HOMICIDE DEATHS IN SUB-SAHARAN AFRICA: A REVIEW 1970±2004 Anne Outwater,1 Jacquelyn C. Campbell, Daniel Webster Johns Hopkins University, School of Nursing, Maryland, USA
Edward Mgaya Mvhimbili University College of Health Sciences, School of Medicine, Dar es Salaam, Tanzania
ABSTRACT Homicide is predicted to increase globally. The foundation for explanatory models of violence in lowincome countries is rudimentary. The objective of this article is to describe peer-reviewed literature about homicide in sub-Saharan African (SSA) from 1970±2004. Searches in PubMed, OVID, and African Journals on-Line (AJOL) using the key words/phrases ``homicide'', ``injury death'', and ``Africa'', identified 37 articles. Articles were excluded if it was impossible to distinguish homicide from suicide or unintentional injury death. Of the 37 articles, 30 contained information about homicide. According to the reviewed literature, throughout Africa, males are 4±10 times more likely to die of homicide than females. While children seem to be protected, adults (20±29 years) are at highest risk. The most common mechanism of homicide appears to be sharp instruments. The estimations depicting Africa as the world's second most violent continent were made from scanty data. Regional variations exist but the lack of data also makes it impossible to answer basic questions related to proportionate mortality, occupation of the deceased, ethnic or national variations, precipitating factors, or victim-perpetrator relationships. Temporal trends cannot be ascertained. In an attempt to bridge the gaps in existing knowledge about homicide, the World Health Organization (WHO) is coordinating and supporting efforts by countries to set up injury mortality surveillance systems, as a first step towards monitoring the incidence and prevalence of violence.
INTRODUCTION AND BACKGROUND The number of deaths due to violence reached unprecedented heights in the twentieth century (Rummel, 1997). The Global Burden of Disease Report supported by the World Bank, WHO, and Harvard University predicted that by 2020 all forms of injury burden would increase dramatically.
ÐÐÐÐÐÐÐÐÐÐÐÐÐÐÐ 1. Please direct all correspondence to Anne Outwater, 62 Overlake Park, Burlington, Vermont 05401, USA; E-mail:
[email protected]
31
Homicide, war and suicide are expected to move to the top 15 causes of death (see Murray & Lopez, 1996). A recent WHO study involving a review of mortality databases 22 countries found that firearmrelated homicides have increased in the last decade (Sharma & Butchart, 2004). Against this backdrop of the global burden of violence, Africa tends to be considered the second most violent continent in the world (Peden, McGee, & Sharma, 2002). However, this assessment is based mostly on extrapolations from South African data and other limited data sources. Accordingly, in the following sections the authors focus on the identified peer-reviewed literature with a view to depicting the epidemiology of homicide in SSA from 1970±2004, and systematically identifying gaps in the body of public health knowledge on homicide across the African continent.
METHODS Homicide is defined in accordance with the International Classification of Diseases that includes intentional injury death by blunt object, sharp object, firearm, fire, poisoning, drowning, strangulation/ hanging, electrocution, exposure to the elements, and use of the physical body. Following this definition the review included searches through PubMed, OVID, and AJOL, using the key words/phrases ``homicide'', ``injury death'' and ``Africa''. The search yielded 37 articles. Full texts of the relevant articles were obtained whenever possible. Since French is an important language in Africa, an effort was made to obtain relevant literature published in French to decrease the bias towards literature published in English. The reference lists of the 37 articles were examined for further literature. The search failed to yield peer-reviewed literature on any aspect of injury for 27 of the 55 SSA countries; 29 of the 37 articles containing information about homicide were eventually selected for the purposes of this article. They included articles from Kenya (2), Mozambique (2), Nigeria (5), Senegal (1), South Africa (18), and Tanzania (1). Ten of the studies, including one case study, were prospective; 17 were retrospective examinations of mortality registers; and the two qualitative studies focused on perpetrators. Articles that were selected for review provided quantitative and qualitative data about homicide in SSA, and included hospital and mortuary reports which have been largely excluded from previous reviews (Aligbe et al., 2002; Amakiri 2000; Muniu et al.,1994). The sparse literature on war, war- and refugee-related camp deaths and reports published by any individual country's Ministry of Health were excluded. The principal reason for excluding an article was if it were organised by type of injury, but it was impossible to distinguish homicide from suicide or unintentional injury death; this criterion especially excluded potentially useful information around head injuries.
RESULTS In 1990 only 1% of all deaths in SSA were noted in vital registration systems, and injury data were based on incomplete police and health facility records (Holder, Peden, Krug, Lund, Gururaj, & Kobusingye, 2001; Kaufman, Asuzu, Rotimi, Johnson, Owoaje, & Cooper, 1997; Krug, Dahlberg, Mercy, Zwi, & Lozano, 2002; Murray, 1987; Nordberg, 1994). Drawing on WHO databases, Butchart
32
and Engstrom (2002) found that in only 61 of the 191 United Nations member countries, and none from Africa, was the homicide data more than 50% accurate in 1995. In some African countries, death registration, including homicide cases, is not enforced in many regions (Amakiri, 2000). Therefore most deaths in these countries are not autopsied which is likely to result in significant undercounts of homicides. In a few countries, however, principally South Africa, all non-natural deaths are required to be autopsied. As these are being performed more frequently, it could emphasise the higher occurrence of non-natural deaths, including homicide, over natural deaths. Below we present the outcome of our review on homicide organised by sex, age and proportionate mortality, mechanism of injury, and civil unrest. Sex Twenty studies reported the number of males and females killed from which ratios were calculated (see Table 1). Ten of these used data from South Africa, including six from Cape Town. These studies suggest that males were 4±10 times more likely than females to be murdered. The only male:female ratios that were close to 1 were those based on a small number of homicides (n = 11) (Amakiri, 2000) or on the number of homicides on children under 15 years (Knobel, De Jilliers, Parry, & Botha, 1984). It is possible that the high male:female death ratios are partially attributable to migration patterns. For example, the fact that men migrate to the cities (where most surveillance is done) in search of work and leave the women behind in rural areas may account for the higher proportion of adult males listed in the death registers than would be expected from life tables (Dgedge, Novoa, Macassa, Sacarlal, Black, Michaud, & Cliff, 2001). High male mortality may also be related to deaths due to civil unrest (Duflou, 1986; Duflou, Lamont, & Knobel, 1988). For instance, in a prospective study involving a low-income population of 60 000 in 20 villages near South Africa's Mozambican border, about a third of the population comprised refugees from the Mozambican war of independence; the age-standardised homicide rate for the years 1992±1995 was found to be very high at 41.9/100 000 that is 74.2 for males and 17.9 for females (Garenne, Kahn, Tollman, & Gear, 2000). This unusually high homicide rate is close to the Reza, Mercy, and Krug (2001) estimate for all Africa (40.2/100 000 total Ð 69.8 for males and 11.2 for females). Only one paper reported on male perpetrators. In Kaduna State, Nigeria, during 1983-1987, a place and time in which men commonly carried knives, 408 murders were reported. Psychiatrists interviewed 72 perpetrators in prison custody in January 1988 (Ogungbemi & Ahmed, 1993) and found the following: 94% of the perpetrators were male; 92% were between the ages of 20 and 40; 30% were unemployed; 18% were ex-military; and 84% of the victims were relatives or friends of the perpetrators. One fifth of the murders were committed for reward (economic, criminal, or psychological) or as a result of mental illness; 80% of the murders reported were for ``socially acceptable'' reasons that included provocation (48%), alleged accident (22%), and self defence (10%).
33
Table 1. Male:Female ratio of homicide deaths in sub-Saharan Africa Country, Region
Author
Period
Number of Male: h o m i c i d e s Female (n) ratio
Sub-Saharan Africa
Reza et al.
Estimate
Nigeria, Ife
Nwosu & Odesanmi
1977-1988
202
4:6
Nigeria, Rivers State
Amakiri
1984-1989
11
1:2
Nigeria, Jos
Mohammed et al.
1995-1999
101
7:4
Nigeria, Benin City
Aligbe et al.
1996-1997
138
4:9
Kenya, Nairobi
Muniu
1986-1990
259
9:0
Tanzania
Moshiro et al.
1992-1998
Not given
4:8
Mozambique, Maputo
Dgedge et al.
1994
188
8:0
South Africa
Wyndham
1970
South Africa
Flisher et al.
1984-1986
South Africa
Matzopoulos
2001
South Africa, Soweto
Byarugaba & Kielkowski
1990-1991
1 470 (
