RELATIONSHIP BETWEEN ECONOMIC WELLBEING AND RISKY SEXUAL BEHAVIOUR AMONG 14-35 YEAR OLD WOMEN IN LIMPOPO PROVINCE, SOUTH AFRICA Mirabel Akwa Nyamboli 0415275H A Research report submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg; in partial fulfilment for the degree of Master of Science in Medicine in Epidemiology & Biostatistics Johannesburg, 2009 Page | II DECLARATION -------------------------------------------------------------------------------------------------------------- I Mirabel Nyamboli declare that this Research Report is my own unaided work. It is being submitted for the degree of Master of Science in Medicine in Epidemiology & Biostatistics, in the University of the Witwatersrand, Johannesburg. It has not been submitted before for any degree or examination at this or any other university. _____________________ Mirabel Akwa Nyamboli 17 September 2009 Page | III DEDICATION -------------------------------------------------------------------------------------------------------------- In memory of my beloved grandmother Mama Kien Magdalene Awah 1919 ? 2007 Those that live in our hearts never die ************************ Page | IV ABSTRACT -------------------------------------------------------------------------------------------------------------- Objective The study examined the relationship between socioeconomic status (SES) and risky sexual behaviour in young women living in the Limpopo Province of South Africa. Methods Quantitative data were drawn from the 2001?2005 Intervention with Microfinance and Gender Equity Study. The analyses were restricted to 705 sexually active young women aged 14?35 years. Logistic regression models were employed to assess the effect of socioeconomic status (estimated from household asset index and employment status, on risky sexual behaviours). Odds ratios (OR), adjusted odds ratios (AOR) and 95% confidence intervals (CI) are reported. Results After controlling for the potential confounding effects of wealth perception, age, level of education, marital status, birth control and HIV knowledge; there was no statistically significant association between asset index, or employment status and most of the risky sexual behaviours. However, students were significantly more likely to have experienced first sex at or before age 17 years (AOR: 3.80, 95% CI: 2.63, 6.11), and significantly less likely to have had more than three lifetime sex partners (AOR: 0.47, 95% CI: 0.29, 0.78). Discussion Given that age had emerged as a definite predictor of the sexual behaviours that were significantly associated with being a student, and that 75% of the students were aged 14-19 years, it is possible that the associations may be due to an age effect, or a cohort effect. That aside, the results suggest that women in general are equally at risk of HIV infection, hence further research is needed to determine other factors that enhance young rural women?s vulnerability to HIV/AIDS. Conclusion The study did not show any strong evidence to suggest that socio-economic status is associated with risky sexual behaviour. The findings underscore the need to re-examine the assumption that poverty drives risky sexual behaviour so that efforts to ensure that HIV prevention messages get across all strata of society. However, it is possible that SES as used in this study did not differentiate people well enough to be able to identify factors associated with risky sexual behaviour. It is recommended that more research be done to establish how other major factors besides wealth may enhance the vulnerability of women to HIV/AIDS. The current study however contributes to the growing evidence that the relationship between wealth and HIV/AIDS is either exaggerated or is very complex and should be considered when designing new policies, programs and interventions to alleviate the growing HIV incidence. Page | V ACKNOWLEDGEMENTS -------------------------------------------------------------------------------------------------------------- Oh, Alpha and Omega. King of kings, Lord of lords, I gratefully thank you. There is none like you. Indeed you alone are Mighty. Be thou GLORIFIED my Father. I am most indebted to the Belgian Technical Corporation for providing me with a scholarship. This much needed financial support enabled me to complete the course. My sincere thanks go to the most patient, understanding and enduring supervisor, Mr. Edmore Marinda. Indeed sir, your mentoring and guidance have been indispensable to me. I am grateful to Dr Paul Pronyk and RADAR for providing me with the opportunity to do this project. Thank you for the well revised and properly designed data sets, monographs, protocols and questionnaires. Special thanks to my special friends; colleagues; and the incredible staff in the School of Public Health. You all provided academic, emotional, and administrative support when needed, and created a family climate in a restricted environment. My greatest appreciation is to my family and friends especially my parents (Clement and Rophina), and my cousins (Mantso, Nsoh and Sylvie) for their much needed support and encouragement. ? And finally to my one and only Brown: Booboo, you do not have to wait much longer now. It is all coming together. I could not have done it without you. ?THANK YOU ? Page | VI TABLE OF CONTENTS -------------------------------------------------------------------------------------------------------------- DECLARATION ??????????????????????????.. II DEDICATION ??????????????????.?????????. III ABSTRACT ???????...????????????????????... IV ACKNOWLEDGEMENTS ?????????????...????????... V TABLE OF CONTENTS ?.?????????????????????..... VI LIST OF FIGURES?.????????????????????????... VIII LIST OF TABLES ?.????????????????????????... IX NOMENCLATURE .....???????????????????????.... X LIST OF APPENDICES ???????????????????????... XI 1 INTRODUCTION ?.??????????????????????? 1 1.1 Background ??????????????????????........... 1 1.2 Problem Statement ?...????????????????????. 2 1.3 Rationale ?????????????????????????... 3 1.4 Literature Review ?..????????????????????... 4 1.4.1 Risky Sexual Behaviour ????????????................... 4 1.4.2 Risky Sexual Behaviour, HIV/AIDS and Rural Poverty ???.... 6 1.5 Aim and Objectives .??????????????...................??. 8 1.6 Outline of Report ......................................................................................... 9 2 METHODOLOGY ??....????????????????????? 10 2.1 Methodological Framework ......????????????????? 10 2.2 The Current Study...........................?????.??????????.. 11 2.3 Study Context .....??????????????????????.. 11 2.4 Study Sample ................................................................................................ 13 2.5 Definitions .................................................................................................... 13 2.5.1 Risky Sexual Behaviour ????????????.????. 13 2.5.2 Socioeconomic Status ??......??..???????????. 16 2.5.2.1 Household Asset Index as a Proxy for SES ..................... 16 Page | VII 2.5.2.2 Employment Status as a Proxy for SES ........................... 17 2.6 Statistical Analyses ??.???????????????????. 17 2.6.1 Data Management ??????????????????? 17 2.6.2 Variables ??????????????????????.. 18 2.6.2.1 Outcome (Dependent) Variables ?????????.. 18 2.6.2.2 Explanatory (Independent) Variables ???????? 18 2.6.3 Data Analyses ????????????????????.. 19 3 RESULTS ????...??????????????????????? 21 3.1 Socio-Demographic Characteristics .??????????????? 21 3.2 Characteristics of Study Sample by Socioeconomic Status ......????.. 26 3.3 Logistic Regression Analyses ????????????.?????.. 30 3.3.1 Univariate Analyses ...?????????????????.. 30 3.3.2 Multiple Logistic Regression Models ....??????????.. 36 4 DISCUSSION & LIMITATIONS .......................................????????. 40 4.1 Discussion ????????????????????????... 40 4.2 Limitations of Study ?????????????????????. 44 5 CONCLUSION AND RECOMMENDATIONS ???????????? 46 5.1 Conclusion ????????????????????????? 46 5.2 Recommendations ?????????????????????? 47 REFERENCES ?????????????????????????.. 49 Page | VIII LIST OF FIGURES -------------------------------------------------------------------------------------------------------------- Fig 2.1 Map showing location of villages included in the IMAGE study .......... 12 Fig 3.1 Food Security Experience of 14-35 year old women of the 2001 IMAGE study ......................................................................................... 24 Page | IX LIST OF TABLES -------------------------------------------------------------------------------------------------------------- 3.1 Characteristics of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study ???????????????.. 22 3.1 Perception of own wealth, outlook for the future and recent crisis of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study ...................................................................................................... 23 3.3 Sexual behaviour characterisitics and HIV knowledge of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study 25 3.4 Socio-demographic and Sexual Behaviour characteristics of 625 young women who participated in the 2001 IMAGE study by Household Asset Index ............. 28 3.5 Socio-demographic and Sexual Behaviour characteristics of 625 young women who participated in the 2001 IMAGE study by Employment Status ................... 29 3.6 Unadjusted Odds ratios of the associations between selected characteristics and sexual behaviours ................................................................................................. 34 3.7 Adjusted Odds ratios of the associations between selected characteristics and sexual behaviours ................................................................................................. 35 Page | X NOMENCLATURE -------------------------------------------------------------------------------------------------------------- AIDS Acquired Immune Deficiency Syndrome AOR Adjusted Odds Ratio BTC Belgian Technical Corporation CI Confidence Interval DHS Demographic Health Survey DSS Demographic Surveillance Systems HIV Human Immunodeficiency Virus IMAGE Intervention with Microfinance for AIDS and Gender Equity n Number OR Odds ratio P Probability RADAR Rural Aids and Development Action Research Program SD Standard Deviation SES Socioeconomic Status UN United Nations UNAIDS Joint United Nations Program on HIV/AIDS WHO World Health Organization WITS University of the Witwatersrand Page | XI LIST OF APPENDICES -------------------------------------------------------------------------------------------------------------- Appendix I Permission to use RADAR data ............................................. 61 Appendix II Ethics clearance certificate permitting analyses of the IMAGE study ...................................................................... 62 Page | 1 Chapter 1 INTRODUCTION -------------------------------------------------------------------------------------------------------------- 1.1 Background The toll of the acquired immunodeficiency syndrome (AIDS) pandemic has surpassed the worst prediction in Sub-Saharan Africa, home to over two thirds of people living with the human immunodeficiency virus (HIV) worldwide (UNAIDS, 2006). South Africa is amongst the countries worst affected by the HIV/AIDS pandemic. In 1990, the South African national antenatal surveillance data documented an HIV prevalence of 0.7% among women attending public antenatal clinics. By 1993, these prevalence figures had risen to 4%, signalling a dramatic shift in the burden of infection (Tawil et al., 1995; Lee and Zwi, 1996; Mann and Tarantola, 1996). Over the years that followed, the epidemic rose sharply with the prevalence among pregnant women doubling every two years (UNAIDS, 1999; Waldo and Coates, 2000; Sumartojo, 2000; Sumartojo et al., 2000). In 2001, it was estimated that about 4.8 million people in South Africa were infected with HIV, with about 1700 new HIV infections occurring on a daily basis (Ramkisson et al., 2004). By 2002, there were more people living with HIV in South Africa than in any other country in the world. The number of people living with the HIV virus in South Africa was 5.5 million as of 2005, with almost 1000 deaths occurring every day (UNAIDS/WHO, 2006). Accordingly, average life expectancy in South Africa dropped dramatically from 64 to 51 years due to HIV/AIDS (Dorrington et al., 2006). These devastating figures paint a picture now all too familiar to many countries Page | 2 in sub-Saharan Africa - one that raises serious questions about the effectiveness and impact of current intervention strategies. 1.2 Problem Statement Studies in developed countries suggests that people still engage in risky sexual behaviour despite their knowledge of the risks associated with spreading and/or contracting HIV (e.g. McDonald et al., 1996; Rhodes, 1997), however, such studies are limited in the South African context. The adoption of safe sex practices such as the ABC (?Abstain, Be faithful, Condomize?) of prevention can significantly reduce the spread of HIV/AIDS (Cleland and Ferry, 1995; Dodoo and Ampofo, 2001). Yet, such strategies have had a limited meaningful role to play in preventing the spread of HIV/AIDS. Such messages assume that sexual behaviour is a matter of rational individual choice, when in reality sexual behaviour itself is influenced by a range of factors, which include social, cultural, economic, political and technological factors. These factors further determine the extent to which people can access and use methods of HIV prevention, such as condoms, abstinence, and mutual faithfulness (van Donk, 2002). Several studies confirm that sexual behaviour involves complex dynamics e.g., it has been argued that condom use is not an easy option for many people in Africa, especially women (Bujura, 2000; Mill and Anarfi, 2002). In view of the aforementioned, it follows that pertinent information about sexual behaviour is essential in the design and assessment of interventions to improve sexual health. The need to predict and prevent transmission of HIV provides a valuable impetus to sexual behaviour studies, as it has in the past two decades (Cleland and Ferry 1995; Hubert et al., 1998). Importantly too, empirical evidence is needed to correct myths in public perceptions of sexual behaviours (Wellings et al., 2006). Page | 3 1.3 Rationale It seems that wealthier populations do better than poorer ones on most measures of health status, including nutrition, morbidity and mortality and healthcare utilisation (Kuate-Defo, 1997; Adler and Newman, 2002; Fotso and Kuate-Defo, 2005). Consistent with this findings is evidence of an inverse relationship between SES and the risk of sexually transmitted infections (Holtgrave and Crosby, 2003; Bukusi et al., 2006). Now with much of these evidence coming from third world countries, it is reasonable to expect that poverty increases vulnerability to HIV in the same manner in low and middle income countries (Mishra et al., 2007). It has been repeatedly argued that poverty is the root cause of AIDS (e.g. Fitzgerald et al., 2000; Fenton, 2004). Fenton (2004) argued that since poverty plays a role in creating an environment in which individuals are particularly susceptible and vulnerable to HIV/AIDS, poverty reduction will undoubtedly be at the core of a sustainable solution to stopping the HIV/AIDS pandemic. Much of the literature on the relationship between poverty and risky sexual behaviour from South Africa derives from studies on young people?s sexual behaviour, where early sexual debut and multiple sexual partnerships have been linked to poverty but elsewhere have produced mixed results (Ainsworth and Semali, 1998; Deheneffe et al, 1998; World Bank, 1998; Desmond, 2001; Zulu et al. 2002; Akwara et al., 2003; Richens et al., 2003; Wojicki 2005). These results indicate that paradoxically, both wealth and economic disadvantage play pivotal roles in HIV transmission, thus challenging the long held assumptions that poverty reduction is the key to the fight against AIDS (Shelton et al. 2005). Accordingly, a better understanding of the risks between poverty and sexual behaviour is important for improving policies and programs to combat the spread of HIV/AIDS. Page | 4 Several reasons account for the lack of clarity between socioeconomic status (SES), risky sexual behaviour and vulnerability to HIV infection. For example, many cross- country or cross-site comparisons of proof on the associations between SES and risky sexual behaviour are complicated by the different measures used. Composite measures are used to describe risky sexual behaviour (section 1.4.1). Typically, some elements of risky sexual behaviour may be more closely related to wealth status than others; but, a more comprehensive picture may be obtained by examining individual components of risky sexual behaviour. Also, restricting the measure of SES to poverty or standard of living may be reason for this lack of clarity (Mishra et al., 2007). Whereas it is true that a gold standard scale for SES is difficult to visualize, the need to consider multiple distinct domains of SES has been increasingly emphasised (Braveman et al., 2005). 1.4 Literature review 1.4.1 Risky Sexual Behaviour Reminiscent of Sub-Saharan Africa, HIV in South Africa is spread primarily through unprotected heterosexual intercourse (Poku, 2001; Ackermann and Klerk, 2002). Risky sexual behaviour is an important determinant of HIV/AIDS, which contributes substantially to the burden of disease (Ezzati et al., 2003, Collumbien et al., 2004; Slaymaker, 2004). In the absence of an effective vaccine, making positive changes in risky sexual behaviours remain fundamental to stemming the spread of HIV (Cleland and Ferry, 1995; Hope, 1995; Caldwell, 2000). In order to devise effective policies and measures that will protect both people living with HIV/AIDS and their sexual partners (and ultimately to control the spread of the epidemic), it is imperative to understand sexual behavioural patterns and the Page | 5 underlying causes of risky behaviour, as this could be essential in curbing the secondary transmission of HIV/AIDS. In detailing risky behaviours in the sexual domain, Langer et al. (2001) defined risk factors of sex-related behaviours as characteristics, conditions, attitudes, and behaviours which significantly increase the possibilities of experiencing undesirable outcomes related to sexual activities. In the same light, many researchers have defined risky sexual behaviours to include but not limited to: having an early sexual debut, the inconsistent or non-use of condoms or other birth control methods with a non-spousal partner, having multiple sexual partners (concurrent or lifetime), the use of alcohol or drugs prior to sexual activity, transactional sex (Langer et al., 2001; Kotchick et al., 2001; Cooper et al., 2003; Gowen et al., 2004; Konguy et al., 2006). Although many of these definitions are commonly used in literature, hardly any studies use a combination of all of them. In view of the above, it seems that the criteria for defining risky sexual behaviour largely depends on the study population (e.g. drug use is hardly included in studies on rural populations especially in Africa); and the sources of data e.g. demographic health surveys, census data or data collected for other purposes would limit the kinds of questions asked on issues such as drug and alcohol use, birth control use, exchange of sex for money and vice versa, lifetime sex partners and condom use. Albeit, based on this information and the data available, risky sexual behaviours have been defined for the current study as early coital debut before the age of 17 years, having had two or more sex partners in the previous 12 months, a high frequency of lifetime sexual partners (three or more), non-use of a condom at last sex with a non-spousal partner, and transactional sex. These risky sexual behaviours are described in detail in section 2.6.2.1. Page | 6 1.4.2 Risky Sexual Behaviour, HIV/AIDS and Rural Poverty In Africa, substantial resources have been mobilized for research and programme development in efforts to combat HIV/AIDS. Over the past decade, researchers have documented some of the structural factors that facilitate HIV transmission and its concentration within particular geographic areas and populations; most of which can be grouped into three interconnected categories, namely: poverty, mobility and gender inequalities [Zwi 1993; Lee and Zwi 1996; Parker 1996; World Bank 1997). Remarkably, in spite of the uniqueness of each local epidemic, the same general structures and processes can be observed in Africa, Asia, Latin America, as well as certain groups and communities in North America (Parker et al., 2000). According to Burtler (2000), poverty may be a contributing factor, but may not be sufficient for the spread of the HIV epidemic. Following the increasing trend in HIV infections in sub-Saharan Africa, primarily made up of third world or ?poor? countries, the topical public debate on the relationship between poverty and HIV/AIDS has generalised substantial interest in South Africa (Horton, 2000; Makgoba, 2000). ?In the early days of HIV transmission, ignorance of the mechanisms of transmission was not associated with poverty? (Burtler, 2000); but, this is no longer the situation. Poverty plays a major part in the large-scale transmission of opportunistic infections such as tuberculosis, which in the HIV/AIDS era has become an important public health concern (Burtler, 2000). South Africa is a compelling case study in how poverty and economic under- development has fuelled the rapidly growing HIV epidemic. Furthermore, economic crisis has driven many women, either formally or informally, to exchange sex for resources as a means of survival (de Bruyn 1992). As a result, in South Africa as elsewhere, women?s economic vulnerability and dependence on men has heightened their vulnerability to HIV Page | 7 by constraining their ability to negotiate condom use, discuss fidelity with their partners, or leave risky relationships (Roa Gupta 2002). It is the young women who are currently the group at highest risk of infection (UNAIDS 1999; Department of Health 2000). Wherever economic need compels women into earlier sexual debut, to have more sexual partners than they otherwise would, and compromise their ability to negotiate condom use, especially with a non spousal partner, there will be concerns regarding HIV infection. As well, sexual and reproductive health has an important role in elucidating the persistence of poverty. Lack of awareness, as a consequence of low income, cause people not to invest in education, and contribute to a lack of family planning (Booysen and Summerton, 2002). The resulting overpopulation and large families intensify land shortages, which in turn feeds into low production and productivity; and low income; thus, closing the vicious cycle of poverty and disease (Burkey, 1996). Moreover, the AIDS epidemic is likely to root this vicious cycle (Bonnel, 2000); particularly in South Africa where the HIV prevalence is currently amongst the highest in the world (UNAIDS 2006). Sub-Saharan Africa remains predominantly rural despite the rapid urban growth (Dodoo et al., 2007) and is the only continent where the number of rural dwellers will not decrease before 2030 (United Nations, 1998). Dodoo and Tempenis (2002) described that on the whole, rural areas are considered ?bastions of traditional life where as urban centres are influenced heavily by modernization and Westernization?. Apparently, rural areas are sheltered from the ?vices? and media representations of such ? presumed to be prevalent in urban areas (Dodoo et al., 2007). However, people are not as isolated as their urban counterparts because lineage-based supports for example, endorse opportunities for straying from typical ?traditional? behaviours such as extramarital activity (Dodoo et al., 2007). The authors further argue that it would not be senseless to assume that even the same economic difficulty might provoke more economic-based deviance in urban than in Page | 8 rural areas where cash is not as indispensable (Dodoo et al., 2007). Still, the rural poor are exposed to the economic deterioration Africa has experienced over the last three decades. Compounding this is the lack of access to social and health facilities (Zulu et al., 2002). Also, economic stresses are associated with low wages ?with women on average earning less than half of what men would earn (United Nations Development Programme, 2002). Unemployment and increasing poverty presumably has caused some women to use sex to generate income for basic needs, provoking early initiation of sexual activity and high incidence of multiple sexual partnerships (Ulin, 1992; Carael and Allen, 1995). These conditions also prompt men to exploit women?s economic vulnerability by paying very little for sex (Oppong, 1995; Ezeh and Gage, 2000). In this regard, the health implications of early age at first sex and multiple sexual partnerships related to economic deprivation (Zulu et al., 2002), are potentially as relevant in rural areas. Evidence from other developing countries suggests that poorer women are also more likely to have casual sexual partners, with whom condom use is significantly lower (Hallman, 2004). 1.5 Aim and Objectives According to Epstein (1981), cultural background does not determine behaviour but it does provide social actors with a framework for making decisions (Epstein, 1981 ? as cited in Hargreaves et al., 2002 p 800); and apparently, this also describes well the potential importance of SES as a risk factor for HIV infection (Hargreaves et al., 2002), resulting from risky sexual behaviour. Socioeconomic status is conceptualised as the social standing or class of an individual group (American Psychological Association, 2007). Several studies have shown the HIV epidemic to be different between urban and rural populations Page | 9 (e.g. van Donk, 2002), and may also be different in terms of SES. HIV has been characterised as a disease of poverty due to its widening diversity in both rich and poor nations (Hargreaves et al., 2002). Several initiatives are in play to address the problem of poverty and HIV (Cameron, 2000) but few studies have investigated the link between SES and HIV infection ? due to risky sexual behaviour, especially within the poorest populations (Hargreaves et al., 2002) such as rural communities. Using data from the Intervention with Micro-finance for AIDS and Gender Equity (IMAGE) study, the current study is an in-depth analysis of the association between two measures of SES and risky sexual behaviours in rural South Africa. It aimed to examine the patterns of risky sexual behaviour among young rural women aged 14-35 years in the Limpopo province of South Africa. The main objectives of the study were: ? To describe the level of risky sexual behaviour in this population ? To examine the relationship between economic status and risky sexual behaviour Page | 10 Chapter 2 METHODOLOGY -------------------------------------------------------------------------------------------------------------- 2.1 Methodological Framework The current study utilizes quantitative data derived from survey questionnaires, which form part of a wider public health intervention: the Intervention with Micro-finance for AIDS and Gender Equity (IMAGE) study. The IMAGE study was conducted in the Limpopo province South Africa, by the Rural AIDS and Development Action Research (RADAR) Programme in 2001. The RADAR programme has been coordinated through the University of the Witwatersrand?s School of Public Health since 2001. The IMAGE study was an integrated, prospective, randomized, controlled, community-matched intervention trial. The evaluation program integrated data from participatory, qualitative and quantitative methodologies to investigate the impact of a combined poverty-alleviation/gender- empowerment program on sexual behaviour, gender based violence and HIV infection rates. Methods used in the IMAGE study are published elsewhere (RADAR, 2002a, 2002b; Hargreaves et al., 2004; Pronyk et al., 2006). Briefly, the study was built around the prospective follow up of three cohorts in eight rural villages. Appropriate comparison groups were recruited from villages where the program was not operating. Thus there were four intervention and four control villages. Participants in the research were fully informed of the nature of the study; including how and why they had been chosen to participate. Informed consent was voluntarily given by all participants before they were allowed to Page | 11 willingly participate in the study. All data were recorded confidentially and participant anonymity maintained. Quantitative data were collected by trained female facilitators through face-to-face structured interviews. Interviews were concluded by providing local information of HIV counselling and additional support services. The study was approved by Ethics Committees at both the University of the Witwatersrand, Johannesburg South Africa (protocol number M991108, approved 31 January 2000), and the London School of Hygiene and Tropical Medicine (Reference number 598, approved 6 September 2000). 2.2 The Current Study The current study utilized cross-sectional data drawn from the baseline survey of the IMAGE study. Through the ongoing data collection ventures, detailed multi-level longitudinal and cross-sectional data, including but not limited to socio-demographic variables, sexual behaviour, HIV knowledge and perception are available. Permission to use the data was obtained from RADAR (Appendix I). Ethical clearance to perform a secondary data analysis of the IMAGE study was granted by the Human Research Ethics Committee (Medical) at the University of the Witwatersrand, Johannesburg, South Africa (approved 25 January, 2008; protocol number: M071148), (Appendix II). 2.3 Study Context The IMAGE study is set in eight rural villages in the Sekhukhuneland region of South Africa?s rural Limpopo province. A map of the study region is shown below in Figure 2.1. Page | 12 Fig 2.1: Map showing location of villages included in the IMAGE study Source: (RADAR 2002a) Limpopo Province is amongst the most deprived provinces in the country (McIntyre, et al., 2002). With an estimated 9500 households and a total population of over 50,000 people, the population characteristics of the Sekhukhuneland region mirror conditions throughout rural South Africa, as well as other rural regions of Southern and sub-Saharan Africa. HIV prevalence amongst women attending antenatal clinics in the region was estimated at 21.5% [95% CI: 18.5 ? 24.6] in 2006 (Department of Health, 2006). Access to health care Page | 13 facilities and education and life expectancy are consistent with other regions of South Africa. Nearly 50% of the population is under 15 years of age with a mere 5% aged 64 years or older (Udjo & Lestrade-Jefferis, 2000). Unemployment runs in excess of 40% (Lestrade-Jefferis, 2000), and there are very high levels of labour migration among both males and females. Up to 60% of adult men and 25% of women reside away from home for more than 6 months every year (Kahn et al., 2003). Fresh water supply is still problematic, with 47% of households collecting water from taps outside their homes, and even so the supply is irregular (RADAR, 2002a). According to the enumeration data from the 1996 population census supplied by Statistics South Africa, 34% of households reported no cash income and 35% of the adults had no education (RADAR, 2002a). 2.4 Study Sample The current analyses are restricted to 705 sexually active young women, aged 14-35 years, who were resident in the villages where the IMAGE study was operating in 2001. Normal reproductive age for women is usually 15 ? 49 years, but the design of the IMAGE study limited data collection on the variables required for analyses in the current study to the 14 ? 35 year old women. 2.5 Definitions 2.5.1 Risky Sexual behaviour The risk behaviours examined in this study are those that put young people at higher risk of HIV infection, the virus that causes AIDS. These are: 1. Early onset of sexual activity, i.e. coital debut at or before 17 years of age. Page | 14 2. Had two or more sexual partners within the previous year 3. Had more than three life time sexual partners 4. Non use of a condom at last sex with a casual partner 5. Provided sexual favours in exchange for money or gifts (transactional sex). Some of the keywords used to describe sexual behaviour in the context of the current study are defined as follows: a) Coitus, sex or sexual intercourse: This refers to either vaginal or anal intercourse b) Partner types: i) A spouse is someone the respondent was married to, or living with as though they were married. ii) A non-spouse, referred to in this study as a ?casual partner? is anyone else the respondent had had sex with. Accordingly, the partners of all respondents not classified as ?married? or ?living as married? are classified as casual. c) Lifetime sexual partners: This is the total number of different people a respondent ever had sexual intercourse with. These are both spousal and casual partners, including partners with whom the respondent had had coerced, forced or consenting sex with, even if only once. d) Condom use: i) In the affirmative, condom use in this study refers to the actual use of a condom as reported by the respondents. It does not exclude situations where the condom Page | 15 was defective (e.g. had expired, had been stored improperly, such that the quality was compromised, was of poor quality, was broken etc); or used improperly (e.g. inside out, slipped off during intercourse, reused, was put on after pre-ejaculation, etc). ii) Similarly, in the negative, condom use refers to a situation in which a condom was not used at all, even if the respondent had intended to, or attempted to use one but ended up not doing so. e) Condom use at last sex: This refers to the last sexual encounter in which a condom was or was not used. If for example a respondent had sex on three occasions in the same night, and used condoms during the first two occasions but not in the last one; regardless of whether or not it was the same partner, condom use at last sex in this situation is negative. But, if a condom had been used on the last occasion, even if the respondent had never used a condom before in her entire life, then condom use at last sex in this case is affirmative. f) Birth control It is defined in this study as any action a respondent may have resorted to in order to delay or avoid a pregnancy. Birth control methods could be pills/tablets, injectables, implants/norplants, IUD/loop, diaphragm/foam/jelly, calendar/mucus method, female/male sterilization, condoms, withdrawal method or herbs. g) Transactional sex These are sexual relationships in which the giving of gifts or services (such as money or material goods) is an important factor. It is different from prostitution in that the Page | 16 transactional sex provides only a portion of the income of person providing the sexual favour. 2.5.2 Socioeconomic Status (SES) 2.5.2.1 The Household Asset Index as a proxy for SES An asset index was created with variables that measure household ownership of consumer durables (e.g., household assets such as TVs, radios, fridges, cars etc); and aspects of housing infrastructure (e.g., dwelling details such as water/electricity supply, ablution facilities etc). Scoring factors for each of these variables were then estimated with the aid of iterated principal factor analysis as described by Filmer (1998). The statistical technique of deriving weights is referred to as a principal component analysis (PCA). This technique summarises information contained in a set of variables to a smaller number by creating a set of mutually orthogonal components of the data. Intuitively, the first principal component is that linear index of the underlying variables that captures the most common variation among them (Filmer and Pritchett, 1998). Mathematically put, the asset index for asset variables 1 to n was represented as Aj = f1 * (aj1 ? m1) / (s1) + fn * (ajn ? mn) / (sn) (Filmer, 1998) Where: A: represents the asset index, f is the scoring factors or coefficients for each asset, a is the household?s score on a particular asset, m is the mean of each asset variable s is the standard deviation of each asset variable. Page | 17 Young women in this study were assigned the score in the asset index for the particular household to which they belonged. For comparative purposes, scores on the asset index were divided into three equal classes of wealth based percentiles (Method adapted from Booysen, 2002) as follows: i. less than 33.33rd percentile (low SES: very poor) ii. 33.33rd to 66.66th percentile (middle SES: poor) iii. more than 66.66th percentile (high SES: least poor) 2.5.2.2 Employment Status as a proxy for Socioeconomic Status To get a measure of SES using employment status, the women were classified in to four groups as follows: i. Employed full time ii. Employed seasonally iii. Unemployed iv. Students 2.6 Statistical Analyses 2.6.1 Data Management The IMAGE Data had been captured in Microsoft Access files, but were converted and analysed using STATA SE ?version 9? statistical software. Data on household assets, demographics, wealth perception and food security, age, marital status, level of education, employment status, sexual behaviour (including age at Page | 18 sexual debut, birth control use, spousal/co-habiting and casual sexual partners, condom use), and HIV knowledge were extracted from the IMAGE database for the current analyses. Data analyses were then limited to 14 ? 35 year old sexually active women, giving a total of 705 women. 2.6.2 Variables 2.6.2.1 Outcome (Dependent) variables Risky sexual behaviours are the outcome/dependent variables. These are defined as: i. Early age (17 years or less) at coital debut ii. Have had two or more sexual partners within previous 12 months iii. Having more three lifetime sexual partners iv. Not using condoms at last sexual encounter with a casual partner in the past year v. Ever received money or material goods in exchange for sex (* The cut-off point for age at coital debut and for the number of lifetime sex partners were based on the median values reported by participating women). 2.6.2.2 Explanatory (Independent) Variables i. Household Asset Index Following the principal component analysis, the young women were grouped into three categories of SES: 1?Very Poor, 2?Poor, 3?Least Poor ii. Employment Status Page | 19 The women were grouped into four categories of SES as follows: 1? Employed full time, 2? Employed seasonally, 3? Unemployed, 4? Student iii. Age Age was determined as at time of interview and then categorised into year age bands: 1? (31-35 years), 2? (26-30 years), 3? (20-25 years), 4? (14-19 years) to account for the different age cohorts. iv. Marital Status Marital status was defined as 1? Married or living as married, 2? Never married, 3? Separated/divorced or widowed. v. Birth Control Whether a respondent had ever used any form of birth control to delay or avoid a pregnancy in the past was dichotomized as 1 ? No, 2 ? Yes. vi. HIV Knowledge HIV knowledge was based on two questions. The correct response for either of these questions is yes. The responses were categorized into 1? Yes, and 2? No or Don?t Know ? Can a healthy looking person be infected with HIV, the virus that causes AIDS? ? Can an HIV-infected pregnant woman transmit the virus to her unborn child? 2.6.3 Data Analyses Data analyses were limited 705 sexually active women aged 14 ? 35 years. Values for categories of the socio-demographic variables were expressed as absolute numbers and Page | 20 percentages. Unadjusted logistic regression analyses were carried out to investigate the relationship between each socio-demographic variable and the risky sexual behaviours. Multiple logistic regression analyses were then carried out to verify which of the characteristics were collectively associated with risky sexual behaviour. In the logistic regression models, the dependent variables were the risky sexual behaviours (early age at coital debut, having had two or more sex partners in the previous year, having had over three lifetime sex partners, non-use of a condom at last sex with a casual partner, and having engaged in transactional sex). The independent variables were SES estimated from the household asset index and education attainment, age (by age groups), employment, marital status, birth control and HIV knowledge. All the categories denoted as ?1? (section 2.6.2) are the reference groups in the logistic regression models. Every significance test was two-tailed and statistical significance was defined at the alpha level of 0.05. Page | 21 Chapter 3 RESULTS -------------------------------------------------------------------------------------------------------------- 3.1 Socio-Demographic Characteristics A total of 1,456 young persons aged 14-35 years were successfully interviewed at baseline in the IMAGE study; 883 (60.7%) of whom were young women. Seven hundred and five (79.8%) of the 883 women were sexually experienced at the time of the study. All statistical analyses performed were restricted to the 705 sexually experienced young women. All results reported below, unless otherwise indicated, are based on the analyses of the 705 women. More than half of the 705 women i.e. 432 (61.3%) had lived in their villages since birth. Of the 705 women, only two (0.3%) were of Mozambican origin, with the rest (99.7%) being of South African nationality. The majority of the women, i.e. 650 (92.2%) spoke Sepedi, while a few i.e. 55 women (7.8%) spoke Tsonga or some other language as their first language. Selected socio-demographic characteristics of the study sample are shown in Table 3.1 below. Many of the women were quite young, up to 60.3% of them were 25 years of age or younger. The mean age of all women was 23.9 years (standard deviation [SD] 5.5 years). The majority of women (79.6%) had never been married; the rest (17.6%) were married, 2.1% were separated or divorced and less than 1% was widowed (0.7%). The mean age of the 124 married women was 28 [SD: 4.7] years. Their mean age at marriage was 21 (SD: 3.7) years, range (15-34 years). About three quarters of the women (71.6%) Page | 22 had at least one child; the mean parity of these 505 mothers was 1.9 (SD 1.2) children. Almost three quarters (70.3%) of the 705 women had attained at least some secondary education. Unemployment amongst the women was in excess of fifty percent. Table 3.1: Characteristics of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study Characteristic Number (percentage) Age group (in years) 14-19 187 (26.4) 20-25 238 (33.8) 26-30 166 (23.6) 31-35 114 (16.2) Employment Employed full time 117 (16.6) Employed occasionally 26 (3.7) Unemployed 386 (54.8) Student 175 (24.9) Marital status Never married 561 (79.6) Married or living as married 124 (17.6) Separated or divorced 15 (2.1) Widowed 5 (0.7) Level of Education No Schooling 14 (2.0) Attended/completed primary (standard 5) 114 (16.2) Attended some secondary (standard 6-9) 495 (70.3) Completed secondary (standard 10) 63 (9.0) Attended some tertiary 18 (2.5) Page | 23 The young women were asked to describe their own perception of their household wealth status. These results are presented in table 3.2 below. Compared to the wealth status of other households, 415 (64.5%) of the women considered their household wealth to be worse off than most people. In comparison to previous years, 455 (65%) of the women indicated that financially things had been going badly. When asked how easy it would be for the household head to get R50 to pay back an official body by month for the household, 355 (51%) of the women reported it could be done with great difficulty, while up to 204 (29%) admitted it would be virtually impossible. Table 3.2: Perception of own wealth, outlook for the future and recent crisis of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study Wealth perception Number (percentage) Description of own household wealth relative to other households within the village A bit better than most people 95 (13.7) About the same as most people 151 (21.7) A bit worse off than most people 415 (64.6) How things have been going financially in the last year compared to other years Going well 90 (13.0) Going normally 153 (22.0) Going badly 455 (65.0) How easy it is for the Household Head to get R50 if desperately needed to pay back an official body by month end for the household No problem 66 (9.4) Possible but inconvenient 75 (10.7) Possible with real difficulty 355 (50.7) Impossible 204 (29.2) Page | 24 Fig 3.1 below indicates that most of the women had food security problems. In the previous month preceding the study, almost half (46%) of the women had frequently eaten a meal consisting of pap (a traditional porridge made from corn meal) alone, bread alone or worse ? nothing, while over a third (37%) had repeatedly gone without food for an entire day due to food shortage! Fig 3.1: Food security experience of 14-35 year old women of the 2001 IMAGE study 31.71% 8% 22.86% 37.43% Never Once only A few times Often Gone without food or had a reduced amount to eat for a day in the past month 22.14% 6.143% 25.57% 46.14% Never Once only A few times Often Had a meal consisting of pap* alone / bread alone or worse in the past month Page | 25 Table 3.3: Sexual Behaviour Characteristics and HIV Knowledge of 705 sexually experienced 14-35 year old women who participated in the 2001 IMAGE study Characteristic Proportion (percentage) Mean Age at coital debut ( all women) 16.8 (SD 2.0) ? Age at coital debut 10 ? 17 years 451 (64.0) 17 ? 25 years 254 (36.0) Number of sexual partners in last 12 months 0-1 652 (92.5) 2-4 53 (7.5) Mean Number of lifetime sex partners (all women) 2.8 (SD 1.9) ? Number of life time sex partners 1 ? 3 life time sex partners 507 (73.8) 4 ? 22 life time sex partners 180 (26.2) Condom use at last casual sex within the previous 12 months n = 531 * Condom use = Yes 141 (26.6) Condom use = No 390 (73.4) Ever received money and or material goods in exchange for sex n = 531* Yes 67 (12.6) No 464 (87.4) Ever used birth control to delay or avoid pregnancy Yes 528 (75.1) No 175 (24.9) HIV knowledge A healthy looking person could be infected with HIV Yes No or don?t know 431 (61.1) 274 (38.9) A pregnant woman who is HIV positive could pass the virus to her unborn child Yes No or don?t know 629 (89.2) 76 (10.8) ? Data are mean and SD (standard deviation) * n = 531 because data on condom use and transactional sex was only for the last 12 months preceding the study, thus excludes women who did not report any sexual activity in the previous 12 months. The sexual behavioural characteristics of the study sample are illustrated in Table 3.3 above. While well over half of the young women (64.0%) had experienced first sex at Page | 26 or before age 17, almost all (94.8%) had done so at or before age 20. The age at coital debut of 18 women was the same as that at marriage. About a quarter of the 705 women (26.2%) had had four or more sex partners in their life time. Of the 531 women who reported at least one casual sex partner in the previous 12 months, nearly three quarters of them did not use a condom at last sex with their most recent casual sex partner. Furthermore, 67 of the 531 women (12.6%) had received some form of payment paid for sex, while 75% of the 705 women reported having used birth control in the past. Only about half (56%) of the 705 women knew that an HIV infected pregnant woman could pass the virus unto her unborn child, and that a healthy looking person could be HIV positive (combined data for HIV not shown in table above). 3.2 Characteristics of Study Sample by Socioeconomic Status Tables 3.4 and 3.5 below, show the distribution and comparisons of socio-demographic characteristics, sexual behaviour, and HIV knowledge by household asset index and employment status respectively. There were no statistically significant differences in socio-demographic and sexual behaviour between SES groups in all comparisons except for level of education and own perception of wealth (Table 3.4). Women who had completed secondary education (46%) were significantly more likely to be least poor, compared to those who had not (8%, P <0.001) completed secondary education. The women who described their household wealth as worse off than other households (42%) were significantly more likely to be very poor compared to those who said their household wealth was better off than most within the village (20%, P < 0.001). Page | 27 Employment status was also used as a crude proxy for SES (Table 3.5). There were significant differences between all variables compared by employment status, with the exception of HIV knowledge, number of sex partners within the previous year, and lack of condom use at last casual sex. The women aged 31-35 years (45%) were significantly more likely to be employed compared to those aged 26-30 years (26%), those aged 20-25 years (8%) and those aged 14-19 years (3%), (P < 0.001). Women who had never been to school (71%) were significantly more likely to be unemployed, compared to those who had completed secondary education (54%), (P < 0.001). Those who had described their household wealth as worse off than most households (57%) were significantly more likely to be unemployed, compared to those who indicated their household wealth was better off than most in the village (52%), (P = 0.005). Women who had used contraception in the past were significantly more likely to be unemployed (57%), significantly more likely to be employed (18%), and significantly less likely to be students (21%), compared to those who had not (51%, 11%, and 36% respectively); (P < 0.001). Women who had engaged in early first sex (48%), compared to those who had delayed their coital debut till after 17 years (68%), women who had had over three lifetime sex partners (63%) compared to those who had had three or less (52%), and women who had engaged in transactional sex (56%) compared to those who had not (38%), were all significantly more likely to be unemployed (P < 0.001, P = 0.001 and P = 0.033 respectively). Page | 28 Table 3.4: Socio-Demographic and Sexual Behaviour characteristics of 625 young women who participated in the 2001 IMAGE by Household Asset Index Characteristics Low SES (Very Poor) n = 210 Medium SES (Poor) n = 208 High SES (Least Poor) n = 207 P-value Age group (years) 14-19 20-25 26-30 31-35 53 (33.5) 63 (28.9) 59 (39.3) 35 (35.4) 46 (29.2) 76 (34.9) 51 (34.0) 35 (35.4) 59 (37.3) 79 (36.2) 40 (26.7) 29 (29.2) 0.242 Level of Education No schooling Attended/Completed Primary Attended Secondary Completed Secondary 10 (76.9) 45 (46.4) 142 (32.4) 13 (17.1) 2 (15.4) 23 (23.7) 155 (35.2) 28 (36.8) 1 (7.7) 29 (29.9) 142 (32.4) 35 (46.1) <0.001* Marital status Never married Married or living as married Separated/divorced or Widowed 160 (32.0) 45 (41.3) 5 (31.3) 168 (33.6) 34 (31.2) 6 (37.4) 172 (34.4) 30 (27.5) 5 (31.3) 0.425 Description of own household wealth relative to other households within the village A bit better than most people About the same as most people A bit worse off than most people 18 (19.8) 27 (19.4) 165 (42.1) 28 (30.8) 45 (32.4) 132 (33.7) 45 (49.4) 67 (48.2) 95 (24.2) <0.001 Ever used birth control to delay or avoid pregnancy Yes No 151 (32.0) 58 (38.4) 159 (33.7) 49 (32.5) 162 (34.3) 44 (29.1) 0.303 HIV knowledge A healthy looking person could be infected with HIV Yes No / don?t know HIV-infected pregnant woman could pass the virus to her unborn child Yes No / don?t know 123 (32.4) 87 (35.5) 180 (32.2) 30 (45.4) 124 (32.6) 84 (34.3) 191 (34.2) 17 (25.8) 133 (35.0) 74 (30.2) 188 (33.6) 19 (28.8) 0.451 0.093 Mean age at coital debut ? Age at coital debut ?17 years >17 years 16.8 (SD 2.2) 135 (34.4) 75 (32.2) 16.9 (SD 1.9) 130 (33.2) 78 (33.5) 16.9 (SD 1.9) 127 (32.4) 80 (34.3) 0.823 All sex partners in past 12 months 0 - 1 2 ? 4 198 (34.3) 12 (25.0) 190 (32.9) 18 (37.5) 189 (32.8) 18 (37.5) 0.422 Mean number of lifetime sex partners ? Total Number of lifetime sex partners 1 - 3 3 ? 22 5.4 (SD 14.4) 153 (34.0) 51 (32.1) 4.5 (SD 11.9) 148 (33.0) 56 (35.2) 5.5 (SD 15.7) 148 (33.0) 52 (32.7) 0.852 Did not use a condom at last casual sex ? Yes No n = 154 18 (34.8) 36 (27.7) n = 157 114 (33.6) 43 (33.1) n = 158 107 (31.6) 51 (39.2) 0.213 Ever received money and or material goods in exchange for sex ? Yes No n = 154 14 (25.5) 140 (33.8) n = 157 22 (40.0) 135 (32.6) n = 158 19 (34.5) 139 (33.6) 0.399 * P is significant at alpha level of 5% ? Data are mean and SD (standard deviation) of the mean ? Data are for 469 women; n for each category of SES I is as indicated in the table Page | 29 Table 3.5: Socio-Demographic and Sexual Behaviour characteristics of 704 young women who participated in the 2001 IMAGE by Employment Status Characteristics Unemployed n = 386 Employed full time n = 117 Employed occasionally n = 26 Student n = 175 P-value Age group (years) 14-19 20-2 26-30 31-35 50 (45.6) 175 (73.5) 109 (66.1) 52 (13.5) 5 (2.7) 18 (7.6) 42 (25.5) 51 (44.7) 0 (0.0) 6 (2.5) 9 (5.5) 11(9.7) 131 (70.4) 39 (16.4) 5 (3.0) 0 (0.0) <0.001* Level of Education No schooling Attended/Completed Primary Attended Secondary Completed Secondary 10 (71.4) 66 (57.9) 266 (53.9) 44 (54.3) 4 (28.6) 17 (14.9) 77 (15.4) 19 (23.5) 0 (0.0) 4 (3.5) 11 (2.2) 11 (13.6) 0 (0.0) 27 (23.7) 141 (28.5) 7 (8.6) <0.001* Marital status Never married Married or living as married Separated/divorced or Widowed 305 (54.6) 72 (58.1) 9 (45.0) 67 (12.0) 40 (32.3) 9 (45.0) 14 (2.5) 10 (8.1) 2 (10.0) 173 (31.0) 2 (1.6) 0 (0.0) <0.001* Description of own household wealth relative to other households within the village A bit better than most people About the same as most people A bit worse off than most people 49 (51.6) 76 (50.7) 256 (56.8) 13 (13.7) 24 (16.0) 78 (17.3) 7 (7.4) 12 (8.0) 7 (1.6) 26 (27.4) 38 (25.3) 110 (23.4) 0.005* Ever used birth control to delay or avoid pregnancy Yes No 197 (56.5) 89 (50.9) 98 (18.4) 19 (10.9) 22 (4.2) 4 (2.3) 110 (20.9) 63 (36.0) <0.001* HIV knowledge A healthy looking person could be infected with HIV Yes No / don?t know HIV infected pregnant woman could pass the virus to her unborn child Yes No / don?t know 234 (54.4) 152 (55.7) 346 (55.2) 40 (52.6) 76 (17.7) 41 (14.7) 107 (16.9) 10 (13.2) 17 (4.0) 9 (3.3) 24 (3.8) 2 (2.6) 103 (24.0) 72 (26.4) 151 (24.1) 24 (31.6) 0.720 0.476 Mean age at coital debut ? Age at coital debut ?17 years >17 years 17.1 (SD 2.1) 215 (47.7) 171 (67.9) 16.7 (SD 2.2) 77 (17.1) 39 (15.5) 17.5 (SD 1.9) 14 (3.1) 12 (4.8) 16.0 (SD 1.5) 145 (31.2) 30 (11.9) <0.001* All sex partners in past 12 months 0 - 1 2 ? 4 357 (54.7) 29 (58.0) 107 (16.2) 10 (20.0) 25 (3.8) 1 (2.0) 165 (25.3) 10 (20.0) 0.706 Mean number of lifetime sex partners ? Total Number of lifetime sex partners 1 - 3 3 ? 22 3.0 (SD 2.1) 262 (51.9) 114 (63.3) 3.1 (SD 1.8) 78 (15.5) 32 (17.8) 3.1 (SD 2.1) 16 (3.2) 8 (4.4) . 2 (1.4) 149 (29.5) 26 (14.4) 0.001* Did not use a condom at last casual sex ? Yes No n = 284 217 (55.9) 67 (47.5) n = 70 54 (13.7) 16 (11.4) n = 15 10 (2.6) 5 (3.6) n = 161 108 (27.8) 53 (37.6) 0.147 Ever received money and or material goods in exchange for sex ? Yes No n = 284 259 (55.9) 25 (37.9) n = 70 55 (11.9) 15 (21.2) n = 15 13 (2.8) 2 (2.0) n = 161 136 (29.4) 25 (37.9) 0.033* * P-values are significant at an alpha level of 5% ? Data are mean and SD (standard deviation) of the mean ? Data are for 530 women; n is as indicated in the table for each category of Employment status, Page | 30 3.3 Logistic Regression Analyses Univariate and multiple logistic regressions were carried out to determine which characteristics were associated with the risky sexual behaviours. In all the logistic regression models, the dependent or outcome variables were the five risky sexual behaviours: having experienced coital debut at or before 17 years of age, having had two or more sexual partners in the previous 12 months, having had four or more lifetime sexual partners, non-use of a condom at last sex with a casual partner, and having engaged in transactional sex. The main independent or explanatory variables were SES (estimated from household asset index and employment status). Other explanatory variables included in the analyses were age, marital status, contraceptive use and HIV knowledge. Five of the 705 women were widowed, thus were included in the separated/divorced category. This did not affect the results significantly. 3.3.1 Univariate Analyses Results of the univariate logistic regression analyses are presented in Table 3.6. Though not significant, there were slight indications that the least poor compared to the very poor women, were less likely to have initiated early sex (OR: 0.88, 95% CI: 0.59, 1.31), were less likely to have used a condom at last casual sex (OR: 0.64, 95% CI: 0.39, 1.06) and more likely to have engaged in transactional sex (OR: 1.37, 95% CI: 0.66, 2.83). Compared to unemployed women, those employed on a full time basis were significantly more likely to have initiated sex at an early age (OR: 1.57, 95% CI: 1.01, 2.42), and also significantly more likely to have engaged in transactional sex (OR: 2.64, 95% CI: 1.29, 5.39). Though not significant, the employed women were more likely to Page | 31 have had two or more partners within the previous year (OR: 1.61, 95% CI: 0.55, 2.4), slightly less likely to have had over three lifetime sex partners (OR: 0.94, 95% CI: 0.59, 1.50) or to have used a condom at last casual sex (OR: 1.02, 95% CI: 0.55, 1.94), in comparison to the unemployed women. When evaluated against women who described their household wealth as worse off than most households, the women who perceived their households to be better off than others within the village were less likely to have engaged in either of the risky sexual behaviours. These omen who thought their households to be better off than others were also less likely to have engaged in early first sex (OR: 0.88, 95% CI: 0.56, 1.39), to have had two or more sex partners in the previous 12 months (OR: 0.68, 95% CI: 0.26, 1.79), significantly less likely to have had over three lifetime sex partners (OR: 0.51, 95% CI: 0.29, 0.93), less likely to have failed to use a condom at last sex with a casual partner (OR: 0.61, 95% CI: 0.35, 1.06), and less likely to have engaged in transactional sex (OR: 0.81, 95% CI: 0.35, 1.89). The odds of transactional sex and had two or more sexual partners in the previous 12 months increased with age, while the odds of condom use at last casual sex, and the odds of having had over three lifetime partners decreased inversely with age. Some of these associations were significant. Compared to the 30-35 year old women, those aged 14-19 were significantly more likely to have experienced first sex at or before 17 years of age (OR: 11.22, 95% CI: 5.49, 22.9), significantly more likely to have had two or more sex partners in the previous 12 months (OR: 4.24, 95% CI: 1.43, 2.53), significantly less likely to have had over three lifetime sex partners (OR: 0.24, 95% CI: 0.14, 0.45), more likely to have failed to use a condom at last casual sex (OR: 1.14, 95% CI: 0.59, 2.18), and more likely to have engaged in transactional sex (OR: 2.18, 95% CI: 0.72, 6.57). Page | 32 There was suggestion of a trend with level of education and risky sexual behaviour. The odds of engaging in either of the risky behaviours decreased with increasing level of education, thus women with no formal schooling were more likely than any of the other women to have engaged in risky sex. However, none of these associations were statistically significant. These women that had never been to school, compared to those who had completed secondary education were more likely to have initiated early first sex (OR: 2.00, 95% CI: 0.58, 6.91), more likely to have had over three lifetime sex partners (OR: 2.20, 95% CI: 0.67, 7.21), more likely to have failed to use a condom at last casual sex (OR: 1.28, 95% CI: 0.24, 6.86), and more likely to have engaged in transactional sex (OR: 2.61, 95% CI: 0.55, 12.4). In comparison to women that had never married, married women were less likely to have engaged in early first sex (OR: 0.78, 95% CI: 0.52, 1.16), less likely to have had over three lifetime sex partners (OR: 0.77, 95% CI: 0.48, 1.24), but more likely to have not used a condom at last casual sex (OR: 1.47, 95% CI: 0.16, 13.26) to have had at least two sex partners in the previous year (OR: 1.36, 95% CI: 0.23, 3.75), and more likely to have engaged in transactional sex (OR: 1.47, 95% CI: 0.16, 13.26. None of these associations were found to be statistically significant. Not having used birth control was not associated with early sex (OR: 1.00, 95% CI: 0.70, 1.43), having had two or more sex partners in the previous year (OR: 1.00, 95% CI: 0.52, 1.93), or having used a condom at last casual sex (OR: 1.01, 95% CI: 0.65, 1.59). However, women who had never used birth control were significantly less likely to have had three or more lifetime sex partners (OR: 0.46, 95% CI: 0.29, 0.71), and more likely to have been involved in transactional sex (OR: 1.27, 95% CI: 0.72, 2.25), compared to women who had used birth control. Page | 33 Although not significant, when evaluated against women who knew that a healthy looking person could be HIV positive, those who did not were less likely to have had two or more sex partners in the previous year (OR: 0.59, 95% CI: 0.31, 1.11), and in their lifetime (OR: 0.79, 95% CI: 0.56, 1.81). These women that were unaware that a healthy looking person could be HIV positive, were also more likely to have not used a condom at last casual sex (OR: 1.22, 95% CI: 0.82, 1.81) and to have involved themselves in transactional sex (OR: 0.87, 95% CI: 0.51, 1.47). When matched against women who were aware that an HIV-infective pregnant woman could pass the virus to her unborn child, those who did not were more likely to have engaged in sex at an early age at or before 17 years of age (OR: 1.53, 95% CI: 0.90, 2.60), more likely to have had two or more sex partners in the previous year (OR: 1.64, 95% CI: 0.74, 3.65), and less likely to have engaged in transactional sex (OR: 0.73, 95% CI: 0.30, 1.77). Still, none of these associations were significant. Moreover, having had over three lifetime sex partners (OR: 1.00, 95% CI: 0.59, 1.13) and having failed to use a condom at last casual sex (OR: 1.02, 95% CI: 0.56, 1.87) were not associated with knowing that an HIV-infected pregnant woman could pass the virus to her unborn child. Page | 34 Table 3.6: Unadjusted Odds Ratios (OR) of the associations between selected characteristics and sexual behaviours Characteristic Coital debut at 17 years or younger (Yes/No) Had at least two sex partners in the last 12 months (Yes/No) Had more than three lifetime sex partners (Yes/No) Did not use a condom at last casual sex (Yes/No) Transactional sex (Yes/No) Household Asset Index Very Poor Poor Least Poor Reference 0.93 [0.62, 1.38] 0.88 [0.59, 1.31] Reference 1.01 [0.50, 2.04] 0.95 [0.47, 1.94] Reference 1.14 [0.73, 1.77] 1.05 [0.67, 1.65] Reference 0.81 [0.48, 1.35] 0.64 [0.39, 1.06] Reference 1.63 [0.80, 3.32] 1.37 [0.66, 2.83] Employment Status Unemployed Employed full time Employed Occasionally Student Reference 1.57 [1.01, 2.42] 0.92 [0.41, 2.06] 3.84 [0.47, 5.98]* Reference 1.61 [0.55, 2.46] 0.49 [0.06, 3.76] 0.74 [0.36, 1.57] Reference Group 0.94 [0.59, 1.50] 1.15 [0.49, 2.76] 0.40 [0.25, 0.64]* Reference 1.02 [0.55, 1.91] 0.62 [0.20, 1.87] 0.63 [0.41, 1.07] Reference 2.64 [1.29, 5.39]* 1.59 [0.34, 7.47] 1.90 [1.05, 3.44]* Description of own household wealth relative to other households within the village A bit worse off than most people About the same as most people A bit better than most people Reference 1.07 [0.72, 1.57] 0.88 [0.56, 1.39] Reference 0.96 [0.48, 1.95] 0.68 [0.26, 1.79] Reference 1.09 [0.72, 1.66] 0.51 [0.29, 0.93] Reference 1.14 [0.69, 1.88] 0.61 [0.35, 1.06] Reference 1.39 [0.76, 2.53] 0.81 [0.35, 1.89] Age group (years) 31-35 26-30 20-25 14-19 Reference 0.72 [0.44, 1.16] 0.78 [0.49, 1.21] 11.22 [5.49, 22.9]* Reference 1.21 [0.35, 4.24] 2.12 [0.70, 6.44] 4.24 [1.43, 2.53]* Reference 0.86 [0.52, 1.45] 0.69 [0.43, 1.13] 0.24 [0.14, 0.45]* Reference 1.79 [0.87, 3.68] 1.28 [0.67, 2.43] 1.14 [0.59, 2.18] Reference 1.89 [0.59, 6.04] 1.98 [0.66, 5.96] 2.18 [0.72, 6.57] Level of Education No schooling Attended/Completed Primary Attended Secondary Completed Secondary 2.00 [0.58, 6.91] 2.24 [01.22, 4.10]* 1.36 [0.84, 2.19] Reference 0.95 [0.35, 4.54] 0.78 [0.29, 2.12] 0.65 [0.29, 1.47] Reference 2.2 [0.67, 7.21] 0.64 [0.34, 1.21] 0.61 [0.36, 1.01] Reference 1.28 [0.24, 6.86] 1.61 [0.71, 3.63] 0.92 [0.49, 1.73] Reference 2.61 [0.55, 12.4] 0.82 [0.31, 2.13] 0.68 [0.31, 1.47] Reference Marital status Never married Married or living as married Separated/divorced/widowed Reference 0.78 [0.52, 1.16] 2.18 [0.72, 6.62] Reference 1.36 [0.23, 3.75] 1.80 [0.51, 6.37] Reference 0.77 [0.48, 1.24] 2.00 [0.79, 5.09] Reference 1.47 [0.16, 13.26] 1.59 [0.45, 5.67] Reference 1.81 [0.20, 16.42] 2.41 [0.75, 7.70] Ever used birth control to delay or avoid pregnancy Yes No Reference 1.00 [0.70, 1.43] Reference 1.00 [0.52, 1.93] Reference 0.46 [0.29, 0.71]* Reference 1.01[0.65, 1.59] Reference 1.27 [0.72, 2.25] HIV knowledge A healthy looking person could be infected with HIV Yes No / don?t know HIV infected pregnant woman could pass the virus to her unborn child Yes No / don?t know Reference 0.93 [0.68, 1.28] Reference 1.53 [0.90, 2.60] Reference 0.59 [0.31, 1.11] Reference 1.64 [0.74, 3.65] Reference 0.79 [0.56, 1.13] Reference 1.00 [0.59, 1.73] Reference 1.22 [0.82, 1.81] Reference 1.02 [0.56, 1.87] Reference 0.87 [0.51, 1.47] Reference 0.73 [0.30, 1.77] Values are Odds Ratios (95% confidence interval); Reference OR = 1 * Odds Ratio is significant at alpha level of 0.05% Page | 35 Table 3.7: Adjusted Odds Ratios (OR) of the associations between selected characteristics and sexual behaviours Characteristic Coital debut at 17 years or younger (Yes/No) Had at least two sex partners in the last 12 months (Yes/No) Had more than three lifetime sex partners (Yes/No) Did not use a condom at last casual sex (Yes/No) Transactional sex (Yes/No) Household Asset Index Very Poor Poor Least Poor Reference 0.97 [0.63, 1.49] 0.89 [0.57, 1.39] Reference 1.93 [0.83, 4.47] 0.88 [0.88, 5.09] Reference 1.22 [0.77, 1.97] 1.03 [0.63, 1.70] Reference 0.84 [0.47, 1.43] 0.64 [0.39, 1.11] Reference 1.59 [0.74, 3.44] 1.47 [0.66, 3.30] Employment Status Unemployed Employed full time Employed Occasionally Student Reference 1.44 [0.89, 2.34] 0.85 [0.36, 2.01] 3.80[2.63, 6.11]* Reference 1.35 [0.58, 3.18] 0.48 [0.06, 4.05] 0.65 [0.23, 1.83] Reference Group 0.74 [0.42, 1.28] 1.39 [0.53, 3.63] 0.47 [0.29, 0.78]* Reference 0.84 [0.41, 1.70] 0.61 [0.18, 2.05] 0.55 [0.30, 1.03] Reference 2.97 [1.19, 7.35] 1.67 [0.13, 9.66] 1.64 [0.17, 3.73] Description of own household wealth relative to other households within the village A bit worse off than most people About the same as most people A bit better than most people Reference 1.25 [0.80, 1.96] 0.95 [0.57, 1.57] Reference 0.88 [0.38, 2.05] 0.79 [0.28, 2.26] Reference 1.17 [0.72, 1.88] 0.50 [0.27, 0.95]* Reference 1.28 [0.73, 2.24] 0.72 [0.39, 1.32] Reference 1.33 [0.65, 2.71] 0.75 [0.28, 2.00] Age group (years) 31-35 26-30 20-25 14-19 Not included in model Reference 0.68 [0.26, 1.75] 0.37 [0.14, 0.98]* 0.28 [0.08, 1.00] Not included in model Reference 2.27 [1.05, 5.09] 2.06 [0.95, 4.52] 2.10 [0.84, 5.30] Reference 9.21 [1.29, 65.47] 13.85 [1.88, 101.84] 9.62 [1.15, 80.52] Level of Education No schooling Attended/Completed Primary Attended Secondary Completed Secondary 2.77 [0.68, 11.36] 1.73 [0.88, 3.40] 1.00 [0.59, 1.70] Reference 1.54 [0.65, 5.79] 1.30 [0.41, 4.11] 0.87 [0.35, 2.15] Reference 1.92 [0.51, 7.22] 0.76 [0.37, 1.56] 0.73 [0.41, 1.30] Reference 0.74 [0.13, 4.25] 1.18 [0.47, 3.00] 0.74 [0.36, 1.52] Reference 4.31 [0.78, 24.68] 1.06 [0.34, 3.36] 0.68 [0.27, 1.68] Reference Marital status Never married Married or living as married Separated/divorced/widowed Reference 1.02 [0.65, 1.61] 2.10 [0.64, 6.90] Reference 0.68 [0.23, 4.57] 0.87 [0.20, 3.82] Reference 1.83 [1.06, 3.17]* 2.24 [0.65, 7.64] Reference 0.73 [0.07, 7.64] 2.03 [0.4, 8.49] Reference 5.79 [1.05, 15.22 ] 7.68 [1.13, 15.35] Ever used birth control to delay or avoid pregnancy Yes No Reference 0.85 [0.56, 1.27] Reference 2.56 [1.28, 5.12]* Reference 0.43 [0.26, 0.71]* Reference 1.10[0.67, 1.83] Reference 1.36 [0.68, 2.70] HIV knowledge A healthy looking person could be infected with HIV Yes No / don?t know HIV infected pregnant woman could pass the virus to her unborn child Yes No / don?t know Reference 0.85 [0.60, 1.21] Reference 1.38 [0.76, 2.49] Reference 0.59 [0.29, 1.19] Reference 2.25 [0.92, 5.49] Reference 0.82 [0.55, 1.22] Reference 1.50 [0.83, 2.73] Reference 1.22 [0.80, 1.91] Reference 0.91 [0.47, 1.78] Reference 0.72 [0.39, 1.34] Reference 0.81 [0.28, 2.29] Values are Odds Ratios (95% confidence interval); Reference OR = 1 * Odds Ratio is significant at alpha level of 0.05% Page | 36 3.3.2 Multiple Logistic Regression Models Table 3.7 above shows the adjusted odds ratios (AOR) and 95% CI from the multiple logistic regression models. All multiple logistic models had household asset index, employment status, perception of household wealth, age (age group), level of education, marital status, birth control, and HIV knowledge included. However, age was not included in the regression models where early age at coital debut and having had more than three lifetime partners were the outcome variables since both are a function of age. Having adjusted for all other variables, the least poor women, compared to very poor women were less likely to have initiated early sex (AOR: 0.89, 95% CI: 0.57, 1.39) or had over one sex partner in the previous year (AOR: 0.88, 95% CI: 0.86, 5.09), less likely to have failed to use a condom at last casual sex (AOR: 0.64, 95% CI: 0.39, 1.11) and more likely to have been paid for sex (AOR: 1.47, 95% CI: 0.66, 3.30). Having had over three lifetime sex partners was not associated with SES, measured from the household asset index (AOR: 1.03, 95% CI: 0.63, 1.70). When compared to unemployed women, those employed full time, were less likely to have had over three lifetime sex partners (AOR: 0.74, 95% CI: 0.42, 1.28) and also less likely to have not used a condom at last casual sex (AOR: 0.84, 95% CI: 0.47, 1.43); but were more likely to have initiated sex at an early age (AOR: 1.44, 95% CI: 0.89, 1.49), more likely to have had two or more partners in the previous year (AOR: 1.35, 95% CI: 0.58, 3.18), and also significantly (up to 7.35 times) more likely to have engaged in transactional sex (AOR: 2.97, 95% CI: 1.19, 7.35). Nonetheless, besides transactional sex, none of the other associations were statistically significant. When matched against women who described their household wealth as worse off than most households, those who perceived themselves as better off than most households Page | 37 within the village were significantly less likely to have had over three lifetime sex partners (AOR: 0.50, 95% CI: 0.27, 0.95); and though not significant, they were also slightly less likely to have engaged in early sex at or before 17 years of age (AOR: 0.95, 95% CI: 0.57, 1.57), less likely to have had two or more sex partners in the previous year (AOR: 0.79, 95% CI: 0.28, 2.26), less likely to have engaged in transactional sex (AOR: 0.75, 95% CI: 0.28, 2.00), and less likely to have not used a condom at last casual sex (AOR: 0.72, 95% CI: 0.39, 1.32). Compared to the 31-35 year old women, 26-30, 20-25, 14-19 year old women were all less likely to have had two or more sexual partners in the previous 12 months, and this association was significant for the 20-25 year old women (AOR: 0.37, 95% CI: 0.14, 0.98) Also, compared to the 31-35 years age group, the women aged 14-19, 20-25 and 26-30 years were all about two times more likely to have to have failed to use a condom at last casual sex, and over nine times significantly more likely to have engaged in transactional sex. In contrast to women that had completed secondary education, those that had never been to school had a higher odds of having had first sex at or before 17 years of age (AOR: 2.77, 95% CI: 0.68, 11.36), were more likely to have had two or more sex partners in the previous year (AOR: 1.54, 95% CI: 0.65, 5.79), and to have had over three lifetime sex partners (AOR: 1.92, 95% CI: 0.51, 7.22); and had over four times the odds of having had transactional sex (AOR: 4.31, 95% CI: 0.78, 24.7); but these women were less likely to have not used a condom at last casual sex (AOR: 0.73, 95% CI: 0.07, 7.64). Yet, none of these associations were statistically significant. Married women, compared to those who had never married, were significantly more likely to have had over three lifetime sex partners (AOR: 1.83, 95% CI: 2.24); and though not significant, married women were also more likely to have engaged in transactional sex Page | 38 (AOR: 5.79, 95% CI: 1.05, 15.22) and used a condom at last casual sex (AOR: 0.73, 95% CI: 0.07.64), but less likely to have had two or more partners in the previous year (AOR: 0.68, 95% CI: 0.23, 4.57). In comparison to women who had used birth control, those who had not were significantly more likely to have had two or more sex partners in the previous year (AOR: 2.56, 95% CI: 5.12), but significantly less likely to have had over three lifetime sex partners (AOR: 0.43 95% CI: 0.26, 0.71); and though none of the following associations were not significant, women who had never used birth control were also less likely to have had early sex at of before 17 years (AOR: 0.85, 95% CI: 0.56, 1.27), but more likely to have not used a condom at last casual sex (AOR: 1.10, 95% CI: 0.67, 1.83), and to have engaged in transactional sex (AOR: 1.36, 95% CI: 0.68, 2.70). HIV knowledge was not significantly associated with either of the risky sexual behaviours. Women who did not know that a healthy looking person could be HIV positive were less likely to have initiated sex at of before 17 years of age (AOR: 0.85, 95% CI: 0.60, 1.21), had two or more sex partners in the previous year (AOR: 0.59, 95% CI: 0.29, 1.19) or over three lifetime sex partners (AOR: 0.82, 95% CI: 0.55, 1.22), and less likely to have engaged in transactional sex (AOR: 0.72, 95% CI: 0.39, 1.34), even though they were more likely to have not used a condom at last casual sex (AOR: 1.22, 95% CI: 0.80, 1.91). Women who knew that an HIV-infected pregnant woman could pass the virus to her unborn child, compared to those who did not, were those who did not were more likely to have engaged in early first sex at or before 17 years of age (AOR: 1.38, 95% CI: 0.76, 2.49), more likely to have had two or more partners in the previous year (AOR: 2.25, 95% CI: 0.92, 5.49) and over three lifetime sex partners (AOR: 1.50, 95% CI: 0.83, 2.73), but were less likely to have failed to use a condom at last casual sex (AOR: 0.91, 95% CI: 0.47, Page | 39 1.78) and less likely to have engaged in transactional sex (AOR: 0.81, 95% CI: 0.28, 2.29). As well, none of these associations were statistically significant. The key findings are discussed in chapter 4. Page | 40 Chapter 4 DISCUSSION & LIMITATIONS -------------------------------------------------------------------------------------------------------------- 4.1 Discussion The current study examined the relationship between socioeconomic status (SES) and risky sexual behaviour among young rural South African women aged 14-35 residing in the Limpopo province of South Africa. Data on SES and HIV risk-related sexual behaviours for 705 sexually active women were examined for significant associations while controlling for other variables. In the univariate logistic regression analyses, the variable ? age, emerged as a definite predictor of early first sex at or before 17 years of age and having had over three lifetime sex partners. It is not clear whether the association between age and these two risky sexual behaviours, was due to a cohort effect or an age group effect. Consequently, the variable age, was not included in the adjusted models where the two afore-mentioned risky sexual behaviours were the outcome variables. In the adjusted logistic regression models, keeping all other variables in the models constant, there was no statistically significant association between SES and most of the risky sexual behaviours. There was a non-significant indication that the least poor compared to the very poor women, were less likely to have engaged in the risky sexual behaviours. The least poor women were less likely to have initiated early sex, had two or more sex partners in the previous year, and less likely to have failed to use a condom at last casual sex; but they were also more likely to have been paid for sex. Similarly, compared to unemployed women, those employed on a full time basis were less likely to have engaged Page | 41 in transactional sex, while unemployed women were more likely to have not used a condom at last casual sex. But the women employed on a full time basis were more likely to have initiated sex at or before 17 years of age, had two or more partners in the previous year but were less likely to have had over three lifetime sex partners. As well, none of these associations were statistically significant. However, students were significantly more likely to have initiated early sex and significantly less likely to have had more than three lifetime sex partners, compared to the unemployed women. Overall, the results indicate that SES did not correlate with risky behaviour on all aspects. Nonetheless, there was some weak indication that higher SES women in this study (the least poor and the employed women) had the highest odds of engaging in transactional sex, but were most likely to have used a condom at last casual sex. It could be inferred that higher levels of economic wellbeing places one at increased risk of HIV infection by exhibiting risky sexual behaviours like transactional sex. This contradicts the long held belief that it is the lower SES women who engage more in risky sexual behaviours. Some authors have argued that having a higher SES may make it possible to afford sexual relationships with many partners (e.g. Kongyuy et al, 2006; Msisha et al., 2008). This explanation would make sense for other risky sexual behaviours such as multiple partners, but not for transactional sex, which in this study is defined as having ever had sex in exchange for money or material goods, thus the respondents in this study are not paying for, but rather are being paid for sex. This leads one to suppose that the higher SES women exhibiting such behaviour, may have became financially better off because they were getting paid for sex, which could also explain why they had more lifetime sex partners. Also, high SES women were less likely to have not used a condom at last casual sex compared to low SES women. This result points to the ability of high SES women to afford condoms (due to their employment status), or that they know about where to get condoms Page | 42 (due to a higher education status); or may use condoms because they engage in these risky sexual behaviours. This is in keeping with other studies that have found similar results (Kapiga and Lugalla, 1996; Bloom et al., 2002). Low SES groups were also associated with early age at marriage, poor HIV knowledge, and non-use of birth control. Some researchers have found similar results in other settings (Hargreaves et al., 2002; Shelton et al., 2005; Fenton, 2004; Kongyuy et al., 2006). Such sexual behaviours will remain important risk factors for infection in a widespread generalised HIV epidemic (Hargreaves et al., 2002). This is particularly true if condom use remains low; and would suggest that those of low SES may be at high risk of acquiring HIV infection (Hargreaves et al., 2002). Other analyses divulged that level of education (which has been used as a proxy of SES in several studies (e.g. Kongyuy et al., 2004), whether or not a person was currently a student and own perception of wealth correlated positively with household wealth and employment status. It fits the norm of wealthier persons usually being more educated, and educated people are usually more likely to have better job opportunities, and hence be wealthier. Using household survey data from Demographic Health Surveys from over 44 surveys in 35 countries, Filmer and Pritchett (1998) showed that people of higher SES households were more likely to enrol in school and attain higher levels of education. As well, a study by Hargreaves et al. (2008) showed school attendance to be protective against risky sexual behaviours. When compared, women with good HIV knowledge were more likely to have protected themselves with a condom at last casual sex have, whereas those with poor HIV knowledge were more likely to a have engaged in early first sex at or before 17 years of age, had two or more sex partners in the previous year and have over three life time partners. This finding suggests that knowledge may have been translated into behaviour Page | 43 change (for condom use). While this may be considered a reassuring finding, it is imperative to use caution when interpreting it. The way these questions were asked during the IMAGE study does not allow one to determine whether the risky sexual encounter behaviour of condom at last casual sex), in fact postdates the acquiring of HIV/AIDS knowledge. Yet, it is plausible that this positive finding is a true reflection of sexual behaviour in the population. A person cannot change when they had their first coital debut; having a new partner is not a rapid occurrence; but having sex with a casual partner, may be a frequent activity. Thus change in sexual behaviour would be more apparent when measured with activities such as condom use that occur more frequently. Another interesting finding was that regardless of SES, all the women who showed increased odds of having engaged in transactional sex did not necessarily have a high prevalence of sexual partners within the previous year, or lifetime. It seems that transactional sex in this population may be more of a permanent arrangement between the partners, for example, a woman may opt to offer sex for a place to stay, thus this partner would be more permanent, or some women may actually be in more polygamous relationships simply for monetary or material gain. More differences were noted between comparison groups when the women were classified by employment status compared to when they were classified by the household asset index. Firstly the household as a basic unit of analysis may be inadequate for much of the research on poverty and risky sexual behaviour. It is often the case that households participate in saving schemes and social networks that reach outside the household. In understanding the impact of SES level of risky sexual behaviour, it is necessary to understand these broader networks as well as the adjustments made within households (IUSSP, 2005). Yet it is noteworthy to mention that SES in the communities where the IMAGE intervention is operating was generally the same at baseline. Although one can Page | 44 range individuals in this group as relatively poor among themselves, this classification may not have differentiated them well enough. It was assumed in this study that relative ranking of individuals in to three SES groups differentiated participants, but it may have been that the individuals in this study were relatively very similar in terms of house hold assets. Employment status seems to have differentiated participants better than the SES groups. 4.2 Limitations of Study The relationship between economic wellbeing and risky sexual behaviour was explored with reference to various demographic and sexual behaviour characteristics. Such comparative analysis, however, has certain limitations: firstly, because the analyses are mainly descriptive and the design of the survey is cross-sectional, thus cannot be employed in the analyses of the causal relationship between economic wellbeing and risky sexual behaviour; and in the second instance, such analyses have only provided an economic perspective on poverty, and does not include parameters of social exclusion as determinants of socio-economic status (World Bank, 2000; Mwabu, 2001). Certain questions required respondents to recall past and recent behaviours. The reliability of the responses received can vary between people in important ways. In the current study, condom use in particular had a high proportion of inconsistent responses. Such findings had been earlier reported by Fenton et al. (2001a), who reported that the accuracy of recall varied by age, number of sexual partners and how far back participants were asked to remember. In the current study, individuals were asked to recall several events which had occurred over a year ago. Some of the participants included in the current study seemed to have struggled to recall how often they used a condom with their partners and whether a condom was used with all partners. Questions on condom use triggered large Page | 45 numbers of inconsistencies in the IMAGE data, where for example individuals reported no condom use in the last year but then did report condom use with an individual partner; or in other cases respondents reported having always consistently used a condom yet when asked if they had used a condom at last sex with a non-spousal partner, the responses were negative for up to three partners. Zenilman et al. (1995) noted that not only do individuals struggle to recall condom use accurately, but they also may report on ?use? rather than ?correct use?. Consequently, condom breakages or slippages, for instance, which would increase HIV risk of HIV acquisition, would not be reported, thus diluting the strength(s) of any association(s). Because sexual behaviour requires the provision of personal and often intimate information, some people may be more willing to participate in the research than others. This can lead to participation bias if there are systematic differences, for example in age, sex and social class between those who agree to participate and those who do not (Fenton et al., 2001a). In this study, males were excluded thus sex cannot be counted as a bias. However, age was a potential bias. The 31-35 year old female formed only 16% of the study sample, which is less than half of the 20-25 age group that made up 33% of the study sample. Even amongst those who agree to participate in a study, not all questions will be answered. Item response bias can arise where people who choose not to answer a question, may have engaged in risky sexual behaviours which are systematically different to that of of those who choose to answer it (Fenton et al., 2001a) Page | 46 Chapter 5 CONCLUSION AND RECOMMENDATIONS -------------------------------------------------------------------------------------------------------------- 5.1 Conclusion The current study did not show any strong evidence to suggest that socio-economic status is associated with risky sexual behaviour. It is possible that women in general (all levels of SES) are at risk of HIV infection, which could attest to the indiscriminate nature of the HIV/AIDS epidemic. The findings underscore the need to re-examine the assumption that poverty drives risky sexual behaviour so that efforts to ensure that HIV prevention messages get across all strata of society. However, it is possible that SES as used in this study did not differentiate people well enough to be able to identify factors associated with risky sexual behaviour. Furthermore, additional research is required to establish how other major factors besides wealth (such as gender, power, partner violence, beliefs and perceptions etc) may enhance the vulnerability of women to HIV/AIDS. The current study however contributes to the growing evidence that the relationship between wealth status and HIV acquisition is either exaggerated or is very complex, and should be considered when designing new policies, programs or interventions to alleviate the growing HIV incidence. Page | 47 5.2 Recommendations It is recommended that several measures of SES be utilised in studies of this nature, rather than individual proxies. Using composite measures or several indicators of SES would be more robust as a way of sensitivity analysis. It may be worthwhile to use a clinically ascertained outcome (e.g. HIV status or other sexually transmitted infections) as opposed to using self-reported risky sexual behaviours for ascertaining the social epidemiology of sexually transmitted infections. It is challenging to coin a standard definition of risky sexual behaviour. For example, how many sex partners should a person have before their behaviour is considered risky? Is it more than one; above three; over five, or ten? Several studies have used very different definitions (e.g. Mertz et al., 2000; Fenton et al., 2001a, 2001b; Latino, 2002; Jaccard et al., 2004; Konguy, 2006 etc). This can make comparisons between studies difficult. A number of definitions of risky sexual behaviour (engaging in coitus at or before age 17; having had two or more sex partners within the previous 12 months; having had over three life time sex partners; non-use of a condom at last casual sex; and having been paid money or material goods for sex) were used in this study. Adopting one, all, or more of these definitions, as with other clinical variables could ensure that when researchers refer to risky sexual behaviour, it is based on the same measure(s). Sonkin and Hinde (2007) advocated that being able to confidently use a single or combination of measures such as those suggested above would make life easier for researchers and participants. It would also ensure that fewer and less personal questions have to be asked. The authors added that intrusion into one?s personal life is really only ethical if it adds substantially to better understanding of the risky sexual behaviours (Sonkin and Hinde, 2007). One would concur with Sonke and Hinde (2007) in that by Page | 48 simply asking people ?Have you been sexually involved with partner A or B? etc?? would give enough information to determine that the person has had sex with that partner rather than probing further into knowing how many times in a month the person has engaged in sexual activity with the partner in question. In summary, the usefulness of the current study may extend beyond shedding light on the link between economic wellbeing and risky sexual behaviour, and simply adopting a simple uniform measure for risky sexual behaviour. It also expands current understanding of the distribution of the risky sexual behaviour within an important group of people, which can in turn inform program or policy efforts to reduce HIV incidence. Page | 49 REFERENCES -------------------------------------------------------------------------------------------------------------- Ackermann L, De Klerk GW. 2002. Social factors that make South Africa women vulnerable to HIV infection. Health Care Women International, 23: 163-172. Adler NE and Newman K. 2002. Socioeconomic disparities in health: Pathways and policies. Health Affairs, 21(2), 60-76. Ainsworth M and Semali I. 1998. Who is most likely to die of AIDS? Socioeconomic correlates of adult deaths in Kagera region, Tanzania. In: Ainsworth, M., Fransen, L., Over, M. (Eds). Confronting AIDS: Evidence from the developing world. Selected background papers for the World Bank Policy Research Report, Confronting AIDS: Public priorities in a global epidemic. European Commission, Brussels, pp. 95-109. http://europa.eu.int/comm/development/aids/limeletter/html/lim06.htm [Accessed 20th May, 2007] Akwara PA, Madise NJ and Hinde A. 2003. Perception of HIV/AIDS and sexual behaviour in Kenya. Journal of Biosocial Science, 35: 385-411 American Psychological Association. 2007. HIV/AIDS and Socioeconomic Status available online: http://www.apa.org/pi/ses/hiv_fact_sheet.pdf [accessed 24 Nov 2008] Beltran ED, Ostrow DG and Joseph JG. 1993. Predictors of sexual behaviour change among men requesting their HIV-1 antibody status: The Chicago MACS/CCS cohort of homosexual/bisexual men, 1985-1986. AIDS Education and Prevention, 5: 185-195. Bloom SS, UrassaM, Isingo R, Ng?weshemi J, Boerma JT. 2002. Community effects on the risk of HIv infection in rural Tanzania. Sexually Transmitted Infections, 78: 261-266 Page | 50 Booysen F le R. 2002. Using Demographic and Health Surveys to measure Poverty: An Application to South Africa. Journal for Studies in Economics and Econometrics, 26(3): 53-70. Booysen F le R and Summerton J. 2002. Poverty, risky sexual behaviour, and vulnerability to HIV infection: evidence from South Africa. Journal of Health and Population Nutrition, 20(4): 285-288. Braveman PA, Cubbin C, Egerter S et al. 2005. Socioeconomic status in health research: one size does not fit all. Journal of the American Medical Association 294: 2879-2888 Bonnel R. 2000.c What Makes and Economy HIV-Resistant? ACT Africa, World Bank. Bujura J. 2000. Risk and trust: unsafe sex, gender and AIDS in Tanzania. In P Caplan (Ed.), Risk revisted pp. 59-84. London Pluto Press. Bukusi EA, Cohen CR, Meier AS, Waiyaki PG, Nguti R, Njeri JN and Holmes KK. 2006. Bacterial vaginosis: risk factors among Kenyan women and their male partners. Sexually Transmitted Diseases, 33:361-7. Burkey S. 1996 People first: A Guide to Self-Reliant, Participatory Rural Development London: Zed Books. Burtler C. 2000. HIV and AIDS, poverty and causation. Lancet, 356:1445. Caldwell JC. 2000. Rethinking the African AIDS epidemic. Population and Development Review, 26: 117-135. Carael M and Allen S. 1995. Women?s vulnerability to HIV/STD in sub-Saharan Africa: An increasing evidence. In Makinwa P. and Jensen AM (Eds.) Women?s position and demographic change in sub-Saharan Africa (pp 2201-2222). Liege, Belgium: International Union for the Scientific Study of population (IUSSP). Page | 51 Cleland J and Ferry B. 1995. Sexual behaviour and AIDS in the developing world. Geneva: World Health Organization Collumbien M, Gerressu M, and Cleland J. 2004. Non use and use of ineffective methods of contraception. In: Ezzati M, Lopez A, Rodgers A, Murray C, (Eds). Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization, 1255?320. Cooper, ML, Wood, PK, Orcutt, HK and Albino A. 2003. Personality and the predisposition to engage in risky or problem behaviours during adolescence. Journal of personality and Social Psychology, 84(2): 390-410. de Bruyn, M. 1992. Women and AIDS in developing countries. Social Science and Medicine, 34(3): 249-262. Deheneffe, J., Carael, M., Noumbissi, A. 1998. Socioeconomic determinants of sexual behaviour and condom use. In: Ainsworth, M., Fransen, L., Over, M. (Eds). Confronting AIDS: Evidence from the developing world. Selected background papers for the World Bank Policy Research Report, Confronting AIDS: Public priorities in a global epidemic. European Commission, Brussels, pp. 95-109. http://europa.eu.int/comm/development/aids/limeletter/html/lim08.htm. [Accessed 20th May, 2007] Department of Health, South Africa. 2000. National HIV and Syphilis Sero-Prevalence Survey of women attending Public Antenatal Clinics in South Africa. Department of Health: Pretoria. Department of Health, 2006. National HIV and syphilis antenatal seroprevalence survey in South Africa 2005. Department of Health: Pretoria. Desmond D. 2001. Joint epidemics: poverty and AIDS in sub-Saharan African. Havard International Review, Fall: 54-58. Page | 52 Dodoo FN and Ampofo AA. 2001. AIDS-related knowledge and behaviour among married Kenyan men: A behavioural paradox? Journal of Health and Human Services Administration, 24(2): 197-231. Dodoo FN and Tempenis M. 2002. Gender, power, and reproduction: Rural-urban differences in the relationship between fertility goals and contraceptive use in Kenya. Rural Sociology, 67(1):46?70. DodooNF, Eliya ZM and Ezeh AC. 2007. Urban-rural differences in the socioeconomic deprivation Sexual behaviour link in Kenya. Social Science & Medicine, 64: 1019-1031 Dorrington RE, Johnson LF, Bradshaw D and Daniel T. 2006. The Demographic Impact of HIV/AIDS in South Africa. National and Provincial Indicators for 2006. Cape Town: Centre for Actuarial Research, South African Medical Research Council and Actuarial society of South Africa. Epstein AL. 1981. Urbanisation and kinship: the Domestic Domain on the Copperbelt of Zambia, 1950-56. Academic Press, London ? [cited in: Hargreaves RJ, Morison, LA, Chege J, Ruteburg N, Kahindo M, Weiss AH, Hayes R, and Buv? A. 2002. Socioeconomic status and risk of HIV infection in an urban population in Kenya. Tropical Medicine and International Health, 7(9): 793-802 pp 800] Ezeh AC and Gage AJ. 2000. Domestic violence in Uganda: Evidence from qualitative and quantitative data. African Population and Health Research Centre Working Papers no 18. Nairobi: Population Council. Ezzati M, Vander Hoorn S, Rodgers A, Lopez AD, Mathers CD, Murray CJL. 2003. Estimates of global and regional potential health gains from reducing multiple major risk factors. Lancet, 362: 271?80. Fenton K, Korovessis C, Johnson A, McCadden A, McManus S, Wellings K, Mercer C, Carder C, Copas A, Nanachahal K, Macdowall W, Ridway G, Field J and Erens B. 2001a. Page | 53 Sexual behaviour in Britain: Reported sexually transmitted infections and prevalent genital Chlamydia trachomatis infection. Lancet, 358: 1851-1854. Fenton K, Johnson A, McManus S and Erens B. 2001b. Measuring sexual behaviour: Methodological challenges in survey research. Sexually Transmitted Infections, 77: 84-92. Fenton L. 2004. ?Preventing HIV/AIDS through poverty reduction: the only sustainable solution?? Lancet, 364: 1186-7. Filmer, D. 1998 Socioeconomic correlates of sexual behaviour: a summary of results from an analysis of DHS data In: Ainsworth, M., Fransen, L., Over, M. (Eds). Confronting AIDS: Evidence from the developing world. Selected background papers for the World Bank Policy Research Report, Confronting AIDS: Public priorities in a global epidemic. European Commission, Brussels, pp. 95-109. (http://europa.en.int/comm/development/aids/limeletter/tml/lim07 [accessed 20th May 2007]. Filmer D and Pritchett L. 1998. Estimating Wealth Effects without Expenditure Data ? or Tears: An Application to Educational Enrollments in States of India. World Bank Policy Research Working Paper no 1994. Washington, DC: Development Economics Research Group (DECRG), The World Bank 6. Fitzgerald DW, Behets F, Caliendo A, Roberfroid D, Lucet C, Fitzgerald JW and Kuykens L. 2000. ?Economic hardship and sexually transmitted diseases in Haiti?s rural Artibonite Valley,? American Journal of Tropical Medicine and Hygiene 62(4): 496?501. Fotso JC and Kuate-Defo B. 2005. Measuring Socioeconomic Status in Health Research in Developing Countries: Should we be Focusing on Household, Communities or Both? Social Indicators Research 72(2): 189-237 Gowen LK, Feldman SS, Diaz R, and Yisrael, DS. 2004. A comparison of the sexual behaviours and attitudes of adolescent girls with older versus similar-aged boyfriends. Journal of Youth and Adolescence, 33(2): 167-175. Page | 54 Hallman K. 2004. ?Socioeconomic Disadvantage and Unsafe Sexual Behaviors Among Young Women and Men in South Africa.? Policy Research Division Working Paper No. 190. New York: Population Council. http://www.popcouncil.org/publications/wp/prd/rdwplist.html#2004 Hargreaves RJ, Morison, LA, Chege J, Ruteburg N, Kahindo M, Weiss AH, Hayes R, and Buv? A. 2002. Socioeconomic status and risk of HIV infection in an urban population in Kenya. Tropical Medicine and International Health, 7(9): 793-802 Hargreaves RJ, Pronyk PM, Kim JC, Makhubele MB, Morison L.A., et al. 2004. The Intervention with Microfinance for AIDS and Gender Equity Study (IMAGE Study): an integrated community-randomised trial of a structural intervention to prevent HIV and gender-based violence in South Africa (03PRT/24). http://www.thelancet.com/journals/lancet/misc/protocol/03PRT-24 [Assessed 17th June, 2007] Hargreaves RJ, Morison LA, Kim JC, Bonelle PC, Porter DHJ, Watts C, Busza J, Phetla G and Pronyk MP. 2008. The association between school attendance, HIV infection and sexual behaviour among young people in rural South Africa. Journal of Epidemiology and Community Health, 62: 113-119. Holtgrave DR and Crosby RA. 2003. Social capital, poverty, and income inequality as predictors of gonorrhoea, syphilis, chlamydia and AIDS cases rates in the United States. Sexually Transmitted Infections, 79:62-4. Hope KR Sr. 1995. The socio-economic context of AIDS in Africa. Journal of Asian and African Studies, 30:80?89. Horton R. 2000. Mbeki defiant about South African HIV/AIDS strategy. Lancet, 356:225 Hubert M, Bajos N and Sandfort T, (Eds). 1998. Sexual behaviour and HIV/AIDS in Europe. London: University College London Press. Page | 55 IUSSP. 2005. A report back on the IUSSP seminar on ?Poverty and HIV/AIDS? held at the University of Cape Town, 12-14 December, 2005. Jaccard J, Macdonald R, Wan C, Guilamo-Ramos V, Dittus P and Quinlan S. 2004. Recalling sexual partners: the accuracy of self-reports. Journal of Health Psychology, 9: 699-712. Kahn K, Collinson M, Tollman S, Wolff B, Garenne M and Clark SJ. 2003. Health consequences of migration : Evidence from South Africa?s rural northeast. In African Migration In Contemporary Perspective Conference, Johannesburg. Kapiga SH and Lugalla JL. 1996. Sexual behaviour patterns and condom use in Tanzania: results from the 1996 Demographic and Health Survey. AIDS Care, 14:455-469. Kongyuy EJ, Wiysonge CS, Mbu RE, Nana P and Konan L. 2006. Wealth and sexual behaviour among men in Cameroon. BMC International Health and Human Rights, 6: 11 http://www.biomedcentral.com/1472-698X6/11. [Accessed 20th October, 2007]. Kotchick BA, Shaffer A, Forehand R and Miller KS (2001). Adolescent sexual risk behaviour: A multi-system perspective. Clinical Psychology Review, 21(4): 493-519. Kuate-Defo B. 1997. Effects of socioeconomic disadvantage and women?s status on women?s health in Cameroon. Social Science and Medicine, 44, 1023-1042. Langer LM, Warheit GJ and McDonald LP. 2001. Correlates and predictors of risky sexual practices among a multi-racial/ethnic sample of university students. Social Behaviour and Personality, 29(2): 133-144. Latino M, Bello L, Lanza A, Leotta E, Tersiev P, Intinis G, Spagnolo E, Smirne C and Grio R. 2002. Chlamydia trachomatis infection among sexually active women in Italy. Sexually Transmitted Infections 78. Last JM. 1988. A dictionary of epidemiology, (2nd ed.) Oxford: Oxford University Press. Page | 56 Lee K and Zwi AB. 1996. A global political economy approach to AIDS: ideology, interests and implications. New Political Economy, 13: 355-373. Lestrade-Jefferis J. 2000. The labor market. In EO Udjo (Ed.). The people of South Africa population census 1996. Pretoria: Statistics South Africa. Makgoba MW. 2000. HIV/AIDS: the peril of pseudoscience. Science, 288: 1171 Mann J and Tarantola D. 1996. AIDS in the world II: global dimensions, social roots, and responses. New York: Oxford University. McDonald TK, Zanna MP and Gong GT. 1996. Why common sense goes out the window: Effects of alcohol on intentions to use condoms. Personality and Social Psychology Bulletin, 22: 763-775. McIntrye D, Muirhead D and Gilson L. 2002. Geographic patterns of deprivation in South Africa: informing health equity analyses and public resource allocation strategies. Health Policy and Planning, 17(1): 30-39. Mertz K, Finelli L, Levine W, Mognoni R, Berman S, Fishbein M, Garnett G and Louis M. 2000. Gonorrhoea in male adolescents and young adults in Newark, Neww Jersey. Sexually Transmitted Diseases, 27: 201-207. Mishra V, Bignami-Van Assche S, Greener R, Vaessen M, Hony R, Ghys P, Ties Boerma J, Van Assche A, Khan S and Rutstein S. 2007. ?HIV Infection does not disporportionately affect the poorer in Sub-Saharan Africa?, AIDS, 21 (supplement 7), s17-s28. Mill JE and Anarfi JK. 2002. HIV risk environment for Ghanaian women: Challenges to prevention. Social Science and Medicine, 54: 325-337. Page | 57 Msisha WM, Kapiga HS, Earls F and Subramanian SV. 2008. Socioeconomic status and HIV seroprevalence in Tanzania: a counterintuitive relationship. International Journal of Epidemiology 37: 1297-1303 Mwabu G. 2001. Health Status in Africa: A Regional Profile. South African Journal of Economics, 69: 319-335. Oppong CA. 1995. high price to pay: For education subsistence or a place in the job. Health Transition Review, 5(suppl): 35-56. Parker RG. 1996. Empowerment, community mobilization and social change in the face of HIV/AIDS. AIDS, 10 (suppl III): s27-s31. Parker RG, Easton D, and Klein CH. 2000. Structural barriers and facilitators in HIV prevention: A review of international research. AIDS, 14 (suppl1): S22-S32. Poku NK. 2001. Africa?s AIDS crisis in context: ?how the poor are dying?. Third World Quarter, 22: 191-204. Pronyk PM, Hargreaves JR, Kim JC, Morison LA, Phetla G, Watts C et al., 2006. Effect of a structural intervention for the prevention of intimate-partner violence and HIV in rural South Africa: a cluster randomized trial. Lancet, 368: 1973-1983. RADAR 2002a. Social Interventions for HIV/AIDS. Intervention with microfinance for AIDS and gender equity. Evaluation monograph number 1 (No.1). Acornhoek: RADAR. RADAR 2002b. Social Interventions for HIV/AIDS. Intervention with microfinance for AIDS and gender equity. Evaluation monograph number 2 (No.1). Acornhoek: RADAR Ramkisson A, Kleinschmidt I, Beksinka M, Smit J, Hlazo J and Mabude Z. 2004. National Baseline Assessment of Sexually Transmitted Infections and HIV services in South African Public Health Sector health Facilities 2002/2003.Reproductive Health Research unit, University of the Witwatersrand, South Africa. Summary Report 2004 Page | 58 Rao Gupta G. 2002. How men?s power over women fuels the HIV epidemic. British Medical Journal, 324: 183-184 Rhodes T. 1997. Risk theory in epidemic times: Sex, drugs and the social organization of risk bahaviour. Sociology of Health and Illness, 19: 208-227. Richens J, Imrie J and Weiss H. 2003. Human immunodeficiency virus risk: is it possible to dissuade people from having unsafe sex? Journal of Royal Statistical Society Series A, 166(2): 207-215. Shelton J, Cassell M and Adetunji J. 2005. ?Is poverty or wealth at the root of HIV?? The Lancet, 366: 1057-1058. Slaymaker E. 2004. A critique of international indicators of sexual risk behaviour. Sexually Transmitted Infections, 80 (suppl II): 13?21. Sonkin B and Hinde A. 2007.Defining risky sexual behaviour in the UK: A latent class approach. Southampton, UK, Southampton Statistical Sciences Research Institute (S3RI Applications and Policy Working Papers, A07/01). http://eprints.soton.ac.uk/43969. [Accessed 13th October 2007] Sumartojo E. 2000. Structural factors in HIV prevention: concepts, examples, and implications for research. AIDS, 14 (suppl1): S3-S10. Sumartojo E, Doll L, Holtgrave D, Gayle H and Merson M. 2000. Enriching the mix: incorporating structural factors into HIV prevention. AIDS, 14 (suppl 1): S1-S2. Tawil OA, Verster K and O'Reilly R. 1995. Enabling approaches for HIV/AIDS prevention: can we modify the environment and minimize the risk? AIDS, 9: 1299-1306. Udjo EO and Lestrade-Jefferis J. 2000. Demographic profile of South Africa. In: EO Udjo (Ed.) the people of South Africa population census 1996, Pretoria: Statistics South Africa. Page | 59 Ulin PR. 1992. African women and AIDS: Negotiating behavioural change. Social Science and Medicine, 34(1): 63-73. UNAIDS 1999. Facts and Figures - World AIDS Campaign. 1999, UNAIDS: Geneva. UNAIDS 2006. Joint United Nations Programme on HIV/AIDS. South Africa: epidemiological fact sheets. http://unaids.org [Acessed 22nd July 2007] UNAIDS/WHO. 2006. 'UNAIDS 2006 Report on the global AIDS epidemic', Annex 2: HIV/AIDS estimates and data, 2005. United Nations. 1998. World urbanization prospects: The 1996 revision. New York: United Nations, Department of Economic and Social Affairs, Population Division United Nations Development Programme. Human development report 2002. New York: Oxford University Press, pp 222-242. Van Donk M. 2002. HIV/AIDS and Urban Poverty in South Africa: Prepared by Mirjam van Donk for the South African Cities Network (August2002). Available: http://www.isandla.org.za/papers/AIDS%20and%20urban%20poverty%20in%20South%20 Africa.pdf Waldo CR and Coates TJ. 2000. Multiple levels of analysis and intervention in HIV prevention science: exemplars and directions for new research. AIDS, 14 (suppl 2): S18- S26. Wellings K, Collumbien M, Slaymaker E, Singh S, Hodges Z, Patel D and Bajos N. 2006. Sexual bahaviour in context: a global perspective. Lancet, 368:1706-1728 Wojicki JM. 2005. Socioeconomic status as a risk factor for HIV infection in East, Central and Southern Africa: a systematic review. Journal of Biosocial Science, 37(1): 1-36 Page | 60 World Bank. Confronting AIDS: Public priorities in a global epidemic. 1997, New York: Oxford University Press. World Bank. Confronting AIDS. 1998. Washington, DC: World Bank World Bank. 2000. Country Reports on Health, Nutrition, Population and Poverty. http://www.worldbank.org/poverty/health/data/intro.htm [Accessed 29th November 2007] Zenilman J, Weisman C, Rompalo A, Ellish N, Upchurch D, Hook E and Celentano D. 1995. Condom use to prevent incident STDs: the validity of self-reported condom use. Sexually Transmitted Diseases, 22: 15-21. Zulu EM, Dodoo FN and Ezeh AC. 2002. Sexual risk taking in the slums of Nairobi, Kenya, 1993-1998. Population studies, 56(3): 311-323. Zwi, A. 1993. Reassessing priorities: Identifying the determinants of HIV transmission. Social Science and Medicine, 36: iii-vii. Page | 61 Appendix I -------------------------------------------------------------------------------------------------------------- Page | 62 Appendix II --------------------------------------------------------------------------------------------------------------