The future of HIV depends on our ability to change the risk calculus for individuals and communities in hyper-endemic countries

David Harrison, MBChB, MSc(Med), MPP

Abstract

The future of HIV globally largely depends on our ability to substantially reduce new infections in the hyper-endemic countries of Southern Africa. The premise of most approaches to HIV prevention in these countries is that HIV will be substantially curtailed through a combination of behaviour change communication and biomedical interventions like male circumcision and condom promotion. The central focus has been on reducing or avoiding the absolute risk of HIV, but insufficient attention has been paid to understanding people’s response to the relative risk of HIV in the context of other perceived threats to wellbeing. This paper argues that the main difference between hyper-endemic countries and the rest of the world is not that specific sexual behaviours are entrenched in Southern Africa, nor that national responses are inadequate, but that people in hyper-endemic countries are more willing to risk HIV transmission. New gains in HIV prevention rest in creating a sense of real and imminent possibility for young people in particular.

INTRODUCTION

About 1.5% of the world’s population lives in the eight countries of Southern Africa that are regarded as having hyper-endemic epidemics. Yet those countries account for almost a third (31.4%) of all people living with HIV [Table 1].

Table 1

Country Estimated HIV prevalence (total) HIV prevalence 15-49 years (%) Estimated number of people living with HIV
Botswana 15.1% 23.9 300,000
Lesotho 9.9% 23.2 200,000
Mozambique 6.9% 12.5 1,500,000
Namibia 9.5% 15.3 200,000
South Africa 11.7% 18.1 5,700,000
Swaziland 16.3% 26.1 190,000
Zambia 8.7% 15.2 1,100,000
Zimbabwe 10.3% 15.3 1,300,000
Total 10.2% 10,490,000

Across the world, governments have had varying success in combating HIV, but the main distinction between hyper-endemic countries and others is not a difference in the quality of governance on HIV/AIDS. In fact, African countries with a relatively good public health system – at least as measured in terms of child mortality rates, per capita public health expenditure and immunisation for TB and measles – have higher rates of HIV infection. Rather than tracking health systems performance, patterns of HIV prevalence seem to follow the trends of ‘social diseases’ such as crime and homicide [1].

PROPOSITION

There have been many attempts at explaining the variation of HIV infection across regions and within countries, drawing on different disciplines from epidemiology to sociology to behavioural economics. They come at it from different perspectives, using different evidence. Yet behind all these theories appears to be a singular reality: unless there are compelling but yet-to-be-discovered biological explanations for the spatial distribution of HIV, what separates hyper-endemic countries from the rest is that – for whatever reason – more people in heavily affected countries are willing to risk HIV infection. This risk tolerance may manifest in a variety of ways – greater predisposition to age-disparate sex, more willingness to engage in concurrent partnerships, failure to use condoms despite knowledge of status discordancy – but the main sociological driver of high risk behaviour must be a willingness to risk HIV infection (b). This proposition gets to the heart of the challenge in hyper-endemic countries. I argue that unless it is confronted, global efforts to strengthen HIV prevention will have limited effect and Southern Africa will continue to be mired in the struggle to contain the effects of an intractable AIDS epidemic. The theoretical and empirical backing for this proposition is discussed below.

2.1 Proximal-distal-determinant constructs of HIV transmission have limited strategic value

Epidemiologists tend to equate the risk of HIV with the statistical probability of acquiring it; thus the design of ‘risk-avoiding’ and ‘risk-reducing’ strategies to decrease the odds of infection. Some of these strategies target the biological determinants of HIV transmission. For example, medical male circumcision aims to reduce the efficiency of transmission per contact. Other strategies aim to reduce the exposure of susceptible individuals to persons who are infected; hence the programmatic focus on multiple and concurrent partnerships. These and other ‘high risk sexual behaviours’ are regarded as the proximal determinants of HIV infection. Consequently, behavioural programmes are mainly designed to inform people of the risk. Frustration follows for programme implementers when they - the target audience – ‘just don’t get it’ and the sticking point is often framed as a ‘failure to internalise the risk’. Yet several studies in sub-Saharan Africa have shown that risk perception (as defined as the perceived likelihood of infection) is generally consistent with self-reported risk behaviour, at least among HIV negative individuals (2,3). In South Africa, most young people were able to explain their self-assessment of risk appropriately [4]. In other words, people hear the risk, they understand the risk to themselves, but many still take the risk. As will be explained later, this decision is not necessarily overt. But, despite knowledge and often-appropriate understanding of personal risk, relatively more people in hyper-endemic areas still take the risk (c).

Within any given population, some people are more willing to take risks than others, because of innate individual predilections to risk or risk aversion [5]. But this does not explain why more people in hyper-endemic countries still take the risk – and its answer points to the most critical leadership imperatives for Southern Africa.

Most observers recognise the influence of social and economic factors in shaping the distribution of both sexual behaviour and HIV across populations [6,7]. But because the mediating pathways are poorly defined, these factors tend to be relegated to the back of the queue and labelled as ‘underlying’ or ‘distal’ determinants [8]. The problem with that construct is that it limits both social mobilisation and the potential for effective leadership - either embedding prevention in local community-empowerment efforts only or demanding large-scale structural change to redress the macro-inequalities that drive HIV infection. While it is true that hard structural factors – unemployment, poor educational attainment and unplanned urban settlements - have to change in order to reduce HIV, the effect of the proximal-distal-determinant model is a tendency to absolve HIV prevention protagonists from responsibility for the ‘distal factors’. Visually, these determinants are often presented as a series of discrete text boxes, woven together by a web of directional arrows and feedback loops [9]. The implication is that HIV transmission is the consequence of a cascade of events and conditions in which government and markets operate upstream; the individual chooses whether or not to play in a pool of infection in the middle; leading to the downstream consequences of AIDS and underdevelopment [10].

The main problem with this model is that it fails to understand the direct effects of structural factors on sexual decision-making. In doing so, it often sets prevention interventions up to fail, for the following reasons:

While many of the responses to HIV can be packaged as programmes and delivered by governments and civil society organisations, much of the dynamic of HIV infection is rooted in the individual and collective psyche. Dislocated from their ‘underlying context’, behaviour change paradigms tend to be so personalized that they fail to create real incentives for change.

Governments try to be responsive to the largely technocratic advice of international agencies that focus on proximal strategies and pay insufficient attention to the connections between life circumstance and sexual behaviour. While structural influences are recognized, the intervention emphasis is on changing norms and attitudes by communicating ‘the message’ of safer sexual behaviour (typically abstinence, condom use, and partner reduction) and biomedical interventions that reduce transmission efficiency - and far less on creating individual and group motivation for change.

This model also prescribes a narrow role for civil society, largely that of holding the feet of governments to the fire and at the same time filling the service delivery gaps in local communities [11]. Non-government organisations end up making their contribution at community level, or face the dilemma of being either a ‘special interest’ activist group or broader movement demanding radical social change [12].

Almost inevitably, these structuralist models of HIV infection pit ‘cultural’ explanations of sexual behaviour against economic ones, and individualist interpretations against sociological ones. This dichotomy is illustrated in debates about multiple concurrent partnerships (MCP), where both the reasons for hyper-endemicity and the cause of the reasons are contested. One side argues that the HIV epidemic is driven, at least to a large extent, by dense and concurrent sexual networks [13]. While accepting the theoretical feasibility of that premise, others question its empirical basis [14]. Many of the protagonists of the ‘MCP theory’ accept that economic factors as well as cultural ones may be driving concurrency [15]. Others have pointed to the significant differentials in HIV infection rates between black and whites (d), claiming that sexual partner concurrency is more inherent in the culture of black African people in Southern Africa than other racial groups and people in other regions (16). Given the sensitivities of race and sexual behaviour, the debate has become increasingly polarised.

Much of this debate is white noise generated by our failure to apply the right analytical tools – either because we don’t have them or because we don’t use those that are available. Quantitative methods may omit many variables that could explain the distribution of HIV across communities, gender and race groups. Socio-demographic measures such as household type and access to electricity are, in survey terms, fairly crude approximations of structural influences; while the social-cognitive measures that have typically been used in quantitative surveys have tended to focus on self-efficacy for protective sexual behaviours [17] or affective associations such as optimism and general happiness about life. Although the latter may be loosely correlated with socio-economic status, associations with high risk behaviour tend to be insubstantial or have little value in explaining overall variation [18]. Few measures have sought to dissect the relationship between HIV infection and response to life circumstance, which might add significant explanatory power to regression models of HIV infection.

This is not to downplay the role that the probability of viral exposure plays in the distribution of HIV, completing a vicious cycle of HIV infection among Africans compared to other racial groups: It is likely concentrated in the African population because most marginalised communities are black and there is relatively little inter-racial sexual contact [19], which limits the probability of exposure to other racial groups. It’s a vicious cycle of culture and transmission dynamics. This is exemplified by patterns of HIV in educators in South Africa: in 2005, the prevalence of HIV among African educators was 16.3%, compared to less than 1% among whites, Indians and people of mixed race. However, the prevalence among more educated and higher paid was significantly lower than among less educated and worse paid teachers [20]. Clearly, being African puts one at higher at higher risk for infection. This fact could be explained by transmission dynamics alone. But being African and relatively poor puts one at even greater risk. That fact – greater susceptibility among poorer Africans - requires additional explanation. Epidemiological understandings of HIV transmission are still important to understand how HIV transmission happens through susceptible population networks [22]; but we need to better understand social and cognitive dynamics to explain why HIV is still transmitted at such high rates in susceptible populations.

2.2 A risk tolerance model of HIV transmission

A leading anthropologist and risk theorist, Mary Douglas, argued that “culture is a dynamically interactive and developing socio-psychic system” [23] – interacting not least with the economic circumstances in which people find themselves. If Douglas was right, a crucial challenge is to combine sociological and psychometric measures of risk in a unified explanatory framework in order to understand how external factors like social fragmentation or unemployment can trigger greater risk-taking behaviour in an individual. “Such a coalition,” said Douglas, “would be like going to heaven” [24]. Subsequently, Paul Slovic and others have tried to go there, developing integrated models of risk that identify the psychological mediators of risk perception across different cultures and in different environments.

They conclude that preferential differences for risky decision alternatives are associated with differing perceptions of the relative risk of choice options, rather than differences in attitudes to specific ‘objective’ risks [25] (e). It is the relative weighting of choice options under conditions of risk, rather than personal or group attitudes to a perceived risk, which determines the decision. In other words, in making choices that involve risk, we all weigh up the risks of one option compared to another. Beliefs matter when risks are held constant, but we don’t hold absolute positions when confronted by risky alternatives. We take risks because – all things considered, at least subconsciously – we think the risk is worth taking. These findings hold true in explaining both group and individual differences, as well as situational differences in risk preference.

Although there is empirical evidence for this risk calculus in respect of a number of areas – including health and environmental risk, financial risk, technological failure and the threat of terrorism [26] – processes of sexual choice in the context of life-threatening risk are not well-described. This fact points to a dearth of information that needs to be addressed, which will in time sharpen our analysis. However, I would argue that there is no obvious reason to assume that sexual decision-making will be fundamentally different from other perceived risks. Some of the choices that come into play in a sexual risk calculus may include: the risk of physical violence to a young woman who is without the protection of a man a few years older than herself; the risk of food insecurity to a household if the daughter gives up the older partner who has some source of income; the risk of being excluded from a society where, in the absence of job and educational opportunity, personal affirmation is linked to motherhood [28,f].

Differing perceptions of the relative risk of choice options may also explain why HIV is higher in more polarised societies, and within informal settlements and other areas where wealth and poverty are juxtaposed: When individuals feel devalued relative to others, the gradients between risk options may be much flatter.

The risk calculation is further complicated by the fact that the consequences of personal behaviour kick in at different points in time. For example, if a woman walks down the street without the protection of an older man, the risk to her personal security and wellbeing may be immediate. On the other hand, the risk of unprotected sex demanded by that man may only have health consequences in five to eight years’ time. Thus, in making her decision – the risk of rape now versus the risk of AIDS later – that woman may heavily discount future health benefits. This description may sound like classical utility theories of rational choice, and certainly it draws on behavioural economics which frames options in terms of their opportunity costs [29]. For example, Becker first noted in 1974 that, as women have greater access to the job market, the opportunity costs of childrearing increase – and fertility rates are likely to decrease as a result [30].

But it is different in two important respects:

First, the premise of exponential utility decay (i.e. a consistently negative social discount rate applied to health benefits over time) is at odds with the fact that poor people in Southern Africa are generally optimistic about their future [4,31]. In fact, this finding is typical of human nature. Lowenstein and Prelec [32] recognized this anomaly between the expected utility of deferred benefits and the utility perceived by survey respondents. They proposed that the utility curve related to inter-temporal choice is hyperbolic, rather than exponential in shape. This could explain why long-term health benefits are not heavily discounted, yet pessimistic short-term horizons lead to risky day-to-day decisions.

The second distinction from classical rational choice theory is that, if decisions are shaped by group perceptions of the relative risk of choice options, then the structure of society contributes to shaping preferences. The Cultural Theory of Risk argues that the response to risk across different societies is essentially determined by just two factors, namely (i) the degree to which choices are circumscribed and (ii) the level of social solidarity [33, g]. Those societies with a high degree of choice and social solidarity tend to be more egalitarian and eschew risk, while fatalistic societies are characterised by limited choice and social polarisation. Although empirical backing for Cultural Theory is disputed, it could help explain the gender distribution of HIV (34,35) and points to potential new sites for intervention. It could also explain why times of transition – school-leaving, migration and temporary accommodations – are associated with higher risk-taking, in that social systems of support are disrupted.

It seems that far from being distal determinants of HIV infection, it is the ‘upstream’ factors that directly intrude - consciously or not - into sexual decision-making. They give relative weight to the choices under consideration. They are factored into the calculus of risk through cognitive mediators such as a sense of inclusion or exclusion and presence or lack of perceived choices in life. This insight is of fundamental importance in combating HIV/AIDS in hyper-endemic countries. It means that people take the risk, because having weighed up the choices, they decide that the risk is worth taking. Obviously, this calculation is not as explicit, nor as conscious as the sentence above implies. On the contrary, most sexual decision-making is subliminal and influenced by both by social and individual factors which cannot easily be reduced to a simple causal chain. But unless we can depict the dynamics of risk tolerance more plainly, the above analysis provides few hooks for new intervention and is just as static as the multi-tiered proximal-distal-determinant framework.

One way of reducing the ‘risk-tolerance-of-polarised-societies model’ to simpler form is to ask: what is the cognitive trigger for high-transmission sexual behaviour among people from marginalized communities? At some point, the sense of social exclusion, the lack of educational and employment opportunities and survival pressures tip the balance in decision-making and prompt high-transmission sexual behaviour. The evidence presented by Slovic and Weber above implies that the weighting matters i.e. the process that calibrates one form of risk against the other. In other words, the cue for decision-making is the determination that one option has less risk and more opportunity than another. Perceived risk may thus be redefined as a lack of perceived opportunity in one choice relative to another. It follows that a lack of perceived opportunity is the cognitive trigger for high-transmission sexual behaviour in the face of structural inequality. This definition needs to be further adjusted to take the hyperbolic theory of inter-temporal choice into account, which suggests that individuals may be optimistic for the long-term future, yet pessimistic about the challenges they will face tomorrow. In sum, the conclusion of this line of reasoning is that the cognitive trigger for high risk tolerance is the perception of a lack of real and imminent possibility [h]. This conclusion provides a way of linking the sociological effects of structural inequality to the individual behavioural factors associated with high risk behaviour. It still accepts that poor knowledge and insufficient internalisation of the absolute risk of HIV can lead to risky sexual behaviour, but it helps explain why more people in hyper-endemic countries that know and understand the risk of HIV infection still take the risk [Figure 1].



Figure 1. Proposed chain of mediators between structural inequality and high risk tolerance


3 Implications of a risk tolerance model for HIV infection

The risk tolerance model of HIV transmission in hyper-endemic countries implies that decisions on risk-taking may be amenable to social and cognitive interventions - even in the context of structural inequality. Put another way, we may not be able to change every life circumstance, but we could change the response to circumstance in ways that swing the risk calculus toward safer sexual behaviour. Key objectives would be to enhance social cohesion, expand access to information for personal growth and development, and create bridges that protect specific target groups through times of transition.

For some, this may be a reactionary conclusion, suggesting that fundamental social change is not required. On the contrary, these theories support the view that it is the high degree of social and economic polarisation in Southern Africa - exacerbated by patterns of migration and domestic fluidity rooted in the fragmenting effects of apartheid - which explains the hyper-endemicity of the region. It also provides new avenues to tackle the HIV epidemic, by focusing more systematically on the causal links between structural inequality and HIV transmission that have not been addressed yet, but may be amenable to change.

Practically, what does this mean? If we could identify ‘nexus strategies’ that address the common roots of HIV infection and other social ills, we could potentially catalyse a strong multiplier effect. The causal model of risk tolerance described by Figure 1 above helps identify such nexus strategies which would enhance the perception of social and economic opportunity. One such strategy is the development of leadership-with-opportunity for young people who have proved their commitment to the public good as volunteers in AIDS-related community programmes. Too often, we use school-based peer educators to convey our ‘messages’, and then drop them as soon as they leave school. Yet they could provide a new cadre of sexual-risk-averse leaders for public innovation. Another might be a branded national social network that bridges school and employment, enabling school leavers to be supported through both cell-phone and direct interpersonal contact as they criss-cross the country looking for work. A further example is the development of new links to information for people in marginalised communities, who are doubly disadvantaged by both the relative lack of opportunity and the lack of access to information about the opportunities for education, health and employment that do exist [37]. Another is a comprehensive national programme to combat alcohol abuse, which contributes substantially to both injuries and high-transmission sexual behaviour in Southern Africa [38]. A final example is a focused initiative to ensure that young people complete school. In South Africa, school attendance has been shown to be associated with lower rates of HIV infection and safer sexual behaviour, yet only about 50% of each cohort of scholars completes 12 years of education [39]. Even as supply-side problems in education are addressed, we should understanding and responding to the community-level factors that constrain demand for education.

Obviously, these strategies do not negate the importance of biomedical interventions such as medical male circumcision or behavioural interventions such as discordant couple counselling. On the contrary, they will provide new incentive for the full range of safer sexual practices. More people will not only know what to change, but will be more likely to want to change.

Footnotes

(a) David Harrison is the Chief Executive Officer of the DG Murray Trust, a South African grant-making foundation focusing on early childhood development, literacy, connection to opportunity and leadership development for young people. From 2000-2009, David Harrison was the Chief Executive Officer of loveLife, a national HIV prevention programme for young people in South Africa (www.lovelife.org.za).

(b) An obvious first retort is that many women experience sexual violence and coercion that is outside of their control. While this is so, their experience which is so common in hyper-endemic countries is the consequence of a society that tolerates high levels of sexual violence, leading to HIV infection.

(c) It is important to make the point that not all forms of risk-taking are destructive, and that public innovation requires entrepreneurship and risk taking. In this paper however, ‘risk tolerance’ refers to the risks associated with HIV infection which limit the potential of both individuals and countries in Southern Africa.

(d) For example, in 2003, the adjusted odds ratio for 15-24 year old black men and women (compared to other race groups) was 2.61(1.25-5.47) and 8.33 (4.15-16.71) respectively, controlling for variables such as some socio-economic factors and programme exposure (21).

(e) Lowenstein et al (27) have argued that emotions can influence decisions of risk through mechanisms that are not cognitive, i.e. that do not subject the desirability or likelihood of possible outcomes of decisions to some form of expectation-based calculus. While this ‘risk-as-feelings hypothesis’ may help explain why sexual behaviour is so resistant to consequentialist approaches to HIV prevention, it doesn’t seem to help explain why the spatial distribution of HIV and hyper-endemicity in Southern Africa. For that reason, this line of thinking will not be pursued further in this paper.

(f) Persson and Sjöstedt (36) have reached a similar, though arguably incomplete conclusion. They maintain that risk communication and government intervention has failed to recognize the material and non-material costs (‘sacrifices’) involved in behaviour change. They describe some of these costs as the inconvenience of condom use, partner reduction, interventions in the process of childbirth and denial of breastfeeding. My criticism is that this definition of costs is too narrow and fails to explain how differing life circumstances may affect decisions through for example, the application of varying social discount rates to future health benefits.

(g) Douglas & Wildavsky (33) first made their arguments in relation to risk-taking for public and private innovation, arguing that individualist societies (high grid/low group) were more likely to take risks than more egalitarian societies (high group/high grid).

(h) It must be reiterated that a perceived lack of real and imminent possibility does not only refer to economic possibility, but also stems an individual’s perception of status in society, likelihood of societal and family support, educational prospects, sense of uncertainty etc. This may explain why the poorest countries do not necessarily have the highest rates of HIV infection, and why a country like Brazil with an income concentration not unlike countries in Southern Africa could have such a low rate of infection (<1%).

Table 1

Country Estimated HIV prevalence (total) HIV prevalence 15-49 years (%) Estimated number of people living with HIV
Botswana 15.1% 23.9 300,000
Lesotho 9.9% 23.2 200,000
Mozambique 6.9% 12.5 1,500,000
Namibia 9.5% 15.3 200,000
South Africa 11.7% 18.1 5,700,000
Swaziland 16.3% 26.1 190,000
Zambia 8.7% 15.2 1,100,000
Zimbabwe 10.3% 15.3 1,300,000
Total 10.2% 10,490,000
References: 

1. Holmqvist G. HIV and income inequality: If there is a link, what does it tell us, International Policy Centre for Inclusive Growth ? Brasilia, Brazil: International Policy Centre for Inclusive Growth. 2009. Working Paper No.54.

2. Kagaayi J, Gray R, Ssetaba A, Nabukenya D, Makumbi F, Kiwanuka N et al. Individual self-perception of HIV risk is associated with HIV acquisition. Poster presented at Conference of the International AIDS Society, July 2009; Cape Town.

3. Shisana, O, Rehle, T, Simbayi, L et al. Third South African national HIV, behaviour and health survey 2008: Outcome Indicators for assessing mid-term progress on the NSP 2007-2011. Report 1: A turning tide among teenagers? Cape Town: HSRC Press. 2009

4. Young South Africans, Broadcast Media and HIV Awareness: Results of a national survey. Kaiser Family Foundation and South African Broadcast Corporation. Johannesburg. March 2007

5. World Health Report 2002: Reducing risks, promoting health life. Geneva. World Health Organisation. 2002

6. Wellings K, Collumbien M, Skymaker E, Singh S, Hodges Z, Patel D, Bajos N. Sexual behaviour in context: a global perspective. The Lancet. 2006. 368: 1706 – 1720. Doi:10.1016/S0140-6736(06)69479-8

7. Gupta G, Parkhurst J, Ogden J, Aggleton P, Mahal A (2008). Structural approaches to HIV prevention. Lancet, 372 (9640):764 – 775. Doi:10.1016/S0140-6736(08)60887-9

8. Blas E, Gilson L, Kelly M, Labonté R, Lapitan J et al. Addressing social determinants of health inequities: what can the state and civil society do? The Lancet. 2008; 372:1684-89. Doi:10.1016/S0140-6736(08)61693-1

9. Boerma J, Weir S. Integrating demographic and epidemiological approaches to research on HIV/AIDS: The proximate-determinants framework. Journal of Infectious Diseases 2005; 191(Suppl 1): S61-67. Doi:10.1086/425282

10. Piot P, Russell S, Larson H. Good politics, bad politics: The experience of AIDS. American Journal of Public Health.2007b; 97 (11): 1934-1936. Doi: 10.2105/AJPH.2007.121418

11. Grebe, E (Centre for Social Science Research, AIDs and Society Research Unit). Networks of influence: A theoretical review and proposed approach to AIDS treatment activism. Cape Town. aids2031 Leadership Working Group. 2008. CSSR working paper 3.

12. Friedman S, Mottiar S. Rewarding engagement: The Treatment Action Campaign and the Politics of HIV/AIDS. Centre for Civil Society and the School for Development Studies, University of KwaZulu-Natal. 2004.

13. Halperin D, Epstein H. Concurrent sexual partnerships help to explain Africa’s high prevalence: Implications for prevention. Lancet. 2004; 364(9428):4-6. 16606-3 Doi: 10.1016/S0140-6736(04)16606-3

14. Lurie M, Rosenthal S. Concurrent partnerships as a driver of the HIV epidemic in sub-Saharan Africa? The evidence is limited. AIDS Behaviour (EPub ahead of print). 2009 Doi: 10.1007/s10461-009-9583-5

15. Halperin D, Epstein H. Letter of response to Wellings et al. Lancet. 2007; 369:557.

16. Kenyon C, Zondo S. Riding HIV’s superhighway. Mail & Guardian November 20, 2009. http://www.mg.co.za/article/2009-11-20-riding-hivs-superhighway. Accessed January 2010

17. Hendriksen E, Pettifor A, Sung-Jae Lee, Coates T, Rees H. Predictors of condom use among young adults in South Africa: the Reproductive Health and HIV Research Unit National Youth Survey. American Journal of Public Health. 2007; 97(7): 1241-1248. Doi: 10.2105/AJPH.2006.086009

18. Kenyon C, Badri M. The role of concurrent sexual relationships in the spread of sexually transmitted infections in young South Africans. Southern African Journal of HIV Medicine. 2009; 33: 29-34

19. Jacobson C, Amoateng A, Heaton T. Inter-Racial Marriages in South Africa. Journal of Comparative Family Studies. 2004; 35: 443-458

20. Zungu-Dirwayi N, Shisana S, Udjo E, Mosala T, Seager J. An audit of HIV/AIDS policies in Botswana, Lesotho, Mozambique, Swaziland and Zimbabwe. Research Report. Pretoria: Human Sciences Research Council. 2005

21. Pettifor A, Rees H, Kleinschmidt I, Steffenson A, MacPhail C, Hlongwa-Madikizela L, Vermaak, Padian N. Young people’s sexual health in South Africa: HIV prevalence and sexual behaviours from a nationally representative household survey. AIDS. 2005, 19:1525-1534

22. Pisani E, Garnett G, Brown T, Stover J, Grassly N, Hankins C, Walker N, Ghys D. Back to basics in HIV prevention: Focus on exposure. British Medical Journal. 2003; 326:1384-1387. Doi:10.1136/bmj.326.7403.1384

23. Douglas, M. “The depoliticization of risk”. In: R. J. Ellis & M. Thompson (Eds.), Culture matters: Essays in honor of Aaron Wildavsky. Boulder, Colo.: Westview Press. 1997. p. 121-32

24. Douglas M. “Traditional culture – let’s hear no more about it”. In Rao V, Walton M (eds). Culture and Public Action. Washington DC: The World Bank. 2004

25. Slovic P, Weber E. Perception of risk posed by extreme events. Paper prepared for: Conference on Risk Management Strategies in an Uncertain World, April 12-13, 2002; New York, USA.

26. Weber E, Milliman R. Perceived Risk Attitudes: Relating Risk Perception to Risky Choice. Management Science, 1997; 43: 123-144. Doi: 10.1287/mnsc.43.2.123

27. Loewenstein G, Hsee C, Weber E, Welch N. Risk as feelings. Psychological Bulletin, 2001; 127(2): 267-286. Doi: 10.1037/0033-2909.127.2.267

28. LeClerc-Madlala S. Intergenerational/ age-disparate sex and young women’s vulnerability in Southern Africa. Presentation prepared for: UNAIDS Technical Meeting on young women in HIV hyper-endemic countries of southern Africa, at ICASA 2008. 18-19 June 2008; Muldersdrift, South Africa

29. Oster E. HIV and sexual behaviour: Why not Africa? University of Chicago and National Bureau of Economic Research Working paper. February 2009

30. Becker G. A Theory of Marriage. In: Schultz T (ed.). Economics of the family: Marriage, Children and Human Capital. Chicago: University of Chicago Press. 1974. p. 299-344.

31. Graham C, Hoover M (University of Maryland). Optimism and poverty in Africa: Adaption or a means to survival? South Africa. Afrobarometer Working Papers. November 2007 Working Paper no. 76

32. Lowenstein G, Prelec D. Anomalies in Intertemporal Choice: Evidence and an Interpretation, The Quarterly Journal of Economics. 1992; vol. 107(2), 573-597.

33. Douglas, M, and Wildavsky A. Risk and Culture: an Essay on the Selection of Technical and Environmental Dangers. Berkeley and Los Angeles: University of California Press. 1983

34. Bajos N, Marquet J. Research on HIV Sexual Risk: Social relations-based approach in a cross-cultural perspective. Social Science and Medicine. 2000; 50(11):1533-1546. Doi:10.1016/S0277-9536(99)00463-3

35. Pettifor A, Measham D, Rees H, Padian N. Sexual Power and HIV risk, South Africa. Emerging Infectious Diseases. 2004; 10: 1996 - 2004

36. Persson A, Sjöstedt M. Government structures and behaviour change in the Politics of HIV/AIDS. The Quality of Government Institute, Department of Political Sciences, Göteborg University. 2008. QoG Working Paper Series 2008:10

37. Krishna A. Active Social Capital: Tracing the Roots of Development and Democracy. Columbia University Press. New York. 2002.

38. Seedat M, van Niekerk A, Jewkes R, Suffla S, Ratele K. Violence and injuries in South Africa: Prioritizing an agenda for prevention. The Lancet. 2009 374: 1011- 1022. Doi:10.1016/S0140-6736(09)60948-X 

39. EMIS. School realities 2008. South Africa. Department of Education. September 2008. EMIS08/2/004