Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 1.
Published in final edited form as: Am J Psychiatry. 2025 May 1;182(5):463–472. doi: 10.1176/appi.ajp.20240232

Randomized Controlled Trial of Bounce Back Now, a Mobile Application to Reduce Post-Disaster Symptoms of Posttraumatic Stress, Depressed Mood, and Sleep Disturbance

Kenneth J Ruggiero 1, Arthur Andrews 2, Tatiana M Davidson 1, Yulia Gavrilova 3, Brian E Bunnell 4, Jennifer Dahne 5,6, Matthew Price 7, Zoe M F Brier 5,8, Gregory Cohen 9, Dean Kilpatrick 5, Ron Acierno 10, Sandro Galea 9
PMCID: PMC12047739  NIHMSID: NIHMS2041505  PMID: 40308112

Abstract

Objective:

There is tremendous public health interest in cost-efficient, scalable interventions to improve post-disaster mental health. We examined the efficacy of Bounce Back Now (BBN), a mobile application, vs. enhanced usual care (EUC).

Methods:

A population-based trial was conducted with a diverse sample of 1,357 adults affected by Hurricane Harvey, Irma, Florence, Maria, or Michael in 2017 and 2018. Inclusion criteria were age ≥18 years, access to an internet-accessible device, English speaking, and living in a hurricane-affected area. BBN is designed to address symptoms of posttraumatic stress, depression, and sleep disturbance using evidence-based techniques grounded in behavioral and cognitive principles. Depressive, posttraumatic stress, and sleep symptoms were measured.

Results:

Access of BBN vs. EUC was similar. Active engagement was greater for BBN (t(1,133.56)=5.79,p<0.001,d=0.31), but BBN users engaged more actively in coping-skills activities than more time-intensive elements designed to promote behavior change. Moderate symptom reduction was observed in both conditions; Cohen’s ds for the 3-month post-baseline ranged from 0.49–0.60 in BBN vs. 0.36–0.41 in EUC. A latent change model of a single-factor depression outcome revealed that BBN users had greater reduction in depression, sleep difficulty, and PTSD symptoms than EUC users (p-values<0.05) and these differences were maintained at 6-month and 12-month post-baseline assessment.

Conclusion:

Population impact is driven by reach and effectiveness. The potential reach of BBN is high, which heightens opportunity for population-level impact, but per-user symptom reduction was modest. Per-user impact may be improved by embedding digital health resources in the context of a broader healthcare strategy.

ClinicalTrials.gov ID:

NCT03403738.

Introduction

Disasters can affect the wellbeing of thousands or millions of individuals simultaneously and pose major risks to public health.1–4 Disasters often trigger secondary stressors including threat of death or injury, loss of loved ones, limited access to basic needs (e.g., food, shelter), and financial strain. Resilience and rapid recovery are common trajectories, but depression, sleep disturbance, and posttraumatic stress disorder (PTSD) are prevalent.4–6 Safe, scalable, cost-efficient interventions are needed to address mental health needs and ensure access to best-practice treatment.7,8

Most adults who develop mental health needs do not seek services.9–12 Stigma, time, transportation, scheduling problems, cost, and competing priorities (e.g., financial, occupational, physical health) are barriers.13–16 Moreover, the “desire to handle the problem on my own” is among the most common attitudinal barriers to seeking mental health care.17,18 Digital mental health resources can overcome many of these barriers and should be evaluated to examine potential for population impact.

Digital self-help interventions are understudied in trauma mental health. Early studies yielded effect size estimates comparable to traditional face-to-face services.19 However, studies often used highly protocolized approaches that are impractical at scale, such as remuneration upon completion of an intervention, highly specified intervention dosages, and proactive lab-/clinic-based recruitment with clear expectations guiding use of the intervention. These procedures likely inflated estimates of engagement, completion, and impact, which have not generalized to real-world settings.20 Digital mental health tools delivered in real-world contexts are more likely to be accessed than professional services, but less likely to deliver a sufficient therapeutic dose.21 It is critical to examine digital mental health interventions in the post-disaster context, when they can be delivered at a time of greatest relevance to the needs of individuals and at a time when it is impractical for healthcare systems to meet the overwhelming and sudden demand for services that disasters create.

Bounce Back Now (BBN) is a mobile application designed to address common post-disaster mental health symptoms. Its development was informed by extensive research and testing on web-based platforms that featured a screening tool and modules targeting symptoms of depression, PTSD, and substance use problems.22–23 Each module included educational content and interactive content relating to behavioral principles (e.g., breaking patterns of unhealthy avoidance, increasing behavioral activation, motivational enhancement).23 Studies demonstrated its feasibility with adult disaster survivors22,24 and its efficacy in reducing symptoms of PTSD and depression among disaster-affected adolescents.25 Most participants in earlier trials actively used BBN,26,27 with usage comparable across racial/ethnic and geographically diverse groups.28 Here, we tested BBN in a mobile application (app) format with a diverse sample of hurricane-affected adults. We hypothesized that participants randomized to BBN would experience improved symptom trajectories of depression, sleep disturbance, and PTSD vs. an enhanced usual care app.

Method

Design

A randomized controlled trial of BBN vs. an enhanced usual care (EUC) app was conducted with adults in geographic areas impacted by major hurricanes in 2017 and 2018. Social media advertisements were used for recruitment. Baseline and 3-, 6-, and 12-month post-baseline assessments were completed via the app.

Participants

Participants were eligible if they lived in a geographic area affected by a major hurricane that made landfall in 2017 or 2018, were at least 18 years of age, had access to an internet-accessible device, and were English speaking (Spanish translation of the app occurred after the study period). Participants included 1,357 adults recruited from zip codes affected by Hurricanes Harvey (Texas, n=158), Irma (Florida, n=58), Maria (Puerto Rico, n=708), Florence (North Carolina, n=80), and Michael (Florida, Georgia, n=353). Participants averaged 42.8 years (Mdn=43, range=18–78) and were predominantly women (75.2%). Nearly half reported Hispanic ethnicity (49.5%) and annual incomes under $20,000 (44.8%) (see Table 1).

Table 1.

Participant Background Information

Overall Sample Full BBN Condition (n=717) Enhanced Usual Care (n=640)
N % N % N %
Race/ethnicity
 Non-Hispanic 459 33.8% 231 32.2% 228 3.6%
White
 Non-Hispanic 75 5.5% 34 4.7% 41 5.7%
Black
 Hispanic/Latinx 672 49.5% 361 50.3% 311 48.6%
 Asian/Asian 4 0.3% 1 0.1% 3 0.5%
American
 Native American 23 1.7% 9 1.3% 14 2.2%
 Native 4 0.3% 3 0.4% 1 0.2%
Hawaiian/Pac Isl
 Non-Hispanic 21 1.5% 13 1.8% 8 1.3%
Other
 Did not report 99 7.3% 65 9.1% 34 5.3%
Gender
 Men 270 19.9% 140 19.5% 130 20.3%
 Women 1021 75.2% 536 74.8% 485 75.8%
 Other 14 1.0% 9 1.3% 5 0.8%
 Did not report 52 3.8% 32 4.5% 20 3.1%
Hurricane/Area
 Maria 708 52.2% 381 53.1% 327 51.1%
 Harvey 158 11.6% 102 14.2% 56 8.8%
 Irma 58 4.3% 32 4.5% 26 4.1%
 Michael 353 26.0% 171 23.8% 182 28.4%
 Florence 80 5.9% 31 4.3% 49 7.7%
Open in a new tab

Bounce Back Now (BBN) App

BBN consists of: (1) a weekly tracking tool for symptom self-monitoring; (2) education, coping tools, and quick tips to normalize survivors’ experience and reduce distress; (3) self-help components targeting mood, PTSD, and sleep; and (4) access to professional support and resources. Earlier versions were tested using web-based platforms.22,25 See Figure 1 for sample screenshots of BBN.

Figure 1: Bounce Back Now Sample Screenshots.

Figure 1:

Figure 1:

Figure 1:

Open in a new tab

The screenshots illustrate some of the key components of the Bounce Back Now mobile application, which also is available in Spanish (Pa’lante Hoy).

Weekly check-up.

Ten questions targeted feelings of depression, nervousness, distress, avoidance, hopelessness, restlessness, sleep quality, worthlessness, perceived effort burden, and worry.29,30 The app prompted users to complete the screen weekly. It graphically displayed results to encourage self-monitoring. Tailored recommendations were provided on the home screen. Users with mild symptoms (i.e., “a little of the time”) were advised to access Coping Tools whereas those with moderate or greater symptoms (i.e., “some,” “most”, “all of the time”) were advised to use the Activate (mood), Sleep (sleep quality), and/or Write (PTSD) elements.

Coping Tools featured psychoeducation and coping exercises. Tailored education (i.e., guided by user clicks) was provided about the impact of disasters, risk factors, and common emotional reactions. Deep breathing, progressive muscle relaxation, and mindfulness activities (~2–3 min) were guided via audio-narration. “Quick tips” educated users about how to cope with emotional reactions.

Activate, Sleep, and Write.

Activate was based on the efficacious Brief Behavioral Activation Treatment for Depression.31,32 Users were asked to identify values in five life areas (i.e., Relationships, Education/Career, Hobbies/Recreation, Health/Spirituality, Daily Responsibilities) and prompted to select fun/functional activities to rate and schedule. Users initially selected up to three values and five corresponding activities but on subsequent visits were permitted to exceed these limits. Users rated the enjoyment, importance, and difficulty of selected activities and scheduled and tracked completion of activities in the app. Daily mood ratings were recorded and graphically depicted in relation to activities completed.

The Sleep component was based on Cognitive Behavioral Therapy for Insomnia (CBT-I), an efficacious treatment for insomnia and related disorders.38,39 Education was provided about sleep, sleep problems, and sleep hygiene. Users were prompted to complete a sleep log capturing when they went to bed, fell asleep, woke up, left the bed, and how much time spent awake in bed. Recommendations to improve efficiency (i.e., proportion of time sleeping to time in bed), were tailored based on sleep logs. The app suggested wake and earliest bedtimes, consistent with CBT-I recommendations.40 Graphical depictions of sleep data were provided.

The Write component included education and rationale supporting the benefits of writing exposure, based on expressive writing paradigms,33,34 including Written Exposure Therapy, a brief efficacious intervention.35–37 Users were advised to write for 30-minute periods about their most upsetting event on five occasions, once or twice per week. The app assisted in scheduling these sessions, provided a timer, and encouraged use of pen and paper for writing. Motivational enhancement content was delivered when users delayed or did not complete writing tasks. Instructions for “sessions” 2–5 emphasized the importance of writing for 30 minutes, focusing on the most distressing parts of the event.

Get Help provided education about mental health resources and access to professional care. The principal focus was the Disaster Distress Helpline, a SAMHSA-administered national helpline that provides to year-round crisis counseling and/or local referrals for individuals in distress before, during, or after disaster. Crisis counselors are available by text or phone, in English or Spanish. “Get Help” also provided links to the American Red Cross (ARC), Federal Emergency Management Agency, Ready.gov, National Institutes of Health, and the Centers for Disease Control and Prevention.

Enhanced Usual Care (EUC) App

The EUC app provided descriptions of and links to disaster assistance and mental health resources consistent with BBN’s Get Help content. This included the Disaster Distress Helpline and links to resources via the ARC, Federal Emergency Management Agency, Ready.gov, National Institutes of Health, and Centers for Disease Control and Prevention websites.

Measures

Baseline, 3-month, 6-month, and 12-month postbaseline assessments were administered via the app.

Demographics.

Participants provided standard biographic data, including age, sex, educational achievement, marital status, racial/ethnic status, and household income.

Impact of disaster.

Impact of disaster was measured with 19 items used in our epidemiologic research assessing personal impact, (e.g., fear of death/injury, job loss), property loss (e.g., damage to home), and service loss (e.g., days without water, shelter, electricity).

Depressed mood.

The Patient-Reported Outcomes Measurement Information System (PROMIS) – Depression Scale (version 8a)42 is an 8-item measure of depressive symptoms.

Sleep.

The PROMIS – Sleep Disturbance Scale (version 8a) is an 8-item measure to assess sleep disturbance and sleep-related impairment.43

Posttraumatic stress.

The PTSD Checklist-Civilian version (PCL) is a 20-item measure of PTSD symptoms with strong psychometrics.41

BBN use.

BBN use was passively tracked and included how many times each participant logged into BBN, how many app components or activities a participant accessed/used, and total number of screens navigated between the baseline and 12-month post-baseline assessments.

Procedure

All aspects of the study were approved by the Institutional Review Board (IRB) at the lead institution. Our initial protocol included plans to triage and recruit disaster survivors on scene in partnership with ARC, but changes to the structure of ARC necessitated an alternative strategy. Geographically targeted Facebook ads were used (e.g., “Were you affected by Hurricane Irma? A new app developed by health care experts may be helpful. Click to receive an access code.”). Individuals who clicked the ad were redirected to the study website, which described the study and provided a download link and randomly generated access code that determined which app (BBN, EUC) was available. Because this was a fully remote trial with no more than minimal risk, participants were provided a Statement of Research and enrolled into the study after indicating agreement.

The intervention was self-directed. The Statement of Research did not give instructions about the use of the app. Once participants entered the BBN app after completing the baseline assessment, they were prompted by the app to complete the Weekly Checkup, which produced tailored recommendations on the home screen. Users were free to use any component flexibly throughout the study period. When post-baseline assessments were due, participants received a push notification through the app as well as an e-mail reminder and link.

Participants received gift codes (i.e., $10 for baseline, $15 for 3- and 6-month assessments, and $25 for 12-month assessment) and were entered into a raffle for an iPad after each assessment. An automated reminder with a direct link prompted participants to complete assessments when due. Use of the app was not incentivized. Data collection began within a few months of each hurricane after an initial assessment of disaster impact and review and approval by the institutional review board and federal funding agency.

Analytic Plan

Data were examined for missingness and analytic assumptions. There was significant attrition (not atypical in fully remote trials) in the experimental (40.8% at 3 months, 54.0% at 6 months, and 60.0% at 12 months) and control conditions (42.7% at 3 months, 52.3% 6 months, and 57.2% at 12 months). However, missing data were minimal at baseline (<10% missing across each variable). Per Little’s MCAR, missing data, including attrition, were not missing at random (p<.01). Age was the sole predictor of attrition, with older adults having greater attrition at each time point (ps<.01). Missing data were estimated using Full Information Maximum Likelihood (FIML), a model-based estimation method that has been shown to reduce bias relative to other methods (e.g., deletion).44 FIML uses all available data points and the specified model to account for missingness, as opposed to imputation methods.45 FIML was used for all tests of treatment outcomes. Several technical (e.g., flagging/blocking of users deemed “suspicious” based on IP/e-mail address, duplication of contact information, speed of baseline completion) and oversight (e.g., review and exclusion of flagged enrollees) protections prevented individuals from enrolling more than once. This process identified 704 entries that were excluded from analyses, resulting in a final sample of 1,357 adults.

First, descriptive analyses of BBN use were examined, including number of logins and the number of completed app components. When possible, use variables were compared across BBN and the control condition using t-tests. Intent-to-treat analyses were used to test efficacy. Latent change score models examined differences in symptom change across condition. Across all variables, latent variables and latent change variables were created from sum total scores. Change between latent variables representing successive post-baseline assessments was used as the primary dependent variable. Condition was examined as a predictor of latent change. Decisions regarding acceptable model fit used the following criteria by Hu and Bentler (1998):46 CFI≥.95, RMSEA≤.06, and SRMR≤.08. Chi-square tests were examined but the associated significance test was not used to make decisions regarding model fit. As an additional indicator of efficacy, Cohen’s d was calculated for all comparisons of BBN and the control condition using unestimated data (Note: BBN use data contained no missing observations as it was automatically tracked). Finally, to assess the extent to which BBN was efficacious in reducing the frequency of potentially diagnostic levels of symptomology, outcomes were also examined using logistic regression with baseline symptom severity as a covariate and condition as the primary predictor across each variable. Dependent variables for logistic regressions were dichotomously coded indicators of whether participants exceeded established clinical cutoffs at each post-baseline assessment (PROMIS depression > 21, PROMIS sleep > 29, PCL-5>31). For PROMIS measures, these correspond to T-score values of 60 or higher and the recommended cutoff score for likely PTSD based on evaluations of the PCL5.47,48 For tests of efficacy, we did not control for multiple comparisons as each test represented a test of a unique hypothesis or set of hypotheses.

Mplus version 8.0 was used for analyses. Analyses were not pre-registered.

Results

Randomization and Baseline Sample Description

Of 1,357 participants, 717 were randomized to BBN and 640 to EUC (an error in the initial coding algorithm resulted in consecutive assignment of some participants to BBN early in the study). Across bivariate comparisons (t-tests and χ2 tests), BBN and EUC participants did not differ demographically or clinically even without controlling for multiple comparisons (ps>0.05). Table 2 describes means and standard deviations for measures of depressive, sleep, and PTSD symptoms; and Table 3 describes the percentage of participants in BBN vs. EUC who exceeded clinical cutoffs on these measures at baseline.

Table 2.

Means and Standard Deviations for PCL-5table, PROMIS Depression, and PROMIS Sleep

Overall Sample Full BBN Condition (n=717) Enhanced Usual Care (n=640)
M SD M SD M SD
PCL-5 raw total scores
 Baseline (n=1,149) 37.20 18.90 36.98 19.13 37.44 18.66
 3-month post-baseline (n=787) 33.01 19.40 32.13 18.83 33.94 19.97
 6-month post-baseline (635) 31.40 20.11 29.98 20.00 32.93 20.14
 12-month post-baseline (561) 28.85 20.23 27.67 20.25 30.17 20.16
PROMIS Depression
 Baseline (n=1,309) 23.45 8.94 23.33 8.97 23.57 8.92
 3-month post-baseline (n=787) 21.81 9.17 21.31 9.08 22.34 9.25
 6-month post-baseline (635) 21.28 9.17 20.45 9.67 22.17 9.33
 12-month post-baseline (561) 20.84 9.75 20.15 9.52 21.61 9.97
PROMIS Sleep
 Baseline (n=1,306) 27.64 8.55 27.67 8.59 27.60 8.51
 3-month post-baseline (n=787) 26.93 8.78 26.54 8.77 27.34 8.79
 6-month post-baseline (635) 26.30 8.80 25.73 8.85 26.91 8.72
 12-month post-baseline (561) 25.91 8.70 25.17 8.76 26.74 8.57
Open in a new tab

Table 3.

Percentage Exceeding Clinical Cutoffs: PCL-5, PROMIS Depression, PROMISE Sleep

Overall Sample Full BBN Condition (n=717) Enhanced Usual Care (n=640)
N % N % N %
Exceeds PCL-5 cutoffs (> 31)
 Baseline (n=1,149) 692 60.2% 358 49.9% 224 52.2%
 3-month post-baseline (n=787) 401 51.0% 202 50.0% 199 52.0%
 6-month post-baseline (635) 307 48.3% 148 44.8% 159 52.1%
 12-month post-baseline (561) 266 43.2% 131 40.3% 135 46.4%
Exceeds clinical cutoffs (> 21)
 Baseline (n=1,309) 776 57.2% 395 55.1% 381 59.5%
 3-month post-baseline (n=787) 409 52.0% 199 49.3% 210 54.8%
 6-month post-baseline (635) 320 50.4% 157 47.6% 163 53.4%
 12-month post-baseline (561) 290 47.1% 144 44.3% 146 50.2%
Exceeds clinical cutoffs (>29)
 Baseline (n=1,306) 592 45.3% 315 43.9% 277 43.3%
 3-month post-baseline (n=787) 321 40.8% 151 37.4% 170 44.4%
 6-month post-baseline (635) 244 18.0% 120 36.4% 124 40.7%
 12-month post-baseline (561) 232 37.7% 105 32.3% 127 43.6%
Open in a new tab

BBN Engagement and Utilization

BBN participants accessed the app an average of 16.8 times (SD=31.5) vs. 15.84 for EUC (SD=23.71), t(1,355)=0.61, p=0.541. BBN participants engaged more actively with the app during logins (M=104.50 actions [i.e., total number of screens navigated within the app across visits], SD=161.68) than EUC participants (M=64.13, SD=88.42), t(1,133.55)=5.79, p<0.001, d=0.31. Among components available only in BBN, most users (84.2%, n=604) accessed the Activate section. From the full BBN condition, users selected an average of 4.14 (SD=7.59) activities under Activate and reported having completed half of them (M=2.06, SD=36.30). Most (n=627, 87.4%) also accessed Coping Tools, an average of 50.65 times (SD=96.50). Participants used the Sleep and Write sections considerably less, with 18.3% and 22.6% accessing Write and Sleep, respectively. Most participants used Get Help (n=417, 58.2%).

Efficacy for Depression Outcomes

Depression scores decreased moderately from baseline to-3-month assessment in the BBN condition (M=4.10, SD=6.83, d=0.60); a smaller effect size was found for EUC participants (M=2.53, SD=6.99, d=0.36), d=0.23. The effect size comparing change scores across the two conditions were small when comparing each post-baseline assessment with baseline (3-month, d=0.12; 6-month d=0.13, and 12-month d=0.10). The latent change model evidenced good fit, χ2(12)=99.89, p<0.001, RMSEA=0.075 [0.062, 0.089], CFI=0.95, SRMR=0.071. Results from the latent change model indicated that the decline in depression symptoms was greater for BBN than EUC participants between baseline and 3-month post-baseline (b=−1.87, SE=0.34, p<0.001), but there were no significant differences between the 3- and 6-month post-baseline (b=−0.50, SE=0.38, p=0.193) nor between the 6- and 12-month post-baseline (b=−0.42, SE=0.41, p=0.303). The two conditions did not differ in their likelihood of exceeding clinical cutoffs at any of the post-baseline timepoints (ps>0.05) (see Table 4).

Table 4.

Regression Results with Condition Predicting Outcomes

Symptom Severity Outcomes from Latent Change Models b SE p
Depression
 Baseline to 3-month post-baseline −1.87 (0.34) <.001
 3-month to 6-month post-baseline −.0.50 (0.38) .193
 6-month to 12-month post-baseline −0.42 (0.41) .303
Sleep
 Baseline to 3-month post-baseline −0.91 (0.26) <.001
 3-month to 6-month post-baseline −0.58 (0.29) .046
 6-month to 12-month post-baseline −0.17 (0.31) .585
PTSD
 Baseline to 3-month post-baseline −4.22 (0.75) <.001
 3-month to 6-month post-baseline −1.47 (0.80) .066
 6-month to 12-month post-baseline −2.37 (0.84) .005
Dichotomous Outcomes of Clinical Cutoffs from Logistic Regression* aOR 95% CI p
Depression
 3-month post-baseline 0.74 0.52–1.05 .092
 6-month post-baseline 0.82 0.57–1.19 .294
 12-month post-baseline 0.80 0.54–1.17 .249
Sleep
 3-month post-baseline 0.60 0.42–0.86 .005
 6-month post-baseline 0.78 0.54–1.15 .207
 12-month post-baseline 0.54 0.36–0.79 .002
PTSD
 3-month post-baseline 1.07 0.75–1.53 .702
 6-month post-baseline 0.80 0.53–1.20 .280
 12-month post-baseline 0.82 0.54–1.24 .344
Open in a new tab

Note:

*

Logistic regressions were completed with dichotomous variables of whether participants exceeded established cutoffs for each measure (PROMIS 8a Depression > 21; PROMIS 8a Sleep > 29; PCL-5 > 31)

Efficacy for Sleep Outcomes

Sleep scores decreased moderately from baseline to 3-month assessment: 3.56 points (SD=7.24, d=0.49) for BBN and 2.86 (SD=6.97, d=0.41) for EUC. The effect size comparing change scores across the two conditions were small when comparing each post-baseline assessment with baseline (3-month d=0.15; 6-month d=0.17, 12-month d=0.18). The latent change model evidenced good fit, χ2(12)=31.51, p=0.002, RMSEA=0.035 [0.020, 0.051], CFI=0.99, SRMR=0.041. Results indicated that BBN participants, compared with EUC participants, evidenced a greater decline in sleep symptoms from baseline to three-month post-baseline (b=−0.91, SE=0.26, p<0.001) and from 3-month to 6-month post-baseline (b=−0.58, SE=0.29, p=0.046), but not between the 6- and 12-month post-baseline (b=−0.17, SE=0.31 p=0.585). After controlling for baseline symptoms and compared with EUC participants, BBN participants were less likely to exceed clinical cutoffs 3-months post-baseline (aOR=0.60, b=−0.51, SE=0.18 p=0.005) and 12-months post-baseline (aOR=0.78, b=−0.63, SE=0.20 p=0.002), but not 6-months post-baseline (aOR=0.54, b=−0.24, SE=0.19 p=0.207) (see Table 4).

Efficacy for PTSD Outcomes

PCL scores decreased moderately from baseline to 3-month assessment: 8.61 points (SD=15.70, d=0.55) for BBN and 6.47 (SD=16.36, d=0.40) for EUC participants. The effect size comparing change scores across the two conditions were very small when comparing each post-baseline assessment with baseline (3-month d=0.04; 6-month d=0.04, 12-month d=0.01). Fit indicators for the latent change model for PTSD approached but did not exceed thresholds for good model fit, χ2(11)=160.64, p<0.001, RMSEA=0.104 [0.090,0.119], CFI=0.91, SRMR=0.08. Results should therefore be interpreted with caution. Results indicated that BBN participants, compared with EUC participants, evidenced greater declines in PTSD symptoms from baseline to 3-month post-baseline assessment (b=−4.22, SE=0.75, p<0.001) and from 6- to 12-month post-baseline assessment (b=−2.37, SE=0.84, p=0.005), but not from 3- to 6-month post-baseline assessment (b=−1.47, SE=0.80, p=0.066). The two conditions did not differ in likelihood of exceeding clinical cutoffs at any post-baseline assessment (see Table 4; ps>0.05).

Discussion

The public health impact of disasters extends well beyond their immediate aftermath. Depression, PTSD, and sleep disturbance are prevalent post-disaster concerns, and scalable strategies are needed. We developed and tested BBN to understand the impact of a wholly self-directed digital mental health intervention designed to address common access barriers such as stigma, time, transportation, cost, and the desire to “handle the problem on my own.” Consistent with our work with disaster-affected adolescents,15 we hypothesized that BBN participants would experience greater reductions in symptoms of depression, PTSD, and sleep disturbance than EUC participants. Moderate symptom reduction occurred in both groups. BBN was associated with greater reduction in depressive, sleep, and PTSD symptoms than EUC (for PTSD, results should be interpreted with caution due to weak model fit). These differences were maintained over time. However, the proportion of participants who exceeded clinical thresholds in post-baseline assessments generally was statistically similar across study arms. These mixed findings underscore both the strengths and limitations of standalone self-guided post-disaster digital health interventions. Because thousands or millions of disaster survivors can be reached by such interventions at minimal cost, even small clinical benefits can have significant post-disaster population health impact. However, deployed as a singular public health solution, such interventions are likely to have weak clinical benefit at the individual level. A coordinated, integrated population health approach that leverages digital mental health resources while also actively connecting highly symptomatic survivors to more intensive, best-practice treatment options is likely optimal. BBN and its new Spanish-translated companion version, Pa’lante Hoy, have been made freely available via the iOS and Google Play app stores, and we have engaged disaster response entities and helplines to improve awareness and capacity to integrate this resource in the aftermath of future disasters.

Modest effect size estimates found in this study were consistent with conclusions drawn elsewhere that digital mental health interventions often do not deliver a sufficient therapeutic dose in real-world contexts.11 BBN and other digital resources are therefore likely better suited for integration into broader, multi-level public health responses. Education and symptom monitoring may be sufficient to support resilience and recovery for low-risk individuals. Others may benefit from brief interventions that address basic needs and distress in the acute and intermediate phases of post-disaster recovery.49,50 Individuals with mild or moderate symptoms may experience benefit from digital mental health interventions when deployed in the context of broader disaster response and healthcare solutions that include intermittent access to and contact with a helping professional or navigator. Others need best-practice treatments addressing PTSD, depression, suicidality, substance abuse, or other common mental health conditions. Individuals with serious mental illness may benefit from a wider range of services that address basic, social, and clinical needs. BBN attempted to touch on many of these elements. However, our use of a flexible app structure – intended to achieve high acceptability and user satisfaction51,52 – resulted in suboptimal dosage of active intervention ingredients, such as the Write (written exposure) and Sleep (CBT-I) components. These active components were included in BBN due to their strong evidence base and relatively uncomplicated structure, but considerable patience (e.g., daily sleep logs), time (e.g., 30-min blocks of writing time), planning (e.g., writing, sleep scheduling), and trust (e.g., perceived accuracy of recommendations offered by the app) was required on the part of the user.

Notably, moderate pre-post effect sizes were identified for PTSD, depressive, and sleep symptoms in both conditions (Cohen’s ds 0.49–0.60 [BBN], 0.36–0.41 [EUC]). Natural symptom recovery is most prominent in the first 4–8 weeks post-trauma, and most participants enrolled in the study at least three months post-disaster. Symptom reductions therefore may, in part, reflect the value of both interventions. Disaster assistance and mental health resources were provided in EUC, accessed an average of 15.8 times by participants, who may have experienced some benefit. Indeed, control groups in clinical trials often feature education and resources that have minimal clinically relevant overlap with the experimental arm. However, our EUC app provided links to numerous disaster response entities (ARC, FEMA, Ready.gov) with valuable education and resources, including the Disaster Distress Helpline. Some EUC participants with significant mental health needs may have benefited from this information that may not have been encountered outside of study participation. We cannot examine the clinical impact of EUC due to the lack of a waitlist-only control group.

There are limitations of note in addition to the lack of a second control group. First, this wholly remote trial cannot answer critical questions relating to the added value of a disaster response clinician or navigator. Second, self-report and app usage data were our only sources of data collection. Third, although our sample size was significant, most participants reported mild symptoms at baseline, which restricts potential clinical benefit. Research should examine changes over time with a larger sample of symptomatic participants in the context of a stepped-care intervention that integrates digital health elements. Fourth, attrition in wholly remote studies examining digital mental health interventions is a significant challenge and this study was no exception. Finally, the demographic composition of our sample (e.g., mostly women, below-average income) may limit generalizability of the findings to other populations.

There is growing need for disaster mental health interventions that are scalable and cost-efficient. Digital mental health interventions likely have a role to play, but such approaches should be deployed in the context of a broader disaster response framework. Stepped-care models show potential for high reach, efficacy, and cost-effectiveness.8 Stepped-care approaches with embedded digital health resources may help to address barriers to care, identify individuals in need of assistance, and educate and connect them to services based on their clinical needs via an evidence-informed matching process.

Acknowledgements.

We thank the digital development team including Sachin Patel, Jameson Burroughs, Phil Smeltzer, PhD, and Spencer Wilder of SpursTech Corporation; and Jonathan Tindall, Bernard Jansen, and Andrew Mathews of the Medical University of South Carolina. We also thank Josh Nissenboim and his staff at Fuzzco for their contributions toward designing and developing earlier iterations of Bounce Back Now.

Funding.

This research was supported by National Institute of Mental Health (NIMH) grants R01 MH107641 (MPIs: KJ Ruggiero, S Galea) and R01 MH119193 (MPIs: S Galea, KJ Ruggiero). Dr. Bunnell was supported by NIMH award K23 MH118482. Dr. Dahne was supported by National Institute on Drug Abuse award K23 DA045766. All views and opinions expressed herein are those of the authors and do not necessarily reflect those of the funding agency or respective institutions. The authors have no conflicts of interest to report.

Bibliography

  • 1.Norris FH, Friedman MJ, Watson PJ, et al. 60,000 disaster victims speak: Part I. An empirical review of the empirical literature, 1981–2001. Psychiatry. 2002;65(3):207–239. [DOI] [PubMed] [Google Scholar]
  • 2.Galea S, Ahern J, Resnick H, et al. Psychological sequelae of the September 11 terrorist attacks in New York City. N Engl J Med. 2002;346:982–987. [DOI] [PubMed] [Google Scholar]
  • 3.Fergusson DM, Horwood LJ, Boden JM, et al. Impact of a major disaster on the mental health of a well-studied cohort. JAMA Psychiatry. 2014;71(9):1025–1031. [DOI] [PubMed] [Google Scholar]
  • 4.Goldmann E, Galea S. Mental health consequences of disasters. Annu Rev Public Health. 2014;35:169–183. [DOI] [PubMed] [Google Scholar]
  • 5.Beaglehole B, Mulder RT, Frampton CM, et al. Psychological distress and psychiatric disorder after natural disasters: systematic review and meta-analysis. Br J Psychiatry. 2018;213(6):716–722. [DOI] [PubMed] [Google Scholar]
  • 6.Bonanno GA. Loss, trauma, and human resilience: Have we underestimated the human capacity to thrive after extremely aversive events? Am Psychol. 2004;59(1):20–28. [DOI] [PubMed] [Google Scholar]
  • 7.North CS, Pfefferbaum B. Mental health response to community disasters: A systematic review. JAMA. 2013;310:507–518. [DOI] [PubMed] [Google Scholar]
  • 8.Cohen GH, Tamrakar S, Lowe S, et al. Comparison of simulated treatment and cost-effectiveness of a stepped care case-finding intervention vs. usual care for posttraumatic stress disorder after a natural disaster. JAMA Psychiatry. 2017;74:1251–1258. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Boscarino JA, Galea S, Ahern J, et al. Utilization of mental health services following the September 11th terrorist attacks in Manhattan, New York City. Int J Emerg Ment Health. 2002;4:143–155. [PubMed] [Google Scholar]
  • 10.Demyttenaere K, Bruffaerts R, Posada-Villa J, et al. Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization World Mental Health Surveys. JAMA. 2004;291:2581–2590. [DOI] [PubMed] [Google Scholar]
  • 11.Kessler RC, Demler O, Frank RG, et al. Prevalence and treatment of mental disorders, 1990 to 2003. N Engl J Med. 2005;352:2515–2523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Wang PS, Lane M, Olfson M, et al. Twelve-Month Use of Mental Health Services in the United States: Results From the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2005;62(6):629–640. [DOI] [PubMed] [Google Scholar]
  • 13.Blais RK, Renshaw KD. Stigma and demographic correlates of help-seeking intentions in returning service members. J Trauma Stress. 2013;26:77–85. [DOI] [PubMed] [Google Scholar]
  • 14.Conner KO, Copeland VC, Grote NK, et al. Mental health treatment seeking among older adults with depression: The impact of stigma and race. Am J Geriatr Psychiatry. 2010;18:531–543. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Kessler RC, Berglund PA, Bruce ML, et al. The prevalence and correlates of untreated serious mental illness. Health Serv Res. 2001;36:987–1007. [PMC free article] [PubMed] [Google Scholar]
  • 16.Kim PY, Britt TW, Klocko RP, et al. Stigma, negative attitudes about treatment, and utilization of mental health care among soldiers. Mil Psychol. 2011;23(1):65–81. [Google Scholar]
  • 17.Andrade LH, Alonso J, Mneimneh Z, et al. Barriers to mental health treatment: Results from the Who World Mental Health surveys. Psychol Med. 2014;44:1303–1317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Mojtabai R, Olfson M, Sampson NA, et al. Barriers to mental health treatment: Results from the national Comorbidity Survey Replication. Psychol Med. 2011;41:1751–1761. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Amstadter AB, Broman-Fulks JJ, Zinzow H, et al. Internet-based interventions for traumatic stress-related mental health problems: A review and suggestion for future research. Clin Psychol Rev. 2009;29:410–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Baumel A, Edan S, Kane JM. Is there a trial bias impacting user engagement with unguided e-mental health interventions? A systematic comparison of published reports and real-world usage of the same programs. Transl Behav Med. 2019;9:1020–1033. [DOI] [PubMed] [Google Scholar]
  • 21.Schueller SM, Torous J. Scaling evidence-based treatments through digital mental health. Am Psychol. 2020;75:1093–1104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Ruggiero KJ, Resnick HS, Acierno R, et al. Internet-based intervention for mental health and substance use problems in disaster-affected populations: A pilot feasibility study. Behav Ther. 2006;37:190–205. [DOI] [PubMed] [Google Scholar]
  • 23.Ruggiero KJ, Davidson TM, McCauley J, et al. Bounce Back Now! Protocol of a population-based randomized controlled trial to examine the efficacy of a web-based intervention with disaster-affected families. Contemp Clin Trials. 2015;40:138–149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Ruggiero KJ, Resnick HS, Paul LA, et al. Randomized controlled trial of an Internet-based intervention using random-digit-dial recruitment: The Disaster Recovery Web project. Contemp Clin Trials. 2012;33:237–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Ruggiero KJ, Price M, Adams Z, et al. Web intervention for adolescents affected by disaster: Population-based randomized controlled trial. J Am Acad Child Adolesc Psychiatry. 2015;54:709–717. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Price M, Yuen EK, Davidson TM, et al. Access and completion of a web-based treatment in a population-based sample of tornado-affected adolescents. Psychol Serv. 2015;12:283–290. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Price M, Gros DF, McCauley JL, et al. Nonuse and dropout attrition for a web-based mental health intervention delivered in a post-disaster context. Psychiatry. 2012;75:266–283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Price M, Davidson T, Andrews JO, et al. Access, use, and completion of a brief disaster mental health intervention among Hispanics, African Americans, and Whites affected by Hurricane Ike. J Telemed Telecare. 2013;19(2):70–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Kessler RC, Barker PR, Colpe LJ, et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry. 2003;60:184–189. [DOI] [PubMed] [Google Scholar]
  • 30.Bunnell BE, Davidson TM, Winkelmann JR, et al. Implementation and utility of an automated text messaging system to facilitate symptom self-monitoring and identify risk for post-traumatic stress disorder and depression in trauma center patients. Telemed e-Health. 2019;25:1198–1206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Lejuez CW, Hopko DR, Acierno R, et al. Ten year revision of the brief behavioral activation treatment for depression: revised treatment manual. Behav Modif. 2011. Mar;35(2):111–161. [DOI] [PubMed] [Google Scholar]
  • 32.Hopko DR, Lejuez CW, Ruggiero KJ, et al. Contemporary behavioral activation treatments for depression: Procedures, principles, and progress. Clin Psychol Rev. 2003;23:699–717. [DOI] [PubMed] [Google Scholar]
  • 33.Smyth J, Helm R. Focused expressive writing as self-help for stress and trauma. J Clin Psychol. 2003;59(2):227–235. [DOI] [PubMed] [Google Scholar]
  • 34.Hirai M, Skidmore ST, Clum GA, et al. An investigation of the efficacy of online expressive writing for trauma-related psychological distress in Hispanic individuals. Behav Ther. 2012;43(4):812–824. [DOI] [PubMed] [Google Scholar]
  • 35.Sloan DM, Marx BP, Lee DJ, et al. A brief exposure-based treatment vs cognitive processing therapy for posttraumatic stress disorder: A randomized noninferiority clinical trial. JAMA Psychiatry. 2018;75(3):233–239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Sloan DM, Marx BP, Resick PA, et al. ; STRONG STAR Consortium. Effect of Written Exposure Therapy vs Cognitive Processing Therapy on increasing treatment efficiency among military service members with posttraumatic stress disorder: A randomized noninferiority trial. JAMA Netw Open. 2022;5(1):e2140911. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Andrews AR III, Acosta LM, Canchila MNA, et al. Perceived barriers and preliminary PTSD outcomes in an open pilot trial of Written Exposure Therapy with Latinx immigrants. Cogn Behav Pract. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Taylor DJ, Pruiksma KE. Cognitive and behavioural therapy for insomnia (CBT-I) in psychiatric populations: a systematic review. Int Rev Psychiatry. 2014;26(2):205–213. [DOI] [PubMed] [Google Scholar]
  • 39.Geiger-Brown JM, Rogers VE, et al. Cognitive behavioral therapy in persons with comorbid insomnia: a meta-analysis. Sleep Med Rev. 2015;23:54–67. [DOI] [PubMed] [Google Scholar]
  • 40.Manber R, Friedman L, Siebern AT, et al. Cognitive Behavioral Therapy for insomnia in Veterans: Therapist manual. Washington, DC: U.S. Department of Veterans Affairs; 2014. [Google Scholar]
  • 41.Bovin MJ, Marx BP, Weathers FW, et al. Psychometric properties of the PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders–Fifth Edition (PCL-5) in veterans. Psychol Assessment. 2016;28(11):1379–1391. [DOI] [PubMed] [Google Scholar]
  • 42.Pilkonis PA, Choi SW, Reise SP, et al. Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): depression, anxiety, and anger. Assessment. 2011;18(3):263–283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Yu L, Buysse DJ, Germain A, et al. Development of short forms from the PROMIS™ sleep disturbance and Sleep-Related Impairment item banks. Behav Sleep Med. 2012;10(1):6–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Enders CK, Bandalos DL. The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling. 2001:8(3),430–457. [Google Scholar]
  • 45.Lee DY, Harring JR, Stapleton LM. Comparing methods for addressing missingness in longitudinal modeling of panel data. J Exper Educ. 2019:87(4),596–615. [Google Scholar]
  • 46.Hu LT, Bentler PM. Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychol Method. 1998:3(4),424–453. [Google Scholar]
  • 47.Healthmeasures. PROMIS score cut points: General guidelines. Accessed on 24 June 2024 at https://www.healthmeasures.net/score-and-interpret/interpret-scores/promis/promis-score-cut-points
  • 48.Bovin MJ, Marx BP, Weathers FW, et al. Psychometric Properties of the PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders–Fifth Edition (PCL-5) in Veterans. Psychological Assessment. 2016:28(11),1379–1391. [DOI] [PubMed] [Google Scholar]
  • 49.Ruzek JI, Brymer MJ, Jacobs AK, et al. Psychological First Aid. J Ment Health Counsel. 2007;29(1):17–49. [Google Scholar]
  • 50.Berkowitz S, Bryant R, Brymer M, et al. The National Center for PTSD & the National Child Traumatic Stress Network, Skills for Psychological Recovery: Field Operations Guide. 2010.
  • 51.Yuen EK, Gros K, Welsh K, et al. Development and preliminary testing of a web-based, self-help application for disaster-affected families. Health Informatics J. 2016;22:659–675. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Ben-Zeev D, Schueller SM, Begale M, et al. Strategies for mHealth research: Lessons from 3 mobile intervention studies. Adm Policy Ment Health. 2015;42:157–167. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES