The MR program is grounded in the biopsychosocial model of health [24]. Just as the risks and threats to well-being in the target population are multidimensional and affect multiple systems (psychological, social, spiritual and physical), multidimensional intervention addressing these multiple systems simultaneously may be beneficial. VA staff treating this comorbid population have coined the term postdeployment multisymptom disorder and noted that treatments focusing on only a single condition or symptom produce “suboptimal outcomes” [25]. Thus, we composed an integrated program of practices supporting psychological (PTSD, stress, depression, resilience, self-compassion), social (mutual support, relationship satisfaction, collaborative participation, compassion), and physical outcomes (physical pain, tension, sleep quality).

The specific methods used are grounded in the evidence bases of mindfulness-based therapies [26,27], massage therapy [28,29], positive emotions [30-32], and caregiver education [33,34]. MR is designed for autonomous, self-directed use without dependence on professional instruction. This makes it a resource for users not accessing formal services, whether due to geographic obstacles, lack of suitable services, or personal preference. It is not designed as a substitute for formal clinical services but can be used to complement or enhance them.

The program delivers the instruction by techniques grouped into three categories— connecting with yourself, connecting with quiet, and connecting with your partner —together comprising the 11 individual activities described in Table 1. Instruction is provided via videos, guided audio exercises, and written materials, all accessible on the program website [35] and mobile device apps (iOS, Android, and Windows Phone versions). The study website was a static design but has been a made responsive design since completion of the study. The information offered on the website and the mobile app is redundant except for the print materials (see the Supporting Materials subsection below), which can be downloaded from the website only.

Aim 1 was to evaluate the impact of MR on mental health outcomes associated with postdeployment readjustment (primary outcomes). Hypothesis 1 was that veterans and partners assigned to MR would report significantly greater improvements in stress, depression, PTSD symptoms, self-compassion, sleep quality, resilience, social support, and relationship satisfaction compared with those in the non-MR comparison groups.

Aim 2 was to evaluate program use and satisfaction among participants in the MR program (secondary outcomes). Hypothesis 2 was that participants assigned to MR would comply with recommendations to use the nonmassage techniques of the program weekly and would average 30 minutes per week of nonmassage practice. Hypothesis 3 was that participants assigned to MR would comply with recommendations to exchange massages weekly and would receive an average of at least one massage per week. Hypothesis 4 was that users of MR receiving massage would report significant positive massage effects.

Aim 3 was to collect exploratory data on change in pain levels associated with the use of MR. There were no specific hypotheses associated with this aim.

Aim 4 was to collect exploratory data on the presence of moral injury. There were no specific hypotheses associated with this aim.

To evaluate outcomes of MR, we planned a 4-arm randomized controlled trial using a standard-of-care comparator to evaluate both comparative and additive effects over a 16-week period. The comparator was the PREP for Strong Bonds program used widely by the military for relationship enhancement and postdeployment reintegration [36]. This is a standardized, evidence-based program conducted during weekend residential retreats with experiential exercises, facilitated by Army Chaplain Corps-trained chaplains. The multidimensional program uses methods from cognitive behavioral therapy and communication-oriented marital enhancement programs developed by Markman et al, Denver University Center for Marital and Family Studies [37]. Thus, the 4 arms to be studied were (1) MR alone, (2) MR+PREP, (3) PREP alone, and (3) a waitlist control. Sample size was determined by power analysis applied to the phase I feasibility data [23]. We planned a sample of 160 dyads, with 40 dyads allocated to each arm.

We used intent-to-treat analysis. For the survey data, paired-sample t tests were performed for each pairwise contrast of times T1 (baseline), T2 (8 weeks), and T3 (16 weeks). Descriptive statistics were used to report the weekly report data for MR arms and non-MR arms. Pre-post massage session effects were evaluated using the Wilcoxon signed rank test for each weekly pairwise contrast. In addition, Kendall tau-b was evaluated across the 8-week set of samples of premassage effects. The 4 study arms were compared with each other for each of the survey outcomes at each time point using 2-sample t tests. The MR and MR+PREP groups were combined to define a common metric concerning frequency and duration of MR activities. This metric was correlated to the survey data and analyzed by Pearson correlation. Improvement in survey scores at time T2 was also predicted from the frequency and use of MR activities with Kendall tau-b. Linear discriminant analysis was performed at baseline to investigate differences in survey outcomes at T2 and T3 between the waitlist control and each of the other 3 comparison groups.

We used a sequentially rejective Sidak Bonferroni-type multiple comparisons procedure to ensure the desired experimentwise type I error rate (P<.05). This procedure has been demonstrated to more efficiently ensure the desired experimentwise P value when compared with Dunn’s procedure [48].

For dual-veteran couples, we performed subanalyses for veterans as a class and nonveteran partners as a class. In dyads where both members met the criteria for veteran, we included both in analyses of veteran data, and not in partner data.

A total of 238 dyads (476 individuals) that met eligibility criteria applied and were assessed for candidacy for 160 slots. This process included submission of a copy of a DD214 form or other official military document that verified their deployment status. Figure 4 shows the reasons prompting exclusion from selection and the flow of participants through the project (see Multimedia Appendix 4 [49] for the CONSORT eHealth checklist). All 320 participants completed their baseline surveys, with 313 completing the 8-week follow-up and 311 also completing the 16-week follow-up. Within the sample of 160 dyads were 21 dyads in which both members were veterans. Thus, the sample comprised 181 veterans and 139 nonveteran partners (hence the differences in numbers in the tables below separately reporting veteran and partner outcomes).

There were no absolute requirements for use of MR practices other than to both give and receive one 20-minute massage per week for the first 8 weeks. We wanted participants in the MR arms to use the range of nonmassage techniques offered in the way they felt was most efficacious for them, knowing that this could vary widely. We were interested to learn which practices participants chose to use most frequently. Thus, we requested that all participants in both MR arms try every practice at least once early on, then use whatever practices they felt helped them most, as often as they wished.

With a complex and self-directed program such as MR, use, or fidelity, is a critical question. If we build it, will they use it? And will those who use it more derive more benefit? With these questions in mind, we tracked use through their weekly reports.

As Table 2 indicates, during weeks 1–8, on average veterans and partners used some aspect of MR 20 times per week, totaling nearly 2.5 hours per week. With an average of more than 17 “uses” of nonmassage practices per week by both veterans and partners, and an average of 1.4 massages per week received by both veterans and partners, the usage data met and surpassed our hypothesized use of the program (hypotheses 2 and 3). The most frequently used practice cluster was connecting with yourself, with morning gratitude being most popular, followed by the mirror greeting. Since this entire cluster involves relatively brief practices, frequency does not correlate with most time spent. Nearly twice as much time was spent in connecting with quiet and even more than that in connecting with your partner. The 2 massages each averaged roughly 20 minutes and the contemplative exercise of seeing each other with fresh eyes averaged nearly 15 minutes. The large standard deviations in this table indicate widely differing usage patterns. Cumulative totals for each cluster indicate that both veterans and partners spent over half the MR time per week in exercises to enhance connection with their partner. (See Multimedia Appendix 6 for MR usage reports for weeks 2, 8, and 16.)

Weeks 9–16 required no weekly reports, but the final survey at T3 included the weekly report questions answered just for week 16, to collect data on any longitudinal pattern changes. Frequency of activity use had decreased somewhat in all 3 MR clusters for veterans, as had time spent, shrinking a bit, yet still remaining above 90 minutes of weekly use on average. Partners, on the other hand, increased their mean total time per week by over 10 minutes, due entirely to increased use of the connecting with yourself and connecting with quiet practices.

Participants in the PREP-alone and waitlist arms employed a range of techniques to help themselves relax and to support their general well-being, ease pain or tension, support their partner and strengthen their relationship, or improve their sleep. These included prescription and nonprescription drugs, sex, meditation, exercise, baths, yoga, and reading. Both veterans and partners reported most frequently engaging in activities for relaxation or well-being, followed by support for the relationship or partner, and easing pain or tension, and least frequently for improving sleep. Veterans averaged 3.6 total activities per week and partners 3.9 (see Multimedia Appendix 7 for a description of activities reported).

While we assigned 40 dyads to each arm, for dual-veteran dyads we analyzed both participants as veterans, leading to sample sizes larger than 40. Table 3 presents veterans’ scores by arm at all 3 testing points. Baseline differences between arms were few and relatively modest except a PainBest difference between arms 1 and 3. P values shown reflect significant differences for T2 and T3 relative to baseline scores.

The most striking differences were between the MR-only and waitlist control arms. The MR-only veterans had improvements at 8 weeks in a broad array of mental health dimensions, indeed, everything other than pain scores, dyadic adjustment scores, and perceived social support. All improvements were sustained at 16 weeks with the exception of sleep improvement (PSQItotal). Perceived ability to respond to stressful events (RSES) had further improved.

The PREP-only veterans (arm 3) showed a modest gain in three important dimensions at T2 (PSS, BDI, and PCL-C), which became stronger by T3. The MR+PREP arm was included to see whether there would be added value by combining the programs. These results indicate that adding MR to PREP produces improvement in more domains than PREP alone (adding improvements in self-compassion and capacity to respond to stress). However, the combination yielded a lower magnitude of improvement in those arenas than resulted from MR alone. Thus, aim 1, our hypothesis that veterans and partners assigned to MR would report significantly greater improvements in a wide range of mental health outcomes, was partially met in that veterans in the MR-only arm reported significant improvements for a broader array of mental health outcomes than veterans in the other arms.

The 4 partner arms were also quite comparable at baseline (Table 7). Differences were that arm 1 had higher perceived stress than arm 4, and arm 3 differed from arms 2 and 4 on the PainBest and PainUsual scores. However, partners differed from veterans in terms of program outcomes.

Examining change within each arm from baseline to T2 and T3, we see no significant changes for either of the non-MR arms. The MR-only arm shows change in the same arenas for partners as for veterans, but a bit less powerfully at T2 and somewhat diminished at T3. For MR+PREP, the changes for partners strengthen from T2 to T3, which may reflect partners’ increased use of the program in weeks 8–16, when veterans had decreased both duration and frequency of use. Despite this within-group change for the MR arms, there were no statistically significant differences between the 4 arms as time progressed, other than for pain.

Group differences for partners at T2 and T3 were not noteworthy and are not reported here (available upon request).

We included partner massage as an important element of MR both to give couples another way to communicate their care and because our own prior studies found that brief massage by family members significantly reduced troublesome symptoms [23,34]. Both veterans and partners reported highly significant reductions in all assessed symptoms of physical pain, tension, irritability, anxiety or worry, and depression shortly after receiving massage (Table 8), thus supporting hypothesis 4. In addition, both veterans and partners reported significant reductions in most premassage symptoms over the 8-week period (Table 9). While the short-term improvements recorded in the pre- and postmassage scores can reasonably be attributed to massage effects, the premassage changes over time cannot. They may be influenced, at least in part, by the MR program use overall, but we did not directly test that.

In view of the encouraging phase I feasibility results [23], after adding a margin of conservatism, we chose a moderately large effect size, corresponding to a value of 0.7 for Cohen d. The analysis of veterans’ within-group change (Table 3), assuming n=40 per arm, and type I error probability alpha=.05, obtained a power of 1–beta=99%. Between-group changes (Table 5,Table 6) yielded a power of 87%. Partners’ within-group changes (Table 7), with n=32 per arm, obtained a power of 96%. Massage session effects (Table 8) yielded a power >99% for both veterans and partners.

On the final survey (T3) we asked MR participants how likely they would be to recommend the program to a friend. On a 0–10 scale, veterans’ mean score was 8.7, and partners’ mean was 9.1, indicating high user satisfaction with the program.

Our primary intention in designing MR was to offer veterans and their partners a flexible form of instruction in simple ways to improve their own well-being. The range of outcomes in which users had significant improvement can be seen as confirmation that MR teaches skills that improved their well-being. The fact that improvements, including reduced PTSD symptoms, and increased self-compassion, were sustained at the 16-week follow-up is particularly promising.

Hypothesis 1 was generally supported by the findings that MR participants had significant improvement on far more mental health outcomes than did participants in other arms of the study. The exceptions in this case were sleep quality (partners using MR had more stable improvements than veterans), and lack of movement in dyadic adjustment and perceived social support. We speculate that the lack of change in relationship variables may have been due to high baseline levels, which may have led to the decision to participate, but this is conjecture and warrants further study. Hypothesis 1 was not fully supported in that, while the 2-sample t tests indicated significantly stronger improvement among MR participants on many mental health outcomes, only a portion of these remained following adjustment for potential experimentwise error. The remaining hypotheses were strongly supported.

It is notable that both veterans and partners in the MR arms used this self-directed program for over 2 hours per week during the initial 8 weeks, surpassing our hypothesized use, and averaged well over an hour per week throughout the 16-week data collection period. We suspect the control each user had over which program elements to use and when contributed to the high use level, underscoring the value of user preference in long-term adherence to, and ensuing effectiveness of, self-care approaches for this population.

Sustained use may also have been enhanced through the inherent support and encouragement of compliance by coparticipation with a significant relationship partner. We note, however, that the majority of MR practices (except for massage) can be used either alone or with others. This adds to the convenience for young parents, allowing one partner to practice while the other attends to their children.

In comparing the trial’s 4 arms, the contrast for both veterans and partners is greatest between MR-only and waitlist control arms, a predictable finding based on phase I results. The contrast in outcomes for the MR-only versus PREP arms, especially at 8 weeks, is notable. While PREP has been found in multiple studies to support significant improvement in variables related to relationship dynamics and intimacy for couples, we chose outcome measures to assess other dimensions of mental health. We selected some MR components based on evidence of health-promoting neurological and neurochemical effects; potential easing of symptoms related to PTSD, stress, and depression; and likelihood of enhancing self-compassion and forgiveness. We expected the physical exercises to amplify energy, improve mood, and reduce susceptibility to the fight-flight-freeze response. While PREP offers some content aimed at relaxation, stress reduction, and forgiveness, it devotes more time to relationship skills per se than does MR.

Thus, one interpretation of the outcomes is that MR performed well at its goals and PREP does less well at the things MR is designed to do. As to whether combining the programs adds value, adding MR to PREP appears to amplify improvement in self-compassion and response to stressful events. Adding PREP did not enhance other results for MR-only.

The program showed strong benefit for both men and women veterans and for their partners, although benefits were strongest for veterans. Based on these findings, which included veterans from all branches of service, and participants both on active duty and retired or separated from service, it appears that the program provides a safe, low-cost self-care intervention that enhances overall well-being. A modified version could be useful for single veterans.

This study evaluated MR in a community-based sample with no inclusion or exclusion criteria related to specific mental or physical health parameters. Thus, the program’s impact in clinically defined populations remains to be assessed. Other limitations are that the follow-up period was limited to 16 weeks and that the program was tested only with dyads, not single veterans. In addition, all participants were required to attend an in-person launch meeting, and it is possible that this excluded potential applicants for whom such attendance was impossible for reasons of time, geography, or something else. Finally, the instructional program as tested did not include video closed captioning or verbatim transcripts of the audio instruction, and thus would not have accommodated users with hearing limitations. These enhancements will be considered in future upgrading.

Four directions for future research on MR for veterans present themselves. First is to test whether adherence and benefit are still present at the 1-year mark or even further with a partnered sample such as we used. Second is to offer MR, minus the partner massage, to nonpartnered veterans to see whether lack of a partner affects use or outcomes. Third, it is important to follow this study with research on diagnostically defined samples (eg, high PTSD) to see what benefit might be provided for a clinically based sample. Fourth, we also recommend that health services research be done to determine how best to offer MR. This study demonstrates that veteran-partner dyads can learn a range of physical and contemplative self-care practices on their own from a media program accessed via mobile app or webstream, and that they can derive great benefit across several dimensions of well-being. Still, it is possible that, especially for a clinically based or nonpartnered sample, use of the program would be enhanced by introduction through in-person group instruction, for example, with 1 to 4 structured sessions. While this would add cost to delivery of the program, it is possible that added benefit could justify the cost. Related to this, one could retest MR, forgoing the launch meetings and substituting Web-based instruction in the provision of weekly report and survey data.

In addition, we recognize that many nonveterans have PTSD, physical pain, lack of self-compassion, etc. MR, as it now exists, or with minor changes, could and should be tested with other samples. This could include others who have experienced severe trauma, such as refugees, youth and adults exposed to mass shootings such as that in an Orlando, FL nightclub in 2016, as well as people with chronic pain and more generalized anxiety.

This study adds to the growing literature on the power of brief, repeated mind- and body-based practices to address physical, psychological, and spiritual well-being [50-52]. The results indicate that MR is a widely accessible, low-cost approach that supports well-being and reduces multiple symptoms among post-9/11 veterans and their partners. Both veterans and partners were able to learn and make sustained use of a range of wellness practices from a media-only source. While the launch meetings offered 90 minutes of human contact, they offered no instruction in any of the practices. These usage findings contrast with others’ findings of high dropout rates for clinic-based programs such as meditation instruction [53]. Home-based, self-directed interventions may be of particular service to veterans who are distant from, averse to, or prohibited by schedule from using professional services. The partner relationship may enhance sustained use of self-directed interventions for this population. Finally, these data suggest MR to be superior to waitlist and PREP for Strong Bonds in both within-group and between-group assessments. Notably, this distinction appears to increase over time.