The World Health Organization estimates that more than 12 million people were infected with the HIV in sub-Saharan Africa at the beginning of 1996(1). In South Africa, national antenatal HIV seroprevalence has risen rapidly, from 0.76% in 1990 to 7.6% in 1994(2,3). KwaZulu/Natal, in the northeast of the country, is the province worst affected: antenatal prevalence was estimated at 14.4% in the 1994 survey(3).
Little has been published on the epidemiology of the HIV epidemic in rural South Africa(2). The only population-based study, done in KwaZulu/Natal in 1990, reported a prevalence of 1.2%(4). There is a need for more detailed epidemiologic study, because most reports quote the national antenatal surveys and actuarial models(2). Anonymous antenatal surveys provide useful information on levels of, and trends in, HIV infection in members of the population conventionally categorized as being at relatively low risk(5). Women attending antenatal clinics are recognized as being reasonably representative of the fertile sexually active population(5). Results of these surveys can guide the design and evaluation of prevention and care programs.
The objective of this study was to describe the increasing prevalence of HIV infection in a rural district in South Africa, to compare this with a nearby urban setting, and to estimate the prevalence of infection in the general population of the rural district. This is important because little has been published on the epidemic from rural South Africa.
KwaZulu/Natal (population approximately 10 million) is the largest of South Africa's nine provinces. The population of the largely rural Hlabisa health district in 1995 was estimated at 205,463 from the 1991 census. Most residents speak Zulu and live in widely scattered kraals, relying on pension remittances, migrant labor, and subsistence farming for money and food. A major national road and trading route crosses the district and a large township is sited on this road. The annual per capita income in KwaZulu/Natal is R5189 ($1730 U.S.), the literacy rate 69%, and the life expectancy is 63 years(6).
Antenatal care is provided by the local 450-bed district hospital, seven village clinics, and two mobile clinic teams. At their first antenatal visit, all women have blood taken to estimate hemoglobin concentration and to test for rhesus and syphilis serology. Approximately 95% of pregnant women in the district receive antenatal care at these clinics(7).
In March 1992 (n = 884), November 1993 (n = 709), and June 1995 (n = 314), three anonymous surveys of HIV infection in women attending the 10 antenatal clinics were conducted. Serum left over from the antenatal blood testing of consecutive women at each clinic over a defined time period, had personal identifiers removed and was marked only with the patient's age and the name of clinic. Serum was stored at 4°C and transported to the regional hospital within 48 hours for testing for antibody to HIV. Two different method enzyme-linked immunosorbent assay (ELISA) tests were used, according to laboratory practice at that time, and in accordance with international guidelines(8). Specimens were deemed HIV-positive if both ELISA results were positive or if one positive ELISA result was confirmed with a Western Blot or with an immunofluorescent assay(8). Confidential and voluntary HIV testing and counseling was available in all clinics at the times of these surveys.
Data from the national antenatal surveys, which provide provincial prevalence estimates, were compared with the Hlabisa surveys. Results of an anonymous antenatal survey of consecutive women booking for antenatal care at the King Edward VIII Hospital, Durban were provided by the Department of Virology, University of Natal, Durban. Durban was chosen because it is the nearest large urban center with such data available.
Sample size calculation for the 1992 local survey was based on estimates of infection in the province reported in the first national antenatal survey. Sample sizes for subsequent surveys were limited by cost and were calculated to have the power to detect a doubling in overall prevalence at the time of each survey. HIV results were categorized by age and clinic. Estimates of crude, age-specific, and clinicspecific prevalence rates with their 95% confidence intervals were calculated using EPI INFO 6.02 (Centers for Disease Control and Prevention, Atlanta, GA, U.S.A.). Small sample size did not allow accurate comparison of rates between individual clinics within the district. Proportions were compared by the χ2 test. To enable valid comparison between surveys direct age, standardization(9) was also done. To estimate the burden of infection in adults in the district, the age-specific prevalence rates from the 1995 survey were applied to the population age distribution estimates for 1995 from the 1991 census; 95% confidence intervals were obtained by using the corresponding intervals for each age-specific prevalence. The male:female ratio of infection was estimated to be 1:1(10). The prevalence of HIV infection among children in the district in 1995 was estimated by assuming a mother-to-child transmission rate of 30%(10), which was applied to the estimated prevalence of infection among antenatal women for each year 1991 to 1995 and the number of women who delivered in the district(extrapolated from health service data). Prevalence in children was discounted each year by 15% to account for their assumed mortality rate(10).
HIV Prevalence in Hlabisa
The prevalence of HIV infection in women attending antenatal clinics in the Hlabisa district increased from 4.2% (95%CI, 3.0-5.7) in 1992 to 7.9% (95%CI, 6.0-10.1) in 1993, and to 14.0% (95%CI, 10.4-18.4) in 1995(Table 1). This increase is highly significant (p < 0.0001).
The pattern of age-specific prevalence was similar each year (Table 1); (Fig. 1). Highest prevalence was in the 20- to 24-year-old age group, increasing from 6.9% in 1992 to 21.1% in 1995 (p = 0.0001). Results were not significantly affected by age standardization and the unadjusted figures are reported here.
Prevalence was not uniform in the different clinics within the district. Higher prevalence was consistently measured in the clinic that serves the township situated on the national road (8.5% in 1992, 10.3% in 1993, 29.5% in 1995).
Comparison With Other South African Surveys
HIV prevalence in Hlabisa was consistently lower than that estimated for the province as a whole from the national surveys (Table 2). Prevalence in the Durban antenatal clinic was 19% (95%CI, 16.5-21.7) in 1995, compared with 14% in Hlabisa (p = 0.045). The difference in these crude rates can be attributed to the much higher age-specific prevalence in women aged 15 to 19 years in Durban (22.4%) than Hlabisa (7.4%, p = 0.004). Other age-specific rates were similar in the two sites.
Estimates of the Prevalence of Infection in the Rural District
Using the estimates and assumptions listed above we estimated that there were 11,571 prevalent HIV infections(5547 women, 5547 men, and 477 children) in Hlabisa district in 1995 (Table 3). This corresponds to 5.6% (95%CI, 3.0%-9.6%) of the total population.
The HIV epidemic in rural South Africa has escalated rapidly. Within 4 years the antenatal seroprevalence in Hlabisa has increased from 4.2% to 14.0%. The 1995 population prevalence was estimated at 5.6%: a substantial increase over the 1990 estimate of 1.2% in a neighboring area(4). This escalation poses a major challenge, in terms of both prevention and care, for South Africa.
An important epidemiologic parameter is the incidence of HIV infection. Although not directly measured here, the epidemic in rural South Africa seems to be driven by a high incidence of infection in young people, because the prevalence is highest and rising fastest in the younger age groups. In this area, one in five rural women aged 20 to 24 years were HIV-positive in 1995. Lower rural than urban HIV prevalence is well recognized in Africa(11) and is probably the result of differences in timing of HIV entry into the population, sexual behavior, prevalence of sexually transmitted disease (STD), and migrancy(5). Our results reflect this and suggest that the difference between rural and urban prevalence in South Africa is mainly a result of much higher rates of infection in the youngest urban women than their rural counterparts(more than one in five urban women aged 15 to 19 years were HIV infected in 1995). Prevalence also varied within the rural district with higher rates occurring in the township close to the national road. We have previously shown that HIV infection is highly associated with migrancy in this area(12). The underlying societal factors that account for these observations need further study to inform prevention activities.
Women attending antenatal clinics are conventionally categorized as being at relatively low risk(5). Because 14% of women tested in 1995 were HIV infected, we believe that this risk categorization is misleading. Although their risk may be lower than that of commercial sex workers or people with STD, for example, these women do clearly suffer a substantial risk and this fact needs to be recognized.
What are the implications of our findings for prevention? Resources-particularly manpower-currently in existence do not allow a truly comprehensive response to the HIV epidemic in our setting. Within the Hlabisa district, despite our best efforts to use all available resources over several years, we feel we are struggling to respond effectively. Could interventions be targeted? In the United States, targeting, although politically sensitive, is now recognized as being a valid strategy. In Thailand, the highly effective "100% condom campaign" was initiated in areas of highest prevalence, which were in turn, the first to show a decrease in HIV prevalence(13). Is targeting appropriate, however, in South Africa? It seems that the greatest impact could be had by targeting interventions at young people, especially before they initiate sexual activity. An important question to ask is, "who is infecting these very young women and under what circumstances?" Risk also seems highest in urban areas, and among migrants and their partners(12). However, to define different risk areas as "urban" or "rural" is simplistic. These terms do not describe the behavior patterns that lead to the different risks, nor help our understanding of how such patterns are formed and influenced. Clarifying these issues will enhance the effectiveness of preventive interventions.
A significant clinical burden of HIV-related disease is already manifest in rural South Africa(14). The area is dominated by tuberculosis and other early manifestations of HIV-related disease. Population-based estimates of HIV infection can be used to estimate the expected rates of disease such as tuberculosis and pneumococcal disease, and this is important for health service planners. Estimating the number of HIV-infected children born each year is similarly of value, because many will become ill and require care within the first 2 years of life.
The demographic, economic, and social implications of the impact of HIV infection in our setting are staggering(6). The population growth rate is expected to fall, the population structure will change, and the increased demand for health care and care of orphans is already being felt(6,14). The impact on the health service is particularly worrying(15,16). No slack is present within hospitals to cope with the increased demand for care. Caring for the sick and their dependents will need to take place in the community. Proactive mobilization, consultation, and planning are vital. In addition, innovative ways of providing effective care for the increasing numbers of people with, for example, tuberculosis will need to be developed(17).
Acknowledgments: This study was supported by the South African Medical Research Council and the Department of Health, KwaZulu/Natal. We wish to thank the Department of Obstetrics and Gynaecology and the Department of Virology, University of Natal, Durban, for allowing us to use the data from the King Edward VIII antenatal clinic. We also thank Drs. Charles F Gilks and Kevin M De Cock for making valuable comments on the manuscript.
1. World Health Organization. AIDS: image of the epidemic
. Geneva: World Health Organization, 1994.
2. McIntyre J. HIV/AIDS in South Africa: a relentless progression? S Afr Med J
3. Department of National Health and Population Development. Fifth national HIV survey in women attending antenatal clinics of the public health services in South Africa, October/November 1994.Epidemiol Comments
4. Abdool KQ, Abdool KK, Singh S, Short R, Ngxongo S. Seroprevalence of HIV infection in rural South Africa. AIDS
5. US Bureau of the Census. HIV/AIDS in Africa.
Research Note 20. Washington, DC: U.S. Government Printing Office, 1995.
6. Whiteside A, Wilkins N, Mason B, Wood G. The impact of HIV/AIDS on planning issues in KwaZulu/Natal
. Pietermaritzburg, S. Africa: Town and Regional Planning Commission, 1995.
7. Wilkinson D, Cutts F, Ntuli N, Abdool Karim SS. Maternal and child health indicators in a rural South African health district. S Afr Med J
8. World Health Organization. Recommendations for the selection and use of HIV antibody tests. Wkly Epidemiol Rec
9. Hennekens CH, Buring JE. Epidemiology in medicine
. Boston: Little Brown, 1987:66.
10. World Health Organization. AIDS in Africa: a manual for physicians
. Geneva: World Health Organization, 1992.
11. Nunn AJ, Kengeya-Kayondo JF, Malamba S, Seeley JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS
12. Colvin M, Abdool Karim SS, Wilkinson D. Migration and AIDS.Lancet
13. Rojanapithayakorn W, Hanenberg R. The 100% condom program in Thailand. AIDS
14. Wilkinson D, Habgood LC. The evolving HIV epidemic in a rural hospital in Zululand from 1989 to 1993. Epidemiol Comments
15. Impact of HIV on delivery of health care in sub-Saharan Africa: a tale of secrecy and inertia [editorial]. Lancet
16. Gilks CF, Haran D, Wilkinson D. Coping with the impact of the HIV epidemic: the Hlabisa-Liverpool Link. S Afr Med J
17. Wilkinson D. High-compliance tuberculosis treatment programme in a rural community. Lancet
Keywords:© Lippincott-Raven Publishers.
HIV prevalence; South Africa, rural.