Prevalence of post-deployment violence in the military

In a recent study, researchers surveyed US troops 3 and 12 months following return from deployment in Iraq using a similar question to that used in our study about getting into a fight with someone and hitting the person (Thomas et al. Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2010). They found that 17.7% reported interpersonal violence at 3 months and this remained roughly the same at 18.4% at 9 months later (Thomas et al. Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2010). Considering their use of a similar self-report measure of physical violence over a similar time-frame, the prevalence of violence post-deployment in our study appears low in comparison. A number of factors may make an impact on the level of reported violence among US troops such as the increased average length of deployment (1 year for the US military compared with 6 months in the UK military), an increased proportion of troops experiencing combat exposure or more traumatic combat experiences (Fear et al. Reference Fear, Jones, Murphy, Hull, Iversen, Coker, Machell, Sundin, Woodhead, Jones, Greenberg, Landau, Dandeker, Rona, Hotopf and Wessely2010).

It is of note that the average age of our study population was 32 years (37 years for officers and 32 years for other ranks), which is slightly older than expected for the average age of regular personnel in the UK military (37 years for officers and 29 years for other ranks) (DASA, 2003). It should therefore be acknowledged that, given that violent behaviour is associated with younger age, this age difference may lead to a slight underestimation of the true prevalence of homecoming violence in UK military personnel.

Sociodemographic and pre-military factors

In this study the demographic factors associated with violence on homecoming were the same as those known to be associated with antisocial behaviour in the general population. Pre-enlistment antisocial behaviour was found to have the strongest association with post-deployment violence indicating that, from a life-course perspective, those already demonstrating antisocial behaviour prior to joining the military are more likely to continue on this trajectory. We have previously shown that self-reported pre-enlistment antisocial behaviour is associated with later negative behavioural outcomes in military personnel, including aggression, alcohol abuse and dangerous driving (MacManus et al. Reference MacManus, Dean, Iversen, Hull, Jones, Fahy, Wessely and Fear2011). This is an important potential source of confounding, as it is well known that recruitment to the military in the UK is not random. Historically, many recruits have come from socially disadvantaged inner-city areas where rates of social problems, including crime, are high (Johnstone, Reference Johnstone1978). Some studies from the USA have demonstrated that pre-adult antisocial behaviour predicted antisocial behaviour in the military as well as or better than combat trauma or wartime experiences (Resnick et al. Reference Resnick, Foy, Donahoe and Miller1989; Fontana & Rosenheck, Reference Fontana and Rosenheck2005). In this study we show that even though pre-enlistment antisocial behaviour had a stronger association with violence on homecoming than all other variables, when we controlled for the confounding effect of pre-enlistment antisocial behaviour, combat exposure and traumatic experiences during deployment were still strongly associated with violence following deployment.

Military factors

Violence on homecoming from deployment was higher for Army personnel. It is noteworthy that the reporting of violence on homecoming from deployment was more common in those who had left the military compared with those still in service. Woodhead et al. (Reference Woodhead, Rona, Iversen, MacManus, Hotopf, Dean, McManus, Meltzer, Brugha, Jenkins, Wessely and Fear2011) also recently reported that post-National Service veterans (largely of the Cold War era) report more violent behaviours than non-veteran males of the same age. On first reading this may not be surprising, since there has been considerable recent media attention regarding the problems faced by ex-service personnel around reintegration back into ‘civvy’ life, such as mental health problems, substance misuse, unemployment, relationship problems and violent crime (King, Reference King2009; James, Reference James2010). However, the questionnaire used in the present study asked participants to report on violence in the immediate post-deployment period, i.e. in the weeks following homecoming, not after leaving the services. The time lag between returning from deployment until completing the questionnaire could vary from 1 month to a maximum of 2 years. There are, therefore, several interpretations of our finding. First, if the observation is accurate, then it could be the result of confounding. Violence on homecoming from deployment and leaving the services could both be associated with a common factor, of which the most likely candidates are pre-enlistment factors such as social adversity. Second, it may be the result of reverse causality, i.e. that violence on homecoming from deployment led to premature leaving (including being compulsorily discharged from the military). Third, the observed results may be the result of recall bias. If ex-service personnel are going through a difficult transition back into civilian life after leaving the armed forces, experiencing more social adversity and engaging in more violent behaviour, they may be more inclined to report violence in the immediate post-deployment period too. Finally, a different recall bias may be operating if those who are still serving perceive it as potentially stigmatizing or detrimental to their career to admit to violent behaviour (even though the study was conducted independently of the military). However, we did control for serving status, i.e. whether personnel were still serving or had left the military, in all the analyses.

Combat experiences

Serving in a combat role during deployment, discharging a weapon in direct combat, experiencing multiple traumatic events and having thoughts that one might be killed were strongly associated with violence on homecoming from deployment. Studies of US Vietnam and Gulf war veterans have shown an association between violent and criminal behaviour in military personnel and combat experience (Yesavage, Reference Yesavage1983; Yager et al. Reference Yager, Laufer and Gallops1984; Black et al. Reference Black, Carney, Peloso, Woolson, Letuchy and Doebbeling2005; Killgore et al. Reference Killgore, Cotting, Thomas, Cox, McGurk, Vo, Castro and Hoge2008). Booth-Kewley et al. (Reference Booth-Kewley, Larson, Highfill-McRoy, Garland and Gaskin2010b ) published a recent paper which aimed to examine factors associated with antisocial behaviour in combat veterans using a sample of Marines enlisted in the US armed forces who had been deployed to Iraq or Afghanistan between 2002 and 2007. They found that combat exposure was positively and significantly associated with antisocial behaviour after adjustment for a range of potential confounders. Killgore et al. (Reference Killgore, Cotting, Thomas, Cox, McGurk, Vo, Castro and Hoge2008) surveyed 1252 Operation Iraqi Freedom veterans regarding different combat experiences immediately on return from deployment and again 3 months later. They controlled for the confounding effects of age, gender and other relevant sociodemographic factors. They found that exposure to combat and high levels of trauma were predictive of greater risk-taking propensity after homecoming. They also found that these combat experiences were predictive of actual risk-related behaviours in the preceding month, including greater alcohol use and increased verbal and physical aggression towards others. This is consistent with the idea that some combatants may develop an ‘invincibility complex’. Anecdotally, military personnel have come back from war with feelings of invincibility. These may be evidenced through increased risk taking, binge drinking, drug use, getting into fights and other antisocial behaviours (Vaughan, Reference Vaughan2006). However, more research is required to help determine whether this is mainly a product of their combat experiences or pre-existing characteristics of the individual.

Armed forces personnel are of course trained to inflict violence when necessary, albeit in a controlled, disciplined fashion. These individuals go through specific training to develop the necessary skills and strengths required for combat, such as ‘targeted aggression’. The results from this study emphasize the association of combat and combat trauma with aggressive behaviour in military personnel post-deployment and suggest that these experiences have an association over and above that of previous aggressive behaviour or socio-economic factors. One problem, however, is that combat is not random. Those in combat roles are invariably younger, for example. It is also possible, and indeed probable, that those in combat roles are more likely to be risk takers, and perhaps have a greater propensity for violence. Educational levels may also differ. In the UK, regiments that are traditionally and for good reason associated with an increased aggressiveness, such as the Parachute Regiment, and who are as a result more often deployed on dangerous combat missions, are also more likely to recruit from areas of social disadvantage and where the recruits are likely to have lower educational attainment. Indeed, we do know that those with a history of pre-enlistment antisocial behaviour are more likely to perform in a combat role on deployment (MacManus et al. Reference MacManus, Dean, Iversen, Hull, Jones, Fahy, Wessely and Fear2011) and whether this is through self-selection or being selected by their seniors/trainers we cannot comment. Some confounders, such as age, can be dealt with statistically, but others, such as temperament, risk taking and propensity to violence, are less amenable to analysis. The possibility that residual confounding accounts for the association between combat exposure and offending remains a real one.

Symptoms of PTSD, common mental disorders and heavy alcohol use

We also found that symptoms of PTSD and common mental disorder were strongly associated with post-deployment violence even when all other factors were controlled for. Whilst it has been suggested on one hand that combat exposure is associated with aggression partly due to the reinforcement and modelling of violence in the military (Gimbel & Booth, Reference Gimbel and Booth1994), several studies have indicated that combat exposure is associated with aggression primarily through its relationship with PTSD symptoms (Byrne & Riggs, Reference Byrne and Riggs1996, Orcutt et al. Reference Orcutt, King and King2003). A recent meta-analysis revealed a strong relationship between PTSD and anger and hostility among trauma-exposed adults (Orth & Wieland, Reference Orth and Wieland2006) and combat exposure showed a stronger association with PTSD compared with other traumatic events. This suggested that PTSD is an important mediator in the association between combat (trauma) and antisocial behaviour. It has been consistently shown that male veterans with PTSD report higher rates of violence and aggressive behaviour than those without PTSD (Kulka et al. Reference Kulka, Schlenger, Fairbank, Hough, Jordan, Marmar and Weiss1990; Lasko et al. Reference Lasko, Gurvits, Kuhne, Orr and Pitman1994; Beckham et al. Reference Beckham, Feldman, Kirby, Hertzberg and Moore1997; Taft et al. Reference Taft, Vogt, Marshall, Panuzio and Niles2007b ). In our study, 14.6% of those who reported violence on homecoming also reported symptoms of PTSD in the month prior to questionnaire completion; this is in comparison with 2.5% of those who did not report violence, suggesting almost a five times increased risk of violence in those with PTSD symptoms compared with those without.

So it appears that there is a relationship between PTSD and violence among military personnel. However, this link is not straightforward. The concept of PTSD is complex and studies have found different aspects of the symptomatology to be specifically related to post-deployment violence, such as hyperarousal (McFall et al. Reference McFall, Fontana, Raskind and Rosenheck1999; Taft et al. Reference Taft, Vogt, Marshall, Panuzio and Niles2007b ), avoidance/numbing symptoms (McFall et al. Reference McFall, Fontana, Raskind and Rosenheck1999), or co-morbid dysphoria (Taft et al. Reference Taft, Vogt, Marshall, Panuzio and Niles2007b ). Elbogen et al. (Reference Elbogen, Wagner, Fuller, Calhoun, Kinneer and Beckham2010) explored the correlates of anger and hostility in Iraq and Afghanistan war veterans. They found that aggressive impulses or urges, difficulty managing anger and perceived problems controlling violent behaviour were each significantly associated with PTSD hyperarousal symptoms. Other PTSD symptoms were less strongly and less consistently linked to anger and hostility (Elbogen et al. Reference Elbogen, Wagner, Fuller, Calhoun, Kinneer and Beckham2010).

Fear et al. (Reference Fear, Jones, Murphy, Hull, Iversen, Coker, Machell, Sundin, Woodhead, Jones, Greenberg, Landau, Dandeker, Rona, Hotopf and Wessely2010) recently showed that the prevalence of PTSD in UK military personnel who had been deployed to Iraq or Afghanistan (since 2003) was low, only 4.0%. This suggests that the amount of violence in the UK military population which could be attributable to PTSD (if causality were to be assumed) would be low. Of course, given the cross-sectional nature of the data collection, we cannot assume causality nor can we dismiss the possibility of reverse causality, i.e. the risk of PTSD symptoms following violent behaviour. The authors are unaware of any studies to date that address the question of causality. In order to do this, prospective data are required to establish the temporal sequence of the combat trauma, development of PTSD and subsequent violent behaviour. All of the studies to date have collected data cross-sectionally and cannot therefore address this issue.

In contrast to the low rates of PTSD, Fear et al. (Reference Fear, Jones, Murphy, Hull, Iversen, Coker, Machell, Sundin, Woodhead, Jones, Greenberg, Landau, Dandeker, Rona, Hotopf and Wessely2010) found that 19.7% of UK military personnel who had been deployed in Iraq or Afghanistan reported symptoms of common mental disorders. In our study, 37% of those who reported violence reported symptoms of common mental disorder in comparison with 16.7% of those who did not report violence (odds ratio 2.90). Given the estimated prevalence of symptoms of common mental disorder in the deployed UK military personnel, such symptoms may be important to consider in the assessment of post-deployment violence. Booth-Kewley et al. (Reference Booth-Kewley, Highfill-McRoy, Larson and Garland2010a ) recently published findings from a longitudinal study of US Marines which found that one of the strongest predictors of antisocial behaviour by military personnel was a psychiatric diagnosis and the relationship was even stronger if that diagnosis was given post-combat.

Similarly, we have demonstrated a strong association between heavy alcohol consumption and post-deployment violence. Almost half of those who reported violence also reported heavy alcohol use in the last year compared with 14.1% of those who did not report violence. Fear et al. (Reference Fear, Iversen, Meltzer, Workman, Hull, Greenberg, Barker, Browne, Earnshaw and Horn2007) previously showed that excessive alcohol consumption is more common in the UK armed forces than in the general population. We know from a large body of research in non-military populations that alcohol misuse is an important risk factor for violence (Swartz et al. Reference Swartz, Swanson, Hiday, Borum, Wagner and Burns1998). This has been shown to be the result of a rather complex interaction of biochemical, psychological, situational and cultural factors. Recent research has shown that risk-related behaviours including both heavier alcohol use and increased verbal and physical aggression towards others were associated with previous combat experience (Killgore et al. Reference Killgore, Cotting, Thomas, Cox, McGurk, Vo, Castro and Hoge2008). This suggests that the impact of military service on subsequent aggression and violence can be mediated indirectly via increased alcohol consumption but that it can also have a direct effect through combat exposure.

Strengths and limitations

The major strength of this study is that it utilized data from a large randomly selected sample of UK military personnel. It is the only study of its kind in the UK and the comprehensive range of information collected enables the exploration of a wide range of factors of interest to the study of aggression and violence in this group of individuals.

This study is based on data obtained from a self-report measure of violence, which introduces the potential for misclassification error. Similarly, the self-report measure of pre-enlistment antisocial behaviour is prone to recall bias. However, while all methods of violence measurement have their limitations with regard to measurement error, self-report has been shown to capture a greater prevalence of violence than other methods (Lidz et al. Reference Lidz, Mulvey and Gardner1993). An objective measure such as violent offending records would have been a useful additional outcome measure. However, offending records exclude violence that did not come to the attention of the police and therefore these measures may underestimate the problem. Similarly, offending records may not reflect childhood or adolescent antisocial behaviour. Therefore, while recognizing their limitations, self-report measures are useful and informative estimates of both post-deployment violence and pre-enlistment antisocial behaviour in this study. The violence outcome measure was also limited, as it did not record details about the violence other than whether it occurred or not. It did not record the level of severity of the violence, the context of the violence, or whether or not it was associated with alcohol intoxication.

The response rate of the military study was 61% and factors associated with non-response were being male and of younger age. Both of these factors are known to be associated with aggression and violence, which suggests that the prevalence of violence identified in this study may be an underestimation of the true prevalence in the reference population. The number of participants who did not answer the violence questions (n=319, 6.5%) is also a potential source of bias. However, we repeated our analyses to examine the possible extent of the bias and there was no notable change to the results.

Finally, cross-sectional data have a number of inherent limitations and we are unable to draw any causal inferences from our analyses, particularly where reverse causality cannot be excluded, as in the case of mental health status and post-deployment violence. Prospective data from the ongoing cohort study will enable these issues to be resolved.