Peer-Supported Mobile Mental Health for Veterans in Primary Care: A Pilot Study

Abstract

One in four veteran primary care patients suffers from a mental health condition; however, most do not receive any treatment for these problems. Mobile health (mHealth) can overcome barriers to care access, but poor patient engagement limits the effectiveness and implementation of these tools. Peers may facilitate patient engagement with mHealth. We designed a protocol for peers to support implementation of mobile mental health tools in primary care and tested the feasibility, acceptability, and clinical utility of this approach. Thirty-nine patients across two VA sites who screened positive for depression during a primary care visit and were not currently in mental health treatment were enrolled. Participants were scheduled for four phone sessions with a peer over eight weeks and introduced to five mobile apps for a range of transdiagnostic mental health issues (stress, low mood, sleep problems, anger, and trauma). Pre/post phone interviews using quantitative and qualitative approaches assessed participants’ self-reported app use, satisfaction with the intervention, symptom change (stress, anxiety, depression, insomnia), and progress with personal health goals. On average, patients reported using 3.04 apps (SD=1.46). Per the Client Satisfaction Questionnaire, global satisfaction with the intervention was high (M=25.71 out of 32, SD=3.95). Pre to post, participants reported significant improvements in their level of stress, based on a quantitative measure (p=.008), and 87% reported progress on at least one personal health goal. Findings support the feasibility, acceptability, and clinical utility of peer-supported mobile mental health for veterans in primary care. A randomized controlled trial of an adaptive version of this intervention is recommended.

One in four patients presenting to a primary care clinic in the Veterans Health Administration (VHA) has a mental health diagnosis, which increases these patients’ risks for an array of adverse health outcomes (Trivedi et al., 2015). The prevalence of specific mental health conditions in this population is highest for depression (13.5%), with over one-half million veterans receiving this diagnosis annually (Trivedi et al. 2015). Depression is also the mental health condition that has the highest rates of comorbidity with other mental health conditions such as posttraumatic stress disorder (PTSD) and other anxiety disorders (Rytwinski et al., 2013) as well as common, transdiagnostic mental health symptoms such as insomnia (Wright et al., 2011) and anger (Gonzalez et al., 2016).

Across the US healthcare system, the VHA has been a model for the integration of mental health services into primary care. In 2007, VHA implemented primary care-mental health integration, which co-located mental health services in primary care (Johnson-Lawrence et al., 2012). In 2010, the VHA also implemented a patient-centered medical home model, referred to as Patient Aligned Care Teams (PACT), to further enhance care coordination for veterans in primary care with and without mental illness (Rosland et al., 2013). Despite these efforts, most veterans who screen positive for a mental health condition in primary care never receive mental health treatment (Mott et al., 2014). Consequently, such patients are at increased risk for emergency department visits, hospitalizations, and premature mortality (Trivedi et al., 2015). Time constraints on providers (Abraham et al., 2017) and costs associated with traveling to clinics for treatment contribute to this problem (Elnitsky et al., 2013), as does stigma about seeking mental healthcare among veterans (Mittal et al., 2013).

Rapid growth in the use of smartphones and development of applications (“apps”) intended for self-care provide new opportunities for delivering services to patients with mental health conditions. Apps have the benefit of overcoming barriers to care access (Donker et al., 2013; Prentice & Dobson, 2014) and provide patients with an option for care that is private, self-directed, and available as needed (Whealin et al., 2014). VHA and Department of Defense (DoD) have invested considerable resources to developing mobile apps for a range of mental health problems (Gould et al., 2019). Evidence for the effectiveness of mobile apps for improving users’ mental health outcomes is emerging (Armstrong et al., 2017). For example, systematic reviews and randomized controlled trials support the effectiveness of apps for reducing symptoms of depression (Mantani et al., 2017). However, given that poor patient engagement and sustainability greatly limit the routine use and effectiveness of mobile apps (Erbes et al., 2014; Lipschitz et al., 2019; Reger et al., 2021), strategies are needed to boost engagement of these promising tools.

To address patients’ challenges of access and engagement in care, healthcare systems have incorporated peers into the delivery of support-enhancing interventions (Chinman et al., 2008). In VHA, peer specialists are veterans with lived experience of mental illness, who are now in recovery and trained to provide services to other veterans actively struggling with their mental health. This role includes providing veterans with social and emotional support, assisting with care navigation in VHA or the community, and serving as a role model for self-care of one’s health problems. Though originally working in substance use and mental health settings, VHA and other healthcare systems have expanded the peer specialist role to primary care to support patients’ self-care efforts and navigation of the care system (Chinman et al., 2017; Daaleman & Fisher, 2015).

Given their recent expansion into primary care services, peers may be an ideal workforce to facilitate patient uptake of apps and implementation of mobile mental health tools into these settings more generally. Peers can enhance patients’ engagement with mobile apps that are intended for self-care of mental health problems by helping to orient patients to these apps and providing technical support and accountability for use (Ray et al., 2017). The use of peers in this regard aligns with research supporting the value of using peers to increase engagement and adherence to e-health interventions in general (Possemato et al., 2019). In terms of mHealth specifically, recent studies indicate strong support among peers and primary care providers for using peers to facilitate patients’ engagement with mobile apps (Miller et al., 2019). For example, while limited awareness and ineffective promotion of mobile mental health tools are barriers to their implementation in primary care (Montena et al., 2021; Reger et al., 2021), peers are viewed as ideal champions of mHealth by sharing their personal experiences with these tools (Montena et al., 2021).

In the present study, we conducted a proof-of-concept test of the use of peers in the implementation of mobile mental health apps in primary care settings. With input from a committee of stakeholders, we designed a phone-based protocol for peers to support mobile mental health app use and then conducted a single-arm trial to evaluate the feasibility, acceptability, and clinical utility of this approach at two VHA medical centers. Given that comorbidity is the norm regarding mental health problems in veterans (Trivedi et al., 2015), we focused on VHA/DoD mobile apps that address common transdiagnostic mental health symptoms – i.e., low mood, anger problems, stress/anxiety, sleep problems, and trauma symptoms. Feasibility was assessed in terms of patients’ intervention engagement (app use, session attendance with peers) and peers’ fidelity to the protocol. Acceptability was assessed via quantitative ratings of satisfaction and helpfulness of the approach as well as feedback from patients and the peers delivering the intervention. Clinical utility was evaluated in terms of the magnitude of within-person change in clinical outcomes pre- to post-intervention and progress with personal health goals.

Methods

Participants

Electronic medical records were reviewed to identify patients at one West Coast and one East Coast VHA medical center who (a) had a primary care appointment at VHA within six weeks of recruitment, (b) screened positive for depression on the 2-item Patient Health Questionnaire (PHQ-2; Kroenke, Spitzer, & Williams, 2003) during their most recent primary care visit, and (c) did not engage in mental health services since their positive depression screen. We chose to limit the sample to only those with positive depression screens because depression is the mental health diagnosis that is most prevalent among veteran primary care patients, has the highest rates of comorbidity with other mental health disorders as well as other health problems that are commonly addressed by VA mobile apps – e.g., insomnia, anger (Trivedi et al., 2015), and is most commonly screened for in primary care settings (Mulvaney-Day et al., 2018). Eligible veterans (n = 504) were recruited between September 2020 and February 2021 through study invitations sent by mail; 318 of these veterans received a follow-up phone call one week later. A total of 307 individuals (60.9%) did not respond to any study contacts, 128 individuals (25.4%) declined to participate, 18 individuals (3.6%) were found to be ineligible because they either did not own a smart device that could support the mobile apps or they reported receiving mental health services in the past six weeks, and 51 individuals (10.1%) were found to be eligible and expressed interest in participating in the study. There were no significant differences between study participants and non-participants in terms of gender, age, or race (ps>.05). In terms of ethnicity, participants were more likely than non-participants to be Hispanic/Latino: χ² (1, 454) = 7.01, p=.008.

Across sites, 39 of the 51 eligible participants were enrolled and completed a baseline interview, and 12 declined to participate or were unable to be reached to complete the interview. Quota sampling was used to ensure that under-represented demographic categories in veteran populations were represented in the final sample (i.e., those < 60 years of age, non-White/Caucasian race/ethnicity, and female). Sample characteristics are provided in Table 1. Participants were mostly male (79.5%) and reported their race/ethnicity as White (84.6%) and Non-Hispanic (71.8%). On average, participants were 53.18 years of age (SD = 15.68) with an annual income of $49,676. Most participants had at least some college education (82.0%), were not currently married (53.8%), owned their own residence (64.1%), and were not employed at the time of enrollment (71.8%). At baseline, 84.6% (n = 33) of participants indicated a history of mental health treatment, mostly on an outpatient basis. Four (10.3%) participants reported receiving prior care from a peer specialist. Most participants (n = 24, 61.5%) had Android smartphones. On a 1 (very uncomfortable) to 5 (very comfortable) scale, participants reported a high level of comfort with both smartphones (M = 4.2, SD = 1.1) and apps (M = 4.1, SD = 1.3). A minority of participants reported any past use of VA apps (n = 8, 20.5%) or non-VA apps (n = 13, 33.3%) to achieve their personal health goals.

Table 1.

Sample characteristics.

Variable	n (%) or M (SD)
Gender
Male	31 (79.5%)
Female	8 (20.5%)
Age (years)	53.18 (15.68)
Race
Asian	1 (2.6%)
Black/African American	2 (5.1%)
White	33 (84.6%)
Mixed Race	1 (2.6%)
Declined to Answer	2 (5.1%)
Ethnicity
Hispanic/Latino	11 (28.2%)
Non-Hispanic/Non-Latino	28 (71.8%)
Years of Education
High School Diploma/GED	7 (17.9%)
Some College/Associate Degree	20 (51.3%)
Bachelor’s Degree	5 (12.8%)
Graduate Degree	7 (17.9%)
Marital Status
Married	18 (46.2%)
Divorced or separated	13 (33.3%)
Never married or widowed	8 (20.5%)
Living Status (past 30 Days)
Own residence	25 (64.1%)
Renting	12 (30.8%)
Living with friends/family	2 (5.1%)
Job Status
Employed Full- or Part-time	11 (28.2%)
Retired/Unemployed	19 (48.7%)
Student/Homemaker/Other	9 (23.1%)
Annual Income a	$49,676 (40,710.69)
Mental Health Treatment History
Inpatient/Residential	9 (23.1%)
Outpatient	32 (82.1%)
Peer Specialist Care	4 (10.3%)

Notes. N = 39.

^aN = 37.

Procedures

Following informed consent, participants completed a baseline interview to obtain sociodemographic information, mental healthcare treatment history, smartphone and app comfort level and use, mental health symptoms, and ratings of personal health goals. Participants were offered use of up to five VA mobile apps for a range of transdiagnostic mental health issues (i.e., stress, low mood, sleep problems, anger, and trauma). Research staff assisted participants with downloading the apps to their smartphones by providing unique invitation codes prior to beginning the peer sessions. Participants were then scheduled to meet with a peer for four sessions over the following eight weeks. Participants were re-interviewed by a research staff member post-treatment (n = 23; 59% retention) to assess their satisfaction with and feedback on the intervention, in addition to completing the same measures used during the baseline interview. Participants were re-interviewed an average of 8.17 weeks (SD = 3.46) after the baseline interview. Participants were compensated $25 for each interview completed. Study procedures were approved by the local institutional review boards at each site.

Attrition analyses were conducted to examine whether completers (i.e., those who completed the follow-up interview; n = 23) differed from the non-completers (n = 16) on sample characteristics and outcome variables at baseline. No significant differences were observed across completers and non-completers on any sociodemographic factors or outcome variables at baseline. The proportion of participants with a history of any mental health residential treatment was significantly higher among completers (M = 0.40, SD = 0.21) than non-completers (M = 0.31, SD = 0.48), t(37)= 2.40, p = 0.022. Conversely, the proportion of participants with a history of prior care from a peer specialist was significantly lower among completers (M = 0.00, SD = 0.00) than non-completers (M = 0.25, SD = 0.45), t(37) = 2.70, p = 0.010.

Mobile applications.

The VA apps offered to patients were: 1) Anger and Irritability Management Skills (AIMS); 2) Insomnia Coach; 3) Mindfulness Coach; 4) Mood Coach; and 5) PTSD Coach. AIMS is used to assess, track, and manage anger. Insomnia Coach allows users to assess, monitor, and self-manage insomnia symptoms. Mindfulness Coach provides mindfulness training to reduce stress and promote emotional balance. Mood Coach improves low mood symptoms through behavioral activation. PTSD Coach provides information, support, and tools to manage trauma symptoms. More information about the mobile apps can be found at https://mobile.va.gov/appstore/. Participants were provided links, QR scan codes, and unique invitation codes to instrumented versions of each app, which allowed the research team to track and monitor participants’ app usage. All apps were available on both iOS and Android platforms, with the exception of Mood Coach, which was only available on the iOS platform.

Peer phone protocol.

The protocol for peers to support patients use of the mobile apps was developed using input received from a committee of stakeholders during two meetings in May and June of 2020. This committee consisted of 10 individuals: 5 researchers (experts on veteran mental health care, primary care and mental health integration, implementation science, peer support, and mHealth); 4 administrators from VA Central Offices overseeing primary care–mental health integration, peer support services, mHealth, and whole health care; and a representative of a Veteran and Family Engagement Council. Using the feedback received from these stakeholders, a four-session manualized protocol was developed. The initial session involved peers (a) assisting participants with identifying their personal health goals, (b) reviewing the benefits of using mobile apps for self-care, (c) assessing participants’ tech literacy and access, (d) reviewing the privacy and security features of VA apps, (e) reviewing the five apps offered in the study and assisting participants with app selection, and (f) scheduling the next session. For the app selection component, the peers used a shared decision-making approach to encourage participants to consider which app(s) could best address their needs and help them reach their personal health goals; neither the peer nor the participant was restricted in how many apps they could recommend or choose, respectively. The follow-up sessions involved peers (a) assessing participants’ app usage since the last session, (b) discussing the content of any apps that were used, (c) discussing the fit of the app(s) used with participants’ personal health goals, (d) providing technical support, and (e) encouraging participants’ ongoing use of the app(s). The final session included guidance for providing closure that would encourage patients’ continued use of the mobile apps.

Sessions were delivered via phone by three VHA-employed peer specialists with lived experience with mental health problems who were now in recovery. Prior to study enrollment, the peers participated in six weekly, one-hour training sessions to review each of the apps, study procedures, and the components of the phone sessions. Over this training period, the peers were asked to download each app to their personal smartphones and to use the app to become familiar with its functionality and features. In addition to the essential components of each session described in the next paragraph, peers were instructed to practice their usual peer role, including sharing their lived experience with mental health problems (as appropriate), providing emotional support, and supporting participants linkage to health care services in the VA or community, as needed. With participants’ permission, phone sessions were audio-recorded. During the active intervention phase of the study, peers attended a weekly one-hour fidelity clinical supervision videoconference with the first author (DB) and last author (KP).

Fidelity monitoring.

Fidelity was rated using distinct scales for each of the four sessions in the protocol. For a given session, criteria for the rating scale were divided into required components and other components. Required components included selecting an app, discussing an action plan, addressing privacy and security concerns, and encouraging participants’ continued use of an app. Other components included peer-specific skills such as identifying strengths and role-modeling a healthy living style. Components were chosen to reflect instructions in the peer manual, core skills for the peer specialist role, and study-specific note templates. Ratings were determined by reviewing the notes entered into participants’ medical records after each session (96 sessions in total). Independent raters from the research team reviewed notes in the medical record for each of the completed peer sessions. Based on these ratings, peers fulfilled the required criteria for these sessions 94% of the time.

Measures

App usage.

Instrumented versions of the mobile apps allowed for collection of objective app use data. However, at least 9 participants encountered technical challenges when attempting to access one or more of the instrumented versions of an app. In these situations, research staff prompted participants to use the publicly-available versions of the app. Three of these participants did not have functional invitation codes for any of the apps; therefore, complete objective app use data were only available for 36 participants. Given the technical challenges with using instrumented versions of the apps for some participants, the objective app use data may provide an underestimate of app usage in the sample. Therefore, participants also reported their app usage for each of the five apps during the follow-up interviews. If a participant indicated that they had used an app, they were asked “How often did you use the app?” (1 = Once, 2 = More than once, but less than weekly, 3 = One to two times per week, 4 = Daily or almost every day), and “How likely are you to continue to use the app after the study?” (1 = Not likely, 2 = Somewhat likely, 3 = Likely, 4 = Very likely”). The participants were also asked if they had used any other VA apps since the last interview. Among the 23 participants who completed the follow-up interviews and reported using at least one app, 19 (82.6%) had objective app use data supporting their use. Among the 18 participants who completed the follow-up interviews and reported using two or more apps, 13 (72.2%) had objective app use data supporting use of two or more apps.

Intervention engagement and satisfaction.

At the follow-up interview, participants rated 11 items regarding the helpfulness of the peer support they received using a 4-point scale (1 = Poor to 4 = Excellent). Participants also used a 4-point scale to rate the perceived flexibility (1 = Not at all flexible to 4 = Very flexible) and how much discussions with the Peer helped them understand and use skills taught in the app(s) (1 = Not at all to 4 = A lot) of peer support of mobile app use. The 8-item Client Satisfaction Questionnaire (CSQ; Attkisson & Zwick, 1982) was used to calculate overall satisfaction with the intervention. CSQ responses of three or greater indicate at least some satisfaction with the intervention. Therefore, a CSQ total score of 24 (i.e., an average response of ‘3’ per item) was selected as the benchmark for acceptability. The internal consistency of the CSQ total scores at follow-up was high (α = .82).

Psychological distress.

The Depression Anxiety Stress Scales (DASS) is a self-report inventory to assess psychological distress related to symptoms of depression and anxiety, and general stress (Lovibond & Lovibond, 1995). We used a short version, which includes 21 items (DASS-21). Participants were asked to rate how much each item applied to them over the past month on a 4-point scale (0 = Did not apply to me at all to 3 = Applied to me very much or most of the time). Item responses were summed to create a total score, which provides a broad index of psychological distress and had excellent internal consistency at baseline (α = .92) and follow-up (α = .90). Internal consistency also ranged from good to excellent for the 7-item DASS-21 stress subscale at baseline (α = .88) and follow-up (α = .82).

Insomnia symptoms.

The Insomnia Severity Index (ISI) is a 7-item self-report measure that assesses problems with sleep based on internationally-recognized sleep disorder criteria (Dieperink et al., 2020). Participants were asked to rate their insomnia symptoms over the past two weeks using a 5-point scale with higher scores indicating greater problems due to sleep concerns. ISI total scores (sum of all item responses) had good internal consistency at baseline (α = .88) and follow-up (α = .85).

Trauma symptoms.

The PTSD Checklist for DSM-5 (PCL-5) is a 20-item self-report inventory that assesses symptoms of PTSD based on DSM-5 criteria (Weathers et al., 2013). Participants were asked to rate how much they are bothered by their PTSD symptoms over the past month on a 5-point scale (0 = Not at all to 4 = Extremely). Total scores (sum of all item responses) had excellent internal consistency at baseline (α = .93) and follow-up (α = .92).

Depression symptoms.

The Patient Health Questionnaire-9 (PHQ-9) is a 9-item self-report measure of symptoms of depression (Kroenke, Spitzer, & Williams, 2001). Participants were asked to rate how often the symptoms described in each item bothered them over the past two weeks using a 4-point scale (0 = Not at all to 3 = Nearly every day). Total scores (sum of all item responses) had acceptable reliability at baseline (α = .72) and follow-up (α =.76).

Personal health goals.

The Goal Attainment Measure (GAM) was developed by researchers from the Bedford VA Medical Center (B. Bokhour, personal communication). Participants were presented with a list of goals and asked to rate each on a 7-point scale in terms of whether it is a goal for them at this time and their current level of progress in meeting the goal (0 = N/A or not a goal at this time, −1 = getting worse, 1 = almost no progress, 2 = a little progress, 3 = some progress, 4 = a lot of progress, 5 = goal reached or almost reached). In this study, participants were asked to rate their interest/ progress on 10 goals: engage in enjoyable or meaningful activities, find greater meaning and purpose in my life, manage my stress, manage my anger, engage in mindfulness activities, improve my sleep and feel more rested, find ways to relax, manage my anxiety or depression, manage my PTSD, and other. The GAM was readministered at the follow-up interview to measure progress on goals that were endorsed at baseline.

Feedback on the intervention.

During the follow-up interview, participants were asked to provide feedback on the intervention, including their perceptions of the mobile apps and the support they received from the peers for using the apps. Feedback was solicited via open-ended questions that were adapted from an interview used in prior studies of peer-supported e-health interventions (Blonigen et al., 2020; Possemato et al., 2019). In addition, following study completion, each of the three study peers was interviewed to provide feedback on the mobile mental health apps that were provided to participants and the app support they provided to participants. The peer interview included quantitative ratings regarding the perceived helpfulness of the intervention and impact on patients, as well as open-ended questions regarding barriers and facilitators to delivering mobile app support to participants, and the feasibility of implementing this approach more routinely within their primary care clinic. Feedback interviews with both participants and peers were audio-recorded with their permission.

Data Analysis

Descriptive statistics were used to quantify measures of intervention engagement and satisfaction, including app usage, peer engagement, peer satisfaction, and overall intervention satisfaction. Repeated measures ANOVAs using complete cases were used to identify the significance of changes in clinical outcomes pre- and post-treatment. Cohen’s d effect sizes were calculated to estimate the magnitude of within person changes in (a) clinical outcomes pre-and post-treatment, and (b) clinical outcomes pre- and post-treatment categorized into subgroups based on those who used specific apps. Specifically, the clinical outcomes selected for the effect size calculations were representative of the symptoms targeted by at least one app (i.e., total scores on the DASS-21, DASS-21 Stress subscale, ISI, PCL-5, and PHQ-9). Among the subgroups of participants who reported using specific apps, analyses did not control for participants’ use of multiple apps. Progress on personal health goals was measured in terms of the proportion of participants who endorsed a goal as being applicable at both time points and who selected a higher rating (i.e., reported greater progress towards goal attainment) at the follow-up assessment.

Post-treatment qualitative responses were analyzed using Rapid Qualitative Analysis (Beebe, 2014; Neal et al., 2015) to identify themes of what participants (a) liked best about the program, (b) liked least about the program, and (c) suggested as improvements for the program. Responses were originally recorded in VA REDCap (Harris et al., 2009) by interviewers and then copied into an Excel matrix to identify preliminary themes for each question (matrix columns) across participants (matrix rows). Three authors (DB, JS, AM) independently reviewed the matrix to identify key themes. The preliminary themes were then reviewed until team consensus was reached, and the final themes were reviewed by all authors. The same methodology was used to examine post-treatment qualitative responses from peers to identify barrier and facilitator themes to their delivery of the intervention with patients.

Results

Intervention Engagement and Satisfaction

Frequencies of participants’ self-reported app usage are provided in Table 2. All participants who completed the follow-up interview (n = 23) reported use of at least one of the five VA apps that were offered to them. The most commonly reported apps used were Mindfulness Coach (n = 17; 73.9%) and Insomnia Coach (n = 16, 69.6%). Use of multiple apps use was common; on average, participants reported using 3.04 (SD = 1.46) of the five apps, and 78.3% (n = 18) reported using at least two apps. Based on the objective app usage data, 26 out of the 36 (72.2%) participants who had functional invitation codes used at least 1 instrumented app during the study period, and 16 (44.4%) used 2 or more apps. Mindfulness Coach (n = 15) and Insomnia Coach (n = 14) were also the most used apps, per the objective data. Objective app use (1 = yes) was positively correlated with study completion (r_s = .41, p = .01) and negatively correlated with DASS-21 total scores at follow-up (r_s = −.42, p = .048). There were no other significant correlations between objective app use and clinical outcomes or progress on personal health goals.

Table 2.

Self-reported app usage.

App	Any use n (%)	Frequency of use M (SD)	Likely to continue M (SD)
AIMS	10 (43.5%)	3.10 (0.88)	2.50 (0.97)
Insomnia Coach	16 (69.6%)	2.63 (1.20)	2.19 (1.38)
Mindfulness Coach	17 (73.9%)	2.53 (1.13)	2.94 (1.29)
Mood Coach	13 (56.5%)	2.23 (1.01)	2.38 (1.26)
PTSD Coach	14 (60.9%)	2.50 (1.02)	2.50 (1.35)

Notes. N = 23. AIMS = Anger and Irritability Management Skills. Frequency of use scale included ratings of 1 (once), 2 (more than once, but less than weekly), 3 (one to two times per week), and 4 (daily or almost every day). Likely to continue scale included ratings of 1 (not likely), 2 (somewhat likely), 3 (likely), and 4 (very likely).

In terms of peer engagement, across all participants (n = 39) a mean of 2.46 phone sessions (SD = 1.73) were completed. Most participants (n = 29; 74.3%) attended at least one session; 25.6% (n = 10) did not complete any sessions, 7.7% (n = 3) completed one session, 10.3% (n =4) completed two sessions, 7.7% (n = 3) completed three sessions, and 48.7% (n = 19) completed all four sessions. Sessions lasted 20.93 minutes, on average (SD = 2.88). Regarding the perceived helpfulness of the peer sessions, on a scale of 1 (quite unhelpful) to 4 (very helpful), average ratings were high in terms of receiving instructions on how to use the apps (M = 3.28, SD = 1.07), setting personal health goals (M = 3.00, SD = 0.87), discussing my symptoms (M = 3.29, SD = 0.77), discussing my values and strengths (M = 3.09, SD = 0.95), discussing the app content to better understand it (M = 3.24, SD = 0.95), discussing the app content to apply it to my own life (M = 3.19, SD = 0.98), and making a plan to practice skills I learned on the app (M = 3.36, SD = 0.63). Participant ratings were also high in terms of the perceived flexibility of the peer support services received (M = 3.82, SD = 0.51; (1 = not at all flexible, 4 = very flexible) and how much they believe peer support helped them to understand the skills taught in the app (M = 3.33, SD = 0.73; 1 = not at all, 4 = a lot). In terms of global satisfaction with the intervention, the average CSQ total score was 25.71 (SD = 3.95), which was above the benchmark for success (i.e., 24). Further, 71.4% (n = 15 of 21) of participants that completed the CSQ at follow-up had a total score above the benchmark for success.

Within-Person Changes in Clinical Outcomes

The results of the ANOVAs and magnitude and significance of changes on the DASS-21, ISI, PCL-5, and PHQ-9 scores from pre- to post-treatment are provided in Table 3. Effect size changes were small to moderate in magnitude across these measures of psychological distress and stress, and symptoms of insomnia, trauma, and depression. The largest improvements were observed for self-reported stress (p < .01). All other pre- to post-treatment changes were nonsignificant. In ancillary analyses, we conducted repeated measures ANOVAs to determine whether pre/post changes in clinical outcomes were moderated by past app use. All interactions were non-significant. We also conducted independent samples t-tests to compare scores on the clinical outcome measures at follow-up between those who did and did not have past app use. There were no significant mean differences between these groups on any outcome measures.

Table 3.

Within-person changes in clinical outcomes (pre/post).

	Baseline		Follow-up		Repeated Measures ANOVA
Measure	M	(SD)	M	(SD)	Cohen’s d	[95% CI]	F(df)	p
DASS-21 Total	48.35	(26.20)	41.04	(24.88)	−0.29	[−0.86, 0.30]	1.95 (1, 22)	.177
DASS-21 Stress	20.09	(11.50)	14.35	(10.01)	−0.53	[−1.11, 0.06]	8.56 (1, 22)	.008
ISI Total	13.87	(7.28)	12.83	(7.28)	−0.14	[−0.72, 0.44]	2.14 (1, 22)	.157
PCL-5 Total	32.48	(17.23)	29.78	(16.72)	−0.16	[−0.74, 0.42]	2.73 (1, 22)	.112
PHQ-9 Total	12.87	(5.36)	11.65	(5.82)	−0.22	[−0.79, 0.37]	2.68 (1, 22)	.116

Notes. N= 23. DASS-21 = Depression Anxiety Stress Scale – 21 item; ISI = Insomnia Severity Index; PCL-5 = PTSD Checklist for DSM-5; PHQ-9 = Patient Health Questionnaire – 9 item.

In terms of personal health goals, participants endorsed an average of 5.92 (SD = 2.18), out of a possible 10, personal health goals at baseline. Majorities of participants endorsed the following goals: “manage my anxiety or depression (n = 29, 74.4%), “engage in an enjoyable or meaningful activity” (n = 27, 69.2%), “find greater meaning and purpose in my life” (n = 25, 64.2%), “manage my stress” (n = 25, 64.1%), “improve my sleep and feel more rested” (n = 25, 64.1%), “manage my anger” (n = 22, 56.4%), “find ways to relax” (n = 21, 53.8%), and “engage in mindfulness activities” (n = 20, 51.3%). At follow-up, participants endorsed an average of 6.3 (SD = 2.65) personal health goals. On average, participants reported progress on 1.91 (SD = 1.00) personal health goals from baseline to follow-up, and 20 out of 23 completers (87%) reported progress on at least one goal over time. Engaging in mindfulness activities was the goal for which the majority of participants (8 out of 12; 66.7%) reported progress from pre- to post-treatment.

Intervention Engagement and Within-Person Changes on Clinical Outcomes

Table 4 provides the results of exploratory analyses estimating the magnitude of pre- to post-treatment change in clinical outcomes among the subgroups of participants who reported using specific apps. In general, improvements were largest and most robust across all subgroups for the outcomes of psychological distress and stress. Improvements in insomnia were moderate in magnitude for those who used the Insomnia Coach app but were small in magnitude among all other app subgroups. Improvements in trauma symptoms were small and of comparable magnitude for all app subgroups. Improvements in depression symptoms ranged from small (PTSD Coach, Mood Coach, and Insomnia Coach users) to moderate in magnitude (AIMS and Mindfulness Coach).

Table 4.

Intervention Engagement and Within-Person Changes on Clinical Outcomes.

		AIMS (n = 10)		INSOMNIA (n = 16)		MINDFULNESS (n = 17)		MOOD (n = 13)		PTSD (n = 14)
Measure	Time	M (SD)	d [95% CI]	M (SD)	d [95% CI]	M (SD)	d [95% CI]	M (SD)	d [95% CI]	M (SD)	d [95% CI]
DASS-21 Total	Baseline	53.80 (27.90)		48.63 (26.57)		46.94 (27.61)		53.54 (28.40)		55.57 (28.15)
	Follow-up	39.00 (25.63)	−0.55 [−01.42, 0.36]	40.75 (21.69)	−00.32 [−1.01, 0.38]	34.00 (24.93)	−00.49 [−01.16, 0.20]	44.00 (26.85)	−00.35 [−01.11, 0.44]	47.29 (26.03)	−00.31 [−01.04, 0.45]
DASS-21 Stress	Baseline	22.20 (10.73)		19.50 (10.55)		18.94 (11.12)		22.77 (10.97)		23.14 (11.03)
	Follow-up	15.40 (9.75)	−00.66 [−01.53, 0.26]	13.13 (7.97)	−00.68 [−01.38, 0.05]	12.24 (9.38)	−00.65 [−01.32, 0.05]	16.46 (10.65)	−00.58 [−01.35, 0.22]	17.14 (9.95)	−00.57 [−01.31, 0.20]
ISI Total	Baseline	13.20 (6.63)		15.25 (5.29)		13.24 (6.96)		14.23 (7.52)		14.71 (7.98)
	Follow-up	12.50 (8.54)	−00.09 [−00.96, 0.79]	12.94 (6.32)	−00.40 [−01.09, 0.31]	12.18 (7.32)	−00.15 [−00.82, 0.53]	13.77 (7.32)	−00.06 [−00.83, 0.71]	14.21 (7.99)	−00.06 [−00.80, 0.68]
PCL-5 Total	Baseline	33.10 (17.07)		32.81 (16.21)		29.59 (16.92)		35.92 (14.71)		38.21 (15.64)
	Follow-up	29.30 (16.91)	−00.22 [−01.09, 0.66]	29.50 (14.57)	−00.22 [−00.91, 0.48]	26.29 (15.06)	−00.21 [−00.88, 0.47]	32.29 (14.65)	−00.25 [−01.01, 0.53]	34.79 (16.67)	−00.21 [−00.95, 0.54]
PHQ-9 Total	Baseline	13.20 (5.09)		13.38 (5.04)		12.88 (5.61)		14.08 (5.82)		13.93 (5.62)
	Follow-up	10.30 (4.79)	−00.59 [−01.46, 0.33]	12.31 (5.49)	−00.20 [−00.89, 0.50]	10.24 (5.31)	−00.48 [−01.15, 0.21]	12.31 (5.15)	−00.32 [−01.09, 0.46]	12.93 (5.97)	−00.17 [−00.91, 0.57]

Notes: N=23. DASS-21 = Depression Anxiety Stress Scale – 21 item; ISI = Insomnia Severity Index; PCL-5 = PTSD Checklist for DSM-5; PHQ-9 = Patient Health Questionnaire – 9 item.

Open-ended Feedback

Participant.

When asked what they liked about the intervention, the number of participants who referenced the human support elements (n = 15) versus aspects of the mobile apps (n = 13) were comparable. In terms of human support, participants highlighted the importance of working with a fellow veteran with lived experience whom they could relate to and trust. Participants also reported liking that the peer kept them accountable and motivated to use the apps (“Meeting with a peer regularly kept me motivated, knowing that I would have to discuss this means I’d have to work through this and kept me accountable.” [ID: 218]). In terms of apps, patients viewed them as user-friendly and liked both the veteran-centric nature of the app content as well as the flexibility of using them whenever they needed (“[The apps were] intuitive, simple, and you can see the differences between them.” [ID: 105]).

When asked what they disliked about the intervention, some participants highlighted the lack of any in-person interaction with peers during the study (“No human connection using the apps. We are already on our phones all the time. In person would have been much more personable.” [ID: 224]). They also reported various technical issues and limitations of the apps, for example, glitches, not available on all platforms, no audio for older adults (“There are some issues with the apps… glitches. That is what I liked least.” [ID: 115]). Some participants also viewed the apps as redundant in their content, and others did not like that the apps were self-contained and could not be linked to their medical records.

Regarding suggestions for improvements, several participants desired more contact and guidance earlier in the intervention that would be in-person and focus on downloading and setting up apps and assisting with initial navigation and understanding how to use the apps (“More contact initially; someone should be there to help download the app in person or by video.” [ID: 102]). Some participants also recommended some mix of in-person and virtual sessions, either by phone or video. Others suggested that the intervention needed to be longer and that four weeks was insufficient to build a relationship with the peer. Finally, some participants preferred more of a match between themselves and the peer in terms of lived experiences (“Find someone who has more similar life experience to the veteran, not just you’re the next one in line.” [ID: 204], “…find other veterans who have had similar experiences to what I have endured.” [ID: 219]).

Peers.

On a scale of 0 (not helpful) to 10 (very helpful), on average, the peers rated the mobile apps as being moderately helpful for patients in terms of learning about mental illness and treatment (M = 6.00, SD = 3.46), tracking mental health symptoms (M = 6.00, SD = 2.65), and managing mental health symptoms (M = 7.00, SD = 0.00), and slightly less helpful for finding support (M = 4.67, SD = 3.22). On a scale of 0 (no impact) to 10 (a lot of impact), on average, the peers rated the mobile apps as having moderate impact on patients’ ability to manage their mental health symptoms (M = 6.67, SD = 1.16) but lower impact on patients’ engagement in additional mental health services after the study (M = 4.00, SD = 4.00).

Using open-ended questions, the peers were asked which mobile apps they found most and least helpful for patients. All peers reported the Mindfulness Coach app as one of the more helpful apps, and two of the three peers reported PTSD Coach as helpful as well. Two of the three peers reported Mood Coach and Insomnia Coach as the least helpful apps for patients. All the peers perceived mHealth support as feasible to implement within their role in primary care settings, and two peers indicated that they have already incorporated mHealth into their PACT role (“It’s do-able. It’s done all the time.” [Peer 3]). Regarding challenges of the program, peers reported that they found it difficult to help veterans consistently use the apps due to additional participant commitments (e.g., work schedule, parenting). Additionally, peers indicated that the lack of in-person interactions with the veterans posed a significant challenge to supporting their app use, as did a lack of tech literacy for some older veterans and those without access to reliable Wi-Fi (“Some people have difficulty with technology and reliable Wi-Fi.” [Peer 2]). For suggested modifications, peers recommended providing some sessions in-person or through video conferencing and providing veterans with more guidance and education about the specific features of each apps (“It was difficult to do sessions over the phone…[would have been] easier to do sessions in person or on VVC.” [Peer 2]).

Discussion

We designed a protocol for peers to support implementation of mobile mental health apps in primary care settings and tested its feasibility, acceptability, and clinical utility with patients across two VA sites. The feasibility of the approach was supported, based on the findings that (a) the majority of completers reported use of multiple apps, and (b) across all participants approximately three-quarters attended at least one peer session and nearly half completed all four sessions. The rate of app usage among completers must be qualified by the low rate of retention for the follow-up interview and potential for selection effects, as well as limitations on the availability of the objective app use data. Comparisons of the frequency of app usage from prior research are also complicated by the fact that all prior studies of VA mobile apps have focused on use of a single app. Feasibility was also supported by peers’ high fidelity to the protocol, suggesting that they were able to adhere to the core elements of the protocol. Patient acceptability of the intervention was found to be high based on global ratings of satisfaction that exceeded an a priori benchmark, as well as ratings of the perceived helpfulness of the peer sessions in terms of both app-specific and general peer functions.

Preliminary evidence for the clinical utility of the intervention was indicated by significant reductions in self-reported stress and the vast majority of participants reporting progress on at least one personal health goal. Reductions in clinical symptoms such as depression, insomnia, and trauma were observed, though they were generally small in magnitude and not significant. These findings are consistent with the peer protocol, which was designed to focus on the personal health goals of participants rather than management of specific conditions. That said, peers were encouraged to use shared decision-making in supporting patients’ app selection and to recommend specific apps as applicable based on the problems and personal health goals patients reported in the first session (e.g., patients who described improved sleep as a personal health goal were encouraged to review the Insomnia Coach app). With the exception of improvements in insomnia symptoms being largest for those who reported using the Insomnia Coach app, app-specific effects for the clinical outcomes were generally not detected. Attempts to interpret whether observed changes in the clinical outcomes were driven by specific apps are complicated by the fact that (a) there was insufficient power to detect such changes in the app subgroups, and (b) the subgroups of app users were not mutually exclusive. Further, as this was a pre-post trial without a comparator, conclusions regarding the efficacy of the intervention cannot be drawn.

Implications for Delivery of Mobile Mental Health in Primary Care

Strong interest among veterans in the use of mobile mental health apps has been established (e.g., Erbes et al., 2014; Lipschitz et al., 2019). How to achieve greater penetration of these tools in patient care, however, remains an ongoing question, particularly for those seen in primary care settings. Reger and colleagues (2021) reported that, despite nearly ubiquitous smartphone ownership among a sample of veteran primary care patients, only 39% had heard of any of the VA/DoD mobile mental health apps and only 14% had ever used one of them. Such low rates were also reported among those who endorsed mental health problems. Lack of awareness of specific apps and concerns about the security and privacy of the app data are common barriers to app use in this population (Lipschitz et al., 2019; Reger et al. 2021). To overcome these barriers, support specialists who are embedded within clinical settings and dedicated to orient and socialize patients to mHealth tools have been proposed (Mohr et al., 2011; Reger et al., 2021). Peers have only recently been tested in such a role, though only for use of a web-based intervention and not within a primary care setting (Possemato et al., 2022). Peers may be ideally suited to this support role, which is theorized to be maximally effective when a coach is seen as trustworthy, benevolent, and having legitimacy in terms of experience in expertise or personal experiences (Mohr et al., 2011). The qualitative findings from this study as well as other research highlights how peers may have unique strengths as a mHealth support specialists (Miller et al., 2019; Montena et al., 2021). The scalability of this approach, however, will be dependent on an ongoing expansion of the peer specialist role in primary care settings (Chinman et al., 2017).

In addition to scalability, the extent to which the model of mHealth support proposed here can be adapted to the different needs of primary care patients may be key to its effectiveness and sustainability going forward. Notably, there was substantial variability in patients’ engagement with peers (i.e., 0 and 4 were the most common frequencies for number of sessions attended), and some patients reported a desire for support beyond eight weeks. It will be critical for future research on this model to determine which patients treated in primary care settings may benefit most from peer support of mHealth, how much support is warranted (a brief, one-time session vs. multiple sessions), when and for how long such support needs to be provided, and for whom a more intensive level of care is needed. An adaptive intervention that could be tailored to the needs of patients and implemented within the context of a stepped-care model of primary care may be ideal (Mohr et al., 2019). For example, for patients who report mild or moderate mental health symptoms or who are unwilling or unable to attend treatment, they could be offered mobile apps and peer support (one session or multiple sessions, depending on their needs and preferences). For patients whose mental health problems do not remit or worsen, they could be stepped up to more intensive forms of treatment, such as outpatient care. Such an approach would align with the Efficiency model of eHealth support, which considers the ratio of benefits for human support of e-health tools to the costs of providing this support (Schueller et al., 2016).

Finally, a unique strength of the approach that was piloted in this study was that it focused on supporting patients’ engagement with a suite of mobile mental health apps from the VA/DoD portfolio. With some exceptions (see Graham et al. 2020), prior studies of coaching of mHealth tools have generally focused on a single app for a specific condition. However, mental health comorbidity is the norm among veterans in primary care (Trivedi et al., 2015), and many of the apps in the VA/DoD app portfolio address transdiagnostic issues that may be applicable to patients with a range of mental health problems (e.g., low mood, anger issues, stress, sleep problems, trauma history). Consistent with a whole health approach to care that is focused on patient-identified health goals rather than treatment of diseases (Collins et al., 2018), it may be ideal for any model of support of mobile mental health in primary care to focus on transdiagnostic, non-disease specific issues that are most important to patients.

Strengths and Limitations

This study had several strengths, including use of both qualitative and quantitative methods to index feasibility, acceptability, and clinical utility of the intervention, recruitment from multiple sites, and testing use of a suite of mental health apps from the VA/DoD portfolio. Some limitations of this research must also be acknowledged. First, although we used quota sampling to ensure representation of different demographic subgroups of veterans, the sample was mostly male, white, older, and retired. Therefore, the results may not apply to veterans who are younger, female, or members of racial/ethnic minorities. The sample is also characterized by a high proportion of participants with a history of mental health care. Although the reasons for this are unclear, it is possible that patients were not currently engaged in mental health treatment because of lower severity of problems than in prior mental health episodes or they currently had barriers to care access that were not present during past episodes. In any event, the current results may not generalize to patients who have no prior history of mental health treatment such that use of a mobile app would be their foray into treatment. In future research, it may be beneficial to enroll sufficient samples of both patients with and without prior mental health care to understand the impact of peer-supported mobile mental health on these distinct populations. The small sample size and low rate of retention of participants at the follow-up time point also limit inferences regarding the clinical utility of the protocol. Further, given that this was a single arm-trial, no conclusions can be made regarding the efficacy of this approach for either app engagement or improvements in clinical outcomes. That is, we cannot determine from the current study whether improvements in any of the clinical outcomes or progress on personal health goals was related to use of the mobile app, the peer support, or the combination of the two. A future trial with multiple conditions, including one without peer support, is needed to isolate the relative impact of peer support vs. app engagement alone vs. the interaction of these components.

Conclusions

The current protocol represents a novel approach to delivering mental health care for primary care patients. The findings suggest that peer-supported mobile mental health is feasible and highly acceptable to patients in primary care who are not currently engaged in mental health treatment. Accordingly, a larger trial of the protocol is warranted. However, given the variability in peer engagement and the qualitative feedback from patients, peer support of mobile apps may function best if it is tailored to the needs of patients over time. Testing an adaptive intervention that could be delivered over a longer time period, incorporates a mix of in-person and virtual sessions, and can be implemented within the context of a stepped-care model in primary care is recommended.

Impact Statement:

Mobile apps can only increase access to mental health treatment if patients actively engage with them. This study found that peers who have a history of mental illness and are trained to support those who are currently struggling with their mental health can help these patients become more engaged with health-related apps.