Adolescent mortality in South Africa: An analysis of unnatural causes of deaths by sex, 2000-2009

Despite South Africa being a peaceful democracy for the past 18 years, mortality due to unnatural and violent causes still occur among the youth, who are aged between 15 and 34 years old (Presidency of the Republic of South Africa 2009). However, this is not specific to South Africa; with developed countries experiencing the same trend. In developed countries, Christoffel (1994) found that death due to unnatural causes, specifically related to violence, had increased especially among the youth. South Africa is now moving towards the same transition that is currently experienced by developed nations in the world: violent deaths as a major cause of death amongst the youth which includes some adolescents (10-19 years old) (Norman, et al, 2007). With this in mind, this paper aims to examine the different levels of male and female adolescent mortality due to five unnatural causes of death. Data from South African Death Notification Forms is analysed for the years 2000-2009. Cause-specific mortality rates and proportional mortality ratios are produced. Results show adolescents are dying from events of undetermined intent, transport accidents’ and self- harm, especially males. The selected causes of death are contributing up to 27% of all adolescent male mortality and almost 12% of all female mortality in 2009. The results of this paper allude to crime, violence and safety issues in South Africa.


Introduction
South Africa is notorious for violence and safety concerns. Urban areas are affected by issues of robberies, assaults and sexual violence (CSVR 2008;UNODC 2002). Mortality related to violent incidents is well-documented in the country (Abrahams et al. 2010;MRC/UNISA 2008). The Medical Research Council (MRC) found that in 2007, violence accounted for 35.8% of all fatal injuries in the country and in certain national cities the percentage was even higher. For example in Cape Town, violence accounted for 46.9% of all fatal injuries in the same year (MRC/UNISA 2008). It has been argued that adult males are most commonly the perpetrators of violent crime worldwide and South Africa is no exception. Masuku (2004) interviewed approximately 116 youth, aged 15-25 years old, in South Africa and found that young adult males are most commonly the perpetrators of crimes such as armed robbery, theft and hijacking (Masuku 2004). With regard to victims of crime in South Africa, Altbeker (2008) found that 87% of homicide victims were male and 13% were female (Altbeker 2008). Further, South African females are most commonly the victims of sexual assaults, but suffer robberies and hijackings too (Jewkes & Abrahams 2002;Schönteich & Louw 2011;Seedat et al. 2009). Among children, Mathews et. al (2012) found that homicide, rape and neglect were the most common causes of violent deaths in South Africa (Mathews et al. 2012). In this study, it was likewise found that assault (homicide) is a leading cause of unnatural deaths among South African adolescents.
Pre-1994 South Africa was beset by violence and mortality due to the oppressive nature of the Apartheid regime (Harris 2001). Post-1994, however, there remains a high prevalence of violence and death due to violence in the country (Seedat et al. 2009). However, this is not specific to South Africa; there have been a number of studies done in the United States of America and other developed countries with regard to this (Brismar & Bergman 1998;Christoffel 1994;Patton 2009;Roy et al. 2004). In all these studies a high prevalence of youth deaths due to unnatural causes, specifically related to violence and accidents, were found. South Africa is moving towards the same transition that is currently experienced by developed nations in the world: violent deaths as a major cause of death amongst the youth (Norman et al. 2007). In 2003 the Medical Research Council revealed that pedestrians accounted for the largest percentage (39.5%) of traffic-related deaths (MacKenzie et al. 2006). Estimates of death due to pedestrian injuries indicate that, during the period, 2001-2004 a ratio of 3.3 male deaths to one female was recorded (MacKenzie et al. 2006). Also Seedat et al (2009) found that the overall injury death rate of South Africans is 157.8 per 100,000 population, which is higher than the global average of approximately 86.9 per 100,000 (Seedat et al. 2009).
Globally adolescents and youth are acknowledged to be at higher risk of unnatural causes of death such as violence, accidents and suicides (Hawkins et al. 2000;Leon et al. 2006). South Africa is experiencing the same prevalent causes of death, unnatural causes such as suicide or accidents, amongst the youth as is many developed countries (Norman et al. 2007). In a study conducted by the Medical Research Council (MRC) on the number of deaths amongst the youth in South Africa between 2000-2008 found that the mortality rates increased during this period: in 2000 youth deaths between the ages of 14-19 years was 8,678 and in 2008 it was 9,890 (Reddy et al. 2010). This includes suicide, accidents and the use of firearms and it was found that in 2010 the percentage of attempted suicides among youth had increased to 32.7% (Reddy et al. 2010).
South Africa's youth comprises 37% of the total population of 50.5 million people (NYDA 2012). This large youth population are essential to the country's future growth and development making them a government priority (NYDA 2012). South African youth are defined as being between the ages of 10 -34 years old. Within this broad age group, lie the adolescent population of 10-19 year olds. The mortality of this particular faction of the youth population, are under-researched in South Africa, yet are of vital importance to the country. Adolescents will become economically active adults. Yet for now adolescents are a diverse group, because some are still wholly dependent on parents and care-givers (10-14 years old) while others are experiencing social and economic autonomy for the first time (15-19 years old). It is during adolescence then, that negative lifestyle and behavioural practices (smoking, drug-abuse, violence, etc.) that will follow this cohort into adulthood is learned. Further, as the lifestyle and behavioural practices of males and females differ, as a result, so too does their mortality. With this in mind, this paper aims to examine the different levels of male and female adolescent mortality due to five specific unnatural or external causes of death.

Data and Methods
Data on causes of death from Death Notification Forms for 2000-2009 are here used (Stats SA, 2012). Adolescents, between the ages of 10 and 19 years old are studied. Both males and females have been included in the analysis. Unnatural causes of death include deaths due to injury, accident from external causes and exclude deaths from disease and pathology (natural causes). Unnatural causes of death have been selected from the frequency at which they have occurred on Death Notification Forms. The causes of death and variable descriptions are seen in Table 1.  Any accident involving a device designed primarily for or being used at the time primarily for, conveying persons or goods from one place to another.
Other external causes of accidental injury  Slipping, tripping, stumbling and falls  Exposure to inanimate mechanical forces  Exposure to animate mechanical forces  Accidental non-transport drowning and submersion  Exposure to electric current, radiation and extreme ambient air temperature and pressure  Exposure to smoke, fire and flames  Contact with heat and hot substances  Exposure to forces of nature  Accidental exposure to other specified factors

Self-Harm
 Suicide  Intentional self-harm Source:http://www.icd10data.com/ICD10CM/Codes Apart from generating frequency distributions, causespecific mortality rates are calculated for males and females separately. Cause-specific mortality rates show the number of deaths from each particular cause of death, per 10,000 populations at risk (Hinde 1998). The formula used for this is: Where nD i x is the number of deaths from a cause and nPx is the total number of adolescents in the population. Proportional Mortality Ratios (PMR) which specifies the contribution of each selected cause of death to overall mortality is used (Mallin et al. 1986). The ratios are expressed as percentages (%) and the formula used is as follows: x is the number of deaths from a particular cause and nDx is the total number of deaths from all causes. Figure 1 shows the percentage distribution of adolescent deaths from unnatural causes by sex and year. The graph shows that overall male and female deaths from unnatural causes have remained over 15% consistently throughout the period. Further, from 2002 to 2005, the percentage of female deaths outnumbered the percentage of male deaths from these causes. However, in 2008-2009 it is seen that there is a marginally higher percentage of adolescent male deaths compared to female deaths. By way of comparison between the sexes, instances of mortality due to events of undetermined intent, transport accidents and self-harm are higher among adolescent females than males over the period. Alternatively, assault deaths are higher among males than females. Other Accidents Self-harm Figure 4 represents the percentage distribution of unnatural causes of death by month of death and sex for the 2000-2009 period combined. Male adolescent deaths are higher compared to those of female adolescent deaths. The graph demonstrates that there are a higher percentage of dead adolescent males occurring during the December period-just over 3.5%. This is followed by higher male adolescent deaths in the months of November, January, October, March and July. The lowest level of adolescent male deaths occurs during the month of May-with approximately 2% dead male adolescents for the 2000-2009 period.

Results
For females a similar pattern can be observed however at a lower percentage. The number of female adolescent deaths for the period 2000-2009 is highest in December, that is just over 1% of all deaths due to unnatural causes occur among this group at this point of the year. The lowest level recorded for females is during the month of June for the period 2000-2009. The occurrence of higher adolescent deaths over these periods in the country, suggest that it is during holiday periods, when adolescents are not in school, that unnatural causes of death occur the most. As a means of comparison between the sexes, it can be said that the percentage of adolescent male deaths are significantly higher than those of adolescent female deaths in South Africa for the 2000-2009 period. Furthermore, this can be compared to the distributions of unnatural causes of death between the sexes. As mentioned above, instances of mortality due to certain unnatural causes of death are higher among adolescent females than males over the period; this is in stark contrast to the percentage distribution by sex and month for the periods combined, in which instance mortality appears to be higher among male rather than female adolescents in South Africa.

Cause-Specific Mortality and Proportional Mortality Ratios:
With regard to the Cause-Specific Mortality Rates (CSMRs), male rates from all causes are consistently higher than female rates over the period. For males specifically in 2002-2003 rates were highest for deaths from events of undetermined intent at approximately 6 deaths per 10 000 population than all other causes for the period. This rate begins to decline from 2004-2009, reaching approximately 1 death per 10,000 population. The cause specific mortality rates for males due to all other causes such as assault, transport accidents, other accidents as well as self-harm, remain at approximately 1 death per 10,000 population for the 2000-2009 period. For females, rates from all selected causes of mortality are consistently lower than that of males within the period. Specifically, the highest deaths were from events of undetermined intent at approximately 2 deaths per 10,000 population. The cause specific mortality rates for females due to all other causes such as assault, transport accidents, other accidents as well as self-harm, remain well below 1 death per 10,000 population for the 2000-2009 period. The lowest cause specific mortality recorded for females was due to self-harm. And this remains the lowest rate of reported cause of unnatural death. Self-harm amongst females increased to 0.05 deaths per 10,000 population in 2004

Discussion
The aim of this study is to examine the different levels of male and female adolescent mortality due unnatural or external causes of death. The general finding that adolescent males die from unnatural causes of death more than adolescent females suggests that the South African context is not very different from what is happening globally (Blum & Nelson-Mmari 2004;Patton 2009). In Latin American countries, homicide mortality peaks among males aged 15-29 years old, making it the fifth leading cause of male mortality in the region (Blum & Nelson-Mmari 2004). In other studies that looked at both who the perpetrators and victims of violent crime are, it was found that young males are both more the victims and perpetrators of violence and crime compared to females (Perren & Hornung 2005). The fact that mortality is higher among males than females does not imply that adolescent females are not victims of violence in the country. It has been found in South Africa, that young females are victims of sexual assault more than young males ). It was found in Cape Town, South Africa, that 40% of interviewed adult females had a history of sexual assault (Kalichman & Simbayi 2004). A second issue, South African women face, and related to sexual assault, is the prevalence of gender-based violence in the country. In a study done in the Soweto region of the country, it was found that 55% of women attending an antenatal clinic had reported physical abuse from their partner (Dunkle, Jewkes, Brown, Yoshihama, et al. 2004). However, the consequences of such acts did not result in a death. This prominent form of violence is not reflected in this study, since other studies are based on self-reports of sexual and physical violence and not mortality. However, since females are more susceptible to being victims of sexual violence, and presumably other forms of physical assault such as gender-based violence, it raises the question, why is it not reflected in the death reports. Possible explanations are twofold: (1) adolescent males may suffer violent acts (such as assault) less frequently than females, yet the severity of these acts are worse, which is why it results in mortality; or (2) the classification of unnatural causes of death among females require more meticulous analysis, so if the assault which caused the death was sexual or gender-based, it should be reflected. The latter, point is related to a general limitation of all mortality data, especially where certain causes of death are considered taboo or moot topics.
While deaths from self-harm remain low for both sexes, cause-specific mortality rates and proportional mortality ratios show that since 2006, other external causes of accidental injury contribute to the mortality of adolescents more than all other causes, including events of undetermined intent. The lowest rates and contribution for both sexes in the period is from causes related to self-harm. Although self-harm causes of death could have possibly been declared events of undetermined intent in the absence of any proof of suicide, such a note left by the deceased. Other studies concerning health outcomes have shown a peak in suicides and attempted suicides during adolescence (Nock et al. 2006;Wasserman et al. 2005). In the United States of America, it was found that lifetime prevalence of having at least 1 suicide attempt was 3.8% (Dube et al. 2001). Literature suggests that this is a result of psychological problems and traumas associated with the transition from child to adulthood (Beautrais 2000). Further in a multi-country study, it was found that suicide rates vary from a low of 2.3 per 100,000 in Kuwait to a high of 27.1 per 100,000 in Finland among youth (Johnson et al. 2000). In developing countries, less research is done on suicide and self-harm. However, research has been done on the similar psychological problems and trauma shared by adolescents globally (Barbarin & Richter 2001;Cluver et al. 2007). These studies have found that the impact of HIV/AIDS and migrant status of the family are associated with increased mental distress among South African adolescents.
Challenges encountered in researching suicide in developing countries, include cultural and traditional beliefs about suicide. In Uganda, suicide (known as 'kikonko') is acknowledged as an extremely dangerous act with consequences for the bereaved community. It is here believed that when a person commits suicide, the family and the entire community will be haunted by the deceased's spirit and innocent children and livestock may be killed out of revenge (Burr et al. 1994;Mugisha et al. 2011). Religious beliefs are another barrier to the acknowledgement of suicide as a cause of death. For example, in orthodox Catholicism it is believed that the soul of a person who commits suicide cannot enter heaven (Stack & Lester 1991). Under such circumstances, it is understandable why self-harm and suicide are rarely reported as such. In light of the similar psychological experiences and trauma that South African adolescents face and the challenges in reporting suicide, it cannot then be assumed that suicide does not happen among adolescents in South Africa. It is then possible that the low figures of selfharm reflected in this study is possibly due to incorrect classification of suicide deaths in the country. Misclassification of causes of death, is not a new occurrence in South Africa. In 2005, it was reported that a number of HIV/AIDS deaths in the country had been incorrectly classified as Tuberculosis and other infectious disease causes of death (Groenewald et al. 2005). So it stands to reason that there are other causes of death which are quite possibly also misclassified.

Limitations
In addition to the limitation of misclassification of causes of death in South Africa, there is another data limitation. Death registration in South Africa takes place at the Department of Home Affairs. After a death is registered, the Department issues a death certificate and where applicable, updates the population register. The forms are then collected by Statistics SA for processing. If the deceased is not in possession of a South African identity number or birth certificate their death cannot be registered on our National Population Register. This leaves a substantial amount of under-reporting of deaths among the poor (who do not have access to the Department of Home Affairs and were thus not issued birth certificates and identity numbers) and foreign nationals who reside in the country (Statistics SA, 2007). Stats SA have attempted to rectify this limitation, by supplementing this data with that which is collected by Census and other nationally representative surveys. This has supplemented the Death Notification Form data used in this study. However, in order for data from the National Population Register to be useful in the study of mortality in the country, the registration of deaths to persons who are not in possession of the correct documentation needs to be done.

Conclusion
Unnatural causes of death allude to existing levels of crime, violence and safety in South Africa. Because these are acts associated with inter-personal violence, the findings of this paper can inform youth programmes on the consequences of violent acts against adolescents in the country. Mortality from unnatural causes of death among South African adolescents have fluctuated over the years. It is evident that as of 2009, rates have increased. Overall this study has shown that unnatural causes of death are an increasing burden among adolescents in South Africa. Despite fluctuations over the years, these causes are contributing up to 27% of all adolescent male mortality and almost 12% of all female mortality in 2009. The paper has also showed that differences in mortality by particular cause of unnatural death between the sexes exist, in particular, when it relates to deaths from assaults and events of undetermined intent. For this reason, sex-specific programmes and interventions need to be developed to avert further increases in mortality. In addition, this study has shown that adolescent mortality from such causes occurs predominantly around holiday periods such as December. Due to adolescents being idle during such periods, since school is closed, it is important that programmes in the country institute 'holiday activities' for adolescents, so as to keep them occupied and away from violence and crime. Finally since these unnatural causes of death are largely preventable through the promotion of safe lifestyle and behavioural practices, these should be key areas of focus in programmes and interventions aimed at South African adolescents.
All named authors have contributed sufficiently to the work submitted and that the content of the manuscript has never been previously published.