Chapter 6: Synthesis and recommendations

This study was the first ever randomised controlled trial of resilience training for emergency services personnel. The goal was to assess the impact of this training in relation to Health and Wellbeing, Reactivity to Trauma and Workplace Functioning. Originally the full dataset was to be collected to 12-month follow-up, but only a subsample was collected to this time point. This chapter brings together what has been learnt from the analyses described above and makes recommendations for moving forward. In essence, a complete long-term follow-up—accessing the full cohort of participants—is warranted.

Major findings

The results of the analyses showed that overall there was no significant difference between the two conditions for resilience across all three domains at six-months. There was a non-significant trend, however, for those who received resilience training to be more likely to show no deterioration or an improvement in Workplace Functioning at six-month follow-up. Those in the resilience condition reported greater satisfaction with the training both at the post-training assessment and at six-month follow-up. Over time, however, the satisfaction of those in the resilience condition decreased, while the satisfaction of those in the control condition increased. We suggest regression to the mean as an explanation.

For the Reactivity to Trauma domain all recruits except two showed resilience, indicating that resilience in this domain is the norm. No significant differences were found between the conditions for drug and alcohol usage at follow-up. Resilience training was not found to have any statistically significant beneficial effects. However, considering the small sample size at 12-month follow-up and the possible moderate effect size of intervention, it is likely that the effects of this resilience training will be most evident when a full follow-up is permitted and at a time point further down the line. It should be noted that others in the field have noted that short follow-ups can sometimes make it difficult to detect effects of any kind (eg Guthrie & Bryant, 2005). Such longer term assessments of full cohorts would also address duty of care responsibilities for police service employees.

The relationship between the primary outcome variables and pre-program attributes

Breaking the global results down further, age was found to significantly correlate with depersonalisation, with older recruits having higher levels of depersonalisation. Older participants were also found to have lower levels of affective distress at six-month follow-up than younger participants. A significant relationship was also found between age and police services access, with older participants accessing a greater number of police services than younger participants.

There was a significant relationship between gender and personal accomplishment: males had significantly higher levels of personal accomplishment at six-month follow-up compared to females. No relationship was found between gender and affective distress or trauma symptomatology. This finding is in line with studies by both Carlier and colleagues (1997), and Hodgins and colleagues (2001), which found no relationship between gender and posttraumatic stress symptoms for police officers.

Those in a relationship at the pre-program assessment were found to have significantly higher levels of general health and personal accomplishment, and lower levels of substance involvement. Single recruits had significantly higher levels of affective distress and trauma symptomatology than those in a relationship.

Those at city stations experienced significantly higher levels of emotional exhaustion and depersonalisation. However, those in country stations reported significantly higher levels of affective distress than those in metropolitan stations.

The CD-RISC is a resilience measure that has previously been found to relate to trauma symptomatology. An examination of survivors of violent trauma found that those who scored more highly on the CD-RISC had less posttraumatic symptomatology and better general health than those with low scores (Connor, Davidson, & Lee, 2003). This finding was partially replicated for the current study. Resilience as measured by the CD-RISC was found to correlate with general health, and higher resilience scores on the CD-RISC to correlate with higher general health scores. CD-RISC score correlated with affective distress: higher resilience scores on the CD-RISC correlated with lower affective distress scores. CD-RISC score also correlated with relationship satisfaction, with greater relationship satisfaction correlating with higher CD-RISC score. However, CD-RISC score was not found to be related to posttraumatic symptomatology.

CD-RISC scores were found to correlate with all three types of burnout (depersonalisation, emotional exhaustion and personal accomplishment), with high scores on the CD-RISC found to be related to high levels of personal accomplishment, low levels of emotional exhaustion and low levels of depersonalisation. These results suggest that the CD-RISC may, in fact, to some degree be measuring burnout.

Currently, no study has been published that has examined the relationship between the CD-RISC and the MBI, or whether the CD-RISC has acceptable content validity. Future research should be directed in this area to ensure that the CD-RISC is in fact evaluating the domains that are claimed to be measured. This is recommended in light of previous research that found trauma measures to actually be a better predictor of burnout in the workplace than measures of post-trauma functioning (Devilly, Wright, & Varker, 2009).

Neuroticism (emotional instability) was found to be associated with general health, affective distress, substance involvement, relationship satisfaction, emotional exhaustion and personal accomplishment at six-month follow-up. For each of these relationships, lower neuroticism scores were correlated with more positive outcomes. No relationship was found between neuroticism and posttraumatic distress. This finding is in contrast to the large number of studies finding a link between these two factors (eg Breslau, et al., 1991; Charlton & Thompson, 1996; Davidson, et al., 1987).

The finding for the current study may be explained by the fact that the vast majority of participants reported extremely low levels of PTSD symptomatology, and many participants reported no trauma symptoms at all. It is most likely that the link between neuroticism and trauma symptomatology is only evident when trauma symptoms are at a detectable level.

Higher levels of optimism correlated with lower levels of affective distress, depersonalisation and emotional exhaustion at six-month follow-up. Those with higher levels of optimism also had higher levels of personal accomplishment. This finding is congruent with the existing literature linking optimism to a variety of positive outcomes, including faster recovery from surgery (Fitzgerald, et al., 1993; Scheier, et al., 1989), lower illness burden after natural disaster (Costello, 1998), less distress and fewer HIV-related concerns in gay men (Taylor, et al., 1992) and less distress in women following a failed IVF attempt (Litt, et al., 1992).

Trait anger was found to correlate significantly with general health, affective distress, emotional exhaustion, depersonalisation and personal accomplishment. A higher level of trait anger was associated with poorer outcome for each of these variables at six-month follow-up. No association was found between trauma symptoms and anger.

Perceived social support was found to correlate with all three types of burnout: depersonalisation, emotional exhaustion and personal accomplishment. Those who reported higher levels of social support reported lower levels of depersonalisation and emotional exhaustion, and higher levels of personal accomplishment. This finding is consistent with a number of other studies that found perceived social support to be negatively related to burnout (eg N. C. Brown, Prashantham, & Abbott, 2003; Greenglass, Fiksenbaum, & Burke, 1994; Koniarek & Dudek, 1996).

Perceived social support was found to correlate significantly with general health, affective distress, relationship satisfaction, and trauma symptomatology at six-month follow-up. This finding is consistent with the emergency services personnel trauma literature, which has found a robust relationship between poor, or few, social supports and trauma symptoms (Carlier, et al., 1997; Regehr, et al., 2000; Stephens, 1996).

Evaluation of the resilience training program

Overall, 37.9 per cent of those in the resilience condition and 32.9 per cent of those in the control condition were found to be resilient across all three domains of resilience (Health and Wellbeing, Reactivity to Trauma, and Workplace Functioning).

Health and Wellbeing

The domain of Health and Wellbeing was assessed by the four factors of affective distress, substance involvement, relationship satisfaction and general health. Overall, there was no significant difference between the resilience and control conditions for resilience for the Health and Wellbeing domain. Approximately half of the recruits demonstrated resilience in this domain (48.5% for the resilience condition, 56.7% for the control condition). Each of the four factors that comprised Health and Wellbeing are discussed in greater detail below.

No significant difference was found between the conditions for affective distress. The majority of recruits in both the resilience and control conditions exhibited either no change or an improvement in their levels of affective distress at follow-up, compared to the pre-program assessment. There was a significant decrease in level of affective distress over time, irrespective of condition. Overall, however, levels of affective distress were low at the pre-program assessment and remained low at six-month follow-up.

The conditions were not found to differ significantly in relationship satisfaction at six-month follow-up. Participants reported high relationship satisfaction at the pre-program assessment, and relationship satisfaction remained high across the two conditions at the follow-up assessment.

In terms of substance involvement, no significant difference was found between the resilience and the control condition at six-month follow-up. However, it should be noted that at the pre-program assessment there was a significant difference between the resilience and the control condition, with a higher percentage of the resilience condition members meeting criteria for total substance involvement use/abuse as compared to the control condition. At six-month follow-up, the percentage of people in the control condition who met criteria for total substance involvement use/abuse was approximately equal to the percentage of people in the resilience condition, meaning that the level of total substance involvement use/abuse increased at a greater level over time for those in the control condition.

No significant differences were found between the resilience and the control conditions for levels of alcohol involvement. At six-month follow-up there was a higher percentage of people in the resilience training condition who demonstrated no change/improvement in their substance involvement in comparison to the control condition, but this difference was not significant.

At six-month follow-up, 51.2 per cent of all participants reported total substance involvement scores which were at risk level (for either substance use/abuse or abuse/dependence), and 56.6 per cent of participants reported alcohol involvement scores which were at risk level (for either alcohol use/abuse or abuse/dependence). These findings support previous research showing a strong normalisation of alcohol consumption within the police service, and that police officers are more likely to drink alcohol at risk levels than members of the general public (eg Davey, et al., 2001; McNeill & Wilson, 1993).

The rates of alcohol consumption for the current study are comparable to a study of NSW police officers, which found 48 per cent of policemen and 40 per cent of policewomen consumed alcohol excessively (Richmond, et al., 1998). The figures reported in the current study are considerably higher than those for the general Australian population, with a 2007 National Drug Strategy survey of households finding that approximately 10.3 per cent of the population consumes alcohol in a way that is considered risky in the long term (Australian Institute of Health and Welfare, 2008).

Research of Queensland Police by Rallings (2000) found that hazardous drinking rates increased from 13 per cent to 22 per cent after commencement of police work. In a later replication study with Queensland new recruits, Rallings, Martin and Davey (2005) found a significant increase in the quantity and frequency of alcohol consumption from the time officers undertook initial training to when they had completed 12 months of operational duties. The percentage of officers who drank more frequently than once a month increased from 47 per cent to 60 per cent, and the percentage of officers who reported consuming six or more drinks once a month increased from 25 per cent to 32 per cent. These findings were replicated in the current study, with the percentage of recruits who drank alcohol at a risky level increasing from 31.8 per cent at the pre-program assessment, to 56.6 per cent at the six-month follow-up. These findings suggest that working as a police officer has a direct impact upon an individual’s substance involvement and alcohol consumption.

Across conditions, 80.75 per cent of recruits reported good general health at six-month follow-up, and general health was found to increase significantly over time, irrespective of condition. There was a trend for the general health of those in the resilience condition to be lower at the pre-program assessment than the general health of those in the control condition. At the follow-up assessment the general health of the resilience condition increased, and the general health of the control condition remained relatively static, to a point where both conditions had approximately equal levels of general health at the six-month follow-up assessment. This interaction was not significant, however; nor was the difference between the two conditions at the pre-program assessment.

As discussed previously, both the control condition and the resilience condition reported very low levels of affective distress (as measured by indices of stress, anxiety and depression) at both the pre-program assessment and at follow-up. Given that there are strong links between stress, anxiety and depression (affective distress), and physical ill health (L. R. Martin, et al., 1995; Rice, 1999), it is logical that those reporting low levels of affective distress would also report good levels of general health.

Reactivity to Trauma

The Reactivity to Trauma domain was only assessed at the follow-up time point, due to the fact that it could only be assessed once the recruits had had the opportunity to be exposed to potentially traumatic policing events. Preliminary analyses showed that there were no significant differences between the conditions in the number of different types of policing critical incidents that they were exposed to; nor was there a significant difference in the number of policing incidents rated as causing significant ‘distress at the time’. This indicates that both conditions were exposed to policing trauma in a similar way. Analyses revealed that there were no significant differences between the conditions in the level of trauma symptomatology and that both conditions reported extremely low levels of trauma symptomatology, with many recruits reporting no trauma symptoms at all.

Clinical classification scores were calculated in order to assess the number of people above the clinical cut-off for trauma symptomatology and the number of people below the cut-off. It was found that all recruits in the sample except two (people from the control condition) obtained trauma symptomatology scores below the clinical cut-off. That means that all but two participants demonstrated resilience for the Reactivity to Trauma domain. This finding adds support to recent research suggesting that resilience, when considered in terms of a lack of trauma symptoms following exposure to a traumatic event, may be the norm rather than the exception.

For example, in a study of September 11 survivors by Bonanno and colleagues (2006), resilience was considered to be having 0 or 1 PTSD symptom following exposure to the events of September 11. Using this definition, 65.1 per cent of participants (where n=2,752) were found to be resilient. In the current study, not all participants were exposed to a traumatic policing event, although it was the case for the majority (87.2%). Of those exposed to a traumatic event, approximately half (49%) rated the event as causing significant distress at the time. Yet of these people only two reported trauma symptoms at a level to meet criteria for PTSD. This finding supports the notion of resilience—in terms of trauma symptoms—being ‘ordinary magic’ (Masten, 2001) and replicates findings by those such as Bonanno and colleagues (2007), who found that the majority of people who are exposed to a traumatic incident do not go on to report trauma symptoms.

Of particular interest was the relationship between stressor exposure and trauma symptomatology. Our intervention was primarily aimed at providing serial approximation to traumatic events and desensitising the recruits to the ‘slings and arrows’ of police duty. Our results showed that, while there remained a strong positive relationship between stressors and symptoms for the control condition, this was greatly reduced or eradicated for the resilience condition. This would suggest that we have, at least, disrupted the relationship between stressor and symptom and would therefore expect trauma effects to become more apparent in the longer term.

Workplace Functioning

The domain of Workplace Functioning was measured in terms of burnout (emotional exhaustion, depersonalisation and personal accomplishment) and access to police and external help services. Burnout and access to services were only measured at follow-up, as these factors would be relevant and meaningful only after the recruits had been working as police officers for six months. Recruits were considered resilient in the domain of Workplace Functioning if they reported low-risk levels for at least two out of the three domains of burnout, and if they had not accessed police or external help services in the preceding six months. No significant difference was found between the two conditions for the domain of Workplace Functioning.

When the resilience and the control conditions were compared to both a clinical and a normative sample, both the resilience and the control conditions were found to have significantly lower levels of emotional exhaustion in comparison to the clinical sample, with the resilience condition having lower levels to a greater degree than the control condition. Those in the resilience condition also had significantly lower levels of emotional exhaustion in comparison to a normal sample.

For depersonalisation, both those in the resilience and those in the control conditions scored significantly lower than the clinical sample. In contrast, both the control condition and the resilience condition had significantly greater depersonalisation in comparison to the normative sample.

Both the resilience and control conditions were found to be significantly lower in personal accomplishment in comparison to the clinical sample, with the control condition reporting scores which were lower than the clinical sample to a greater degree than the resilience condition. Both the resilience and the control conditions reported significantly greater personal accomplishment than the normal sample.

Comparisons were made between the resilience condition and the control condition, using previously published risk cut-off scores, for each of the three domains of burnout. No significant differences were found between the two conditions for any of these domains.

The degree to which police services and external services were accessed in the preceding six months was also assessed. Those in the control condition were found to have accessed police services to a significantly greater degree than those in the resilience condition. There was also a significant difference between the conditions for the degree to which external services were accessed. Those in the control condition accessed a greater number of external services than those in the resilience condition.

These results suggest that those recruits who were provided with the resilience training may have learnt important skills and strategies that helped them to deal with workplace stress, which in turn led to a significantly lower degree of seeking police and external help services. These results may indicate that, rather than needing to seek professional advice, the recruits in the resilience condition were able to deal with problems on their own using the coping strategies that they were taught–or they talked to friends, family and colleagues to receive help and advice in dealing with any stressors or problems that they may have faced.

Alternatively, help-seeking behaviours may actually reflect good adjustment, and not seeking help may lead to poor outcomes in the longer term. This can only be assessed by a longer term assessment of outcomes in years rather than months. This is demonstrated quite clearly in the 12-month follow-up, which produced conflicting results (albeit based on a small number of participants). For now, we can only know that there is a short-term improved cost benefit for resilience training, in comparison to training as normal, through fewer help services being accessed.

In a study of resilience in physicians, Keeton, Fenner, Johnson and Hayward (2007) reverse-scored the emotional exhaustion domain of burnout, renamed it ‘emotional resilience’ and claimed that high levels of emotional exhaustion were representative of low levels of resilience. Although this approach has not been replicated by other researchers (and is not being advocated here), it does raise the question of whether resilience as conceptualised by some researchers may not in fact be representative of low levels of burnout. In the current study, high correlations were found between the CD-RISC measure of resilience and burnout, adding support to this argument.

Resilience across the three domains

When resilience was measured across a number of domains, the picture changed with the majority of recruits failing to display resilience for the domains of Health and Wellbeing and Workplace Functioning. Only approximately one-third of recruits in the resilience condition (37.9%) and one-third of those in the control condition (32.9%) showed resilience across all three domains of resilience. In areas such as affective distress, substance involvement, relationship satisfaction, general health, burnout and use of help services, the majority of recruits were shown overall to have deteriorated when their responses were compared to the responses given six months earlier, before they began working as police officers. This finding is consistent with the theory of Layne and colleagues (2007), who suggest that resilience is a multidimensional process and that an individual can display resilience in one domain and be non-resilient in another.

There was a significant difference between the conditions for how helpful they considered the training handouts to be, with those in the control condition considering the handouts to be more helpful than those in the resilience condition. There was no significant difference between the conditions in how helpful they considered the handbook to be, and those in the resilience condition reported that they practised the breathing and muscle exercises they were taught only to a very small degree.

There was no significant difference between the conditions in memory of the training modules, both immediately after the end of the training program and at six-month follow-up, although at both time points there was a non-significant trend for those in the resilience condition to have greater memory for the training content.

Intervention satisfaction

An assessment of participant satisfaction with the training content showed that those in the resilience condition had greater satisfaction with the training both at the post-training assessment and at six-month follow-up.

There was also a significant interaction between Time and Condition for participant satisfaction. At both time points, the satisfaction of those in the resilience condition was greater than the satisfaction of those in the control condition. However, satisfaction of those in the resilience condition decreased over time, while satisfaction of those in the control condition increased over time.

Similarly, and as would be expected given the findings for satisfaction, an interaction that was approaching significance (p=0.05) was found for the importance of the training content. Immediately after completion of the training program, those in both the resilience condition and control condition rated the importance of the training content approximately equally. However, at follow-up those in the resilience condition increased their rating of importance of the training, while those in the control condition decreased their importance rating. This result suggests that, once the recruits had been working for six months, those in the resilience condition realised (or at least ‘rated’) the importance of the training in light of their working experience. Those in the control condition did not consider the training they received to be as important once they were able to consider it in relation to their policing experience.

General exploration of the outcome variables

A significant relationship was found between trauma symptomatology and substance involvement (alcohol and drug use) at six-month follow-up, with those with higher substance involvement scores more likely to have higher trauma symptomatology. No relationship was found between pre-program substance involvement and follow-up trauma symptomatology. These findings add to the body of evidence suggesting that drugs and alcohol are used to self-medicate trauma symptoms (eg Davidson, et al., 1991; Kessler, et al., 1995; Reed, et al., 2007).

Significant differences were found between the two conditions in credibility and expectancy. Those in the resilience condition found the training to be significantly more credible than those in the control condition. They also had significantly greater expectancy than those in the control condition. Those who found the intervention more credible immediately after the program were found to report significantly lower levels of depersonalisation at six-month follow-up. Those who rated the intervention as more credible immediately after the program reported greater feelings of personal accomplishment at six-month follow-up.

A small correlation was also found between credibility and affective distress: those who found the training more credible experienced slightly higher levels of affective distress. Expectancy correlated with depersonalisation to a small degree, with those with greater expectancy having lower depersonalisation. Credibility and expectancy were not found to significantly correlate with any of the other major variables.

12-month follow-up subgroup

Exploratory analyses of 12-month follow-up data for a subgroup of 92 participants showed that some of the key findings from the six-month follow-up analyses persisted over time and that on other, important, indices the effects became more marked.

Within the domain of Health and Wellbeing there were no significant differences between the conditions for affective distress, relationship satisfaction, substance involvement and general health. However, if a full dataset had been collected, as planned, and the effect size differences were maintained, then there would have been significant differences in affective distress and relationship satisfaction—with the resilience-trained group being lower on affective distress and higher on relationship satisfaction. Likewise, there was no significant difference within the domain of Workplace Functioning (p<0.05). However, if the full sample had been assessed and the trend maintained, there would have been significant differences in the domain of workplace burnout, with emotional exhaustion being lower in the resilience group.

There were also no significant differences between the conditions for Reactivity to Trauma. However, as with the other indices, means displayed a trend for resilience superiority. Had a full complement of subjects been obtained, this would have reached significance. However, it should be borne in mind that participants in both conditions had very low levels of trauma symptomatology at 12-month follow-up. One member of the control condition met criteria for PTSD at 12-month follow-up.

Similarly to the six-month follow-up results, those in the resilience condition showed less of a relationship between stressor exposure (number of stressors and type of stressors) and trauma symptomatology, suggesting that the stressor–symptom disruption from the resilience training appeared to have persisted to this time point.

At 12-month follow-up 49.4 per cent of participants had total substance involvement scores at risk level (for either use/abuse or abuse/dependence), while 61.8 per cent of participants reported alcohol involvement at risk level (for either use/abuse or abuse/dependence).

Substance involvement was found to correlate moderately at 12-month follow-up with trauma symptomatology: those with higher substance involvement scores were more likely to have high trauma symptomatology scores.

General health levels were stable across time, with the majority of participants (80%) reporting good general health at 12-month follow-up.

Limitations of the study

This study contained several limitations which may have affected the results of the evaluation of the resilience training program. The first limitation was that the full sample follow-up was relatively soon (in six months) and would not be an accurate assessment of sustained change. Although 12-month follow-up data was collected for a subgroup, the number of participants in this group was insufficient to repeat the full host of resilience analyses that were conducted on the six-month follow-up data for the complete sample.

Recently, Layne and colleagues (2007) called for an increase in sophisticated methods for measuring risk and adaptation across multiple domains of functioning and recommended that more longitudinal studies be undertaken (although these authors also noted the formidable logistical challenges of such an approach). They suggested that study designs which comprise at least four waves of data collection over an extended period (often more than two years) may be needed to shed light on the mechanisms and processes that underpin positive adaptation.

Although a short follow-up was necessary in this study because of time constraints, there is an opportunity for future research to include assessment at a longer interval, such as at the five year time-point, to further investigate sustained change. It is expected that only in the longer term, once the police officers have been exposed to numerous high impact stressors, can a reliable test of the intervention be made.

A second limitation was the sample size. Although the sample was relatively large (n=281), a larger sample may have detected very small effects, such as those that may be expected to result from a training program. If all squads who had been two to three years at the police academy were included in a resilience training program, it could have the effect of changing the culture of an entire generation of police officers. This presents an interesting opportunity for future research.

Although it was originally planned that reliable change scores would be calculated for all of the major variables, this was not possible because test–retest reliabilities had not been reported for some of the measures, or for some of the measures the means and standard deviations associated with test–retest reliabilities had not been reported. This represented the third limitation of this study. Due to the absence of these statistics, cut-offs had to be used for several analyses instead. This meant that although the numbers of people who had changed categories could be observed and counted, the number of people who reliablychanged could not be accounted for in several instances.

Although limited in several ways, the results of this study are valuable and provide the first comprehensive development, implementation and evaluation of a resilience training program. The limitations presented by this study are common to training programs and, while researchers and practitioners need to be aware of them, their impact is not significant enough to discredit its findings.

Areas for future research

As previously mentioned, the present study did not comprise the very large sample size needed to detect very small effect sizes for a resilience training program. An area for future research would therefore be to conduct such a training program again, but using a sample of approximately 500.

Also mentioned above was the limited follow-up used for the current study. Effects only started appearing, as expected, at 12-months follow-up, but the data available to us came only from a very small sample (n=92). Longitudinal research provides us with valuable information, and the true test of the intervention would best be gauged by a longer term—for example, five or 10-year—follow-up of the complete sample.

An important area of research that is beyond the scope of the current study is the consideration of predictors of who will be best served by resilience training. Participants were assessed for a number of known vulnerabilities to developing trauma symptomatology and substance abuse at the pre-program assessment to ensure that each group was of equal composition at intake. A novel and exciting area for future researchers would be to look at whether certain characteristics make individuals more likely to display enhanced resilience following resilience training. This area of research could have particularly important implications for the emergency services and the armed forces, where it is important to enhance individuals’ resilience as much as possible. However, again, such an approach necessitates long-term assessments.

Summary and implications

Most studies of resilience to traumatic events consider it solely in terms of a single dimension of psychological function (eg trauma symptoms, depression or anxiety). Rarely is resilience considered in terms of a number of domains, although this approach was recommended by researchers such as Layne and colleagues (2007). In the current study resilience was considered in terms of three domains, so a much broader range of psychological measures was utilised. This has enabled resilience to be broken down and evaluated in terms of specific elements in a way that has never been undertaken before. This approach highlighted the fact that, as has previously been suggested (eg Masten, 2001), the vast majority of recruits were resilient to exposure to traumatic events. Only two participants, from the control condition, reported trauma symptomatology at six-month follow-up.

A non-significant trend was observed between the conditions in the number people who exhibited resilience for the domain of Workplace Functioning. Those in the resilience condition were more likely to show resilience for Workplace Functioning, suggesting that this domain received the most significant impact of the resilience training. Psychological injuries, which include occupational stress claims, make up 8 per cent of workers’ compensation claims in Australian Government agencies but 29.1 per cent of the total claim costs (Australian Government Comcare, 2007). The average lifetime cost of claims for psychological injuries sustained in 2005–06 for Australian Government premium-paying agencies was $115,000, compared to $27,000 for a non-psychological claim (Australian Government Comcare, 2006). Therefore, interventions that reduce psychological injuries can have a significant impact on the economy as well as the community.

It has been noted that intervention programs designed to enhance resilience tend to be based primarily on speculation, ‘pet’ theories, clinical experience and intuition. This carries a number of disadvantages, including the development of an intervention program that lacks adequate scope, effectiveness or efficiency or contains therapeutically inert or potentially harmful components (Layne, et al., 2007).

The current study represents the first time a resilience training program for emergency services personnel has been based upon solid, empirical evidence (Varker, Cook, & Devilly, In press), theories and recommendations (eg Foa & Rothbaum, 1998; Keyes, 1995; Kozak, et al., 1988; National Health and Medical Research Council, 2001; Rapee, 1985). This resilience training program has been shown to contain no noxious or harmful elements, and the modular design means that it can be used for a wide range of professions in a variety of settings. The training program is also a manualised intervention, meaning that it can be disseminated with fidelity, evaluated rigorously and replicated by other independent researchers.

The study found that at six-month follow-up 51.2 per cent of all participants reported total substance involvement scores at risk level (for either substance use/abuse or abuse/dependence) and 56.6 per cent reported alcohol involvement scores at risk level (for either alcohol use/abuse or abuse/dependence). In light of such high percentages, policies and procedures must be put in to place to identify and support those with either substance or alcohol use problems. There must be a clear, comprehensive substance use policy that is widely known about and equitably applied to all (P. Martin, Davey, & Mann, 1998) that clearly outlines employee and employer responsibilities regarding alcohol and drugs. Procedures for dealing with specific circumstances of misuse are essential first steps in providing support (Mann, 2006).

Supervisors would benefit from training in identifying employees who may be at risk of substance use, in recognising emerging problems and in developing strategies for timely referral to support services (Mann, 2006). In 2007 it was announced that Victorian police officers would be subject to routine testing for alcohol and drug use after critical incidents such as police shootings or high-speed chases that result in injuries, and that the Chief Commissioner would be given the power to order tests to protect the ‘good order or discipline of the force’ (Silvester, 2007). This opens the door for possible targeted or random testing—including of whole squads or stations. Given that the stakes are so high in terms of officers’ professional careers and health, it is important that these high levels of drug and alcohol consumption be addressed by Victoria Police. The current resilience study during academy training suggests beneficial effects on workplace burnout, negative affect, relationship satisfaction and trauma symptomatology, which appear to be increasing in strength over time.