PTSD in the New Zealand Police

* - Preliminary versions of this paper have been presented at the International Society for Stress and Trauma Studies Annual Conference, San Francisco, November, 1996 and at the British Psychological Society, Scottish Branch, Annual Conference, Crieff, November, 1996.

Abstract

Organisations whose workers risk traumatic exposure, are increasingly interested in preventing sequelae such as posttraumatic stress disorder (PTSD). A common intervention is the provision of psychological debriefing following trauma. In accordance with recent criticisms of this approach, Harvey (1996) has proposed a multidimensional model of trauma recovery. The present study tested some of the person, event and environmental variables in the model with 527 New Zealand police officers who responded to a questionnaire survey of trauma and social support at work. The results showed no differences in PTSD symptoms between officers who had and those who had not been debriefed. However, greater social support and opportunities to talk about traumatic experiences and their emotional impact, with others in the work place, were shown to be related to fewer PTSD symptoms. The findings support current suggestions that organisational post-trauma interventions should be developed take into account social environmental factors and recovery needs over time.

Although psychological debriefing has theoretical rationale and strong support from many practitioners, much of this support is anecdotal and there is a need for further empirical study to examine the immediate and long-term effects of such interventions (Shalev, 1994). Bisson and Deahl (1994) also call for more rigorous evaluative research before debriefing is provided routinely. The results of the evaluation of debriefing to date are mixed. Bisson & Deahl cite examples of studies that support the effectiveness of early psychological intervention, but have been flawed for at least one of several reasons: they assessed only the subjective, immediate reactions of participants and no long term outcomes; there were no control groups; or their results were thrown into doubt by other contradictory findings. There are some more recent studies that have employed comparison groups of people who have experienced the same event and have not attended a debriefing. These studies have not found any differences in psychological stress symptoms for those who were debriefed, at two weeks (Hytten & Hasle, 1989), six months (Brom, Kleber & Hofman, 1993), nine months (Deahl, Gillham, Thomas, Searle & Srinivasan, 1994), or up to two years (Kenardy et al., 1996) after the traumatic event. Griffiths & Watts (1992) found that emergency workers who had been debriefed following bus crashes had significantly higher symptom scores one year later, than those who were not debriefed. These findings, in combination with findings of psychological sequelae in emergency service personnel, six months following debriefing (Sloan et al., 1994), suggest that debriefing alone, is not effective in preventing adverse outcomes. Other aspects of prevention are suggested by Thompson & Solomon (1991), who found that a body recovery team had lower symptom scores compared to other similar teams. They attributed this to careful selection, training, and ongoing managerial support as well as critical incident debriefing sessions, which were managed as part of the group routine. Such findings support recent suggestions (e.g. Raphael, Meldrum & McFarlane, 1995) that, whatever the contribution of psychological debriefing to the reduction of posttraumatic stress symptoms, there are other variables in the environment which also contribute to recovery and must be considered by responsible organisations and health professionals.

In accordance with these suggestions, Harvey (1996) has proposed a multidimensional definition of trauma recovery. She suggests that the efficacy of an intervention depends upon its fit with the recovery environment and accordingly, provides an ecological model of trauma recovery that includes person, event and environmental factors. The present study is an initial test of some aspects of Harvey's model (see Figure 1) in the context of the recovery environment of officers in the New Zealand Police. It examines the main effects of some person, event and environment characteristics on PTSD symptom outcomes, taking into account whether the officers have ever attended a debriefing or not. The personal characteristics suggested by previous research (Smith & Ward, 1986) to have impact on behaviour in stressful situations in police officers are gender, length of service, membership of a branch of the service, and educational qualifications. The event characteristics are the number of traumatic events experienced (Vrana & Lauterbach, 1995). The environment characteristics are, the ease of talking about trauma at work, the attitudes to expressing emotion in the work place, social support from peers and from supervisors. These have been proposed in a related report (Stephens & Long, 1996) to be important recovery environment characteristics in an organisational context. These aspects of the model to be tested are shown in Figure 2 and according to this model it is predicted that the event, person and environment characteristics will show important effects on PTSD symptoms, and these effects will be moderated by the experience of debriefing.

Measures
1. Demographic variables. The variables included because of their demonstrated importance in relation to police stress and health outcomes were: age, length of service (in years), educational qualifications (none, secondary school or tertiary), gender, and the branch of the service (general duties or traffic and CIB).
2. PTSD. The Civilian Mississippi, a version of the Mississippi-PTSD (M-PTSD; Keane, Caddell & Taylor, 1986) was used to measure PTSD symptom scores during the past month across a 5 point Likert scale. Lower values indicate no evidence of PTSD symptoms and higher values indicate the presence of symptoms. A total score is obtained by summing across the scores on all items with a possible range of 35 to 175. The coefficient alpha estimate of reliability was .90. To identify respondents as PTSD cases, for use as discriminant groups, the least conservative cut-off score of 94 on the Military M-PTSD (Watson, 1990) was used. There has been little published work to date employing the Civilian version of the Mississippi Scale and less use of cut-off points to identify respondents with high scores as PTSD cases. A recent study resolved the problem, of the differences between the Military and Civilian versions of the Mississippi Scale. Eustace (1994) calculated the number of standard deviations from the mean, for the cut-off score of 94, on M-PTSD data from a New Zealand sample of Vietnam Veterans (Long, Chamberlain & Vincent, 1992). This number of standard deviations was applied to a community sample of respondents to the Civilian Mississippi. This was done with the approval of the author of the scale (T. M. Keane, personal communication, December, 1993). When this method was applied to the present sample, the resulting cut-off score was 96 (z = 1.07); the same cut-off score which Eustace had calculated using her 1994 sample. At the cut-off score of 96, 69 respondents were classified as PTSD cases (13.6%, N = 508). The least conservative score for classificatory purposes was chosen as being more conservative for the purpose of making statistical comparisons between groups on related variables.
3. Traumatic stressors. The traumatic stress schedule (Norris, 1990) was used as a basis for the collection of data on past events likely to be traumatic. The 9 items in this instrument were supplemented with 6 items relating specifically to police duties. The sum of the Events experienced while on duty as a police officer (range from 0-9) was used as the measure of traumatic stress.
4. Social support from peers and supervisors. Two scales of 4 items each from Caplan, Cobb, French, Van Harrison & Pinneau (1975) were used to measure perceived emotional social support. Each item was measured on a 5-point Likert scale and each set of 4 items was summed into an index of support from that source. The possible range of scores was from 1 to 5 with higher scores indicating stronger perceptions of support. The coefficient alpha estimates of reliability were .80 (peers) and .88 (supervisor).
5. The ease of talking about trauma at work. Two items in which respondents rated, on a 3 point scale, how easy it is to talk about traumatic experiences - details or feelings - at their work place. The items were combined into one measure named `Talk' with a coefficient alpha of .81. Possible scores ranged from 0 to 4, with a higher score indicating greater ease of talk.
6. Attitudes to expressing emotion at work. This scale (`Attitude') comprises four items that describe incidents related specifically to police work. The three possible responses to each item ranged from acceptance of the expression of personal emotions (score = 2), through avoidance techniques such as humour and more acceptable emotions such as generalised anger (score = 1), to physical avoidance and suppression of feelings (score = 0). Scores on each item were summed to provide an expression of emotion index for each individual. Higher scores indicate greater acceptance of the expression of personal emotions at work. The coefficient alpha estimate is .52.
7. Psychological support. Officers were asked whether they had attended a trauma policy debriefing or not.

Analysis
The statistical package, SPSS/PC (Norusis, 1992) was used to run the following analyses:
A factorial ANOVA was used to examine the effects of trauma and debriefing on PTSD symptoms. The N varies for each analysis owing to missing data on some variables.
Direct discriminant analyses were run to test the model by determining which set of variables would contribute the most to the variance in PTSD symptoms for those who had and those who had not experienced debriefing. To minimise loss of information in this multivariate analysis, missing data on each variable, except for the Debriefing and PTSD scores, was replaced with the sample mean for that variable (N = 507). Dummy dichotomous variables (Education1 and Education2) were created to test the level of Educational Qualifications.

The distribution statistics for both dichotomous and continuous variables are shown in Tables 1 and 2.

ANOVA
A factorial ANOVA (2 X 10) showed that there were significant differences on Civilian Mississippi mean scores between groups who had experienced different numbers of events, F (9, 497) = 5.94, p=<.001. There were no differences in Civilian Mississippi mean scores for those who had attended a debriefing, or for an interaction between Events and Debriefing. Table 3 provides the mean scores for the groups (No. of Events is compressed into 3 groups) and shows that Civilian Mississippi scores are higher as the number of events experienced while on duty increases. Although the means for those who had experienced a debriefing are slightly higher than for those who had no debriefing, these differences were not significant.

Discriminant Analysis
First Analysis. As an initial test of all predictor variables, a direct discriminant function analysis was performed using five demographic variables (Gender, Age, Service, Education and Branch), five recovery environment variables (Attitude, Talk, Social Support from peers and supervisors, and Debriefing), and Events as predictors of membership in two groups. Groups were those classified as PTSD cases (Civilian Mississippi score equal to or greater than 96; N = 69) and non-PTSD cases (Civilian Mississippi less than 96; N = 436). The discriminant function showed a significant association between the groups and predictors, X²(12) = 70.13, p<.001. The predictors together accounted for 13% of the variance (Canonical R = .36). The loading matrix of correlations between the predictors and the discriminant function (see Table 4) shows that the best predictors for distinguishing between the PTSD and non-PTSD cases were Talk, Events, Social Support from peers and supervisors, Attitude and Education1 (which compares no educational qualifications with secondary and tertiary qualifications). As the ANOVA analysis has shown, Debriefing adds nothing to the prediction. A Box's M test of homogeneity of variance-covariance matrices was significant at p<.001, which indicates that, with numerous DVs and discrepancy in cell sample sizes, the significance test may not be robust (Tabachnick & Fidell, 1989). Accordingly, a second test was performed to test the model using only the most important predictors.

Figure 1. An ecological model of psychological trauma, adapted from Harvey (1996).

Figure 2. The aspects of Harvey's (1996) model to be tested.

Second and Third Analyses. As a test of the model (see Figures 1 & 2), two direct discriminant function analyses were performed, using six variables (Talk, Events, Social Support from peers and supervisors, Attitude and Education1) as predictors of PTSD and non-PTSD cases. To include the possible moderating effect of debriefing that is described in the model, separate analyses were performed for those who had been debriefed (N = 117) and those who had not been debriefed (N = 390). The results were similar for both groups and are reported together, with those for the non-debriefed group in brackets. The discriminant function was again significant, X²(6) = 14.13, p<.05, (X²(6) = 58.09, p<.001). A Box's M test was non-significant at p<.05, indicating homogeneity of variance-covariance matrices, for both sets of variables. The predictors accounted for 12% (14%) of the variance (Canonical R = .34 (.37)). Table 4 shows the correlations between the predictors and the discriminant function for each of the three analyses. The contribution of Events to the discriminating function is notably reduced for those who have been debriefed. However the results of the ANOVA reported above show that there was no significant interaction between the effects of Events and Debriefing on Civilian Mississippi scores. Table 5 shows the differences between the mean Civilian Mississippi scores for each continuous variable in the discriminant functions and the results of a univariate F-test. The mean score for Attitudes, Talk and Social Support from peers and supervisors is significantly higher for those classified as non-PTSD cases. Those officers who had any educational qualifications were more likely to be classified as non-PTSD cases than those who had none, but this difference was not significant.

As predicted, the environment characteristics (but only education among the person characteristics) were related to PTSD symptoms, but these effects were not moderated by the experience of debriefing. The effects of the environmental variables were the same for the debriefed and non-debriefed groups. There were no main effects on PTSD symptoms of debriefing although the number of potentially traumatic events did have a significant effect on PTSD symptoms. The more events that an officer had experienced, the more likely they were to have higher symptom scores. When the effects of the number of events on symptoms was taken into account, along with a number of other demographic and environmental variables, debriefing did not predict whether an officer would be classified as a PTSD case or not. The best predictors of PTSD, as shown in Figure 3, were: the ease of talking about traumatic experiences at work; attitudes to expressing emotion in the work place; social support from peers; social support from supervisors; and whether officers had educational qualifications or not. The differences between those classified as PTSD cases or non-cases were such that, those with higher PTSD symptoms reported less social support from peers or supervisors, less positive attitudes to expressing emotion and less ease in talking about trauma at work. PTSD cases were more likely to have no educational qualifications in this sample.

Figure 3. The aspects of Harvey's (1996) model that were shown to predict PTSD in the present study.

Harvey (1996) suggests that the efficacy of post-trauma interventions depends on the degree to which they enhance person-community relationships and achieve ecological fit. Although debriefing aims to enhance social support and to encourage talk about trauma and the expression of emotions, it is possible that the provision of debriefing by health professionals from outside the organisation does not fit with the ongoing recovery needs of police officers who have experienced trauma. Police officers, who participated in the present study, expressed appreciation of the provision of support by the organisation and they also spoke of the value of talking about traumatic experiences. However, they also expressed a preference to talk with others who had shared their experiences in their own time and a resentment of compulsory group debriefings, as in the following example:

    Officer: Well, the guys might not feel like talking about it by then ...She got very upset during it, she was crying and that sort of thing, and I just don’t - she was very um annoyed with the guy, she didn’t feel like talking about it at the time and she felt like she was compelled to and he was sort of pressuring her into it.
    Interviewer: So you think it was more upsetting for her to actually have the debriefing.

    Officer: Yeah, from what I understand, people, they told me, is that they had to go round the room and talk about specific sorts of things and it wasn’t sort of voluntary, they sort of felt they had to actually say something, and as far as I was concerned if I was there and I didn't feel like doing it I wouldn't do that, I'd tell him to get stuffed.

This type of reaction is not uncommon in emergency service workers. Mitchell and Dyregrov (1993) warn that mental health professionals who are unknown to emergency workers may be treated with mistrust, resistance and anger. Burns and Harm (1993) speak of the reluctance of emergency nurses to participate in debriefings. Deahl, et al. (1994) suggest that the effectiveness of debriefing is enhanced if delivered by members of the organisation involved in the work, and others (e.g. Paton, 1994) recommend peer support models.

This study tests only part of Harvey's model and supports recent calls for more rigorous evaluation of debriefing methods, rather than providing definitive results. There are several other recovery environment and demographic variables that could be included in future tests of the model. For example, other variables suggested by Harvey (1996) are initial distress level of the victim, intelligence and personality. Raphael et al., (1995) include personal coping resources, cognitive impairments, past psychological morbidity and other life stresses. Additionally, recent research has shown that organisational factors, such as shift work or unfair work practices, have the most important impact on psychological outcomes for police officers and may interact with traumatic experiences (Sloan et al., 1994; Stephens, 1996). One notable limitation, in terms of the variables that may contribute to PTSD symptoms, is the measurement of trauma used here. The number of events, although it has been shown to have some impact on subsequent symptoms, does not include any indication of the severity or salience of the event to the person. This aspect along with others such as the frequency, severity and duration of the events, should be included in future studies and must be taken into account in the provision of debriefing or other forms of intervention (Busuttil, 1995). Furthermore, this study does not explore in any depth the interactions between event and person characteristics which the model proposes as mediators between the trauma and the response. Others (e.g. Atchison, 1995) would view these interactions as constituting the trauma itself. Finally, the use of PTSD symptoms as an index of recovery is seen by Harvey as useful but very limited. She proposes a multidimensional definition of recovery which the present study does not take into account at all. Another type of limitation to this study is the problems caused by comparing groups of unequal size. The use of the large scale survey, in an organisation in which events are similar in type and many officers for a variety of reasons are not included in debriefings, seems a useful approach to evaluation. In this situation, the selection of the participants for study could be made with specific attention to the two groups. Future explorations of this model must also take into account the development of symptoms across time.

Although limited, this use of Harvey's (1996) model has proved fruitful in demonstrating the importance of a number of aspects, other than debriefing, of the post-trauma environment. Social support and opportunities to talk about traumatic experiences and their emotional impact with others in the work place, have been shown to be related to PTSD symptoms. These findings support current suggestions (e.g. Busuttil, 1995; Gillham, 1995; Jones, 1995) that post-trauma interventions at work should be developed to fit with local requirements, take into account social environmental factors, and the ongoing needs of workers.

Atchison, M. (1995). Measuring stressors in post traumatic stress disorder. Paper presented at the Australasian Society for Traumatic Stress Studies Annual Conference, Hobart, Tasmania, Australia.

Bisson, J. I., & Deahl, M. P. (1994). Psychological debriefing and prevention of post-traumatic stress: More research is needed. British Journal of Psychiatry, 165, 717-720.

Brom, D., Kleber, R. J., & Hofman, M. C. (19930. Victims of traffic accidents: Incidence and prevention of post-traumatic stress disorder. Journal of Clinical Psychology, 49, 131-140.

Burns, C., & Harm, N. J. (1993). Emergency nurses' perceptions of critical incidents and stress debriefing. Journal of Emergency Nursing, 19, 431-436.

Busuttil, A. (1995). Psychological debriefing (Letter). British Journal of Psychiatry, 166, 676-677.

Caplan, R. D., Cobb, S., French, J. R. P., Jr., Van Harrison, R., & Pinneau, S. R., Jr. (1975). Job demands and worker health. Washington, D. C.: U.S. Department of Health, Education and Welfare.

Deahl, M. P., Gillham, A. B., Thomas, J., Searle, M. M., & Srinivasan, M. (1994). Psychological sequelae following the Gulf War: Factors associated with subsequent morbidity and the effectiveness of psychological debriefing. British Journal of Psychiatry, 165, 60-65.

Eustace, K. L. (1994). Cyclone Bola: A study of the psychological after-effects. Unpublished master's thesis, Massey University, Palmerston North, New Zealand.

Gillham, A. B. (1995). Psychological debriefing (Letter). British Journal of Psychiatry, 167, 407.

Griffiths, J. A., & Watts, R. (1992). The Kempsey and Grafton bus crashes: The aftermath. University of New England, Australia: Instructional Design Solutions.

Harvey, M. R. (1996). An ecological view of psychological trauma and trauma recovery. Journal of Traumatic Stress, 9, 3-23.

Hytten, K., & Hasle, A. (1989). Fire fighters: A study of stress and coping. Acta Psychiatrica Scandinavia, 355 (suppl.), 50-55.

Jones, L. (1995). Response to stress is not necessarily pathological (Letter). British Medical Journal, 311 (August 19), 509-510.

Keane, T. M., Caddell, J. M., & Taylor, K. L. (1986). The Mississippi scale. Boston, MA: VA Medical Centre.

Kenardy, J. A., Webster, R. A., Lewin, T. J., Carr, V. J., Hazell, P. L., & Carter, G. I. (1996). Stress debriefing and the pattern of recovery following a natural disaster. Journal of Traumatic Stress, 8, 37-50.

Long, N., Chamberlain, K., & Vincent, C. (1992). The health and mental health of New Zealand Vietnam war veterans with posttraumatic stress disorder. The New Zealand Medical Journal, 105, 417-419.

Mitchell, J. T., & Dyregrov, A. (1993). Traumatic stress in disaster workers and emergency personnel: Prevention and intervention. In J. P. Wilson & B. Raphael (Eds.), International handbook of traumatic stress syndromes. New York: Plenum Press.

Norris, F. H. (1990). Screening for traumatic stress: A scale for use in the general population. Journal of Applied Social Psychology, 20, 408-418.

Norusis, M. J. (1992). SPSS/PC+ Advanced statistics V5.O. Chicago: SPSS Inc.

Paton, D. (1994). Dealing with traumatic incidents in the workplace. (2nd ed.). Queensland Australia: Gull Publishing.

Raphael, B., Meldrum, L. & McFarlane, A. C. (1995). Does debriefing after trauma work? British Medical Journal, 310 (June 10), 1479-1480.

Shalev, A. Y. (1994). Debriefing following traumatic exposure. In R. J. Ursano, B. G. McCaughey & C. Fullerton (Eds.), Individual and community responses to trauma and disaster (pp.201-219). Cambridge: Cambridge University Press.

Sloan, I. H., Rozensky, R. H., Kaplan, L., & Saunders, S. M. A shooting incident in an elementary school: Effects of worker stress on public safety, mental health and medical personnel. Journal of Traumatic Stress, 7, 565-574.

Smith, B. L., & Ward, R. M. (1986). Maladaptive behaviors among military police personnel and implications for administrative policy. Journal of Criminal Justice, 14, 307-317.

Stephens, C. (1996). Organisational factors contributing to the psychological effects of traumatic stress in the police. In Health and Well-Being in a Changing Work Environment: Proceedings of the 1996 National Occupational Stress Conference. Brisbane, Australia: Academic press.

Stephens, C., & Long, N. (1996). Traumatic Stress in the New Zealand Police: Testing a theoretical approach to social support in the recovery environment. Manuscript submitted for publication.

Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics. (2nd ed.). New York: Harper & Row.

Thompson, J., & Solomon, M. (1991). Body recovery teams at disasters: Trauma or challenge? Anxiety Research, 4, 235-244.

Vrana, S., & Lauterbach, D. (1994). Prevalence of traumatic events and post-traumatic psychological symptoms in a nonclinical sample of college students. Journal of Traumatic Stress, 7, 289-302.

Watson, C. G. (1990). Psychometric post-traumatic stress disorder measurement techniques: A review. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 460-469.

Copyright

Comments to Psych.Webmaster@massey.ac.nz

Department of Psychology, Massey University , New Zealand

URL: http://www.massey.ac.nz/~trauma/

last changed Monday, 3 February 1997

Copyright © 1997 Massey University