Secondary Logo

Journal Logo

Demographic and socioeconomic impact of AIDS: taking stock of the empirical evidence

Zaba, Basiaa; Whiteside, Alanb; Boerma, J Tiesc

Original articles

From the aLondon School of Hygiene and Tropical Medicine, London, UK; bUniversity of KwaZulu-Natal, Durban, South Africa; and cWorld Health Organization, Geneva, Switzerland

Correspondence to: Basia Zaba, Centre for Population Studies, London School of Hygiene and Tropical Medicine, 49-51 Bedford Square, London WC1B 3DP, UK.

Back to Top | Article Outline


Six years have passed since AIDS published a supplement on the demographic impact of AIDS [1]. That volume reviewed the evidence of the impact of AIDS on mortality and fertility in the early and mid-1990s. Community studies showed increases in adult and child mortality and a fertility reduction among HIV-infected individuals; demographic and health surveys and national censuses showed substantial increases in adult mortality in several countries, and models showed the large effects of AIDS on mortality and moderate impacts on population growth.

Since 1998, a large amount of new evidence has become available from specialist community-based studies, national surveys and censuses. Various analytical techniques have been developed that allow us to use these data in new ways. In addition to demographers and epidemiologists, social scientists and economists are now starting to look at empirical evidence on the effects of HIV, at the family, community and national level. Although modelling was the primary basis for an assessment of the size and future of the epidemic in the past, increasingly use is made of real data to inform policy and to evaluate national responses to the epidemic. In this context it is timely and appropriate to re-visit this problem area, to summarize our wider knowledge of the demographic and social impacts of the epidemic, and to broaden the range of effects that we scrutinize critically to include macro and micro-economic impacts. These impacts of AIDS are having an increased impact on policy making: the health services are stretched; the education system is losing teachers; and numbers of orphans are growing. Policy needs to be advised by good data and rigorous analysis. A scientific meeting on the Demographic and Socioeconomic Impact of AIDS was held in Durban, South Africa, on 26–28 March 2003, and forms the basis for the reviews in this supplement.

Most evidence of the impact of the epidemic presented here relates to sub-Saharan Africa. With an estimated 26.6 million HIV infections out of the global total estimate of 40 million by end-2003, and with an adult HIV prevalence at least 10 times higher than in most other parts of the world [2], it is clear that the magnitude of the impact in sub-Saharan Africa on the population and economies is bound to be of a different scale. Also within Africa there are huge and widening differences in the spread of HIV, which implies that the consequences of the epidemic will differ substantially, irrespective of differences in non-AIDS conditions [3]. HIV prevalence among pregnant women in southern Africa is as much as five times higher than in western Africa and three times higher than in eastern Africa.

Back to Top | Article Outline

Adult mortality

The most dramatic impact of the HIV epidemic is on adult mortality. In the worst affected countries of eastern and southern Africa, the probability of a 15-year old dying before reaching the age of 60 years has risen dramatically, from a range of 10 to 30% in the mid-1980s, to a range of 30–60% at the turn of the century [4]. Two of the review papers in this volume focus on the topic of adult mortality. The first, by Porter and Zaba [5] examines the direct evidence furnished by community-based studies in which the serological status of individuals is measured, allowing comparisons to be made between the mortality of infected and uninfected individuals. The second, by Blacker [6], looks at the indirect evidence afforded by national level statistics, in which serostatus is unknown, but time trends and age patterns strongly suggest a major impact of the HIV epidemic.

In five community-based studies in eastern Africa, mortality in HIV-infected adults was between 10 and 20 times higher than in uninfected individuals after standardizing for age. There is a characteristic age pattern in the relationship between mortality rates for infected and uninfected individuals, so that the maximum impact is usually observed between the ages of 20 and 40 years, earlier for women and later for men, a reflection of typical age patterns of incidence, which peak 5–10 years earlier for women. Porter and Zaba [5] examine a variety of methods for calculating the population-attributable fraction of mortality to obtain estimates of how much lower overall mortality would be without HIV-related deaths from a variety of data sources. In the seven studies reviewed, crude adult prevalence ranged from 4 to 16%, attributable mortality ranged from 24 to 74%, generally the population-attributable fraction is approximately six times prevalence, the exact relationship depending on epidemic maturity and background mortality level. A few developing country cohort studies have furnished evidence about mortality by duration since infection; the most reliable estimates come from the Masaka study [7], and suggest that median survival time post-infection is of the order of 9 years, approximately 2 years less than observed survival in developed country cohort studies before the advent of effective antiretroviral therapy.

Blacker [6] reviews the usefulness of sources such as vital registration systems, national censuses, demographic surveys and demographic surveillance systems for furnishing information on adult mortality trends in the context of the HIV epidemic. He provides an incisive discussion of the weaknesses of each source when it comes to furnishing concrete proof of the impact of AIDS, and describes some useful methods for adjusting and correcting estimates on the basis of these data. Census and survey data from Kenya, Malawi and Zimbabwe showed rising adult mortality in the 1990s: in Kenya the probability of dying between the ages of 15 and 60 years (45q15) increased from 18% in the early 1990s to 28% by the end of the decade; in Malawi it is now over 45%, having been below 30% in the early 1980s; in Zimbabwe by 1997 45q15 was 50% in women and 65% in men. In Zimbabwe, these results are confirmed by trends in mortality rates based on vital registration, after making allowances for improved coverage of the death registration system. Registration data in Thailand and Trinidad also furnish evidence for mortality increases, even though HIV prevalence is considerably lower than in most of Africa. In Thailand, the crude mortality rate for young adults (aged 15–49 years) increased from 2.8 to 5.4 per thousand between 1987 and 1996; in Trinidad AIDS has become the leading cause of death in the 15–44 year age group. Given the imperfections in the data sources and the periodic nature of censuses and large-scale population surveys, the general population mortality impacts of HIV only become apparent some 10 years or more after HIV prevalence reaches significant levels. Indirect evidence of mortality impact should be sought primarily in those age groups (20–40 years), which the direct studies have identified as having the most disparate mortality rates between infected and uninfected and a high prevalence of infection.

Back to Top | Article Outline

Child mortality

Since the beginning of the HIV pandemic, nearly four million children under 15 years of age worldwide have been infected with HIV, and in the year 2003 alone an estimated 700 000 children were newly infected [2], nearly always through mother-to-child transmission, which can occur before, during and after delivery, including breastfeeding. In the absence of vital registration and reliable cause of death information, evidence on the impact of HIV infection on child mortality is limited to data from either relatively small hospital-based studies or to more extensive population-based data that have been manipulated with specific assumptions to arrive at estimated rates of child mortality caused by HIV/AIDS. The review by Newell et al. [8] focuses on survival among HIV-infected children, factors affecting survival among HIV-infected children, survival among children of HIV-infected mothers compared with children of HIV-negative mothers, and child mortality trends and estimates from population-based surveys.

Before the introduction of antiretroviral therapy, the progression of disease among HIV-infected children in Europe and the USA was considerably slower than that observed in a few studies in Africa. Data from the studies in west and east Africa indicate that the median survival time is below 2 years of age compared with well over 5 years of age in industrialized country studies. There is increasing evidence that the survival time of children is affected by the stage of infection of the mother, maternal survival, the use of antiretroviral therapy, background mortality levels and possibly the timing of infection. Currently, there is no consistent evidence that children who acquire infection in utero or intrapartum have a higher mortality rate than those who become infected through breastfeeding.

Data from population-based longitudinal studies with HIV testing of adults in Malawi, Tanzania and Uganda permit an assessment of child mortality among children of HIV-positive mothers. Allowing for geographical and temporal differences in background levels of mortality, for the child's sex and for the mother's age, children born to HIV-infected mothers were estimated to be three times more likely to die than children born to uninfected mothers, with the effect lasting throughout the first 5 years of childhood. The results also indicate that having a dead or dying HIV-negative mother is associated with similar adverse consequences to having a healthy HIV-infected mother.

Trend data for child mortality are available for many countries from vital registration, census, or more recently, household surveys. In many countries there is evidence of a reversal of the child mortality decline during the 1990s, especially in some of the more severely AIDS-affected countries, and modelling indicates that AIDS is a significant contributor to the changes in many countries. There are, however, large differences between African countries, and both disparate trends in HIV prevalence and varying levels of non-HIV-associated child mortality will ensure very different impacts in different countries. It has been estimated that HIV/AIDS was the primary cause of approximately 10% of under-five deaths in sub-Saharan Africa in 2003 [9].

Back to Top | Article Outline


In a review of six community studies, it was concluded in 1998 that in societies with low contraceptive use, HIV-infected women have at least 25% lower fertility than HIV-uninfected women, except for the youngest age group in which selection for early sexual activity results in higher fertility for the HIV infected. Lewis et al. [11] examined new evidence from 19 community studies on the estimates of the fertility rate ratio and population-attributable change in total fertility, and concluded that the earlier estimates are still valid. A few studies contributed multiple observations over time, which is of particular interest for the assessment of biases associated with the mainstay of HIV surveillance in generalized epidemics: monitoring trends among women attending antenatal clinics. No clear patterns of change in the association between fertility and HIV over time have emerged.

The direct fertility impact of the AIDS epidemic at the population level is small but not negligible if the epidemic is severe. Lewis et al. [11] concluded that a 1% change in HIV prevalence is associated with a population-attributable change of 0.37% in total fertility, and also showed that Uganda had 6% fewer births during the past two decades because of AIDS. Little is known about the fertility behaviour of the much larger HIV-negative population in the era of AIDS. National surveys do not reveal any association between fertility intentions and the stage of the epidemic. It is clear, however, that better data are needed.

Back to Top | Article Outline

Population and family structure

The impact of the AIDS epidemic on adult mortality may affect the welfare of affected families and social institutions. Heuveline [12] considered such compositional changes, changes in the age structure of the population or in the composition of households, which may ultimately have far-reaching social and economic consequences. Changes in the population age–sex structure have hitherto been limited to specific age groups, and no dramatic changes have occurred in most countries. Focusing on the most severely affected parts of Africa, Heuveline [12] found that demographic and social processes have diffused the impact of the epidemic throughout the entire population and spread it over all households. Limited evidence of the negative impact of the epidemic has begun to appear, but overall family systems of the most affected populations have been quite resilient. However, there are good reasons to continue and step up the monitoring of households and communities affected by the epidemic, as the next decade is likely to see further increases in adult mortality, even if access to antiretroviral therapy increases rapidly.

National household surveys are an essential large-scale monitoring tool. Monasch and Boerma [13] analysed data from 40 recent national surveys in sub-Saharan Africa to assess the impact of AIDS on the prevalence of orphanhood and to describe orphan care patterns. Among children under 15 years of age in the 40 countries, representing nearly 97% of the region's population, 9% had lost at least one parent and 1% had lost both parents. The prevalence of orphanhood increased during the 1990s in countries more severely affected by HIV/AIDS. In a region where child fostering is very common irrespective of the survival status of the parents, most orphans continue to live with a family member, most often with the grandparents if not with the surviving parent. The surveys present some evidence for the negative association between education and orphanhood among older children.

Back to Top | Article Outline

Accuracy of estimates of impact

Model-based estimates are the most commonly used source of information on the size and impact of the AIDS epidemic. Wide margins of error are usually given, because of the limitations of the available data. The availability of an easy and cheap diagnostic test (an HIV antibody test) has greatly helped the measurement of HIV prevalence. The growing body of evidence on the association between AIDS, adult and child mortality and orphanhood allows an assessment of the accuracy of demographic and epidemiological estimates.

Stover et al. [14] compared the ratio of mortality among HIV-positive and HIV-negative populations, the proportion of all adult deaths attributable to AIDS, and the number of orphans between research studies in countries with generalized epidemics with estimates published by UNAIDS. The differences between the demographic indicators from the two sources are generally within the uncertainty range of plus or minus 25–35%, which was used by UNAIDS for the end-2001 estimates. In other words, these results are encouraging, and indicate that the demographic estimates based on surveillance data and demographic models are reasonably accurate and adequate for advocacy and medium-term planning. For short-term planning it is important to recognize that there is a wide uncertainty range around the estimates.

Current improvements in surveillance, including population-based surveys with HIV testing and the extension of antenatal clinic-based surveillance into the rural population will contribute to the improvement of the accuracy of the estimates by providing better epidemiological estimates of HIV prevalence. On the other hand, it is not likely that more data will become available to assess the distribution of survival time [5]. Increasing access to antiretroviral therapy is likely to widen the uncertainty ranges around the mortality impact of the epidemic.

Back to Top | Article Outline

Socioeconomic impact

This special issue of AIDS is built on findings presented at a conference that addressed both the demographic and socioeconomic impact of AIDS. The socioeconomic sections of the conference included sessions and papers on family welfare, family and household structure, empirical evidence of impacts in the private sector and empirical evidence for impact in the public sector. Of the 41 papers presented, 19 were about the impact of the disease on welfare or economics, although in some cases the impact was rather peripheral. Despite this, the seven review papers in the volume do not include one on socioeconomic impact. The last part of this introductory article will discuss why there are no economic and sociological reviews in the volume. It will also draw together some of the more recent information on the socioeconomic effects of the HIV epidemic and draw some conclusions on the socioeconomic impact.

HIV/AIDS usually affects prime-age adults at the peak of their economically productive years and incapacitates them for significant periods of time before causing their death. This changing pattern of mortality is well illustrated in the demographic section of the supplement. From an economic perspective, it is at the level of individuals and their households that the epidemic has its most significant impact. Where a large number of households are affected, as in many countries in sub-Saharan Africa and some countries in Asia and the Caribbean, the economic impact of HIV/AIDS then spreads to the communities, particular economic sectors, and finally entire national economies. In economic terms, the macro is made up of the micro, each illness and death causes misery and increased poverty to the families and households of those infected, but it is the many thousands of illnesses that may cause economic growth to falter.

There are a number of reasons why empirical papers on the socioeconomic impact of AIDS have not been included in this volume, which include the following.

Back to Top | Article Outline

Lack of empirical data

The first reason is the lack of data. Demographers have access to censuses, vital registration data and Demographic and Health Surveys: large national databases. Economists do not have this information available. It is impossible to say what the macroeconomic or social implications will be, although there are a growing number of household and private sector studies, which are measuring the current situation. Sectoral and household impact data are not collected routinely in national surveys, so we are dependent on a small patchwork of surveys, often with incomplete data, sometimes added on to other instruments. There has never been a study of a national economy which attempts to measure the impact of AIDS and it would be very difficult to perform one, although one might hope for more imaginative use to be made of data collected at the household level for the cost of living, labour force and national accounts surveys. It is not feasible to link such information to individual HIV status, but it might at least be possible to link it to information on household experience of mortality and morbidity.

Back to Top | Article Outline

Time scales

It is very difficult to look at the socioeconomic impact because most of the consequences of increased adult morbidity and mortality have not yet been felt, let alone measured; in many countries they will not be felt for many years. The nature of the disease means that illness and death takes place between 8 and 10 years after infection, and the consequences will evolve in the years after this. In many African countries HIV prevalence is still increasing. There is evidence from Uganda to show that although the prevalence of HIV began to decline in the early 1990s, the number of orphans in the country was still increasing in 2003. This means that if HIV prevalence is nearing its peak in, say Botswana, the number of orphans will increase for many more years, and the social problems of coping with this increase will continue to escalate.

Back to Top | Article Outline

Process versus event

Demography is predominantly interested in measuring and explaining vital events, such as births and deaths. Economists and sociologists recognize that process is also crucial in looking at AIDS. For individuals living with or impacted by AIDS the process of illness, struggling to cope, and impoverishment may be as important as the deaths and orphaning. Indeed for governments facing illness among staff it is ironically easier to plan for early retirement or death and replacement than for protracted periods of illness. Firms are likely to be impacted by decreased productivity among their workforces. Morbidity and mortality are the determinants of socioeconomic impacts on families and households and thus in the longer term on national economies. We are rather better at measuring mortality than morbidity.

Back to Top | Article Outline

Units of measurement

An additional problem for economists and sociologists is that in some instances we may be trying to measure something that is not there. The worst affected households dissolve, and if a survey is carried out at a household level then the ones that have gone will be missed. Papers in this volume [12,13] show that AIDS deaths are associated with an increase in household dissolution and child migration. Although the children will generally end up in another household, survey questions addressed to the new household head may not shed light on the events that led to the relocation of the children, because these events occurred in another household. Another example of this phenomenon at the national level is that we believe that AIDS and the perception of some parts of the world as ‘plague areas’ has reduced flows of foreign direct investment, but how do you measure money that has not come to a country?

Back to Top | Article Outline

The relative value of lives

In looking at the impact of illness and death on households and economies the problem is that not all lives are, in economic terms, equally valuable. Developing world economies may grow more slowly and their structures change. This could mean that there will be fewer jobs and less money to spend on healthcare. If, however, people's deaths did not affect the functioning of the economy, then at the macroeconomic level and in purely economic terms the survivors could be better off. The issue of the relative value of lives is going to be very important as treatment is rolled out; who will access the expensive therapy and how will decisions on this be made? These are ethical issues, but unless grasped as such, economists may have an inordinate influence.

Most publications that look at the impacts of AIDS adopt similar formats. They address the impact on households, the effects on various public sector activities (normally health and education because this is where the limited data are available), the private sector and finally macroeconomic impact. A number of review articles and books have done this [15–17]. Rather than going through the same process here we will highlight some of the main findings presented at the conference and events or publications that have taken place since then.

Back to Top | Article Outline

Conference presentations

The conference presentations with the most empirical evidence on impact came from the Centre for International Health at Boston University's School of Public Health, the source of two papers. The first presentation on the cost of HIV/AIDS to businesses showed that AIDS is equivalent to an additional payroll cost [18]. Six formal sector enterprises in South Africa and Botswana provided detailed human resource, financial, and medical data and carried out voluntary, anonymous HIV seroprevalence surveys. The authors then estimated the present value of incident HIV infections with a 9-year median survival and 7% real discount rate. HIV prevalence in the workforces studied ranged from 7.9 to 25.0%. It was estimated that HIV/AIDS among employees added 0.4–5.9% to the companies’ annual salary and wage bills. Costs varied across firms and among job levels within firms depending on the levels and stability of employee benefits, HIV prevalence, and the contractual status of unskilled workers.

The second paper presented brand new data on the effect of AIDS on productivity on a tea estate in Kenya [19]. The authors noted that there is very little research on productivity because in most settings neither the health nor the productivity of an individual worker can be directly observed. On this commercial agriculture estate workers are paid by the amount harvested each day, and receive healthcare from on-site, company-owned medical facilities. There is thus a fortunate coincidence of data on both the daily output and the health of each worker. The team reviewed records of health and output over a 5-year period. They found that during their last 3 years of life, tea pluckers who died as a result of AIDS were absent from work almost twice as often as other tea pluckers. Most (58%) of this difference consisted of unpaid (and unauthorized) leave. Output began to fall as early as 3 years before death for those infected with HIV. Over the last 3 years of life, individuals who were to die of HIV infection averaged only 91% of the ‘full’ productivity (of controls). During the last year before death, the productivity fell sharply to 82% in the last year of life and 77% in the last 3 months.

The public sector presentations were data poor. Information is being collected on a range of education indicators through the HEARD Education Mobile Task Team and using a District Education Management and Monitoring Information System [20]. The data on educator mortality in KwaZulu-Natal was presented, but so far extends only over a short period, and although it suggests disquieting trends, with increased mortality, the main benefit may be as a model for data collection.

There is real concern about the impact of AIDS on the health sector, but by and large this is unquantified. There will be an increased demand for healthcare and reduced capacity as healthcare workers fall ill and die. Since the conference was held in March 2003, the new Director General of the World Health Organization has made a commitment that three million people will be accessing antiretroviral therapy by 2005 (the 3 by 5 initiative). The implications of this for the health sector are huge and this is an area that urgently needs data. Comments made in the conference showed how little is currently available.

Another huge gap was the potential impact of AIDS on government expenditure and revenue collection. This lacuna has been brought to prominence by the recent statement by the President of Botswana that the huge budget deficit in his country in 2003/2004 is caused by the cost of dealing with AIDS [21]. The government had predicted a surplus, instead they face a US$396 million deficit.

The economic impact of HIV/AIDS on households is so significant because most individuals living with HIV/AIDS in highly affected countries are parents and workers providing for their households. Moreover, the impact is compounded because unlike other diseases, HIV/AIDS-related illness and death tend to be clustered within selected households with two or more adults affected as one member passes on the virus to his or her partner(s). The impact on households is felt in a number of different ways. Household income is affected as earnings are foregone during periods of prolonged illness and as a result of untimely deaths. Household expenditures are affected as medical, transportation and funeral expenses tend to increase and expenses on other categories of goods and services are decreased. Household assets and savings are affected as households use them to meet their augmented needs. Finally the dissolution of households after the deaths of parents means that orphans are usually fostered out to family members or left to fend for themselves.

Death and sicknesses were shown to be expensive and increase poverty. In South Africa many of the affected households live below the R220 per capita poverty line. In Kenya the death of a male household head is associated with a 68% reduction in the net value of the household's crop production; much less severe effects were found in the case of the death of other household members [22]. What was alluded to frequently was that AIDS might increase inequality and affected households could be pushed into deep poverty. Households with HIV/AIDS tend to spend relatively more on certain categories of goods and services than do unaffected households. In Free State Province, significant differences were found between expenditures on food within affected households and non-affected households. Per capita expenditures on food were 23 and 32% less among urban and rural affected households than they were among unaffected urban and rural households [23].

Another impact of HIV/AIDS on households is that it leads to decreases in household assets and savings and increases in borrowing after the reduced income resulting from incapacitation as a result of prolonged illness or the death of a household member. Looking at a variety of savings options including bank savings, insurance policies and stocks, the study in Limpopo Province found that affected households saved approximately 36% less than did unaffected households [24]. Although unaffected households in the study borrowed more from banks and burial services agencies than did affected households, affected households were significantly more likely to borrow from relatives.

There is no empirical evidence for the macroeconomic impact of HIV/AIDS. All statements on the likely impact of AIDS are based on macroeconomic models of various types, but until recently they always predicted economic growth with and without AIDS. The models have been based on two key assumptions: that there will be a loss of production from the labour force as individuals fall ill and die; and second, that the costs of these illnesses and deaths will be funded from money that would otherwise go to investment [24].

Most models show a small impact on macroeconomic growth, of the order of between 0.2 and 1.4% less per annum. Although this is quite significant over time, at any given point it is very small. There are other influences on economies (drought, oil prices and political instability, which may have greater impacts). Furthermore, the models suggest that although total output (gross domestic product) may go down, the impact on the per capita income is less clear; indeed, per capita income may even go up depending on how many individuals die and the impact of this mortality on savings. In other words, the absolute wealth may be smaller but the number of individuals among whom it is divided will be even fewer: a ‘perverse’ effect that is generally not discussed because of its political and ethical implications [25].

One major difficulty with models is that they have not taken into account possible feedback loops. The most recent attempt to understand the complexity of the epidemic is the recent World Bank report [26]. The authors argue that the long-run economic costs of AIDS are almost certain to be very much higher than those predicted to date and may even be devastating. Their paper emphasizes the importance of human capital in an ‘overlapping generations’ model. This model when applied to South Africa warns of economic collapse in three generations [26].

Back to Top | Article Outline


This is the second conference on the social, demographic and economic impact of AIDS that the authors of this article have jointly organised in Durban. The first was in February 1997. That meeting resulted in two publications, the demographers produced the supplement to AIDS referred to at the beginning of this article [1]. The social scientists and economists produced an edited book [27]. The concluding chapter noted ‘this book is intended to begin the dialogue between planners and demographers’ (p. 138). Nearly 6 years on, we need to ask what progress has been made.

One thing is clear, and that is that the social sciences will have difficulty in producing the type of empirical data that demographers are used to working with. This is a reflection of the nature of the epidemic and its evolution. There are those who would argue that we will always miss the true impact of this insidious disease when we apply economic and social tools. We can identify orphans and assess what effect orphaning has on their life chances and education, but psychosocial and emotional impacts are much more difficult to measure. We could measure the coping strategies in terms of household resource allocation in families that take in orphans, but it would be harder to measure the impact of the presence of orphans on the development of other children in these families, or the effect of depressed and uncared for children in a classroom. How do we cost a cuddle?

At the same time, the evidence base of demographic evidence has broadened substantially during recent years. Much of the evidence is, however, generated by a small number of longitudinal studies in sub-Saharan Africa, and there is a clear need for more studies to capture the impact of the epidemic as it unfolds. Such impact data will be crucial to mitigate the impact of the AIDS epidemic through an effective response.

Back to Top | Article Outline


1. Caraël M, Schwartlander B (editors). Demographic impact of AIDS.AIDS 1998; 12 (Suppl. 1):S1–S2.
2. UNAIDS WHO. Epidemiological update December 2003. Geneva; UNAIDS WHO; 2003.
3. WHO, Regional Office for Africa. HIV/AIDS Epidemiological Surveillance Update for the WHO African Region. Harare, Zimbabwe, 2003.
4. Timaeus I, Jasseh M. Adult mortality in sub-Saharan Africa: evidence from the Demographic and Health Surveys.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, 26–28 March 2003.
5. Porter K, Zaba B. The empirical evidence for the impact of HIV on adult mortality in the developing world: data from serological studies.AIDS 2004; 18(Suppl 2):S9–S17.
6. Blacker J. The Impact of AIDS on adult mortality: evidence from national and regional statistics.AIDS 2004; 18 (Suppl. 2): S19–S26.
7. Whitworth J, Shafer LA, Mahe C, van der Paal L. Survival since infection and its relation to background mortality in the Masaka general population cohort.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, 26–28 March 2003.
8. Newell ML, Brahmbhatt H, Ghys PD. Child mortality and HIV infection in Africa: a review.AIDS 2004; 18 (Suppl. 2):S27–S34.
9. Walker N, Schwartländer B, Bryce J. Meeting international goals in child survival and HIV/AIDS.Lancet 2002; 360:284–289.
10. Zaba B, Gregson S. Measuring the impact of HIV on fertility in Africa.AIDS 1998; 12 (Suppl.):S41–S50.
    11. Lewis JJC, Ronsmans C, Ezeh A, Gregson S. The population impact of HIV on fertility in sub-Saharan Africa.AIDS 2004; 18 (Suppl. 2):S35–S43.
    12. Heuveline P. Impact of the HIV epidemic on population and household structure: the dynamics and evidence to-date.AIDS 2004; 18 (Suppl 2):S45–S53.
    13. Monasch R, Boerma JT. Orphans and child care patterns in sub-Saharan Africa: an analysis of 40 national surveys.AIDS 2004; 18 (Suppl. 2):S55–S65.
    14. Stover J, Ghys PD, Walker N. Testing the accuracy of demographic estimates in countries with generalized epidemics.AIDS 2004; 18 (Suppl. 2):S67–S73.
    15. Opuni M, Bertozzi S. The global impact of HIV and AIDS, Chapter 41. In: Wormser GP (editor). AIDS and other manifestations of HIV infection, 4th ed. Elsevier Science; 2003.
    16. Barnett T, Whiteside A. AIDS in the twenty-first century: disease and globalisation. London: Palgrave; 2002.
    17. Kelly K, Parker W, Gelb S. HIV/AIDS, economic and governance in South Africa: key issues in understanding response. Johannesburg: Centre for AIDS Development Research; July 2002.
    18. Fox MP, Simon JL, Bii M, Wasunna KM, MacLeod WB, Foglia G, Rosen S. The impact of HIV/AIDS on labor productivity in Kenya.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, 26–28 March 2003.
    19. Rosen S, MacLeod WB, Vincent JR, Fox MP, Thea D, Simon JL. Investing in the epidemic: the value to the private sector of HIV/AIDS prevention and treatment.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, 26–28 March 2003.
    20. Badcock-Walters P. Educator mortality in KwaZulu-Natal: a consolidated study of impact and trends.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, 26–28 March 2003.
    21. BBC World Service. Webnews: AIDS cause of Botswana crisis. 21 September 2003.
    22. Yamano T, Jayne TS. Measuring the impacts of prime-age adult death on rural households in Kenya.Paper presented at Conference on Empirical Evidence for the Demographic and Socioeconomic Impact of AIDS. Durban, South Africa, March 26-28, 2003.
    23. Booysen F. Financial responses of households in the Free State Province to HIV/AIDS-related morbidity and mortality.S Africa J Economics 2002; 70:1193–1215.
    24. Oni A, Obi CL, Okorie, Thabede D, Jordan A. The economic impact of HIV/AIDS on rural households in Limpopo Province.S Africa J Economics 2002; 70:1173–1192.
    25. Barnett T, Whiteside A. AIDS in the twenty-first century: in disease and globalisation, Chapter 11. Basingstoke: Palgrave; 2002.
      26. Bell C, Devarajan S, Gersbach H. The long-run economic costs of AIDS: theory and an application to South Africa. Washington DC: The World Bank; 2003.
      27. Whiteside A (editor). Implications of AIDS for demography and policy in Southern Africa. Pietermaritzburg, South Africa: University of Natal Press; 1998.
      © 2004 Lippincott Williams & Wilkins, Inc.