Abstract
Psychological responses to traumatic events vary depending on the type of disaster and the type ofvictim [1, 2]. The present study focused on a train disaster (a collision between a freight train and a post office train) and community residents who lived on both sides of the track where the collision occurred. Using the definition of Figley and Kleber [3], these community residents were not primary victims, in that they had not been on either of the trains, nor secondary victims, in that they were not the relatives, families or significant others of the dead or injured. Neither were they the helpers involved in the rescue operation. Nevertheless, they were exposed to a train collision, the impact of which could consequently lead them to develop some degree of traumatic stress.
In disaster research in the community, the traumatic stress of the present type of community sample has by and large been neglected, while much emphasis has been placed on community samples of primary victims, for example, the Buffalo Creek disaster [4], the Mount St. Helen disaster [5], the Lockerbie disaster [6, 7], and the Bijlmermeer air crash [8]. The aims of the present study were to (i) examine the traumatic stress, resulting from the Stafford train disaster, of community residents who were neither primary nor secondary victims; and (ii) identify the relationship between the traumatic stress and the coping strategies of these residents.
Although descriptions of the traumatic effects of train disasters date back to Victorian times [9–11], there has not been a steady increase in research in this area, unlike research on war or sexual abuse [12]. The existing research on the effects of train disasters has, as was mentioned, mostly focused on primary and secondary victims. In 1977, in Granville, Australia, an early morning commuter train came off the rails, demolished a bridge stanchion and consequently caused a massive concrete span to crash into two train carriages. Eighty-three people were killed. We learned from this disaster that traumatic reactions of passengers (primary victims) included numbness, feelings of detachment and unreality, uncontrollable crying and extreme fear, avoidance, anxiety, irritability, insomnia, nightmares, depression, survivor guilt, anger and phobia of travelling or hearing extraordinary sounds [13].
With regard to secondary victims, 15–18 months later, the level of functioning and psychological wellbeing of bereaved families were poor. Bereavement counselling was offered to the high-risk families and research showed that those who had received such counselling were better than those who had not [14]. Similarly, helpers who were involved in the rescue operations of disasters could possibly be affected [15, 16]. Indeed, the helpers involved in the Granville train disaster developed extreme stress, feelings of helplessness, anxiety, depressed feelings, disturbed sleep and strain. Four out of 13 (31%) helpers, followed up 1 year later, scored between 6 and 20 on the GHQ-20 [17, 18]. The social workers involved in the Clapham rail crash were found to experience a high level of somatization, obsessive–compulsive disorder, interpersonal sensitivity, depression and anxiety [19]. Eighteen per cent and 10%, respectively, of the rescue workers involved in a major rail accident in 1988 in Denmark, scored at or above the cutoff point of the GHQ-28 and the Impact of Event Scale (IES) at 7 months; 6% developed posttraumatic stress disorder (PTSD) of low to moderate severity at 7 months [20].
However, we know that the effects of disasters, especially large-scale disasters such as plane and train crashes, can reach the whole community, from the bereaved relatives to the members of the infrastructure, and not simply primary and secondary victims [21, 22]. Raphael andcolleagues [17, 18,23–27] also emphasized that, in addition to offering professional help to primary and secondary victims, one must not ignore the potential traumatic effects upon other members of the community, who also need to be reached out to and educated about possible and likely traumatic reactions. Psychosocial care also needs to be provided. Such community members could also experience anxiety, anger, sadness, despair, helplessness and guilt, and ultimately, could also become victims, despite not being classified as primary or secondary victims.
Recent disaster research has indeed told us very little about the traumatic stress of community members who do not fall in the categories of primary or secondary victims. However, one could say that the study on the Hillsborough Football Stadium disaster in 1989 (95 spectators were crushed to death at a football match) has given us a glimpse of how such community members could be affected. Wright, Binney and Kunkler [28] investigated the effects of the disaster upon some community members, who were neither primary nor secondary victims, living half a mile from the stadium. They found that 22% of the community sample were considered to suffer from acute PTSD according to DSM-III. The same percentage of the sample scored above the cutoff point of the GHQ-28, that is, they were considered to be psychiatric cases. Based on the IES, the sample showed a high level of traumatic stress symptoms, with at least half reporting frequent intrusive thoughts.
Not only do we know little about the traumatic stress of the community members who are not in the category of primary or secondary victims, we also know very little of the ways in which they cope with their stress. However, coping has been measured in numerous ways [29] mostly among individuals who were involved in warrelated trauma [30–32], natural and technologicaltrauma [33–37] or who suffered interpersonal and personal losses [38].
We recognize that there are different ways of coping with the effects of disasters, such as monitoring and blunting [39], and personality hardiness [19, 40, 41]. Problem-focused and emotion-focused ways of coping are certainly the most widely discussed, though it is not always easy to categorize different strategies neatly into problem or emotion-focused coping [29, 42]. Problemfocused coping refers to efforts undertaken to manage or alter the troubled person–environment relationship that is the source of stress, while emotion-focused coping refers to efforts undertaken to regulate stressful emotions.
On the whole, a great deal of research findings have suggested that emotion-focused coping was a predictor to psychological distress and psychiatric symptomatologies. For example, emotion-focused coping was associated with severe PTSD among Israeli soldiers from the Lebanon war [43], and with PTSD, state anxiety and physical symptoms among soldiers from the Persian Gulf war [32, 44]. To cope with the threat of missile attack during the Gulf War, the Israelis who used emotionfocused coping tended to manifest anxiety, physical and somatic symptoms and pessimism [45]. Also, veterans who sought mental health treatments for their PTSD tended to use emotion-focused coping [46]. Some studies also suggested that there was a positive relationship between emotion-focused coping and long-term psychiatric symptoms. Problem-focused coping however, moderated the detrimental effects of emotion-focused coping on mental health [39].
Based on the victims of transportation accidents, industrial and domestic accidents, terrorism and violent crime, Charlton and Thompson [47] recently found that most coping strategies, measured by the Ways of Coping Checklists (WOC), were associated with severe psychological distress, particularly escape-avoidance (one type of emotion-focused coping), though distancing and positive reappraisal were associated with positive psychological outcome. However, emotion-focused coping was thought not to be associated with avoidance for some victims of boat sinkings [48]. One study found that all coping strategies were equally associated with the presence of PTSD. They included problem-focused coping, wishful thinking, detachment, seeking social support, focusing on positive things, self-blame, tension reduction, and keeping to oneself [49].
In this study, we hypothesized that there would be a high level of traumatic stress, characterized by the impact of the event and general health, among the community residents exposed to the train disaster, and that coping strategies classified as emotion-focused coping would be predictors of distress.
Method
The disaster
On 8 March 1996, a railway disaster occurred in Rickerscote, Stafford, U.K. At 23.08 h, a freight train, carrying liquid carbon dioxide tank wagons, derailed due to the complete fracture of an axle fitted to one of the wagons. Derailed wagons subsequently blocked the adjacent line and were hit by a post office train running in the opposite direction. Only the driver was in the freight train, while 21 employees, including the driver, were in the post office train. Although the driver of the freight train did not appear to be injured, all of the employees on the post office train were. One was killed in the gangway area between the first and second coaches, which had suffered the most damage during the impact [50].
The present community residents were living on both sides of the track where the collision occurred. Some of them were in a ‘near-miss’ situation, in that a 30-foot long locomotive had been pushed up the embankment as a result of the collision, coming to a halt just one inch away from an end-of-terrace house. One carriage landed on top of residents' garages, and carbon dioxide tanks lay just yards from residents' gardens. One locomotive was interrupted by an overhead electrification mast and so did not push into the kitchen wall of a resident's house. A line of trees on the embankment saved several houses from being scythed by the wreckage of the mail train.
Subjects
Sixty-six community residents (M = 23, F = 43) participated in the study. They were all Caucasian and the average age was 54, ranging from 22 to 91. On average, they had lived in Rickerscote for 7 years. Forty-eight per cent were married; 31% were widows or widowers; the rest were single (10%) or divorced (11%). Many were retired (44%) and 3% were unemployed. The rest were working as clerks or administrators (10%), factory workers (9%), educators (12%) and others.
Procedure
The present study was a cross-sectional survey with a retrospective design, and commenced approximately 7 months after the disaster. The researchers designated an area of the housing estate, on both sides of the track, which was thought to be nearest to the crash site. This area, containing 90 households, was within 30–100 feet of the crash site. The total number of adult residents (excluding children) was 95. Letters explaining the purpose of the research were distributed to all 90 households within the designated area. Fifteen residents refused to participate in the study, the main reason being that they felt that they had nothing to contribute to it. Thirteen residents were not at home when visited and revisited two or three times. One had moved out. The remaining 66 were interviewed, giving a 69% response rate. In the interview, the residents were asked to describe some of their initial and aftermath traumatic responses, and were assessed using the IES, the GHQ-28 and the WOC. In order to find out about initial and aftermath traumatic responses, residents were asked questions regarding: any early signs prior to the disaster; their immediate feeling as they experienced the first impact of the crash; whether they felt that they would be killed; whether they offered help or felt helpless throughout the whole event; whether they had expected such a disaster to happen one day; their feeling when they saw the wreckage of the train in daylight the next morning; whether they received professional help; and whether they felt anxious, worried about their safety or angry about the whole event. These residents' traumatic responses have been reportedelsewhere [51].
Measures
The IES is a 15-item, four-point scale (0 = not at all, 1 = rarely, 3 = sometimes, 5 = often), self-report instrument which aims to measure intrusive thoughts related to the traumatic event and consequent avoidance behaviour. The questionnaire was developed on the basis of two samples. One consisted of patients (n = 66) who attended a psychotherapy outpatient clinic as a result of reactions to a serious life event. These patients suffered from stress response syndromes. The other consisted of medical students (n = 110) who had dissected their first cadaver. For the intrusion subscale, the mean scores for male and female patients were 21.2 (SD = 12.5) and 21.4 (SD = 8.6), and for male and female students were 2.5 (SD = 3.0) and 6.1 (SD = 5.3), respectively. For the avoidance subscale, the mean scores for male and female patients were 14.1 (SD = 12.0) and 20.6 (SD = 11.3), and for male and female students were 4.4 (SD = 5.3) and 6.6 (SD = 7.0), respectively [52].
The GHQ-28 [53] was designed as a screening instrument which attempts to estimate the likelihood of subjects being assessed as psychiatric cases at interview. Four subscales comprise the questionnaire: somatic, anxiety, social dysfunction and depression items. As the total GHQ-28 score exceeds the recommended cut-off point of 4, the probability of becoming a psychiatric case increases.
The WOC [42] is a 67-item, four-point scale (0 = not used, 1 = used somewhat, 2 = used quite a bit, 3 = used a great deal) which aims to explore the role of coping in the relationship between stress and adaptational outcomes. The items on the original WOC were classified on the basis of problem-focused or emotion-focused functions of coping. Problem-focused coping consists of confrontive coping, accepting responsibility, and planful problem solving. Emotion-focused coping consists of distancing, self-controlling, seeking social support, escape–avoidance and positive reappraisal.
Data analysis
Descriptive statistics were calculated on the IES, followed by a series of t-tests comparing the mean IES scores of the samples with Horowitz's standardized Stress Clinic samples [52]. Descriptive statistics were also calculated on the GHQ-28 which revealed the percentage of psychiatric cases among the samples. Further descriptive statistics were performed to calculate the mean scores of the WOC, followed by a series of t-tests comparing the means of the samples with the standardized samples. Stepwise multiple regression was the final analysis measuring the association between coping (i.e. WOC) and traumatic stress (i.e. IES and GHQ-28). The data were analysed using SPSS 9.0 (SPSS, Chicago, IL, USA).
Results
The results of the IES showed that for intrusion (mean = 12.61, SD = 10.84), 32% of the community residents often had waves of strong feeling about the disaster and 29% said that reminders brought back feelings about the disaster to them. Just over 20% said that other things often made them think about the disaster and pictures about it popped into their minds. Just under 20% often thought about it when they did not mean to and had trouble falling asleep or staying asleep. However, only 5% had dreams about it.
In avoidance (mean = 11.16, SD = 10.30), 29% often found themselves trying to remove the disaster from memory. Otherwise, lessthan 20% avoided letting themselves get upset, and tried not to think and talk about it. They were often aware of the fact that they had lots of feelings about the disaster which they had not dealt with. Eleven per cent felt as if the disaster had not happened and was not real. Ten per cent often found themselves staying away from reminders of the disaster and felt numb about it.
To compare with Horowitz's standardized Stress Clinic samples, the results showed that the present community residents scored significantly lower in intrusion (t = −4.78, p < 0.001), avoidance (t = −4.04, p < 0.001) and the IES total (t = – 4.91, p < 0.001).
The results of the GHQ-28 revealed that anxiety (mean = 1.75, SD = 2. 54) seemed to be the major problem for this community sample, followed by somatic problems (mean = 1.40, SD = 2.01). Otherwise, social dysfunction (mean = 0.69, SD = 1.60) and in particular, depression (mean = 0.37, SD = 1.19), did not seem to be a problem. The total score (mean = 4.21, SD = 6.17) was just slightly above the cut-off point for psychiatric caseness. Twenty-two (35%) residents had total scores of 4 or above.
The results of the WOC revealed that the distancing (mean = 6.69, SD = 5.06) coping strategy was most widely used, followed by selfcontrolling (mean = 4.21, SD = 3.89) and seeking social support (mean = 3.83, SD = 4.20). Otherwise, the extent to which community residents used confrontive (mean = 3.24, SD = 3.27), escape-avoidance (mean = 3.16, SD = 3.39), planful problem solving (mean = 3.45, SD = 4.65) and positive reappraisal (mean = 2.59, SD = 3.03) coping strategies was similar. The least-used strategy was accepting responsibility (mean = 1.27, SD = 1.96).
Comparing the results with the standardized samples showed that there was no significant difference in use of the confrontive (t = −1.85, NS) and escape-avoidance (t = −0.04, NS) coping strategies. However, apart from distancing (t = 7.69, p < 0.001), the community residents scored significantly lower in the rest of the subscales (self-controlling t = −3.21, p < 0.01; seeking social support t = −3.40, p < 0.001; accepting responsibility t = −2.45, p < 0.02; planful problem solving t = −7.84, p < 0.001; positive reappraisal t = −1.96, p < 0.05).
To make an association between traumatic stress and coping strategies, stepwise multiple regression analyses were performed. Table 1 shows that distancing, seeking social support, accepting responsibility and escape-avoidance were the coping strategies predicting the IES intrusion. Escape-avoidance was the only predictor of the IES avoidance. Confrontive coping, distancing, accepting responsibility and escape-avoidance were the coping strategies predicting the GHQ-28 total.
Multiple stepwise regression with coping strategies predicting IES and GHQ-28
Further analyses were carried out to examine whether there were associations between traumatic stress, coping strategies and demographic variables of gender, age, marital status, occupation and distance from the crash site. Results showed no significant associations.
Discussion
The results seem to suggest that, contrary to our hypothesis, the impact of the event was not severe, at least when compared with Horowitz's primary victims who suffered from stress response syndromes, due to bereavement, or sustained personal injuries resulting from accidents, violence, illness or surgery. However, when we compared the IES total of the community residents with that of Wright et al 's [28] community samples (mean = 24.06, SD = 6.93), we found no significant differences. Also, when we compared the IES total with that of Andersen et al 's [20] rescue workers in a rail disaster, who were followed up 7 months after the incident (mean = 7.24, SD = 9.09), we found that the community sample scored significantly higher (t = 6.45, p < 0.001). This could be explained by the fact that the rescue workers might have recovered from the impact of the event to a significant degree within 7 months, or alternatively that their professional training had helped them to deal with the impact of the event successfully. Indeed, various studies have demonstrated the value and importance of training as a way of coping with distress [54–57] though Paton [58] recently showed that the training and experience of fire-fighters did not prepare them for major disaster work.
With regard to the level of general health, the number of GHQ-28 case-scorers among the present residents was similar to that found by Andersen et al. [20] (χ2 = 1.76, NS) but significantly higher than that of Wright et al. [28] (χ2 = 7.74, p < 0.01). To confirm previous studies by Raphael et al. [17, 18] and Hodgkinson et al. [15, 19] anxiety and somatization appeared to be problematic for the community residents. However, depression did not appear to be a problem, the reason being, one might speculate, that the depression items in the GHQ-28 mainly concentrated on measuring suicidal intent and, as such, were perhaps not appropriate for these community residents.
In terms of coping strategies, our hypothesis was partially supported in that, while emotion-focused coping strategies (i.e. in this case escape-avoidance and distancing) were found to be predictors of distress, problemfocused coping strategies (i.e. in this case confrontive coping and accepting responsibility) were also found to be predictors. This cast a different light on the previous general findings recorded in the literature which point to emotion-focused coping as being the main predictor.
It was understandable that accepting responsibility was the least used coping strategy. The majority of community residents did not feel that they were responsible for the disaster, nor that they should criticise themselves or had brought the problem on themselves and therefore should apologise for what had happened. Instead, they believed that it was the fault of the company which owned the freight train or of the company which maintained the rails. There was a tendency that the less they used the coping strategy of accepting responsibility, the lower the scores were in both impact of event and general health.
Conversely, escape-avoidance was used to some extent, in that residents had fantasies about how things might have turned out, or refused to believe that such a disaster had actually happened. However, one could suspect that such a strategy would not really help residents to cope successfully, as they lived so near to the track and were so often reminded of the disaster by the passing trains. This could be why anxiety was the most reported symptom of the GHQ-28. In fact, at the interview, about 40% of the residents reported that they felt anxious about their safety and they also felt anxious when trains passed. There was a tendency that the more the residents tried to avoid facing up to the disaster, the more distress symptoms, in terms of the impact of event and general health, they experienced.
Instead of using the escape-avoidance strategy, over 60% reported that they had to distance themselves from what had happened. They submitted to fate and believed that it was just bad luck, and it was best to pretend that nothing had happened and look on the bright side. Over 50% reported that they tried to forget the whole thing and that they refused to think about it too much. However, similar to escape-avoidance, there was a tendency that the more they tried to distance themselves from the disaster, the more symptoms they experienced. On the other hand, the results also suggested that the more residents confronted the disaster (i.e. confrontive coping) by expressing anger at the train company, venting their feelings and trying to get the train company to accept responsibility, the more symptoms they experienced.
The conclusion was that our hypotheses were not entirely supported. The severity of the impact of the event was less than that of the standardized samples. Just over one-third were thought to be psychiatric cases. Both emotion-focused and problem-focused coping strategies seemed to play a role in predicting the impact of the event and general health.
To close the paper, it is important to comment on a methodological issue concerning the retrospective nature of the study. We realized that it was possible that the present community sample, who had developed psychological distress, might recollect their coping strategies in the context of their current mental state rather than on a more objective assessment of the way in which they actually used coping strategies at the time of the trauma. The best we could do to address this problem, though far from being a perfect solution, was to ask the sample to respond to each WOC question according to the way in which they coped with the actual trauma and to emphasize the importance of answering with this in mind. Also, before they embarked on filling in the questionnaire, a brief statement explaining precisely the above had to be read. We felt confident that, with such guidance, the sample did indeed try to address the WOC questions according to how they used the coping strategies at the time of the trauma. Another methodological issue worth mentioning is the fact that the present study could have been strengthened by data comparison with a community control group of residents not exposed to the disaster. This would certainly help in terms of interpreting the results of the GHQ-28. We expected that the present community residents would have scored higher than those not exposed to the disaster in not only the individual GHQ-28 items but also the overall number of psychiatric cases.
Footnotes
Acknowledgements
The authors are grateful to the Royal Society for providing a travel grant enabling dissemination of the results of this study at the XXVII International Congress of Psychology in Sweden.