Discussion
In this meta-analysis, we found that parents of children who have had a traumatic brain injury report more severe internalizing and externalizing symptoms than parents of children who have not had a TBI. This was true across injury severity. Children who had a TBI exhibited more internalizing symptoms than children who had an orthopedic injury or no injury at all. Based on the results of the meta-analysis alone, severe TBI had the most severe internalizing problems, with moderate TBI the least. The small number of studies that included moderate TBI groups may explain the small effect size difference found in this analysis.
Consistent with prior research, children who had sustained a TBI were at greater risk for developing psychiatric symptoms as measured by self-reports, most often completed by a parent. Previous studies that were not included in this meta-analysis often looked at the diagnosis of a novel psychiatric disorder. One such study by Max and colleagues followed a cohort to determine what percentage of the children developed a novel psychiatric disorder following a TBI compared to those who sustained an orthopedic injury. In this sample, the TBI groups consistently had a higher percentage of participants qualifying for a novel psychiatric diagnosis. The results of this meta-analysis are consistent with these findings. Additionally, these results suggest that the difference between the TBI group and the control group was nearly equivalent for both internalizing and externalizing symptoms. This may suggest that children who experience a TBI are equally likely to see increases in both types of symptoms. This does not mean that symptom severity is equivalent in internalizing and externalizing symptoms, but that both cases of symptoms are worse for children with a TBI than for those with no injury or an orthopedic injury. While the current meta-analysis did not seek to establish whether children met criteria for a psychiatric disorder, other studies have demonstrated that the most common psychiatric conditions following TBI are personality changes, depressive disorders, and ADHD (Bloom et al., 2001).
A goal of this meta-analysis was to determine the effect of time since injury and age at injury on the severity of internalizing and externalizing symptoms in children with traumatic brain injury when compared to peers who did not experience a head injury. The results of both of these meta-regressions were null, indicating that age at injury and time since injury were not significant predictors of the effect size differences present in the samples. However, severity of injury was included as a covariate for these analyses, and was a significant predictor for both internalizing and externalizing symptoms. This may indicate that severity of injury is more closely tied to psychiatric symptoms following a TBI than time since injury or age at injury. Thus, it may be that the largest risk for developing psychiatric symptoms after a TBI is the severity of the TBI. It is also possible that the studies included did not include a large enough range of data to demonstrate the effects of age at injury or time at injury in the samples included in the studies.
It is possible that children who sustain a TBI are more likely to demonstrate internalizing and externalizing symptoms prior to injury. Most studies include an orthopedic injury control group to account for pre-injury behavioral characteristics such as hyperactivity and impulsivity. Some studies indicate that while TBI and OI groups do not differ on family history of psychiatric illness, the TBI group continue to show larger effect sizes indicating more severe symptomatology or greater numbers of individuals diagnosed with a novel psychiatric disorder (Max et al., 1998).
This meta-analysis has several limitations. First, as in all meta-analyses, there is inherent risk of publication bias. Publication bias is the possibility that only those studies that found significant effects were published. This could mean that a large number of studies that did not find significant differences between TBI groups and controls were never published, and their null findings or small effect sizes were not included in the results of this meta-analysis. We addressed this issue by including results from the classic fail-safe N test, which indicates the number of studies with null findings required to make the results of this study nonsignificant. Second, due to the stringent inclusion criteria, only a small number of studies (14) were included in the meta-analysis, leaving the study susceptible to the results of findings from additional studies. This is likely to be especially true for the moderate TBI group, which had the least amount of studies representing this group. Third, all of the studies included in this study represented behaviors seen by parents. It is possible that parents are not the most reliable source when reporting problematic behaviors in their children, and these findings may be subject to bias in the parents. Future studies should include multi-method models of assessing behavior in children, such as including teacher and self report. Finally, the results of this meta-analysis are contingent upon the methodologies used in the source studies. We acknowledge these potential limitations in our study.
Conclusion
Regardless of TBI severity and in the context of the study's limitations, the results of this meta-analysis indicate that parents of children who sustain a TBI report more severe symptomatology than parents of children who were not head injured. Furthermore, the effect size was largest for children who acquired severe TBI than for those with mild and moderate TBI. Age at injury at time since injury did not appear to be significant predictors of psychiatric symptoms following head injury.