Intimate femicide–suicide in South Africa: a cross-sectional study
Shanaaz Mathews a, Naeemah Abrahams a, Rachel Jewkes a, Lorna J Martin b, Carl Lombard c, Lisa Vetten d
Intimate femicide, the killing of a woman by her intimate partner, is considered the most extreme form and consequence of intimate partner violence. Until recently, not much has been known about this phenomenon in South Africa. The first national study on female homicide estimated that the intimate-femicide rate in 1999 was 8.8 per 100 000 women aged 14 years and older.1 This rate is higher than other reported rates worldwide, with the only comparison being North Carolina in the United States of America (USA), which has reported a rate of 3.46 per 100 000 women aged 15 years and older.2
Internationally, between 18% and 40% of perpetrators of intimate femicide commit suicide afterwards.3–7 The past decade has seen an emergence of studies from developed countries such as Australia and the USA that have explored the type of perpetrator and the associated risk factors.3,4,6 However, there is an absence of information from developing settings. More commonly, this phenomenon has been examined in homicide-suicide research, where the murder victims are both men and women and perpetrators may be either an intimate partner or not.
The South African national study on the epidemiology of female homicide provided the opportunity to describe the epidemiology of intimate femicide–suicide for the first time. This has not been described previously in a developing setting since such settings have limitations with the availability of reliable death data, thus placing a constraint on the range of variables available.
The aim of this paper is to describe the incidence and patterns of intimate femicide–suicide and the factors associated with an increased risk of suicide after intimate femicide.
This was designed as a cross-sectional mortuary-based national retrospective study of female homicide victims aged over 13 years who presented at a medical legal laboratory (MLL) between 1 January 1999 and 31 December 1999. In South Africa, all unnatural deaths are required to undergo a postmortem at an MLL to determine cause of death. All MLLs operating in 1999 formed part of the sample and were stratified based on the number of postmortems performed per annum; small < 500 bodies, medium 500–1499 bodies and large ≥ 1499 bodies. The approximate ratio of allocation between the three strata was 8:5:12 (8 large mortuaries, 5 medium mortuaries and 12 small mortuaries), which was based on optimal allocation fitting a sample of 25 mortuaries.8 Data were collected 3 years after the murder, so as not to compromise the criminal investigation. Ethical approval for the study was granted by the ethics committee of the South African Medical Research Council.
Data were collected from three sources using a pretested data collection sheet, which was designed based on an assessment of various instruments and was finalized after the pilot study. The primary data source was the death register at the MLLs so as to identify cases of homicide. The second data source was the postmortem report, which was entered onto the data collection sheet by the forensic pathologist on the team. Interviews with the investigating officer or police record reviews were the third data source. Police data allowed us to confirm a homicide, victim–perpetrator relationship and whether the perpetrator committed suicide.
The following definitions were used:Intimate femicide: The killing of a woman by an intimate partner. This includes the woman’s husband, boyfriend (dating or cohabiting), ex-husband (divorced or separated) or ex-boyfriend, same sex partner or a rejected would-be lover. Intimate femicide–suicide: An intimate femicide followed by the suicide of the perpetrator within a week of the homicide. Intimate femicide–non suicide: The killing of a female by her intimate partner without subsequent suicide of the perpetrator.The perpetrator was defined as the person whom the investigating officer considered as the primary person responsible for the homicide. Cases were classified into intimate femicide and non-intimate femicide, and then intimate-femicide cases were sub-classified into intimate femicide–suicide and intimate femicide–non suicide.
Classification bias was minimized through the use of two data sources; data from the investigating officer and the outcome of the inquest court inquiry.
Data collected from MLL records included information on police case details and victim information such as age, race, date and time of death. The pathology reports provided data on injuries, manner of death and primary cause of death. Data collected from the police included demographic details of perpetrator, victim perpetrator relationship and relationship status, previous history of violence, events leading to the murder, the legal outcome of the case or the death of the perpetrator. Data on race were collected: inequalities imposed by apartheid have had a lasting public health impact and must be considered by health researchers. Race was used based on the apartheid classification system as it is still documented in all official records. Race of the victim was determined through mortuary records, while perpetrator race was determined via police records. Race is accurately documented at both these sources as it is based on the person’s identification documentation.
Data were analysed using Stata version 8 (StataCorp LP, College Station, TX, USA). The analysis took into account the survey design, including the stratification and weighting of the sample. Incidence rates for intimate femicide–suicide were calculated for victims and perpetrators using population estimates from the 1996 South African Census Report,9 adjusted to reflect the year under investigation. Descriptive statistics were used to compare intimate femicide–suicide and intimate femicide–non suicide cases. Significant differences between the two groups were tested using the chi-square test. Unadjusted odds ratios and 95% confidence intervals (CI) were calculated to describe the association between intimate femicide–suicide and selected variables. A logistic regression model was built to investigate the factors associated with intimate femicide–suicide. A backward stepwise model-building process was followed. Candidate variables for the model included perpetrator’s race, victim’s age, perpetrator’s age, perpetrator’s occupation, legal gun ownership, relationship status, events leading to the homicide, primary cause of death and mechanism of death. The final model contained the independent variables that remained significant at ≤ 0.05 level. Finally, the attributable fraction for legal gun ownership and intimate femicide–suicide was calculated using the adjusted odds ratio for gun ownership.10
A total of 3793 (95% CI: 2693–4894) estimated cases (weighted) were identified via death registers (Fig. 1). Complete police data were collected on 86.7% (3296; 95% CI: 2440–4152) of cases, of which 18.6% (95% CI: 13.9–24.2) had an unknown perpetrator and were excluded from further analysis. Overall 11.4% (95% CI: 7.8–15.0) of the perpetrators died in the 3 year follow-up period, with this figure increasing to 22.2% (95% CI: 15.2–31.3) for intimate-femicide perpetrators, suicide being the leading cause of death (86.6%; 95% CI: 73.6–99.4) in this group.
Fig. 1. Flowchart of all cases identified
This study found an estimated 261 (95% CI: 155–368) intimate femicide–suicide cases (i.e. 19.4% of intimate-femicide cases; 95% CI: 11.9–26.8), giving an intimate femicide–suicide fatality rate of 1.7 per 100 000 (95% CI: 1.0–2.4) women aged 14 years and older and a perpetrator fatality rate of 2.0 per 100 000 (95% CI: 1.2–2.8) males 15 years and older for 1999.
Table 1 presents the demographic characteristics of victims. The median age of victims of intimate femicide–suicide cases was 26 years and median age of victims of intimate femicide–non suicide cases was 30 years. Victims of intimate femicide–suicide cases were usually younger than their perpetrators, who were significantly less likely to commit suicide as the victim’s age increased.
Table 2 shows that perpetrators of intimate femicide–suicide were more likely to be of white or Indian than of African racial background and least likely to be of mixed race; more likely to be younger; more likely to be employed; and more likely to be working in the security industry (police, army or as a guard) than a blue-collar worker.They were also more likely to own a legal gun.
Significant differences were found between the two intimate-femicide groups in relation to the preceding events to the murder (P = 0.03). In the intimate femicide–suicide group, women were more likely to have ended the relationship.
No significant differences were found in the primary cause of death. However, mechanism of death shows that intimate femicide–suicide cases were significantly more likely to have been killed by a gun and less likely to have been killed by blunt force, compared to intimate femicide–non suicide cases.
Table 3 shows a logistic regression model for factors associated with femicide–suicide perpetrators. This was more likely if the perpetrator was a professional or white-collar worker (odds ratio, OR: 37.3; 95% CI: 5.82–238.93) rather than a blue-collar worker; owned a legal gun (OR: 45.26; 95% CI: 8.33–245.80) and was racially classified as white (OR: 5.8; 95% CI: 1.21–27.85) rather than African. Calculating the attributable fraction reveals that 91.5% of the deaths of legal gun-owning perpetrators and their victims may have been averted if legal gun ownership had been restricted among this group.
Table 1. Characteristics of victims
Table 2. Characteristics of perpetrators
Table 3. Logistic regression analysis model: factors associated with intimate femicide–suicidea
In South Africa during 1999, 19.4% of perpetrators of intimate femicide committed suicide. This proportion appears to be at the lower end of the range reported internationally (18–40%),3–10 with similar rates reported for Australia (21%)5 and North Carolina, USA (24%).6 Yet, the intimate femicide–suicide fatality rate for women (1.7 per 100 000) was higher than the only comparable rate in North Carolina, USA, which reported an increase from 0.67 to 1.06 per 100 000 over a 5-year period.6 This increase of almost two times is similar to the overall difference in intimate-femicide rates between South Africa (8.8 per 100 000)1 and North Carolina (3.46 per 100 000).2 These high rates of intimate femicide and femicide–suicide appear to be related to the excessive levels of interpersonal violence prevalent in South Africa, with homicide being the second leading cause of death.11 We do not completely understand why South Africa has a lower proportion of intimate femicide–suicide, but it could be linked to the racial distribution of male suicides. Our overall suicide rate is comparable to global rates, however, we have an overrepresentation of white males.12 This difference suggests that a different dynamic exists within South Africa and could partly be explained by the prevailing high level of interpersonal violence and gender inequality.
Intimate femicide–suicide has often been considered as a distinct phenomenon, differing from both female homicide and intimate femicide.13 However, few studies have been conducted to explore its associated factors. Risk factor studies from the USA have shown that a history of domestic violence is an important risk factor for both intimate femicide and femicide–suicide.4,6 An important limitation of this study is that history of domestic violence was not routinely investigated by the police. This information was only known for one-third of intimate-femicide cases, with no data on previous violence for femicide–suicide cases as they were investigated only as inquest enquiries.
Theories on the cause of intimate femicide–suicide have a common theme of jealousy and possessiveness by the women’s male partner.5,13,14 Some authors have argued that it is due to depression in the perpetrator, which results in morbid jealousy and delusions that lead him to kill his partner.5,15,16 Intimate femicide–suicide is often premeditated and carefully planned.13,14,17 This is demonstrated by the short period of time between the two acts, as well as suicide notes and a history of stalking.13 Graser17 suggests that this is primarily an extended suicide (i.e. the primary aim is suicide), which is rational and carefully planned, with the homicide being an act of taking the family with him. Intimate femicide may also occur as a spontaneous act of murder occurring in a fit of jealousy, with the suicide being an act of remorse or stemming from fear of the consequences. The extent of premeditation, depression in the perpetrator or spontaneity of the act could not be determined in our study as the police had limited information in their files.
Suicide among intimate-femicide perpetrators was more likely if the perpetrator was white; employed as a professional or white-collar worker; and owned a legal gun. Given South Africa’s political history, race is often a loose indicator of socioeconomic status and social group membership, or culture.18 The increased suicide risk for white men matches national suicide patterns in South Africa that show the rate of suicide overall is highest in this group.12 Therefore we suggest that the social meaning of intimate femicide, and the perpetrator’s reaction afterwards, may differ between racial groups. Interestingly, overall intimate-femicide rates were found to be lowest among white males.1
Studies that have explored the factors associated with intimate femicide–suicide4,5,7 suggest that it is a “middle-class phenomenon”, more common among married men who are employed. Kozoil-McLain et al.4 report that perpetrators of intimate femicide–suicide were more “conventional”, employed, married, did not abuse drugs or abuse their partner during pregnancy. This study has also found that perpetrators of intimate femicide–suicide were of a higher socioeconomic status and would perhaps have more to lose after killing an intimate partner. This suggests that the possibility of negative consequences, such as the shame attached to incarceration and imprisonment, might result in the perpetrator choosing to commit suicide.
Legal gun ownership has been shown to be strongly associated with an increased risk of intimate femicide–suicide in other studies.3,4,7,14,15 This almost certainly is due to the lethality of guns and the ease with which they can be used in suicide. The greater proportion of perpetrators employed in the police, army or private security industry reflects the easier access to guns in these professions. Two-thirds of intimate femicide–suicide perpetrators owned a legal gun and the attributable fraction shows that a large proportion of these deaths might have been prevented should gun ownership within this group have been restricted. The strong association found between legal gun ownership and intimate femicide–suicide suggests that access to legal guns in South Africa requires urgent attention. In the USA, it was found that restricting gun access for abusers who had restraining orders decreased rates of intimate partner killings, while confiscation and domestic violence laws proved ineffective.19 This suggests that, although there has been huge emphasis on illegal gun ownership in South Africa, legal gun ownership and its relationship with intimate partner violence clearly poses an important public health problem and highlights the need to restrict gun ownership.
An important limitation in this study is the choice of comparison group. Comparing intimate femicide–suicide to intimate-femicide cases results in the contribution of common risk factors being rendered invisible. This factor is compounded by data being obtained from police sources. This could result in data being incomplete while also being limited in scope at times.
South Africa has a rate of intimate femicide–suicide that greatly exceeds reported rates elsewhere and clearly reflects the high rate of intimate femicide. We have shown that intimate-femicide perpetrators are more likely to commit suicide if they are white or of a higher socioeconomic status. Gun ownership is a very important risk factor for intimate femicide–suicide. The World report on violence and health reports on the risks posed by guns in the home for both intimate femicide and suicide. It recommends legislation on restricting ownership of guns as a form of primary prevention.20 This study once again highlights the public health risk posed by legal gun ownership and the importance of restricting access to guns. ■
Funding: The study was funded from the South Africa Medical Research Council baseline research funds.
Competing interests: None declared.
- Mathews S, Abrahams N, Martin LJ, Vetten L, van der Merwe L, Jewkes R. Every six hours a woman is killed by her intimate partner: a national study of female homicide in South Africa [Report no. 5]. Cape Town: Medical Research Council; 2004.
- Moracco KE, Runyan CW, Butts JD. Female intimate partner homicide: a population based study. J Am Med Womens Assoc 2003; 58: 20-5 pmid: 12553639.
- Easteal P. Homicide-suicides between adult sexual intimates: an Australian study. Suicide Life Threat Behav 1994; 24: 140-51 pmid: 8053008.
- Koziol-McLain J, Webster D, McFarlane J, Block C, Curry MA, Ulrich Y, et al., et al. Risk factors for femicide–suicide in abusive relationships: Results from a multi-site case control study. Violence Vict 2006; 21: 3-21 doi: 10.1891/vivi.21.1.3 pmid: 16494130.
- Cooper M, Eaves D. Suicide following homicide in the family. Violence Vict 1996; 11: 99-112 pmid: 8933707.
- Morton E, Runyan CW, Moracco KE, Butts JD. Partner homicide-suicide Involving female homicide victims: a population based study in North Carolina, 1988-1992. Violence Vict 1998; 13: 91-106 pmid: 9809390.
- Lund LE, Smorodinsky S. Violent death among intimate partners: a comparison of homicide and homicide followed by suicide in California. Suicide Life Threat Behav 2001; 31: 451-9 doi: 10.1521/suli.31.4.451.22046 pmid: 11775720.
- Foreman EK. Survey sampling principles. New York: Marcel Dekker Inc.; 1991.
- The people of South Africa population census, 1996. Pretoria: Statistics South Africa;1998.
- Mitchell HG, Benichou J. Encyclopedia of epidemiological methods. New York: Wiley Press; 2000.
- Norman R, Matzopoulos R, Groenewald P. Bradshaw. The high burden of injuries in South Africa. Bull World Health Organ 2007; 85: 695-702 doi: 10.2471/BLT.06.037184 pmid: 18026626.
- Flisher A, Liang H, Laubsher R, Lombard C. Suicide trends in South Africa, 1968-90. Scand J Public Health 200432-.
- Marzuk PM, Tardiff K, Hirsch CS. The epidemiology of murder-suicide. JAMA 1992; 267: 3179-205 doi: 10.1001/jama.267.23.3179 pmid: 1593740.
- Dawson M, Gartner R. Male proprietariness or despair? Examining the gendered nature of homicides followed by suicides. American Society of Criminology; 1998.
- Bourget D, Gagne P, Moamai J. Spousal homicide and suicide in Quebec. J Am Acad Psychiatry Law 2000; 28: 179-82 pmid: 10888185.
- Rosenbaum M. The role of depression in couples involved in murder-suicide and homicide. Am J Psychiatry 1990; 147: 1036-9 pmid: 2375438.
- Graser RR. A study of selected cases of family murder in South Africa. Human Sciences Research Council; 1992.
- West ME, Boonzaaier EA. Population groups, politics and medical science. S Afr Med J 1989; 76: 185-6 pmid: 2772762.
- Vigdor ER, Mercy JA. Do laws restricting access to firearms by domestic violence offenders prevent intimate partner homicide. Eval Rev 2006; 30: 313-46 doi: 10.1177/0193841X06287307 pmid: 16679499.
- World report on violence and health. Geneva: WHO; 2000.
- Gender and Health Unit, Medical Research Council, PO Box 19070,Tygerberg 7505, South Africa.
- Division of Forensic Medicine and Toxicology, University of Cape Town, Cape Town, South Africa.
- Biostatistics Unit, Medical Research Council, Tygerberg, South Africa.
- Centre for the Study of Violence and Reconciliation, Braamfontein, Johannesburg, South Africa.