Mortality of Women from Intimate Partner Violence in South Africa: A National Epidemiological Study
Abrahams, Naeemah, Jewkes, Rachel, Martin, Lorna J., Mathews, Shanaaz, Vetten, Lisa, Lombard, Carl, Violence and Victims
The purpose of this article is to describe mortality of women from intimate partner violence (IPV) in South Africa using a retrospective national study in a proportionate random sample of 25 mortuaries. Homicides identified from mortuary, autopsy, and police records. There were 3,797 female homicides, of which 50.3% were from IPV. The mortality rate from IPV was 8.8 per 100,000 women. Mortality from IPV were elevated among those 14 to 44 years and women of color. Blunt force injuries were more common, while strangulation or asphyxiation were less common. The national IPV mortality rate was more than twice that found in the United States. The study highlights the value of collecting reliable data across the globe to develop interventions for advocacy of which gender equity is critical.
Keywords: female homicide; femicide; intimate partner violence; South Africa
Death is the most extreme consequence of intimate partner violence (IPV) and is often the culmination of extended periods of abuse (Campbell, Sharps, & Glass, 2001; Campbell et al., 2003; Daly & Wilson, 1988; McFarlane et al., 1999; Moracco, Runyan, & Butts, 1998). The health consequences of IPV have been extensively researched, but there has been little focus on mortality. It is a much less common health outcome than morbidity, but it is nonetheless particularly important. The United States has national crime databases that routinely collect statistics, and these are presented in a manner that shows the victim-perpetrator relationship and that enables mortality rates from IPV to be studied. Globally, this is very uncommon (Campbell, Glass, Sharps, Laughon, & Bloom, 2007). In most other settings, mortality from IPV can be described only through research, and very few studies have been done (Arbuckle et al., 1996; Campbell et al., 2003; Moracco et al., 1998).
Routine data sources in South Africa do not enable murder to be studied by victim-perpetrator relationship, and therefore the burden of mortality due to IPV has not been explored. We conducted a national study to describe the epidemiology of mortality from IPV, and we present and discuss the mortality rates in this article.
This was a retrospective national study. All 225 mortuaries in South Africa that were operating in 1999 were included in the sampling frame. These were stratified by size based on the number of autopsies performed per annum (small = <500 autopsies, medium = 500-1,499 autopsies, large = >1,499 autopsies). A stratified random sample of 25 medicolegal laboratories was drawn using proportional allocation (see Table 1). Within each sampled laboratory, all women aged 14 and older who had been killed by another person in circumstances that were not accidental, between January 1 and December 31, 1999, and whose bodies were taken to mortuaries were identified. This study took 14 as the youngest age, as below this age very few women have intimate partners.
The primary data source was the death registers at the sampled mortuaries. Where death registers were incomplete or not available (n = 4 mortuaries), cases were identified from diaries of the forensic medical examiners or from other police record-keeping systems. The underlying cause of death was usually recorded in the register. All gunshots injuries, head injuries, poisonings, hangings, decomposed bodies, or any cases where the cause of death was "undetermined" or "unknown" were initially included, while clearly recorded suicides and transport accidents were excluded at this stage. For each of these cases, the autopsy report was photocopied. Cases were finally classified as homicide or nonhomicide after review of the autopsy report (second data source) and the interview with the police or docket review (third data source). In a small number of cases (n = 34) where there were discrepancies between the autopsy report and police information, the research team discussed each case, and if there was any doubt that the case was a homicide, it was excluded. …