Understanding the Southern African ‘Anomaly’: Poverty,
Endemic Disease and HIV
Larry Sawers and Eileen Stillwaggon
ABSTRACT
The epicentre of the global HIV epidemic is southern Africa. Previous explanations point to migration patterns and highly skewed income distribution,
both thought to promote risky sexual behaviour. This study emphasizes the
importance of common infectious and parasitic diseases that increase the
likelihood of HIV transmission by increasing contagiousness and vulnerability to infection. Using multiple regression analysis on country-level data, the
authors find that socio-economic variables explain statistically only one-tenth
of the difference in HIV prevalence between southern Africa and other lowand middle-income countries. Measures of five cofactor infections together
with the socio-economic variables, however, explain statistically about twothirds of the southern Africa difference in HIV prevalence. They conclude
that the relative affluence of countries in southern Africa and historical migration patterns have tended to mask the vulnerability of the majority of
their populations who are poor and who have very high prevalence of infectious and parasitic diseases. Those diseases replicate a cycle of poverty
that produces biological vulnerability through coinfections. An important
implication of this research is that integrating treatment of endemic diseases
with other HIV-prevention policies may be necessary to slow the spread
of HIV.
INTRODUCTION
Average adult HIV prevalence in 2007 was 18 per cent in the nine countries of
southern Africa (South Africa, Swaziland, Lesotho, Zimbabwe, Botswana,
Mozambique, Namibia, Zambia and Malawi). Elsewhere in sub-Saharan
Africa, HIV prevalence averaged 2.7 per cent in 2007; in other low- and
middle-income countries, HIV prevalence averaged 0.5 per cent. The epicentre of the global HIV epidemic is southern Africa. Understanding the
nature of HIV requires that we understand the region where the epidemic is
most intense.
We would like to express our gratitude to the anonymous referees of Development and Change.
Their challenging comments and helpful suggestions have greatly improved the article.
C International Institute of Social Studies
Development and Change 41(2): 195–224 (2010). 2010. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and
350 Main St., Malden, MA 02148, USA
196
Larry Sawers and Eileen Stillwaggon
HIV is an opportunist virus that may have found in southern Africa
a laboratory with the most propitious combination of factors promoting
transmission. Various explanations have been proposed for the region’s vulnerability to HIV, but the present study is the first to attempt a comprehensive
account of the southern Africa anomaly and to test it statistically. Explaining
HIV in southern Africa may give us a key to understanding how social, biological and ecological factors interact to produce catastrophic results. That is
particularly instructive for understanding a pathogen that has had relatively
little impact in most other parts of the world. We examine the relevance of
numerous factors considered in the literature and then construct statistical
models of HIV prevalence. We find that, controlling for a combination of
socio-economic and biological factors, the levels of HIV infection in southern Africa are not anomalous. They conform to what is known about HIV
and reveal important vulnerabilities generally overlooked in discussions of
southern Africa.
BACKGROUND: HISTORY, ECONOMICS AND HEALTH
IN SOUTHERN AFRICA
Some of southern Africa’s historical features are unique, but many of its
socio-economic, ecological and health characteristics are similar to those in
the rest of Africa or in other low- and middle-income countries. The severity
of the HIV epidemic in southern Africa is most probably the result of a
combination of attributes it shares with other poor countries, especially in
Africa, and features unique to southern Africa. The literature on HIV and
southern Africa suggests a number of factors that might have aggravated the
spread of HIV in the region.
Labour Migration and Apartheid
Perhaps the most obvious characteristic of southern Africa that sets it apart
from the rest of the world is its history of apartheid and the earlier system of
internal and international labour migration that this reinforced. Even before
the National Party decreed strict residential segregation in South Africa in
1948, cross-border and internal labour migration supplied the labour for
the country’s mines and factories. An extensive literature dating back to the
1940s addressed the health effects of migration in southern Africa, including
the spread of tuberculosis and sexually transmitted infections (STIs) from
migrants to their home communities (see Horwitz, 2001 for a summary of
the early literature).
There is also a substantial literature that points to the role of migration
in the spread of HIV, and many of those works focus on southern Africa in
Poverty, Endemic Disease and HIV
197
particular.1 High prevalence among migrants seems to support the assumption that long separation of spouses in circular migrant streams, single-sex
hostels and higher levels of commercial sex work around mines and factories promote the spread of HIV (Lurie, 2004; see also Brummer, 2002;
Clark et al., 2007; and Kanyenze, 2004, although without data). Migration
has also been associated with HIV transmission in other parts of the world.
In Ecuador, for example, over 80 per cent of the people infected with HIV
in the first decade of the epidemic had worked in the United States or were
partners of migrants to the USA. Circular migration between the USA and
the Dominican Republic has also been associated with the spread of HIV in
the Dominican Republic (Stillwaggon, 2006). Not all studies, however, have
found that labour migrants are more likely to be HIV-infected. In Southeast
Asia, Oppenheimer and colleagues found that labour migrants were more
risk-averse and less likely to be infected (Oppenheimer et al., 1998, cited in
Skeldon, 2002: 3).
Labour migration has played a very important role in the economies
of southern Africa and very likely has aggravated the HIV epidemic. But
migration is not unique to the region and may be insufficient to explain
the difference in prevalence between southern Africa and West Africa, the
Caribbean, the Middle East or Southeast Asia. In combination with other
characteristics of southern Africa, however, migration may have been an
important factor in the complex aetiology of the epidemic and should be
included in a model of HIV transmission.
In the post-apartheid years, there have been significant changes in the
southern African economies. Migration within South Africa and to South
Africa from most neighbouring countries has increased, but the composition
of the migrant labour force has changed significantly. Women are migrating
in increasing numbers, both with and without male partners. Rising levels
of male unemployment have changed the support relationships in the region
(Posel and Casale, 2003). Mark Hunter argues that the rapid increase and
enormous scale of the HIV epidemic in South Africa (and one might extend
that analysis to other countries in the region) was not an inevitable result of
a history of apartheid, although he acknowledges that labour migration and
the enforcement of family separation in apartheid did set the stage. Rather,
the decision to institute market-oriented reforms and austerity measures
in the 1990s was disastrous for a large proportion of the population. The
unemployment rate has risen to over 40 per cent, marital rates have fallen
sharply and inequality has increased substantially in the last two decades
(Hunter, 2007). Women’s migration has increased and informal settlements
have mushroomed around both large and small cities (Crush et al., 2005;
1. See for example, Brummer (2002); Clark et al. (2007); Crush et al. (2005); Gillespie,
Kadiyala and Greener (2007); Hargreaves et al. (2007); Horwitz (2001); Hunter (2002,
2007); ILO (1998); Kanyenze (2004); Lurie (2004); Lurie et al. (2002); Marks (2002);
Thahane (1991); Zuma et al. (2003).
198
Larry Sawers and Eileen Stillwaggon
Posel, 2003; Posel and Casale, 2003). Hunter draws much-needed attention
to the gap in health care provision that mirrors the stark inequality in income
and wealth in South Africa (Hunter, 2007). (For more on inequality in health
care, see Booysen, 2002, 2003).
Cross-border and internal migration is socially disruptive and affects
health in numerous ways. The literature on HIV and migration in southern Africa, however, focuses exclusively on the effect of migration on risky
sexual behaviour, often without data, to the exclusion of other health-related
factors (Brummer, 2002; Kanyenze, 2004). While the literature of social
medicine in South Africa of the last half century has always included some
mention of the miserable working and living conditions and inadequate nutrition of migrants (Horwitz, 2001; Marks, 2002), the sound bite that has
emerged in AIDS policy discourse emphasizes ‘the constant movement of
large numbers of sexually active men from town to countryside and back’
(Marks, 2002: 18). Certainly the behavioural implications of separating men
from their families and communities were very important in the spread of
STIs, and later HIV, in southern Africa. Nevertheless, that emphasis on
sexual behaviour reinforces the prevailing explanation for high HIV prevalence in sub-Saharan Africa that focuses almost exclusively on risky sexual
behaviour.
The literature on HIV and migration in southern Africa overlooks the
direct biological impact of deprivation on migrant health and its role in
accelerating the HIV epidemics in the region. There is a substantial body
of evidence that poor sanitation,2 nutrition,3 and inadequate health care
2. Poor sanitation spreads numerous bacterial, viral and parasitic infections. Borkow and
Bentwich (2006) survey a large number of studies and trials investigating the connection
between helminth infections and cell-mediated immunity. They argue that the majority of
observational studies provide evidence for interaction between HIV and helminth infection,
but there are many studies that do not show a connection and a few that suggest a protective
effect of certain helminth infections. Almost all of those studies, however, are observational
studies. Walson, Herrin and John-Stewart (2009) analysed almost 7,000 studies of HIV
patients treated with anti-helminthics. They find only three randomized controlled trials,
all three of which confirm the HIV–helminth connection. See also Bentwich et al. (2008);
Chan (2005); Fawzi and Hunter (1998); Friis and Michaelsen (1998); John et al. (1997);
Landers (1996); Montresor et al. (2001); Nacher (2002); PPC (2002); Semba et al. (1994);
Walson et al. (2008); Wolday et al. (2002); Woodward (1998). Additional studies are cited
in Stillwaggon (2006). In contrast to other helminths, the hypothesized link between HIV
and S. hematobium has to do with its effect on epithelial integrity and not only on its effect
on cell-mediated immunity. See the in-text discussion of S. hematobium below. In our
statistical analysis presented below, we do include any measure of helminth infection other
than schistosomiasis.
3. Protein, fats, calories and most of the 40-plus micronutrients necessary for human health
affect or modulate the immune system in some way and thus affect a person’s vulnerability
to infectious disease. There is a ‘vast’ scientific literature showing that malnutrition undermines the immune system and that vitamin supplementation improves immune response
(Mehta and Fawzi, 2007: 358). HIV is an infection and thus poor nutrition is likely to
make people more susceptible to HIV infection. Malnutrition may also shorten the lives of
Poverty, Endemic Disease and HIV
199
for people living in single-sex barracks and squatter settlements4 can increase the transmission of HIV. A fully specified model of the HIV epidemics in southern Africa must include variables that reflect what is known
about the interactions between HIV and other infectious and parasitic
diseases.
Poverty in Poor Countries
From the beginning of the HIV epidemics in sub-Saharan Africa, researchers
noted that HIV prevalence in many countries was higher in upper income
and wealth brackets. (Gillespie, Kadiyala and Greener, 2007 is the most
comprehensive survey of that literature; other reviews of the literature include Gillespie and Suneetha, 2005; Greener, 2004.) Most research on the
connection between income, wealth and HIV looks at a single country or
district within a country, but Mishra et al. (2007) use nationally representative surveys from eight sub-Saharan African countries to examine the issue.
They find that HIV prevalence is higher in the wealthiest quintile than in the
poorest quintile in all eight countries, usually by a substantial margin. Since
antiretroviral treatment in poor countries is more readily available for the
affluent than for poor people, wealthy people infected with HIV live longer
on average and that biases comparisons of prevalence across wealth levels.
Mishra et al. (2007) were not able to control for this effect, so the higher HIV
prevalence of the wealthy is at least in part a result of the greater longevity of
the wealthy rather than a reliable indicator of cumulative incidence. Moreover, wealthier people are more likely to live in cities and be better educated,
both of which are associated with higher HIV prevalence. Wealthier men
in the eight countries were also more likely to report two or more sexual
partners in the previous year (except in Tanzania), more lifetime sexual
those already infected with HIV and make them more contagious (Chan, 2005; Gillespie
and Suneetha, 2005; Stillwaggon, 2006). The most comprehensive and up-to-date survey
of studies and trials that examine the nutrition–HIV interaction (Mehta and Fawzi, 2007)
sums up the evidence as follows: vitamin A supplementation reduces morbidity and delays
mortality for HIV-infected children; multivitamin supplementation reduces adverse pregnancy outcomes in HIV-infected women, reduces mother-to-child transmission for poorly
nourished mothers, and slows the progression to AIDS for HIV patients. On the other
hand, the evidence from trials is that vitamin A supplementation alone does not protect
against mother-to-child transmission of HIV except for pre-term babies (see Coutsoudis
et al., 1999: 1522; Fawzi et al., 2001 provides corroborating evidence) and may even increase transmission (possibly due to dosing with massive amounts of the wrong form of the
vitamin). The evidence about whether vitamin A protects adults from contracting HIV is
mixed. In the statistical analysis below, we do not include any measure of nutrition because
of the lack of suitable measures at the national level.
4. Poor housing cannot provide adequate protection from disease vectors such as mosquitoes
that spread malaria and flies that spread trachoma. The connections between HIV and
malaria, and HIV and trachoma are discussed below.
200
Larry Sawers and Eileen Stillwaggon
partners, and sex with non-regular partners. On the other hand, the wealthy
in the surveys are more likely to report having information about HIV and using condoms. Using a multiple regression analysis and controlling for these
and other factors, Mishra et al. (2007) find that the wealth difference in HIV
prevalence becomes statistically insignificant for both men and women in
all but one country. Furthermore, there is some evidence that HIV incidence
in the region is falling among the wealthy and better educated (Gillespie,
Kadiyala and Greener, 2007: S12; Lopman et al., 2007). Nevertheless, we
do not make a case that HIV is primarily a disease of poverty since, even
with the above adjustments to the data, it is clear that it is not mostly the
poor who contract HIV in poor countries.
The argument that this article makes, however, is that HIV is a disease
of poor countries because widespread poverty increases prevalence for both
poor and wealthy people. It is widely believed — though there is little direct
evidence — that poverty can force women into transactional sex and into
sexual relations where they do not have the power to negotiate safe sexual
practices (Gillespie, Kadiyala and Greener, 2007: S6). Moreover, affluent
men can afford to pay for the sexual services of poor and desperate women.
The evidence from Mishra et al. (2007) is that the wealthy are more likely
to engage in sexual relations outside of monogamous unions. In countries
with more desperately poor women, both well-to-do men and poor women
may be more likely to become infected with HIV. Members of both groups
can then pass HIV on to their regular sex partners. If that were the case, we
would expect poor countries to have higher HIV prevalence than wealthier
countries.
Poor countries are also less likely to provide safe and effective medical care than higher income countries. Higher income people (especially
women), better-educated people and people living in cities have greater
access to physicians, clinics and hospitals, possibly explaining the higher
HIV prevalence among those groups. There is substantial evidence that
effective infection-control procedures (such as sterilizing syringes or specula) are not followed as consistently in poor countries as in higher income
countries.5 In the countries that Mishra et al. (2007) studied, other than
5. There is substantial evidence that an important proportion of the HIV infections in subSaharan Africa and elsewhere result from iatrogenic/nosocomial transmission. For example,
see Deuchert and Brody (2006, 2007a); Chan (2005); Gisselquist (2008). In addition, most
of Volume 20 of the International Journal of STD and AIDS (2009) is devoted to the topic.
Potterat (2009) authors the review article that leads the volume. Southern Africa seems to
be unusually derelict in providing safe medical care, even when compared with other subSaharan African countries. Five of the nine countries in southern Africa use no autodisable
syringes in child vaccination programmes. Elsewhere in sub-Saharan Africa, only 15 per
cent of countries fail to do so (see Table 5). Another important reason for an association
between poor health care and HIV is that diseases that may promote HIV transmission,
such as schistosomiasis, malaria and STIs, are less likely to be treated or treated effectively
and safely in countries without good health care.
Poverty, Endemic Disease and HIV
201
Lesotho, wealthy men were more likely to be circumcised than poor men.
Circumcision is thought to reduce HIV transmission, but in a poor country
where unsafe medical practices are widespread, circumcision can spread
HIV rather than protect against it. In sum, iatrogenic/nosocomial transmission of HIV will be high among the poor and perhaps even higher
among the affluent within poor countries when compared with wealthy
countries.
Poverty may push some people into risky sexual behaviour or into contact
with unsafe and inadequate medical practices, but it also forces people
into other behaviours that pose risks for their health. Poor countries on
average have a higher share of the population that is poorly nourished,
lives in dwellings that do not protect against disease vectors, and resides
in communities with inadequate sanitation. Poor people thus are burdened
with high rates of infectious and parasitic diseases. The health of the affluent
in poor countries is also affected since they are surrounded by sick people
and thus more likely to become ill than in countries where disease is less
prevalent. For example, malaria cases exceeded 25 per cent of the population
in twenty-seven African countries in 2006; schistosomiasis prevalence was
25 per cent or more in twenty-two African countries (WHO, 2008b, 2009).
Living in a country where one is surrounded with disease raises the likelihood
that any mosquito bite or contact with ground water will bring an infection.
To the extent that some of those diseases promote the transmission of HIV,
HIV is a disease of poor countries.
Inequality
All southern African countries except Malawi have above average income
inequality. Namibia has the highest income inequality in the world and
Botswana is third highest. South Africa, Lesotho, Zambia, Swaziland and
Zimbabwe have Gini coefficients above 50, well above the average in the rest
of Africa and in other low- and middle-income countries. There is a widely
noted correlation across countries between income inequality and HIV prevalence (Gillespie et al., 2007; Mahal, 2002; Over, 1998; Stillwaggon, 2000,
2002; Tsafack and Bassolé, 2006), which appears to be related to the foregoing arguments about the connection between poverty and HIV. For a
country at any given level of income, a higher level of income inequality
means that the poor are a greater share of the population. If a large proportion of poor people in a country is an important driver of HIV epidemics,
then a country with high income inequality should have a higher HIV prevalence. South Africa, Botswana, Namibia and Swaziland are middle-income
countries (and Lesotho is an enclave surrounded by South Africa) and those
countries have the highest HIV prevalence in the world. At the most fundamental level, in the four middle-income countries of southern Africa, stark
inequality means that the relatively high per capita GDP masks the profound
202
Larry Sawers and Eileen Stillwaggon
poverty of the majority of the population and the direct impact of that poverty
on health.
Gender Inequality
Another dimension of inequality is gender inequality. Many researchers argue — though with little direct evidence — that the oppression of women,
by reducing their incomes, assets and power, makes it difficult for them to
negotiate safe sexual practices and encourages them to engage in transactional sex (Brummer, 2002; Gillespie, Kadiyala and Greener, 2007; Hunter,
2007; Over, 1998; Tsafack, 2006). In that case, a country with greater gender
inequality would have higher rates of HIV. On the other hand, greater gender
equality could mean fewer traditional restrictions for women on sexual behaviours, some of which might be risky. At any rate, the evidence shows that
variations in sexual behaviour between countries do not explain the higher
prevalence of HIV in sub-Saharan Africa.
Risky Sexual Behaviours
The previous sections present the arguments that long-distance labour migration, poverty, income inequality and gender inequality could all lead to
higher prevalence of risky sexual behaviours. As more data from survey
research have become available, it has been increasingly clear — by almost
all measures of risky sexual behaviour, including having multiple sexual
partners — that sub-Saharan Africa is little different from other regions
around the world. The countries with the highest measured rates of risky
sexual behaviour are in North America and Western Europe, where HIV
prevalence is very low, not in southern Africa, where HIV prevalence is
very high (see Stillwaggon, 2006; Wellings et al., 2006).
Confronted with that evidence, defenders of the notion that risky sexual
behaviour accounts for the higher HIV prevalence in sub-Saharan Africa
have narrowed their argument to a single kind of sexual behaviour — overlapping ongoing relationships or concurrency (see especially Epstein, 2007;
Halperin and Epstein, 2004, 2007; Mah and Halperin, 2008, 2009). Nevertheless, the supporters of the concurrency hypothesis have not shown that
concurrency is unusually common in sub-Saharan Africa generally or southern Africa specifically, that HIV spreads faster with concurrency compared
with other high-risk forms of multiple sexual relationships, or that the prevalence of concurrency is in fact correlated with HIV prevalence (Kretzschmar
et al., 2009; Lurie, 2009; Lurie, Rosenthal and Williams, 2009; Stillwaggon
and Sawers, 2010).
The supporters of the concurrency hypothesis argue that concurrency is far
more common in sub-Saharan Africa than in other regions, but they give data
Poverty, Endemic Disease and HIV
203
for only a few African countries and even fewer countries elsewhere.6 Some
of the surveys they present as evidence of concurrency do not even measure
concurrency, they misreport data, they cite studies that actually show low
rates of concurrency in Africa, and none of their information on attitudes
toward concurrency is about non-African countries and is thus irrelevant
to a hypothesis that is essentially comparative (Stillwaggon and Sawers,
2010).
Mishra and Bignami-Van Assche (2009) compile all DHS surveys that
measure concurrency using the same definition. In sub-Saharan Africa (ten
countries for men and eight countries for women), 8.7 per cent of men and
0.8 per cent of women had had concurrent partners in the previous year (ibid.,
derived from Tables 5, 8 and 9). In contrast, in the USA, 11 per cent of men
had had a concurrent relationship in the previous year and 12 per cent of
women had had a concurrent relationship in the previous five years (Adimora,
Schoenbach and Doherty, 2007; Adimora et al., 2002). The most comprehensive study of concurrency shows that it is not especially high in sub-Saharan
Africa.7
Concurrency cannot explain the higher prevalence of HIV in sub-Saharan
Africa unless it is more dangerous than other forms of multiple partnering found elsewhere. Supporters of the concurrency hypothesis point to
one mathematical model of disease transmission, which shows that certain
kinds of concurrent relationships under certain conditions can spread HIV
infections far faster than serial monogamy (Kretzschmar and Morris, 1996;
Morris and Kretzschmar, 1997). The model, however, makes several crucial
but completely unrealistic assumptions: that women and men are equally
likely to have concurrent partners, that all partners have sexual contact with
every partner every day, that partnerships on average last only a few months,
and that each sexual contact has a 5 per cent chance of transmitting HIV.
Every survey of concurrency for women shows that women are far less likely
to have concurrent partners than men. When Morris and Kretzschmar (2000)
recalibrate their model using actual survey data on male and female concurrency in Rakai, Uganda (and use realistic assumptions about the length
of relationships), they find that concurrency leads to only 26 per cent more
cases of HIV when compared with serial monogamy (not exponentially
larger HIV prevalence). They achieve that result even though their Rakai
model still uses the preposterous assumption of daily sex with all partners.
6. Halperin and Epstein (2004, 2007) and Mah and Halperin (2008, 2009), relying mostly on
data from one city in Latin America, one country each in North America and Europe, three
countries and two cities in Asia, repeatedly state that concurrency is less common outside
of Africa.
7. The supporters of the hypothesis dismiss the study because it is based on household surveys
(although most of the data they offer in support of their hypothesis are also from household
surveys) and because the surveys do not measure concurrency (although they do measure
concurrency) (Mah and Halperin, 2009; Stillwaggon and Sawers, 2010).
204
Larry Sawers and Eileen Stillwaggon
The Rakai model also continues to assume a per-act transmission rate of .05
(from a study of Thai soldiers, nearly half of whom had STIs). Transmission among otherwise healthy adults is about one fiftieth of that rate, but
transmission rates are far higher if either partner has STIs, schistosomiasis
and/or malaria (Boily et al., 2009; Chan, 2005; Deuchert and Brody, 2007b).
Table 5 shows that the burdens of all of those diseases are far higher in
sub-Saharan Africa than elsewhere. The model produces the rapid spread of
HIV with concurrency by assuming transmission rates that are only found
among unhealthy people. In other words, it is not concurrency that is spreading HIV in sub-Saharan Africa; it is cofactor infections that, by promoting
the transmission of HIV, are driving the HIV epidemics in the region.
Even if concurrency was both an especially dangerous form of multiple
partnering and also more prevalent in southern Africa than elsewhere, it
would be difficult to argue that concurrency explains the high prevalence
of HIV in the region without demonstrating that the prevalence of HIV and
concurrency are indeed correlated. The risk posed by concurrency is not
what one does, but what one’s partner is doing (which is often unknown),
so the only direct way to test the concurrency hypothesis is to map sexual networks.8 The network study offered as evidence for the concurrency
hypothesis (Helleringer et al., 2009) maps a sexual network, but it cannot
show a statistically significant correlation because only 22 per cent of persons interviewed were willing to be tested for HIV and because it failed
to distinguish between serially monogamous and concurrent relationships.
Moreover, the study was conducted on a remote island and may tell us little
about sexual networks elsewhere. A correlation within a population between
self-reported concurrency (not that of their partner) and HIV prevalence cannot provide direct evidence, but at least that correlation would be consistent
with the concurrency hypothesis. There have been four such ecological studies and none of them finds a statistically significant correlation (Jewkes et al.,
2006; Lagarde et al., 2001; Mattson et al., 2007; Mishra and Bignami-Van
Assche, 2009).
In short, the concurrency hypothesis is built on unconvincing empirical
evidence and mathematical models that make critically important but completely unrealistic assumptions. Efforts to establish a statistical correlation
between concurrency and HIV have failed. The stress on concurrency in
sub-Saharan Africa is another stage in the two-decades-old practice of explaining HIV epidemics in Africa without empirical evidence, driven only
by presumptions about the exceptional character of African sexuality.9
8. Mah and Halperin (2009) also argue that two surveys (unnamed and without citation) of
newly infected people thought their partners had other partners. Those findings tell us
only that the respondents believed the message they repeatedly heard from HIV-prevention
programmes for adults, which talk only about safe sex. Those who believe the message and
have HIV will surely suspect that their partners are having sex with someone else.
9. See Stillwaggon (2003) for an explanation of why social scientists and policy makers have
assumed, without evidence, that African sexuality is exceptional.
Poverty, Endemic Disease and HIV
205
Cofactor Infections
Between otherwise healthy adults in developed countries, HIV has very
low rates of heterosexual transmission (Chan, 2005; World Bank, 1997).
In the developing world, including southern Africa, however, the majority
of the population is not ‘otherwise healthy’. Field studies and trials as well
as laboratory research have found substantial evidence for a connection
between the transmission of HIV and several diseases highly endemic in
southern Africa. The relative affluence of southern Africa and the emphasis
on risky sexual behaviour have led most policy makers to overlook the
biological vulnerability of the majority of southern Africa’s population to
infectious disease.
There seems to be little doubt that sexually transmitted infections (STIs)
increase transmission of HIV.10 Some STIs lead to genital ulceration creating
entry points for the HIV virus, thereby making those who are not infected
with HIV more vulnerable to infection and making infected persons more
likely to transmit HIV. Both ulcerative and non-ulcerative STIs also make
those who are not infected with HIV more vulnerable to HIV infection by
producing genital inflammation, which draws CD4+ cells to the infected
tissues. Those cells are the channel through which HIV infects the body.
Furthermore, both ulcerative and non-ulcerative STIs make those who are
infected with HIV more contagious by increasing viral shedding, which is
the discharge of virus in the genital tract. In poor populations, bacterial STIs
are more likely to be untreated because of lack of access to health care (Sturm
et al., 1998). The health burden of untreated STIs is far higher in African
countries than in other low- and middle-income countries (see Table 5).
Together, these facts suggest that one of the reasons for the high prevalence
of HIV in southern Africa is the high prevalence of untreated STIs.11
Urinary schistosomiasis (S. hematobium), which afflicts more than 100
million people in sub-Saharan Africa (and almost nowhere else; WHO,
2008a), also can act as a cofactor of HIV transmission much in the same way
as do STIs. Worms and ova of S. hematobium infect the reproductive tract
10. See Fleming and Wasserheit (1999) for their review of more than eighty trials; see also
Boulton and Gray-Owen (2002); Chan (2005); Corbett et al. (2002); Grosskurth et al.
(1995); Hayes et al. (1995); Laga et al. (1993); Rotchford et al. (2000); Wang et al. (2001);
Wolday et al. (2004).
11. Trachoma might seem unrelated to HIV since it is an eye infection, but trachoma is caused
by the same bacteria (C. trachomatis) as the STI chlamydia. As many as 70 per cent of
women and 25 per cent of men with chlamydia are asymptomatic (Webster et al., 1993).
That suggests the possibility that trachoma sufferers are coinfected genitally without the
knowledge of the patient or medical personnel. Alternatively, high prevalence of trachoma
may be merely a proxy for poor health care, since it is easily treated in its early stages
and extracts a very high toll in blindness that would not be tolerated in a country with an
adequate health system. Consequently, we include the trachoma variable in some of the
regressions presented below because we think it suggests an important factor worthy of
further investigation.
206
Larry Sawers and Eileen Stillwaggon
of both men and women, producing lesions and inflammation of the genital
area (Attili et al., 1983; Feldmeier et al., 1995; Leutscher et al., 1998; Marble
and Key, 1995). The genital ulcers of S. hematobium are indistinguishable
without biopsy from those of ulcerative STIs. A recent trial in Zimbabwe
found that genital lesions of schistosomiasis increased HIV risk in women
three-fold compared to women in the same villages without genital lesions of
schistosomiasis (Kjetland et al., 2006).12 The WHO Department of Control
of Neglected Tropical Diseases recently identified the interactions of HIV
with endemic parasitic and infectious diseases as a priority for research
(WHO, 2007: 45), but AIDS policy organizations have not adopted protocols
for treating coinfections of these locally endemic diseases.
Malaria, which is especially prevalent in sub-Saharan Africa, interacts
with HIV, increasing viral load up to ten-fold. Viral load remains elevated
in HIV-infected persons for six weeks after malarial episodes (Abu-Raddad
et al., 2006; Hoffman et al., 1999; Kublin et al., 2005). Elevated viral load not
only increases individual contagiousness of HIV-infected persons, it also affects the dynamics of the epidemic at a population level (Abu-Raddad et al.,
2006; Bloland et al., 1995; Corbett et al., 2002; Herrero et al., 2007; Hoffman
et al., 1999; Whitworth et al., 2000; Xiao et al., 1998; for additional discussion and sources, see Stillwaggon, 2006). Individuals in malaria-endemic
areas have a higher probability of sexual contact with persons who have
high viral load due to coinfection with malaria and who are therefore more
contagious. In the AIDS literature, there is increasing and warranted attention to the first few months after infection with HIV because people have
high viral loads at that time and are thus more likely to transmit HIV to their
partners (Brenner et al., 2007; Cohen and Pilcher, 2005; Pilcher et al., 2004;
Quinn et al., 2000). What also needs to be considered is that malaria and
other coinfections produce recurring elevation of viral loads over the lifetime
of the HIV-infected person, not just in the first few months following HIV
infection.13
12. A recent study shows that rhesus monkeys infected with S. mansoni are more susceptible to
SIV (simian immunodeficiency virus) infection and have elevated viral load compared to
monkeys without schistosomiasis (Chenine et al., 2008). Human trials with S. mansoni have
not looked for differential incidence (whether coinfection with schistosomiasis makes one
more vulnerable to HIV) but have examined viral load and CD4+ counts of coinfected HIV
patients (that is, whether HIV patients are more contagious) and how they are affected by
treatment with antihelminthics. Those trials have produced mixed results. Kallestrup et al.
(2005) support the hypothesis that schistosomiasis coinfection makes HIV patients more
contagious by raising viral load, but others do not (Brown et al., 2005; Elliott et al., 2003;
Lawn et al., 2000; Mwinzi et al., 2001). There are 54 million people in sub-Saharan Africa
afflicted with S. mansoni. The argument of this article is that the genital manifestations
of S. hematobium make it different from other species of schistosomiasis as well as other
helminths.
13. We have provided a rationale for a long list of independent variables in our model, but we
should point out a variable that we are not including. Numerous colleagues have suggested
we add clade to our analysis. The subtype of HIV-1 that is dominant in southern Africa
Poverty, Endemic Disease and HIV
207
METHOD
This research extends the work of authors who use cross-national data and
multiple regression analysis to estimate the effect of factors thought to
drive the HIV epidemic (Bonnel, 2000; Deuchert and Brody, 2007a; Drain
et al., 2004; Mahal, 2002; Over, 1998; Stillwaggon, 2000; Stillwaggon,
2002; Tsafack, 2008). All but one of those studies focus on a specific issue
that the author(s) believe especially important: risky sexual behaviours,
gender discrimination, human development, income inequality, economic
growth, iatrogenic transmission of HIV, and nutrition. Those authors all use
the same or similar socio-economic control variables and then add other
variables relevant to the specific issue addressed in their studies. The most
common control variables used in those studies are income inequality, per
capita income (level or growth rate), urbanization (level or growth rate),
proportion of the population that is Muslim, age of the epidemic, a measure
of gender discrimination, literacy or school enrolment rates, and one or
more binary variables for region. Deuchert and Brody (2007a) examined the
iatrogenic or nosocomial spread of HIV. Their study found a measure of
unsafe injection practices to be positively and significantly correlated with
HIV prevalence, so we also include that variable in our equations.
To those control variables we add one not used in previous studies. We
include a measure of cross-border migration because of the special importance of migration in the discussion of HIV in southern Africa. We call all
of these control variables ‘the basic model’. Finally, in the present study,
which extends our previous research (Sawers, Stillwaggon and Hertz, 2008),
is clade C, but that subtype is relatively rare in most other countries (the most important
exceptions being Bangladesh, Papua New Guinea, India and Nepal). That fact suggests
to many observers that clade C is more fit for transmission than other clades and several
contributors to the virological literature have speculated just that (Chan, 2005; Laurent et al.,
2007; Ping et al., 1999; Rainwater et al., 2005; Spira et al., 2003; Tebit et al., 2007). There
is, however, little evidence to confirm that suspicion. One study finds that vaginal shedding
in pregnant women is greater with clade C (John-Stewart et al., 2005), but another study
found no difference in mother-to-child transmissibility of the virus among different clades
(Tapia et al., 2003). A third study finds no evidence of different fitness for heterosexual
transmission of the different clades (Morison et al., 2001). The most complete data on
HIV clades are from the HIV Sequence Database provided by the Los Alamos National
Laboratory at http://www.hiv.lanl. gov/content/sequence/HIV/mainpage.html, from which
one can obtain the per cent of HIV infections of clade C for seventy-five of the ninety-one
countries in our dataset. That variable is strongly and positively correlated with both HIV
prevalence and the southern Africa variable in our dataset, but when the variables of the
basic model are added to the regression, the regression coefficient on clade C becomes
unexpectedly negative and significant at only the 96 per cent level. If one then adds the
cofactor infections variables to the regression, the regression coefficient on the clade C
variable loses its statistical significance (the t statistic is only 0.40, significant at the 0.691
level). Accordingly, our research provides no support for the notion that differences in HIV
sequences between countries can explain the higher HIV prevalence in southern Africa
using multivariate analysis.
208
Larry Sawers and Eileen Stillwaggon
we add to the basic model measures of cofactor infections that are thought
to promote the transmission of HIV.
The regional binary variable in our model takes the value of one for the
nine countries in southern Africa and zero for other countries. A positive
and significant coefficient on the southern Africa variable tells us that HIV
prevalence is higher in southern Africa than elsewhere, but does not tell us
why. Our objective is to explain differences in HIV prevalence among lowand middle-income countries (not just find statistical associations and a high
R2 ). Location is not a policy-sensitive variable. Although using a regional
binary variable raises the predictive value of the model in a technical sense,
it begs the question of what is causing high prevalence in the region. Accordingly, we measure our success in explaining the variation in HIV among
countries by how far we can reduce the size of the regression coefficient on
the southern Africa variable.
Using data for ninety-one developing and transition countries,14 we estimate a number of ordinary least squares multivariate regressions on the
log of adult HIV prevalence in 2007.15 All of the regressions reported here
include the variables of the basic model: a binary variable for the nine countries of southern Africa, the Gini coefficient (the most widely used measure
of income inequality), log of per capita income on a purchasing power parity
basis, per cent of adults who are literate,16 per cent of the population living
in urban areas, age of the epidemic, a measure of gender discrimination (one
minus the ratio of female economic participation to male economic participation), per cent of the population who are Muslim, workers’ remittances
as a percent of GDP, and a binary variable that takes the value of one if
the country used no autodisable syringes in child vaccination programmes.
Table 1 lists the source for each of the variables. The first step is to regress
14. Countries with per capita income in 2003 of less than US$ 12,000 on a purchasing power
parity basis were included in the analysis. Rather than imputing missing values, we dropped
countries with missing data from the analysis, leaving a dataset of ninety-one countries.
See Appendix for a list of countries in the dataset. The present study follows established
practice by using the log form of HIV prevalence, per capita income and, where possible,
measures of cofactor infections, which reduces the influence of outliers and generally leads
to more robust and efficient estimates when analysing highly skewed variables such as
these.
15. It would have been preferable to use HIV incidence to describe how the epidemic is
unfolding. Prevalence includes historical information and is confounded by greater survival
in countries with effective antiretroviral treatment programmes, but incidence data are
generally not available.
16. All measures of literacy and school enrolment for our sample are highly correlated and
perform similarly as regressors. In our sample, the adult literacy rate and the female-tomale literacy ratio are virtually identical in a statistical sense (with a simple correlation of
0.99). Using the latter variable, however, gives the misleading impression that we are using
it to measure gender discrimination; hence the present study uses the more straightforward
variable, adult literacy, to avoid that confusion.
Poverty, Endemic Disease and HIV
209
Table 1. Sources of Data
Variable
Adult HIV prevalence,
2006
Gini coefficient
Income per capita
Urbanization
Age of the epidemic
Adult literacy
Female participation
ratio
Muslim per cent
Remittances
Autodisable syringe
use
STIs, schistosomiasis,
trachoma
Malaria prevalence
Definition
Per cent of persons aged 15 to
49 with HIV
Measure of income inequality,
2004
Per capita GDP (PPP basis), 2003
Per cent of population in urban
areas, 2003
Years since first diagnosis of HIV
Per cent of adult population who
are literate, 2003
One minus ratio of female
economic activity rate to male
economic activity rate, 2005
Per cent of the population who are
Muslim
Workers’ remittances as a per cent
of GDP
Did not use autodisable syringes in
child vaccination programmes
DALYs (Disability-adjusted life
years) for each disease per
100,000 people, 2002
Cases as a per cent of population
Source
UNAIDS (2008)
World Bank, World Development
Indicators
World Bank, World Development
Indicators
World Bank, World Development
Indicators
World Bank (1997: 318–24)
UNDP, Human Development
Indicators, 2005
UNDP, Human Development
Indicators, 2007
CIA World Factbook
www.cia.gov/library/publications/
the-world-factbook/ accessed
March 24, 2009
World Bank, World Development
Indicators
WHO (2005: Annex Table 7)
WHO (2004)
WHO (2008b)
only the variables of the basic model on HIV prevalence. Next, we add
each cofactor infection variable, one at a time, to a regression with the basic
model variables. Lastly, we regress all of the cofactor infections together
with the basic model variables. Estimation of all of the regressions is robust
to heteroskedasticity.
RESULTS
In the equation presented in Table 2, only the variables of the basic model
are regressed on HIV prevalence. The R2 , that is, the per cent of the variance
in the log of adult HIV prevalence associated with the independent variables
taken together, is 64.0 per cent. Six of the ten independent variables are
significant at the 95 per cent level or better and another is significant at
the 90 per cent level. The southern Africa variable, the Gini coefficient,
and the age of the epidemic are positively correlated with HIV prevalence.
The Muslim per cent, adult literacy, the gender discrimination index and
income per capita variables are negatively associated with HIV prevalence.
The regression coefficient on each independent variable in the model tells us
the change in the log of HIV prevalence — after accounting for the effects
Larry Sawers and Eileen Stillwaggon
210
Table 2. Multiple Regression Analysis of HIV Prevalence with Basic Model
and without Cofactor Infections (N = 91)
Variable
Southern Africa
Gini coefficient
Income per capita (log)
Urbanization
Age of epidemic
Adult literacy
Female participation ratio
Muslim
Remittances
Autodisable syringe use
Constant
Coefficient
t Statistic
Probability
2.7901
.0291
−.4862
.0083
.1003
−.0189
−.0195
−.0089
.0083
.2527
−.2009
6.35
1.71
−2.01
0.87
2.41
−2.16
−2.51
−2.56
0.50
0.94
−0.10
0.000
0.091
0.047
0.387
0.018
0.034
0.014
0.012
0.618
0.351
0.919
Beta
Coefficient
.4717
.1521
−.2478
.0951
.1777
−.2551
−.2047
−.1926
.0294
.0715
of the other independent variables — when the value of the independent
variable changes by 1. Since the southern Africa variable takes the value
of either 1 or 0, its coefficient tells us the difference in the log of HIV
prevalence between southern Africa and elsewhere after taking into account
the effects of the other basic model variables.
Average HIV prevalence in southern Africa is about 16 times the average prevalence in other low- and middle-income countries in our dataset.
The regression coefficient on the southern Africa variable is 2.79 in the
regression using only the variables in the basic model. To find the ratio
of HIV prevalence in southern Africa and elsewhere that is not explained
by the basic model variables, one takes the antilog (with an adjustment
for heteroskedasticity) of the coefficient on the southern Africa variable.17
The regression coefficient on southern Africa tells us that HIV prevalence
in southern Africa not explained by the model is 14.9 times the level in
other low- and middle-income countries. In other words, the basic model
‘explains’ in a statistical sense less than 10 per cent of the nearly 16-fold
differential in HIV prevalence between the southern African and other lowand middle-income countries.
17. The coefficient on the southern Africa variable estimates the difference between the mean
log of HIV prevalence in southern Africa compared to other countries in the dataset. In order
to convert this to a statement about the difference in mean HIV prevalence (as opposed
to mean log prevalence), one must take the antilog of that coefficient, but also correct
for the non-zero mean of the antilogged error terms. That correction takes into account the
heteroskedasticity in error terms between southern Africa and the other countries and yields
a close approximation of the actual ratio of means. The ratio of the predicted means is the
antilog of [the regression coefficient on the southern Africa variable plus (half the difference
between the variance of the regression residuals in southern Africa and the variance of the
regression residuals outside of southern Africa)]. Thomas Hertz devised this calculation.
Poverty, Endemic Disease and HIV
211
Table 3. Multiple Regression Analysis of HIV Prevalence with Basic Model
and each Cofactor Infection Separately (N = 91)
Variable
Chlamydia (log)
Gonorrhoea (log)
Syphilis (log)
STIs (log)
Schistosomiasis
Malaria
Coefficient
t Statistic
Probability
.8932
.8894
.4426
1.0076
.0074
.0689
4.33
5.70
5.37
6.21
4.21
5.89
0.000
0.000
0.000
0.000
0.000
0.000
The next step in the analysis is to add each of the five cofactor infection
variables — one at a time — to a multiple regression in which the other independent variables are those in the basic model. Schistosomiasis, chlamydia,
syphilis and gonorrhoea are measured by DALYs per 100,000 population,
and malaria is measured by the annual number of cases as a per cent of the
population.18 The results of those five regressions are reported in Table 3.19
All of the cofactor infection variables are positively correlated with HIV
prevalence, and all of the cofactor infection variables are significant at the
99.9 per cent level.
The regressions reported in Table 3 do not measure the true importance of
each cofactor infection since they are subject to missing variable bias, that
is, they do not include the other cofactor infections that are correlated with
HIV prevalence. On the other hand, the five cofactor infections are highly
collinear. (The simple correlation [R] in the sample between each pair among
the five cofactor infections ranges between 0.56 and 0.90 — and exceeds
0.68 for all pairs excluding malaria.) We entered a variety of different
combinations of the five diseases into regressions with the basic model
variables, but they produced different results depending on which cofactor
infections were entered and how they were combined. Collinearity prevents
us from determining the separate effect of each infection. Nevertheless, the
objective of our research is not to show the effect of any particular disease,
but to show that the burden of certain diseases taken together explains
statistically the variations in HIV prevalence across countries. Accordingly,
we add together the four cofactor infections that are measured the same
way (in DALYs per 100,000 population) and are thus additive. We call
this variable ‘combined cofactors variable’. (Malaria is entered separately
18. Because the vast majority of deaths from malaria are in infants, the DALYs associated with
malaria are extremely high. That measure, however, does not reflect the burden of malaria
among sexually active adults, unlike the DALYs from other diseases. We therefore use
malaria prevalence as our measure.
19. For the same reason (statistical efficiency) that we use the log of HIV prevalence and
per capita income, we also use the log of chlamydia, syphilis and gonorrhoea in these
regressions. There are many countries in the dataset with no schistosomiasis or malaria and
one cannot calculate a logarithm of zero.
212
Larry Sawers and Eileen Stillwaggon
since it is measured differently.) Combining the four cofactor infections
into a single variable eliminates the collinearity problem, but also makes
epidemiological sense since it measures the combined burden of the four
cofactor infections, all of which are thought to promote HIV transmission
through the same mechanisms. All southern African countries have very high
levels of chlamydia, gonorrhoea and schistosomiasis (110 or more DALYs
per 100,000 population) and only African countries have very high levels of
those three infections. Five southern African countries also have very high
levels of syphilis and only three countries outside Africa have levels that
high.
Accordingly, we estimate a regression with the basic model variables
together with the malaria prevalence and the combined cofactors variable.
That raises the R2 from 66.8 to 80.1 per cent and raises the F statistic for the
equation from 22.3 to 55.1. Both the combined cofactors variable and malaria
prevalence are significant at the 99.9 per cent level and the F statistic for those
two variables taken together is 42.1, significant at the 99.99 per cent level.
That test confirms our hypothesis that cofactor infections are significantly
correlated with HIV prevalence after controlling for other variables.20 The
beta coefficients for the two variables measuring cofactor infections are
larger than for any other variable (.490 and .442) and substantially larger than
either the southern Africa variable (.320) or the per cent Muslim variables
(–.206). The beta coefficients measure which independent variables have a
greater statistical impact on the dependent variable after taking into account
the different units by which the variables are measured.
The regression coefficient on the southern Africa variable in the equation
including the cofactor infection variables is 1.89, which tells us that the HIV
prevalence not ‘explained’ by the other variables in the equation is 5.5 times
that of other low- and middle-income countries. In other words, adding the
cofactor infections to the model allows us to ‘explain’ in a statistical sense
about 70 per cent of the 16-fold differential in HIV prevalence between
southern Africa and other low- and middle-income countries, substantially
more than the model without the cofactor diseases, which explains less than
10 per cent of the southern Africa differential.
One can compare the regression coefficient and t statistic on the log of per
capita income in the regression with only the variables of the basic model
(in Table 2) to those statistics in the regression that includes the cofactor
20. As noted above, there is a substantial medical literature that supports our contention that
STIs and S. hematobium are cofactors with HIV, but there is no such literature connecting
trachoma to HIV. Nevertheless, it is plausible that persons infected with trachoma also have
unrecognized genital infection with the same bacteria, that is, they have chlamydia. We
re-estimate the regression presented in Table 4, replacing the combined cofactors variable
with one that includes trachoma. The two regressions are almost identical except that when
trachoma is included, the coefficient on the southern Africa variable falls from 1.89 to 1.76.
In other words, including trachoma enhances our ability to ‘explain’ the southern Africa
differential.
Poverty, Endemic Disease and HIV
213
Table 4. Multiple Regression Analysis of HIV Prevalence with Basic Model
and Cofactor Infections (N = 91)
Variable
Coefficient
t Statistic
Probability
Combined cofactors
Malaria prevalence
Southern Africa
Gini coefficient
Income per capita (log)
Urbanization
Age of epidemic
Adult literacy
Female participation ratio
Muslim
Remittances
Autodisable syringe use
Constant
.7127
.0492
1.8924
.0284
.2932
−.0023
.0274
.0134
−.0075
−.0095
.0436
.3165
−10.4735
4.61
3.74
5.94
2.09
1.58
−0.27
0.90
1.81
−1.01
−3.41
2.73
1.38
−5.90
0.000
0.000
0.000
0.040
0.118
0.788
0.373
0.074
0.315
0.001
0.008
0.173
0.000
Beta
Coefficient
.4905
.4423
.3200
.1487
.1494
−.0259
.0485
.1808
−.0787
−.2055
.1536
.0896
infections in Table 4. The coefficient on per capita income in the regression without the cofactor infection variables is −0.486 (significant at the
95 per cent level), that is, lower income countries, other things being equal,
have higher HIV prevalence. Adding the cofactor infection variables to the
equation turns the coefficient on per capita income positive and eliminates
its statistical significance. That is consistent with our argument that what
is important about the inverse association between HIV prevalence and per
capita income is the high disease burden of the poor.
Limitations
The statistical analysis presented above shows that after controlling for a
number of factors including the presence of cofactor infections, location in
southern Africa by itself accounts for less than a third of the 16-fold difference in HIV prevalence between that region and other low- and middleincome countries in our sample. Two points should be emphasized. That
the coefficient on the southern Africa variable remains significantly positive means that our model suffers from missing variable bias. The southern
Africa variable is without causal content. If it remains statistically significant in the regression equations, it means that something else sets southern
Africa apart from the rest of the world that we have not yet discovered or
found a way to measure adequately. Some examples of variables that one
would like to add to the equation include condom use, a better measure
of injection safety practices in medical and non-medical settings, various
measures of risky sexual behaviour,21 prevalence of circumcision (not just
21. The most widely available measure of risky sexual behaviour is Age at First Sex for
Females, but that variable has not been measured for almost half of the countries in our
214
Larry Sawers and Eileen Stillwaggon
per cent Muslim), and cross-border traffic with countries that have high HIV
prevalence.22
DISCUSSION
In retrospect, it is clear that, in southern Africa, HIV was an epidemic
‘waiting to happen’ (Marks, 2002). The factors that made southern Africa an
ideal setting for HIV to flourish are numerous. What follows is a discussion
of how southern Africa compares with the rest of sub-Saharan Africa and
with other low- and middle-income countries. Table 5 presents means of
the variables used in this analysis for four groups of countries — southern
Africa, central Africa (the belt of a dozen contiguous countries that stretches
across Africa from Angola to Kenya), the rest of sub-Saharan Africa, and
other low- and middle-income countries outside sub-Saharan Africa.
Southern African shares many characteristics with the rest of sub-Saharan
Africa that set them apart from other low- and middle-income countries.
Excluding South Africa, the proportion of the population who live on less
than US$ 2 a day (not shown in Table 5 because it was not used in the
regressions) is the same in southern Africa (71 per cent) as elsewhere in
sub-Saharan Africa.23 That compares with an average of 40 per cent in other
dataset. Twenty-two of the forty-six countries for which there are such data are in Africa, so
those forty-six countries are a representative sample of neither African countries nor lowand middle-income countries. Sexual behaviour surveys reflect a bias in data collection
common in AIDS research. The assumption that the AIDS epidemics in sub-Saharan
Africa result from something distinctive about African sexuality means that there are more
surveys about sexual behaviour in African countries than in other developing countries. For
a discussion of bias in data selection in Africa, see Stillwaggon (2003).
22. We know that HIV can spread from country to country across land borders. Truck routes
are frequently mentioned as conduits for the spread of HIV in sub-Saharan Africa and
prevalence is higher along highways than in more remote parts of the same country. In
an attempt to capture those neighbourhood effects, we measure, for every country in the
sample, the average HIV prevalence of all countries with which it shares a land border.
Adding that variable to the regression in Table 4 reduces the coefficient on southern Africa
to insignificance (the t statistic is 0.20 and the p is 0.789) while leaving the coefficients on
the cofactor infection variables significant at the 98 per cent level. In that sense, we have
completely explained the southern Africa differential in HIV prevalence. Nevertheless, we
have merely shifted our ignorance back one step: in a statistical sense, we know why each
country in southern Africa has a high level of HIV, but we have not explained why all
of its neighbours have high prevalence. Furthermore, the southern Africa variable and the
HIV prevalence in neighbouring countries variable are so highly correlated (R = 0.92) that
they are practically the same variable. That collinearity makes it difficult to tease out the
separate effects of the two locational variables. In short, there are some very complicated
spatial dynamics to the spread of HIV across sub-Saharan Africa and around the world that
cannot be usefully analysed with an OLS multiple regression on cross-section data.
23. The proportion of the population living on US$ 2 a day would be an appropriate variable
to use in the regression analysis except for missing data. That measure was not available
for thirty-two out of the ninety-one countries in the data set.
Poverty, Endemic Disease and HIV
215
Table 5. Means of Variables in Low- and Middle-income Countries by Region
Sub-Saharan Africa
Variable
Chlamydia DALYs per 100,000
Gonorrhoea DALYs per 100,000
Syphilis DALYs per 100,000
Schistosomiasis DALYs per 100,000
Trachoma DALYs per 100,000
Combined cofactors (STIs and
schistosomiasis)
Malaria prevalence, % of population
HIV prevalence, % of adult population
Gini coefficient
Income per capita PPP in US$
Urbanization, per cent of population
Age of epidemic in years
Adult literacy, per cent of population
Female economic participation ratio
Muslim, per cent of population
Remittances as per cent of GDP
Autodisable syringe, per cent did not use
Other Low- and
Middle-income
Countries
Southern
Central
Other subSaharan Africa
120
156
120
238
149
638
119
162
285
236
149
809
119
168
521
277
187
1095
56
42
29
14
10
145
12.8
18.3
54.1
4391
33.8
21.6
73.2
.721
9.8
5.9
.556
32.4
4.3
42.3
1729
39.0
21.2
69.2
.833
16.0
2.5
.222
32.2
1.72
40.2
1348
33.0
20.2
36.8
.715
58.9
3.1
0.0
0.9
0.6
41.5
5139
51.9
19.6
84.6
.649
24.8
5.9
41.7
a
a
Angola, Gabon, Nigeria, Cameroon, Congo, Dem. Rep. of the Congo, Central African Republic, Rwanda,
Burundi, Kenya, Uganda and Tanzania.
low- and middle-income countries. Consequently, the cofactor infections
analysed in this study (either diseases of poverty or diseases less likely to be
treated because of poverty) are found at roughly similar levels throughout
most of sub-Saharan Africa, but at far lower levels outside Africa. STIs are
epidemic around the world, but their burden in sub-Saharan Africa is more
than triple the level in other low- and middle-income countries. Moreover,
the difference for the other cofactor diseases is far greater. The burden
of schistosomiasis, for example, is between 18 and 21 times as great in
sub-Saharan Africa as it is in other low- and middle-income countries.
Recall that S. hematobium, which affects the genital tract by creating lesions,
inflammation and viral shedding in ways likely to promote HIV transmission,
is found almost exclusively in Africa. To the extent that poverty and the
diseases that poverty promotes lead to high HIV prevalence, then southern
Africa is similar to the rest of sub-Saharan Africa and much more vulnerable
than other low- and middle-income countries.
Other factors that could promote the spread of HIV set southern Africa
apart from the rest of sub-Saharan Africa, but make it more similar to
other developing and transition countries. A number of authors have noted
the positive correlation of educational attainment and literacy with HIV
(for example, Brent, 2006). Adult literacy (and the ratio of female to male
literacy, which is not shown in Table 5) in southern Africa is a little lower
than in low- and middle-income countries outside of sub-Saharan Africa
216
Larry Sawers and Eileen Stillwaggon
and a little higher than in central Africa, but is much higher than in the rest
of sub-Saharan Africa. In addition, the level of migration (as measured by
remittances as a per cent of GDP) is the same in southern Africa as it is in
low- and middle-income countries outside Africa and more than twice the
level in the remainder of sub-Saharan Africa.
Income inequality puts southern Africa in a category by itself. Gini coefficients in most of southern Africa are among the highest in the world, but in
the rest of sub-Saharan Africa, Gini coefficients average about the same as in
low- and middle-income countries elsewhere. Furthermore, southern Africa
on average has the lowest percentage of the population who are Muslim
among the four groups of low- and middle-income countries in Table 5.
In sum, why has HIV reached its highest levels in southern Africa —
one-quarter of the adult population in the most gravely affected countries?
Perhaps what is unique about southern Africa is that it faces higher risk
from so many of the factors that promote HIV epidemics. With regard to
the level of poverty and the burden of cofactor infections, it is more like
the rest of sub-Saharan Africa than countries elsewhere. With regard to
literacy and migration, southern Africa is more like low- and middle-income
countries outside of Africa than those within Africa. With regard to high
levels of income inequality and low proportion of Muslims, southern Africa
sets itself apart from the rest of the world. All of these factors come together
in southern Africa to produce the most intense epidemics of HIV in the
world.
CONCLUSIONS
AIDS prevention policy has failed to stem the spread of HIV in southern
Africa and neighbouring countries. The research presented here suggests
that policy makers have targeted too few factors in their attempts to slow
or reverse the epidemics. The socio-economic variables in our basic model
that are statistically associated with HIV prevalence, however, are not easily
affected by changes in policy. Levels of inequality have been shown to be
highly resistant in the short run to changes in the policy regime. Even the
most successful efforts to promote economic development produce singledigit annual increases in per capita income. Migration patterns, literacy or
female participation in the economy are slow to change. Moreover, after
decades of trying to convince people to avoid risky sexual behaviour, there
appears to be only scattered success.
In contrast, the variables correlated with HIV prevalence in this study that
are the cheapest and easiest to change are the cofactor infections. Dramatic
reductions in all of the cofactor infections can be produced rapidly and at
low cost. Some of the infections can be cured by medicines that cost pennies
per patient and can be distributed by schoolteachers or health care providers
with little training. They are highly effective, can be easily tolerated and
Poverty, Endemic Disease and HIV
217
are heat-stable with long shelf lives. Arguably, the easiest and fastest way
to promote economic development, to alleviate poverty and even to slow or
reverse the spread of HIV is to attack the cofactor infections analysed in this
study. The present research provides support for integrating HIV prevention
policy with efforts to address the multitude of health issues that beset Africa.
Treating or preventing diseases endemic in developing countries with
antihelminthics, antibiotics and other interventions is a pragmatic, safe, ethically sound and relatively inexpensive strategy. Such interventions have
substantial beneficial outcomes on their own, in better health, better school
performance and higher productivity at work. Moreover, the improvement
in overall well-being in itself promotes healthier behaviours in healthier
people. Some HIV treatment protocols now include food for people taking
antiretrovirals and treatment for tuberculosis because failing to do so wastes
lives and resources. Nutritional supplementation for HIV-infected individuals may also reduce viral loads (making them less contagious), postpone
the onset of AIDS, prevent opportunistic infections and prolong the efficacy
of antiretrovirals. Similarly, our research supports the arguments of medical
researchers that redirecting some of the funds designated for HIV prevention
to deworming, sanitation, STI treatment, mosquito control and safe water
supports, rather than undermines, HIV prevention, care and treatment. Treating locally endemic infections and the conditions that cause those infections
in order to control HIV is not mission creep; rather it may be essential to the
central mission of AIDS programming.
Appendix. Countries in Dataset
Albania
Algeria
Armenia
Azerbaijan
Bangladesh
Belarus
Benin
Bolivia
Botswana
Brazil
Bulgaria
Burkina Faso
Burundi
Cambodia
Cameroon
Chile
China
Colombia
Congo
Costa Rica
Cote d’Ivoire
Croatia
Dominican Republic
Ecuador
Egypt
El Salvador
Ethiopia
Gabon
Gambia
Georgia
Ghana
Guatemala
Guinea
Guinea-Bissau
Guyana
Haiti
Honduras
India
Indonesia
Iran
Jamaica
Jordan
Kazakhstan
Kenya
Kyrgyzstan
Laos
Latvia
Lesotho
Lithuania
Madagascar
Malawi
Malaysia
Mali
Mauritania
Mexico
Moldova
Mongolia
Morocco
Mozambique
Namibia
Nepal
Nicaragua
Niger
Nigeria
Pakistan
Panama
Papua New Guinea
Paraguay
Peru
Philippines
Poland
Romania
Russian Federation
Rwanda
Senegal
Sierra Leone
South Africa
Sri Lanka
Swaziland
Tajikistan
Tanzania
Thailand
Togo
Trinidad and Tobago
Tunisia
Turkey
Uganda
Ukraine
Viet Nam
Yemen
Zambia
218
Larry Sawers and Eileen Stillwaggon
REFERENCES
Abu-Raddad, L., P. Patnaik and J.G. Kublin (2006) ‘Dual Infection with HIV and Malaria Fuels
the Spread of Both Diseases in Sub-Saharan Africa’, Science 314: 1603–06.
Adimora, A.A., V.J. Schoenbach and I.A. Doherty (2007) ‘Concurrent Sexual Partnerships
among Men in the United States’, American Journal of Public Health 97(12): 2230–
37.
Adimora, A.A. et al. (2002) ‘Concurrent Sexual Partnerships among Women in the United
States’, Epidemiology 13(3): 320–27.
Attili, V.R., S. Hira and M.K. Dube (1983) ‘Schistosomal Genital Granulomas: A Report of 10
Cases’, British Journal of Venereal Disease 59: 269–72.
Bentwich, Z., C.L. Teicher and G. Borkow (2008) ‘The Helminth HIV Connection: Time to
Act’, AIDS 22(13): 1611–14.
Bloland, P. et al. (1995) ‘Maternal HIV Infection and Infant Mortality in Malawi: Evidence for
Increased Mortality Due to Placental Malaria Infection’, AIDS 9(7): 721–6.
Boily, M.C. et al. (2009) ‘Heterosexual Risk of HIV-1 Infection per Sexual Act: Systematic
Review and Meta-analysis of Observational Studies’, Lancet Infectious Diseases 9(2): 118–
29.
Bonnel, R. (2000) ‘HIV/AIDS and Economic Growth: A Global Perspective’, South African
Journal of Economics 68(5): 820–55.
Booysen, F.l.R. (2002) ‘Provincial Inequalities in Maternal and Child Health Care in South
Africa’, Acta Academica 34(3): 53–80.
Booysen, F.l.R. (2003) ‘Urban-rural Inequalities in Health Care Delivery in South Africa’,
Development Southern Africa 20(5): 659–74.
Borkow, G. and Z. Bentwich (2006) ‘HIV and Helminth Coinfection: Is Deworming Necessary?’,
Parasite Immunology 28: 605–12.
Boulton, I. and S. Gray-Owen (2002) ‘Neisserial Binding to CEACAM1 Arrests the Activation
and Proliferation of CD4+ T Lymphocytes’, Nature Immunology 3: 229–36.
Brenner, B.G. et al. (2007) ‘High Rates of Forward Transmission Events after Acute/Early
HIV-1 Infection’, Journal of Infectious Diseases 195(7): 951–9.
Brent, R.J. (2006) ‘Does Female Education Prevent the Spread of HIV-AIDS in Sub-Saharan
Africa?’, Applied Economics 38: 491–503.
Brown, M. et al. (2005) ‘Treatment of Schistosoma Mansoni Infection Increases Helminthspecific Type 2 Cytokine Responses and HIV-1 Loads in Coinfected Ugandan Adults’,
Journal of Infectious Diseases 191(10): 1648–57.
Brummer, D. (2002) ‘Labour Migration and HIV/AIDS in Southern Africa’. Pretoria: International Organization for Migration, Regional Office for Southern Africa.
Chan, D. (2005) ‘Factors Affecting Sexual Transmission of HIV-1: Current Evidence and Implications for Prevention’, Current HIV Research 3: 223–41.
Chenine, A.L. et al. (2008) ‘Acute Schistosoma Mansoni Infection Increases Susceptibility to
Systemic SHIV Clade C Infection in Rhesus Macaques after Mucosal Virus Exposure’, PLoS
Neglected Tropical Diseases 2(7): e265.
Clark, S.J., M.A. Collinson, K. Kahn, K. Drullinger and S. Tollman (2007) ‘Returning Home
to Die: Circular Labour Migration and Mortality in South Africa’, Scandinavian Journal of
Public Health 35(Supplement 69): 35–44.
Cohen, M.S. and C.D. Pilcher (2005) ‘Amplified HIV Transmission and New Approaches to
HIV Prevention’, Journal of Infectious Diseases 191: 1391–3.
Corbett, E.L. et al. (2002) ‘HIV-1/AIDS and the Control of Other Infectious Diseases in Africa’,
Lancet 359(9324): 2177–87.
Coutsoudis, A., K. Pillay, E. Spooner, L. Kuhn and H.M. Coovadia (1999) ‘Randomized Trial
Testing the Effect of Vitamin A Supplementation on Pregnancy Outcomes and Early Motherto-Child HIV-1 Transmission in Durban, South Africa. South African Vitamin A Study
Group’, AIDS 13(12): 1517–24.
Poverty, Endemic Disease and HIV
219
Crush, J., V. Williams and S. Peberdy (2005) ‘Migration in Southern Africa’. Geneva: Policy
Analysis and Research Programme of the Global Commission on International Migration.
Deuchert, E. and S. Brody (2006) ‘The Role of Health Care in the Spread of HIV/AIDS
in Africa: Evidence from Kenya’, International Journal of STD & AIDS 17(11): 749–
52.
Deuchert, E. and S. Brody (2007a) ‘Lack of Autodisable Syringes Use and Health Care Indicators
Are Associated with High HIV Prevalence: An International Ecologic Analysis’, Annals of
Epidemiology 17(3): 199–207.
Deuchert, E. and S. Brody (2007b) ‘Plausible and Implausible Parameters for Mathematical
Modeling of Nominal Heterosexual HIV Transmission’, Annals of Epidemiology 17(3):
237–44.
Doherty, I.A., A. Shiboski, J.M. Ellen, A.A. Adimora and N.S. Padian (2006) ‘Sexual Bridging
Socially and Over Time: A Simulation Model Exploring the Relative Effects of Mixing and
Concurrency on Viral Sexually Transmitted Infection Transmission’, Sexually Transmitted
Diseases 33(6): 368–73.
Drain, P.K., J.S. Smith, J.P. Hughes, D.T. Halperin and K.K. Holmes (2004) ‘Correlates of
National HIV Seroprevalence: An Ecologic Analysis of 122 Developing Countries’, Journal
of Acquired Immune Deficiency Syndrome 35(4): 407–20.
Elliott, A.M. et al. (2003) ‘Associations between Helminth Infection and CD4+ T Cell Count,
Viral Load and Cytokine Responses in HIV-1-infected Ugandan Adults’, Transactions of the
Royal Society of Tropical Medicine and Hygiene 97(1): 103–8.
Epstein, H. (2007) The Invisible Cure: Africa, the West, and the Fight Against AIDS. New York:
Farrar, Straus and Giroux.
Fawzi, W.W. and D.J. Hunter (1998) ‘Vitamins in HIV Disease Progression and Vertical Transmission’, Epidemiology 9(4): 457–66.
Fawzi, W. et al. (2001) ‘Predictors of Intrauterine and Intrapartum Transmission of HIV-1 among
Tanzanian Women’, AIDS 15(9): 1157–65.
Feldmeier, H., G. Poggensee, I. Krantz and G. Helling-Giese (1995) ‘Female Genital Schistosomiasis. New Challenges from a Gender Perspective’, Tropical and Geographical Medicine
47(2 Suppl): S2–15.
Fleming, D. and J. Wasserheit (1999) ‘From Epidemiological Synergy to Public Health Policy
and Practice: The Contribution of Other Sexually Transmitted Diseases to Sexual Transmission of HIV Infection’, Sexually Transmitted Infections 75: 3–17.
Friis, H. and K.F. Michaelsen (1998) ‘Micronutrients and HIV Infections: A Review’, European
Journal of Clinical Nutrition 52: 157–63.
Ghani, A.C. and G.P. Garnett (2000) ‘Risks of Acquiring and Transmitting Sexually Transmitted
Diseases in Sexual Partner Networks’, Sexually Transmitted Diseases 27(10): 579–87.
Gillespie, S. and K. Suneetha (2005) HIV /AIDS and Food and Nutrition Security: From Evidence
to Action. Washington, DC: International Food Policy Research Institute.
Gillespie, S., S. Kadiyala and R. Greener (2007) ‘Is Poverty or Wealth Driving HIV Transmission?’, AIDS 21 Suppl 7: S5–16.
Gillespie, S., R. Greener, A. Whiteside and J. Whitworth (2007) ‘Investigating the Empirical Evidence for Understanding Vulnerability and the Associations between Poverty, HIV Infection
and AIDS Impact’, AIDS 21 Suppl 7: S1–4.
Gisselquist, D. (2008) Points to Consider: Responses to HIV /AIDS in Africa, Asia, and the
Carribean. London: Adonis and Abbey.
Greener, R. (2004) ‘The Impact of HIV/AIDS on Poverty and Inequality’, in M. Haacker (ed.)
The Macroeconomics of HIV /AIDS, pp. 167–81. Washington, DC: International Monetary
Fund.
Grosskurth, H. et al. (1995) ‘Impact of Improved Treatment of Sexually Transmitted Diseases on
HIV Infection in Rural Tanzania: Randomised Controlled Trial’, Lancet 346(8974): 530–36.
Halperin, D.T. and H. Epstein (2004) ‘Concurrent Sexual Partnerships Help to Explain Africa’s
High HIV Prevalence: Implications for Prevention’, Lancet 364(9428): 4–6.
220
Larry Sawers and Eileen Stillwaggon
Halperin, D.T. and H. Epstein (2007) ‘Why is HIV Prevalence so Severe in Southern Africa?’,
Southern African Journal of HIV Medicine (March): 19–25.
Hargreaves, J.R. et al. (2007) ‘Explaining Continued High HIV Prevalence in South Africa:
Socioeconomic Factors, HIV Incidence and Sexual Behaviour Change among a Rural Cohort,
2001–2004’, AIDS 21 Suppl 7: S39–48.
Hayes, R.J., K.F. Schulz and F.A. Plummer (1995) ‘The Cofactor Effect of Genital Ulcers on
the Per-exposure Risk of HIV Transmission in Sub-Saharan Africa’, Journal of Tropical
Medicine and Hygiene 98(1): 1–8.
Helleringer, S., H-P. Kohler and L. Kalilani-Phiri (2009) ‘The Association of HIV Serodiscordance and Partnership Concurrency in Likoma Island (Malawi)’, AIDS 23: 1285–90.
Herrero, M.D., P. Rivas, N.I. Rallon, G. Ramirez-Olivencia and S. Puente (2007) ‘HIV and
Malaria’, AIDS Review 9(2): 88–98.
Hoffman, I.F. et al. (1999) ‘The Effect of plasmodium falciparum Malaria on HIV-1 RNA Blood
Plasma Concentration’, AIDS 13(4): 487–94.
Horwitz, S. (2001) ‘Migrancy and HIV/AIDS: A Historical Perspective’, South African Historical Journal 45(November): 103–23.
Hunter, M. (2002) ‘The Materiality of Everyday Sex: Thinking beyond “Prostitution”’, African
Studies 61(1): 99–120.
Hunter, M. (2007) ‘The Changing Political Economy of Sex in South Africa: The Significance
of Unemployment and Inequalities to the Scale of the AIDS Pandemic’, Social Science &
Medicine 64: 689–700.
ILO (1998) ‘Labour Migration to South Africa in the 1990s’. No 4. Harare: International Labour
Office, Southern Africa Multidisciplinary Advisory Team.
Jewkes, R.K. et al. (2006) ‘Factors Associated with HIV Sero-Positivity in Young, Rural South
African Men’, International Journal of Epidemiology 35:1455–60.
John, G.C. et al. (1997) ‘Genital Shedding of Human Immunodeficiency Virus Type 1 DNA
during Pregnancy: Association with Immunosuppression, Abnormal Cervical or Vaginal
Discharge, and Severe Vitamin A Deficiency’, Journal of Infectious Diseases 175(1): 57–62.
John-Stewart, G.C. et al. (2005) ‘Subtype C Is Associated with Increased Vaginal Shedding of
HIV-1’, Journal of Infectious Diseases 192(3): 492–6.
Kallestrup, P. et al. (2005) ‘Schistosomiasis and HIV-1 Infection in Rural Zimbabwe: Effect of
Treatment of Schistosomiasis on CD4 Cell Count and Plasma HIV-1 RNA Load’, Journal
of Infectious Diseases 192(11): 1956–61.
Kanyenze, G. (2004) ‘African Migrant Labour Situation in Southern Africa’. Paper presented at
the ICFTU-AFRO Conference on Migrant Labour, Nairobi, Kenya (March).
Kjetland, E.F. et al. (2006) ‘Association between Genital Schistosomiasis and HIV in Rural
Zimbabwean Women’, AIDS 20(4): 593–600.
Kretzschmar, M. and M. Morris (1996) ‘Measures of Concurrency in Networks and the Spread
of Infectious Disease’, Mathematical Biosciences 133: 165–95.
Kretzschmar, M., R. White and M Caraël (2009) ‘Concurrency is More Complex than it Seems’,
AIDS 23: 1–3.
Kublin, J.G. et al. (2005) ‘Effect of Plasmodium falciparum Malaria on Concentration of HIV1-RNA in the Blood of Adults in Rural Malawi: A Prospective Cohort Study’, Lancet
365(9455): 233–40.
Laga, M. et al. (1993) ‘Non-ulcerative Sexually Transmitted Diseases as Risk Factors for HIV-1
Transmission in Women: Results from a Cohort Study’, AIDS 7(1): 95–102.
Lagarde, E. et al. (2001) ‘Concurrent Sexual Partnerships and HIV Prevalence in Five Urban
Communities of Sub-Saharan Africa’, AIDS 15(7): 877–84.
Landers, D. (1996) ‘Nutrition and Immune Function II: Maternal Factors Influencing Transmission’, Journal of Nutrition 126: 2637S–40S.
Laurent, C., M. Peeters and E. Delaporte (2007) ‘HIV/AIDS Infection in the World with a
Special Focus on Africa’, in M. Tabayrenc (ed.) Encyclopedia of Infectious Diseases: Modern
Methodologies, pp. 45–56. Hoboken, NJ: John Wiley & Sons.
Poverty, Endemic Disease and HIV
221
Lawn, S.D. et al. (2000) ‘The Effect of Treatment of Schistosomiasis on Blood Plasma HIV-1
RNA Concentration in Coinfected Individuals’, AIDS 14(16): 2437–43.
Leutscher, P. et al. (1998) ‘Clinical Findings in Female Genital Schistosomiasis in Madagascar’,
Tropical Medicine and International Health 3(4): 327–32.
Lopman, B. et al. (2007) ‘HIV Incidence and Poverty in Manicaland, Zimbabwe: Is HIV Becoming a Disease of the Poor?’, AIDS 21 Suppl 7: S57–66.
Lurie, M.N. (2004) ‘The Epidemiology of Migration and AIDS in South Africa’. Working Paper.
Oxford: University of Oxford, Centre on Migration, Policy and Society.
Lurie, M.N. and S. Rosenthal (2009) ‘Concurrent Partnerships as a Driver of the HIV Epidemic
in Sub-Saharan Africa? The Evidence is Limited’, AIDS and Behavior 14(1). Published
online 2 June 2009. DOI 10.1007/s10461-009-9583-5.
Lurie, M., S. Rosenthal and B. Williams (2009) ‘Concurrency Driving the African HIV Epidemics: Where is the Evidence?’, Lancet 374(9699): 1420.
Lurie, M.N. et al. (2002) ‘The Impact of Migration on HIV-1 Transmission in South Africa’.
Working Paper. Durban: South African Medical Research Council, HIV Prevention and
Vaccine Research Unit.
Mah, T.L. and D.T. Halperin (2008) ‘Concurrent Sexual Partnerships and the HIV Epidemics in
Africa: Evidence to Move Forward’, AIDS Behavior 14(1). Published online 22 July 2008.
DOI 10.1007/s10461-008-9433-x.
Mah, T.L. and D.T. Halperin (2009) ‘The Evidence for the Role of Concurrent Partnerships in
Africa’s HIV Epidemics: A Response to Lurie and Rosenthal’. AIDS and Behavior 14(1).
Published online 9 October 2009. DOI 10.1007/s10461-009-9617-z.
Mahal, A. (2002) ‘The Human Development Roots of HIV and Implications for Policy: A
Cross-country Analysis’, Journal of Health and Population in Developing Countries 4:
43–60.
Marble, M. and K. Key (1995) ‘Clinical Facets of a Disease Neglected Too Long’, AIDS Weekly
Plus (7 August): 16–19.
Marks, S. (2002) ‘An Epidemic Waiting to Happen? The Spread of HIV/AIDS in South Africa
in Social and Historical Perspective’, African Studies 61(1): 13–26.
Mattson, C., R. Bailey, K. Argot, J. Ndinya-Achola and S. Moses (2007) ‘A Nested Case-Control
Study of Sexual Practices and Risk Factors for Prevalent HIV-1 Infection among Young Men
in Kisumu, Kenya’, Sexually Transmitted Diseases 34(10): 731–36.
Mehta, S. and W. Fawzi (2007) ‘Effect of Vitamins, Including Vitamin A, on HIV/AIDS
Patients’, Vitamins and Hormones 75: 355–83.
Mishra, V. and S. Bignami-Van Assche (2009) ‘Concurrent Sexual Partnerships and HIV Infection: Evidence from National Population-Based Surveys’. Working Paper. Geneva: USAID.
Mishra, V. et al. (2007) ‘HIV Infection Does Not Disproportionately Affect the Poorer in SubSaharan Africa’, AIDS 21 Suppl 7: S17–28.
Montresor, A. et al. (2001) ‘Extending Anthelminthic Coverage to Non-enrolled School-Age
Children Using a Simple and Low-Cost Method’, Tropical Medicine and International Health
6(7): 535–7.
Morison, L. et al. (2001) ‘HIV-1 Subtypes and the HIV Epidemics in Four Cities in Sub-Saharan
Africa’, AIDS 15 Suppl 4: S109–116.
Morris, M. and M. Kretzschmar (1997) ‘Concurrent Partnerships and the Spread of HIV’, AIDS
11: 641–8.
Morris, M. and M. Kretzschmar (2000) ‘A Microsimulation Study of the Effects of Concurrent
Partnerships on the Spread of HIV in Uganda’, Mathematical Population Studies 8(2): 109–
33.
Mwinzi, P.N., D.M. Karanja, D.G. Colley, A.S. Orago and W.E. Secor (2001) ‘Cellular Immune
Responses of Schistosomiasis Patients Are Altered by Human Immunodeficiency Virus
Type 1 Coinfection’, Journal of Infectious Diseases 184(4): 488–96.
Nacher, M. (2002) ‘Worms and Malaria: Noisy Nuisances and Silent Benefits’, Parasite Immunology 24: 391–3.
222
Larry Sawers and Eileen Stillwaggon
Oppenheimer, E., M. Bunnag and A. Stern (1998) ‘HIV/AIDS and Cross-border Migration:
A Rapid Assessment of Migrant Populations along the Thai-Burma (Myanmar) Border
Regions’. Bankok: Asian Center for Migration, Chulalongkorn University.
Over, M. (1998) ‘The Effects of Societal Variables on Urban Rates of HIV Infection in Developing Countries: An Exploratory Analysis’, in M. Ainsworth, L. Fransen and M. Over
(eds) Confronting AIDS: The Evidence from the Developing World, pp. 39–51. Luxembourg:
Office for the Official Publications of the European Communities.
Pilcher, C.D. et al. (2004) ‘Brief but Efficient: Acute Infection and the Sexual Transmission of
HIV’, Journal of Infectious Diseases 189: 1785–92.
Ping, L.H. et al. (1999) ‘Characterization of V3 Sequence Heterogeneity in Subtype C Human
Immunodeficiency Virus Type 1 Isolates from Malawi: Underrepresentation of X4 Variants’,
Journal of Virology 73(8): 6271–81.
Posel, D. (2003) ‘Have Migration Patterns in post-Apartheid South Africa Changed?’. Paper
presented at the African Migration in Comparative Perspective, Johannesburg, South Africa
(4–7 June).
Posel, D. and D. Casale (2003) ‘What Has Been Happening to Internal Labour Migration in
South Africa, 1993–1999?’, South African Journal of Economics 71(3): 455–79.
Potterat, J. (2009) ‘AIDS Epidemiology in Africa: A Changing of the Guard’, International
Journal of STD and AIDS 20: 813–15.
PPC (2002) ‘Rome Meeting Notes’. Geneva: Partnership for Parasite Control. http://www.who.
int/wormcontrol/abo_utus/en/mtgnotes/apri12002.pdf (accessed May 2002).
Quinn, T.C. et al. (2000) ‘Viral Load and Heterosexual Transmission of Human Immunodeficiency Virus Type 1’, New England Journal of Medicine 342(13): 921–9.
Rainwater, S. et al. (2005) ‘No Evidence for Rapid Subtype C Spread within an Epidemic in
which Multiple Subtypes and Intersubtype Recombinants Circulate’, AIDS Research and
Human Retroviruses 21(12): 1060–65.
Rotchford, K., A.W. Strum and D. Wilkinson (2000) ‘Effect of Coinfection with STDs and of
STD Treatment on HIV Shedding in Genital-tract Secretions: Systematic Review and Data
Synthesis’, Sexually Transmitted Diseases 27(5): 243–8.
Sawers, L., E. Stillwaggon and T. Hertz (2008) ‘Cofactor Infections and HIV Epidemics in
Developing Countries: Implications for Treatment’, AIDS Care 20(4): 488–94.
Semba, R. et al. (1994) ‘Maternal Vitamin A Deficiency and Mother-to-Child Transmission of
HIV-1’, Lancet 343: 1593–7.
Skeldon, R. (2002) ‘Population mobility and HIV Vulnerability in South East Asia: An Assessment and Analysis’. Bangkok: United Nations Development Programme.
Spira, S., M.A.Wainberg, H. Loemba, D. Turner and B.G. Brenner (2003) ‘Impact of Clade
Diversity on HIV-1 Virulence, Antiretroviral Drug Sensitivity and Drug Resistance’, Journal
of Antimicrobial Chemotherapy 51(2): 229–40.
Stillwaggon, E. (2000) ‘HIV Transmission in Latin America: Comparisons with Africa and
Policy Implications’, South African Journal of Economics 68(5): 985–1011.
Stillwaggon, E. (2002) ‘HIV/AIDS in Africa: Fertile Terrain’, Journal of Development Studies
38(6): 1–22.
Stillwaggon, E. (2003) ‘Racial Metaphors: Interpreting Sex and AIDS in Africa’, Development
and Change 34(5): 809–32.
Stillwaggon, E. (2006) AIDS and the Ecology of Poverty. New York: Oxford University
Press.
Stillwaggon, E. and L. Sawers (2010) ‘Twenty Years is Enough: Sexual Behavior and HIV in
Africa’. Working Paper. Washington, DC: American University.
Sturm, A.W., D. Wilkinson, N. Ndovela, S. Bowen and C. Connolly (1998) ‘Pregnant Women as
a Reservoir of Undetected Sexually Transmitted Diseases in Rural South Africa: Implications
for Disease Control’, American Journal of Public Health 88(8): 1243–5.
Tapia, N. et al. (2003) ‘Influence of Human Immunodeficiency Virus Type 1 Subtype on Motherto-Child Transmission’, Journal General Virology 84(Pt 3): 607–13.
Poverty, Endemic Disease and HIV
223
Tebit, D.M., I. Nankya, E.J. Arts and Y. Gao (2007) ‘HIV Diversity, Recombination and
Disease Progression: How Does Fitness “Fit” into the Puzzle?’, AIDS Review 9(2):
75–87.
Thahane, T.T. (1991) ‘International Labor Migration in Southern Africa’, in D.G. Papademetriou
and P.L. Martin (eds) The Unsettled Relationship: Labor Migration and Economic Development, pp. 65–88. New York: Greenwood Press.
Tsafack, C. (2006) ‘Gender Inequality and HIV/AIDS Epidemic in Sub-Saharan Africa’. Working Paper. Auvergne: CERDI-CNRS, Université d’Auvergne.
Tsafack, C. (2008) ‘What Drives HIV/AIDS Epidemic in Sub-Saharan Africa’. Working Paper.
Auvergne: CERDI-CNRS, Université d’Auvergne.
Tsafack, C. and L. Bassolé (2006) ‘Income Inequality and HIV/AIDS Epidemic in Sub-Saharan
Africa’. Working Paper. Auvergne: CERDI-CNRS, Université d’Auvergne.
UNAIDS (2008) 2008 Report on the Global AIDS Epidemic. Geneva: UNAIDS.
Walson, J., B. Herrin and G. John-Stewart (2009) ‘Deworming Helminth and HIV Coinfected Individuals to Delay Disease Progression: A Systematic Review and Meta Analysis’, Cochrane Database of Systematic Reviews 3. Published online 8 July 2009. DOI:
10.1002/14651858.CD006419.pub3.
Walson, J.L. et al. (2008) ‘Albendazole Treatment of HIV-1 and Helminth Co-infection: A
Randomized, Double-blind, Placebo-controlled Trial’, AIDS 22(13): 1601–09.
Wang, C.C. et al. (2001) ‘The Effect of Treatment of Vaginal Infections on Shedding of Human
Immunodeficiency Virus Type 1’, Journal of Infectious Diseases 183(7): 1017–22.
Webster, L., J. Greenspan, A. Nakashima and R. Johnson (1993) ‘An Evaluation of Surveillance
for Chlamydia trachomatis Infections in the United States, 1987–1991’, Morbidity and
Mortality Weekly 42(SS-3): 21–7.
Wellings, K. et al. (2006) ‘Sexual Behaviour in Context: A Global Perspective’, Lancet
368(9548): 1706–28.
Whitworth, J. et al. (2000) ‘Effect of HIV-1 and Increasing Immunosuppression on Malaria
Parasitaemia and Clinical Episodes in Adults in Rural Uganda: A Cohort Study’, Lancet
356(9235): 1051–6.
WHO (2004) Global Burden of Disease Estimates, 2004. Geneva: World Health Organization.
http://www.who.int/healthinfo/global_burden_disease/2004_report_update/en/index.html
WHO (2005) World Health Report 2005. Geneva: World Health Organization. http://www.
who.int/whr/2005/en/
WHO (2007) ‘Report of the Global Partners’ Meeting on Neglected Tropical Diseases: 2007:
A Turning Point’. Geneva: World Health Organization, Department of Control of Neglected
Tropical Diseases. http://whqlibdoc.who.int/hq/2007/WHO_CDS_NTD_2007.4_eng.pdf
(accessed 10 February 2010).
WHO (2008a) ‘Current Estimated Total Number of Individuals with Morbidity and Mortality
Due to Schistosomiasis Haematobium and S. Mansoni Infection in Sub-Saharan Africa’.
Geneva: World Health Organization. http://www.who.int/schistosomiasis/epidemiology/
table/en/index.html (accessed 8 November 2008).
WHO (2008b) ‘World Malaria Report’. Geneva: World Health Organization. www.who.int/
malaria/wmr2008/malaria2008.pd (accessed 23 September 2008).
WHO (2009) ‘PCT Databank’. Geneva: World Health Organization. www.who.int/neglected_
diseases/en/ (accessed June 2009).
Wolday, D. et al. (2002) ‘Treatment of Intestinal Worms is Associated with Decreased HIV
Plasma Viral Load’, Journal of Acquired Immune Deficiency Syndromes 31: 56–62.
Wolday, D. et al. (2004) ‘The Impact of Syndromic Treatment of Sexually Transmitted Diseases
on Genital Shedding of HIV-1’, AIDS 18(5): 781–5.
Woodward, B. (1998) ‘Protein, Calories, and Immune Defenses’, Nutrition Reviews 56(1, Part
2): S84–92.
World Bank (1997) Confronting AIDS: Public Priorities in a Global Epidemic. Oxford: Oxford
University Press.
224
Larry Sawers and Eileen Stillwaggon
Xiao, L., S.M. Owen, D.L. Rudolph, R.B. Lal and A.A. Lal (1998) ‘Plasmodium falciparum
Antigen-Induced Human Immunodeficiency Virus Type 1 Replication Is Mediated through
Induction of Tumor Necrosis Factor-alpha’, Journal of Infectious Diseases 177(2): 437–45.
Zuma, K., W. Gouws, B. Williams and M. Lurie (2003) ‘Risk Factors for HIV Infection among
Women in Carletonville, South Africa: Migration, Demography and Sexually Transmitted
Diseases’, International Journal of STD & AIDS 14(12): 814–7.
Larry Sawers is Professor of Economics at American University in Washington, DC and is the corresponding author; he can be contacted at
[email protected]. His research interests are in the field of economic
development and he has published books and articles on Argentina, Ecuador,
Tanzania and Lithuania. His research in recent years has focused on the HIV
epidemics in Africa.
Eileen Stillwaggon is Professor of Economics and Harold G. Evans Eisenhower Professor at Gettysburg College in Pennsylvania. She has written a
number of scholarly articles and books on poverty, health and development
and on HIV epidemics in poor populations.