Education and HIV/AIDS A report prepared for the UNESCO Global Monitoring Report 2005 by Matthew Jukes and Kamal Desai. Section 1 – The relationship between Education and HIV prevalence As the HIV epidemic shows few signs of slowing in Africa, and threatens to gather momentum in much of Asia and Eastern Europe, an effective prevention response is desperately sought. An educated population may be an important foundation for this response. This paper examines the relationship between schooling, sexual behaviour and HIV prevalence and investigates the role expanded primary school access and increased literacy can play in tackling the HIV epidemic. The following sections consider the relationship between literacy and HIV at the national level and then examines data concerning sexual behaviour and HIV infections and its relationship with education. National level associations between Education and HIV. At the national level in Africa there is a positive relationship between literacy rates and HIV infection rates (Figure 1)1: More literate countries have higher rates of HIV infection. More literate African countries tend to be the most developed on the continent and they share a number of features that make them vulnerable to high rates of HIV infection. First the most developed countries often have the largest income disparities between men and women, a factor associated with HIV infection rates (World Bank, 1997). Similarly, employment in the formal sector is associated with HIV infection (Barongo, Borgdorff, Mosha, Nicoll, & al., 1992; Serwadda et al., 1992). Second, increased migration and improved transport infrastructure can facilitate the spread of HIV (Caldwell & Caldwell, 1993). Similarly, urban residence is associated with higher levels of HIV infection (Barongo et al., 1992; Boerma, Urassa, Senkoro, Klokke, & Ng'weshemi, 1999; Fylkesnes et al., 1997; 1 Permission not yet sought for reproduction of graphs. 1 Serwadda et al., 1992). Finally, as discussed below, higher levels of education per se are associated with higher infection rates. Although various demographic and socioeconomic conditions put the educated at greater risk of HIV infection it has been hypothesized (Simon Gregson, Waddell, & Chandiwana, 2001; Over & Piot, 1993) that they will be more likely to change their behaviour in response to information about the epidemic. One way in which this trend may manifest itself at the national level is in the weakening of the relationship between literacy and HIV infection rates as the epidemic matures. In the early stages of the epidemic the most educated will be at great risk and a strong relationship will be expected between literacy and HIV infection rates. In later stages of the epidemic, when information is available about prevention methods and education campaigns have begun, the educated are more likely, it is argued, to change their behaviour and this mitigates against socioeconomic factors putting them at greater risk of infection. We would therefore expect the relationship between national HIV infection and literacy to be weaker in later stages of the epidemic. This hypothesis has been tested (Simon Gregson et al., 2001) by comparing the relationship between adult HIV prevalence in 1999 and adult literacy in 1998 in three different regions of sub-Saharan African. In two regions the epidemic is more mature and the relationship between HIV prevalence and literacy is relatively weak. In West and Central Africa, a 1% increase in national literacy rates is associated with a 0.077% rise in national HIV prevalence; in East Africa a 1% rise in literacy rates is associated with a 0.033% rise in HIV prevalence. For example, the literacy rates in Kenya (~80%) were much greater than in Ethiopia (~40%) in 1998 but HIV prevalence estimates at the time were similar for the two countries (~12% and ~10% respectively). This stands in contrast with the relatively recent epidemic in Southern Africa where a 1% rise in literacy rates is associated with a 0.35% rise in HIV prevalence. For example, Zimbabwe has a higher literacy rate (~85%) than Malawi (~60%) and similarly Zimbabwe has one of the highest prevalences in region (~23%) and Malawi one of the lowest (~17%). Thus, the evidence suggests that there is a strong relationship between education and HIV prevalence in the early stages of an epidemic, but a weaker relationship or no relationship in more mature epidemics. This analysis is consistent with the hypothesis that education prepares individuals better to mount a response to the HIV/AIDS 2 epidemic but for a more fine-grained assessment of the hypothesis we look at the relationship between HIV infection and education at the level of the individual. Education and HIV at the level of the individual Education and sexual behaviour As with the picture at the national level, an individual level analysis suggests that educated individuals should be at greater risk of infection at the early stages of an HIV epidemic but should be better equipped to change their sexual behaviour when faced with the facts of HIV transmission. There are a number of aspects of the sexual behaviour of more education individuals that initially puts them at great risk of infection. More educated individuals change partners more rapidly, in part because they are more mobile and because they have greater control over their own sexual behaviour (A. Blanc, 2000). The higher socioeconomic status of more educated men gives them a greater disposable income, increase leisure time, increased ability to travel and to use commercial sex partners (Berkley et al., 1989; Dallabetta et al., 1993; Quigley et al., 1997). In addition, more educated women start having sex later but delayed marriage to an even great extent. This leads to them being single and sexually active for a longer period of time and thus to them having a greater number of sexual partners (A. K. Blanc & Way, 1998). Choice of contraceptives may also be influenced by education. Although educated people may be more likely to use contraception overall, they are also more likely to choose methods, such as the contraceptive pill, which do not protect against sexually transmitted diseases such as HIV. Taken together, and in the absence of any response to an epidemic, these factors increase the vulnerability of more educated individuals to HIV infection in the early stages of an epidemic. However, evidence and theory suggest that educated people should be more likely to adopt safe sexual practices in response to health promotion (Fylkesnes et al., 1997) or from other sources of information about HIV/AIDS. Social cognitive models point to several key determinants of sexual behaviour. In all theories, knowledge and understanding of a behaviour and its consequences is a necessary but not sufficient condition for performing the behaviour and underpins the perception of (Rosentock, 1990) and attitudes (I Ajzen, 1985) towards that behaviour. In the context of HIV, 3 understanding transmission routes and methods to block them are essential for the adoption of safe sexual behaviour. More educated people are more likely to be exposed to such information as part of formal schooling and also through the media (S. Gregson, Zhuwau, Anderson, & Chandiwana, 1998). For example, a study in 32 countries found that literate women were three times more likely than illiterate women to know that a healthy looking person can have HIV, and four times more likely to know the main way s to avoid AIDS (Vandemoortele & Delamonica, 2000). Greater levels of education may also provide a framework of biological knowledge and understanding of causality into which HIV prevention messages can be assimilated. For example, children with a deeper understanding of the biological mechanisms of viruses are more resistant to myths about HIV transmission (Keselman, Kaufman, & Patel, 2004). A second key determinant of behaviour is the perceived control one has over the behaviour. This includes self-efficacy, one’s belief in one’s capabilities to perform a specific action required to attain a desired outcome (Bandura, 1977), the perceived personal power one has over the behaviour (I Ajzen, 1985) and the actual personal power one has over a behaviour (I. Ajzen, 2002). Evidence suggests that education is associated with increased self-efficacy in general (Bandura, 1977) and in the context of the HIV epidemic in sub-Saharan Africa in particular (Lindan et al., 1991). In addition, more educated people are more likely to believe they have control over their own behaviour, rather than another individual or fate, and they are more likely to have actual control over their own behaviour. For example, educated women are more able to negotiate safe sex with a partner (GCE, 2004). This analysis suggests that education should lead to a greater adoption of safe sexual behaviour in response to the HIV epidemic. Data from Demographic and Health Surveys (DHS) in 11 countries (GCE, 2004) showed that women with primary school education were more likely than those with no education to report using a condom at last sex. In nine of these countries, secondary education was associated with a further increase likelihood of using a condom at last sex. Another study in Zimbabwe (Simon Gregson et al., 2001) found that women with secondary education were less likely to report having had unprotected casual sex. A study in the four African cities of Cotonou in Benin, Ndola in Zambia, Yaoundé in Cameroon, and Kisumu in Kenya found that education led to less risky sexual behaviour. Condom use was more common amongst more educated individuals in all four cities (Lagarde et al., 2001). 4 Exchange of money for sex was less likely amongst educated women in all four cities and amongst more educated men in Yaoundé. Non-marital sex without a condom was less prevalent among more educated women in all four cities and among more educated men in Cotonou and Kisumu. In Yaoundé, more educated men and women were less likely to have sex with a casual partner on the day of meeting, and in Ndola, for both men and women, not knowing a partner’s age was much more common among those with little schooling (Glynn et al., 2004). Other behaviours that reduce HIV infection are also more common among the educated. For example, more educated people are more likely to seek treatment for other sexually transmitted diseases which would otherwise increase their chances of becoming infected with HIV (A. Blanc, 2000). Overall, the evidence suggests that more educated people are at a greater risk of HIV infection in the early stages of an epidemic but tend to adopt less risky sexual behaviours in response to the epidemic. How is this reflected in empirical data on HIV prevalence and its relationship with education? Education and HIV prevalence The majority of studies investigating this issue have found a positive relationship between education and HIV infection. That is, HIV prevalence is higher among educated individuals. This was found at the population level in data from Rakai, Uganda in 1990 and 1992 (Kirunga & Ntozi, 1997; Smith et al., 1999); from Mwanza, Tanzania in 1991-1996 (Grosskurth et al., 1995; Quigley et al., 1997; Senkoro et al., 2000) and amongst women attending ante-natal clinics in Fort Portal, Uganda in 1991-4 (Kilian et al., 1999) and in Zambia in 1994 and 1998 (Fylkesnes et al., 2001; Fylkesnes, Ndhlovu, Kasumba, Musonda, & Sichone, 1998). In all cases, the relationship was adjusted for age, sex and setting (urban or rural). Five population based studies found the opposite trend. Education had a protective effect against HIV among young women of Manicaland, Zimbabwe from 1998-2000 (Simon Gregson et al., 2001), among men and women in Masaka district in Uganda in 2000 but not in 1990 (De Walque, 2002), among women in Youndé, Cameroon and men in Cotonou, Benin (Glynn et al., 2004) and against HIV-2 infection in the Gambia (Wilkins et al., 1991). Two Ethiopian studies focusing on sub-populations found opposing trends. Education was related to lower HIV prevalence amongst sugar estate workers 5 (Fontanet et al., 2000), but higher HIV prevalence in male army recruits in rural areas (Abebe et al., 2003). Several large studies of HIV prevalence among 21-year old Thai army recruits found that those with more years of education had lower levels of HIV infection (Carr et al., 1994; Mason et al., 1998; Mason et al., 1995; Nelson et al., 1993; Sirisopana et al., 1996) although no relationship between HIV and education was found in studies from Northern Thailand where the prevalence is highest (Celentano et al., 1996; Dobbins et al., 1999; Nopkesorn et al., 1993). Several other studies found that HIV prevalence and education were not statistically related, including studies in Zimbabwe (S. Gregson et al., 2001), in Ndola, Zambia and Kisumu, Kenya (Glynn et al., 2004) and in 7 of the 27 studies reported in a review (Hargreaves & Glynn, 2002). There is no consistent pattern in these results. This inconsistency may represent the combination of two opposing trends in the data: the initial increased vulnerability of educated individuals to HIV infection followed by their more rapid behavioural change once informed about the epidemic. The studies reported do allow us several opportunities to try and pick apart these two trends from their data by analyzing the evolution of these trends with epidemic maturity. Changing relationship between HIV and education with epidemic maturity It is notable that the majority of studies conducted in Africa until the mid 1990s found higher HIV prevalence among the more educated whereas later studies were more likely to find the opposite. This is consistent with the view that educated Africans were more at risk of HIV infection initially and were only equipped with the information required to mount a response to the epidemic years later. In Thailand the picture is different. The HIV epidemic was monitored by existing mechanisms and was initially confined to high-risk groups. This allowed the spread of information about the epidemic before the spread of the disease to the general population. This is perhaps why the most educated individuals are protected from HIV from the early 1990s. More direct evidence addresses the hypothesis that the relationship between HIV and education evolves as the epidemic matures. The clearest evidence comes from a serial 6 cross-sectional survey in the rural areas of Masaka district, Uganda (De Walque, 2002). The national prevalence of HIV in the adult population declined from its peak of 14% in the early 1990s to around 5%, largely due to a strong prevention campaign (Stoneburner & Low-Beer, 2004). As illustrated in Figure 2, the rate of decline in prevalence was greater for those with secondary education and those with primary education showed faster decline in prevalence than those with no education. The chances of contracting HIV infection during this period was reduced by 6.7% for each year spent in education (De Walque, 2002) and those with no education were 2.2 times more likely to become infected than those who had completed primary education. These analyses were restricted to those born after 1971 most of whom would have become sexually active after the beginning of the information campaigns (in 1986). For older Ugandans, the more educated were actually more likely than less educated people to become infected during the period of the study, suggesting that population behaviour change is driven by young people. On the basis of these figures, the Global Campaign for Education argue that less educated young adults will experience a disproportionately large number of new infections. They estimate that the 36% of young adults in low-income countries without a complete primary education will experience 55% of new infections – 1.3 million of the total of 5 million new infections in the whole population every year. The estimates imply that achieving Universal Primary Education could reduce the number of new infections in this group by 700,000 a year (GCE, 2004). The data from Uganda demonstrate the importance of schooling in an individual’s response to a prevention campaign. It also demonstrates the evolving nature of the relationship between HIV and education. In 1990 there was no relationship between HIV prevalence and education. In 2000, having completed primary education was associated with a 5.1% reduction in the risk of HIV infection and secondary education was associated with an 8.8% reduction in risk. This relationship between HIV and education was found for women but not men. Similarly, in Rakai, Uganda, HIV infection was associated with increased levels of education in 1990 and 1992 but not by 1994 (Kelly et al., 1999; Smith et al., 1999). These finding supports the thesis that more educated individuals are better able to mount a response to the HIV epidemic. However, the reversal of the epidemic in Uganda is not typical of African countries. What evidence is there that education 7 protects against HIV in the absence of a successful national prevention campaign? In fact, data from a number of countries show a similar evolution in the relationship between HIV and education. In the following studies, this evolution is evident in comparisons between younger and older age groups at one time point, and in comparisons between similar populations over time. In a population-based study in Zimbabwe, men and women aged 17-19 were at a lower risk of HIV infection if they had secondary education. The benefit of education was less for those aged 20-24 and there was little or no protective benefit for those aged 25 and over (Simon Gregson et al., 2001). In Fort Portal, HIV prevalence amongst women aged 15-49 attending an antenatal clinic was highest for those with secondary education in 1991-1994 but by 1995-97 older illiterate women had the highest prevalence (Kilian et al., 1999). Prevalence reduced to the greatest extent amongst women with secondary education and among young women. Similarly, there was a positive association between education and HIV infection amongst women attending an antenatal clinic in 1994 but not by 1998 (Fylkesnes et al., 1997; Fylkesnes et al., 2001). Again, largest reductions were seen amongst younger more educated women. Similar patterns were seen in northern Malawi (Crampin et al., 2003) but there was no evidence of a changing association between HIV and education in Blantyre, Malawi (Taha et al., 1998) or in Kagera, Tanzania (Kwesigabo et al., 1998). The data presented so far have been concerned with the static relationship between HIV and education. An important policy question concerns the impact of increasing levels of education on the epidemic. There is no experimental evidence addressing this issue, but one study has estimated this relationship by analysing longitudinal data from 20 regions in Tanzania over 8 years (Brent, 2005). This study estimated that an increase of 1% in female primary school enrolment would be responsible for a 0.15% reduction in HIV prevalence in this group, corresponding to 1,408 infections in the period 1994-2001. A further analysis of these data suggest that the investment in increased school enrolment is justified by the averted cases of HIV and the earning potential of these individuals, with a cost-benefit ratio of between 1.3 and 2.9. Urban-rural differences in the HIV-education relationship 8 Three studies have collected comparable data from urban and rural areas on the relationship between HIV infection and education. The finding from all three is that education is a greater risk factor for infection in rural areas compared to urban areas. In Kagera, Tanzania (Kwesigabo et al., 1998), secondary educated individuals were 3.3 times more likely than those with no education to be infected with HIV in rural areas. Conversely, in urban areas secondary education was associated with a reduction in HIV risk. In Zambia in 1994 (Fylkesnes et al., 1997), having more than 10 years of education was associated with an increased risk of being HIV infection, compared to those with less than 4 years of education, in both rural and urban areas. However, the increased risk was greater in rural areas (odds ratio of 4.2) than in urban areas (odd ratio of 2.5). By 1999 figures from Zambia (Fylkesnes et al., 2001) showed that education was now associated with a decrease in HIV prevalence in urban areas but there was no relationship between education and infection in rural areas. A similar pattern of results was found in Rakai, Uganda, in 1990 (Smith et al., 1999). Increased education was associated with an increased risk of HIV in rural villages but there was no relationship in roadside trading centres or in trading villages. Similarly, in Mwanza, Tanzania, in 1991-2, HIV infection was positively associated with education levels in rural villages and in roadside trading villages but not in urban centres. All studies show a stronger relationship between HIV infection and education in rural areas compared to urban areas. This should be considered with the finding that urban residence is associated with a higher risk of HIV infection overall (Barongo et al., 1992; Boerma et al., 1999; Fylkesnes et al., 1997; Serwadda et al., 1992). Taken together these findings suggest that, at least in the early stages of the epidemic, educated individuals were similarly at risk of infection in urban and rural areas. The less educated were at a lower risk of HIV infection in urban areas and to a greater extent in rural areas (Fylkesnes et al., 1997). Sex differences in the HIV-education relationship Data on sex differences are limited by the common use of infection rates in women visiting ante-natal clinics as a proxy for population infection rates. Nevertheless, four studies have looked at gender differences in the HIV-education relationship (Barongo et al., 1992; Fylkesnes et al., 2001; Grosskurth et al., 1995; Smith et al., 1999). In all four studies the increased risk of HIV infection associated with education was very 9 similar for men and women, suggesting there are no sex differences in the relationship between HIV and education. Sex differences in the HIV-education relationship are complicated by consideration of the partner’s education. Evidence shows that a women’s risk of HIV infection is increased with higher levels of education in their partner (Allen et al., 1991; Dallabetta et al., 1993). Considering evidence from education interventions, where sex differences are found they consistently suggest that HIV-related knowledge is more strongly associated with sexual behaviour in women, compared to men (Jukes, in preparation). Therefore, although there is no evidence of sex differences in the HIV-education association at present, we might expect more educated women to change their behaviour more rapidly than men in response to the HIV epidemic. Education and social capital Studies in Uganda show how more educated people are more likely to change their behaviour in response to an HIV prevention education campaign. One study in Manicaland, Zimbabwe shows how more educated women are also able to benefit more from other protective measures. This study looked at membership of social groups related to, among others, churches or political parties. Women who were members of a well-functioning social group were 1.3 times more likely to avoid HIV infection than those who were not in such groups or who were in groups with which they were dissatisfied. Education played a key role in the protective effect of the social groups. Women with secondary education were more likely to belong to such groups and among women with secondary education, those who were members of well functioning groups were 1.5 times less likely to be infected with HIV, whereas women with no education received no such benefits from group membership (S. Gregson, Terceira, Mushati, Nyamukapa, & Campbell, 2004). Explaining the relationship between HIV and education In trying to understand the relationship between HIV and education two questions are of interest. What is it about educated people that enables them to change their behaviour in response to the epidemic? And, what aspects of behaviour change are 10 responsible for the reduction in HIV prevalence in this group? Few studies have addressed these questions directly. In response to the first question, analyses from Uganda (De Walque, 2002) and elsewhere (Vandemoortele & Delamonica, 2000) suggest that parental and individual income are not explanatory factors. This supports the view that the increased knowledge, understanding or self-efficacy that comes with education is responsible for behaviour change. One study in Zimbabwe addresses the question of which behaviours change amongst educated individuals. The finding was that the secondary educated women acquired HIV infection at a slower rate and have both a later sexual debut and less unprotected casual sex (Simon Gregson et al., 2001). Subsequent analyses suggest that the delayed age of sexual debut is primarily responsible for the relationship between education and HIV infection (James Lewis, pers. comm.). This relationship may result from the impact of sexual behaviour on girls’ education (through pregnant young women dropping out of school) as well as the impact of education on sexual behaviour. Conclusion Overall there is convincing evidence that education better equips individuals to respond to the HIV epidemic. Although education is associated with higher HIV prevalence in the early stages of an epidemic, in the later stages more educated individuals have less risky sexual behaviour and are less likely to be HIV positive. This is true in many settings but is particularly evident in Uganda, where a national prevention campaign has successfully reduced HIV prevalence. There have been few estimates of the likely impact of increasing primary school completion on the HIV epidemic. One analysis based on the Uganda data suggests that universal primary education could save 700,000 young adults from HIV infection. Another analysis from Tanzania suggests that investments in expanded school enrolment for girls is cost effective purely in terms of the effect this increased enrolment will have on the HIV epidemic. Taken together, there is a strong case for making expanded primary education and improved literacy a central part of the global response to the HIV/AIDS epidemic. 11 Section 3. Education and Antiretroviral Therapy Treatment adherence is a critical issue in the promotion of antiretroviral therapy in developing countries. Poor adherence is associated with reduced viral suppression (Bangsberg et al., 2000; McNabb et al., 2001; Paterson et al., 2000), progression to AIDS (Bangsberg et al., 2001), the development of multidrug resistance and death (Hogg et al., 2002; Stephenson, 1999). Adherence is perceived as a significant barrier to the delivery of ART in sub-Saharan Africa in particular (Harries, Nyangulu, Hargreaves, Kaluwa, & Salaniponi, 2001). Poor education is often implied as a reason why adherence is unlikely to be high in sub-Saharan Africa. A spokesman for USAID is quoted as saying ‘Ask Africans to take their drugs at a certain time of day and they do not know what you are talking about’ (cited in Orrell, Bangsberg, Badri, & Wood, 2003). Here we review the evidence that education and literacy are related to treatment adherence. Treatment adherence in Africa. Initial studies of treatment adherence in Africa have allayed some fears of poor adherence on the continent. A study in South Africa looked at adherence in a poor HIV-positive cohort receiving ART through a public sector hospital (Orrell et al., 2003). Adherence of the cohort was 93.5%. A composite measure of socioeconomic status, including education and income, was not related to adherence. Those speaking English at home were more than twice as likely to adhere to the treatment regimen. Older patients and those required to take fewer doses were also more likely to adhere to their regimen. One study in Uganda looked at adherence in 304 HIV-infected individuals on a selffinanced regimen of antiretroviral therapy. The study population was relatively well educated and wealthy. Monthly income was greater than $250 for 87.8% of the sample and 63.2% had post-secondary education. Those with a monthly income of less than $50 were 2.8 times more likely to miss 5% or more of doses. Seventy five percent gave lack of money to buy drugs as a reason for lapses in adherence. Those who were single (never married) were 2.9 times as likely to miss 5% or more of doses. There was no significant relationship between adherence and post-secondary 12 education. An analysis by years of education was not reported (Byakika-Tusiime et al., 2005). A relationship between education and adherence was found in a study of 109 patients attending private clinics in Botswana (Weiser et al., 2003). However, the relationship was in the unexpected direction. Those who had not completed secondary education were 3.9 times as likely to adhere to their treatment as those with higher education. In this study, those who perceived cost or side effects as a barrier to treatment were less likely to adhere to treatment; those who had disclosed their HIV status to others were more likely to adhere to their treatment. Treatment adherence in resource rich settings. Studies in the West show that more educated individuals are more likely to adhere to treatment with combination ART (Chesney, 2000; Chesney, Ickovics, Hecht, Sikipa, & Rabkin, 1999). For example, in one study in America, high school completion was associated with a 5.5% increase in adherence to a regimen of combined anti-retroviral therapy (Golin et al., 2002). However, education does not emerge as a major determinant of treatment adherence in the West. Alcohol and drug abuse are the key patient characteristics affecting adherence. Other factors include depression, anxiety, extreme pain, low self-efficacy and being young, male, or from an ethnic minority (Chesney, 2000). In a review of adherence to highly active antiretroviral therapy (HAART) similar factors emerged as key determinants of adherence. In 14 studies that assessed the relationship between education and adherence, 10 found no relationship and 4 found higher levels of adherence amongst the more educated. This relationship disappeared when controlling for other demographic factors in all 4 studies (Ammassari et al., 2002). Other studies show a relationship between adherence and knowledge, beliefs and other cognitive factors of the patient that may be related to their level of education. A patient's beliefs about their illness and the effectiveness of medication are predictive of adherence. A patient's level of knowledge about HIV disease (Eldred, Wu, Chaisson, & Moore, 1995; Eraker, Kirscht, & Becker, 1984; Morse, Simon, & 13 Walker, 1995) a belief that HAART is effective (Klosinski & Brooks, 1998; Wenger, Gifford, Liu, Chesney, & Golin, 1999) and prolongs life (Stone et al., 1998), and a recognition that poor adherence may equal viral resistance and treatment failure (Wenger et al., 1999) all impact favourably upon a patients ability to adhere. Conversely, a lack of interest in becoming knowledgeable about HIV (Kammann, Williams, Chesney, & Currier, 1999) and a belief that HAART may in fact cause harm, impede adherence (Brigido et al., 1998; Horne, Pearson, Leake, Fisher, & Weinman, 1999; Johnston, Ahmad, Smith, & Rose, 1998). Conclusion There is insufficient evidence to make strong conclusions about the relationship between education and ART treatment adherence in low-income countries. Of the three studies conducted so far, only one was with relatively poor patients in a public hospital. This found that those speaking English at home were more likely to adhere to their treatment. It is possible this relationship was mediated through improved education or literacy of those receiving treatment. The other two studies were conducted with more wealthy individuals paying for their own treatment. These studies found either no associated between adherence and education or found that more education was associated with poorer adherence. However, these studies compared those with or without complete secondary education and their findings may not be relevant to questions of basic education and literacy. Findings from the West suggest that there is a link between education and literacy. It seems that improving education levels in developing countries is likely only to improve ART treatment adherence and, perhaps more importantly, greater education levels may serve to allay widespread fears amongst policy makers about the problems of adherence in Africa. 14 15 Adult HIV Prevalence %, 1999 (UNAIDS, 2000) 40 y = 0.2483x - 4.7922 2 R = 0.3509 30 20 10 0 0 Figure 1: 10 20 70 60 50 40 30 Adult Literacy %, 1998 (UNESCO, 2000) 80 90 100 HIV prevalence in adults aged 15-49 years by level of adult literacy for 40 countries in sub-Saharan Africa. (From Simon Gregson et al., 2001) 16 Figure 2. HIV prevalence by education category, Rural Uganda, 1990-2001. Individuals aged 18-29. (De Walque 2002). 17