Pilot Randomized Clinical Trial of an App-Based Early Intervention to Reduce PTSD and Alcohol Use Following Sexual Assault

Abstract

Objective:

Sexual assault is associated with high risk for posttraumatic stress disorder (PTSD), and PTSD often co-occurs with alcohol misuse. Most sexual assault survivors do not access early preventative interventions for such conditions. App-based interventions are a promising means to extend the reach of early interventions and thereby reduce risk of chronic PTSD and alcohol misuse.

Method:

This study was a pilot randomized clinical trial of an app-based early intervention with phone coaching (THRIVE) for survivors of past-10-week sexual assault (NCT#: NCT03703258). Intended active components of the THRIVE app are daily cognitive restructuring, daily activity scheduling, and as-needed relationally-focused exercises, supported by coaching calls. Forty-one adult female survivors of recent sexual assault with elevated PTSD and heavy drinking were randomized to intervention or control (coached symptom-monitoring app). Participants were encouraged to use the app for 21 days and completed self-report symptom assessments at baseline, post-intervention, and 3-month follow-up.

Results:

At 3 month follow-up, the between-group effect size favored intervention for PTSD (d = −0.70), intoxication frequency (d = −0.62), and drinking hours per week (d = −0.39). More participants evidenced reliable change in intervention versus control for PTSD (OR = 2.67) and alcohol problems (OR = 3.05) at 3 months.

Conclusions:

The general direction of effects indicates that THRIVE, coupled with coaching, reduces risk for PTSD and alcohol outcomes beyond coached monitoring. These findings suggest that apps like THRIVE may provide an option for early intervention for sexual assault survivors.

Epidemiological data suggest that 27–44% of women experience sexual assault in their lifetime and 2.2–4.7% of women experience sexual assault each year (Breiding et al., 2014; Smith et al., 2018). These experiences are associated with elevated risk for posttraumatic stress disorder (PTSD) and alcohol misuse (Dworkin, 2020). Unfortunately, these conditions are highly comorbid (Zinzow et al., 2012) and are thought to be interrelated (Hawn et al., 2020). This is concerning in light of evidence that individuals with comorbid PTSD and alcohol use disorder are more difficult to treat (Back et al., 2009) and have greater healthcare utilization (Ouimette et al., 2006) and poorer treatment engagement (Smith et al., 2019) than those with either disorder alone. Thus, there is an urgent public health need to identify interventions to address these harmful and interrelated conditions among sexual assault survivors.

The Development and Maintenance of PTSD and Alcohol Misuse After Sexual Assault

Trauma-related symptoms are common and generally seen as non-pathological in the early aftermath of sexual assault (Foa et al., 2006; Foa & Rothbaum, 1998). These symptoms often resolve naturally, but approximately 41% of survivors still meet criteria for PTSD at 12 months post assault (Dworkin et al., 2021). Several factors may interfere with naturally-occurring recovery processes and contribute to the development of PTSD (Whealin et al., 2008). First, avoidance of trauma-related cues (e.g., thoughts, emotions, activities, situations) through maladaptive coping strategies such as alcohol use can reduce opportunities for survivors to learn that distressing trauma-related cues are safe and tolerable (Pittig et al., 2018), reduce opportunities for positive reinforcement (Blackledge, 2004), and negatively reinforce alcohol use and other forms of further maladaptive coping (Hawn et al., 2020). Second, trauma survivors commonly develop negative appraisals of the cause or implications of the trauma, can which generate negative emotions and lead to alcohol use and other maladaptive coping strategies (Dunmore et al., 1999; Ehlers & Clark, 2000; Whealin et al., 2008).

Early Post-Sexual-Assault Interventions to Prevent PTSD and Alcohol Misuse

Intervening in the first months after sexual assault could reduce incidence of chronic PTSD and alcohol use. This, in turn, could lessen the long-term public-health burden of this form of trauma (Dworkin & Schumacher, 2016). Indeed, early interventions have successfully reduced PTSD and alcohol misuse following trauma (Giummarra et al., 2018). Effective early interventions have involved cognitive-behavioral strategies (e.g., targeting avoidance coping and/or negative appraisals) and have enrolled individuals with elevated symptoms (Guay et al., 2019; Kornør et al., 2008; Roberts et al., 2019). Though promising, the impact of these interventions is reduced by low rates of service-seeking after sexual assault (Tjaden & Thoennes, 2006) for reasons including privacy concerns and shame/embarrassment (Walsh et al., 2010).

mHealth as a Strategy to Increase the Reach of Early Interventions

mHealth (i.e., delivery of interventions via web platforms or smartphone apps) (Lal & Adair, 2014) is a promising strategy to increase access to empirically-supported preventative interventions following trauma. As a result of the pervasiveness of smartphone ownership (Pew, 2021), most individuals could have immediate, private, low-barrier access to an intervention at the time and place that it is needed. Although several mHealth interventions for PTSD and alcohol misuse and have been tested in non-recent trauma survivors (e.g., Brief et al., 2013, see Dworkin et al., 2019 for a review), no mHealth early intervention has been tested for PTSD or alcohol misuse following sexual assault.

Several mHealth early interventions for general or non-sexual-assault trauma have been tested. First, Bounce Back Now is a fully self-guided cognitive-behavioral intervention with modules for PTSD, alcohol use, and other concerns, which reduced PTSD symptoms but not alcohol use in a trial of adolescent natural disaster survivors (Ruggiero et al., 2015). Second, an internet-delivered version of Prolonged Exposure with daily therapist support was associated with significant reductions in PTSD symptom severity in a pilot trial with adult survivors of recent trauma (Bragesjö, 2021). Third, an mHealth cognitive-behavioral early intervention called Trauma TIPS did not reduce PTSD symptom severity in adult injury patients (Mouthaan et al., 2013). Given the higher risk for mental disorders (Dworkin, 2020) and greater stigma (Kennedy & Prock, 2018) associated with sexual assault relative to other forms of trauma, there is a need for efficacious mHealth early interventions for sexual assault survivors.

There are currently no validated mHealth early interventions for PTSD and alcohol misuse among sexual assault survivors. However, two mHealth treatments for sexual assault survivors have been tested. Both were web-based, used cognitive-behavioral strategies, and targeted college women with a history of non-recent sexual assault. First, a 2-week self-guided intervention reduced drinking quantity, heavy episodic drinking, and PTSD at post-treatment relative to self-monitoring control (Stappenbeck et al., 2021). Second, in a trial of a 14-week therapist-facilitated intervention compared to a psychoeducational website, decreases in PTSD were observed but did not differ by condition (Littleton et al., 2016).

The Current Study

THRIVE (Tools for Health and Resilience after Interpersonal Violence Exposure) was designed to address the lack of validated mHealth early interventions for sexual assault. THRIVE involves an app that is intended to be used for 21 days, starting within 10 weeks of sexual assault, supported by brief weekly coaching calls. The intended active components of THRIVE are two daily cognitive-behavioral exercises (cognitive restructuring and activity scheduling) and relationally-focused exercises intended to be used as needed. THRIVE also involves daily self-monitoring of PTSD symptoms and alcohol use and supportive elements (e.g., resources, encouraging messages banner). In this manuscript, we report on efficacy results from a pilot randomized clinical trial of THRIVE among women with elevated PTSD and alcohol misuse who had experienced sexual assault within the prior 10 weeks. We hypothesized that participants in the intervention condition would evidence decreased PTSD symptom severity (primary outcome; H1), alcohol problems (primary outcome; H2a), and alcohol use (secondary outcome; H2b) at post-intervention and 3-month follow-ups than participants in the control condition. We also tested potential mechanisms of the intervention effects (i.e., reductions in drinking to cope, behavioral avoidance, and negative posttraumatic appraisals) on an exploratory basis.

Method

This study was registered on http://www.ClinicalTrials.gov (NCT#: NCT03703258). Procedures received approval from the University of Washington Institutional Review Board.

Participants

The sample size was set primarily for feasibility reasons. An a priori power analysis determined that our target sample size of N = 40 would yield sufficient power to detect a 40% higher count of alcohol problems in the intervention group. Inclusion criteria were: 1) informed consent, 2) self-identification as female, 3) unwanted sexual experience within the last 10 weeks, 4) age ≥18, 5) English fluency, 6) smartphone and internet access, 7) at least 1 alcoholic drink in the past month, 8) either more than 3 drinks on a given day or more than 7 drinks in a given week (reflecting the NIAAA heavy drinking criteria for women; NIAAA, nd) in the past 6 months, and 9) at least 3 symptom clusters endorsed on the PTSD Checklist. Participants were excluded if they endorsed active suicidality or psychosis.

Procedures

Figure 1 summarizes participant flow and retention through various study procedures. Participants were recruited from January 13^th, 2021 to August 24^th, 2021 via mass emails sent to registered students at the University of Washington, flyers, social media ads, and provider referrals. Study staff conducted a brief phone call with interested participants to describe the purpose of the study (i.e., test the THRIVE intervention with and without daily coping exercises) and preliminarily assess eligibility criteria. When participants were preliminarily eligible, the staff member guided participants in installing the app, scheduled the participant for a phone call with a coach, and sent a link to the consent form and baseline survey to confirm eligibility.

Figure 1:

CONSORT Diagram

See Table 1 for a comparison of treatment elements by condition. All eligible participants were encouraged to use the app daily for 21 days and attend weekly ~10–20 minute coaching calls (4 total) with the first author, a PhD-level licensed clinical psychologist. The sole difference between intervention and control was the inclusion of cognitive-behavioral and relationally-focused app modules and discussion of these modules during coaching calls. Topics related to these modules (e.g., scheduling activities, negative appraisals) were actively avoided in control-condition coaching calls. Thus, the study tested the impact of coached intervention modules, over and above self-monitoring, supportive app elements, and weekly coach contact alone.

Table 1:

Intervention Elements by Condition

Component	Condition
	Intervention	Control
Daily app use for 21 days
Intervention modules (daily cognitive restructuring, daily activity scheduling, as-needed relational module)	x
Self-monitoring (i.e., daily survey of PTSD symptoms and alcohol use, graph of PTSD and alcohol use)	x	x
Supportive elements (e.g., encouraging message banner, resource list)	x	x
Weekly coaching calls for 3 weeks (4 total)
Assistance with intervention exercises (~5–10 minutes)	x
Mood/safety monitoring (~1–2 minutes)	x	x
Assistance with app access and technological difficulties (~1–2 minutes)	x	x
Offers of referrals (~1–2 minutes)	x	x

Participants who were eligible at baseline were randomized 1:1 (unstratified) to intervention or control via a computer-generated randomization algorithm. Participants were presented with a message following baseline stating which version of the app they had been selected to receive (i.e., “the version of the app that involves daily surveys and coping exercises” for intervention and “the version of the app that involves a daily survey to check in about how you’re doing” for control). App access was granted automatically after randomization. The app was programmed as a progressive web application (i.e., a website accessed via a home screen bookmark that functioned like a native app). It contained surveys that functioned as interactive activities. App data were stored securely on study servers without identifying information.

On day 21 post-baseline, participants were sent the post-intervention assessment, and on day 84, participants were sent the 3-month follow-up assessment. Participants were paid $20, $40, and $60 in gift cards for completing baseline, post-intervention, and 3-month follow-up, respectively. They were also paid $3 for every day of self-monitoring they completed, with a $10 bonus for completing 7 consecutive days. Total possible payment was $213 for both conditions.

Intervention

THRIVE is a multi-component early intervention for sexual assault survivors delivered via mobile app and brief weekly calls with a coach. THRIVE was developed by the study team. The app was programmed by an external company that was not involved in data analysis.

Activity scheduling module.

The activity scheduling module of THRIVE targets avoidance generally and drinking to cope specifically. Activity scheduling is a cognitive-behavioral strategy that is an element of interventions that are effective for PTSD (e.g., behavioral activation; Gros et al. 2012; Wagner et al. 2019) and substance use (Daughters et al., 2018). Day 1 of THRIVE’s activity scheduling module guides the user in creating a list of valued and/or avoided daily activities, presents brief psychoeducation about the risks of drinking to cope, and offers the option to append strategies to reduce heavy drinking to users’ selected activities. Days 2–21 of the activity scheduling module involves setting daily goals to complete an activity from their list and then marking activities as completed.

Cognitive restructuring module.

The cognitive restructuring module of THRIVE targets negative appraisals (e.g., self-blame). Cognitive restructuring is a cognitive-behavioral strategy that is a common element across evidence-based PTSD treatments (Cusack et al., 2016; Schnyder et al., 2015) and relapse prevention for alcohol use (Larimer et al., 2003). Day 1 of the cognitive restructuring module teaches participants to identify negative posttraumatic appraisals and guides the user in creating a list of their own negative appraisals. Days 2–21 of the cognitive restructuring module involves users selecting a negative appraisal from their list and using a guided cognitive restructuring activity to identify new, more balanced thoughts.

Relationally-focused module.

The THRIVE app also includes a relationally-focused module with five exercises targeting common issues with relationships that can arise following sexual assault: assessing social resources, expressing needs, weighing the costs and benefits of disclosure, coping with negative disclosure reactions, identifying strategies to maintain social connections. Exercises are intended to be completed on an as-needed basis.

Self-monitoring.

Daily self-monitoring of PTSD symptoms and alcohol use are available via a brief daily survey. App users have continual access to a combined graph of their alcohol use and PTSD symptoms, plotted against their personal maximum.

Supportive elements.

THRIVE includes several elements intended to provide a positive and supportive user experience. Encouraging statements from other app users are displayed continuously on a rotating basis at the bottom of the app screen. In addition, resource lists of referral options are continually available within the app.

Coaching calls.

THRIVE involves four weekly 10–20 minute coaching calls to encourage and assist with exercise completion, troubleshoot app access, encourage daily self-monitoring, monitor mood and safety risk, and provide referrals.

Measures

PTSD (primary outcome).

We used the 20-item Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5; Blevins et al., 2015) to assess past-month DSM-5 PTSD symptom severity and provisional diagnostic status. Participants answered items in relation to the focal assault on a scale from 0 (not at all) to 4 (extremely). Sum scores range from 0 to 80. PTSD diagnosis was calculated by treating each item scored as 2 or higher as “endorsed” and applying DSM-5 scoring rules. The PCL-5 had strong internal consistency, test-retest reliability, and convergent and discriminant validity in a sample of primarily female undergraduate students (Blevins et al., 2015). Internal consistencies were α = 0.87 (baseline), α = 0.92 (post), and α = 0.94 (3 months).

Alcohol problems (primary outcome).

We used the 16-item Short Rutgers Alcohol Problem Index (RAPI-S) (Earleywine et al, 2008) to assess past-month alcohol-related problems. Items were rated on a 5-point Likert scale from 0 (never) to 4 (10 or more times). We computed a sum score (range: 0 to 64). In a large sample of college students, the RAPI-S showed strong internal consistency and less gender bias than the full scale (Earleywine et al., 2008). Internal consistencies were α = 0.84 (baseline), α = 0.90 (post), and α = 0.79 (3 months).

Alcohol use (secondary outcome).

We used the 14-item Daily Drinking Questionnaire (DDQ; Collins et al., 1985) to assess past-month typical alcohol use. Participants recorded the number of standard drinks they consumed on each day of a typical week and over how many hours they typically drank alcohol. Responses were used to calculate typical drinks per week, typical drinking hours per week, and typical drinking days per week.

We used two items from the National Institute on Alcohol Abuse and Alcoholism Task Force to assess past-month peak drinks (0 = 0 drinks to 10 = 36 drinks or more) and frequency of binge drinking (0 = never to 7 = every day). We added an additional item to assess intoxication frequency based on a previous trial (Doumas et al., 2011): “During the past month, how often did you get drunk from alcohol?” assessed on a scale from 0 = never to 7 = every day.

Drinking to cope (hypothesized mechanism).

The 3-item drinking to cope subscale of the Drinking Motives Questionnaire – Revised Short Form (Kuntsche & Kuntsche, 2009) assessed how often participants drank to cope with negative affect over the last 30 days. Participants indicated how often (1 = Almost never/Never to 5 = Almost always/Always) they engaged in drinking behavior based on coping motivations. Sum scores range from 3–15, with higher scores indicating more drinking to cope. This subscale has strong internal consistency (α = 0.83) and concurrent validity among undergraduate drinkers (Harbke et al., 2019). Internal consistencies were α = .72 (baseline), α = 0.86 (post), and α = 0.91 (3 months).

Behavioral avoidance (hypothesized mechanism).

We used the 3-item Avoidance subscale of the Behavioral Activation for Depression Scale Short-Form (BADS-SF; Manos et al., 2011) to assess past-month behavioral avoidance. Items were rated on a 7-point Likert scale (0 = Not at all to 6 = Completely). Sum scores range from 0–18; higher scores represent increased behavioral avoidance. The BADS-SF avoidance subscale had good internal consistency in a sample of community members and undergraduates with a history of depression (Manos et al., 2011). Internal consistencies were α = 0.64 (baseline), α = 0.79 (post), and α = 0.73 (month 3).

Negative posttraumatic appraisals (hypothesized mechanism).

The Posttraumatic Cognitions Inventory (PTCI; Foa et al., 1999) assessed negative posttraumatic appraisals via 33 items rated on a 7-point scale (1 = Totally disagree to 7 = Totally agree) in relation to the index sexual assault. Sum scores range from 33 to 231; higher scores indicate more negative appraisals. The PTCI has shown strong internal consistency in a community sample (Foa et al., 1999). Internal consistencies were α = 0.93 (baseline), α = 0.97 (post), and α = 0.97 (month 3).

Analyses

As this study was not powered to detect all effects of interest, we rely primarily on the direction and magnitude of effect sizes to interpret results and conduct significance tests in an exploratory manner. All analyses were conducted with the intent-to-treat (ITT) sample.

Effect sizes.

Within-group Cohen’s d effect sizes represented the mean-level difference in scores from baseline to post-intervention and 3 month follow-up for each condition divided by the standard deviation of the change scores; pairwise deletion was used for missing data in these calculations. Between-group ds represented the difference between the within-group ds. Paired sample t-tests evaluated the significance of within-group mean changes to both follow-ups. We also used the reliable change index to understand whether observed change from baseline to 3-month follow-up was unlikely to be attributable to measurement error or day-to-day symptom fluctuation (Jacobson & Truax, 1991) and report odds ratios for differences in reliable improvement by condition.

Mixed-effects models.

We tested time by condition interactions in exploratory linear mixed-effects models using the R package lmer (Bates et al., 2015) to understand potential differences between conditions. We used likelihood ratio tests to explore distributions for count variables and ultimately selected negative binomial distributions for all count variable models. Maximum likelihood estimation was used to account for missing data.

Results

Forty-one eligible survivors were randomized 1:1 to either intervention or control. All participants completed at least two days of daily monitoring (M = 15.85 of 21 days, SD = 4.97) and at least one coaching call (M = 3.71 of 4 calls, SD = 0.81). In intervention, all but one participant completed an intervention module exercise on at least one day (M = 10.40 of 21 days, SD = 6.52). All participants completed at least one follow-up assessment; one participant did not complete post-intervention and a different participant did not complete 3-month follow-up.

There were no significant differences by condition in demographics or baseline values of outcomes or mechanisms (Table 2). Participants were all adults assigned female sex at birth, most were students, 56.1% identified as a member of a racial/ethnic minority group, and 43.9% identified as a sexual minority. Most assaults involved force (53.7%) and/or penetration (58.5%). Assaults were perpetrated by friends (32.5%), acquaintances (25.0%), romantic partners (17.5%), strangers (12.5%), or someone they were hooking up with (5.0%).

Table 2:

Baseline descriptives

	Full sample (N = 41)	Intervention (n = 20)	Control (n = 21)	Difference by condition
Demographics
Ethnicity: n (%)				X2(1) = 0.67, p = .41
Hispanic/Latina	6 (14.63%)	4 (20.00%)	2 (9.52%)
Not Hispanic/Latina	33 (80.49%)	16 (80.00%)	17 (80.95%)
Race: n (%)				X2(3) = 5.18, p =.16
Black/African American	5 (12.20%)	2 (10.00%)	3 (14.29%)
White/Caucasian	18 (43.90%)	12 (60.00%)	6 (28.57%)
Asian/Asian American	6 (14.63%)	1 (5.00%)	5 (23.81%)
Multiracial	12 (29.27%)	5 (25.00%)	7 (33.33%)
Sexual orientation: n (%)				X2(3) = 4.02, p = .26
Heterosexual/straight	23 (56.10%)	12 (60.00%)	11 (52.38%)
Lesbian	3 (7.32%)	0 (0.00%)	3 (14.29%)
Bisexual	14 (34.15%)	7 (35.00%)	7 (33.33%)
Other	1 (2.44%)	1 (5.00%)	0 (0.00%)
Student status: n (%)				X2 (2) = 2.23, p = .33
Not a student	4 (9.76%)	2 (10.00%)	2 (9.52%)
Part-time	2 (4.87%)	2 (10.00%)	0 (0.00%)
Full-time	35 (85.37%)	16 (80.00%)	19 (90.48%)
Age: M (SD)	20.78 (3.40)	21.45 (4.31)	20.14 (2.15)	t(27.61) = -1.22, p = .23
Number of weeks since SA: M (SD)	3.50 (2.23)	3.36 (2.20)	3.63 (2.31)	t(39) = 0.37, p = .71
Primary outcomes
Provisional PTSD diagnosis: n (%)				X2(1) = 1.38, p = .24
Meets criteria	34 (82.93%)	18 (90.00%)	16 (76.19%)
Does not meet criteria	7 (17.07%)	2 (10.00%)	5 (23.81%)
Posttraumatic stress: M (SD)	43.12 (12.19)	46.60 (11.45)	39.81 (12.22)	t(39) = -1.83 , p = .07
Alcohol problems: M (SD)	11.68 (7.39)	12.40 (8.40)	11.00 (6.43)	t(39) = -0.60, p = .55
Secondary outcomes
Typical drinks per week: M (SD)	10.24 (7.05)	9.35 (7.38)	11.10 (6.79)	t(39) = 0.08, p = .44
Typical drinking hours per week: M (SD)	8.24 (6.87)	8.15 (8.03)	8.33 (5.76)	t(39) = 0.79, p = .93
Typical drinking days per week: M (SD)	2.83 (1.64)	2.90 (1.83)	2.76 (1.48)	t(39) = -0.27, p = .79
Peak drinks: M (SD)	6.22 (3.04)	5.00 (1.52)	5.19 (1.25)	t(39) = 0.44, p = .66
Binge frequency: M (SD)	2.07 (1.27)	1.90 (1.25)	2.24 (1.30)	t(39) = 0.85, p = .40
Intoxication frequency: M(SD)	2.90 (1.32)	3.05 (1.40)	2.76 (1.26)	t(39) = -0.69, p = .49
Hypothesized mechanisms
Drinking to cope: M (SD)	2.52 (1.12)	2.44 (1.05)	2.59 (1.20)	t(39) = 0.41, p = .68
Behavioral avoidance: M (SD)	7.61 (3.94)	6.60 (3.63)	8.57 (4.07)	t(39) = 1.63, p = .11
Negative appraisals: M (SD)	123.07 (30.20)	128.95 (27.28)	117.48 (32.39)	t(39) = -1.22, p = .23

Preliminary Efficacy

Table 3 displays effect size estimates and significance tests for baseline to post-intervention and 3-month follow-up, and Table 4 summarizes the percent of participants experiencing reliable change at 3 months. Table 5 reports results from mixed-effects models.

Table 3:

Descriptive statistics and effect sizes for changes in symptoms

	Intervention		Control		Comparison
	Paired samples t	Within-group d	Paired samples t	Within-group d	Between-group d
Primary Outcomes
Posttraumatic stress
Baseline to post-intervention	t(18) = −2.62, p = .02	−0.60	t(20) = −2.05, p = .05	−0.45	−0.15
Baseline to 3 months	t(19) = −8.15, p = .000	−1.82	t(19) = −5.03, p = .000	−1.12	−0.70
Alcohol problems
Baseline to post-intervention	t(18) = −1.09, p = .29	−0.25	t(20) = −2.79, p = .01	−0.61	0.36
Baseline to 3 months	t(19) = −3.41, p = .003	−0.76	t(19) = −3.39, p = .003	−0.76	−0.01
Secondary Outcomes
Typical drinks per week
Baseline to post-intervention	t(18) = −1.60, p = .13	−0.37	t(20) = −1.01, p = .33	−0.22	−0.15
Baseline to 3 months	t(l9) = −1.84, p = .08	−0.41	t(19) = −1.86, p = .08	−0.41	0.00
Typical drinking hours per week
Baseline to post-intervention	t(18) = −1.09, p = .29	−0.25	t(20) = −0.96, p = .35	−0.21	0.04
Baseline to 3 months	t(19) = −1.30, p = .21	−0.29	t(19) = 0.44, p = .67	0.10	−0.39
Typical drinking days per week
Baseline to post-intervention	t(18) = −1.49, p = .16	−0.34	t(20) = −1.03, p = .32	−0.22	−0.12
Baseline to 3 months	t(19) = −1.24, p = .23	−0.28	t(19) = −0.55, p = .59	−0.12	−0.16
Peak drinks
Baseline to post-intervention	t(18) = −2.84, p = .01	−0.62	t(20) = −2.23, p = .04	−0.41	−0.21
Baseline to 3 months	t(19) = −03.15, p = .005	−0.50	t(19) = −2.07, p = .05	−0.46	−0.05
Binge frequency
Baseline to post-intervention	t(18) = −3.99, p = .001	−0.91	t(20) = −0.36, p = .72	−0.08	−0.84
Baseline to 3 months	t(19) = −1.99, p = .06	−0.44	t(19) = −1.31, p = .21	−0.29	−0.15
Intoxication frequency
Baseline to post-intervention	t(18) = −2.73, p = .01	−0.63	t(20) = −1.86, p = .08	−0.41	−0.22
Baseline to 3 months	t(19) = −3.46, p = .003	−0.77	t(19) = −0.68, p = .51	−0.15	−0.62
Hypothesized Mechanisms
Drinking to cope
Baseline to post-intervention	t(18) = −2.75, p = .01	−0.63	t(20) = −0.20, p = .84	−0.04	−0.59
Baseline to 3 months	t(19) = −4.26, p = .000	−0.95	t(19) = −1.60, p = .13	−0.36	−0.59
Behavioral avoidance
Baseline to post-intervention	t(18) = −2.41, p = .03	−0.55	t(20) = −0.82, p = .42	−0.18	−0.38
Baseline to 3 months	t(19) = −3.99, p = .001	−0.89	t(19) = −4.00, p = .001	−0.89	0.00
Negative appraisals
Baseline to post-intervention	t(18) = −2.81, p = .01	−0.64	t(20) = −1.91, p = .07	−0.42	−0.23
Baseline to 3 months	t(18) = −7.53, p = .000	−1.73	t(19) = −3.07, p = .006	−0.69	−1.04

Note. Grey shading denotes that the magnitude and direction of effects favor intervention over control (i.e., d ≤ −0.20).

Table 4:

Reliable Change to 3 Month Follow-Up

	Reliable Change Benchmark	Reliable Improvement* at 3 Month Follow-Up
		THRIVE	Control	OR	95% CI
Primary Outcomes
Posttraumatic stress	12.19 points	80.0%	60.0%	2.67	0.65 to 10.97
Alcohol problems	8.33 points	35.0%	15.0%	3.05	0.66 to 14.14
Secondary Outcomes
Typical drinks/week	11.70 drinks	15.0%	10.0%	1.59	0.24 to 10.70
Typical drinking hrs/week	9.96 hours	10.0%	10.0%	1.00	0.13 to 7.89
Typical drinking days/week	2.67 days	15.0%	10.0%	1.59	0.24 to 10.70
Hypothesized Mechanisms
Drinking to cope	1.22 points	35.0%	25.0%	1.62	0.41 to 6.34
Behavioral avoidance	3.34 points	40.0%	28.6%	1.67	0.45 to 6.13
Negative appraisals	21.50 points	73.7%	30.0%	6.53	1.61 to 26.47

^*Calculated using the Reliable Change Index (Jacobson & Truax, 1992).

Table 5:

Results from Mixed Effects Models

Model	Intercept B (SE)	Condition (Intervention) B (SE)	Time Main Effects B (SE)		Condition x Time Interactions B (SE)
			Post-intervention	3 months	Intervention condition x Post-intervention	Intervention condition x 3 months
Primary Outcomes
Posttraumatic stress	3.65 (0.09)***	0.17 (0.13)	−0.13 (0.09)	−0.50 (0.10)***	−0.12 (0.13)	−0.28 (0.14)*
Alcohol problems	2.26 (0.18)***	0.13 (0.25)	−0.45 (0.18)*	−0.78 (0.19)***	0.14 (0.26)	−0.02 (0.27)
Secondary Outcomes
Typical drinks/week	2.30 (0.17)***	−0.29 (0.25)	−0.18 (0.17)	−0.41 (0.17)*	−0.05 (0.25)	0.01 (0.25)
Typical drinking hours/week	2.04 (0.22)***	−0.26 (0.32)	−0.24 (0.24)	−0.06 (0.25)	0.06 (0.35)	−0.22 (0.36)
Typical drinking days/week	0.97 (0.15)***	0.03 (0.21)	−0.11 (0.18)	−0.11 (0.18)	−0.08 (0.26)	−0.12 (0.27)
Peak drinks	6.29 (0.68)***	−0.14 (0.97)	−0.86 (0.69)	−1.33 (0.70)	−0.48 (1.00)	−0.57 (1.00)
Binge frequency	2.24 (0.31)***	−0.34 (0.44)	−0.10 (0.28)	−0.41 (0.29)	−0.57 (0.41)	−0.24 (0.41)
Intoxication frequency	2.76 (0.32)***	0.29 (0.46)	−0.52 (0.27)	−0.20 (0.28)	0.01 (0.39)	−0.85 (0.39)*
Hypothesized Mechanisms
Drinking to cope	2.59 (0.23)***	−0.15 (0.32)	−0.05 (0.23)	−0.45 (0.24)	−0.42 (0.34)	−0.43 (0.34)
Behavioral avoidance	9.43 (0.87)***	1.97 (1.25)	−0.67 (1.03)	−3.77 (1.05)***	−2.65 (1.49)	−1.38 (1.49)
Negative appraisals	4.74 (0.07)***	0.11 (0.11)	−0.13 (0.07)	−0.19 (0.07)**	−0.10 (0.10)	−0.36 (0.10)***

^*p < .05,

^**p < .01,

^***p < .001

PTSD (primary outcome).

Within-group effect sizes suggesting symptom decreases in PTSD were observed in both conditions at both follow-ups. These changes were larger in intervention than control at 3 months but not post-intervention, as evidenced by a between-group d of −0.70 and a significant condition × time interaction in the mixed-effect model. More participants experienced a reliable decrease in PTSD symptoms in intervention by 3 months, and no participants evidenced a reliable increase. Provisional PTSD diagnosis decreased from 90% at baseline to 25% at 3 months in intervention compared to a change from 70% at baseline to 46% at 3 months in control (OR = 3.06 for loss of diagnosis).

Alcohol problems (primary outcome).

Within-group effect sizes indicating decreases in alcohol problems were observed in both conditions at both follow-ups. The between-group d favored control at post-intervention but was negligible at 3 months, and no significant condition × time interaction was observed in the mixed-effect model. However, more participants in intervention than control had reliable decrease in alcohol problems at 3 months (OR = 3.05) and no participants evidenced a reliable increase.

Alcohol use (secondary outcome).

In the intervention condition, small to large within-group effect sizes for all alcohol use outcomes were observed at both follow-ups. In control, small within-group effect sizes were observed at post-intervention for most alcohol use outcomes, but small decreases were maintained to 3 months for only typical drinks per week and peak drinks. The difference in magnitude of these changes favored intervention for drinking hours per week (3 months only), peak drinks (post-intervention only) and intoxication frequency (post-intervention and 3 months). A significant time × condition interaction was observed for intoxication frequency at 3 months but not post-intervention in the mixed-effect model.

Hypothesized mechanisms.

At post-intervention, medium-sized within-group effect sizes suggested improvement in all mechanisms in the intervention condition, while the control condition had small improvements in only one mechanism (i.e., negative cognitions). At 3 months, the intervention condition evidenced large within-group improvements on all mechanisms, and the control condition evidenced small to large within-group improvements. Comparisons of between-group effect sizes by condition favored intervention for drinking to cope (post-intervention and 3 months), behavioral avoidance (post-intervention only), and negative appraisals (post-intervention and 3 months). The time × condition interaction for negative appraisals was significant at 3 months only. Reliable change metrics also indicated that significantly more participants improved on negative appraisals in the intervention condition compared to control at 3 months. Somewhat more participants in intervention than control improved on drinking to cope and behavioral avoidance at 3 months, although these odds ratios were not significant.

Discussion

The goal of this study was to conduct a pilot evaluation of THRIVE, an mHealth early intervention intended to reduce risk for PTSD and alcohol misuse after sexual assault. We compared a full version of THRIVE, involving a coached app with daily cognitive-behavioral exercises, as-needed relationally-focused exercises, daily assessments, and supportive elements, to a control condition involving a coached app with daily assessments and supportive elements only. Thus, this comparison reflects the added value of coached daily cognitive-behavioral exercises and as-needed relationally-focused exercises, over and above app use, self-monitoring, and coach contact alone. Consistent with hypotheses, THRIVE outperformed control in reducing PTSD (H1), alcohol outcomes (H2a & H2b), and hypothesized mechanisms at 3 months (more than 2 months after the end of the intervention). These results indicate that THRIVE may be the first mHealth early intervention for recent sexual assault survivors with efficacy in prospectively reducing risk for PTSD and suggest that further testing of THRIVE is warranted.

In addition, THRIVE is the first mHealth intervention for recent trauma survivors with efficacy in reducing alcohol misuse, albeit in a small sample. At 3 months, the change in intoxication frequency and drinking hours per week observed in intervention was larger than that observed in control, and more intervention participants had reliable decreases in alcohol problems. Although a prior mHealth intervention for survivors of lifetime sexual assault reduced alcohol misuse (Stappenbeck et al., 2021), the only other mHealth early intervention, to our knowledge, that targeted alcohol use (i.e., Bounce Back Now) did not demonstrate reductions in binge drinking over time (Ruggiero et al., 2015). It is notable that THRIVE included much less explicit emphasis on alcohol misuse than Bounce Back Now. Both conditions in the current study self-monitored daily drinking, which is an active intervention for alcohol use and may have helped reducing drinking (McCambridge et al., 2011). It is also possible that that targeting avoidance coping broadly rather than alcohol use specifically could have decreased participant resistance to reflecting on alcohol use. However, a fully-powered test of THRIVE is needed to determine what aspects of the intervention were active components in reducing drinking.

A major strength of this study is its control group, which made it possible to understand benefits of the intervention over natural recovery and allowed us to isolate the effects of the in-app exercises and additional coaching support beyond app use and coach contact alone. Also, unlike most other studies on PTSD prevention (Roberts et al., 2019), our sample was mostly comprised of individuals who identified as racial/ethnic minorities, and nearly half of our sample endorsed a sexual minority identity. Individuals from racial/ethnic minority groups and sexual minorities are at higher risk for sexual assault (Coulter et al., 2017) and associated negative outcomes (Sigurvinsdottir & Ullman, 2016), but racial/ethnic minorities are frequently underrepresented in PTSD clinical trials and sexual orientation is rarely reported in trial demographics (Madnick & Spokas, 2022). Despite these strengths, this study also had limitations. First, we relied exclusively on self-reports of symptoms rather than clinical interviews. Second, our use of an all-female, primarily college-student sample limits generalizability to the broader population. Third, this study was underpowered to detect all statistically significant effects or to examine heterogeneity of treatment effects by demographic groups (eg. Race/Ethnicity/Sexual Orientation). Fourth, this study was conducted during a period of active COVID-19 restrictions, which might have affected observed efficacy. Fifth, all coaching calls were conducted by the first author, who was not blind to study condition, and fidelity in coaching calls was not monitored; these factors could have introduced bias. Sixth, payment for daily survey completion might have artificially inflated engagement. Seventh, there were multiple outcome variables and tests; although analyses focused primarily on the direction and magnitude of effects rather than statistical significance, this could have increased the risk of type I error.

These preliminary results suggest that THRIVE may be an effective way to improve outcomes among sexual assault survivors. A larger randomized clinical trial is needed to confirm these effects and test moderators (e.g., race, sexual orientation, assault type, time since assault). If efficacy is confirmed, THRIVE would be a promising, low-barrier strategy to increase the reach of evidence-based practices among this high-need population.

Clinical impact statement:

This pilot study found that a new app for recent sexual assault survivors might be effective in preventing PTSD and alcohol misuse.