A Pilot Randomized Controlled Trial of the PTSD Coach App Following Motor Vehicle Crash-Related Injury

Abstract

Objective.

Posttraumatic stress disorder (PTSD) symptoms (PTSS) are common after minor injuries and can impair recovery. We sought to understand whether an evidence-based mobile phone application with self-help tools (PTSD Coach) could be useful to improve recovery after acute trauma among injured emergency department (ED) patients. This pilot study examined the feasibility, acceptability and potential benefit of using PTSD Coach among acutely injured motor-vehicle crash (MVC) patients.

Methods.

From September 2017-September 2018, we recruited adult patients within 24 hours post-MVC from the EDs of two Level 1 trauma centers in the United States. We randomly assigned 64 injured adults to either the PTSD Coach (n = 33) or treatment as usual (TAU; n = 31) condition. We assessed PTSD symptoms (PTSS) and associated symptoms at 1-month (83% retained) and 3-months (73% retained) post-enrollment.

Results.

Enrollment was feasible (74% of eligible subjects participated) but usability and engagement were low (67% used PTSD Coach at least once, primarily in week 1); 76% of those who used it rated the app as moderately to extremely helpful. No differences emerged between groups in PTSS outcomes. Exploratory analyses among Black subjects (n = 21) indicated that those in the PTSD Coach condition (versus TAU) reported marginally lower PTSS (95% CI: −0.30; 37.77) and higher PTSS coping self-efficacy (95%CI: −58.20; −3.61) at 3-months.

Conclusions.

We demonstrated feasibility to recruit acutely injured ED patients into an app-based intervention study, yet mixed evidence emerged for the usability and benefit of PTSD Coach. Most patients used the app once and rated it favorably in regard to satisfaction with and helpfulness, but longitudinal engagement was low. This latter finding may explain the lack of overall effects on PTSS. Additional research is warranted regarding whether targeting more symptomatic patients and the addition of engagement and support features can improve efficacy.

Minor injuries are common reasons for emergency department (ED) care [1]. Among the 4 million patients who seek care for motor vehicle crashes (MVCs) annually in the US, over 90% are discharged to home [2–4]. Despite being categorized as “minor”, these events can still result in poor outcomes. Many MVC patients develop chronic pain [5], disability [6], poor health-related quality of life [7] and posttraumatic stress disorder (PTSD) [8]. Specifically, up to 26% of patients with predominantly mild MVC-related injuries report PTSD and reduced quality of life throughout 2 years post-MVC [7, 8]. Moreover, MVC patients with PTSD at 6-months are 2.3 times more likely to have a comorbid psychiatric disorder at 2-years post-injury, suggesting that early PTSD increases vulnerability to long-term problems[8].

Even in the acute aftermath of ED care for minor injury, hyperarousal symptoms of PTSD are associated with daily pain [9]. Additionally, PTSD is prevalent and increases risk for recurrent health problems and mortality among patients seeking emergency treatment and evaluation for acute medical events such as acute cardiac syndrome (12%;[10, 11]), transient ischemic attack, and stroke (25% [11]). Notably, PTSD symptoms (PTSS) contribute to distress-related impairments [12–15] and poor outcomes [6, 8, 16] in the absence of full-blown PTSD. Addressing the acute psychological response to a stressful event and providing guidance on self-management strategies to reduce the negative impact of PTSS could lessen negative downstream effects on health and well-being.

No current evidence-based tools exist to assist ED patients in self-managing acute post-injury PTSS. In other settings, cognitive-behavioral therapies (CBT [17]) can reduce PTSS after injury [18–20]. However, these therapies are difficult for acutely injured patients to access due to barriers to mental health service use, e.g.: stigma, shame, and rejection [21, 22], low mental health literacy, lack of time, knowledge or resources [21], trauma-specific barriers (e.g., concerns about re-experiencing; avoidance symptoms), and post-injury physical limitations [23]. Nontraditional mental healthcare modalities, such as mobile digital technology (i.e., mHealth), may overcome barriers to acute post-injury care [22], increase access to and opportunities for mental health care [24–26], offer 24/7 availability, affordability, equity, immediate support, anonymity, and offer a tailored approach that provides users with control based on their symptomology [27, 28].

One such digital self-help tool is PTSD Coach which uses CBT principles to improve mental health literacy, motivate treatment seeking behaviors, and empower traumatized populations [29, 30]. In Veterans and civilians with established PTSD, it has been shown to reduce PTSD and associated symptoms [30–33]. Approximately 79% of ED patients in Level 1 trauma centers own smartphones [34]; Given this increased access, and that Hispanic and black individuals are more likely to rely on their mobile phone (versus other devices) to access the internet and health-related information (see [35]), PTSD Coach may present a unique opportunity to engage minority populations in mental health care and provide accessible and sustainable mental health care to underserved populations [36]; however, few evidence-based digital tools are designed to prevent post-injury PTSD [27](for exception, see [37]), and no prior study has tested a PTSD-focused self-management mHealth app for ED patients with minor injuries.

In this pilot study, we randomly assigned patients with minor MVCs to use PTSD Coach as desired for 1-month post-injury, or to a treatment as usual (TAU) condition to determine process and scientific outcomes. Primary outcomes include the: 1) feasibility of recruiting and retaining acutely injured patients into an mHealth RCT; and 2) usability and acceptability of PTSD Coach acute MVC patients. We also explored the potential benefit of the app to reduce post-injury sequelae. We hypothesized that: 1) >50% of ED patients would agree to screening (based on our prior ED recruitment rates [38]) and > 70% would be retained; 2) >80% of subjects would use PTSD Coach at least once a week and report it as moderately satisfying and helpful; and 3) PTSD Coach (vs. TAU) subjects would report lower mean PTSS and higher PTSS coping self-efficacy (SE) at follow-up. This pilot study was designed to inform future mHealth interventions aimed at improving PTSS after ED care for minor injury.

Method

Subjects.

We recruited adults (ages 18–65) in the EDs of two Level I trauma centers within 24 hours of an MVC or motorcycle crash (MCC)-related injury (including pedestrian or bicycle crash). Additional eligibility criteria included: English-speaking patients; musculoskeletal injury; smartphone ownership and the ability to download apps; self-reported life threat or exposure to actual or perceived serious injury (i.e., Criterion A of the PTSD diagnosis [39, 40]); and pain score ≥4 using the verbal numerical score (e.g.: “On a scale of 0–10, how severe is your pain?”). Exclusion criteria included: major lacerations resulting in significant damage to subcutaneous tissue and specific nerve injury; moderate to severe cognitive impairment secondary to head injury; self-inflicted injury; initiation of benzodiazepines, psychotropic medications, or psychotherapy in the ED; active pharmacological treatment or psychotherapy for PTSD; presence of neurological disease; and MVC/MCC caused by a medical condition (e.g. syncope).

Procedure.

The Human Research Protection Office of the University of Pittsburgh [PRO16010595] approved all procedures. We used convenience sampling and recruitment based on research assistant (RA) availability on weekdays (10am-6pm). RAs identified patients through the triage report viewed from the ED tracking software, then a member of the treating team confirmed initial eligibility. RAs then approached permission-granting patients in their private treatment room during breaks in medical care. The RA briefly described the study to interested patients, and confirmed eligibility criteria upon verbal consent for screening. After obtaining informed consent, randomization and the baseline (T1) assessment occurred prior to ED or hospital discharge; follow-up assessments occurred at 1-month (T1) and 3-months (T3) post-injury.

Randomization.

Subjects were randomly assigned to either the PTSD Coach or the TAU condition (parallel fashion) via the randomization module in the University of Pittsburgh’s REDCap program [41]; the PI used random researcher.org to create the randomization list (1:1 allocation ratio). Subjects were not blinded as the intervention was compared with usual care.

Intervention.

The RA assisted patients in downloading the instrumented version of PTSD Coach, which allows for app usage data capture. Subjects were instructed to use PTSD Coach at their own discretion for symptom management [33] for 1-month post-injury. The RA guided each subject through a 10-minute introduction session, then gave them a handout that summarized app navigation through the four primary psychosocial modules: Learn (psychoeducational material), Track (monitor symptoms), Manage Symptoms (CBT tools), and Find Support (crisis services). We sent reminder text messages and prompts about whether they had used PTSD Coach once per week beginning at 7 days post-T1 via the secure EZ texting (up to 13 total messages throughout 3-months).

TAU.

TAU subjects received identical treatment as the PTSD Coach subjects in the ED (excluding the app-related information) and received the same surveys at T2 and T3 (excluding the app-related items). TAU subjects also received a single text message at the same intervals that served as a reminder of the number of weeks/days until T2 (up to 9 total messages).

Surveys.

Subjects in both conditions completed the T1 survey in the ED on an iPad (via REDCap); the RAs assisted with survey completion if subjects were too injured. The T1 survey captured demographics, PTSS, PTSS coping SE, and pain-related items. At T2 and T3, a hyperlink for the follow-up surveys was emailed to subjects. Subjects received $20 for completing T1-T3; all subjects in the PTSD Coach condition received an extra $20 for the additional time burden during the 30-day intervention period.

Measures.

Demographic Information.

At T1, items included: age, sex, race (recoded into two categories reflecting white vs. non-white subjects), and income.

Medical Record Review.

We coded injury details from the medical record: role during the crash (e.g., driver, passenger, pedestrian, cyclist), hospital admission details, and information to calculate the injury severity score (ISS), a numeric description of injury severity (range: 1–75) from multiple trauma [42, 43].

PTSS.

In the ED, pre-injury PTSS related to a prior stressful event was assessed with the 8-item PTSD Checklist- Short Version [44]. Responses were summed to create a total score; A cutoff of 19 was used to determine probable PTSD status. At T2 and T3, the 20-item PCL-5 was used to assess PTSS stemming from the index MVC [45, 46]); a cutoff of 33 was used to determine probable PTSD status.

Mental and Behavioral Health Service Use.

At T1, subjects responded yes/no to whether they received current or lifetime treatment for 1) alcohol or drugs and 2) emotional or mental health problems. Responses were re-coded and dichotomized to reflect any history of mental or behavioral health services, and any current use of services.

PTSS Coping SE.

We assessed self-efficacy for managing PTSS at T1 (e.g., I can handle situations that remind me of the injury) with a modified 9-item scale ([32]; Instructions read: “Please rate your degree of confidence by recording a number between 0 (cannot do at all) - 100 (highly certain can do).”

Process Measures:

We determined the rates of recruitment and retention, and reasons for refusals, exclusion, and dropout. To determine usability of PTSD Coach, we: 1) objectively monitored app usage with an instrumented version of PTSD Coach that enabled such data to be captured and then transmitted to a research server; subjects were categorized as users if they opened at least 1 module during the 1-month post-injury monitoring period; 2) used a self-report survey whereby subjects selected the PTSD Coach modules they found to be most helpful and issues that made app use difficult; subjects could select more than 1 response. Further, acceptability, perceived helpfulness of and user satisfaction with the app was assessed with the 15-item self-report PTSD Coach Survey [47], plus an additional item about pain (item #4; see Table 2 for items). Responses ranged from (0 = not at all to 4 = extremely).

Table 2.

Satisfaction with PTSD Coach (n = 21)

	Endorsed Moderately or Greater Responses (%)	Mean (SD)
Helping me learn about symptoms of PTSD	14 (67%)	2.14 (1.42)
Helping me learn about treatments for PTSD	15 (71%)	2.19 (1.33)
Helping me find effective ways of managing my PTSD symptoms	16(76%)	2.57 (1.21)
Helping me find effective ways of managing my pain-related symptoms	10 (48%)	1.81 (1.63)
Helping me feel comfortable in seeking support	11(52%)	1.95 (1.46)
Helping me feel there is something I can do about my PTSD symptoms	17 (81%)	2.67(1.28)
Helping me track my symptoms	15 (71%)	2.43(1.43)
Helping me know when I’m doing better or when I’m doing worse	15 (71%)	2.29 (1.35)
Increasing my access to additional resources	16 (76%)	2.43 (1.36)
Providing practice solutions to the problems I experience	16 (76%)	2.43 (1.25)
Helping me overcome the stigma of seeking mental health services	15 (71%)	2.10 (1.34)
Helping me better understand what I have been experiencing	15(71%)	2.29 (1.35)
Enhancing my knowledge of PTSD	16 (76%)	2.52 (1.33)
Helping to clarify some of the myths about PTSD	15 (71%)	2.29 (1.45)
Providing a way for me to talk about what I have been experiencing	14 (67%)	2.05 (1.32)
Overall, how satisfied are you with the PTSD Coach app?	16 (76%)	2.62 (1.16)

Data Analysis

We conducted statistical analyses using SPSS version 24.0 (IBM, 2016) and defined significance using an alpha level of 0.05. We did not conduct a priori sample size calculations or power analysis due to the pilot nature of the study (e.g., to generate point estimates of effects versus a test of effects).

Process Outcomes.

We used frequencies and summary statistics to examine feasibility and usability. To test whether the randomization produced roughly equivalent groups, we calculated associations of group with variables using chi square for categorical variables and one-way ANOVAs for continuous variables. We used similar analyses to examine differences between completers and those who were not retained at T2 and T3, and bivariate correlations to evaluate the strength of associations between continuous variables.

Estimates of Benefit.

Given the pilot nature of this study, we did not conduct hypothesis testing, but instead examined means for patterns of change over time and between conditions. Specifically, we used paired samples t-tests to examine mean change over time for PTSS and PTSS Coping SE. To obtain mean differences in outcomes by condition status, we conducted independent samples t-tests; 95% confidence intervals (CIs) that do not cross zero indicate meaningful differences over time and between groups. Given the proximity of the ED assessment to the point of injury (< 24 hours), we did not assess baseline PTSS stemming from the MVC, and only examine differences from 1 to 3-months post-injury. We also used descriptive crosstab analyses to determine group differences in probable PTSD status.

Exploratory Subgroup Analyses.

To further explore race differences in outcomes, subgroup analyses (using paired samples and independent samples t-tests) were limited to black subjects (n = 27) as no subjects identified Hispanic or Latino, and few identified as “other (n = 5)” or “multiracial (n = 4)”.

Results.

Table 1 shows baseline characteristics of the 64 enrolled patients (33 in the app condition; 31 to TAU) from September 2017-September 2018. Subjects were on average 37 years old (SD = 13), primarily female (62%), non-white (56%), and had a modal annual income of ≤ $20,000. Most patients suffered from mild injuries (84% not requiring hospitalization). Randomization appeared successful as no large differences emerged between conditions on baseline or demographic factors. For the self-report outcomes, internal consistency as measured via Cronbach’s alpha was high from T1-T3 (⍺ range = 0.84– 0.97).

Table 1.

Descriptive statistics of the randomized sample (N = 64)

Demographics		App Group (n = 33)	TAU Group (n = 31)	Total (N = 64)
		N (%)/M (SD)	N (%)/M (SD)	N (%)/M (SD)
Age	Range: 18–65	37.3 (13.37)	36.90 (12.23)	37.0 (12.71)
Sex	Male	14 (42%)	10 (32.3%)	24 (38%)
	Female	19 (57.6%)	21 (67.7%)	40 (62.5%)
Race	White	15 (45.5%)	13 (42%)	28(44%)
	Black	14 (42.5%)	13 (42%)	27 (42%)
	Multiracial	2 (6%)	2 (6%)	4 (6%)
	Other	2 (6%)	3 (10%)	5 (8%)
Annual Income	≤ $20,000	14 (42%)	11 (35%)	38 (60%)
	$20,000–40,000	8 (24%)	12 (39%)	15 (23.5%)
	$40,000–60,000	5 (15%)	3 (90%)	8 (12.5%)
	≥$60,000	6 (18%)	5 (16%)	11 (17%)
Role During Accident	Driver	21 (64%)	23 (74%)	44 (69%)
	Passenger	4 (13%)	4 (13%)	10 (15.6%)
	Pedestrian or Bicyclist	6 (18%)	6 (18%)	10 (15.6%)
Pre- Injury Daily Pain	Yes	3 (9%)	6 (19.4%)	55 (86%)
	No	30 (91%)	25 (81%)	9 (14%)
Life Threat During Injury	Yes	15 (45.5%)	17 (54.8%)	32 (50%)
	No	18 (54.5%)	14 (45.2%)	32 (50%)
Admitted to Hospital	Yes	5 (15%)	5(16%)	10 (16%)
	No	28 (85%)	26 (84%)	54 (84%)
a Initial Pain Score	Range: 4–10	7.15 (2.06)	6.74 (1.93)	6.95 (1.99)
+ Injury Severity Score	Range: 1–9	2.19 (1.53)	2.00 (1.79)	2.10 (1.64)
Pre-Injury PTSS	Range: 0–33	9.61 (7.32)	10.29 (10.86)	9.92 (8.70)
Lifetime Mental/Behavioral	Yes	16 (48.5%)	16 (53.3%)	32 (48.4%)
Health Treatment	No	17 (51.5%)	14 (46.7%)	31 (50%)
	Missing	-	-	1(1.6%)
Current Mental/Behavioral	Yes	3 (9%)	2 (6.5%)	5 (8%)
Health Treatment	No	30 (91%)	29 (93.5%)	59 (92%)

^aInitial Pain Score = pain score reported during the screening assessment.

^*Note. Continuous variables are presented as M (SD); dichotomous variables are presented as n (%).

⁺Injury severity score was only available for 58 participants; n = 6 participants did not have codeable injuries to create this score.

^**Significance tests (chi square between categorical variables and one-way ANOVA between dichotomous and continuous variables) reveal no difference between groups for any descriptive

Process Outcomes: Feasibility of Recruitment & Retention.

Recruitment.

Figure 1 shows the CONSORT diagram. Of 244 ED patients approached, more than half (57%) agreed to screen for eligibility and 43% declined, primarily due to lack of interest (56%) and pain (22%) (Figure 2). Of those screened, 62% were eligible and 38% did not meet inclusion criteria, primarily due to low pain score < 4 (34%; n =18) and a lack of perceived serious injury or life threat (34%; n = 18); see Figure 3. Younger age was associated with eligibility (M = 36.76; SD = 12.74) versus non-eligibility (M = 42.73; SD = 14.21) (F (1, 131) = 6.19; p = 0.014), and non-white patients screened eligible at a higher rate (73%; n = 46) than white patients (53%; n = 38) (χ² (1, n = 135) = 5.86; p = 0.02).

Figure 1.

Consort diagram of subject flow throughout the protocol.

Figure 2.

Reasons for Ineligibility (N = 53)

Figure 3.

Reasons for refusals (N = 126) at both the pre-screening (n = 105) and post-screening stage (n = 21).

Sixty-five (76%) of the 86 eligible patients were interested in the study and signed the consent form. One subject was withdrawn prior to protocol initiation due to pregnancy. The primary reasons for declined participation at the post-screening phase was too much pain (38%; n = 8) (Figure 2). Analyses comparing eligible subjects who chose to enroll compared to those who declined enrollment revealed a greater percentage of non-white (56%) versus white subjects (44%) enrolled in the study (χ² (1) = 4.49; p = 0.034). Lastly, no demographic differences emerged among the final sample compared to patients who declined participation at either pre- or post-screening.

Retention.

Retention rates were moderately high, with 98% (n = 63) of patients completing T1 (one subject dropped out before completing T1 due to injury-related complications), and 83% (n = 53) and 73% (n = 47) completing T2 and T3, respectively. There were no large differences in baseline characteristics between those retained versus not retained at follow-ups. Of note, two subjects who were not retained at T2 completed T3. Relatedly, one subject in the TAU condition only partially completed the T3 survey due to a diagnoses of breast cancer; her data was complete for the primary outcomes, and therefore was included in the final analyses.

Process Outcomes: Usability and Acceptability of PTSD Coach.

Feasibility and App Usability.

App data revealed that 22 subjects (76% of the 29 retained at T1; 67% of those in the app group) used PTSD Coach by visiting at least one module within 1-month post-injury. App engagement per person ranged from lowest of 7 to highest of 156 unique actions over 30 days. See Supplemental Table 2 for app indicators: most app engagements (57.9% of total actions) occurred in week 1 by 17 subjects, 12.5% actions occurred in week 2 (12 subjects), 16.3% in week 3 (12 subjects), and 13.2% in week 4 (by 8 subjects). Additionally, 15.7% of interactions (by 10 subjects) occurred past the monitoring period in week 5 to 13. Most engagement occurred on Fridays (25.7%) and Wednesdays (18.5%), and after 5pm (5–9pm; 38.5% and 9pm-midnight; 12/5%) or early afternoon between 12–5pm (29.8%). The most accessed module was “Manage Symptoms” (33.2%), closely followed by self-assessment (30.6%), Learn (24.9%) and Get Support (11.4%). Most subjects (91%) accessed the Learn, Self-Assessment, and Manage Symptoms modules at least once. Similarly, most subjects accessed all four modules throughout 30 days (54.5%), and 31.2% accessed at least three modules.

Acceptability (n = 21).

Complete PTSD Coach outcome data was available for 21 subjects; one subject did not finish the self-report T2 due to a family emergency. The “Manage Symptoms” module was rated the most helpful by 57% (n = 12) of subjects, followed by the “Learn” (38%; n = 8) and “Self-Assessment “(24%; n = 5) modules; only 1 subject (5%) found the “Find Support” module to be the most helpful. Regarding factors that made it difficult to use the app (see Figure 4), “not enough time” was selected by most users (67%).

Figure 4.

Barriers to using PTSD Coach (n = 21)

On average, subjects rated PTSD Coach between moderately to very helpful (M = 2.30; SD = 1.11); See Table 2 for item-level details. Over three-quarters (76%) of subjects were at least moderately satisfied with the app overall, and 73% of the helpfulness items (shaded in grey, Table 2) were rated at least moderately helpful by >70% of subjects.

Primary Analyses.

Descriptive Analyses.

Table S1 shows bivariate correlations. Pre-injury (T1) PTSS were moderately to strongly associated with PTSS coping SE (negatively) throughout 3-months, and age was positively correlated with PTSS coping SE in the ED. One-way ANOVAs revealed that white versus non-white subjects had higher pre-injury PTSS (M = 12.50; SD = 8.83 versus M = 6.61; SD = 7.44; F[1,62] = 8.02, p = 0.006), T3 PTSS (non-significant trend: M = 24.38; SD = 22.13 versus M = 13.39; SD = 15.83; F[1,45] = 3.36, p = 0.074), and lower PTSS coping SE at both T2 (M = 57.56; SD = 31.65 versus M = 72.56; SD = 20.64; F[1,51] = 4.09, p = 0.048) and T3 (M = 64.46; SD = 32.48 versus M = 81.42; SD = 16.71; F[1,45] = 4.71, p = 0.035).

Probable PTSD Status.

At T2, TAU subjects screened positive for PTSD at a lower rate (40%; n = 8) than PTSD Coach subjects (60%; n = 12). However, at T3, the opposite pattern emerged, and fewer PTSD Coach subjects screened positive for PTSD (17%; n = 4) compared to those in TAU (30%; n = 7).

PTSS and PTSS Coping SE.

See Tables 3 and 4 for patterns of change over time and by condition. PTSS decreased from T2 to T3 (95% CI: 2.98– 9.82), but the group means were similar across conditions. No additional effects of time or condition were present.

Table 3.

Reduction in symptoms over time among the full sample (n = 47) and the exploratory subsample of black patients (n = 21).

Outcomes: Full Sample (n = 47)	Baseline Mean (SD)	30-day Follow-up (SD)	Change over time Mean (SD)	95% CI	90-Day follow-up Mean (SD)	Change over time Mean (SD)	95% CI
PTSS	-	25.73 (19.96)	-	-	19.33 (18.97)	6.40 (11.38)	2.98; 9.82
PTSS Coping Self- Efficacy	70.12 (19.26)	65.34 (27.53)	4.78 (23.05)	−1.64; 11.19	71.55 (27.72)	−0.85 (25.70)	−8.48; 6.78
Outcome: Subsample (n = 21)
PTSS	-	30.00 (21.20)	-	-	23.21 (12.71)	6.79 (11.18)	1.40; 12.17
PTSS Coping Self- Efficacy	64.31 (21.26)	53.99 (31.93)	10.32 (28.15)	−2.50; 23.13	61.43 (33.11)	4.62 (32.47)	−10.16; 19.41

Note. The change sore for PTSS represents the change from 30 to 90 days given that PTSS stemming from the index injury were not assessed at baseline. The change scores for PTSS Coping Self-Efficacy represent the change from baseline to 30-days, and the change from baseline to 90-days.

(SD) = Standard deviation; PTSS = Posttraumatic Stress Symptoms.

^*P <.05

^**P <.01

^***P <.001.

Table 4.

Group Differences in PTSS and PTSS Coping Self-Efficacy from Baseline to 90-days post-injury.

	Baseline Mean (SD) (n = 63)	Mean Difference (standard error)	95% CI	30-day Follow-up M(SD) (n = 53)	Mean Difference (standard error)	95% CI	90-Day follow-up M (SD) (n = 47)	Mean Difference (standard error)	95% CI
PTSS					−0.43 (5.83)	−12.15; 11.28		1.62 (6.09)	−12.15; 11.28
PTSD Coach	-			26.72 (17.77)			19.37 (16.29)
TAU	-			26.29(24.65)			21.00 (24.48)
PTSS Coping SE
PTSD Coach	70.29 (17.75)	−1.46 (4.78)	−11.02; 8.09	66.19 (24.72)	−3.87 (7.79)	−19.51; 11.78	76.00 (22.94)	−11.57 (8.35)	−28.26; 5.11
TAU	68.83 (20.15)			62.32 (32.01)			64.42 (33.16)

(SD) = Standard deviation; PTSS = Posttraumatic Stress Symptoms.

^*P <.05

^**P <.01

^***P <.001.

Exploratory subgroup analysis.

Upon limiting the sample to the 27 black subjects, 21 (78%) completed T2 and T3 (Tables 3 and 5), PTSS decreased from 1- to 3-months post- injury (95% CI: 1.40; 12.17); Black subjects in the app condition (compared to black TAU subjects) reported marginally fewer PTSS at T3 (the CI slightly crosses zero; 95% CI: −0.30; 37.77), and higher PTSS coping SE at T3 (95%CI: −58.20; −3.61).

Table 5.

Group Differences in PTSS and PTSS Coping Self-Efficacy from Baseline to 90-days post-injury among the subgroup of black patients (n = 21).

Outcomes	Baseline Mean (SD)	Mean Difference (standard error)	95% CI	30-day Follow-up M(SD)	Mean Difference (standard error)	95%CI	90-Day follow-up M (SD)	Mean Difference (standard error)	95% CI
PTSS					4.58 (9.07)	−14.40; 23.56		18.74(8.82)	−0.30; 37.77
PTSD Coach	-			28.82 (15.39)			14.90 (10.45)
TAU	-			33.40 (25.42)			33.63 (27.13)
PTSS Coping SE		−9.24 (7.66)	−25.03; −6.54		19.61 (13.59)	−48.05; 8.83		−30.91 (13.04)	−58.20; −3.61
PTSD Coach	69.29 (20.82)			63.33(28.24)			77.62 (22.25)
TAU	60.04 (18.85)			43.72 (34.00)			46.72(35.32)

(SD) = Standard deviation; PTSS = Posttraumatic Stress Symptoms.

^*P <.05

^**P <.01

^***P <.001

Discussion

This pilot RCT was the first to determine whether PTSD Coach is feasible to use among ED patients with minor injury. We found that it is feasible to enroll and retain acutely injured patients into an app-based intervention study within 24-hours post-MVC (retention ≥ 73%). However, our finding that 43% of those approached refused to be screened is slightly higher than the refusal rates of trials recruiting acutely injured patients for internet-based or in-person PTSD treatment (22–37%) [19, 37], potentially due to the proximity of the initial ED assessment to the point of injury (< 24 hours). Contrary to prior findings that minorities are typically less likely to seek treatment or engage in post-trauma symptom monitoring [48, 49], minorities (and younger vs. older patients) were more likely to screen as eligible, and eligible minorities also chose to participate at a higher rate compared to eligible white patients. Notably, a lack of smartphone or inability to download apps (≈ 4%) was not a limiting factor for enrollment.

Although we did not screen for or include initial PTSS as eligibility criteria, 22% of patients screened positive for significant pre-injury PTSS, and half of the subjects endorsed life threat during the crash; this latter finding indicates individuals with minor injury should not be overlooked for screening and monitoring purposes. Additionally, minorities reported higher pre-injury and T3 PTSS, and lower PTSS coping SE at T2 and T3, suggesting that our sample reflects a typically underrepresented population who may be particularly vulnerable to post-injury sequelae. Linking those with initial elevated risk (e.g., via pre-injury symptoms or reactions to the index injury) to screening and monitoring services in the ED, and eventual mHealth platforms may be a critical step towards increasing access to services, accelerating recovery, and/or mitigating long-term sequaelae of minor MVCs.

Usability and Acceptability.

We found mixed evidence in support of usability. Approximately two-thirds of the patients in the PTSD Coach condition used the app at least once, indicating that injured MVC patients are at least willing to try a self-management support app. Although most subjects visited at least 3 of the 4 primary modules, and more than half visited all four modules at least once, longitudinal engagement was less-than-optimal. The majority of app use occurred in week 1; the number of app interactions steeply dropped off from almost 60% to under 17% through weeks 2–4. As such, we did not find evidence to support our hypothesis that at least 80% of patients would use the app weekly. Additionally, less than half of the subjects used the app after week 1. This data indicate that our intervention suffered from both adoption (many users may have been simply experimenting with the app in week 1) and attrition. Notably, 10 subjects continued to use the app following the monitoring period, yet actions were low throughout 5–13 weeks.

The lack of higher usage may indicate that many may not be intrinsically motivated to engage in self-care related to PTSS after trauma and/or do not have time or interest in the app. Although the “Manage Symptoms” module was the most accessed, longitudinal support may be necessary. Additionally, PTSD Coach may not be ideally designed to maximize engagement, as users must interact with the app first rather than tailored app content being “pushed” to its users. Prompts to increase engagement are critical to determine app efficacy moving forward.

Consistent with hypotheses regarding acceptability, many acutely injured patients who used the app and provided follow-up found value in PTSD Coach, as over three-quarters of subjects were at least moderately satisfied with the app overall. Similar to Veterans and to trauma-exposed community members [30, 47], acutely injured patents rated PTSD Coach as moderately-very helpful, and were equally satisfied with the app. Consistent with objective data, the Manage symptoms module was self-reported as the most helpful feature. However, the items with the highest ratings of helpfulness involved a broad range of increasing knowledge, resources, and management of PTSS. These results are notable, as psychoeducation is critical to facilitate mental health service use and provide patients with options for therapy [21], yet rarely adequately provided to patients during acute medical care [50, 51]

Magnitude of Benefit.

Although PTSS declined over time in both groups, PTSD Coach did not serve to improve outcomes relative to controls. However, contrary to the literature regarding the adverse effects of psychological debriefing [52, 53], there is no indication that PTSD Coach is harmful after injury. The exploratory subgroup analyses, although preliminary and based on a small sample, suggest that minority patients may have benefited from access to PTSD Coach to a greater extent than White patients. At 3-months post-injury, PTSS reported by black subjects in the app condition were less than half the level of those reported by TAU black subjects. Similarly, levels of PTSS coping SE were dramatically improved among black subjects in the app condition. Additionally, black TAU subjects were the only subgroup to report average PTSS at the cutoff level for probable PTSD (33 points).

Although black individuals have the highest lifetime prevalence of PTSD [49, 54], injured black men admitted to a Level 1 trauma center cited financial constraints, limited access to, and fear of judgement from mental healthcare professionals as primary obstacles to seeking treatment [55]. As such, access to a private self-management tool such as PTSD Coach may be viewed more positively by vulnerable populations; by nature, an app-based tool eliminates the stigma and implicit bias (from both patient and provider) associated with face-to-face encounters, and subjects may be more receptive to its content. Further, Black individuals are more likely than White individuals to use smartphones as their primary mode of internet access[35, 36]. To this end, mHealth tools may be well-positioned to reduce racial disparities and provide accessible and sustainable mental health care to black patients [36]. Alternatively, given that minority subjects had high PTSS and room to improve, mHealth tools such as PTSD Coach may be more efficacious for those with high symptomology (regardless of race).

Consistent with our overall null findings, Mouthaan and colleagues (2013) failed to find support for a self-guided internet and CBT-based intervention to prevent PTSD onset after injury. Possible explanations include: 1) small sample size and lack of power to detect a small effect among a heterogeneous sample of patients with varied PTSS; 2) insufficient levels of user engagement to produce a clinically significant effect; 3) inappropriate timing of app introduction (< 24-hours post-injury) among patients with low PTSS. Additionally, PTSD Coach: 4) may be insufficient as a stand-alone treatment or to address the full spectrum of PTSS [56] and/or ED patients may have different needs immediately post- injury; 5) may only benefit those at high-risk for persistent PTSS; and 6) may not be ideally designed to maximize effects in this population. Future research addressing these factors may aid in determining the true efficacy of PTSD Coach among acutely injured patients.

Future research is also warranted to determine methods to improve engagement with the app and efficacy in managing post-injury sequelae: 1) Increasing motivation for patients with minor injury to engage in PTSS self-care; 2) Directing patients to the learn and symptom management modules during the times when they may benefit the most, and pushing useful content or reducing the time within the app to find material, 3) Adapting the Find Support module to feature resources about injury recovery, and 4) Personalizing content that is relevant based on specific deficits and strengths. Paraprofessional telephone support [56], and in-person and telephone clinician support [33] have also shown promise to improve engagement and utilization among Veterans. Additionally, a new version of PTSD Coach was recently released (January, 2020) with added features regarding personalization of app content, and learning about and managing symptoms; it is also available in Spanish.

Further, given high rates of comorbidity between PTSD, depression and anxiety disorders (ranging from 20.8%–21.1% through 2-years post-injury), future work is needed to determine effective interventions to address this dynamic and comorbid symptomology[8]. Similarly, age-related motivational, physical and cognitive barriers may impact mHealth usage[57, 58]; although beyond the scope of this work, these interactions should be considered for future studies, particularly given that ages 40–49 and 65+ are often associated with poorer post-injury recovery[59]; yet, increasing age was correlated with greater ED levels of PTSD coping SE in the current study.

Although this study was limited to acutely injured patients, future work may determine whether PTSD Coach is beneficial to other samples at-risk for PTSS (e.g., acute medical events)[10, 11]. Given that many patients in this trial had low symptoms, our enrolled sample, as a whole, was not necessarily considered high-risk, nullifying the need for a symptom management tool: 19% of subjects felt that they did not need PTSD Coach. Moving forward, it may be critical to target a sample who reacted to the injury with high initial distress. Further, increasing the sample size and the period of app usage to 3-months, and extending the outcome assessments to at least 6-months may improve the likelihood of detecting a small effect [30, 32] and allow for a determination of long-term effects. Lastly, PTSD Coach may be better positioned for use as an adjunct or first tier within a stepped care model to improve post-injury recovery [37], with initial ED screening and symptom monitoring after injury to maintain communication with and identify those with high symptoms in the weeks and months to follow [9, 38, 60].

Limitations.

This pilot study had a small sample size, a limited 1-month time-period of suggested app use, and a limited follow-up period of 3-months post-injury, the earliest time when PTSS is considered persistent [19]. Notably, the validity of our findings, particularly those regarding group differences for black patients, is threatened by the small sample size and a lack of a priori planned analyses or sample size calculation; future research is warranted to replicate these findings using adequately powered pre-planned samples. Additionally, qualitative feedback was not obtained, and app-based customizations/personalization were not included; this information would help to inform improvements regarding app feasibility, usability, and engagement. Further, our liberal definition of app usage (e.g., those navigating to at least 1 content area in 1-month) is not ideal; we are lacking granular details regarding how subjects interacted with the app (e.g., how much time was spent in the app and each module; how and why modules were most frequently used; whether modules were completed) that would be useful in informing user engagement and efficacy. Additionally, enrollments were only possible during weekdays (versus 24/7 ED coverage), and no patients identified as Hispanic or Latino, potentially due to the English-speaking inclusionary criteria. Our sample was also limited to minor-moderate MVC-related musculoskeletal injury and to those with a minimum pain score of 4, thereby limiting generalizability. Finally, given that app group patients were compensated an additional amount and received weekly text message prompts, it is unknown whether and/or how acutely injured patients would use the app in real world circumstances, outside of the study context.

Conclusion.

The results from this pilot trial provide mixed evidence for using PTSD Coach among acutely injured patients without diagnosis of PTSD. Future efforts may focus on examining larger and more PTSD-affected patient populations and improving app engagement through tailored feature development. Although self-guided PTSD Coach did not reduce symptoms immediately following injury among the full sample, large mean differences emerged in the subgroup analyses, suggesting that black adults may have benefited from having access to an mHealth app for mental health. This trial serves to advance the extant literature among this understudied population; Optimally-designed remote digital support for self-managing post-injury PTSS could reduce the public health burden of ongoing health issues related to chronic symptoms.