A comparative effectiveness trial tested 2 parent-based interventions in improving the psychosocial recovery of hospitalized injured children: (1) Link for Injured Kids (Link), a program of psychological first aid in which parents are taught motivational interviewing and stress-screening skills, and (2) Trauma Education, based on an informational booklet about trauma and its impacts and resources.
A randomized controlled trial was conducted in 4 children’s hospitals in the Midwestern United States. Children aged 10 to 17 years admitted for an unintentional injury and a parent were recruited and randomly assigned to Link or Trauma Education. Parents and children completed questionnaires at baseline, 6 weeks, 3 months, and 6 months posthospitalization. Using an intent-to-treat analysis, changes in child-reported posttraumatic stress symptoms, depression, quality of life, and child behaviors were compared between intervention groups.
Of 795 injured children, 314 children and their parents were enrolled into the study (40%). Link and Trauma Education was associated with improved symptoms of posttraumatic stress, depression, and pediatric quality of life at similar rates over time. However, unlike those in Trauma Education, children in the Link group had notable improvement of child emotional behaviors and mild improvement of conduct and peer behaviors. Compared with Trauma Education, Link was also associated with improved peer behaviors in rural children.
Although children in both programs had reduced posttrauma symptoms over time, Link children, whose parents were trained in communication and referral skills, exhibited a greater reduction in problem behaviors.
Up to a third of hospitalized injured children develop posttraumatic stress, with little support for their psychosocial recovery.
Two parent-based programs (Link for Injured Kids and Trauma Education) were comparably associated with reduced psychosocial stress and increased pediatric quality of life over time. Link, which involves motivational interviewing, was furthermore associated with improved child behaviors.
Unintentional traumatic injury, the leading cause of death and morbidity among children in the United States, results in >225 000 hospitalizations and 9 million emergency department visits each year.1,2 With advancements in medical care, a majority of children survive their physical injuries. However, 20% to 36% develop posttraumatic stress symptoms (PTSSs), 17% develop depressive symptoms 1 to 2 months after an injury, and up to 30% have PTSSs and 4% to 11% experience depressive symptoms that may persist after 3 to 6 months.3–7 Untreated PTSSs may progress into full-blown posttraumatic stress disorder if symptoms persist for >1 month. Intervening early on PTSSs at the interpersonal level with parents or guardians could enhance children’s emotional recovery after injury.
Parents or guardians, if provided adequate skills, may facilitate 3 processes known to promote resiliency and support recovery posttrauma: social support, meaning making, and reestablishing normalcy.8 The Child and Family Traumatic Stress Intervention provides family social support and coping skills by encouraging parent-child communication and parent-supported behavioral skills.9 Cognitive behavioral intervention approaches have been effectively delivered to families via the internet10 and as a brief face-to-face or telehealth intervention.11 A drawback of these programs is that they involve formal health care or mental health providers, who have limited ongoing contact with patients after discharge, especially in rural settings.
Psychoeducation and motivational interviewing (MI), in contrast, may be delivered directly to injured children after discharge by parents or guardians themselves. MI is a person-centered communication style that taps into one’s intrinsic strength to foster positive behavioral change.12,13 Psychoeducational interventions, rooted in MI and delivered postinjury, have been shown to improve behaviors,14 reduce symptoms of anxiety,15 and increase use of coping strategies by parents with their injured child.16 One school-based intervention that incorporated MI with education about child stress was associated with increased connectedness between adults and traumatized children and reduced PTSSs and depressive symptoms.17 A parent-based version of this program was developed for injured children in pediatric hospital settings but has yet to be tested for its effectiveness.18
We conducted a comparative effectiveness trial to test a novel intervention based on MI against a psychoeducational program with some evidence of effectiveness. The 2 interventions tested were (1) Link for Injured Kids (hereinafter referred to as Link), a psychological first-aid program that integrates MI and stress-screening strategies, and (2) “So You’ve Been in An Accident” (hereinafter referred to as Trauma Education), a trauma psychoeducational program. We compared the effect of these 2 programs on the progression of PTSSs and depressive symptoms and pediatric quality of life (PedsQL) in children admitted for unintentional traumatic injury in 4 children’s hospitals in the Midwest.
Methods
Participants
The study population was composed of children aged 10 to 17 years who were admitted for a new unintentional injury into 4 pediatric trauma centers (the University of Iowa Stead Family Children’s Hospital, Blank Children’s Hospital, Children’s Mercy Kansas City, and Children’s Minnesota). Intentional injuries; complications of an unintentional injury; diagnoses of a severe intellectual, neurologic impairment (including those resulting from the traumatic injury); psychosis; suicide attempts in the last year; law enforcement involvement; pregnancy; residential treatment; foster care; and non–English-speaking patients were excluded because interventions would require tailoring and translation beyond the scope of this study.
Random Assignment and Masking
Dyads, composed of an injured child and 1 parent, gave consent and were recruited during their hospital stay and block randomly assigned into 1 of 2 treatment groups: Link or Trauma Education. Recruitment personnel opened envelopes containing computer-generated random assignments after participants’ enrollment into the study.
Procedures
Link for Injured Kids
Parents received both video and a brief live training in MI skills (ie, open questions, reflective listening skills, and linking to additional resources) to use when communicating with their child and the use of the Kessler-6 screening scale (K-6)19 to identify symptoms of stress in their child. A research team member trained in MI conducted the training. After the training, parents were given (1) A Link pocket card and separate brochure, both containing information about basic Link skills and a copy of the K-6 screener, (2) a magnet containing a list of medical resources and/or crisis hotlines, and (3) a copy of the Link instructional DVD. Training sessions lasted between 20 and 45 minutes, reflecting the varied time needed by parents to absorb material. At 6 weeks postdischarge, a telephone booster session was conducted with the Link-trained parent to review content from the initial training and to assess the use and helpfulness of the Link skills.
Trauma Education
The study team provided parents the educational booklet So Your Child Has Been in an Accident… A Book for Parents about Dealing with Accidents. The booklet contained information about (1) children’s normal reactions to being injured, (2) how long reactions last, (3) what parents can do to help the child, (4) coping strategies for parents, and (5) a list of medical resources and/or crisis hotlines.20 At 6 weeks postdischarge, a telephone survey was conducted with parents to review intervention materials and assess the use and helpfulness of the booklet.
Data Collection and Sources
Parent and child participants each separately completed 3 follow-up questionnaires via US mail or Internet over the course of the study (6 weeks, 3 months, and 6 months postenrollment). Participants were compensated up to $200 per dyad for the completion of all questionnaires.
Outcomes
Baseline and follow-up questionnaires assessed two primary outcomes: child symptoms of posttraumatic stress using the Child PTSS Symptom Scale (CPSS)21,22 and child symptoms of depression using the Center for Epidemiologic Studies Depression scale (CES-D).23 Secondary outcomes were nonspecific psychological distress (K-6 screener),19 PedsQL,24 and childhood emotional/behavior problems (parent- and child self-reported) by using the Strengths and Difficulties Questionnaire (SDQ).25
Baseline questionnaires also contained basic demographics (sex, age, grade, race and/or ethnicity, education, income, and employment) and residence zip code, which was used to classify rural and/or urban residence.
Link training sessions offered to parents were audio recorded, and 20% of audio files were reviewed by using an adapted version of the Behavioral Change Counseling Index for adherence of trainers to the MI style of communication.26 Nine constructs of MI were scored on a continuous scale of 0 to 4 (in which 0 = not at all and 4 = a great extent).
Statistical Analysis
Generalized linear mixed models were constructed to estimate the effect of the interventions on each outcome of interest: PTSSs, depression, quality of life, and strengths and difficulties. The generalized linear mixed models were based on a γ distribution because of the right-skewed empirical distributions of the outcomes and used a log link function. Models included a covariate for group membership (Link or Trauma Education), as well as the interaction between group membership and time. Because surveys were not completed at exact time points (ie, 6 weeks, 3 months, and 6 months), time was represented in models as the number of weeks postbaseline. Baseline measurements were excluded from final models as their inclusion induced nonlinearity, and sensitivity analyses indicated no meaningful difference in results with their inclusion in models. Intervention effects were quantified as (1) the multiplicative change in the mean outcome for 1 of the intervention groups over a 3-month time period (values of <1 are indicative of a reduction over time) and (2) the multiplicative change in the ratio of the 2 mean outcomes over a 3-month period, in which the mean for the Link group is represented in the numerator and the mean for the Trauma Education group in the denominator. (For intervention effect 2, values of <1 are indicative of greater improvement in the Link group.) The multiplicative mean change in the outcome may be understood as the percentage change in the mean over the specified span of time. Analyses comparing the specification of subject-level and hierarchical (site- and subject-level) random effects were also performed, indicating negligible within-site correlation (on the order of 1e–12). Final models included only subject-level random effects.
A subanalysis of a 3-way interaction between intervention, time, and rural and/or urban status was used to examine heterogeneity of effects for rural and/or urban patients. A post hoc analysis was used to compare the effects of our interventions among subjects with elevated PTSSs (≥11),21 depression (≥12,)23 SDQ27 (overall ≥14; emotional subscale ≥4, conduct subscale ≥3, hyperactivity subscale ≥6, peer subscale ≥4, and prosocial subscale ≥4), K-628 score (≥13), and PedsQL24 levels (≥12.5 [the sample median]). Multiple imputation with 300 iterations (200 burn-in) and Jeffreys prior was used to impute data missing from partially completed surveys. Missing data accounted for 2% to 8% of all response data. Sensitivity analyses were used to compare results from the imputed data to those obtained through a complete case analysis, and showed negligible differences.
Study procedures were approved by the University of Iowa Institutional Review Board (IRB) (IRB ID no. 201111728, reliance IRB no. 14110518), Blank Children’s Hospital (IRB ID no. IM2014-029) and Children’s Minnesota (IRB ID no. 1602-010). This study is registered at www.clinicaltrials.gov (identifier: NCT02323204, unique protocol identifier: 201111728).
Results
Study Participants
Of the 1688 injured patients aged 10 to 17 years admitted into 1 of the 4 children’s hospitals, 893 children were excluded because they did not meet study criteria (Fig 1). Of the 795 eligible for inclusion, 314 children and their parents enrolled into the study and were randomly assigned into Link (n = 155 dyads, 49.4%) or Trauma Education (n =159, 50.6%). During hospitalization, Link parents were trained in the intervention, whereas the Trauma Education parents received program materials. During follow-up, 72% of Link parents and 77% of Trauma Education parents completed booster sessions. Fidelity to MI communication styles during initial and booster training sessions was found to be high, with a mean Behavioral Change Counseling Index score of 25.8 and a range of 13 to 34 of 36 possible points.
Consort flow diagram. a Eligible patients often met multiple reasons for not being recruited, and 25 were counted more than once across categories. b Ineligible patients often met multiple exclusion criteria (significant psychiatric history, intentional injury, law enforcement involvement, pregnant, language barrier, unable to complete study procedures, death, foster care, short stay, admitted for complication, injury occurred >2 weeks before admission, and/or patient turned 18 years old).
Consort flow diagram. a Eligible patients often met multiple reasons for not being recruited, and 25 were counted more than once across categories. b Ineligible patients often met multiple exclusion criteria (significant psychiatric history, intentional injury, law enforcement involvement, pregnant, language barrier, unable to complete study procedures, death, foster care, short stay, admitted for complication, injury occurred >2 weeks before admission, and/or patient turned 18 years old).
For the Link arm, follow-up questionnaires were completed by 75.3% of parents and 71.7% of children at 6 weeks, 67.8% of parents and 62.8% of children at 3 months, and 65.8% of parents and 66.2% of children at 6 months. For the Trauma Education arm, completion rates were 76.3% for parents and 72.7% for children at 6 weeks, 71.8% for parents and 73.4% for children at 3 months, and 71.8% for parents and 73.4% for children at 6 months. Completion rates of Link and Trauma Education parents were not statistically different from each other.
Link and Trauma Education groups exhibited similar characteristics at baseline (Tables 1 and 2). The mean age (SD) was 14.04 (2.16) years among Link children and 13.82 (2.57) years among children in the Trauma Education group. In both Link and Trauma Education, children were mostly male, but parents were mostly female. The majority of participants identified as white and non-Hispanic and/or non-Latino, and >60% of families resided in urban locations. At baseline, average PTSS and depression scores were similar between the intervention groups. In both groups, ∼30% of children had acute PTSSs and 25% had somewhat or very elevated depressive symptoms.
Demographics and Outcome Measures at Baseline (N = 299)
| Link, n = 145 | Trauma Education, n = 154 | |
|---|---|---|
| Age, y | ||
| Mean (SD) | 14.04 (2.16) | 13.82 (2.57) |
| Range (minimum, maximum) | 8 (10, 18) | 9 (9, 18) |
| Sex, n (%) | ||
| Female | 52 (35.9) | 60 (39.0) |
| Male | 93 (64.1) | 94 (61.0) |
| Ethnicity, n (%) | ||
| Hispanic or Latino | 9 (6.2) | 14 (9.1) |
| Not Hispanic or Latino | 136 (93.8) | 139 (90.3) |
| Missing | 0 (0) | 1 (0.7) |
| Race, n (%) | ||
| American Indian or Alaskan Native | 2 (1.4) | 0 (0) |
| Asian American or Pacific Islander | 2 (1.4) | 3 (2.0) |
| African American | 6 (4.1) | 7 (4.6) |
| Multirace | 5 (3.5) | 5 (3.3) |
| Other | 1 (0.7) | 1 (0.7) |
| White | 125 (86.2) | 132 (85.7) |
| Missing | 4 (2.8) | 6 (3.9) |
| Urbanicity, n (%) | ||
| Rural | 51 (35.2) | 58 (37.7) |
| Urban | 94 (64.8) | 95 (61.7) |
| Missing | 0 (0) | 1 (0.6) |
| Link, n = 145 | Trauma Education, n = 154 | |
|---|---|---|
| Age, y | ||
| Mean (SD) | 14.04 (2.16) | 13.82 (2.57) |
| Range (minimum, maximum) | 8 (10, 18) | 9 (9, 18) |
| Sex, n (%) | ||
| Female | 52 (35.9) | 60 (39.0) |
| Male | 93 (64.1) | 94 (61.0) |
| Ethnicity, n (%) | ||
| Hispanic or Latino | 9 (6.2) | 14 (9.1) |
| Not Hispanic or Latino | 136 (93.8) | 139 (90.3) |
| Missing | 0 (0) | 1 (0.7) |
| Race, n (%) | ||
| American Indian or Alaskan Native | 2 (1.4) | 0 (0) |
| Asian American or Pacific Islander | 2 (1.4) | 3 (2.0) |
| African American | 6 (4.1) | 7 (4.6) |
| Multirace | 5 (3.5) | 5 (3.3) |
| Other | 1 (0.7) | 1 (0.7) |
| White | 125 (86.2) | 132 (85.7) |
| Missing | 4 (2.8) | 6 (3.9) |
| Urbanicity, n (%) | ||
| Rural | 51 (35.2) | 58 (37.7) |
| Urban | 94 (64.8) | 95 (61.7) |
| Missing | 0 (0) | 1 (0.6) |
Demographics and Outcome Measures at Baseline, Continued
| Overall Analyzed | Link Children, n = 145 | Trauma Education Children, n = 154 |
|---|---|---|
| PTSSs (CPSS), mean (SD)a | 8.97 (7.91) | 8.95 (8.31) |
| Depressive symptoms (CES-D), mean (SD)a | 8.51 (6.08) | 7.98 (5.79) |
| Quality of life (PedsQL), mean (SD)a | 88.94 (11.13) | 86.53 (13.32) |
| Psychological distress (K-6), mean (SD)a | 5.24 (4.26) | 5.11 (4.72) |
| Child emotional/behavioral problems, mean (SD)a | ||
| SDQ | ||
| Total | 7.24 (5.75) | 8.13 (6.32) |
| Emotional symptoms | 1.58 (2.01) | 1.76 (2.20) |
| Conduct problems | 1.47 (1.68) | 1.55 (1.54) |
| Hyperactivity | 2.91 (2.33) | 3.08 (2.33) |
| Peer relationship problems | 1.39 (1.53) | 1.75 (1.87) |
| Prosocial behavior | 8.24 (1.88) | 8.20 (1.88) |
| Minimal depressive symptoms, n (%)b | 94 (75.81) | 100 (75.76) |
| Somewhat elevated depressive symptoms, n (%)b | 25 (20.16) | 28 (21.21) |
| Very elevated depressive symptoms, n (%)b | 5 (4.03) | 4 (3.03) |
| Overall Analyzed | Link Children, n = 145 | Trauma Education Children, n = 154 |
|---|---|---|
| PTSSs (CPSS), mean (SD)a | 8.97 (7.91) | 8.95 (8.31) |
| Depressive symptoms (CES-D), mean (SD)a | 8.51 (6.08) | 7.98 (5.79) |
| Quality of life (PedsQL), mean (SD)a | 88.94 (11.13) | 86.53 (13.32) |
| Psychological distress (K-6), mean (SD)a | 5.24 (4.26) | 5.11 (4.72) |
| Child emotional/behavioral problems, mean (SD)a | ||
| SDQ | ||
| Total | 7.24 (5.75) | 8.13 (6.32) |
| Emotional symptoms | 1.58 (2.01) | 1.76 (2.20) |
| Conduct problems | 1.47 (1.68) | 1.55 (1.54) |
| Hyperactivity | 2.91 (2.33) | 3.08 (2.33) |
| Peer relationship problems | 1.39 (1.53) | 1.75 (1.87) |
| Prosocial behavior | 8.24 (1.88) | 8.20 (1.88) |
| Minimal depressive symptoms, n (%)b | 94 (75.81) | 100 (75.76) |
| Somewhat elevated depressive symptoms, n (%)b | 25 (20.16) | 28 (21.21) |
| Very elevated depressive symptoms, n (%)b | 5 (4.03) | 4 (3.03) |
Reported means (SDs) are based on the following sample sizes (sizes listed in order of the corresponding groups as they appear in the table columns): PTSSs (133, 136); depressive symptoms (124, 132); QoL (119, 121); psychological distress (141, 150); SDQ (128, 134); SDQ emotional (140, 148); SDQ conduct (137, 148); SDQ hyperactivity (138, 145); SDQ peer (137, 144); SDQ prosocial (143, 146). Reduced sample sizes are a result of unreported (missing) data.
Minimal depressive symptoms is defined as having a depression score of <12; somewhat elevated depressive symptoms corresponds to having a score of ≥12 and <21; very elevated depressive symptoms corresponds to a score of ≥21.
Outcomes
There was a general decrease in the prevalence of elevated posttraumatic stress and depressive symptoms in both Link and Trauma Education children (6 weeks: PTSSs = 34%, CES-D = 18%; 3 months: PTSSs = 26%, CES-D = 16%; 6 months: PTSSs = 22%, CES-D = 17%). Both programs were associated with reduced child PTSSs and depressive symptoms and nonspecific stress, with no evidence of a differential mean effect by intervention (PTSSs = 0.963, 95% confidence interval [CI]: 0.83 to 1.12 [P = .63]; depression = 1.019 [0.89 to 1.17] [P = .79]; and K-6 = 1.029 [0.90 to 1.17] [P = .68]) (Table 3). The mean reductions over a 3-month period in PTSSs, depressive symptoms and nonspecific stress were 21.1%, 11.3%, and 17.2%, respectively, for Link children and 24%, 9.6%, and 14.8%, respectively, for Trauma Education children. Both interventions were also associated with improved quality of life over time (Link: 0.642; Trauma Education: 0.663). For SDQ, a measure of problematic childhood emotional conduct behaviors, Link trended toward improvement over time for all constructs (emotional, conduct, peer, and prosocial), whereas Trauma Education showed no overall effect (Link: 0.925 [0.86 to 1.00] versus Trauma Education: 0.999 [0.93 to 1.07]). Link was notably associated with reduced problematic emotional behaviors (SDQ emotional 0.894 [0.83 to 0.97] versus 0.987 [0.92 to 1.06]), although this difference was marginal. For problematic conduct and peer behavior, the Link group had also slighter greater mean improvement over time, as compared with the Trauma Education group (SDQ conduct [0.914 (0.84 to 0.99)]; P = .03 and SDQ peer (0.899 [0.82 to 0.99)]; P = .03). Supplemental Table 4 shows parameter estimates for all models, including estimates for the intercept, main effects, and interactions. In our post hoc analyses, subjects with elevated baseline levels showed larger effects relative to those displayed in Table 3, but additional differential effects between interventions were not found (see Supplemental Table 5).
Estimated Time Effects by Intervention, Child-Reported Outcomes
| Outcome | Trauma Education (n = 131) | Link (n = 117) | Ratio (Link/Trauma Education) |
|---|---|---|---|
| PTSSs (CPSS) | 0.789 (0.71–0.88)* | 0.760 (0.68–0.85)* | 0.963 (0.83–1.12) |
| Depressive symptoms (CES-D) | 0.887 (0.81–0.97)* | 0.904 (0.82–1.00) | 1.019 (0.89–1.17) |
| Quality of life (PedsQL) | 0.663 (0.59–0.75)* | 0.642 (0.56–0.73)* | 0.968 (0.81–1.16) |
| Psychological distress (K-6) | 0.828 (0.76–0.91)* | 0.852 (0.77–0.94)* | 1.029 (0.90–1.17) |
| Child behaviors (SDQ) | |||
| Total | 0.999 (0.93–1.07) | 0.925 (0.86–1.00)* | 0.926 (0.83–1.03) |
| Emotional | 0.987 (0.92–1.06) | 0.894 (0.83–0.97)* | 0.906 (0.82–1.01) |
| Conduct | 1.027 (0.97–1.09) | 0.942 (0.88–1.00) | 0.914 (0.84–0.99)* |
| Hyperactivity | 0.987 (0.92–1.05) | 0.986 (0.92–1.06) | 1.000 (0.91–1.10) |
| Peer | 1.074 (1.00–1.15)* | 0.966 (0.90–1.04) | 0.899 (0.82–0.99)* |
| Prosocial | 0.960 (0.89–1.03) | 0.930 (0.86–1.01) | 0.969 (0.87–1.08) |
| Outcome | Trauma Education (n = 131) | Link (n = 117) | Ratio (Link/Trauma Education) |
|---|---|---|---|
| PTSSs (CPSS) | 0.789 (0.71–0.88)* | 0.760 (0.68–0.85)* | 0.963 (0.83–1.12) |
| Depressive symptoms (CES-D) | 0.887 (0.81–0.97)* | 0.904 (0.82–1.00) | 1.019 (0.89–1.17) |
| Quality of life (PedsQL) | 0.663 (0.59–0.75)* | 0.642 (0.56–0.73)* | 0.968 (0.81–1.16) |
| Psychological distress (K-6) | 0.828 (0.76–0.91)* | 0.852 (0.77–0.94)* | 1.029 (0.90–1.17) |
| Child behaviors (SDQ) | |||
| Total | 0.999 (0.93–1.07) | 0.925 (0.86–1.00)* | 0.926 (0.83–1.03) |
| Emotional | 0.987 (0.92–1.06) | 0.894 (0.83–0.97)* | 0.906 (0.82–1.01) |
| Conduct | 1.027 (0.97–1.09) | 0.942 (0.88–1.00) | 0.914 (0.84–0.99)* |
| Hyperactivity | 0.987 (0.92–1.05) | 0.986 (0.92–1.06) | 1.000 (0.91–1.10) |
| Peer | 1.074 (1.00–1.15)* | 0.966 (0.90–1.04) | 0.899 (0.82–0.99)* |
| Prosocial | 0.960 (0.89–1.03) | 0.930 (0.86–1.01) | 0.969 (0.87–1.08) |
Results shown are based on the imputed data. Provided estimates for each intervention group reflect the multiplicative change in mean outcome per 3 months. Values of <1 are indicative of a reduction over time. Ranges presented in parentheticals are 95% CIs. The ratio (Link/Trauma Education) column provides the mean ratios of the per-3-month treatment effect between Link and Trauma Education (eg, 0.760 / 0.789 = 0.963). Values of <1 suggest greater improvement in the Link group. The above are based on the following model: log(E(Y) = α+β1(Intervention = Link)+β2Time+β3Time*(Intervention = Link), in which Y follows a γ distribution. See Supplemental Table 5 for corresponding model estimates.
Significant at α = 0.05.
Figure 2 displays the model-predicted trajectories for each intervention group. For depressive symptoms and PTSSs, a gradual reduction in scores is observed with each intervention, and the predicted lines are mostly parallel. The trends for SDQ conduct and SDQ peer demonstrate a gradual increase in score for Trauma Education (which is indicative of a worsening in condition), whereas Link exhibits a decreasing (improving) trajectory. Results were generally consistent between parent-reported and child-reported data.
Mean trajectories of selected child-reported posttraumatic stress and depressive symptoms and conduct and peer relationship problems, comparing Link and Trauma Education interventions.
Mean trajectories of selected child-reported posttraumatic stress and depressive symptoms and conduct and peer relationship problems, comparing Link and Trauma Education interventions.
Heterogeneity of Treatment Effect: Rural and/or Urban Status
For child-reported data, the 3-way interaction term was significant only for the SDQ peer outcome (P value = .04). The ratio of the 3-month estimated mean reduction (Link to Trauma Education) among children in rural locations was 0.78, whereas the estimated mean reduction among urban-residing children was 0.96, suggesting greater efficacy of Link in rural settings. For the parent data, there was no evidence in support of a differential rural and/or urban treatment effect.
Discussion
Approximately 25% to 30% of hospitalized injured children in our sample were symptomatic for posttraumatic stress or depression, which is comparable with other samples of injured youth.7 We evaluated 2 parent-based interventions of different intensities in improving the psychological recovery of hospitalized, injured children. Link required ∼20 to 45 minutes to train parents in MI techniques, impacts of trauma, screening for stress, and linkage to services. Trauma Education is a low–resource intensive program composed of reading materials about trauma and its effects and linkage to services. Overall, both Link and Trauma Education were found to be similarly associated with reduced symptoms of child posttraumatic stress, depression, and nonspecific stress, as well as improved child quality of life. Link was also modestly associated with improved emotional, peer, and conduct behaviors; in contrast, Trauma Education was not.
The Link and Trauma Education groups’ reduced symptoms of psychological harm over time were noteworthy. When translated to a 6-month follow-up period, these mean reductions were substantial, with >18% and 40% of reduced symptoms of depression and posttraumatic stress, respectively. Findings are consistent with the reduced symptoms of stress resulting from previous use of Link in a school setting17 and of Trauma Education in a hospital setting.18 Without intervention, up to 30% of children have PTSSs 4 to 6 months after a hospitalized injury.5
The improvements in emotional, peer, and conduct behaviors observed in Link children were modest but not unexpected. At baseline, participants on average were within the US normative range for low problem behaviors, leading to a ceiling effect in this sample.29 Yet, children whose parents received Link still had a 15% mean reduction in problem behaviors over a 6-month period, whereas children whose parents received Trauma Education had no meaningful improvements overall. For peer, conduct, and emotional behaviors, Link demonstrated 6-month mean reductions of 6.8%, 11.6%, and 22%, respectively. Findings support the possible wide-reaching utility of MI-based interventions like Link in improving outcomes through parenting communication skills.
Notably, children who received Trauma Education showed an unexpected slight increase in peer behavioral problems. It is unlikely that Trauma Education increases behavioral issues, but rather this pattern reflects the effects of untreated trauma.30 The traumatic roots of problem behaviors have been well documented.31,32 We recommend studying Link as a potential intervention for children who exhibit significant problem behaviors caused or exacerbated by trauma.
Study Limitations
Study findings are limited to the clinic populations from the 4 children’s hospitals in the Midwest. We also cannot generalize our findings to less severely injured patients from emergency department or outpatient settings. Although mostly white, our sample had a reasonable representation of rural patients (35%), which allowed for testing of intervention effectiveness by rurality.
Across follow-up, retention rates ranged from 66% to 76%, which are comparable to rates in intervention studies of pediatric trauma patients.9,15,20 Despite this limitation, no appreciable differences in characteristics were found between those at baseline and the sample at each follow-up. To address potential recall bias or underreporting and/or overreporting, we conducted surveys with both parents and children and found consistent intervention effects between data sources. Although we conducted fidelity testing of training sessions, we were unable to measure fidelity of actual intervention sessions delivered by parents to their injured children.
As a comparative effectiveness study, all subjects received intervention. This patient-centered design responded directly to the preferences of our patient partners.18,33 However, without an untreated control group, causal inferences are limited, and all findings must be interpreted as associations. Thus, we cannot determine with certainty if reduction in stress and depressive symptoms and behavioral problems could have occurred through the natural course of recovery from trauma. The reverse is also possible but not testable wherein stress symptoms and problem behaviors could have worsened in the absence of any supports. Previous research suggests that our interventions may be better than no intervention. A randomized controlled trial of Trauma Education that used a no-treatment control group reported improved anxiety in similarly injured children in Australia.20 Furthermore, without intervention, descriptive studies suggest that up to 38% of hospitalized injured children report PTSS34,35 rates that are much higher than those in our study subjects after receiving intervention.
Conclusion
Findings have important implications for translating Link and Trauma Education in real-world settings. There are costs associated with the time and expertise needed to train families in Link. In contrast, Trauma Education is simple to provide, and the only costs associated with its use would be in tailoring resources and/or referrals specific to a hospital. On the basis of our research, Trauma Education may be reasonably offered as a first step to families of all traumatically injured children in any hospital across the country. However, using Link among traumatically injured children who exhibit behavior problems is a stepped-care approach that may maximize resources.
Acknowledgments
The opinions in this publication are solely the responsibility of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute and its Board of Governors or Methodology Committee.
Dr Ramirez conceptualized and designed the study, coordinated and supervised data collection, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Flores and Cavanaugh conducted statistical analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Woods-Jaeger designed the study and data collection instruments, supervised data collection, and reviewed and revised the manuscript; Dr Peek-Asa conceptualized and designed the study and study instruments and reviewed and revised the manuscript; Ms Branch coordinated and supervised data collection, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Bolenbaugh designed the data collection instruments, collected data, and reviewed and revised the manuscript; Drs Chande, Pitcher, Ortega, and Randell designed the data collection instruments, supervised data collection, and reviewed the manuscript for important intellectual content; Ms Wetjen designed the study and data collection instruments and protocols and reviewed the manuscript for important intellectual content; Ms Roth developed the data collection instruments and protocols for data collection, acquired the data, and reviewed the manuscript for important intellectual content; Dr Kenardy conceptualized and designed the study and reviewed and revised the manuscript; and all authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of this work are investigated and resolved, and all authors approved the final manuscript as submitted.
Individual participant data, including data dictionaries, that underlie these results in this study, after deidentification, are available to researchers by request to mramirez@umn.edu. Data will be available beginning 3 months and ending 5 years after the publication date. Researchers must present a sound protocol and research aims. To gain access, researchers will be asked to sign a data use agreement.
This trial has been registered at www.clinicaltrials.gov (identifier NCT02323204, unique protocol identifier 201111728).
FUNDING: Supported by the Patient-Centered Outcomes Research Institute, CER-1306-02918. The Patient-Centered Outcomes Research Institute required the 2-arm design and requested the removal of a no-treatment arm.
Comments