RESEARCH REPORT

doi:10.1111/j.1360-0443.2009.02565.x

Drugs and violent death: comparative toxicology of homicide and non-substance toxicity suicide victims
Shane Darke1, Johan Duou2 & Michelle Torok1
National Drug and Alcohol Research Centre, University of New South Wales, Australia1 and Department of Forensic Medicine, Sydney South West Area Health Service, School of Medical Sciences, University of New South Wales and Department of Pathology, University of Sydney, NSW, Australia2

ABSTRACT Aims To determine the comparative toxicology of death by homicide and suicide by means other than substance toxicity. Design Cross-sectional (autopsy reports). Setting Sydney, Australia. Cases A total of 1723 cases of violent death were identied, comprising 478 homicide (HOM) cases and 1245 non-substance toxicity suicide (SUI) cases. Findings Substances were detected in 65.5% of cases, and multiple substances in 25.8%, with no group differences. Illicit drugs were detected in 23.9% of cases, and multiple illicit in 5.3%. HOM cases were signicantly more likely to have an illicit drug [odds ratio (OR) 2.09] and multiple illicits (OR 2.94), detected, HOM cases being more likely to have cannabis (OR 2.39), opioids (OR 1.53) and psychostimulants (OR 1.59) present. HOM cases were, however, signicantly less likely to have benzodiazepines (OR 0.53), antidepressants (OR 0.22) and antipsychotics (OR 0.23) present. Alcohol was present in 39.6% of cases (median blood alcohol concentration = 0.12), with no group difference in prevalence. Conclusions The role drugs play in premature death extends far beyond overdose and disease, with illicit drugs associated strongly with homicide. Keywords Dependence, homicide, mortality, suicide, toxicology, violence.

Correspondence to: Shane Darke, National Drug and Alcohol Research Centre, University of New South Wales, NSW 2052, Australia. E-mail: s.darke@unsw.edu.au Submitted 11 August 2008; initial review completed 16 December 2008; nal version accepted 6 February 2009

INTRODUCTION Substance-dependent populations have substantially higher mortality rates than matched peers [1,2]. Alcohol dependence, for example, is associated with an estimated twofold increase in all-cause mortality [1]. Similarly elevated rates have been reported for other substances, including benzodiazepines (twofold) and opioids (13-fold) [1,3]. Of particular note here, across all substances excess mortality rates due to unnatural causes far exceed those due to natural causes [1]. The major causes of death associated with substance dependence are disease [e.g. human immunodeciency virus (HIV), hepatic cirrhosis] and overdose (particularly among opioid users) [4]. Signicant proportions of substance user deaths, however, are due to violence, whether perpetrated by others upon the person (homicide) or by the person upon themselves (suicide) [4]. The reasons for such elevated rates of violent death differ. In terms of suicide, substance-dependent populations are known to

have high levels of suicide risk factors, including psychopathology, childhood abuse, social isolation and unemployment [4]. There are also the proximal effects of intoxication that may increase the likelihood of an attempt [5]. Elevated rates of homicide also may be due to the proximal effects of the drug in question (e.g. disinhibition from alcohol intoxication, paranoia from psychostimulant use) that may lead to circumstances of violence [68]. Illicit drug use also increases the distal risk of violence, as high levels of crime are performed to support such use or to protect drug-dealing networks [9]. To date, relatively few studies have examined levels of drugs among cases of homicide [1014] or completed suicide [1519]. Among both homicide and suicide victims, psychoactive substances are present in more than half of cases. Among homicide victims, the most commonly seen substances appear to be alcohol and illicit drugs [1014]. Among suicide cases, alcohol is also by far the most commonly seen substance, and illicit drugs are also seen in large proportions of cases, but
Addiction, 104, 10001005

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

Drugs and violent death

1001

pharmaceuticals such as benzodiazepines or antidepressants appear to be seen more commonly than in homicide [1519]. It should be noted, however, that the global gures from suicide studies include deaths due to selfadministered drug overdose, which may be of licit substances, such as antidepressants, or illicits such as heroin. While casecontrol studies indicate an elevated risk of violent death among chronic substance users [20], no study to date has examined the comparative toxicology of drugs in deaths due to violence per se, and no study has compared directly the toxicology of drugs in these different forms of violence. The focus here is the toxicology of violence: by denition, a deliberate overdose of a drug will produce positive toxicological ndings. Such deaths involve a series of different questions than those studied here, such as the comparative toxicology of deliberate and accidental overdose. It is the extent of the involvement of substances in deaths that are due solely to acts of violence that is at question here. Thus, to what extent are substances involved in violent death not due to poisoning? As such, deliberate overdose is not included in this analysis of violent behaviours. It is worth noting in passing that dependent substance users do not typically use their drug overdose as a means of suicide, and violent means are frequently employed [4]. The current study aimed to examine the toxicology of death by violent means across a 10-year period, and to compare the toxicology of violent suicide and homicide to determine the comparative roles of licit and illicit substances. For the purposes of the current study, violent suicide included suicide by: hanging/asphyxia, falling from a height, gunshot, cuts/stabbing, drowning, jumping in front of a vehicle, re and electrocution. Deaths due to substance toxicity (i.e. self-administered drug overdose, self-poisoning, gassing) were excluded. Specically, the current study aimed to: 1 Determine the toxicology of death by violent means; and 2 Determine the comparative toxicology of death by homicide and violent suicide.

METHODS Case identication Autopsy reports and police summaries of all nalized cases of homicidal and suicidal death aged between 15 and 60 years who underwent autopsy at the New South Wales (NSW) Department of Forensic Medicine between 1 January 1997 and 31 December 2006 were retrieved. This age range was selected as it represents the range that encompasses almost all illicit drug use [21]. All cases of death due to murder or manslaughter were included in the homicide group (HOM). All cases of death due to

suicide by violent means were included in the suicide (SUI) group, with deaths due to substance toxicity (i.e. self-administered drug overdose, poisoning, gassing) excluded. By way of providing a context, the homicide rate in Australia is 1.5 per 105, with stabbing (34%), blunt force injury (28%) and gunshot (14%) the most common causes of such deaths [22]. The suicide rate is 11.2 per 105, with hanging (51%), self-poisoning (16%; drugs 12%, gassing/other poisons 16%) and gunshot (7%) being the most common methods [23]. The NSW Department of Forensic Medicine is located in central Sydney, and is the primary forensic pathology centre in NSW, conducting approximately 2000 autopsies per year. Permission to inspect the les was received from the Sydney South West Area Health Service human research ethics committee. Case lists were generated by J.D. All cases were reviewed by S.D. and M.T., with specialist comment on the toxicology and circumstances of death provided by J.D. In NSW a case must be reported to the Coroner where a person dies a violent or unnatural death. It is mandatory for all cases of suspected suicide and homicide to be reported to the Coroner for medicolegal death investigation. All such cases, including all those presented in this study, undergo a standardized forensic autopsy, including full quantitative toxicology and histology, as part of the Coroners investigation. Cause of death is determined by the forensic pathologist on the basis of circumstances of death, the comprehensive autopsy ndings and the toxicological analyses. Specic data retrieved from the autopsy and police reports included: cause of death, demographic characteristics, location of fatal incident, time and date of the fatal incident and the toxicological ndings for each case. Quantitative toxicological analysis of all major drug groups is performed in all non-natural deaths. Toxicological data were reported for blood alcohol concentration (BAC), cannabis [determined by the presence of D-9tetrahydrocannabinol (THC) and/or THC acid], morphine (the major metabolite of heroin), methadone, methamphetamine, cocaine (determined by the presence of cocaine itself and/or the presence of benzoylecgonine, the major metabolite of cocaine), 3,4-methylenedioxymethamphetamine (MDMA), benzodiazepines, antidepressants and antipsychotic medications. For the purpose of analysis, the presence of illicit substances was dened by the presence of any of the following: cannabis, morphine, methamphetamine, cocaine or MDMA. It was not possible from toxicology to determine whether the use of pharmaceutical was prescribed, illicit or in excess of the prescribed dosage. All presented toxicological analyses were of blood, and quantitative data are presented only for alcohol. In cases where there was prolonged hospitalization prior to death, antemortem toxicology was reported where available, otherwise toxicology was not
Addiction, 104, 10001005

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

1002

Shane Darke et al.

Table 1 Case characteristics of deaths due to homicide and violent suicide. Homicide (n = 478) Age (years) Gender (% male) Married/de facto (%) BMI Cause of death (%) 35.9 72.2 35.3 24.7 Cuts/stabbing (34.9) Blunt force injury (29.4) Gunshot (26.1) Strangulation (7.7) Drowning (2.1) Fall from height (1.7) Fire (1.3) Hanging/asphyxia (0.8) Poison (0.6) Suicide (n = 1245) 36.2 80.7 26.9 24.7 Hanging/asphyxia (65.2) Fall from height (17.7) Gunshot (5.2) Cuts/stabbing (3.9) Drowning (3.4) Train (3.3) Fire (1.2) Electrocution (0.8) All (n = 1723) 36.1 78.4 28.9 24.7

BMI: body mass index.

reported. In all cases, drugs administered by hospital and medical staff were excluded. Police summaries were available for all cases. Statistical analyses Where distributions were highly skewed, medians and inter-quartile ranges (IQR) were reported, otherwise means were presented. For bivariate comparisons, t-tests or odds ratios (OR) with 95% condence intervals (CI) were reported. In order to determine whether group (HOM versus SUI) was an independent predictor of the presence of substances, a series of logistic regressions were conducted, controlling for age and gender. All analyses were conducted using SPSS for Windows (release 14.0) [24].

were hanging/asphyxia, falling from a height and gunshot, with multiple causes noted in 0.7% of cases. Toxicology Global Toxicology was available for 1693 cases (HOM 467, SUI 1226). Antemortum blood results were reported in 180 cases (HOM 109, SUI 71). Substances were detected in two-thirds of cases and multiple substances in a quarter, with no group differences (Table 2). Logistic regression indicated that group remained a non-signicant predictor of the presence of a psychoactive substance, with younger age (adjusted OR 0.99, CI 0.981.00) and being male (adjusted OR 1.41, CI 1.111.79) being independent predictors. Alcohol

RESULTS Cases A total of 1723 cases were identied, comprising 478 HOM cases and 1245 SUI cases (Table 1). The mean age 36.1 years [standard deviation (SD) 11.5, range 1560 years], with no group (HOM 35.9 versus SUI 36.2 years) or gender (males 35.9 versus females 36.6 years) differences. More than three-quarters were male, with the SUI group having a slight, but signicantly higher, proportion of males (OR 1.64, CI 1.282.08). A signicantly higher proportion of the HOM group was married/de facto (OR 1.62, CI 1.262.06). The mean body mass index (BMI) was 24.7, with no group difference. The most common causes of death among the HOM group were cuts/stabbing, blunt force injury and gunshot, with multiple causes noted in 6.7% of cases. The most common causes of death among the SUI group Alcohol was present in 39.6% of cases, with no group difference. Alcohol was more common among males (42.9 versus 27.4%, OR 1.99 CI 1.542.56). The median BAC among alcohol-positive cases was 0.12 g/100 ml (IQR 0.12, range 0.010.41 g/100 ml), with no gender difference (0.12 versus 0.12 g/100 ml). Among alcoholpositive cases, BACs were signicantly higher among the HOM group (0.14 versus 0.11 g/100 ml, U = 38,228.5, P < 0.001). Logistic regression indicated that SUI group membership remained a non-signicant predictor of the presence of alcohol, with being male (adjusted OR 2.06, CI 1.59 2.66) the only signicant predictor. Illicit drugs Illicit drugs were detected in a quarter of cases, and multiple illicit drugs in a twentieth of cases (Table 2). HOM
Addiction, 104, 10001005

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

Drugs and violent death

1003

Table 2 Comparative toxicology of homicide and violent suicide. Homicide (n = 467) % Global Substance detected Multiple substances Illicit drug(s) detected Multiple illicit drugs Alcohol Cannabis D-9-THC THC acid Opioids Morphine Methadone Codeine Psychostimulants Methamphetamine Cocaine/benzoylecgonine MDMA Pharmaceuticals Benzodiazepines Antidepressants Antipsychotics Suicide (n = 1226) % All (n = 1693) %

Group comparisons

64.0 26.8 34.3 9.9 42.0 22.1 20.6 5.6 12.2 11.6 2.4 6.2 10.5 6.0 4.5 1.5 11.6 8.6 3.0 1.3

66.0 25.5 20.0 3.6 38.7 10.6 8.2 8.3 8.3 5.2 1.7 4.8 6.9 4.4 1.7 1.5 26.5 14.9 12.6 5.5

65.5 25.8 23.9 5.3 39.6 13.8 11.6 7.6 9.4 7.0 1.9 5.2 7.9 4.8 2.5 1.5 22.4 13.2 9.9 4.3

NS NS OR 2.09 (CI 1.652.65) OR 2.94 (CI 1.914.50) NS OR 2.39 (CI 1.803.17) OR 2.91 (CI 2.153.95) NS OR 1.53 (CI 1.092.16) OR 2.37 (CI 1.633.47) NS NS OR 1.59 (CI 1.102.31) NS NS NS OR 0.36 (CI 0.270.50) OR 0.53 (CI 0.370.77) OR 0.22 (CI 0.120.38) OR 0.23 (CI 0.100.52)

NS: not signicant; referent group for ORs: suicide; MDMA: 3,4-methylenedioxymethamphetamine; THC: tetrahydrocannabinol.

cases were signicantly more likely to have an illicit drug, and multiple illicit drugs, detected. Logistic regression results showed that membership of the HOM group remained a signicant independent predictor of the presence of an illicit drug (adjusted OR 2.21, CI 1.732.82), with younger age (adjusted OR 0.96, CI 0.950.97) and being male (adjusted OR 1.35, CI 1.011.80) also independent predictors. HOM cases were signicantly more likely to have cannabis, opioids and psychostimulants present (Table 2). Specically, HOM cases were more likely to have D-9-THC present, indicating an active inuence of the drug at the time of death. In terms of opioids, the HOM group was signicantly more likely to have morphine (the major metabolite of heroin) present. While there was an overall group difference in the presence of psychostimulants, there were no differences for any individual psychostimulant. Pharmaceuticals Pharmaceutical preparations were detected in nearly a quarter of cases (Table 2), with females signicantly more likely to have such drugs present (30.7 versus 20.0%, OR 1.77, CI 1.372.29). The SUI group were signicantly more likely to have a pharmaceutical preparation detected (Table 2). Specically, SUI cases were more

likely to have benzodiazepines, antidepressants and antipsychotics present. Logistic regression indicated that membership of the SUI group remained a signicant independent predictor of the presence of pharmaceutical preparations (adjusted OR 2.99, CI 2.184.10), as were older age (adjusted OR 1.02, CI 1.011.03) and being female (adjusted OR 1.97, CI 1.512.58). DISCUSSION The current study provided the rst comparative toxicological data on psychoactive substance use and death due to violent suicide/homicide. Moreover, all autopsies were conducted at the same forensic institute, and all toxicological analyses were conducted at the same laboratories, using the same analytical techniques. Two major ndings emerged. First, consistent with previous research [1019], psychoactive substances were ubiquitous among victims of violent death. Secondly, despite global similarities, substantial differences exist between the toxicology of homicide and violent suicide cases. It should be emphasized that, unlike previous research, the gures for the violent suicide group were not inated by deliberate substance poisonings, but reected the role of drugs in acts of self-violence. The strength of the association between substance use and violent death was illustrated by the nding that twoAddiction, 104, 10001005

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

1004

Shane Darke et al.

thirds of cases were positive for a psychoactive substance, and a quarter for multiple substances. Furthermore, there were no differences between homicide and violent suicide cases in the global prevalence of substances. This is not to suggest, of course, that intoxication was the proximal cause of all of these deaths. In all probability, in many cases it was, as previous research has demonstrated a proximal link between intoxication and violent death [5,6]. The link between substance use and violent death is strong, whatever the circumstance. The strength of this association is emphasized when we examine population gures on daily drug use. On the day they died, 40% of cases had alcohol present. By contrast, 8% of the Australian population drink daily [20]. Taking daily use as a comparative point prevalence estimate, cases were ve times more likely to have alcohol present. Similarly, using the same comparisons, we see substantial overrepresentation of cannabis (14 versus 2%), methamphetamine (5 versus 0.4%) and heroin (7 versus 0.1%). Again, it should be emphasized that the current study concerns solely violent death. Drugs, by denition, play a role in all deliberate psychoactive substance self-poisonings. The current study illustrates graphically the extent of alcohol involvement in violent death. Alcohol was ubiquitous and, importantly, there was no group difference in prevalence. The levels of intoxication were clinically signicant, with the median concentration being the equivalent of approximately six to eight standard drinks in men. While levels were comparable across groups, the intoxication of HOM cases was signicantly higher than that of SUI cases. The implications of this nding are unclear. It may reect, however, the effects of heavy intoxication on increased situational risk to involvement in violence, e.g. intoxicated involvement in street ghts [10]. While there were no global differences between violent suicide and homicide cases in the presence of psychoactive substances, or of alcohol in particular, there were marked differences in the presence of illicit drugs. While present in substantial proportions among both groups, homicide victims were more than twice as likely to have illicit drugs (and multiple illicit drugs) present at the time of death. Homicide cases were more likely to have cannabis, opioids (morphine in particular) and psychostimulants present. The data indicate that the link between illicit dugs and homicide is stronger than with violent suicide. The data also indicate the risks associated with illicit drug use in terms of exposure to violence, whether through disinhibition or the broader life-style risks of illicit drug use. In contrast, all three classes of pharmaceuticals were present among signicantly higher proportions of suicide cases. Given that these drugs are prescribed therapeutic

substances, in terms of violent death these ndings suggest a far larger role for psychopathology in violent suicide than is the case for homicide: anxiety disorders (benzodiazepines), mood disorders (antidepressants) and psychosis (antipsychotics). These data are consistent with those from longitudinal studies on suicide as an outcome for psychopathology across a wide range of disorders [1,2]. As noted previously, it was not possible in this study to distinguish between the therapeutic use or abuse of these substances at the time of death. As with all studies, caveats need to be borne in mind. First, in all countries there are inherent difculties in determining cases of suicide and homicide. Suicide may not be obvious, and there may be attempts to conceal the fact due to social stigma. Similarly, offenders may attempt to conceal cases of homicide (as accidents, for example). In all Australian jurisdictions, determination of suicide and homicide in the Coronial process is based upon standard medicolegal Coronial procedures that include full autopsy, witness testimony, police reports and toxicology. Despite a standardized process, it is always possible that some cases are misattributed, and do not appear in this case-series. Caution should be also exercised in extrapolating these results to other suicide and homicide cases. While this is the case, however, the demographics, toxicology and means of death of both the homicide and suicide case-series are consistent with the characteristics of Australian homicide and violent suicide cases [22,23]. While care should also be taken in comparing the toxicology of the two groups, as noted above it was a major advantage of the current study that all autopsies and toxicological analyses were conducted at the same institutes. In summary, psychoactive substances were present in the majority of violent deaths in this case-series. There were, however, meaningful differences in the toxicology of other- and self-inicted violent death. The role played by drugs in premature death extends far beyond overdose and disease. Acknowledgement This research was funded by the Australian Government Department of Health and Ageing. Declarations of interest None. References
1. Harris E. C., Barraclough B. Suicide as an outcome for mental disorders. Br J Psychiatry 1997; 170: 20528. 2. Wilcox H. C., Connor K. R., Caine E. D. Association of alcohol and drug use disorders and completed suicide: an empirical review of cohort studies. Drug Alcohol Depend 2004; 76S: S1119.
Addiction, 104, 10001005

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

Drugs and violent death

1005

3. Hulse G. K., English D. R., Milne E., Holman C. D. J. The quantication of mortality resulting from the regular use of illicit opiates. Addiction 1999; 94: 2219. 4. Darke S., Degenhardt L., Mattick R. Mortality amongst Illicit Drug Users: Epidemiology, Causes and Intervention. Cambridge: Cambridge University Press; 2007. 5. Crombie I. K., Pounder D. J., Dick P. H. Who takes alcohol prior to suicide? J Clin Forensic Med 1998; 658. 6. Athanasiadus L. Drugs, alcohol and violence. Curr Opin Psychiatry 1999; 12: 28186. 7. Boles S. M., Miotto K. Substance abuse and violence. A review of the literature. Aggress Violent Behav 2003; 8: 155 74. 8. Darke S., Kaye S., McKetin R., Duou J. The major physical and psychological harms of methamphetamine use. Drug Alcohol Rev 2008; 27: 25362. 9. Flynn P. M., Joe G. W., Broome K. M., Simpson D. D., Brown B. S. Recovery from opioid addiction in DATOS. J Subst Abuse Treat 2003; 25: 17786. 10. Darke S., Duou J. Toxicology and circumstances of death of homicide victims in New South Wales, Australia 1996 2005. J Forensic Sci 2008; 53: 44751. 11. Galea S., Ahern J., Tardiff K., Leon A. C., Vlahov D. Drugs and rearms deaths in New York City, 19901998. J Urban Health 2002; 79: 7086. 12. Gill J. R., Lenz K. A., Amolat M. J. Gunshot fatalities in children and adolescents in New York City. J Forensic Sci 2003; 48: 83235. 13. Tardiff K., Marzuk P. M., Leon A. C., Hirsch C. S., Stajic M., Portera L. et al. Cocaine, opiates and ethanol in homicides in New York City. J Forensic Sci 1995; 40: 38790. 14. Tardiff K., Wallace Z., Tracy M., Piper T. M., Vlahov D., Galea S. Drug and alcohol use as determinants of New York City homicide trends from 1990 to 1998. J Forensic Sci 2005; 50: 47074.

15. Centers for Disease Control. Toxicology testing and results for suicide victims13 states, 2004. MMWR 2006; 55: 12458. 16. Darke S., Duou J., Torok M. Toxicology and circumstances of completed suicide by means other than overdose. J Forensic Sci 2009; 54: 4904. 17. Dhossche D. M., Rich C. L., Ghani S. O., Isacson G. Patterns of psychoactive substance detection from routine toxicology of suicides in Mobile, Alabama, between 1990 and 1998. J Affect Disord 2001; 64: 16774. 18. Marzuk P. M., Tardiff K., Leon A. C., Hirsch C. S., Stajic M., Hartwell N. et al. Use of prescription psychotropic drugs amongst suicide victims in New York City. Am J Psychiatry 1995; 152: 15202. 19. Shields L. B. E., Hunsaker D. M., Hunsaker J. C., Ward M. K. Toxicologic ndings in suicide. A 10-year retrospective review of Kentucky medical examiner cases. Am J Forensic Med Path 2006; 27: 10612. 20. Rivara F. P., Mueller B. A., Somes G., Mendoza C. T., Rushforth N. B., Kellerman L. Alcohol and illicit drug abuse and the risk of violent death in the home. JAMA 1997; 278: 56975. 21. Australian Institute of Health and Welfare. 2004 National Household Survey: Detailed Findings. Drug Statistics Series no. 16. Canberra: Australian Institute of Health and Welfare; 2005. 22. Mouzos J. Homicide in Australia: 20032004 National Homicide Monitoring Program (NHMP) Annual Report. Canberra: Australian Institute of Criminology; 2005. 23. Australian Bureau of Statistics. Suicides, Australia 2005. Report No. 3309.0. Canberra: Australian Bureau of Statistics; 2007. 24. SPSS Inc. SPSS for Windows, 14.0. Chicago: SPSS Inc.; 2005.

2009 The Authors. Journal compilation 2009 Society for the Study of Addiction

Addiction, 104, 10001005