Posttraumatic Stress Among School Students Following the 2014 Hazelwood Mine Fire

ABSTRACT Children and adolescents are among the most vulnerable to developing posttraumatic stress in the aftermath of a disaster. This study investigated posttraumatic stress among students attending surrounding primary and secondary schools following the 2014 Hazelwood mine fire in the Latrobe Valley region of Australia. Posttraumatic stress levels of students (n = 323) attending smoke-exposed schools (n = 20) were assessed 16–21 months after the event using the Children’s Revised Impact of Events Scale (CRIES). The mean CRIES total score of students attending these schools was 17.22, suggesting an overall moderate level of posttraumatic stress associated with the mine fire. Across the sample, 22% of the students (n = 23) had a CRIES score of ≥30, indicative of posttraumatic stress at a level warranting clinical concern. Students attending schools located closer to the mine (Morwell Schools) experienced significantly greater levels of posttraumatic stress (CRIES M = 21.63) compared with students attending schools located further away (Non-Morwell Schools; CRIES M = 14.70). This difference appeared to have been primarily driven by higher scoring among Morwell students on the Arousal symptom domain of the CRIES. The mean CRIES score of primary school students (M = 23.18) was over double that of secondary school students (M = 10.30) regardless of the proximity of their school to the smoke event; however, this difference did not reach statistical significance. These findings highlight the need for public health responses following disasters that include targeted mental health support for child and adolescent age-groups.

Although most children and adolescents exhibit resilience in the aftermath of disaster (Lai et al., 2017), these age-groups are recognized as among the most vulnerable to mental health impacts after disaster, being susceptible to a range of psychiatric conditions associated with trauma that impair functioning, learning, and development (Braun Lewensohn, 2015;Danese et al., 2020).The acute vulnerability of children and adolescents in times of disaster may be attributable to their dependence on caregivers for safety and support throughout the event, and interruptions caused to the daily social and educational activities of young people (Peek, 2008).Exposure to disaster in childhood may result in mental health impacts enduring into adulthood and persisting decades later (McFarlane & Van Hooff, 2009;Shonkoff et al., 2012).Posttraumatic stress in childhood may be associated with adverse outcomes in many domains across the lifespan, including physical health, mental health, interpersonal relationships, education, and employment (Hadi et al., 2006).
Posttraumatic stress is the most commonly recorded mental health issue that presents among children and adolescents exposed to disaster (Furr et al., 2010;Pfefferbaum et al., 2019), and around 16% of children and adolescents exposed to a traumatic event will go on to develop Posttraumatic Stress Disorder (PTSD) (Alisic et al., 2014).Posttraumatic stress reactions in children and adolescents typically include over-arousal and problems with sleep (arousal symptoms), uninvited or unwanted thoughts about the traumatizing event (intrusion symptoms) and engaging in behaviors to evade or escape reminders of the event (avoidance symptoms) (Carrion et al., 2002).Research demonstrates that the prevalence and severity of posttraumatic stress is related to the type, proximity, and duration of a traumatizing event, with more intensive exposure to disaster related to more profound posttraumatic effects among children and adolescents (Mels et al., 2010;Wang et al., 2012).
The stages of neurological development and cognitive capacity to code, retrieve, interpret, and communicate information about life experiences and emotions are central to how posttraumatic stress responses manifest in children (Pervanidou et al., 2020;Salmon & Bryant, 2002), though it is unclear whether vulnerability to posttraumatic stress in childhood differs depending on age.As children mature, they typically acquire greater competence in the management of their thinking and emotion, which may enhance their capacity to cope and adapt to potentially traumatic events (Pynoos et al., 1996).Trickey et al. (2012) meta-analysis suggested a weak association between being younger at the time of exposure to a potentially traumatic event and PTSD diagnosis.The meta-analysis indicated that this age effect was only in response to traumas impacting groups and not to individual exposures.Accordingly, differential posttraumatic stress outcomes among younger-aged children might be expected in the context of a disaster impacting broadly across a community.

The hazelwood mine fire
In February 2014 bushfire embers spotting into the large open-cut brown coal mine adjacent to the Hazelwood Power Station located in the Latrobe Valley, Australia, caused it to catch fire.The subsequent mine fire burnt for 6 weeks, covering the nearby town of Morwell and other surrounding localities in heavy plumes of smoke and ash for a protracted period of time.The Commonwealth Scientific and Industrial Research Organisation (CSIRO) modeled smoke distribution and deposition over the duration of the mine fire period and determined that levels of particulate matter (PM 2.5 ) in Morwell, the most exposed residential area, exceeded national air quality standard for 23 days of the 45-day mine fire period (Luhar et al., 2020).The widespread distribution of smoke and ash generated considerable community concern about the immediate and long-term health impacts of the event (Wood et al., 2015).Subsequent research has shown the detrimental impacts the event has had on adult mental health in the community (Broder et al., 2020;Carroll et al., 2022;Maybery et al., 2020;C. L. Smith et al., 2023).
The mine fire was particularly disruptive for children and adolescents living and studying within smoke-exposed communities.The severely deteriorated air quality necessitated temporary closures and relocations of some schools (Teague et al., 2014).Initial qualitative investigations of the impact of the mine fire within educational contexts suggested that the event had a detrimental effect on the engagement and wellbeing of school staff and students (Berger et al., 2018(Berger et al., , 2020)).A longitudinal investigation of academic achievement among students attending these schools found delays in academic progress attributable to both smoke exposure during the event and to posttraumatic stress associated with that exposure (Berger et al., 2021;Gao et al., 2023).
The present study aimed to ascertain the presence and severity of posttraumatic stress symptoms among children and adolescents who were attending primary and secondary schools in localities subject to smoke exposure during the 2014 Hazelwood mine fire.The objectives of the present study were to (1) determine levels of posttraumatic stress experienced by primary and secondary students attending schools exposed to smoke during the mine fire and (2) examine differences between the levels of posttraumatic stress experienced by students attending schools located in closest proximity to the mine compared with attending schools located further away.In line with existing theory and research, it was hypothesized (1) that students attending schools located in closest proximity to the mine fire event (Morwell schools) would report higher levels of posttraumatic stress compared with students attending schools in surrounding localities less exposed to the event (Non-Morwell schools), and (2) that younger students (those in primary school year levels) would report higher levels of posttraumatic stress compared with older students (those in secondary school year levels).

Participants
In total, 323 students participated in the survey.This represented an overall participation rate of 15.1% among the eligible student population, with the participation rates within the Morwell and Non-Morwell school-groups being 25.0% and 12.0%, respectively (see Table 1).The sample comprised 157 (48.6%) male and 166 (51.4%) female students and 17 (5.3%)students had an Aboriginal or Torres Strait Islander background.
Participants were recruited from 20 schools across the Latrobe Valley region, comprising government (public-funded) and non-government (public and privately funded; religious-affiliated) primary and secondary schools, that had agreed to participate.Within the Australian system, primary schools are those catering for the elementary grades with, children around 5-11 years of age, and secondary schools are at the intermediate level, catering for students from 12 to 18 years of age.
Participants were categorized into two levels of exposure which distinguished between students attending schools in the town with greatest exposure to smoke during the event (Morwell) and those attending schools situated outside of the town, where the density of dispersed smoke was known to be lower (Non-Morwell).All seven schools (six primary schools and one secondary school) located in Morwell, each situated within 1-4 km of the mine, participated in the study; the 13 participating Non-Morwell schools were located 6-15 km from the mine.The study required four levels of approval: (1) the State education department; (2) each individual school; (3) parental consent; and (4) student consent.Each school distributed the explanatory information and consent forms to parents who returned them to the school for students to participate.

Material
Posttraumatic stress symptoms associated with the mine fire were assessed using the Children's Revised Impact of Events Scale (CRIES; Children and War Foundation, 2005), which is an adaptation of the Impact of Events Scale (IES; Horowitz et al., 1979).The CRIES has good internal consistency and construct validity (Giannopoulou et al., 2006), with a Cronbach's alpha of 0.87 (subscale range: 0.70 to 0.82).The CRIES has also demonstrated good testretest reliability (Dow et al., 2012;P. Smith et al., 2003;Verlinden et al., 2014).The CRIES comprises 13 items each describing a typical posttraumatic stress symptom after exposure to a potentially traumatizing event (e.g., "Do you startle more easily or feel more nervous than you did before it happened?").Items are organized into three subscales corresponding to the symptomdomains of PTSD: "Intrusion" (four items; e.g., the experience of uncontrollable thoughts about the event); "Avoidance" (four items; e.g., engaging in behaviors to evade things that prompt reminders of the event); and "Arousal" (five items; e.g., experiencing sleep difficulties as a result of the event).
Respondents rate how frequently they have experienced each of the example symptoms over the preceding 7 days on a 4-point scale (0 = "Not at all"; 1 = "Rarely"; 3 = "Sometimes"; 5 = "Often").A CRIES total score is the sum of responses to all items (scoring range: 0-65), with higher scoring indicating greater event-related posttraumatic stress.Symptom-domain scores are the sum of responses to all items applicable to the subscale.Perrin et al. (2005) have prescribed a CRIES score of ≥30 as a threshold for identifying posttraumatic stress at a level of clinical concern among children and adolescents.The 2,138 students enrolled in primary year-levels 3 and 5 (children aged approximately 8-11 years at the time of the survey and 6-9 years at the time of the mine fire) and secondary year-levels 7 and 9 (adolescents aged approximately 12-16 years at the time of the survey and 10-14 years at the time of the mine fire) in these schools were invited to complete the survey.This focus on year-levels 3, 5, 7, and 9 was done to facilitate an analysis of academic performance on the annual Australian National Assessment Program -Literacy and Numeracy (NAPLAN), which is completed by these year-levels; the investigation of NAPLAN outcomes among the sample has been published (see Berger et al., 2021).Latrobe Valley region 2015 school enrollment data for year-levels 3, 5, 7, and 9 were obtained from the Victorian Government Department of Education and Training and used to calculate the poststratified survey weights based on year-level, school-type, and gender.
Multiple imputation by chained equations (Royston & White, 2011) was used on all symptom-domain subscales to create 20 imputation sets.Due to the presence of covariate-dependent missingness, age, and gender were included in the imputation process.Weighted imputed means and standard errors were used to investigate differences in CRIES scoring across two separate divisions of the sample: (1) Morwell school students compared with non-Morwell school students, and (2) primary school students compared with secondary school students.Multiple linear regression was performed, with adjustments applied to account for possible confounding effects of age, gender, and school-type.The mean and regression results accounted for weighting and clustering effects of students within schools across year-levels and stratification across school-types.Data transformations and analysis were performed using Stata version 17 (StataCorp, 2017) and the Stata user-written ICE command for multiple imputation (White et al., 2011).

Results
In total, 323 students participated in the survey.This includes 126 students from Morwell (105 primary and 21 secondary students) and 197 students from non-Morwell schools (98 primary and 99 secondary students).This represented an overall participation rate of 15.1% among the eligible student population, with the participation rates within the Morwell and Non-Morwell school-groups being 25.0% and 12.0%, respectively (Table 1).The sample comprised 157 (48.6%) male and 166 (51.4%) female students and 17 (5.3%)students had an Aboriginal or Torres Strait Islander background.More primary school students (n = 203) participated in the study compared with secondary school students (n = 120); there were also more Non-Morwell school students (n = 197) than Morwell school students (n = 126) who participated.
Across the sample, the weighted mean CRIES total score was 17.22 (95% CI [13.19-21.25])and 22% of all students (n = 23) recorded a score exceeding the ≥30 threshold indicative of clinically concerning posttraumatic stress.Distributions of CRIES total scores by school level and by location suggest that younger students and those from Morwell were more likely to report higher levels of posttraumatic stress (Figure 1).
Statistical comparisons between Morwell and Non-Morwell students' CRIES outcomes showed that Morwell students scoring significantly higher than Non-Morwell students (mean diff = 3.05, 95% CI [0.49-5.62],p = .022)(Table 2).This appeared to be primarily associated with the significant  ¥Crude differences between weighted means are not provided as this information is potentially misleading due to unbalanced proportions of primary and secondary students between Morwell and Non-Morwell schools; this issue is primarily because of there being only one secondary school in Morwell.The adjusted mean differences compensate for these unbalanced proportions and, accordingly, are more accurate indicators of between-group differences.Ϯ the SE < .01 for weighted mean here is because there is only one Morwell secondary school difference in scoring on the Arousal symptom-domain (mean diff = 1.94, 95% CI [0.47-3.40],p = .013).Similarly, when secondary school students were compared in isolation, the CRIES total scores of Morwell students were significantly higher than those of Non-Morwell students (mean diff = 2.39, 95% CI [1.41-3.36],p = <.001),again appearing to be primarily associated with higher scoring on the Arousal symptom-domain (mean diff = 2.50, 95% CI [1.48-3.51],p = <.001).Importantly, whilst other comparisons shown in Table 2 did not reach statistical significance, with the exception of the Avoidance symptoms among secondary students, all were in the direction of higher scoring amongst Morwell students compared with Non-Morwell students.
The results presented in Table 2 suggest that, regardless of school location, primary school students scored higher than their secondary school counterparts on the CRIES, prompting an examination of differences by school-level.The analysis presented in Table 3 shows that the mean CRIES total score of primary school students was more than twice that of secondary students, and their Avoidance symptom-domain scores were, on average, threefold greater.However, while a clear trend of greater posttraumatic stress among primary school students can be observed, these school-level differences did not reach statistical significance.

Discussion
This study explored the impacts of the 2014 Hazelwood mine fire on children and adolescents, comparing posttraumatic stress outcomes for students in schools in the most exposed community with those of students attending schools in surrounding communities.Overall, the sample-wide mean CRIES total score of 17.22 suggests students across the region were experiencing, on average, moderate levels of posttraumatic stress associated with the event.It is important to note, however, that approximately one in five participating students recorded a CRIES total score indicating they were experiencing a clinically concerning level of posttraumatic stress.As hypothesized, students from Morwell schools, situated 1-4 km from the mine, reported significantly higher levels of posttraumatic stress compared with students attending schools elsewhere in the Latrobe Valley region, situated 6-15 km away.This effect appeared to have been primarily driven by impacts in the Arousal symptomdomain among Morwell secondary students.A trend toward higher posttraumatic stress among primary school-aged students compared with secondary students was apparent in the data but did not reach statistical significance.
The overall moderate level of posttraumatic stress experienced by students may reflect the unusual nature of the mine fire event, where the somewhat unclear threat posed by the heavy distribution of smoke, rather than an immediate threat to life and property, was the primary source of concern.The average CRIES total score of Morwell students (21.63) is lower than average scores of children exposed to other potentially traumatizing events, such as 25.76 after the Wenchuan earthquake (Wang et al., 2012), 27.84 after a major debris flood (Chen et al., 2012), and 29.87 among children of parents with acquired brain injury (Kieffer-Kristensen et al., 2011); it is also lower than mean CRIES scores (37.8 and 42.48) observed in two clinical samples (Perrin et al., 2005;Verlinden et al., 2014).About 20% of the students recorded clinically concerning CRIES scores, which importantly highlights that disasters involving less severe threat can still have considerable impacts on the mental health of children and adolescents.The prevalence of clinically concerning posttraumatic stress levels among students found in the study is more than double that observed among adults exposed to the Hazelwood mine fire event (Abramson et al., 2017;Broder et al., 2020), which further evidences the vulnerability of children and adolescents to adverse mental health outcomes in the context of disaster.
The finding that students attending schools in Morwell reported significantly higher levels of posttraumatic distress than students attending schools in less impacted towns is consistent with existing theory and research (see Mels et al., 2010;Wang et al., 2012) indicating that proximity to a disaster is an important determinant of posttraumatic distress outcomes among children and adolescents.The greater posttraumatic stress experienced by Morwell students is likely to reflect the cumulative effect of having been subject to greater exposure to the smoke itself in conjunction with increased disruption to daily routines due to the mine fire, including some Morwell schools being closed and relocated for periods of time during the event.This proximity effect has been observed in other research showing that the academic achievement of Morwell students was similarly impacted by the mine fire in comparison with schools elsewhere in the region (Berger et al., 2021;Gao et al., 2023).The influence of parental mental wellbeing and functioning on the mental health of their children following disaster is also well-recognized (Cobham et al., 2016).Students would have received information about the circumstances of the mine fire from a variety of other sources, including other family members, teachers, classmates, and social media.Yeung et al. (2018) have found that distressing images portrayed in media increases the risk of PTSD among school-aged children, and Pfefferbaum et al. (2000) found positive correlation between the level of posttraumatic stress experienced by primary school-aged students and their level of exposure to media about a prominent distressing event occurring at the time.Uncertainty about what was transpiring, how the smoke might affect people's immediate and longer-term health, when the protracted event might conclude, and if and when life would return to normal may underlie why posttraumatic stress impacts appeared predominantly within the Arousal symptom-domain.
Although the difference in posttraumatic stress between primary and secondary school students found here did not reach statistical significance, the weighted mean CRIES score of primary school students (23.18) was more than double that of secondary students (10.30).Accordingly, as noted above, this lack of a significant difference is likely due to the small sample size.This clearly observable trend of greater vulnerability to posttraumatic distress among the study's younger age-group is consistent with existing theory regarding agerelated differences in the presentation of trauma in childhood (see Pervanidou et al., 2020;Pynoos et al., 1996;Salmon & Bryant, 2002) and may reflect the relative capacities of each group to comprehend and process information portrayed in the media about the mine fire, the severity of the event and their personal proximity to its epicenter, the range of risks it posed, and, in the events' aftermath, greater understanding about the likelihood of it reoccurring.

Limitations and directions for future research
The low survey participation rate (15% of the eligible cohort of students), which introduces the possibility of sample biases and limits the statistical power to detect effects, is a limitation of the study.For instance, the nonsignificant p-value returned in the comparison of primary and secondary students' CRIES scores may potentially be due to the number of secondary school students who participated being lower than desirable for this particular analysis.This participation rate is similar to rates found in previous studies of students in disaster-impacted communities, which have ranged from 15% to 22% (Pullins et al., 2005;Salloum & Overstreet, 2012), and research that has sought to recruit students via schools where active parental consent is required (Stein et al., 2007).The lower participation rate observed among secondary school students is also consistent with previous disaster research recruiting via schools, where older students have similarly been less likely to participate than younger students (Pullins et al., 2005).Wordsworth et al. (2021) have asserted that low participation rates in post-disaster research may be attributable to crisis fatigue, the burden of engaging in recovery efforts, and desire in the community to move on from the event.In addition, anecdotal evidence suggests that the lower participation rate of students outside of Morwell was likely due, in some part, to impacts of the mine fire not being as severe or disruptive in those localities, making the study seem less relevant to these school populations.Another limitation of this study is that other factors that may have also impacted students' CRIES scores were not measured, such as the proximity of students' homes to the event, their parents' levels of stress and students' prior exposure to potentially traumatizing events or disaster.It was not within the scope of this study to capture these additional risk factors.
As previously noted, the analysis to ascertain differences in levels of posttraumatic stress symptomology between primary-aged and secondary-aged students was not optimal due to participant numbers in the study.However, the trend in the data collected suggested that these two age-groups very likely differed in how they responded to the mine fire and this topic warrants further attention to be better understood.Future research could look to investigate posttraumatic stress responses differences between child and adolescent agegroups, as well as the potential underlying causes of differences, within the context of disaster.

Clinical implications
These findings suggest that children and adolescents could benefit from supports that foster a clear understanding of what is happening around them, their own position within the event, and how issues of safety and disruption to normality are being resolved.Initiatives that may help mitigate the impacts of disaster exposure on school-aged children and adolescents could include the provision of age-appropriate information about the disaster event catering for the need for knowledge of circumstances among these agegroups.This might take the form of news and social media content about the event that is shaped and delivered in ways that are accessible and engaging across various stages of childhood development.This content could focus on proximity and risk so that children can better understand where they are personally situated in relation to what is occurring, as well as providing a clear description of what is being done to ensure their personal safety, as well as that of others, during the event.
A recent systematic review (Kerbage et al., 2022) identified that a key component of efficacious early intervention preventing childhood PTSD after exposure to a potentially traumatic event was the involvement of caregivers in the process.Accordingly, resources could be developed that assist parents, family members, and school teachers to communicate with children and adolescents about disaster events for the purpose of promoting adaptive processing of disaster exposure.Masten (2021) recommended that strategies to support and promote posttraumatic growth in younger children should focus on ensuring emotional security and reliability of care, whilst strategies to support and promote posttraumatic growth in adolescents should focus on providing prosocial peer activities and opportunities to contribute to their community's recovery.
Schools and teachers are uniquely situated to act as centers of support for children and adolescents in preparing for and responding to disaster (Elangovan & Kasi, 2015;Pacheco et al., 2021).However, attending to childhood trauma is a relatively new frontier in the education system and evidence-based trauma-informed practices for implementation in schools are emerging (Berger, 2019;Thomas et al., 2019).Schools operating in disaster-prone or disaster-impacted communities should look to adopt practices specifically targeted at supporting the mental health of students in tandem with maintaining the trajectories of their academic progress.

Conclusion
This study demonstrates that the 2014 Hazelwood mine fire was a traumatizing event for many students attending primary and secondary schools in impacted communities and that the event continued to trouble some students almost 2 years later.The study provides further evidence that children and adolescents are particularly vulnerable to experiencing posttraumatic stress, which has implications for their physical and mental health and development.Accordingly, emergency and public health responses need to recognize and respond to the potential for posttraumatic experiences specifically within school-aged populations, and this need is likely to grow considerably within the context of more extreme environmental conditions driven by climate change.

ProcedureA
quantitative survey was conducted to assess levels of posttraumatic stress associated with the 2014 Hazelwood mine fire among students attending schools in smoke-exposed localities across the Latrobe Valley region.The survey was administered during the 2015 school year, approximately 16-21 months after the mine fire in February-March 2014.Ethics approval was obtained from the Monash University Human Research Ethics Committee (approval number 5834).Permissions to conduct the research were obtained from the Victorian Department of Education and Training, the Catholic Education Office of Sale, and the principals of each school involved.Parental consent was obtained for each students' participation.Protocols aligning with Victorian State Government guidelines for research in schools and early childhood settings (Department of Education and Early Childhood Development, 2013) and each school's student welfare policy were implemented to support student wellbeing during participation.

Figure 1 .
Figure 1.Distribution of students' CRIES total scores: Comparisons by school level and school location.

Table 1 .
Participation rates for the CRIES survey presented by school location and school level.

Table 2 .
CRIES outcomes: Comparisons of students by school location.

Table 3 .
CRIES outcomes: Comparison of students by school level.