Among US veterans, the prevalence of comorbid posttraumatic stress disorder (PTSD) and major depressive disorder is estimated to be roughly twice the rate of PTSD alone [1], and the symptoms of each of these disorders have been associated with deficits in emotion regulation [2,3] and psychosocial functioning [4,5]. Meanwhile, veterans experience numerous practical and cultural barriers to accessing evidence-based mental health treatment. These include the need to travel long distances to facilities, financial constraints, health conditions, inflexible work schedules, and caregiver responsibilities [6-8], as well as veterans’ desire to be self-reliant and the perceived stigma of receiving mental health care [9,10]. There are, in addition, systemic barriers including a shortage of mental health providers in many parts of the United States [11]. Mobile health (mHealth) programs, including web-based and app programs, address these difficulties and have been found to be effective in the treatment of symptoms of PTSD and depression [12-15]. However, it has been shown that mHealth programs with human support consistently provide greater benefits than entirely self-guided programs [16,17]. Given the burdens on clinicians and other licensed professionals in effectively serving patient needs, the introduction of peer-supported mHealth programs holds the promise of reducing the burden on the health care system while maintaining and potentially increasing access to care and supporting patients in obtaining optimal outcomes.

Peer specialists are persons with lived experience of recovery from mental health difficulties who are trained to support patients with similar problems in engaging in and completing mental health programs [18]. Peer specialists bring particular strengths to their roles, including their ability to develop strong rapports with participants and reduce stigma [19,20]. Literature reviews have indicated that the presence of peers increases engagement in treatment across persons with a variety of mental health difficulties and disorders, treatment modalities, and delivery strategies [18,21]. In particular, the use of peer support in mHealth has been found to increase engagement among veterans with unmet mental health needs [22]. However, randomized controlled trials assessing the impact of peers on symptom change are relatively few in number [23] including for mHealth interventions [21].

Studies integrating peer specialists into Veteran Affairs (VA) mHealth programs have strongly supported their acceptability and feasibility [22,24-26]. Randomized controlled trials of peer-supported, web-based programs have reported moderate to modest symptom change outcomes on PTSD and depression when compared to waitlist [27] and found that utilization of peer support was associated with greater reductions in depression when compared to self-directed program usage [22]. Additional trials are needed to assess mental health outcomes, particularly symptom change and psychosocial functioning to better understand the optimal contribution of peer-supported mHealth programs to mental health care. This study evaluates a peer-supported, transdiagnostic mHealth program for trauma-exposed veterans. Given the frequency of comorbid symptomatology among trauma-exposed populations [1,28], a transdiagnostic approach, if successful, provides the opportunity for broad use across various trauma-exposed populations with diverse needs. Moreover, different types of symptoms share issues with emotion regulation and psychosocial functioning, making these difficulties important treatment targets.

The program evaluated, web-based Skills Training in Affective and Interpersonal Regulation (webSTAIR), is a web-based program adapted from Skills Training in Affective and Interpersonal Regulation (STAIR), a manualized evidence-supported cognitive behavioral intervention for trauma-exposed persons with symptoms of PTSD or depression that focuses on improving psychosocial functioning by building emotion regulation and interpersonal skills. Investigations of 10-12 session STAIR have reported large effect sizes in the reduction of PTSD and depression as well as in emotion regulation and psychosocial impairment [29,30]. Similar outcomes have been obtained in a briefer 6-session version, including in a randomized controlled trial in primary care [31] and in an open trial delivered by peers in a low-income, primary care community service [32]. In parallel to the STAIR studies, investigations of webSTAIR assessed a 10-module version with weekly coaching sessions from licensed mental health providers. These studies have included 2 open trials [33,34] and 1 comparison trial [35]. The open trials found significant improvements with moderate to large effect sizes for PTSD, depression, emotion regulation difficulties, and psychosocial impairment, and the comparison study, with a noninferiority design, found that reducing coaching support to biweekly (5 coaching sessions) from weekly (10 coaching sessions) resulted in noninferior benefits on all outcomes.

Given the evidence supporting the success of brief, peer-supported STAIR in reducing a range of symptoms, along with growing support for the effectiveness of web-based mental health programs, a 6-session peer-supported version of webSTAIR was developed. This abbreviated version of the webSTAIR protocol can be completed in a shorter time frame than the 10-session webSTAIR and integrates support from trained veteran peers at the end of each module. The potential benefits of this program, if successful, is the availability of a low-burden, brief mHealth program that is effective in reducing PTSD and depression symptoms as well as improving emotion regulation and psychosocial functioning.

This study aimed to assess the benefits of Brief Peer-Supported (BPS) webSTAIR compared to a waitlist control condition among US veterans with symptoms of PTSD, depression, or both. We hypothesized that participants assigned to the BPS webSTAIR condition would experience significantly greater improvement in PTSD symptoms and depressive symptoms (primary outcomes) as well as in emotion regulation and psychosocial functioning (secondary outcomes) compared to the waitlist, and that these gains would be maintained at an 8-week follow-up. Exploratory analyses were conducted to determine whether the number and type of peer interactions influenced outcomes. This study was funded by the National Center for PTSD Dissemination and Training Division within the VA Palo Alto Health Care System.

Candidates for study participation were recruited via social media advertisements and directed to complete a short set of online screening questions. Following the screening, study candidates were contacted via phone and scheduled for a 30- to 60-minute phone assessment with this study’s coordinator to assess inclusion and exclusion criteria, receive an explanation of the program, and complete (verbal) informed consent. If eligible for this study, the participants were provided access to the program. Randomization was allocated at a 2:1 ratio (webSTAIR to waitlist) in blocks of 12 and was computer generated via an algorithm developed by staff otherwise not involved in this study. Participant randomization assignment was indicated upon opening the program URL. Participants in the BPS webSTAIR condition received instructions to start the program within 2 weeks following the assessment and to complete the 6-module program within 10 weeks. Posttreatment assessment (10-week mark) and an 8-week follow-up were conducted via phone by study staff who were unaware of the condition and assessment period. Participants in the waitlist condition were given the information that they would have access to the webSTAIR program in 10 weeks and completed assessments at the pretreatment and posttreatment time points (10-week mark). Waitlist participants were then given access to webSTAIR. Recruitment began on October 16, 2020, and follow-up ended on September 28, 2022. This study is registered at ClinicalTrials.gov (NCT04286165).

Persons were eligible if they were veteran or military personnel; aged older than 21 years; reported experiencing at least 1 traumatic event; reported symptoms of PTSD or depression, or both, as indicated by a score of ≥3 on the Primary Care Posttraumatic Stress Disorder Screen for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition [36] or ≥2 the Patient Health Questionnaire-2 item (PHQ-2) [37]; were able to read and write in English; had a PC and stable internet connection; and resided in the United States. Exclusion criteria were determined during the phone assessment and were (1) the presence of significant suicidality as indicated by the presence of a plan and means, (2) the presence of cognitive difficulties or active psychosis that would indicate a low likelihood of benefiting from treatment, and (3) current participation in trauma-focused treatment.

The BPS webSTAIR intervention is a shortened, 6-module version of webSTAIR adapted from brief STAIR [31]. The 6 modules of BPS webSTAIR focus on emotional awareness; emotion management through the body, thought, and behavior channels; distress tolerance; and self-compassion. The program integrates text, audio, and video delivery of psychoeducation with interactive exercises and worksheets to help patients understand and internalize the concepts introduced.

At the end of each module, participants were prompted to complete a chat session with a veteran peer coach to help clarify program content and answer any questions they might have about how to implement the skills presented in the module. Interaction with the peer occurred via secure instant messaging on the program platform. Participants could message peers as often as they liked while they were on the platform. It was not expected that the participant would receive a response from the same peer. The peer program was committed to responding to text messages within 24 hours. Due to changes in the availability of veteran peer coaches over this study’s period, 3 distinct cohorts were identified based on participant start date. In cohort 1 (from October 16, 2020, to July 9, 2021), coaches were available for 24/7 support, and participants were required to complete a chat session with a coach at the end of each module to move forward to the next module. In cohort 2 (from July 10, 2021, to November 22, 2021), coach availability was reduced to 8 hours a week, with 4 hours at the beginning of the day; 8:30 AM-12:30 PM) and 4 hours at the end of the day and into the evening (4:30 PM). In cohort 3 (from November 23, 2021, to July 20, 2022), coach availability remained reduced, but participants were able to independently skip any chat sessions they chose. For all 3 cohorts, the research coordinator was available as backup for logistical and content support Monday through Friday by phone call or secure text message during an 8-hour workday (8 AM to 4 PM Pacific) with responses completed within 24 hours or less.

All veteran peer coaches were VA-certified peer counselors, which includes a working knowledge of VA guidelines for assessing suicidal risk and referral to the veteran suicide hotline. In addition, for this project, the peers received training from the developers of the web-based platform about navigation of the system (Vets Prevail) and received a manual and a workshop consisting of two 2-hour meetings taught by webSTAIR staff (MC and SS) which described the interventions and provided suggestions about how to connect the program material to ongoing stressors in the life of the veteran client and to recommend the practice of skills between sessions.

Analyses were conducted using R (version 4.2.2; R Foundation for Statistical Computing). All participants who met eligibility requirements and completed the pretreatment assessment were included in the final dataset, regardless of whether they completed the program or participated in subsequent assessments. Consequently, all analyses were conducted using the intention-to-treat (ITT) sample of randomized participants. Categorical pretreatment demographic variables and responses to individual LEC items were compared between the treatment and control groups using chi-squared tests of independence. The mean LEC scores for the 2 groups were compared using independent samples 2-tailed t tests. Rates of missingness were 0% (0/178) at baseline, 38.2% (68/178) at posttreatment assessment with 47% (55/117) missing for webSTAIR and 21% (13/61) missing for waitlist, and 41.9% (49/117) at follow-up (webSTAIR condition only). Missing data were addressed using multiple imputation. Joint multiple imputation of missing values at posttreatment and follow-up assessments was carried out using the panImpute function in package mitml. Due to (1) some of the presently reported analyses being specific to webSTAIR participants, while other analyses were relevant to the full sample, (2) differences in the assessment schedule between the treatment and control conditions (ie, only webSTAIR participants were assessed at follow-up), (3) the unbalanced randomization design, and (4) differences in rates of missingness between conditions, multiple imputation procedures were carried out separately for the data that would be used to examine within-group change for webSTAIR participants across the 3 time points and between-group differences in change from pre- to posttreatment assessments. In other words, the between-group imputed datasets consisted of baseline and posttreatment data for all participants, whereas the within-group imputed datasets included baseline, posttreatment, and follow-up data for only the participants randomized to webSTAIR. In each case, 100 datasets were generated with missing values imputed. Before imputing the data, chi-square tests and 2-tailed t tests were used to identify predictors of missingness and baseline characteristics related to nonmissing values on outcome variables. The missing data correlates identified from these preliminary analyses (employment status, education level, service era, and gender) were included as auxiliary variables in the imputation models. To ensure that any interactions between time and condition were appropriately preserved in the between-group imputed datasets, this interaction term was included as a predictor in the between-group imputation model. The results were pooled across imputed datasets using the test estimates function in the mitml package.

For the main analyses of our primary and secondary outcomes, linear mixed-effects models were fitted using restricted maximum likelihood estimation with the lmer function in the package lme4. These models included main effects for time and condition, a time-by-condition interaction, as well as fixed effects to adjust for potential differences between the cohorts that may have resulted from changes in peer accessibility and chat requirements during this study. To control for potential cohort effects, 2 dummy variables were created to identify the participants in cohorts 2 and 3, with cohort 1 as the reference group. A random intercept was included in all models to account for repeated measurements within participants. The time by condition interaction term provided a test of the difference in change from pre- to posttreatment assessments between the webSTAIR and waitlist conditions. Cohen d between-group effect sizes for each measure were calculated by dividing the estimated interaction term by the SD of pretreatment scores in the total sample. Similar models were constructed to measure within-group change among webSTAIR participants across all 3 time points. The time predictor in these models was specified so that posttreatment assessment was the reference category, which enabled us to evaluate the magnitude and statistical significance of changes that occurred from pre- to posttreatment assessments, as well as whether any improvements during treatment were retained during the follow-up period (ie, as indicated by a nonsignificant estimate for change between the posttreatment and follow-up assessments). In the within-group analyses, Cohen d effect sizes for change from pre- to posttreatment assessment and from posttreatment to follow-up assessment in the BPS webSTAIR group were calculated by dividing the estimated coefficients for time by the SD of pretreatment scores for the BPS webSTAIR condition.

To evaluate the impact of program completion on outcomes, module completion was categorized as “none” (did not complete any web-based content, including the welcome module), “some” (completed the welcome module and no more than module 2), or “moderate to complete” (completed module 3 or higher). Within each completion group, linear mixed-effects models were constructed for each outcome measure to evaluate within-group change across time. Time predictor dummies were specified such that posttreatment assessment was the reference category, and a random intercept was included in each model. Cohen d effect sizes for pre- to posttreatment change within each completion group were calculated by dividing the estimated coefficient for time by the SD of pretreatment scores for BPS webSTAIR participants in the respective completion group (eg, the SD of pretreatment PCL-5 scores among BPS webSTAIR participants in the “moderate to complete” group).

Exploratory analyses were conducted to better understand the impact of the type (required versus optional) and number of peer chat sessions in the different cohorts. For each participant, we calculated the total number of chat messages exchanged with a peer support coach and the average number of chat messages exchanged per module of the program completed. Descriptive statistics for chat messages were examined by cohort. Comparisons were drawn between cohort 1, in which chat sessions were required to advance to the next module, and cohort 3, in which chat sessions were not required for module advancement, to assess whether chat engagement and completion rates differed in these 2 contexts. We did not include cohort 2 in these analyses, given that this was a transitional phase when participants’ experiences with the chat feature may have varied in ways that are not quantifiable (eg, experienced highly variable wait times in connecting with a peer).

Finally, the relationship of chat messages with treatment outcomes was explored by fitting linear regression models that included average messages exchanged per module and pretreatment scores for the outcome as predictors of pre- to posttreatment change in the outcome. Due to differences between cohorts in the availability of peer coaches and program instructions for module completion, the relationships between messages exchanged per module and changes in outcomes could not be interpreted in the same way across cohorts. Given these differences in context, separate models were estimated for each cohort so that relationships between chat engagement and outcomes could be interpreted according to the specific circumstances of each cohort. Additionally, differences in the number of modules completed by participants were taken into account by examining the relationships between messages exchanged per module (vs total messages exchanged) and change in outcomes.

Institutional review board approval for the study was obtained from Stanford University (Protocol 51076). All participants provided verbal informed consent. Participants were paid US $40 for the pretreatment assessment and US $60 for the posttreatment and follow-up assessments. Data were collected online and stored in a secure VA-approved environment.

The results of this study indicated that as compared to the waitlist, peer-supported webSTAIR is effective in improving symptoms of PTSD and depression, as well as emotion regulation and psychosocial functioning, with gains maintained at 8-week follow-up. The within-group pre- to posttreatment effect sizes for the webSTAIR participants ranged from 0.48 to 0.64, somewhat larger than those found for other peer-supported mHealth programs [18,21]. BPS webSTAIR offers a convenient and potentially more scalable option for veterans who have limited time available, lack access to a mental health provider, or prefer to work through the content in a self-guided manner.

The outcome varied depending on the number of modules completed. Among those who completed 3 or more modules, the mean reduction in PCL-5 score was over 10 points, and for the PHQ-8, it was nearly 5 points, reductions that are typically considered clinically meaningful [47,48]. Furthermore, our findings suggest that even partial completion of BPS webSTAIR (completing at least the welcome module but not past the second module) is associated with improved PTSD and depressive symptoms, emotion regulation, and psychosocial functioning relative to the waitlist. Therefore, the program may be of some benefit to patients who are unable to fully complete it due to low functional status or other demands on their time.

Overall, only about half (62/117, 53.0%) of the veterans engaged with peers, with the highest level of nonengagement (28/40, 70%) occurring in cohort 3. Analyses of chat data may provide insights about the role of peer engagement in mHealth. Based on our analyses of chat data from cohort 1, requiring a peer chat to move to the next module of the program was associated with a greater number of chat sessions and more messages exchanged, per module and in total. However, it did not contribute to program completion. In fact, the average number of modules completed in cohort 1 and the proportion of program completers was significantly lower than that in cohort 3. The introduction of a required engagement with the peer was intended to prompt the participants into peer engagement that otherwise might not have happened and provide an experience of contact that would lead to more engagement with the program and greater completion rates; however, this did not happen. Rather, the higher module completion rate within cohort 3 suggests that participants were motivated to work through the content even without support. Similar to this study, Possemato and colleagues [22] reported that 55% of veterans provided with the option to engage with a peer chose not to and suggested that some veterans might not need or want peer support. We speculate that a flexible, personalized approach to peer use leads to better program use and potentially better outcomes. The analyses indicated that while greater module completion was associated with the cohort with flexible peer contact, and increasing module completion was associated with a better outcome, the number of contacts (peer chats) was not associated with a better outcome in either cohort. This may be due to the potential moderating effects of patient preference regarding the use of peer support. For example, only patients who need and want peer support may benefit from it while others might do equally well or better without it. It would be valuable to identify the baseline characteristics of clients who might substantially benefit from peer support versus those who will not. In addition, it may be that clients who will be helped by peer support will experience greater benefits if the peer is proactive in reaching out to the client. In qualitative interviews, mental health coaches have pointed to the value of including proactive contact initiated by coaches rather than placing the onus on participants to engage [49]. In this study, the peers were available to the veterans to chat, but the veterans initiated the contact. Future efforts to enhance the impact of peers may involve modification regarding how proactive they are.

This study has limitations. First, we relied fully on self-report measures administered by phone. Second, for ethical reasons, we did not maintain the waitlist group during the follow-up period and so were unable to conduct between-group analyses of change from pretreatment to follow-up. Third, this study design did not intentionally include an implementation variation regarding required versus optional engagement peers. As a result, the cohort analyses are post hoc and we have limited information about the veterans’ experience. We do not know, for example, the average wait time for a chat, the range of wait time, or the frequency of contact with the research coordinator. Differences in these factors across the cohorts could have contributed to the observed differences in retention rates. Furthermore, chat transcripts were not accessible and thus we were limited in our ability to explain why more chat messages were not associated with better outcomes. It could be that the conversations were routine and pragmatic (eg, resolving technical issues) versus meaningful guidance about the content of webSTAIR. Lastly, participants were veterans recruited via social media advertisements which may limit generalization of findings beyond for veterans or persons who do not interact with social media.

This study adds to existing literature on mHealth by demonstrating that a brief, transdiagnostic, peer-supported, web-based program can provide moderate to large improvements in PTSD and depression symptoms as well as in emotion regulation and psychosocial functioning. Given the potential of peer-supported mHealth programs to expand access to evidence-based care for symptoms of PTSD and depression, future research should evaluate how best to deliver peer support, how much to deliver, and for whom the support benefits most. Future studies may also explore alternative formats for the delivery of peer support; for instance, proactive outreach from peer coaches or supplementary channels of communication (eg, phone contacts) to increase flexibility in ways in which a peer relationship can be built. The particular strengths of peer support for specific and varied aspects of mental health care such as treatment engagement, program completion, and symptom reduction need to be determined to develop maximally efficient and effective mental health services.