From the aAfrica Centre for Health and Population Studies, University of KwaZulu-Natal, Durban, South Africa
bDepartment of HIV and Genitourinary Medicine, King's College London School of Medicine at Guy's, King's College and St Thomas' Hospitals, London, UK
cDepartment of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK
dRobert Koch Institute, Department of Infectious Disease Epidemiology, Berlin, Germany
eUniversity College London, London, UK.
Received 2 October, 2006
Revised 11 January, 2007
Accepted 6 February, 2007
Correspondence to Tanya Welz, Africa Centre for Health and Population Studies, PO Box 198, Mtubatuba, 3935, South Africa. E-mail: firstname.lastname@example.org
South Africa has one of the highest HIV infection rates in the world. The HIV prevalence among pregnant women attending public sector antenatal clinics in 2004 was approximately 30% nationally and 41% in KwaZulu-Natal, the most populous province . Several national household surveys have also reported very high HIV prevalence: 12% among men and 20% among women aged 15–49 years in 2005  and 5% among men and 16% among women aged 14–24 years in 2003 . Detailed information on rural areas is, however, scarce. The Joint UN Programme on HIV/AIDS (UNAIDS) acknowledges that antenatal sentinel sites ‘outside major urban areas’ are not located in strictly rural districts , and the largest population-based survey to date, the Human Sciences Research Council National Household Survey, 2005, overrepresented the formal urban population . The household surveys also did not have reliable information on the size of the target population, failed to enumerate or trace individuals temporarily away from home and had little or no information on non-participants.
We estimated HIV prevalence and its association with sociodemographic factors in a large well-characterized rural population in KwaZulu-Natal, which has been under demographic surveillance since 2000 and where data on individuals' HIV-related risk factors and presence patterns were available for several years preceding the HIV serosurvey.
The study was conducted between June 2003 and November 2004 in the Africa Centre Demographic Surveillance Area in the rural district of Umkhanyakude in northern KwaZulu-Natal. The area of 435 km2 includes deep rural areas, a township and peri-urban informal settlements. In 2003, approximately 86 000 individuals lived in 11 000 households. ‘Households’ range from extended families living in isolated homesteads to single people in tenant blocks . Resident members are those who report living with the household, i.e. they keep their daily belongings there and spend most nights at the homestead. Approximately 30% of all household members are non-resident or ‘migrant’, i.e. they have their main residence elsewhere but maintain connections with the household through periodic visits .
Since January 2000, every household has been visited twice a year to update individual data on births, deaths, demographic and socioeconomic characteristics. Individuals who move or belong to more than one household (e.g. men in polygamous marriages) are tracked at each household to prevent double-counting or loss to follow-up .
All men aged 15–54 years and women aged 15–49 years recorded as resident in the area on 18 June 2003 were eligible for the HIV serosurvey; 12.5% of non-residents were randomly selected in equal numbers from each of 10 strata, defined according to sex and frequency of return visits to the area, for example, every weekend, month end etc.
No major HIV prevention or treatment studies among adults were conducted in the area before 2004. The government antiretroviral treatment programme started in September 2004 and approximately 100 patients were on treatment by December 2004.
Trained fieldworkers visited all eligible individuals at home. If necessary, they arranged workplace visits or out-of-hours appointments. Residents who were away for long periods were visited on their return. Fieldworkers made at least four visits to the household before registering a non-contact. A special tracking team made up to 10 attempts to find individuals who had moved within the area and also tracked non-residents as far as Durban (250 km) and Johannesburg (750 km).
Written informed consent was obtained for HIV testing. Blood was collected by finger prick on filter paper (Schleicher and Schuell, 903 Guthrie cards) according to standard operating procedures . HIV status was determined by antibody testing with a broad-based HIV-1/HIV-2 enzyme-linked immunosorbent assay (ELISA, Vironostika; Organon Teknika, Boxtel, the Netherlands) followed by a confirmatory ELISA (GAC–ELISA; Abbott, Abbott Park, Illinois, USA). Testing was ‘linked anonymous’ but participants could access their results at any of 19 community-based counseling centres using a secret code system and receive appropriate post-test counseling [7,8].
Ethical approval was obtained from the Nelson Mandela Medical School Research Ethics Committee, University of KwaZulu-Natal, Durban, South Africa. No incentives were provided to participants.
Analyses were performed using Stata 8.0 (College Station, Texas, USA). Categorical data were compared with chi-squared tests.
Unconditional logistic regression was used for multivariable analyses. Variables associated with HIV infection at a significance level of P < 0.15 in univariable analyses or that were theoretically important (e.g. calendar period), were tested in both a forward and backward stepwise manner, and were included in the final model if they remained significant at 1% (likelihood ratio test). Some results of borderline significance are also reported.
Analyses of HIV prevalence among non-residents took into account different sampling probabilities in the 10 strata using Stata 8.0 svy methods .
A total of 12 903 men and 15 986 women (total 28 889) residents were eligible for the study; 1925 residents (7%) moved away or died and 191 (1%) aged out of the eligible age range before their households could be visited. A total of 515 (2%) were lost to follow-up for at least one year before the HIV testing visit. Of the remaining 11 505 men and 14 396 women, 8325 men (72%) and 11 542 women (80%) were traced. Of these, 4692 men (56%) and 6859 women (59%; total 11 551) consented to HIV testing.
Fifty-seven of 2009 sampled non-residents (3%) became ineligible (died, severed links with the area, etc.) before their households were visited. Some 46% of men (453/982) and 48% of women (463/970) were successfully traced, of whom 57% (257/453) and 64% (294/463), respectively, consented to HIV tests.
Forty-nine participants (0.4%) tested were not assigned a conclusive HIV status because the result was indeterminate (40) or could not be matched to a specific individual (nine).
HIV prevalence in residents
HIV prevalence peaked at 51% [95% confidence interval (CI) 47–55%] among women aged 25–29 years and 44% (95% CI 38–49%) among men aged 30–34 years (Table 1), with the highest infection rates of 57.5% (95% CI 49–66%) among 26-year-old women. Overall, 21.5% of residents aged 15–49 years were HIV infected.
The female: male infection ratio for ages 15–49 years was 2.0 (95% CI 1.9–2.2) and for ages 15–19 years it was 13.0 (95% CI 7.7–21.9).
HIV prevalence in non-residents
HIV prevalence among non-residents peaked at 63% (95% CI 50–76%) among women aged 25–29 years and 56% (95% CI 34–78%) among men aged 35–39 years (Table 1). Some 25% of male non-residents below 30 years compared with 6% of residents were HIV infected (P ≤ 0.0001). Among women below 30 years of age, 40% of non-residents and 24% of residents were HIV infected (P ≤ 0.0001). The age-adjusted odds ratio for HIV infection among non-residents compared with residents was 1.8 (95% CI 1.3–2.4) for men and 1.5 (95% CI 1.2–2.0) for women.
Effect of non-participation
The characteristics of residents tested and non-participants (i.e. test refusers and absentees who could not be traced) and adjusted odds ratios for HIV infection by characteristic are shown in Table 2. Nine out of 12 factors positively associated with HIV infection [age, urban or peri-urban residence, living alone, earning an income, having electricity, having sanitation, being harder to trace (three or more visits required), being away from the household for 10 or more nights in the preceding 4 months, and not attending school (≤ 18 years only)] were also positively associated with non-participation among men, women or both.
Test refusers generally had sociodemographic characteristics intermediate between those tested and persistent absentees. For example, 33% of individuals tested lived in an urban area compared with 39% of test refusers and 55% of absentees (P ≤ 0.0001). Test refusers also required more visits to trace [median 2; interquartile range (IQR) 1; 3] than residents who consented to test (median 1; IQR 1; 2; P ≤ 0.0001 Mann–Whitney U test).
This survey, the first of its kind in South Africa to investigate the prevalence of HIV in a rural area and in relation to mobility and migration, shows some of the highest population-based infection rates yet documented worldwide.
Forty-six per cent of the South African population live in rural communities where ‘circular’ migration for work is the norm [10,11]. Migration is a generally accepted risk factor for HIV infection in Africa [12–16]. Cross-sectional studies in migration destinations may, however, miss migrants who are mobile and are more likely to live in non-household dwellings. To our knowledge, no other HIV survey has tracked non-resident members of rural households to their migration destinations, and few studies have looked at female migrants [16,17].
Pre-existing data from longitudinal demographic surveillance provided a complete sampling frame that virtually eliminated the likelihood of problems affecting cross-sectional surveys, e.g. errors with household listing and selection [18,19], and also allowed us to interpret the effects of non-participation on HIV prevalence estimates.
Almost all the sociodemographic characteristics associated with an increased risk of HIV occurred significantly more frequently among non-participants, suggesting that refusal to test and absence are likely to have resulted in an underestimate of HIV prevalence in this study. Adjusting our estimates to reflect the age structure of the population , for example, increased the HIV prevalence among women aged 15–49 years from 26.8 to 29.3% and that among men aged 15–49 years from 13.2 to 16.9%.
Our findings support previous suggestions that cross-sectional surveys underestimate the prevalence of HIV because of the non-inclusion of mobile individuals [21–24], and that absence during HIV surveys may represent passive refusal to participate by high-risk individuals . Residents who ultimately refused testing required more visits to trace and were more similar to absentees than residents who consented to test. Also, whereas 50% of residents reported having been away for 10 or more nights in the preceding 4 months, a disproportionate 22% were reported not to have slept in the household the previous night. This suggests that those who did not want to test were reported as temporarily absent. This would increase selection bias in surveys enumerating only individuals who slept in the household the previous night.
UNAIDS recently revised many countries' HIV prevalence estimates downwards on the basis of new information from national household surveys, particularly on remote rural areas, which had been underrepresented in antenatal surveillance [19,25–27]. The 2003 estimates for South Africa have also retrospectively been revised downwards from 20.9 to 18.6%, with the 2005 estimate at 18.8% . Our prevalence of 21.5% among residents of this rural district where two-thirds of the population live in sparsely populated areas, combined with evidence that non-participants may be at a higher risk of HIV, and the much higher infection rates among mobile non-residents who make up 30% of household members, suggests that the burden of HIV in rural areas of South Africa may be higher than previously estimated.
In conclusion, effective monitoring of the HIV epidemic in South Africa and elsewhere in Africa should include efforts to strengthen sentinel surveillance in rural areas and strategies for the surveillance of migrants and mobile individuals who are at an increased risk of HIV and may be missed by cross-sectional surveys. The finding that half of women aged 25–29 years were HIV infected suggests the scale on which antiretroviral treatment will be needed and the urgency to allocate adequate resources for HIV prevention and treatment towards rural areas.
The authors acknowledge the important contributions of all field staff, members of the Africa Centre Population Studies Group and staff at the Africa Centre Laboratory, Durban, South Africa.
Sponsorship: This work was supported by the Wellcome Trust, UK, through grants to the Africa Centre Demographic Information System, the Africa Centre for Health and Population Studies and the Africa Centre HIV Surveillance programme (grants 050510, 065377, 65377, 50534).
1. Department of Health. National HIV and syphilis ante-natal sero-prevalence survey in South Africa 2004. Pretoria: South African Department of Health; 2005.
2. Shisana O, Rehle T, Simbayi LC, Parker W, Zuma K, Bhana A, et al. South African national HIV prevalence, HIV incidence, behaviour and communication survey, 2005. Cape Town: HRSC Press; 2005.
3. Pettifor AE, Rees HV, Kleinschmidt I, Steffenson AE, MacPhail C, Hlongwa-Madikizela L, et al. Young people's sexual health in South Africa: HIV prevalence and sexual behaviors from a nationally representative household survey. AIDS 2005; 14:1525–1534.
4. UNAIDS/WHO Working Group on Global HIV/AIDS and STI Surveillance. UNAIDS/WHO epidemiological fact sheets on HIV/AIDS and sexually transmitted infections, 2004 update – South Africa. Geneva: UNAIDS and WHO; 2004.
5. Hosegood V, Timæus IM. Household composition and dynamics in KwaZulu Natal, South Africa: mirroring social reality in longitudinal data collection, Chapter 4. In: van der Walle, editor. African households: an exploration of census data. New York: M.E. Sharpe Inc.; 2005. ISBN 0-7656-1619-X.
6. UNAIDS/WHO Working Group on Global HIV/AIDS and STI Surveillance. Guidelines for using HIV testing technologies in surveillance: selection, evaluation and implementation. Geneva: UNAIDS and WHO; 2001.
7. Welz T, Thabethe N, Radebe S, Herbst K. Broadening access to VCT: community-based counseling centers in a large HIV surveillance programme in rural KwaZulu-Natal, South Africa. In: XVth International AIDS Conference. Bangkok, Thailand, 11–16 July 2004 [Abstract TuPeD5150].
8. Welz T, Herbst J. A new model of HIV counseling and confidential results distribution in population-based HIV research using handheld computers. In: South African AIDS Conference. Durban, South Africa, 3–6 August 2003 [Oral presentation].
11. Lurie M. Migration and AIDS in Southern Africa: a review. South Africa J Sci 2000; 96:343–347.
12. Pison G, Le Guenno B, Lagarde E, Enel C, Seck C. Seasonal migration: a risk factor for HIV infection in rural Senegal. J Acquir Immune Defic Syndr 1993; 6:196–200.
13. Mbizvo MT, Machekano R, McFarland W, Ray S, Bassett M, Latif A, Katzenstein D. HIV seroincidence and correlates of seroconversion in a cohort of male factory workers in Harare, Zimbabwe. AIDS 1996; 10:895–901.
14. Nunn AJ, Wagner HU, Kamali A, Kengeya-Kayondo JF, Mulder DW. Migration and HIV-1 seroprevalence in a rural Ugandan population. AIDS 1995; 9:503–506.
15. Lurie MN, Williams BG, Zuma K, Mkaya-Mwamburi D, Garnett G, Sturm AW, et al. The impact of migration on HIV-1 transmission in South Africa: a study of migrant and nonmigrant men and their partners. Sex Transm Dis 2003; 30:149–156.
16. Zuma K, Gouws E, Williams B, Lurie M. Risk factors for HIV infection among women in Carletonville, South Africa: migration, demography and sexually transmitted diseases. Int J STD AIDS 2003; 14:814–817.
17. Brewer TH, Hasbun J, Ryan CA, Hawes SE, Martinez S, Sanchez J, et al. Migration, ethnicity and environment: HIV risk factors for women on the sugar cane plantations of the Dominican Republic. AIDS 1998; 12:1879–1887.
18. Calleja JM, Marum LH, Carcamo CP, Kaetano L, Muttunga J, Way A. Lessons learned in the conduct, validation, and interpretation of national population based HIV surveys. AIDS 2005; 19(Suppl. 2):S9–S17.
19. Boerma JT, Ghys PD, Walker N. Estimates of HIV-1 prevalence from national population-based surveys as a new gold standard. Lancet 2003; 362:1929–1931.
20. Joint United Nations Programme on HIV/AIDS and World Health Organisation Working Group on Global HIV/AIDS and STI Surveillance. Guidelines for measuring national HIV prevalence in population-based surveys. Geneva: WHO and UNAIDS; 2005.
21. Gregson S, Garnett GP. Contrasting gender differentials in HIV-1 prevalence and associated mortality increase in eastern and southern Africa: artefact of data or natural course of epidemics? AIDS 2000; 14(Suppl. 3):S85–S99.
22. Nunn AJ, Kengeya-Kayondo JF, Malamba SS, Seeley JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS 1994; 8:81–86.
23. Serwadda D, Wawer MJ, Musgrave SD, Sewankambo NK, Kaplan JE, Gray RH. HIV risk factors in three geographic strata of rural Rakai District, Uganda. AIDS 1992; 6:983–989.
24. Zaba B, Marston M, Isingo R, Urassa M, Ghys PD. How well do cross-sectional population surveys measure HIV prevalence? Exploring the effects of non-participation. In: XVth International AIDS Conference. Bangkok, Thailand, 11–16 July 2004 [Abstract LbOrC23].
25. Walker N, Grassly NC, Garnett GP, Stanecki KA, Ghys PD. Estimating the global burden of HIV/AIDS: what do we really know about the HIV pandemic? Lancet 2004; 363:2180–2185.
26. Joint United Nations Programme on HIV/AIDS. 2006 Report on the global AIDS epidemic. Geneva: UNAIDS; 2006.
27. Garcia-Calleja JM, Gouws E, Ghys PD. National population based HIV prevalence surveys in sub-Saharan Africa: results and implications for HIV and AIDS estimates. Sex Transm Infect 2006; 82(Suppl. 3):iii64–iii70.
© 2007 Lippincott Williams & Wilkins, Inc.