A Randomized Controlled Trial of a Brief Online Intervention to Reduce Alcohol Misuse and Promote Treatment Initiation among Veterans

Abstract

Objective.

Post-9/11 U.S. veterans exhibit elevated rates of hazardous drinking, alcohol use disorder, and co-occurring mental health conditions, yet many do not access behavioral health treatment. We evaluated BRAVE, a single-session, mobile phone-delivered intervention integrating personalized normative feedback, motivational interviewing, and values-based strategies to reduce alcohol use and promote treatment engagement.

Method.

Post-9/11 young adult veterans aged 18–40 who screened positive for hazardous drinking and were not engaged in behavioral health care (N=748) were recruited via targeted social media ads and randomized to BRAVE or an interactive resources-only control condition. Participants completed baseline and 3-, 6-, 9-, and 12-month follow-up surveys assessing alcohol use, alcohol-related consequences, mental health symptoms, treatment receipt, and readiness for care.

Results.

Compared with control, BRAVE produced a significant sustained reduction in alcohol-related consequences over 12 months (B=−0.07). Post-hoc analyses showed significant decreases in physical and interpersonal consequences from drinking. Though outcomes of drinking frequency, drinking quantity, binge drinking frequency, readiness to seek care, and behavioral health treatment receipt favored BRAVE, none of these treatment effects were significant and effect sizes were very small in magnitude. In post hoc analyses restricted to veterans screening positive at baseline for alcohol use disorder, anxiety, depression, or posttraumatic stress disorder, we observed that among veterans screening positive for anxiety disorders, BRAVE participants were notably 15% more likely to receive treatment during follow-up.

Conclusions.

Findings indicate that this brief, mobile intervention can yield meaningful reductions in alcohol-related harm. Broader dissemination and exploration of multi-session or peer-supported enhancements may strengthen long-term impacts.

Public Health Statement:

Hazardous alcohol use and co-occurring mental health concerns are prevalent and commonly co-occur among post-9/11 veterans. This study demonstrates that BRAVE, a one-time, mobile phone-based intervention, can meaningfully reduce alcohol-related problems. The findings highlight the potential of scalable, technology-delivered approaches to reach veterans who might not otherwise engage in treatment.

Alcohol use disorder (AUD) is prevalent among Americans who served in the U.S. military during or after September 11, 2001, and have since been discharged (post-9/11 veterans). It is estimated to affect upwards of 10% of all veterans (Davis et al., 2025; Seal et al., 2011), with rates of AUD highest among younger veterans (i.e., under age 40) and those who served in combat in Iraq and Afghanistan (Hoggatt et al., 2017; Robinson et al., 2023). Post-9/11 veterans also exhibit concerning rates of mental health disorders, including high rates of depression, anxiety disorders, and posttraumatic stress disorder (PTSD) (Boakye et al., 2017; Fulton et al., 2015; Trivedi et al., 2015). Co-occurrence of AUD with depression, anxiety, and PTSD is also comparatively prevalent; it is more common than not for veterans with AUD to also meet criteria for an additional mental health diagnosis (Dworkin et al., 2018; Knowles et al., 2019; Norman et al., 2018; Teeters et al., 2017). Beyond an AUD diagnosis, a concerning number of veterans (upwards of one-fifth across studies) drink at hazardous levels, a pattern of heavy drinking that places the veteran and others at risk for alcohol-related negative consequences, such as physical health complications (e.g., liver disease, cardiovascular issues, injuries), strained relationships, difficulties in maintaining employment, and increased healthcare utilization and societal costs (Davis et al., 2025). Heavy drinking and co-occurring mental health disorders often emerge during military service but can become exacerbated upon discharge into civilian life (Ayer et al., 2022), suggesting a heightened importance for veterans to initiate evidence-based care to prevent the development of chronic behavioral health conditions that can significantly and negatively impact the course of their lives.

Although evidence-based care for hazardous alcohol use and co-occurring mental health disorders exists for veterans both within the Veterans Affairs Healthcare System (VA) and in the community (Borsari et al., 2011; Pedersen et al., 2020; Perry et al., 2022), upwards of half of the veterans that could likely benefit from care for their symptoms do not seek services from the VA or elsewhere (National Academies of Sciences, 2018; Schell & Marshall, 2008). Indeed, a recent estimate pooling large-scale epidemiological studies with veterans indicated that only about 14% of all veterans received past-year mental health treatment, with just 1.4% receiving substance use treatment during that time (Robinson et al., 2023). Structural, logistical, and belief-related barriers often preclude veterans from accessing services that could be beneficial. Such barriers include a limited understanding about what eligible services are available, difficulties attending appointments (e.g., lack of transportation, long wait times, scheduling conflicts), perceived or actual stigma from others (e.g., belief that colleagues, family, and friends would respect them less), fear of repercussions or career harm (e.g., concerns that their career would not progress if they seek treatment), beliefs that they can handle their problems on their own, or perceptions that available treatments are not effective (Hoge et al., 2004; National Academies of Sciences, 2018; Schell & Marshall, 2008; Vogt, 2011). Addressing the pressing behavioral health needs and barriers of post-9/11 veterans requires designing outreach and intervention programs that connect with veterans outside traditional care settings. Outreach efforts, especially for young adult veterans, who are at highest risk for heavy drinking (Hoggatt et al., 2017; Robinson et al., 2023), can engage them in interventions they might not otherwise receive and address problems early after discharge.

Internet- and mobile phone-based approaches to treatment are promising and can help veterans overcome barriers to accessing care (Belsher et al., 2024; Dedert et al., 2014; Doherty et al., 2017; Zhou et al., 2021). VetChange, a web- and mobile-based self-help program for veteran alcohol use, represents a notable example of this approach and has demonstrated efficacy in reducing drinking among veterans (Brief et al., 2011, 2013). However, these approaches are often lengthy, require substantial commitment, suffer high dropout rates, and assume veterans are ready to address drinking and mental health issues. To address shortcomings in existing brief alcohol interventions for veterans, we designed and pilot-tested a stand-alone, single-session online intervention for young adult veterans (ages 18 to 34) recruited online via social media (Pedersen et al., 2016a; 2017). The intervention targeted veterans outside of care settings who drink alcohol, aiming to reach those not formally seeking treatment for alcohol-related concerns. It was designed to correct veterans’ misperceptions of their peers’ drinking, since these misperceptions, common among young adults more generally and among veterans, are associated with their own drinking behavior (Cox et al., 2019; Garnett et al., 2015; Pedersen et al., 2016b). Importantly, studies of veteran samples demonstrate that perceived drinking norms among “other veterans” are consistently inflated and more strongly predictive of personal alcohol use than actual norms, suggesting a clear and modifiable intervention target (Pedersen et al., 2016a; 2016b). Correcting misperceived norms is among the most widely used and efficacious components of brief interventions for young adults (Saxton et al., 2021). In our pilot personalized normative feedback (PNF) intervention with veterans, small but significant effects were observed one month after the intervention across all pre-specified drinking outcomes: past 30-day drinks per week (d = 0.47), average number of drinks per occasion (d = 0.17), binge drinking days (defined at 5 drinks in a row for males and 4 drinks in a row for females; d = 0.18), and alcohol-related consequences (d = 0.17) compared to control participants (Pedersen et al., 2017). Although promising, our pilot study only assessed a single month of intervention effects. Brief online approaches such as PNF are meaningful because they can reach large audiences and promote at least short-term changes to one’s drinking (Dedert et al., 2014), while lasting change may require more intensive methods. PNF can also help veterans initiate care. In our pilot study (Pedersen et al., 2016c), nearly twice as many veterans in the PNF group as in the control group reported increased motivation to seek care. Thus, while brief interventions can encourage reflection and consideration of treatment, both the motivation to do so along with early reductions in drinking may not be maintained without additional strategies to help promote treatment engagement.

The Present Study

Building upon the pilot trial (Pedersen et al., 2016a; 2017) and to maximize the impact of this single-session online intervention, we enhanced the PNF in several important ways. We detail the enhanced intervention, which we named BRAVE, in prior work (see Pedersen et al., 2024) and summarize the enhancements here. In brief, we incorporated nonjudgmental motivational interviewing (MI) language that emphasized autonomy, and participants explored their reasons for seeking care as a means of helping to build self-efficacy (Miller & Rollnick, 2023). We also incorporated principles and exercises from Acceptance and Commitment Therapy (Hayes et al., 2011), which encourages individuals to move forward with behavioral changes in accordance with their values. Participants completed a values-clarification exercise and set behavioral goals, which supported motivation to seek formal treatment to maintain drinking behavior change and manage co-occurring mental health problems. In the current study, we describe the longer-term effects (i.e., 3-, 6-, 9-, and 12-month outcomes) of a randomized controlled trial of the brief, online intervention on drinking and treatment (i.e., motivation and preparation for; actual initiation) outcomes. We hypothesized that, compared to those who received a control condition with veteran-specific resources, BRAVE participants would drink less frequently and consume fewer drinks per occasion, experience fewer alcohol-related consequences, engage in more treatment, and report greater motivation to seek care in the follow-up months after the intervention.

Methods

Procedures and Participants

BRAVE was designed for a young adult, post-9/11, veteran cohort who drink at hazardous levels and report co-occurring mental health problems, with the goal of preventing the escalation of symptoms into chronic, lifelong problems. Eligibility criteria therefore included (1) being a United States veteran who served during or after the September 11^th attacks and who have been discharged or separated from the Air Force, Army, Navy, Marine Corps, or Coast Guard, (2) aged 18 to 40 years, (3) a score of 4 or greater (men) or 3 or greater (women) on the 3-item Consumption subscale of the Alcohol Use Disorders Identification Test (AUDIT-C [Bush et al., 1998]), which represents cutoff scores to identify those who may benefit from interventions to reduce alcohol misuse, and (4) not have any upcoming appointments at the VA or elsewhere to receive treatment for a behavioral health concern (such as depression, PTSD, or AUD). This latter criterion helped to focus efforts on a group not actively receiving behavioral health care and to understand the impact on care receipt. Participants were permitted to be a member of the reserves or guard units provided they were not on active duty at the time of enrollment. Although we initially intended to recruit only those veterans with no appointments in the past six months for behavioral health treatment at a VA or elsewhere, in an effort to reach participants outside of traditional care settings (i.e., those who may need care but who are not receiving it), we modified this eligibility criterion to include those who had received care in the past month but who had no future treatment plans for treatment (i.e., they reported no appointments scheduled for treatment in the next month) and were still drinking at levels that could benefit from additional treatment. The sample was generally not recently active in care; only 15.4% of the full sample reported treatment of any behavioral health care appointments in the past year, and just 6.8% reported care in the past 3 months.

Participants were recruited for a study on “veteran mental health” via social media ads facilitated by BuildClinical, a technology company specializing in targeted social media and online ads to populations meeting eligibility criteria for studies. The recruitment period lasted from March 2023 through January 2024, with follow-up assessments lasting into January 2025 (i.e., 12 months past the final baseline date). Figure 1 provides a CONSORT diagram for recruitment, reasons for ineligibility, and follow-up rates. An initial 6,479 potential participants clicked on online ads and completed a screening questionnaire in which they completed items related to eligibility and demographic information. Participants who were not screened out by the questionnaire were contacted via phone by a research assistant who confirmed their eligibility and used a series of additional validation checks to limit fraudulent participants (e.g., ensuring phone responses matched survey responses to items such as military branch and rank). As with many other online studies, this study encountered attempts at fraudulent participation by individuals and artificial intelligence bots posing as eligible veterans to receive monetary incentives (Glazer et al. 2021; Pozzar et al., 2020). These validation checks therefore helped to limit enrollment of these fraudulent responders and provided confidence in the final enrolled veterans in the study.

Figure 1.

RCT Flow.

A total of 752 participants who met eligibility criteria and passed verification checks were randomized to either receive BRAVE or a control condition, the latter of which consisted of an interactive website with veteran-focused information and resources about treatment for alcohol use and mental health. Randomization was conducted within the survey distribution program, Qualtrics, immediately after baseline completion. Participants completed 20-minute follow-up online surveys at 3-, 6-, 9-, and 12-month post-baseline. As part of our verification protocol, we also monitored survey responses during follow-up and removed four participants due to responses that were inconsistent with those reported at baseline (e.g., basic demographic information changed) resulting in a final sample of 748 veterans whose data were included in analyses. Participants received a $20 gift card redeemable at multiple online retailers for each completed survey. After each survey, participants received an email reminding them that they could visit their respective condition website (BRAVE or control) to review the content at any time. The study was approved by the local Institutional Review Board and registered at ClinicalTrials.gov (NCT04244461).

Intervention Conditions

BRAVE intervention.

Detailed procedures for developing the BRAVE intervention, with an in-depth description of intervention components, are documented in a published protocol paper (Pedersen et al., 2024). A copy of the intervention content and manual is available upon request to the first author. Briefly, BRAVE is a multicomponent intervention designed for use on mobile phones, tablets, or computers with Internet capabilities, that targets young adult veterans. It builds upon our prior PNF intervention work with veterans (Pedersen et al., 2016a; Pedersen et al., 2017) by presenting data comparing their own drinking behaviors and perceived peer drinking with real veteran-specific drinking norms, helping them recognize that heavy drinking is less common than they may have estimated. In BRAVE, the PNF is integrated into a 20 to 30-minute interactive, multicomponent program led by narrators who are also young adult veterans. BRAVE begins with a values clarification exercise, in which veterans identify personal values (e.g., honor, courage, empathy) to guide reflection on meaningful life choices, including health behaviors. Content then addresses barriers to seeking care by challenging misconceptions about treatment, offering practical information, and providing personalized resources for accessing services. Throughout the program, veterans are encouraged to consider how seeking behavioral health care and/or reducing their alcohol use aligns with their values. They are guided to set immediate, short-term, and long-term goals related to personal change and that reflect their values. The program concludes with a personalized summary, resources tailored to participants’ location of residence (e.g., nearest VA and twelve-step meeting), and encouragement to revisit the intervention content throughout the duration of the study. This integrated approach combines cognitive-behavioral (Beck, 2020; Marlatt & Donovan, 2007), motivational interviewing (Miller & Rollnick, 2012; Miller & Sovereign, 1989), and values-based strategies (Hayes et al., 2013; Hayes et al., 2011; Miller et al., 2001) to promote healthier behaviors and engagement in care. Participants were asked to complete the intervention during a single session, and they were given periodic reminders throughout the study period regarding their continued access to intervention content and resources.

Control condition.

On the interactive control website, veterans could find care centers, peer support groups, and veterans service organizations. Participants were asked to explore the content on the website for about 20 to 30 minutes during their initial visit, to match the length of the BRAVE intervention. They were also reminded over the follow-up period that they could return to access the website at any time.

Measures

Demographics and military characteristics.

Participants responded to several demographic items assessing age, sex at birth, gender, race, ethnicity, income level, employment status, military branch at discharge, number of deployments, and combat exposure while in the military (Schell & Marshall, 2008).

Primary outcomes.

We prespecified four primary past 30-day drinking outcomes (frequency: number of days on which any alcohol was consumed, quantity: average number of drinks per drinking occasion, binge drinking frequency: number of days on which 4 (or 5) drinks were consumed, and negative alcohol-related consequences) and one primary treatment outcome (any use of behavioral health treatment in the past three months). For binge drinking frequency, participants indicated how many days they drank 4 or 5 drinks or more (for females and males, respectively). Participants completed the Short Inventory of Problems (Kiluk et al., 2013), which is modified from the Drinker Inventory of Consequences (Miller, 1995), to assess negative consequences experienced in the past 30 days. The measure asked participants to indicate how often they experienced each of 15 consequences from 0 (“never,” “not at all,” or “no” depending on item) to 3 (“daily or almost daily,” “very much,” or “yes, more than once” depending on item). The scale was reliable in the current study (α = 0.93 to 0.95 across waves). A total sum score was used in analyses, with a possible range of 0 to 45.

The treatment outcome was measured by any use of behavioral health services in the time period after the intervention (i.e., post-baseline). Participants were asked if they visited a provider at a VA or non-VA care setting for therapy, counseling, support groups, hospitalization, or medications for any concerns related to alcohol or substance use (e.g., if they sought help to drink less or attended groups to stay abstinent from cannabis or other drugs) or mental health concerns (e.g., to help with stress, anxiety, depression, nightmares) since their last survey. Providers were defined for participants as a mental health or behavioral health provider (e.g., psychiatrist, psychologist, social worker, mental/behavioral health nurse, or other provider), a general medical provider (e.g., primary care doctor, physician assistant, nurse practitioner), or an addiction specialist (e.g., addiction counselor). Care was specified as either in-person or via telehealth.

Secondary outcomes.

Because care receipt can be a lengthy process, from making the decision to pursue care to actually engaging in care, we assessed changes in readiness to seek care in the future for alcohol use and for mental health problems using measures based on principles of MI. Two self-report rulers included in the surveys assessed participants’ readiness to seek care within the next 30 days using a scale from 0 (not ready at all) to 10 (absolutely ready). One measured readiness to engage in alcohol-related care and the other measured readiness to engage in mental health-related care. Lastly, at each survey wave, we asked participants if they had made any preparations to receive care over the past three months. These behaviors included looking at options online; discussing with a partner, family member, or friend the options available; talking with a partner, family member, or friend about their own experiences with treatment; contacting a provider to learn more about the options; contacting the insurance provider to see what options are available; and contacting a provider to make an appointment. Participants indicated at each wave whether (no = 0, yes = 1) they engaged in each of these behaviors in the past three months. For analyses, we used a 0 or 1 indicator to indicate whether they did at least one of the behaviors at that wave.

Behavioral health symptoms.

To describe the sample at baseline, we assessed for the most prevalent mental health disorders among post-9/11 veterans using validated screening measures and established cutoff scores indicating likelihood of meeting criteria for the respective clinical disorder. Participants completed the 8-item Patient Health Questionnaire (PHQ-8; Kroenke et al., 2009) to assess symptoms of depression in the past two weeks. Sum scores on the PHQ-8 range from 0 to 24; reliability for the PHQ-8 in this sample was adequate (α = 0.90). We used a score of 10 as a cutoff for a positive depression screen (Kroenke et al., 2009). Participants also completed the Generalized Anxiety Disorder – 7 item scale (GAD-7; Spitzer et al., 2006; Rutter et al., 2017) to assess for symptoms of anxiety. We used a cutoff score of 10 as a positive screen for an anxiety disorder (α = 0.90; Spitzer et al., 2006). Lastly, participants completed the 20-item PTSD Checklist (PCL-5) to assess for the symptoms of PTSD (Blevins et al., 2015; Bovin et al., 2016). A score of 33 indicates a positive screen for PTSD in veteran samples, which we used as a cutoff to describe PTSD prevalence among the sample (α = 0.96; Blevins et al., 2016). Participants also completed the AUDIT-C (Bush et al., 1998) at screening to assess for eligibility, but they completed the full 10-item AUDIT at baseline to determine baseline severity of AUD symptoms (Saunders et al., 1993). Cutoff scores of 16 indicated probable AUD, while scores of 8 indicated hazardous drinking (Babor et al., 2001; Wood et al., 2024).

Analytic Plan

To assess differences in continuous primary drinking outcomes (drinking days, average drinks per occasion, binge drinking days, alcohol-related consequences) and secondary outcomes (readiness for alcohol or mental health care), latent growth models were estimated in Mplus version 8.10 (Grimm et al., 2016; Muthen & Muthén, 2017). To assess between-group differences in the intercept and growth factors, we regressed each on a dummy variable representing the intervention (BRAVE = 1) and control (control = 0) groups. Of note, for all count variables, an alpha parameter (dispersion parameter) was estimated to determine if a negative binomial or Poisson distribution best fit the data. To assess differences in the primary treatment outcome (i.e., any use of alcohol or behavioral health treatment in the past three months) and the secondary preparatory behaviors for treatment outcome, we used basic logistic regression at 3-, 6-, 9-, and 12-months. To understand practical significance, we calculated standardized mean differences (Cohen’s d) at all four follow-up waves, which provided descriptive indicators of effect size for both between-group and within-group differences. For the secondary outcomes including readiness to change drinking and readiness to seek care for alcohol use, we followed the same procedures for latent growth modeling described above.

Data analyses followed the plan according to our published paper detailing the prespecified analytic plan and outcomes (Pedersen et al., 2024). All models were fitted using the full information maximum likelihood estimator available in Mplus (Muthen & Muthén, 2017), treating all observed predictors as single-item latent variables. As such, all individuals contribute all their available data (including those with missing time-invariant X variables). In so doing, we invoked the assumption that missing data were conditionally random, after adjusting for the other variables included in the likelihood function (i.e., missing at random, see below for attrition analysis). The plausibility of this assumption was strengthened by our inclusion of variables that are likely to serve as proxies for unobserved missingness mechanisms. Our final analytic sample was 748 participants. To control the family-wise error rate across the five primary outcomes, we applied a Bonferroni-adjusted significance threshold of p < .01 across the five prespecified primary outcomes (drinking frequency, drinking quantity, binge drinking frequency, alcohol-related consequences, and treatment receipt). The two secondary outcomes (readiness to seek alcohol-related care and readiness to seek mental health-related care) were evaluated at the conventional p < .05. Data are available by request from the first author, as is a manual detailing the specific components of the BRAVE intervention. A data use agreement will be available.

Power analysis.

We designed the study to achieve 80% power to detect small treatment effects between the groups; specifically, d = 0.22 at 3- and 6-month follow-ups and d = 0.23 at 9- and 12-month follow ups, with standard p < .05 two-tailed testing. For reference, our observed intervention effect sizes at 1-month post-intervention in the pilot study were d = 0.17 for average drinks, d = 0.18 for binge drinking days per month, and d = 0.19 for consequences (Pedersen et al., 2017). We expected effect sizes to be small, albeit slightly larger than those in our prior work, given the enhancements to the intervention and that we included a sample screening positive for hazardous alcohol use that could benefit from treatment. Prior work had targeted general college students and young veterans drinking at more moderate levels. As noted, given that this intervention is tailored towards those who may never have received care otherwise, small effects seen in brief interventions such as this can be clinically significant.

Attrition analysis.

Among all participants, 8.6% did not have any follow-up data after the initial baseline; 85% completed the 3-month follow-up survey, 82% completed the 6-month follow-up survey, 80% completed the 9-month follow-up survey, and 80% completed the final 12-month follow-up assessment. To assess potential differences between individuals lost to follow-up and those who completed at least one follow-up assessment, attrition analyses were conducted on primary and secondary outcomes. A single difference emerged: those who did not provide any follow-up data reported a greater number of past 30-day average drinks per drinking day (lost to follow up = 5.7 drinks; at least one follow-up = 4.4 drinks; t = −2.89 (742), p = .002).

Results

Participant Description.

Table 1 provides detailed sample characteristics. Participants were, on average, 33 years old, mostly male (92.1%) and non-Hispanic White (80.7%). Most (65.1%) had experienced some combat during military service, with 75.8% reporting deployment(s). Participants reported a mean of 6.40 (SD = 3.61) years on active duty service prior to discharge. Participants lived in all but four of the U.S. states (Louisiana, Mississippi, Rhode Island, Vermont), with approximately half from California, Texas, New York, Florida, and Illinois. At baseline, the average AUDIT score was 12.77 (SD = 7.07), with 31% (n = 235) meeting the cutoff (score of 16 or higher) for probable AUD and 72.6% (n = 542) meeting the cutoff (score of 8 or higher) for hazardous drinking. Veterans also reported drinking alcohol on 14.94 (SD = 9.11) days in the past month and reported 6.31 (SD = 7.93) binge drinking episodes in the past month. Participants reported substantial symptoms of mental health disorders, with 38.8% (n = 290) meeting screening criteria for at least one disorder: depression (31.4% of the sample), generalized anxiety (27.1%), or PTSD (21.4%). Regarding completion of their assigned condition, 90% of participants assigned to BRAVE completed the full program, with an additional 2% completing at least half. In the control condition, 88% opened the assigned website.

Table 1.

Sample Characteristics

	Full Sample N = 748	BRAVE N = 376	Control N = 372
	Mean (SD) or N (%)	Mean (SD) or N (%)	Mean (SD) or N (%)
Age	33.33 (4.99)	33.46 (5.07)	33.20 (4.91)
Sex
Male	689 (92.1%)	348 (92.6%)	341 (91.7%)
Female	58 (7.8%)	27 (7.2%)	31 (8.3%)
Other	1 (0.1%)	1 (0.2%)	--
Race / Ethnicity
White	607 (81.1%)	306 (81.4%)	301 (80.9%)
Black/African American	38 (5.1%)	17 (4.5%)	21 (5.6%)
Hispanic/Latino(a)	34 (4.5%)	18 (4.8%)	16 (4.3%)
Asian	32 (4.3%)	15 (4.0%)	17 (4.6%)
American Indian/Alaskan Native	23 (3.1%)	12 (3.2%)	11 (3.0%)
Native Hawaiian/Pacific Islander	3 (0.4%)	1 (0.3%)	2 (0.5%)
Multiple Race or Other	11 (1.5%)	7 (1.9%)	4 (1.1%)
Employment
Full time	77 (10.3%)	40 (10.6%)	37 (9.9%)
Part time	544 (72.7%)	276 (73.4%)	268 (72.0%)
Unemployed	67 (9.0%)	31 (8.2%)	36 (9.7%)
Retired	10 (1.3%)	5 (1.3%)	5 (1.3%)
Full time student	50 (6.7%)	24 (6.4%)	26 (7.0%)
Branch of Military
Army	319 (42.6%)	142 (37.8%)	177 (47.6%)
Navy	127 (17.0%)	60 (16.0%)	67 (18.0%)
Marine Corps	232 (31.0%)	132 (35.1%)	100 (26.9%)
Air Force	65 (8.7%)	38 (10.1%)	27 (7.3%)
Coast Guard	5 (0.7%)	4 (1.1%)	1 (0.3%)
Any combat exposure	487 (65.1%)	241 (64.1%)	246 (66.1%)
Number of deployments	1.52 (1.55)	1.55 (1.64)	1.49 (1.45)
Years of military service	6.40 (3.61)	6.34 (3.72)	6.46 (3.50)
Alcohol use
AUDIT	12.77 (7.08)	12.86 (7.12)	12.68 (7.04)
Past 30-day alcohol frequency	14.94 (9.11)	14.67 (8.93)	15.21 (9.31)
Past 30-day alcohol quantity	4.55 (3.38)	4.64 (3.40)	4.46 (3.37)
Past 30-day binge frequency	6.31 (7.93)	5.94 (7.52)	6.68 (8.32)
Past 30-day drinking consequences	8.83 (8.47)	8.77 (8.44)	8.90 (8.52)

Primary Drinking Outcomes.

For all primary alcohol use outcomes, dispersion parameters indicated a negative binomial distribution fit the data best. Final model estimates can be found in Table 2. A significant effect emerged for a reduction in drinking consequences among BRAVE participants (B = −0.07, p = 0.01). There were no significant differences in past-30-day alcohol use frequency, quantity, or binge drinking frequency.

Table 2.

Parameter estimates for primary drinking (past 30 day) and secondary outcomes (next 30 days).

	Alcohol frequency	Alcohol quantity	Binge drinking frequency	Drinking consequences	Readiness to seek alcohol care	Readiness to seek mental health care
Model parameters
Intercept	1.75 (0.21)	1.43 (0.19)	0.77 (0.40)	2.23 (0.33)	3.54 (0.73)	3.02 (0.91)
Slope	−0.32 (0.07)	−0.11 (0.07)	−0.36 (0.11)	−0.46 (0.12)	0.42 (0.23)	0.28 (0.09)
Effect of BRAVE
Intercept	−0.06 (0.05)	0.001 (0.05)	−0.11 (0.10)	0.04 (0.08)	−0.003 (0.18)	0.20 (0.22)
Slope	−0.01 (0.02)	−0.02 (0.02)	0.004 (0.03)	−0.07 (0.03)	0.04 (0.06)	0.02 (0.07)
Residual variance
Intercept	0.40 (0.21)	0.24 (0.02)	1.38 (0.11)	1.01 (0.08)	3.91 (0.35)	5.92 (0.52)
Slope	0.03 (0.003)	(0.01) (0.002)	0.04 (0.01)	0.08 (0.01)	0.10 (0.04)	0.35 (0.06)

Primary Treatment Outcome.

Figure 2 depicts the percentage of participants by treatment group that reported receipt of past three-month treatment at each follow-up. BRAVE participants generally reported higher rates of treatment at each time point, apart from the 9-month follow-up. Controlling for age, sex, deployments, race, ethnicity, and receipt of care at baseline, there were no significant differences in treatment receipt between BRAVE and control participants at any time point. Although no significant differences emerged for treatment receipt, we provide odds ratios for effect size estimation (see Figure 2 note).

Figure 2. Treatment Receipt at Baseline and Follow-up for all Participants.

Note: Effect sizes for BRAVE compared to control: 3-month follow-up OR = 1.21 (95% CI [0.65, 2.23]); 6-month follow-up, OR = 1.22 (95% CI [0.78 1.89]); 9-month follow-up OR = 0.86 (95% CI [0.55 1.35]); and 12-month follow-up OR = 1.03 (95% CI [0.67 1.57]).

Secondary alcohol and treatment outcomes.

Secondary alcohol outcome results can be found in Table 2 (readiness to seek care for alcohol use and readiness to seek care for mental health concerns). There were no significant effects of the intervention on readiness to seek alcohol care or readiness to seek mental health care over the 12-month study period. Figure 3 displays the percentage of participants by intervention and control group that reported any preparatory behaviors at the four follow-up time points. When exploring our secondary treatment outcome, preparatory behaviors for treatment, no significant differences emerged in preparatory behaviors at any time point. We provide odds ratios for effect size estimation in Figure 3 (see note).

Figure 3. Treatment Preparatory Behaviors at Baseline and Follow-up for all Participants.

Note: Effect sizes for BRAVE compared to control: 3-month follow-up OR = 1.22 (95% CI [0.83, 1.78]); 6-month follow-up OR = 0.94 (95% CI [0.68, 1.29]); 9-month follow-up OR = 0.87 (95% CI [0.62 1.22]); and 12-month follow-up OR = 0.87 (95% CI [0.63 1.20]).

Effect size estimates and post-hoc analyses

Table 3 provides effect size estimates (Cohen’s d) at each time point for the continuous outcomes. For primary alcohol outcomes, effect sizes generally favored the intervention condition but were all very small in magnitude (range −0.02, −0.17). For the secondary treatment outcomes, we again observed very small effects sizes favoring BRAVE at all but one time point for readiness to seek alcohol use care.

Table 3.

Means, standard deviations, and effects sizes (within, between) for BRAVE and control participants

		Baseline			3 months				6 months
		Mean	SD	Cohen's d (Between)	Mean	SD	Cohen's d (Within)	Cohen's d (Between)	Mean	SD	Cohen's d (Within)	Cohen's d (Between)
Primary alcohol outcomes (past 30 days)
Alcohol frequency	BRAVE	14.67	8.93	−0.06	12.00	8.77	−0.30	−0.15	11.33	9.04	−0.08	−0.14
	Control	15.20	9.30		13.35	9.28	−0.20		12.60	9.29	−0.08
Alcohol quantity	BRAVE	4.64	3.40	0.05	3.70	2.93	−0.30	−0.15	3.71	3.18	0.00	−0.09
	Control	4.46	3.37		4.20	3.61	−0.07		4.02	3.62	−0.05
Binge drinking frequency	BRAVE	5.94	7.52	−0.09	4.39	5.83	−0.23	−0.17	4.40	6.57	0.00	−0.09
	Control	6.68	8.32		5.50	7.33	−0.15		5.01	7.31	−0.07
Drinking consequences	BRAVE	8.77	8.44	−0.02	6.63	7.55	−0.27	−0.12	6.02	7.67	−0.08	−0.09
	Control	8.90	8.52		7.59	8.74	−0.15		6.72	8.45	−0.10
Secondary outcomes (next 30 days)
Readiness to seek alcohol use care	BRAVE	1.71	2.63	−0.07	2.06	2.98	0.12	0.17	2.06	3.10	0.00	0.11
	Control	1.90	2.69		1.60	2.49	−0.12		1.75	2.73	0.06
Readiness to seek MH care	BRAVE	3.37	3.25	0.05	3.60	3.64	0.07	0.08	3.94	3.75	0.09	0.14
	Control	3.20	3.20		3.31	3.39	0.03		3.42	3.45	0.03

		9 months				12 months
		Mean	SD	Cohen's d (Within)	Cohen's d (Between)	Mean	SD	Cohen's d (Within)	Cohen's d (Between)
Primary outcomes (past 30 days)
Alcohol frequency	BRAVE	11.19	9.20	−0.02	−0.04	10.86	8.97	−0.04	−0.06
	Control	11.54	9.15	−0.11		11.42	9.37	−0.01
Alcohol quantity	BRAVE	3.47	2.91	−0.08	−0.04	3.56	2.96	0.03	−0.11
	Control	3.58	3.29	−0.13		3.96	4.24	0.10
Binge drinking frequency	BRAVE	4.09	6.60	−0.05	−0.04	3.98	5.96	−0.02	−0.07
	Control	4.32	6.53	−0.10		4.42	6.94	0.01
Drinking consequences	BRAVE	5.43	7.82	−0.08	−0.07	5.39	7.69	−0.01	−0.10
	Control	5.98	8.31	−0.09		6.24	8.68	0.03
Secondary outcomes (next 30 days)
Readiness to seek alcohol use care	BRAVE	1.81	3.07	−0.08	−0.04	2.40	3.54	0.18	0.07
	Control	1.94	3.07			2.16	3.22	0.07
Readiness to seek MH care	BRAVE	3.42	3.67	−0.14	0.01	3.89	3.97	0.12	0.09
	Control	3.40	3.63	−0.01		3.55	3.63	0.04

Although not prespecified, we conducted a post-hoc analysis for alcohol consequences on the Short Inventory of Problems by exploring treatment effects across five subscales of: physical (e.g., “my physical health has been harmed by my drinking”), interpersonal (e.g., “my family has been hurt by my drinking”), intrapersonal (e.g., “my drinking has gotten in the way of my growth as a person”), impulse control (e.g., “I have had an accident while drinking or intoxicated”), and social responsibility (e.g., “I have spent too much or lost a lot of money because of my drinking”). Results indicated significant reductions in both physical consequences (B = −0.06, SE = 0.03, p = 0.04) and interpersonal consequences (B = −0.10, SE = 0.05, p = 0.04). No other significant effects emerged for changes in intrapersonal consequences, impulse control, or social consequences.

To address whether participants in each condition made significant changes over time, we re-estimated the latent growth models separately for BRAVE and control participants (see Table 4). Across all four primary drinking outcomes, both BRAVE and control participants showed significant within-group declines over the 12-month follow-up. Reductions in alcohol use frequency, quantity, binge drinking frequency, and alcohol-related consequences were statistically significant in both conditions, with slope estimates ranging from −0.20 to −0.85 per 3-month interval. For secondary outcomes, readiness to seek alcohol use care increased significantly over time in both BRAVE (B = 0.12, SE = 0.04, p = .004) and control (B = 0.08, SE = 0.04, p = .038) participants. Readiness to seek mental health care trended upward in both groups but did not reach significance (BRAVE: B = 0.10, SE = 0.05, p = .077; Control: B = 0.08, SE = 0.05, p = .112).

Table 4.

Within-Group Latent Growth Model Results: Intercept and Slope Estimates by Condition

Outcome	Group	Intercept (Baseline Level)		Slope (Rate of Change)
		Mean	SE	B	SE	p
Primary Drinking Outcomes (past 30 days)
Alcohol frequency	Control	14.72	0.46	−0.836	0.107	<.001
	BRAVE	13.93	0.44	−0.848	0.104	<.001
Alcohol quantity	Control	4.43	0.17	−0.198	0.049	<.001
	BRAVE	4.38	0.16	−0.258	0.042	<.001
Binge drinking frequency	Control	6.31	0.40	−0.474	0.089	<.001
	BRAVE	5.29	0.35	−0.353	0.083	<.001
Drinking consequences	Control	8.59	0.43	−0.630	0.096	<.001
	BRAVE	7.98	0.41	−0.752	0.104	<.001
Secondary Outcomes (next 30 days)
Readiness to seek alcohol care	Control	1.75	0.13	0.084	0.040	.038
	BRAVE	1.75	0.13	0.115	0.040	.004
Readiness to seek MH care	Control	3.23	0.16	0.082	0.052	.112
	BRAVE	3.41	0.16	0.095	0.054	.077

Note. Models estimated using maximum likelihood with full information maximum likelihood for missing data. Intercept mean reflects estimated baseline level; slope mean (B) reflects estimated linear rate of change per 3-month interval across the 12-month follow-up. Both conditions show significant within-group declines across all primary drinking outcomes. Readiness to seek alcohol care increased significantly in both groups; readiness to seek mental health care trended upward but did not reach significance in either group. Bold values indicate p < .05

In addition, because a substantial number of participants screened positive for mental health disorders, we descriptively examined the percentage of participants screening positive for each of the mental health disorders (i.e., AUD, depression, anxiety, PTSD) who reported any receipt of treatment in the past three points at any time during the follow-up period. As seen in Figure 4, treatment receipt rates were quite similar between BRAVE and control participants, with slightly fewer BRAVE participants within each of the diagnostic groups reporting treatment receipt. However, there was a notable difference between BRAVE and control participants who screened positive for GAD, with 15% more participants in BRAVE than control reporting receipt of treatment at some point during follow-up.

Figure 4.

Treatment Receipt at Any Point during Follow-up for Participants Screening Positive for Mental Health Disorders

Discussion

This RCT was designed to test the efficacy of a single-session, brief mobile phone-based intervention with young adult veterans. Importantly, both the BRAVE intervention and control participants demonstrated significant within-person reductions across all four primary drinking outcomes over the 12-month follow-up, suggesting that engagement with either condition, an active intervention or distribution of accessible resources, was associated with meaningful improvements in drinking behavior. Through the provision of BRAVE, this multicomponent, interactive, self-delivered intervention featuring evidence-based components from cognitive behavioral therapy, values-based strategies, and motivational interviewing (Beck, 2020; Hayes et al., 2013; Hayes et al., 2011; Marlatt & Donovan, 2007; Miller & Rollnick, 2012; Miller & Sovereign, 1989), combined with practical personalized resources aimed towards reducing veterans’ drinking behavior and promoting receipt of behavioral health care, we found a significant between-group effect on alcohol-related consequences over 12 months, with participants reporting an approximate 39% reduction in consequences at the 12-month follow-up. Post-hoc analyses indicated significant reductions over time favoring BRAVE specifically for physical and interpersonal consequences, such as health or close relationships being harmed because of drinking. These significant reductions in alcohol-related consequences over 12 months are notable, particularly given that BRAVE was a brief, single-session intervention. Brief alcohol interventions generally produce small and often short-lived effects, and when effects are observed, they are more commonly found for alcohol-related consequences than for overall consumption (Doherty et al., 2017; Huh et al, 2015; Mun et al., 2022). It is possible that given the harm-reduction focus of BRAVE and other brief alcohol interventions that affect consequences more directly (Huh et al., 2015), these interventions help individuals change how they drink, not necessarily how much they drink, by focusing on cognitive behavioral skills that can reduce problems like blackouts, injuries, and interpersonal conflict, even if overall drinking frequency does not change much.

A focus of BRAVE was on increasing readiness to seek care and encouraging actual behavioral health treatment receipt. However, treatment receipt rates were similar for veterans in both conditions. As with most of the drinking outcomes, very small, non-significant effects favored BRAVE overall, except for one follow-up 9-months post-intervention. It is unclear why this time point differed from the others, but because recruitment was on a rolling basis, it was likely unrelated to seasonal factors. Post hoc analyses showed that veterans who screened positive at baseline for depression, PTSD, or AUD (i.e., those who would likely benefit from behavioral health treatment) received care at similar rates across conditions. However, among those screening positive for an anxiety disorder, substantially more BRAVE participants reported receiving behavioral health treatment during follow-up than controls (approximately 47% vs. 31%). The mental health-focused content of the intervention, which included clarifying values, setting behavioral goals, and countering stigma, may have resonated more with anxious veterans. Audiovisual segments of the intervention led by veteran peer narrators may have also contributed. Prior research shows that veteran peers can provide significant support for fellow veterans with mental health disorders (Aitken et al., 2024; Turner et al., 2022) and can help increase use of services and mobile mental health interventions (Blonigen et al. 2021; 2023; Hamblen et al., 2019).

As mentioned, small effects such as the ones observed here are common in brief interventions (Doherty et al., 2017; Huh et al., 2015; Mun et al., 2022), but it is uncommon for them to persist over a 12- month span. Still, in this study, the largest effects occurred in the short-term (three- and six-months post-intervention), consistent with prior research. Although most of the BRAVE effects on treatment receipt were non-significant, small impacts on treatment receipt, particularly from such a brief single session intervention, can be very meaningful from a public health perspective. At baseline, just 15% reported any use of behavioral health services in the past year (~7% reported past 3-month use), despite nearly one-third meeting screening criteria for current AUD, depression, anxiety, or PTSD and almost three-quarters drinking at hazardous levels. At follow-up, nearly one-third of participants reported receiving behavioral health care across both BRAVE and control conditions. Although BRAVE was designed to increase treatment motivation, differences in readiness to seek care were small and not statistically significant. Similar rates of care receipt in the control condition suggest that simply reaching veterans and providing accessible resources may function as a low-intensity intervention. It is also possible that repeated symptom monitoring across follow-up assessments increased awareness and consideration of care in both groups. The absence of larger or statistically significant BRAVE effects indicates that additional support may be necessary to move veterans from contemplating care to actively initiating treatment. BRAVE may be better conceptualized as an initial motivational catalyst -- a low-barrier entry point that reduces alcohol-related harm and builds readiness, but perhaps it may be more effective when followed by more sustained programming or recovery support. VetChange, for example, offers a more extended, skills-based web intervention that has demonstrated efficacy in reducing drinking over time (Brief et al., 2011, 2013) and could serve as a logical next step for veterans who engage with BRAVE but need continued support to maintain gains. Future enhancements to BRAVE, such as peer or family involvement, booster sessions, or ongoing coaching, may similarly strengthen long-term impact. Qualitative interviews with BRAVE intervention completers could further clarify which components were most salient and inform future adaptations.

Limitations

This study has limitations. First, it focused on relatively young post-9/11 U.S. veterans (aged 18 to 40), so results may not generalize to older veterans, particularly those less comfortable with technology who may struggle to engage with an intervention like BRAVE. Second, we relied on self-report measures of alcohol use, symptom severity, and treatment utilization. Although behavioral health care was defined in the assessments, some participants may have misinterpreted items or reported on experiences not actually involving behavioral health treatment (e.g., primary care visits with no discussion of mental health). Access to administrative data from care records would have strengthened our analyses by providing details on treatment type, adherence, and quality. Third, because BRAVE is a multicomponent intervention, it remains unclear whether the small observed effects were driven by the program as a whole or by specific components. Dismantling studies could be beneficial in this regard to isolate and identify the active components. Relatedly, although approximately 90% of those assigned to their condition completed it, we did not collect data on whether participants returned to either program after initial viewing. Participants were encouraged to view the program material again and they may have visited BRAVE or the control resources site during a time when they were feeling motivated and ready to initiate care. The proportion of women in the sample (7.8%) is relatively low compared to samples in other online alcohol intervention trials (Riper et al., 2018; Simpson et al., 2022). Thus, we did not have sufficient power to examine sex as a moderator of intervention effects. Research suggests that women veterans face distinct barriers to care and may respond differently to alcohol interventions than men (Livingston et al., 2021), underscoring the need for future trials to oversample women veterans to adequately power sex-stratified or moderation analyses. Finally, we developed BRAVE based on the assumption that all veterans could benefit from some form of behavioral health care or resources, not only those with diagnosed disorders. Although a striking one-third of the participants in our sample met criteria for AUD, depression, PTSD, or anxiety, we did not conduct a more comprehensive assessment of the myriad behavioral health conditions or other challenges that may disproportionately affect veterans, such as relationship problems or life transition difficulties. A more in-depth assessment of participants’ needs could have supported more subgroup analyses.

Conclusion

In this first trial of BRAVE, the intervention showed promise in engaging a hard-to-reach group of younger veterans outside of traditional care settings. Leveraging mobile technology, it integrates multiple intervention strategies to address the needs of post-9/11 veterans with drinking and mental health concerns. Although several aspects of BRAVE appear to have effectively modestly hit their mark, reaching veterans beyond research settings is an essential next step for our team and other intervention developers. Broader dissemination and implementation of evidence-based Internet and mobile interventions are key components to improving behavioral health care access for veterans, including those who may not otherwise seek care, and other underserved populations. Given the ubiquity of mobile technology (Bush & Wheeler, 2015) and the relatively low cost and burden of mobile health tools (Gentili et al., 2022), more resources and research devoted to reaching large numbers of veterans through these advancements is worthwhile.