Pilot Randomized Controlled Trial of a Novel Web-Based Intervention to Prevent Posttraumatic Stress in Children Following Medical Events

Abstract

Objective To assess feasibility and estimate effect size of a self-directed online intervention designed to prevent persistent posttraumatic stress after acute trauma. Methods Children aged 8–12 years with a recent acute medical event were randomized to the intervention (N = 36) or a 12-week wait list (N = 36). Posttraumatic stress, health-related quality of life, appraisals, and coping were assessed at baseline, 6, 12, and 18 weeks. Results Most children used the intervention; half completed it. Medium between-group effect sizes were observed for change in posttraumatic stress severity from baseline to 6 weeks (d = −.68) or 12 weeks (d = −.55). Exploratory analyses suggest greatest impact for at-risk children, and a small effect for intervention initiated after 12 weeks. Analysis of covariance did not indicate statistically significant group differences in 12-week outcomes. Conclusions This pilot randomized controlled trial provides preliminary evidence that a self-directed online preventive intervention is feasible to deliver, and could have an effect in preventing persistent posttraumatic stress.

Exposure to potentially traumatic events is an unfortunately common experience for children, with population samples estimating that two-thirds of children have had exposure to at least one traumatic event by age 16, including violence, injury, illness, disasters, and fire (Copeland, Keeler, Angold, & Costello, 2007). Events related to injury, acute medical illness, and medical treatment are among the most common traumatic experiences of children. In the United States, injuries to children result in more than eight million emergency room visits per year (Grossman, 2000), and nearly six million US children each year are admitted to hospitals under adverse, often life-threatening, circumstances (Agency for Healthcare Research and Quality, 2012). For children exposed to potentially traumatic medical events, psychosocial sequelae appear to play an important role in determining health and functional outcomes. In particular, studies have documented connections between posttraumatic stress symptoms (PTSS) and poor health and functional outcomes including health-related quality of life (HRQoL), even after controlling for injury or illness severity (Holbrook et al., 2005). Widely accessible interventions for secondary prevention of PTSS as a consequence of medical events and other acute traumas in the lives of children could thus have a large public health impact (Kazdin & Blase, 2011; Zatzick, Koepsell, & Rivara, 2009).

There are now a small number of published evaluation studies of early interventions for children that aim to prevent persistent PTSS after trauma exposure; few have found clear evidence of main effects on PTSS. Optimal early preventive interventions would both reduce acute PTSS and prevent the development of new or more severe PTSS over the initial post-trauma period; reducing early PTSS has the dual purpose of addressing immediate distress and helping to prevent longer-term sequelae (Kassam-Adams, 2014). A recent meta-analysis found preliminary evidence that preventive intervention could be helpful, but study heterogeneity precluded drawing clear conclusions about which elements are effective (Kramer & Landolt, 2011). In terms of broad reach, Web-based or mobile interventions could provide accessible, timely, and cost-effective tools to help children in the aftermath of an acute medical event or other trauma. There is increasing empirical support for the efficacy of Web-based interventions that deliver cognitive behavioral treatment to children for behavioral aspects of pediatric medical conditions and for child anxiety (March, Spence, & Donovan, 2009; Ritterband et al., 2013). To our knowledge, the only Web-based preventive intervention for child PTSS after acute trauma that has been examined in a randomized trial is “Kids and Accidents” (Cox & Kenardy, 2010), a psychoeducational intervention combining print information for parents with a basic, informational Web site (kidsaccident.psy.uq.edu.au) for youth. An initial randomized controlled trial (RCT; N = 56) showed a reduction of anxiety symptoms, and a trend for reduced PTSS among higher-risk children. Children’s self-reported use of the Web site was low, suggesting a need for a more interactive and engaging online intervention (Cox & Kenardy, 2010).

Description of the Coping Coach Intervention

The development of the Coping Coach intervention has been described in detail elsewhere (Marsac et al., 2013, 2015). Based on the empirical literature on etiology of PTSS in children, we identified key targets for an interactive child-directed online preventive intervention that would be likely to prevent development of persistent PTSS: (1) promoting adaptive cognitive appraisals, (2) decreasing excessive early avoidance coping, and (3) promoting use of social support (Kassam-Adams, 2014). To address these intervention targets, our team developed content and worked with professional Web and game developers to create an attractive and engaging game-like structure for intervention activities. Feedback from child users was incorporated at each stage of development. Audio is provided throughout the game (characters speak, instructions are voiced out loud) to minimize reading demands for the user.

The intervention is structured as an interactive game with an engaging storyline (i.e., the child user has to help the townspeople when their emotions have been “zapped,” keep the airship moving upwards, fix the weather machine that has made the world cloudy). Because Coping Coach is designed for children with a range of types of acute trauma exposure, we intentionally included game characters who have experienced different types of events. Each character also demonstrates key responses (emotional reactions, appraisals, avoidance behaviors) that are common across events. The user interacts (in any order) with four school-age child characters whose experiences are woven throughout the game: (1) a girl who has had a frightening asthma attack, (2) a boy who witnessed his brother getting beaten up in the community park, (3) a girl who experienced a car crash, and (4) a boy who experienced a house fire. Intervention activities include skills practice and interactions with game characters as the child user progresses through three levels of the game (pilot data suggested that each level requires approximately 20 min to complete). In Level 1, the user interacts with game characters to identify the characters’ feelings during and after acute traumatic events and then the user’s own feelings. In Level 2, the user learns about connections between feelings, thoughts, and behaviors; helps game characters recognize helpful/unhelpful thoughts; identifies his/her own unhelpful thoughts; and generates helpful, adaptive appraisals. In Level 3, the user helps game characters who are avoiding trauma reminders that seem scary (but are realistically safe) and identifies pros and cons of avoiding and approaching such reminders. The user also completes an “adventure log” that reinforces intervention activities. Users earn points for completion of game activities, and can repeat activities as many times as they wish.

Study Objectives

The aims of the current study were (a) to assess the feasibility of delivering a novel online preventive intervention to school-age children in the early aftermath of a potentially traumatic medical event, initiating the intervention while children were still in hospital, and (b) to estimate the effect size of the intervention for child PTSS to help frame the study design and sample size for a future large-scale trial. The use of a wait list condition rather than a no-treatment control condition allowed us to collect additional pilot data concerning the potential impact of a later intervention time frame.

Method

The study protocol was published (Marsac et al., 2013) and registered at clinicaltrials.gov (NCT01653288). Online intervention content and delivery mechanism remained stable across the course of the randomized trial, and there were no serious technical difficulties that prevented participants from using the intervention.

Participant Enrollment and Follow-Up

We enrolled children who had been admitted to the hospital for treatment of an acute medical event, and one parent per child. We defined an acute medical event as a sudden, unexpected, and new medical event for the child (i.e., new injury or illness diagnosis, or a sudden exacerbation of a chronic condition). Inclusion criteria were: (1) aged 8–12 years; (2) acute medical event within the past 2 weeks which the child perceived as potentially traumatic; (3) child’s medical record indicated a current Glasgow Coma Scale score greater than 12, i.e., they were awake and aware; (4) child spoke English well enough to complete measures and understand the intervention; and (5) child had access to the Internet at home. Children were excluded if their current medical condition or apparent cognitive limitations precluded participation in an interview, if the acute medical event was due to family violence or suspected child abuse, or if either child or parent was arrested or subject to legal proceedings related to the medical event.

Figure 1 depicts study flow from eligibility determination through follow-up assessments. In an institutional review board-approved study protocol, we identified eligible children admitted to the hospital’s intensive care unit, or general pediatric or surgical services, and approached parents to explain the study and invite their and their child’s participation in an initial brief screening assessment. Child assent was obtained after parents gave consent. Comparing demographic characteristics of the 83 eligible children who were enrolled for initial screening versus 241 eligible children not enrolled, no significant differences were observed for child gender or race/ethnicity; however, enrolled children were slightly younger on average (mean of 9.7 vs. 10.1 years; t (322) = 2.21, p = .03).

Figure 1.

CONSORT diagram of study screening, enrollment, randomization, and follow-up.

In the initial screening assessment, we determined whether the child met the additional inclusion criterion of perceiving the index event as potentially traumatic, by means of a validated four-item screen derived from the Acute Stress Checklist for Children (Kassam-Adams, 2006). If the child remained eligible, we obtained consent/assent for full study participation, participants completed baseline measures, and they were then randomized to either the immediate intervention (N = 36) or wait list (N = 36) condition. Table I shows demographics, type of medical event, and prior trauma exposure for the 72 participants enrolled in the RCT and randomized. Participants had experienced a range of types of acute medical events, and had hospital admissions ranging from 1 to 23 days; 65% were admitted for 3 days or less. Despite random assignment, children in the immediate intervention condition had higher mean PTSS severity at baseline than children in the wait list condition (Table I).

Table I.

Demographic and Event Characteristics and Trauma History, by Intervention Condition

Variable	Wait list (N = 36)	Immediate intervention (N = 36)
Child age in years (mean, SD)	9.8 (1.3)	9.8 (1.5)	NS
Child sex – male (N (%))	18 (50)	21 (58)	NS
Child race/ethnicity (N (%))			NS
Black	12 (33)	12 (33)
White	24 (67)	21 (58)
Other race/ethnicity	0 (0)	3 (9)
Parent education (N (%))			χ2 = 9.94, (df = 2), p = .007
Attended or completed high school	7 (19)	7 (19)
Attended college	8 (22)	20 (56)
Completed college or graduate degree	21 (58)	9 (25)
Annual household income (N (%))			NS
Less than $30,000	6 (17)	10 (29)
$30,000 to less than $50,000	7 (19)	8 (23)
$50,000 to less than $75,000	4 (11)	4 (11)
$75,000 to less than $100,000	4 (11)	5 (14)
$100,000 or more	15 (42)	8 (23)
Type of acute medical event (N (%))			NS
Appendicitis	18 (50)	13 (36)
Asthma-related	2 (6)	4 (11)
Abdominal pain	3 (8)	3 (8)
Acute joint pain or arthritis	3 (8)	1 (3)
Other acute medical illness	6 (17)	9 (25)
Injury	4 (11)	6 (17)
Length of hospital stay (days; mean, SD)	4.4 (4.2)	3.3 (2.6)	NS
Prior trauma exposure
Any prior trauma (N (%))	25 (71)	27 (75)	NS
Prior interpersonal trauma (N (%))	3 (9)	7 (19)	NS
Prior non-interpersonal trauma (N (%))	24 (69)	27 (75)	NS
Count of prior trauma types (median, range)	1, 0–3	2, 0–6	NS

Follow-up assessments were conducted by phone at 6, 12, and 18 weeks post-baseline by trained research interviewers blinded to participant condition. Children and parents each received a $10 incentive for completing each research assessment and the child received an additional $10 incentive if she/he completed the entire intervention at least once. Among the 72 RCT participants, 85% completed at least one follow-up assessment; the follow-up rate was lower in the immediate intervention group (72%) than in the wait list group (97%). We explored other potential predictors of follow-up completion: child gender, race/ethnicity, prior trauma exposure, baseline PTSS, and family income level. None of these factors were associated with follow-up completion.

Intervention Procedures

Immediately after baseline assessment and randomization, children assigned to the immediate intervention condition were introduced to the Coping Coach intervention. Research staff assisted the child in logging in to begin to use Coping Coach activities. Children were encouraged to complete the remaining activities online over the next month, and to replay the game as many times as they wished. Children assigned to the wait list condition received log-in information for Coping Coach following completion of the 12-week research assessment, and were given the same instructions to complete all online intervention activities over the next month and replay the game. In each group, parents were given a separate username and password and invited to log-in to “play” the game if they wished to see its content and activities. For both immediate intervention and wait list groups, during the 6-week period during which the child was allocated to use the Coping Coach intervention (i.e., between baseline and 6 weeks for the immediate intervention group, between 12 and 18 weeks for the wait list group), parents and children received tailored weekly reminders via email or text (according to participant preference) encouraging the child to complete any remaining Coping Coach activities, and to replay the game if she/he had completed it.

Measures

We gathered demographic information from parents and children, and abstracted diagnostic and treatment information related to the index medical event from the medical record.

Intervention Usage

Participants’ use of the online Coping Coach intervention was tracked automatically, with date and time stamps for each log-in and for interactions within the intervention, indicators for completion of key intervention tasks and activities, and capture of specific participant responses to questions and activities integrated within the interactive intervention.

Outcome Measures

The following measures of key outcomes were included in baseline, 6-, 12-, and 18-week assessments.

Child PTSD Symptom Scale (CPSS)

This 24-item self-report measure assesses PTSS (anchored to refer to the index trauma, in this case the medical event) (Foa, Johnson, Feeny, & Treadwell, 2001). CPSS item ratings can be summed to yield a symptom severity score (possible range: 0–51), based on 17 items that correspond to DSM-IV posttraumatic stress disorder (PTSD) symptom criteria; seven additional items assess impairment from those symptoms but are not included in the current analyses. The CPSS has well-established reliability and validity as a measure of posttraumatic stress symptomatology in children (Foa et al., 2001; Kassam-Adams, Marsac, & Cirilli, 2010). Internal consistency (Cronbach’s alpha) was excellent (.88) in the current sample.

Pediatric Quality of Life Inventory (PedsQL)

The PedsQL is a well-validated 23-item measure of child HRQoL (Varni, Seid, & Rode, 1999). At baseline, parents reported on child HRQoL before the index medical event, at subsequent assessments they rated current child HRQoL. Internal consistency was excellent (.92) in the current sample.

Measures of Hypothesized Mechanisms

The following measures were included in baseline, 6-, 12-, and 18-week assessments as indicators of hypothesized mechanisms of action.

Child Posttraumatic Cognitions Inventory (CPTCI)

This 25-item child self-report measure was adapted from the adult Post-Traumatic Cognitions Inventory (Foa, Ehlers, Clark, Tolin, & Orsillo, 1999) and validated in a large sample of children and adolescents aged 6–18 years (Meiser-Stedman et al., 2009). The CPTCI scale has good internal consistency, test–retest reliability, and convergent and discriminant validity in the acute phase and several months after a trauma (Meiser-Stedman et al., 2009). Internal consistency for the CPTCI was excellent (.88) in the current sample.

How I Coped Under Pressure Scale (HICUPS)

This self-report measure assesses children’s use of coping strategies with regard to a specific event (Ayers, Sandler, West, & Roosa, 1996). The HICUPS has well-established reliability and validity, and has been used with children of a range of different ethnicities and socioeconomic status facing a variety of stressors (Ayers et al., 1996; Ollendick, Langley, Jones, & Kephart, 2001). Internal consistency for each HICUPS subscale was excellent (.83–.89) in the current sample. We administered three HICUPS subscales that correspond to specific coping strategies targeted by the Coping Coach intervention, to assess the extent to which the intervention would increase Positive Cognitive Restructuring and Support Seeking scores and decrease Avoidance Coping scores.

Other Measures

The following measures are also included in the current analyses.

Traumatic Events Screening Inventory – Brief Parent Report Version (TESI-P Brief)

This 12-item measure was administered at baseline, asking parents about the child’s prior exposure to a variety of traumatic events (e.g., disaster, medical event, accident, violence, and physical and sexual abuse) (National Center for PTSD, 1996).

Health & Recovery Questionnaire

This brief questionnaire was administered at 12 weeks to gather standard parent-reported information about help-seeking and services used (including mental health care and informal psychosocial support) after the index medical event. The questionnaire (available upon request) was developed by our team, and has been used in multiple past studies.

Data Analysis

Sample Size Considerations

A primary goal of this initial small RCT was to estimate effect sizes for a later full-scale RCT. Within the constraints of a pilot study, we determined that an enrolled sample of 70, with 60 retained at each follow-up assessment, would provide reasonable power (80%), with alpha set at .05, to detect a clinically meaningful effect size of .50. We enrolled 72 children, and retained 50–58 at each follow-up assessment point (Figure 1).

Analytical Approach

We first used descriptive analyses to examine demographic and medical event characteristics, intervention usage, outcome measures, and potential covariates. All analyses of intervention effects were conducted on an intent-to-treat basis, with all cases included as randomized, regardless of whether the child received the intervention to which she/he was allocated. Primary outcomes were evaluated at 6 and 12 weeks; we also present 18-week outcomes as exploratory. As per the published study protocol (Marsac et al., 2013), we estimated effect sizes (Cohen’s d) for change in PTSS severity, and then examined intervention effects on child health outcomes (PTSS and HRQoL) at 12 weeks and proximal outcomes (i.e., hypothesized mechanisms: appraisals and coping) at 6 weeks, by means of analysis of covariance (ANCOVA). In each ANCOVA, study condition was the grouping factor and the corresponding baseline score for the outcome measure was included as a covariate. To inform potential targeting of the intervention based on initial risk status, we then conducted exploratory analyses of intervention outcomes and estimated effect sizes separately for children whose baseline assessment indicated higher risk for persistent PTSS versus children who were at lower risk.

Results

Intervention Usage

In the immediate intervention group, 35 of 36 participants (97%) used the Coping Coach intervention (one child did not feel well enough to log-in immediately upon randomization, and never subsequently logged in) and 19 (53%) completed the entire intervention at least once. The mean number of logins (sessions of intervention use) was 2.6 (SD = 1.8; range: 1–9); mean time spent on the intervention across all sessions was 52.2 minutes (SD = 36.9; range: 0–199 min); 50% used the intervention within a single day (days from first to last login ranged from 1 to 102). In the wait list group, 28 of 36 (78%) participants completed a 12-week research assessment and were then invited to use the Coping Coach intervention. Of these 28, 19 (68%) used the intervention, and 15 of 28 (54%) completed the entire intervention at least once. The mean number of logins was 2.7 (SD = 2.2; range 0–11); mean time spent on the intervention across all sessions was 51.5 min (SD = 30.7); 41% used the intervention within a single day (days from first to last login ranged from 1 to 39). Over the course of the study, 12 of 72 parents logged in with their own username and password; none completed the entire game.

Analysis of Potential Intervention Effects

Key measures of outcomes (PTSS, HRQoL) and hypothesized mechanisms (appraisals and coping) for each group at each assessment point are summarized in Table II. The top portion of Figure 2 depicts mean PTSS severity for children in each group at each assessment point.

Table II.

Mean Scores for Key Outcomes and Hypothesized Mechanism, by Intervention Condition

Measure	Wait list (Mean, SD)	Immediate intervention (Mean, SD)
Posttraumatic stress symptoms (CPSS)
Baseline	13.1 (7.3)	18.4 (11.8)	t (69) = −2.28; p = .03
6 weeks	12.5 (10.0)	13.0 (12.1)	NS
12 weeks	14.6 (12.0)	13.9 (14.3)	NS
18 weeks	11.0 (10.4)	15.3 (12.5)	NS
Health-related quality of life (PedsQL)
Baseline	82.1 (15.8)	78.0 (17.3)	NS
6 weeks	81.4 (17.5)	75.1 (19.1)	NS
12 weeks	85.1 (15.1)	86.8 (18.3)	NS
18 weeks	90.4 (11.2)	78.5 (23.4)	t (48) = 2.13; p = .04
Maladaptive posttrauma cognitive appraisals (CPTCI)
Baseline	40.5 (11.6)	45.4 (12.1)	NS
6 weeks	43.8 (10.5)	45.9 (10.6)	NS
12 weeks	41.3 (11.5)	39.3 (9.0)	NS
18 weeks	41.1 (11.8)	40.8 (12.6)	NS
Cognitive restructuring coping (HICUPS subscale)
Baseline	35.9 (8.9)	37.9 (9.0)	NS
6 weeks	41.0 (7.3)	41.4 (9.3)	NS
12 weeks	40.2 (8.0)	37.3 (11.8)	NS
18 weeks	40.0 (9.5)	39.7 (8.5)	NS
Avoidance coping (HICUPS subscale)
Baseline	29.3 (7.4)	32.4 (7.3)	NS
6 weeks	31.7 (5.9)	35.4 (7.9)	NS
12 weeks	31.7 (5.7)	30.6 (9.8)	NS
18 weeks	31.4 (6.9)	33.0 (7.6)	NS
Support-seeking coping (HICUPS subscale)
Baseline	22.3 (6.0)	23.3 (6.8)	NS
6 weeks	21.4 (4.0)	22.1 (5.5)	NS
12 weeks	21.9 (4.9)	20.5 (7.3)	NS
18 weeks	22.1 (6.5)	23.2 (6.1)	NS

Note. CPSS = Child PTSD Symptom Scale; PedsQL = Pediatric Quality of Life Inventory; CPTCI = Child Post-Traumatic Cognitions Inventory; HICUPS = How I Coped Under Pressure Scale.

Figure 2.

Mean posttraumatic stress symptom severity at each study assessment point, by intervention condition and baseline risk status. Note. *Child PTSD Symptom Scale (CPSS) score: potential range 0–51, scores ≥15 suggest clinically significant symptoms of posttraumatic stress.

Because our main interest is in the change in PTSS severity over time that might be attributable to the intervention, we calculated effect size (Cohen’s d) as the between-group difference in mean change score from baseline to 6 weeks or to 12 weeks, standardized by the pooled SD of the baseline scores. Standardizing effect size estimates in baseline score units for a widely used measure such as the CPSS can aid interpretability (Cumming, 2012). (The alternative, standardizing by the pooled SD of the change scores, would not change the results substantially, as these SDs were quite similar in the current study.) We observed medium effect sizes in favor of the immediate intervention group for between-group differences in change in PTSS severity from baseline to 6 weeks (d = −.68) and from baseline to 12 weeks (d = −.55). Looking at each group separately, the immediate intervention group showed mean decreases in PTSS severity score from baseline to 6 weeks and baseline to 12 weeks, while the wait list group, on average, was stable from baseline to 6 weeks and showed a slight increase from baseline to 12 weeks. In exploratory analyses, we examined 18-week PTSS outcomes for the wait list group (i.e., including the period in which children in this group were able to use the intervention), and observed decreases in mean PTSS severity score for this group from baseline to 18 weeks, which would translate to a small pre-post effect size (d = −.27 in baseline SD units for that group).

Per our study protocol, we conducted ANCOVA for each outcome of interest with intervention condition as the qualitative factor, and baseline score included as a covariate (Table III). These analyses revealed no statistically significant main effects for intervention condition on PTSS or HRQoL at 12 weeks, nor on trauma-related cognitive appraisals, cognitive restructuring coping, avoidance coping, or social support seeking at 6 weeks. We explored other factors that might moderate intervention effects (e.g., age, gender, race/ethnicity, low family income, prior trauma exposure), by including these factors one at a time as covariates in each ANCOVA. None of these factors modified the ANCOVA results.

Table III.

Results of ANCOVA Comparing Immediate Intervention Group With Wait List Group on Key Study Outcomes, Controlling for the Corresponding Baseline Score

Dependent variable	Independent variables
	Baseline score	Intervention condition
12-week child health outcomes
Posttraumatic stress symptoms (CPSS)	F(1, 47) = 35.51, p = .000	F(1, 47) = 2.24, p = .14
Health-related quality of life (PedsQL)	F(1, 41) = 17.11, p = .000	F(1, 41) = 0.41, p = .53
6-week proximal outcomes (mechanisms)
Trauma-related appraisals (CPTCI)	F(1, 55) = 7.68, p = .008	F(1, 55) = 0.001, p = .98
Cognitive restructuring coping (HICUPS)	F(1, 55) = 16.42, p = .000	F(1, 55) = 0.09, p = .77
Avoidance coping (HICUPS)	F(1, 55) = 13.47, p = .001	F(1, 55) = 1.14, p = .29
Social support seeking (HICUPS)	F(1, 55) = 12.04, p = .001	F(1, 55) = 0.003, p = .96

Additional Analyses

At-Risk Versus Not-at-Risk

Our team’s prior data from more than 400 children across several studies indicate that a score of 15 or higher on the CPSS in the acute aftermath of trauma is a good indicator not only of clinically significant current symptoms, but of risk for persistent PTSS months later. A total of 36 children across groups had a baseline CPSS score ≥ 15, including 22 (61%) of the immediate intervention group and 14 (39%) of the wait list group (χ² = 3.56, df = 1, p = .06). In exploratory analyses, using baseline CPSS score ≥ 15 as an indicator of later PTSS risk, we examined potential intervention effects separately among the 36 participants “at-risk” and then among the 36 participants “not-at-risk”. The lower portion of Figure 2 depicts mean PTSS severity for those at-risk versus not-at-risk, by intervention condition. Effect sizes (Cohen’s d) were calculated, as above, as the standardized between-group difference in mean change score. Among those at-risk, between-intervention-group effect sizes were medium to large for baseline to 6 weeks (d = −.84) and for baseline to 12 weeks (d = −.68). For children not-at-risk, these effect sizes were small for baseline to 6 weeks (d = −.15) and for baseline to 12 weeks (d = −.24).

Seeking Professional Assistance

At 12 weeks (N = 50), very few parents reported that they had sought assistance for their child from a mental health professional (7 out of 50, 14%) or from their pediatrician or family doctor (9 out of 50, 18%) for concerns related to coping with the medical event. There was no significant difference between the immediate intervention group and wait list group in the proportion who sought help from either of these sources.

Discussion

This pilot randomized trial provides preliminary evidence that a novel interactive, game-like, Web-based preventive intervention for school-age children exposed to acute trauma is feasible to deliver and engaging for children, and that such an intervention could have an effect in preventing persistent posttraumatic stress. When offered the self-directed Coping Coach intervention soon after a potentially traumatic medical event, children engaged in the intervention (averaging nearly an hour interacting with its activities), just over half of them completed it, and there were no notable technical difficulties in children’s use of this online game. Effect sizes were promising regarding the intervention’s potential impact in reducing PTSS from baseline levels (i.e., preventing persistent symptoms), especially within the first 6 weeks post-event. This preliminary finding of intervention effects is tempered by the ANCOVA results. Observed coefficients for PTSS severity were in the hypothesized direction, but the ANCOVA did not indicate statistically significant group differences in 12-week PTSS severity, even controlling for baseline symptoms.

This study adds to the literature on early interventions to prevent persistent posttraumatic stress after acute trauma. Prior studies have addressed the impact of psychoeducational interventions delivered in-person or online, and of brief in-person interventions delivered by trained personnel in medical or other settings (Kassam-Adams, 2014; Kramer & Landolt, 2011). Only one prior study has documented an impact on PTSS, in a four-session face-to-face parent–child intervention delivered to children and youth with a variety of types of recent trauma exposure (Berkowitz, Stover, & Marans, 2011). Some in-person interventions, such as “debriefing” models, have not shown any effects (Kassam-Adams, 2014); the Coping Coach intervention is distinct from these models not only in its mode of delivery but also in its targeted mechanisms of action. The only other online preventive intervention investigated in a randomized trial is the Kids and Accidents psychoeducational Web site, which demonstrated an impact on anxiety symptoms and potential impact on PTSS (among children at higher risk), but had technical limitations precluding assessment of children’s actual usage of the Web site (Cox & Kenardy, 2010). The current pilot trial provides new information about children’s actual usage of a self-directed online preventive intervention in the aftermath of acute trauma, and supports ongoing development of online or mobile interventions to interrupt the development of persistent PTSS.

This pilot RCT included children with a particular type of acute traumatic experience – acute medical events. It is possible that the effect of the intervention is influenced by the extent to which the child user is able to identify his or her own experiences as similar to those of the game characters. However, we believe that including a range of stories can help children identify their own reactions to trauma as normative and potentially manageable, because they see them embodied in a range of different experiences. This is supported by results from the development and pilot testing phases of this intervention: 73% of a US pilot sample and 71% of an Australian pilot sample (all were children with acute medical events) reported the “Coping Coach characters are kids like me” (Marsac et al., 2015).

Given the difficulty of reaching children exposed to medical events and other acute trauma with in-person or online interventions, it would be useful to know whether preventive interventions for PTSS after acute trauma can be flexible as to timing. Depending on the nature of the event and the child’s physical/medical condition, some children may not be ready or able to engage in an online intervention immediately post-event. We interpret this study’s results regarding the period from 12 to 18 weeks with caution. Nevertheless, the uptake of the intervention by two-thirds of the children who were offered it after 12 weeks, and the observed PTSS severity changes in this wait list group between baseline and 18 weeks, are both consistent with the potential feasibility and effectiveness of a later intervention time frame. Future studies should incorporate designs that can more fully address questions regarding the time frame for effective preventive intervention.

Our exploratory findings regarding the impact of the intervention for at-risk versus not-at-risk participants suggest that any effect of the intervention was seen primarily in those children most at risk. Future studies should either target only children at higher risk, or stratify randomization so that higher- and lower-risk participants are distributed equally among conditions.

We did not observe hypothesized intervention effects on children’s HRQoL. Over the course of the study, the wait list group showed a gradual increase (toward better HRQoL), while the immediate intervention group PedsQL scores were similar at baseline and 18 weeks. One reason for this different pattern of HRQoL findings, compared to PTSS findings, may be that HRQoL was rated by parents rather than by children themselves. The original study design called for self-reported HRQoL, but based on initial feasibility testing with children, we opted for parent report to reduce child participant burden in the baseline assessment. Future studies could incorporate a brief child self-report measure of HRQoL.

Findings from this study do not provide evidence that the intervention as currently designed and utilized was able to shift maladaptive trauma-related appraisals or specific adaptive or maladaptive coping strategies that are hypothesized as mechanisms of action by which the intervention can reduce PTSS. Prior to a full-scale trial, our team will explore adaptations of intervention activities and intervention structure to increase engagement, repeated use of the intervention, and skills practice – as a means to increase the potency of the intervention to address specific appraisal and coping skills targets. In addition, we will explore additional measurement strategies for appraisals and coping strategies that may be more sensitive to intervention-induced changes.

Limitations

It is unfortunate that the intervention groups in this study differed in initial PTSS severity, despite random assignment to condition. An alternate explanation for the effect sizes we observed is that children with higher symptoms showed a greater degree of early natural recovery. However, examining intervention effects within only the 36 study participants who reported more severe baseline PTSS (and correspondingly higher risk for ongoing PTSS) suggests that this is unlikely. Even among those with initially higher symptoms, we observed a medium to large effect size at 6 and at 12 weeks in favor of the immediate intervention group.

As planned, this pilot study did not have sufficient power to detect statistically significant group differences smaller than 0.5 standard deviations in key measures. While our emphasis in this pilot work was on estimating effect sizes rather than solely on detecting statistically significant group differences, smaller effects may have been missed. Lower follow-up rates among the immediate intervention group may have hampered our ability to detect group differences at 6 and 12 weeks. A unique feature of online interventions that can be accessed repeatedly by child users is the possibility of incorporating brief assessments within the game itself. Building a basic, recurring assessment of symptoms, appraisals, and coping strategies directly into the online intervention game structure, and encouraging children’s repeated use of the game, could help address the challenges of collecting follow-up assessment data.

Implications for Research and Practice/Implementation

The number of children exposed to acute, potentially traumatic events is enormous, and highlights the need to consider the population-level impact of our preventive intervention approaches. Addressing this need will likely require a continuum of stepped care options that begins with low-intensity technology-based interventions that do not require direct professional contact (Comer & Barlow, 2014). A preventive intervention for traumatic stress that produces smaller effect sizes but is accessible to more individuals may achieve wider reach and public health impact than an intervention that has a larger effect but is hard to access (Zatzick et al., 2009). These approaches are consistent with recent calls for the field of mental health to develop a larger portfolio of delivery methods and intervention models, including those that can yield even small effects in very large numbers of people (Kazdin & Blase, 2011).

For all Web-based interventions (as for in-person interventions), issues of uptake, usage, and participant dropout are of concern; Web-based interventions offer the advantage of automated tracking of usage details so that naturalistic usage can be better understood and optimized. The current results support the feasibility of successful engagement of 8–12-year-old children soon after a traumatic event, but additional work is needed to optimize engagement and “stickiness” of the intervention to promote repeated use and skills practice. The game-like format of this intervention allows flexibility in optimizing incentives (e.g., points and rewards) within the game structure, and its modular format allows additional content or activities to be added or “unlocked” as a user progresses through the intervention.

The public health impact of an online intervention would be increased if it can be made widely available with minimal need for initial screening. The extent to which an online intervention can be disseminated broadly to trauma-exposed children will depend in part on (a) whether it is suitable for all children or only for those at higher risk, and (b) whether it must be introduced in a specific critical period for PTSS development or can be efficacious across a wider time frame post-trauma. Future research should test universal versus targeted implementation models in children with medical events and other acute trauma, and utilize designs that examine the flexibility of the time frame during which preventive intervention could be useful.

Funding

Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) grant R21HD069832, and a Mentored Career Award from the National Institute of Mental Health (NIMH) (K23MH093618, to M.L.M.).

Conflicts of interest: None declared.