participating in online yoga or exercise and thought if the study included an online social support group they would have been more likely to continue. Social support is a well-known behavior strategy to increasing compliance and attendance to behavioral interventions [50] Social support is critical for women after stillbirth as many report feeling isolated and their grief is often disenfranchised [51]. It is a key component in coping with loss and trauma, and those who report lower levels of social support have reported more adverse health outcomes [52] Social support may also play a role in the effects of yoga on mental health outcomes as it has been reported that social support provided during classes may be a coping mechanism for stress [53] Because the kind of support that would naturally occur in an in-person class was not present in this online intervention, there is a need to determine feasible, alternative support strategies.

Overall, participants were satisfied with the intervention, reporting they enjoyed the intervention and instructors. Similar to those who dropped out of the intervention, most participants reported they would have participated more if there had been an online social support group. Social support has been used in various physical activity interventions in order to increase adherence and enjoyment, and in some populations has been thought to mediate the effects of the intervention [54].

In the participants who became pregnant during the study and completed the intervention, many reported discomfort and being afraid of hurting the baby as barriers to participation. These women also did not have the opportunity to receive advice for modifications as they would have in an in-person prenatal yoga class. In order to progress the feasibility of online yoga interventions, future studies should consider including an online mode of delivery for a yoga instructor to provide feedback and assistance in order to prevent injury and offer modifications. For example, participants could use a real-time video software to engage with an instructor while actually performing the poses. Yoga instructors may offer support and a feeling of safety, ensure proper techniques are employed, and increase motivation.

Another obstacle with online yoga is the potential for technical issues. Most participants who reported experiencing technical issues said it was due to slow internet connections. One way to address this in future studies might be to provide instruction and recommendations for the participants to download the videos before watching them to complete their sessions.

When asked what barriers kept them from participating in yoga (on a weekly basis), women reported similar barriers to the women that dropped out of the study (i.e., work, caring for others, mood, time). In our study, we provided a barriers workbook (i.e., tips for overcoming barriers to participation) and weekly reminders to participate. The literature is not conclusive regarding overcoming barriers to exercise, but reminders seem to be effective for some populations [55]. Because of the high dropout rate and number of women who reported barriers, there is a need to develop strategies for fidelity to an online yoga intervention. We utilized reminders to participate in yoga or STC but did not include any behavioral strategies or language to overcome barriers within these reminders. Future studies could include reminders with tips that are specific to women who have had a stillbirth (i.e., sensitivity to grief, awareness of other responsibilities such as children). Additionally, because social support is necessary for improving health outcomes after a stillbirth, future research may utilize social support as a means to overcome barriers. For instance, women in the study could connect with other women who are participating and discuss ideas for overcoming their barriers such as scheduling time and childcare, and provide emotional support for each other.

Most women in the MD group reported 150 min per week to be too much and, not surprisingly, did not meet the benchmarks. Interestingly, although most participants in the low dose group reported a dose of 60 min per week of yoga to be appropriate, less than half of participants met the benchmarks for participation. Low fidelity is not surprising considering the high distress within this population [56] Other studies in populations with PTSD symptoms have reported similar compliance rates [57] Schottenbauer and colleagues found that it was common for dropout rates to be around 50% in high posttraumatic stress populations [58]. Reinhardt and colleagues also reported low retention rates in a yoga intervention for veterans with PTSD [57]. In a yoga intervention for survivors of domestic violence, Clark and colleagues reported a 30% dropout rate [59]. More research is needed to determine intervention strategies to increase fidelity specific to this population.

Because most participants did not meet the benchmarks for minutes of yoga participation per week, preliminary efficacy results are interpreted with average minutes of actual participation per week (LD = 44, MD = 77) rather than prescribed minutes per week (LD = 60, MD = 150). There was a significant reduction in posttraumatic stress and depressive symptoms at post-intervention but not at follow-up. Other studies have shown short-term effects of yoga on depressive and PTSD symptoms (e.g., 20) but more research is needed to determine the long-term effects of yoga on these outcomes. The physical domain of self-rated health decreased over the course of the intervention. Our findings are inconsistent to self-rated health outcomes in yoga interventions overall [60]. However, women who have experienced a stillbirth report physical health declines and may be more likely to develop chronic illness [61]. Although women in the STC and LD groups experienced decreases in their physical self-rated health, women in the MD group did not have any changes. This suggests yoga may have mitigated the typical decline in self-rated health seen in mothers who experience stillbirth. More research is needed in this population, specifically, to determine implications of self-rated health as it relates to post traumatic stress and other negative psychological and physical health outcomes.

The significant decreases in depression and perinatal grief in the MD group compared to the STC group suggests that 77 min of yoga per week may be enough to improve these outcomes. Other studies have shown similar improvements in depressive symptoms with 60–75 min of yoga per week compared to control groups (e.g., [62, 63]) It should be noted that most yoga interventions have been conducted using in-person formats as opposed to the online, home-based format in the current study. Therefore, discrepancies between number and length of sessions are inherent despite the potentially identical number of minutes per week. More research is needed to determine the differences and efficacy between the two methods. Additionally, although yoga may decrease depressive symptoms and grief, both states may also be barriers to participation and less likely to participate [64].

Although not significant, trends in improvements of post traumatic stress symptoms, perinatal grief, and sleep quality in both LD and MD groups, compared to the STC group, also suggest the intervention may contribute to these changes regardless of dose, a concept further supported when additionally considering regression results. Decreases on the IES-R in both intervention groups was the largest relative change among all outcome measures, suggesting a potential for yoga to decrease symptoms at doses of both 44 and 77 min.

Our dose-response findings, while also non-significant, suggested potential decreases in post traumatic stress and depression scores for every 10-min increase of yoga per week. Taken together, these trends suggest low doses of yoga (approximately 20 min on average) may be enough to elicit positive changes but more minutes per week may yield greater magnitudes of improvement. That said, these findings should also be considered in light of the feasibility findings previously discussed. That is, even though there may be an increase in benefits with additional participation, participants reported 60 min to be feasible whereas 150 min was too much. More research is needed to determine the point at which benefits are balanced with feasibility.

This was the first study to determine the feasibility of online yoga in women who have experienced stillbirth and therefore had limitations, many related to the online nature of the intervention. First, there were inconsistencies with the software used to track participation (i.e., viewing the videos) and data for some participants was not accurately recorded. Therefore, the research team utilized self-report data and the results were subject to biases that are inevitable with this type of data (e.g., social desirability). In the satisfaction surveys, participants reported technical issues that limited viewing the videos (e.g., slow internet) and may have inhibited their participation and the ability to track the minutes appropriately. Future studies should consider ways to overcome such issues that are likely inherent with online interventions. Second, the sample was not necessarily representative of the population of the women who have had stillbirths in the US, as most were white and higher income. Although in a previous study we asked minority women who had experienced stillbirth about cultural barriers to recruitment and about their perceptions and opinions on yoga [18], additional recruiting methods are needed for representative samples to ensure generalizability of the results. Third, this study had a high number of dropouts. Participants had within the last 2 years, experienced a traumatic event, and may be coping with a number of stressors. Future studies should consider strategies recommended above (i.e., social support) to increase retention and adherence to an online yoga intervention. The nature of social media recruitment may have led to a self-selection bias in this study. Finally, this was a feasibility study and thus our sample size was not powered to determine efficacy. That is, conclusions cannot be made about the effect of yoga on any of the preliminary outcome measures (e.g., PTSD, anxiety, depression). Though findings showed trends in the hypothesized directions, future studies warrant sample sizes powered to determine efficacy.