Up to date and accurate data on HIV prevalence and incidence are needed to monitor the progress of the HIV epidemic, to plan AIDS control programmes and to assess the impact of interventions . Many countries rely on HIV seroprevalence data collected from sentinel surveillance sites such as antenatal clinics (ANC) or blood banks [1–9]. ANC attenders have the advantage of being sexually active, easily definable and accessible, and are assumed to be fairly representative of the general population. Furthermore, it has been suggested that measuring HIV prevalence among young ANC attenders may be useful for monitoring the incidence of HIV infection in the general population . Routine collection of blood from ANC attenders for syphilis screening provides an opportunity to perform anonymous unlinked HIV testing using the same sample [1,6,11,12].
Conflicting results have been reported on the degree to which HIV prevalence among ANC attenders is comparable with that of women in the general population. A lower HIV prevalence was reported in ANC attenders than in women from the general population in Uganda, Tanzania and Zambia [4,8,13,14], but the opposite was reported in Ethiopia .
Such discrepancies may have various causes. First, the catchment population of ANC clinics is often ill-defined, and discrepancies with population prevalences may sometimes reflect a mismatch between the two populations . Second, pregnant women attending ANC services may be more health conscious and have different socio-economic characteristics from the pregnant women who do not attend . Third, women who are infertile, who are not sexually active or who use contraceptives will be under-represented in ANC clinics and the prevalence of HIV may be different in such women [4,6]. There is good evidence that women with HIV have lower fertility and are less likely to become pregnant, and hence will be under represented in ANC surveys .
Various methods have been suggested to correct the HIV seroprevalence rate in ANC attenders so that it more accurately reflects the prevalence in the general population [4,17]. Nicoll and colleagues derived relative inclusion ratios (RIR) to adjust for the relative probability of HIV-infected women attending ANCs . This method aims to account for all factors leading to discrepancies, but it may be difficult to collect these data in many developing countries.
Zaba and colleagues suggest a method for adjustment in which the HIV prevalence is determined separately for ANC attenders with and without children . A separate correction factor is applied to each group, based on the HIV prevalence ratio for ANC attenders versus women in the general population among parous and nulliparous women. This method was shown to work well in two African populations with low contraceptive use (in rural Uganda and in Mwanza town, Tanzania); but these two populations were used to derive the correction factors. In the present study, we have attempted to validate this method in an independent population in rural Mwanza Region.
Mwanza Region is on the southern shores of Lake Victoria, and has a population of approximately 3 million people. Between 1991 and 1994, a community-randomized trial was conducted in 12 rural communities. Detailed information on the study population and the design of the trial have been reported elsewhere [19,20]. Prior to the implementation of the programme, a baseline survey was conducted in a cohort of approximately 1000 adults aged 15–54 years randomly selected from each of the 12 study communities . In a separate study, approximately 100 sequential women attending ANCs were recruited in each of the 12 communities over a 2-week period . Two rounds were conducted, the first from April 1992 to January 1993 and the second continuing to December 1993.
All consenting subjects were interviewed, and those in the ANC survey had a full gynaecological examination. Subjects with confirmed sexually transmitted disease syndromes were treated according to national guidelines, as were those who were found to have a positive syphilis screening test.
Sera from the general population cohort were tested for HIV antibodies by ELISA (Vironostika HIV MIXT Microelisa; Organon, Boxtel, The Netherlands). A second independent ELISA was used for confirmation (GACELISA, Wellcozyme; Murex, Dartford, UK). Specimens with discrepant and indeterminate ELISA results were further tested using a Western blot test (HIV-1 Westernblot; Epitope, Beaverton, Oregon, USA). Sera from ANC attenders were tested for HIV antibodies in 1999, 6 to 7 years after the original survey. HIV testing procedures were identical, but UNIFORM II HIV 1+2 (Vironostika; Organon) was used instead of Vironostika HIV MIXT Microelisa, which was no longer available. A study on 200 sera stored under the same conditions confirmed that the results of the HIV tests performed 7 years later matched the original HIV results.
The prevalence of HIV from the ANC study was analysed (i) without adjustment and (ii) age standardized by applying the raw prevalences for each age band to the number of women seen in these age groups in the general population sample. HIV prevalences were then adjusted for parity using the correction factors given by Zaba et al. , but without use of data on birth interval. The observed prevalence in nulliparous ANC attenders was multiplied by 0.7, whereas that in parous ANC attenders was multiplied by 1.5. These numbers represent the estimated ratio of HIV prevalence in the general population of women aged 15–44 years to that in ANC attenders among nulliparous and parous women respectively. Finally, adjusted prevalences in these two groups were used to obtain an overall estimate of prevalence, based on the expected proportions of nulliparous and parous women aged 15–44 years in the general population (25 and 75% respectively). The variances of the parity-specific estimates were obtained by multiplying the corresponding binomial variances by 0.72 and 1.52 , and the variance of the overall estimate was obtained as the variance of the weighted average, thus enabling a test of the significance of the difference between adjusted ANC and general population prevalences to be performed. Analyses were performed using STATA 5.0 software (Stata Corp., College Station, Texas, USA).
A total of 5675 women aged 15–44 years were included in the general population study, representing approximately 85% of those eligible . In the ANCs, 2265 women (1048 in round 1 and 1217 in round 2) were included. Women attending ANC were younger (median age 24 versus 26 years, P < 0.001), more likely to be married (88 versus 72%, P < 0.001) and more likely to have been educated to standard 4 or beyond (62 versus 54%, P < 0.001) than women in the general population.
HIV prevalences in women in the general population and in women attending ANC are shown by age in Table 1. The peak HIV prevalence was in the 25–29 years age group, being 6.9% for women in the general population and 4.8% for women attending ANCs. The overall HIV prevalence in women attending ANC [3.6%; 95% confidence interval (CI), 2.8–4.4%] was significantly lower (P = 0.025) than in the general population (4.7%; 95% CI, 4.2–5.3%). HIV prevalence was lower in ANC attenders in every age group except the 15–19 years group, although these differences were not statistically significant.
Age standardization of the HIV prevalence in women attending ANC to the age distribution of women in the general population gave an overall HIV prevalence of 3.2% in women aged 15–44 years. Thus, the age standardized HIV prevalence for women attending ANC was 32% lower than the prevalence observed in the general population. Using the adjustment based on parity gave an adjusted HIV prevalence in ANC attenders of 4.6%, which was not significantly different from that observed in the general population (4.7%;P = 0.95).
Figure 1 illustrates the strength of the correlation between HIV prevalences in ANC attenders and in the general female population in the 12 communities. The prevalence in ANC attenders was highly correlated with the prevalence in the general population (R = 0.76;P < 0.005).
The overall prevalence of HIV in ANC attenders was lower than in the general population, as was found in other studies in Mwanza, and in Uganda and Zambia [4,8,13,14]. The opposite was found in the Ethiopian study , perhaps because of a 2-year interval between data collection from the general population (1994) and ANC (1996) surveys.
In the youngest age group (15–19 years), the prevalence of HIV was higher in the ANC attenders than in women from the general population. Although sexual behaviour data were not collected in these studies, in this population not all women aged 15 to 19 years are sexually experienced. The prevalence of HIV in this age group in the general population includes data from both sexually experienced women and from virgins. By definition women attending ANC are sexually active and on sexual debut young women are exposed to pregnancy, HIV and other sexually transmitted diseases.
In the older age groups the prevalence of HIV was lower in ANC attenders than in women from the general population. HIV prevalence in older women represents the cumulative total of infection over a longer period of time. It has been suggested that as HIV infection progresses in women there is a loss of fertility in HIV-positive women compared with HIV-negative women . This effect would be greater in women in the older age groups.
The method of adjustment proposed by Zaba et al. takes into account the high HIV prevalence in childless ANC attenders relative to childless women in the general population, and the low HIV prevalence in parous ANC attenders relative to women with children in the general population observed in African populations with low contraceptive use. The close agreement between the observed HIV prevalence in the general population, and the adjusted prevalence in ANC attenders in the present study suggests that this method is generally valid for such populations, and confirms the importance for surveillance purposes of collecting information on parity from women attending ANC. However, it can only be applied to populations in which the proportion of childless and parous women is known, and correction on the basis of parity may be less useful in areas where contraceptive use is high or changing.
Our findings indicate that in rural Tanzania, unadjusted estimates of HIV prevalence based on ANC attenders are likely to underestimate prevalences in women aged 15–44 years in the general population. Others have argued that if such selection biases remain the same over time, serial data from ANC sentinel surveillance will provide a reliable basis for analysis of HIV trends in the population. However, selection biases may change over time, in which case trends recorded from ANC surveillance data become more difficult to interpret.
Our study, conducted in rural Mwanza at one point in time, has shown that adjustment of ANC data using parity-based correction factors provided an accurate estimate of population prevalence. Further data are needed to examine the validity of this method in different populations, and at different stages of the HIV epidemic.
We wish to thank the Permanent Secretary, Ministry of Health, the Programme Manager of the National AIDS Control Programme (NACP) and the Director General of the National Institute for Medical Research (NIMR), Tanzania for permission to carry out and publish the results of this study. We extend our appreciation to the team leaders and field staff of the two studies for their assistance in field data collection. We also thank the laboratory staff for their help with laboratory testing. We are grateful to Dr. Basia Zaba for helpful discussion of the results and their interpretation.
1. UNAIDS/WHO. Report on the global HIV/AIDS epidemic
. Geneva: UNAIDS; 1997.
2. Chin J. Public health surveillance of AIDS and HIV infection. Bull WHO 1990, 68: 529–536.
3. Berkley S, Naamara W, Okware S. et al
. AIDS and HIV infection in Uganda: are more women infected than men? AIDS 1990, 4: 1237–1242.
4. Zaba B, Carpenter L, Boerma T. et al
. Adjusting antenatal clinic data for improved estimates of HIV prevalence among women in sub-Saharan Africa. AIDS 2000, 14: 2741–2750.
5. Mulder D, Nunn A, Kamali A. et al
. Decreasing HIV-1 seroprevalence in young adults in a rural Ugandan cohort. BMJ 1995, 311: 833–836.
6. Jackson D, Ngugi E, Plummer F. et al
. Stable antenatal HIV-1 seroprevalence with high population mobility and marked seroprevalence variation among sentinel sites within Nairobi, Kenya. AIDS 1999, 13: 583–589.
7. Wawer M, Serwadda D, Gray R. et al
. Trends in HIV-1 prevalence may not reflect trends in incidence in mature epidemics: data from the Rakai population-based cohort, Uganda. AIDS 1997, 11: 1023–1030.
8. Fylkenes K, Ndlovu Z, Musonda R. et al
. Studying dynamics of the HIV epidemic: population based data compared with sentinel surveillance in Zambia. AIDS 1998, 12: 1227–1234.
9. National AIDS Control Programme. HIV/AIDS/STD surveillance. Report No. 13
. Dar es Salaam, Tanzania: Ministry of Health; 1999.
10. Zaba Z, Boerma T, White R. Monitoring the AIDS epidemic using HIV prevalence data among young women attending antenatal clinics: prospects and problems. AIDS 2000, 14: 1633–1645.
11. Ades A, Parker S, Berry T. et al
. Prevalence of maternal HIV-1 infection in Thames regions: results from anonymous unlinked neonatal testing. Lancet 1991, 337: 1562–1565.
12. Gwin M, Pappaioanou M, George JR. et al
. Prevalence of HIV infection in childbearing women in the United States.Surveillance using newborn blood samples.
JAMA 1991, 265: 1704–1708.
13. Kigadye R, Klokke A, Nicoll A. et al
. Sentinel surveillance for HIV-1 among pregnant women in a developing country: 3 years’ experience and comparison with a population survey. AIDS 1993, 7: 849–855.
14. Borgdorff M, Barongo L, Van Jaarsveld E. et al
. Sentinel surveillance for HIV-1 infection: how representative are blood donors, outpatients with fever, anaemia, or sexually transmitted diseases, and antenatal clinic attenders in Mwanza Region, Tanzania? AIDS 1993, 7: 567–572.
15. Fontanet A, Messele T, Dejene A. et al
. Age- and sex-specific HIV-1 prevalence in the urban community setting of Addis Ababa, Ethiopia. AIDS 1998, 12: 315–322.
16. Zaba B, Gregson S. Measuring the impact of HIV on fertility in Africa. AIDS 1998, 12 (suppl. 1): S41–S50.
17. Carpenter L, Nakiyingi J, Ruberantwari A, Malamba S, Kamali A, Whitworth J. Estimates of the impact of HIV infection on fertility in a rural Ugandan population cohort. Health Transition Review 1997, Suppl. 7 : 113–126.
18. Nicoll A, Stephenson J, Griffioen A. et al
. The relationship of HIV prevalence in pregnant women to that in women of reproductive age: a validated method for adjustment. AIDS 1998, 12: 1861–1867.
19. Hayes R, Mosha F, Nicoll A. et al
. A community trial of the impact of improved STD treatment on the HIV epidemic in rural Tanzania: 1.Design.
AIDS 1995, 9: 916–926.
20. Grosskurth H, Mosha F, Todd J. et al
. A community trial of the impact of improved STD treatment on the HIV epidemic in rural Tanzania: 2.Baseline survey results.
AIDS 1995, 9: 927–934.
21. Mayaud P, Grosskurth H, Changalucha J. et al
. Risk assessment and other screening options for gonorrhoea and chlamydial infections in women attending rural Tanzanian antenatal clinics. Bull WHO 1995, 73: 621–630.
22. Gray R, Wawer M, Serwadda D. et al
. Population based study of fertility in women with HIV-1 infection in Uganda. Lancet 1998, 351: 98–103.