Abstract
This paper reports the first post-disaster intervention program known to the authors, where mental health services were provided to children and adolescents based on the results of extensive, school-based, use of psychological testing. This research extends previous efforts to identify children who would benefit from an intervention following exposure to war [1] and a maritime disaster [2].
In the post-disaster environment, the concept of screening is attractive to the service provider. Screening may allow many to be reassured, while individuals at greatest need receive the scarce mental health resources available. Fletcher suggests that for a condition to warrant a screening approach, the prevalence of the condition would need to be ‘sufficiently high’, the screening procedure would have to be acceptable to the public and other professionals, and screening would need to be relatively economical and followed by an efficacious intervention [3]. It is established that a significant proportion of children will experience symptoms of posttraumatic stress disorder (PTSD) following diverse traumatic events: civil violence [4], hostage situations [5,6], natural disasters [2,7,8], and war [9]. No unitary PTSD prevalence would be expected given the diversity of traumatic events. Reports from questionnaire studies following natural disasters include MacFarlane's [10] study reporting 13% of children experienced dreams or nightmares about a bushfire disaster 8 months after the event. Using a self-report PTSD instrument with established discriminant validity [11], a 5% prevalence of PTSD was reported 3 months after Hurricane Hugo [7], and 3 months following Hurricane Andrew 25% of children were in the ‘severe’ PTSD symptom range and 5% in the ‘very severe’ range [12]. Events with greater actual or potential loss of life report higher PTSD prevalence rates: 100% experiencing ‘psychic trauma’ after a child hostage event [6], 90% diagnosed with PTSD after a sniper attack on a school [13] and 40% in the clinical range on an instrument of emotional distress after a shipping disaster [14].
The meaning of these prevalence figures requires some discussion. All studies thus far lack criterion validation; indeed as yet there is no technique to empirically validate PTSD. Construct validation involves finding a significant positive correlation between the PTSD measure used and another form of PTSD measurement, or a similar relationship between the PTSD measure and an anxiety disorder or anxiety symptoms. The studies of Yule [15] and Shannon [7] both demonstrated such a correlation between the measure of PTSD and the Revised Manifest Anxiety Scale (RMAS) [16]. A pilot study by Earls and colleagues, with greater methodological rigour in that it used a diagnostic interview, reported many children with PTSD symptoms; however, no child reached diagnostic caseness for PTSD from a sample of 39 individuals who experienced a flood disaster [17]. In summary, it would be expected that many children would experience symptoms of PTSD after a bushfire disaster and that the prevalence of ‘cases’ of PTSD would be dependent upon event-related variables including trauma exposure and the perceived threat of the event.
A possible post-disaster PTSD prevalence of 5–10%, although elevated, would not warrant screening if parent identification of child distress and subsequent provision of therapy for the child was also high. However, it was hypothesised that parents and teachers would be unable to accurately identify distressed children. The principal argument for possible low-parent identification is that although PTSD in children has some recognisable behavioural symptoms (posttraumatic play, hypervigilance, an exaggerated startle response, nightmares and sleep disturbance), the cognitive re-experiencing symptoms and avoidance and numbing symptoms are more akin to an internalising disorder of childhood. A robust finding in the child and adolescent mental health field is the low parent to child concordance of internalising symptoms. This has been reported when employing symptom checklists such as the Child Behaviour Check-List and the Youth Self-Report [18,19], and with structured diagnostic instruments such as the Diagnostic Interview Schedule for Children [20] or the Child Assessment Schedule [21]. Achenback and colleagues reported the average correlation between child and parent symptom report at 0.25 [22].
Findings specific to the trauma field include many children consciously not divulging to their parents the level of their distress as they do not want to burden their parents with further concerns [15], or generally having difficulty talking to their parents following trauma [23,24]. From a parent perspective, parents may deny the presence of symptoms in their children [25], or parents may also experience post-disaster PTSD. Parents experiencing symptoms such as emotional numbing may lead to the decreased ability to identify emotional distress in their child. Similarly, teacher identification of child distress may also be decreased, especially given some children's ability to function at a reasonable level in school despite emotional distress [9] (although with increasing symptom severity academic progress is impaired [26]). The combination of reported prevalences of post-disaster PTSD of 5–10% and possible low parent report are arguments for proactive screening.
Widespread, proactive post-disaster service delivery organised around screening is in a developmental stage. There is a lack of consensus as to the optimal type of screening instruments. There is evidence that traumatised children report symptoms that generally equate to the DSM-IV [27] criteria for PTSD, including the age-specific features: generalisation of fears, posttraumatic play, a sense of foreshortened future, somatised distress and ‘omen formation’. Further, there appears to be symptom stability over time. Persistent symptoms of PTSD were reported 5–9 months after a shipping disaster [14], 1 year after a sniper attack on a school [13], 26 months after bushfires [10], and 10 years after surviving the ‘killing fields’ of Cambodia [26].
There is evidence that a screening procedure should also include an assay of depressive symptoms. In a sample of adolescent girl survivors of a shipping disaster, depression symptoms measured on the Birleson Depression Scale (BDS) [28], were significantly higher in the trauma group than control and ‘near miss’ groups. Using a BBS cut-off of 16, five of the 24 adolescent girls in this sample were considered to have significant symptoms of depression 10 days post disaster; symptoms were still present at a 5-month follow-up [14,15]. Warheit and colleagues also reported a significant, although small, relationship between immediate post-hurricane stress and post-hurricane depression and suicidal ideation [29].
Demographic and event-related factors might influence the choice of screening questions. Female [14,30,31] and young children [7] have been reported to experience more PTSD symptoms, although the gender finding is equivocal with some studies finding small [12] or no gender effects [4]. Event-related factors associated with increased emotional distress include the degree of trauma exposure and severity of the initial post-trauma reaction [4]. The degree of perceived threat to self has been associated with increased symptoms in both an adult population [32] and in children [33]. Psychological constructs intrinsic to the child are also influential in post-disaster adjustment. The child's attempts to cope with the event [12], a subjective sense of self-efficacy and positive social support, are associated with lower levels of post-disaster distress [30], while internal causal attribution predicts poorer outcome [34]. An ideal screening instrument would combine an assay of PTSD and depression symptoms; event-related factors, and coping and vulnerability factors.
The general aim of the Sutherland Bushfire Trauma Project (SBTP) was to identify children experiencing significant emotional distress and depression 6 months after a bushfire disaster, and to offer such individuals a psychological intervention. Identification at 6 months post disaster was chosen to allow the resolution of post-disaster acute stress reactions. It was hypothesised that a proactive program would identify a greater number of distressed children than would present to school psychologists and the regional child and adolescent mental health team during the 6 months prior to the project. Other hypotheses included: that the prevalence of emotional distress would remain significant 6 months post-disaster; that distress would be correlated with the individual's exposure to the traumatic event; and that depressive symptoms would be elevated in the group exposed to the bushfires, especially in individuals who experienced home damage and loss.
Method
The 1994 New South Wales bushfire disaster involved fires burning over a 5-day period: 600 000 ha of land was burnt; 185 houses were destroyed; 113 further homes were damaged; four people died. In the Sutherland Shire of southern Sydney, the site of the Sutherland Bushfire Trauma Project (SBTP), 86 houses were destroyed, one primary school was completely razed, two others were damaged and, miraculously, only one life was lost. Government mental health resources in this area consisted of one child and adolescent mental health team and Department of School Education psychologists. Approximately 3600 children, grades 4–12, attended public-funded schools in the fire-affected area. Extensive school-based screening for post-disaster distress was one of several service provision strategies that attempted to provide a meaningful response to this large scale Australian disaster.
Participants
The study population consisted of all children in grades 4–6 attending public school education in the designated disaster area and meeting inclusion criteria. The inclusion criteria were: school was destroyed or damaged by fire; or homes of children attending the school were destroyed or damaged by the fire; or large areas of blackened, fire-damaged land needed to be traversed to attend school. Eleven primary schools fulfilled one or more of these criteria. One primary school principal refused participation of students or staff from his school in this project. School size varied greatly from small cottage schools of fewer than 30 students to schools with several hundred students.
Procedure
Project personnel were either school psychologists, members of the local child and adolescent mental health team or volunteers studying psychology at a tertiary level. Student testing was standardised and protocol directed. All personnel involved completed a brief training course. Self-report questionnaires were given to students in their classroom in groups of 20–30 children. All children had questions read to them to standardise for reading ability.
Self-report instruments
The self-report battery used included the Impact of Event Scale (IES) [35], the Revised Manifest Anxiety Scale (RMAS) [16] and the BDS [28]. The IES is a 15-item scale with four answer fields measuring a total trauma score and intrusion and avoidance symptom subscales. Acceptable subscale reliability has been reported [35] and discriminant validation of traumatised versus non-traumatised groups in a range of settings has been demonstrated in adult populations [36]. The scale has been used with adolescents [14] and children as young as 8 years of age [15]. Yule and Udwin, investigating young adolescent schoolgirl survivors of a cruise ship disaster, used an IES (15-question version) score of above 40 to indicate a group at ‘high risk’ for psychological problems [2]. During the pilot phase for this project (n = 60 children) two IES questions (‘I was aware that I still had a lot of feelings about the bushfire, but I did not deal with them’ and ‘My feelings about the bushfire were kind of numb’) needed clarification by greater than 50% of younger students. Accordingly, these questions were removed and subjects answered an IES-13 questionnaire. Consistent with fewer IES questions, an IES-13 cut-off score of 36 was used to identify a population ‘at risk’ for poor psychological outcome. The RMAS is a 37-item trait anxiety measure with a True/False format. It was chosen to measure trait anxiety given evidence of acceptable psychometric qualities including reliability [37], consistency of factor structure and convergent validity [38]. Symptoms of depression were measured with the BDS, an 18-item questionnaire with three answer fields.
Given the service provision nature of this project resources were not available to collect the breadth of data usually sought in dedicated research projects. However, the project consent form did contain individual items for parent report. These included: yes/no responses regarding the child's location on the day of the bushfire; the child's verbalised perception of threat to self and parents; home damage; evacuation experience and residential dislocation experience.
Statistical analysis
Gender (male = ‘1’, female = ‘0’) was analysed as a binary covariate. Age, IESSUM (total IES-13 score), RMASSUM (total RMAS score) and BDSSUM (total BDS score) were analysed as continuous variables. School attended was analysed as a factored (nominal) variable. All other variables were analysed as binary (1//0) covariates. In addition to explaining variables of primary interest, such as the total distress score (IESSUM), models adjusted for the influence of potential confounders, such as age and gender.
Generalised linear models (linear regression) [39] were used to model the effects of multiple covariates on the continuous outcomes IESSUM and BDSSUM. A forward stepwise modelling procedure was used to select a useful subset of independent predictors of an outcome of interest. An F-statistic for each predictor not in the current model was first calculated. If the F-statistic for any predictor was greater than 2.0, the predictor with the highest F-statistic was entered into the model. This iterative process was repeated until no variables met the criteria for addition to the model. A manual process also checked the validity of models.
The forward stepwise regression models constructed included the following variables as initial possible explanatory covariates: age; gender; grade at school; exposure to bushfires; experience of evacuation procedure; perceived threat to self; perceived threat to parent(s); separation of child from parents; damage to home; and RMASSUM. To investigate any cluster effect of school attended, this variable was included as an initial possible explanatory covariate. The BDSSUM index was included as an initial possible explanatory covariate in the model with IESSUM as the outcome and vice versa. Minitab for Windows v12 (Minitab Inc.) and S-Plus v4.5 (Mathsoft Inc.) were used to manipulate and analyse data. Statistical significance was defined at the standard 5% level.
Results
The proportion of students from the 11 schools studied who returned signed parental consent forms to their schools and subsequently participated in screening was 76%. Those parents not giving consent to screening usually cited being ‘out of the area’ during the fires and therefore the project did not ‘apply’ to them. Little other information concerning the group of non-participants exists.
The study population comprised 601 primary grade students, aged 8–12 years (mean = 10.1 years, SD = 0.9 years): 54.6% (n = 322) female, 45.4% (n = 279) male. Participants lived in an area where 81% of residents were Australian born and 0.03% were Australian aboriginal or Torres Strait Islanders. The majority of non-Australian born residents emigrated from an English-speaking country and comprised 8.8% of the total population of this area. Immigrants from a non-English speaking European country or Asia comprised 3.3% and 1.7% of the area's population, respectively. There were relatively few single-parent households (10.4%). The area is considered ‘middle class’, consistent with 46% of residents having obtained some post-school qualification and a median annual household income of A$40 000–50 000 [40].
Post-disaster emotional distress and depression symptoms
Characteristics of the study population (n = 601)
The mean depression scale score was 18.09 (SD = 14.02). Using the BDS cut-off of 16 reported by Yule and Udwin [2], 4.7% (n = 44) of the sample reported symptoms consistent with a depressive illness. The items most frequently validated were: ‘I feel very bored’ and ‘I am (not) easily cheered up’. Two other items frequently validated by this sample, ‘I have horrible dreams’ and ‘I get tummy aches’, are not specific to depression and could have been reported by individuals with PTSD. The mean anxiety scale score was 9.46 (SD = 7.35). Using the RMAS cut-off of 19 cited by Yule and Udwin [2], 14.1% (n = 85) of the sample reported symptoms consistent with high trait anxiety.
Symptoms of emotional distress (IESSUM): multivariate analysis
The RMASSUM index, the BDSSUM index, experience of evacuation procedure, and perceived threat to parents significantly predicted the IESSUM index independently of other possible covariates (Table 2). Higher symptom scores for anxiety (RMASSUM), depression (BDSSUM) and earlier school grade were associated with increased symptom scores for emotional distress. The experience of an evacuation procedure and the presence of a perceived threat to one or both of the parents of a child were also both associated with increased symptom scores for emotional distress. Neither gender nor age was significantly associated with IESSUM independently of the other covariates modelled. No significant cluster effect of school attended was observed for this outcome.
Results of generalised linear modelling
Symptoms of depression (BDSSUM): multivariate analysis
The RMASSUM index, the IESSUM index, and perceived threat to parents significantly predicted the BDSSUM index independently of other possible covariates (Table 2). Higher symptom scores for anxiety (RMASSUM) and emotional distress (IESSUM) were associated with increased symptom scores for depression. Neither gender nor age was significantly associated with BDSSUM independently of the other covariates modelled. No significant cluster effect of school attended was observed for this outcome.
Discussion
This research reports a prevalence of emotional distress of 12% in a large, representative sample of primary grade children. Despite differences in the timing of post-disaster assessment and screening instruments used, our reported prevalence of emotional distress is similar to the report of 13% of children experiencing trauma-related dreams or nightmares 8 months following a bushfire disaster [10] and 5% of school children fulfilling the more stringent PTSD diagnosis 3 months following a hurricane disaster [7]. Our reported prevalence is lower than the reports on individuals experiencing events with greater threat potential and trauma exposure, such as adolescent girl survivors of a ship sinking [14] and children in a schoolyard subject to sniper fire [13].
The prevalence of depression in this sample was of similar magnitude to community prevalence studies in non-traumatised child and adolescent populations. A review by Angold found most studies reported depression rates of < 5%, with a cluster around the 2–3% prevalence rate [41]. Goodyer's review of 12 studies, generally reporting 6-month prevalence data, found a child and adolescent depression rate of 1.8–8.9% [42]. In the bushfire sample, no gender difference in rates of depression in children less than 12 years of age was found, a similar finding to community studies [43, [44]. The absence of elevated levels of depression in this sample probably reflects the low experience of loss of life, bereavement and home destruction reported by individuals. When home damage and destruction was reported, after controlling for a number of potentially confounding variables, depressive symptoms did significantly cooccur with PTSD symptoms. This may be related to depressive symptoms being a direct result of the traumatic experience. However, in non-traumatised child and adolescent samples significant comorbidity between depression and anxiety disorders [45] has also been reported.
Multivariate analysis suggested a significant relationship between increased emotional distress and female gender, younger students, elevated trait anxiety and increased depressive symptoms. A novel finding was that the child's perception their parent may have died during the bushfire was a more significant factor in explaining the variance of emotional distress scores than the perception that they themselves may have died. Methodological problems require caution in interpreting this finding; parental reporting of a child stating they feared for their, or their parents' lives is likely to be a substantial underestimate when compared to direct child self-report. However, it is possible that children in this age group consider their own death as an unlikely abstraction, whereas the death of a parent may be more conceivable, threatening and trauma inducing to the child. Further research using child self-reporting is required. A simplistic exposure question, ‘where was the child on the day of the fire?’, did not differentiate the distressed from the non-distressed group. However, the experience of evacuation on the day of the fire, probably a proxy measure of more severe fire exposure, was significantly associated with increased emotional distress scores independently of the other possible covariates. While weak bivariate relationships were demonstrated between increased separation experience and home damage and subsequent emotional distress, these relationships did not remain significant when other variables were adjusted for in multivariate analysis.
Multivariate modelling of the depression scores found significant independent relationships with total distress score, trait anxiety and perception of threat to the parents. The substantial overlap in significant independent contributors to the IESSUM and BDSSUM model and the fact that each outcome was significantly predicted by the other, suggests that either symptoms of emotional distress and depression are a central feature of children's responses to traumatic events or pre-existing symptoms of depression confer a vulnerability to emotional distress following a traumatic event.
From a service provision perspective, case note review identified fewer than 20 children who presented to either a school psychologist or a child and adolescent mental health professional during the 6 months prior to the school-based testing procedure. Following the program, 72 children with IES-13 scores in the severe range were offered an intervention. Parents of children with mild to moderate IES-13 scores were invited to discuss their child's needs with treatment staff; many of these children and families were provided some therapeutic assistance. The process of testing for distress itself has benefits in the post-disaster environment. Communication of the child's emotional status to parents encouraged a dialogue about the distressed child and the child's emotional management. School-based testing presented the added advantages of employing existing resources in the ‘ownership’ of a post-disaster intervention model and focusing, usually scarce resources on individuals at greatest need. Such arguments, along with the dissonance between the number of children receiving treatment pre- and post-testing suggests widespread, school-based, proactive, post-disaster investigation for emotional distress is an undertaking worthy of further research and development.
Due to lack of resources, this project did not employ formal screening using a ‘gold standard’ measure (e.g. a structured diagnostic instrument) that could establish ‘caseness’, age and gender appropriate case cut-off scores and investigate the specificity and sensitivity of the screening battery employed. Such an approach is clearly needed in post-disaster work with children. Even if formal screening was presently possible, it is acknowledged that screening has a potential cost. False positive status may confer unnecessary treatment, cost and inconvenience, and label a normal child and encourage illness behaviour. Conversely, false negative status may provide a barrier to later care-seeking behaviour. Even when screening identifies a child who warrants intervention, screening at 6 months or later may delay early treatment for that child. Further, it is probable that any post-disaster screening will increase demands for therapy. To service this need requires the co-opting of suitable numbers of trained staff or service delivery methods that can assist large numbers of children; guided therapy workbooks [46] and group therapy are possible solutions.
This study suggests that in the type of disaster described, 12% of children aged 8–12 years experienced significant symptoms of emotional distress, symptoms that persist for at least 6 months. The service provision response used post-disaster testing for psychopathology. The program was economically viable and was well accepted by mental health and education staff. Test results were used to identify at-risk children or as a means to discuss children's emotional needs with their parents. While this project confirms large-scale, school-based screening is a potentially valuable service provision strategy, further study is required, to relate self-report emotional trauma questionnaires with diagnoses of posttraumatic stress disorder and related disorders, case controlled investigations with non-directly exposed children, longitudinal study of distress symptoms and investigation of the child's impairment in the family, school and social domains. To achieve these aims, research in the post-disaster environment needs to be facilitated, in part by research and development funding incorporated in any state or federal government response to a given disaster.
Footnotes
Acknowledgements
The authors wish to acknowledge the dedication of Janet Cross and school psychologists in the Sutherland Shire of New South Wales and the vision of school principals and school authorities with responsibility in this area. Their participation made this project possible. S. Einfeld gave valuable advice on research design and instrument selection. Mary Batik gave advice on earlier drafts of this paper.