The course of the HIV-1 epidemic needs to be monitored to assess the possible impact of AIDS control programmes and interventions and provide the basis for necessary adjustment and future planning. Conducting studies to estimate the incidence of HIV-1 is the best way to monitor HIV-1 epidemic trends. However, these studies are difficult to perform, very costly and only feasible in the frame of scientific studies . Therefore, HIV-1 prevalence data continue to be the most relevant source of information. Anonymous unlinked serial surveillance of sentinel groups and sites is widely used, with pregnant women visiting antenatal clinics representing the preferred group [2–5].
In this article we present the results from antenatal clinic-based serial HIV-1 sentinel surveillance conducted in four different geographic areas of Mbeya region in south-west Tanzania from 1988 until 2000 and Rukwa region in the west of the country from 1991 until 1999 in order to access the continued evolution of the epidemic since 1988 and the diverse patterns of spread. We examine changes in indicators for sexual behaviour and biological risk factors to assess whether these changes, possibly resulting from the implemented prevention programme could have contributed to the observed decline in new HIV-1 infections. The comparison between data from areas with an ongoing comprehensive programme and an area with limited interventions should assist in evaluating the impact of the programme.
Since its introduction in 1988 the AIDS Control Programme in Mbeya region in Tanzania has developed into a unique record of the HIV-1 epidemic. It consists of the serial sentinel surveillance of HIV-1 prevalence and syphilis in antenatal clinic attendees and the data have been collected continuously from 1988 to date. Following a request from the Ministry of Health, sentinel surveillance of HIV-1 prevalence in antenatal clinic attendees has also been performed since 1991 in the neighbouring Rukwa region. A national sentinel surveillance system is not in place and data from other areas of Tanzania are lacking.
The data were collected in Mbeya region in south-west Tanzania. The region is densely populated with 2 million inhabitants and borders Malawi and Zambia. A highway crosses the region, connecting Dar-es-Salaam with southern African countries. The majority of the population are farmers. Fifteen percent of the population lives in the regional capital. There are a few industrial enterprises, coffee and tea farms. The eight districts of the region enjoy a well-developed physical health infrastructure.
Since late 1988, a district-based AIDS control programme has been implemented and supported as part of the Tanzanian–German technical co-operation. Activities implemented in the context of the regional AIDS control programme aim at influencing factors relevant for the further spread of HIV-1 [6–9]. The main components of the programme are interventions to promote behavioural change including a peer education programme in primary schools, work-place programmes, a peer education programme involving sex workers and their clients, theatre for development performances in the communities and development of information material.
Condom use is promoted and social marketing of condoms is supported all over the region in both urban and rural areas. At all health facilities well-trained health workers offer STD case management. Routine screening and treatment of pregnant women and their partners for syphilis is a component of STD control. Community-based voluntary counselling and HIV testing is offered, and home-care teams support patients living with AIDS. The laboratory allows reliable testing, ensures blood safety and the realization of operational research . Interventions are carried out continuously and cover all eight districts of the region to the same extent.
HIV-1 prevalence data were also collected in the neighbouring Rukwa region in the west of Tanzania with 1 117 500 inhabitants. The region is divided into four districts. Eight percent of the population lives in the regional capital Sumbawanga . The main source of income is agricultural production and fishing activities. The latter includes intensive travelling and life in seasonal fishing camps. Unlike Mbeya there is no major development of private companies and no major cash crops are produced, making it a poorer area than Mbeya region. Limited resources and no external support have prevented the implementation of a variety of prevention activities continuously in this region.
Antenatal clinic-based sentinel surveillance
In 1988, a sentinel surveillance system for HIV-1 infection and syphilis in pregnant women was developed for Mbeya region, following the internationally recommended procedures [12,13]. Nine sentinel sites were selected to reflect different geographic areas and connected risk factors, and to provide precise information about patterns in spread of the HIV-1 infection. Three urban sites are located in the town of Mbeya. The road connecting Southern African countries with the harbours of Dar-es-Salaam and Mombasa passes through the town, making it a common stopover for truck drivers. The growing number of private companies increasingly attracts the rural population. The rural area is represented by three remotely situated hospitals serving an isolated population. Two roadside sites in which bars and guesthouses for truck drivers attract many women to earn their money as bar and sex workers were chosen. Sexual networking between the mobile population of the sites and the neighbouring villages might put the village population at risk. A site in a town bordering Malawi was selected to represent the special circumstances of a border town.
All first attendees for antenatal care at the selected sites were enrolled until a sample size of 200 was reached, at 6-month intervals between 1988 and 1993, and 12-month intervals since 1994. Unlinked anonymous HIV-1 testing was carried out for blood samples collected for routine procedures. Sera were tested for syphilis at the site using the venereal disease research laboratory (VDRL) slide flocculation test (Murex Diagnostics Ltd., Dartford, Kent, UK) until 1994, when the methodology was changed to a ‘rapid plasma reagin’ test PLASMATEC (Plasmatec Laboratory Products Ltd, Bridport Dorset, UK). Women were informed about the result of the syphilis screening and treated accordingly.
Information collected included age, parity and location of residence of the women. Sera were tested as a batch at the laboratory of the Mbeya Referral Hospital using an enzyme-linked immunoassay (ELISA) test. Up to October 1994 this was the Enzygnost Anti-HIV1+ HIV2 test and thereafter the Enzygnost Anti-HIV-1/2 test (BehringwerkeAG, Marburg, Germany) was used. As the testing was anonymous and unlinked, HIV-1 serostatus was based on a single ELISA test [2,14]. The laboratory repeated the syphilis tests using the Treponema pallidum haemaglutination test (TPHA; Shield Diagnostics, Luna Parc, Dundee, Scotland). Individuals who had positive results in both tests were classified as having active syphilis.
Antenatal clinic-based sentinel surveillance was also supported annually between 1991 and 1999 in the neighbouring Rukwa region and included one urban site in Sumbawanga, the regional capital.
Statistical analysis was performed using the Statistical Package for Social Sciences (SPSS 10.0 for Windows; SPSS Inc., Chicago, Illinois, USA) and MS-Excel (Microsoft Office 1997; Microsoft Corporation, Redmond, Washington, USA).
Women were grouped according to age in three groups from 15–24, 25–34 and 35–40 years. Women aged < 15 years (n = 26) and > 40 (n = 118) were excluded from the analysis because of the small number represented. Age-specific HIV-1 prevalence rates and binominal 95% confidence intervals (CI) were calculated for antenatal clinic attendees of the four different demographic strata: urban, rural, roadside and border and pooled data from all sites. The statistical significance of the trend in prevalence over time was estimated using the non-parametric trend test of Mann  and the Kendall statistic. Prevalence from 1988 to 2000 was adjusted within age groups using direct standardization, with the total sample serving as the standard population. Rate ratio (RR) and Taylor 95% CI for RR was used to test differences in proportions between years and strata.
Monitoring factors influencing spread of HIV-1
The programme uses indicators, which are simple and easy to verify in order to avoid high costs and ensure sustainability of the monitoring system. As an indicator of condom use and change to safe sexual behaviour, since 1995 districts summarize and report monthly condom sales figures and the number of condoms distributed free of charge.
All health facilities document the monthly attendance rate of patients with a STD. An increasing number of patients in STD clinics with the increasing accessibility of the service and its promotion at a community level might indicate an awareness of the importance of treatment and, as a consequence, an increase of treatment and reduction in the number of people with untreated STDs, which are well known as a factor facilitating the spread of HIV-1.
Two cross-sectional studies were performed in 1995 and 1999 to assess possible changes of behaviour and knowledge of primary school children exposed to a peer education programme, which started in 1996. The study population were 12- to 15-year-old pupils, selected through stratified random cluster sampling. Interviews were performed using self-administered questionnaires. [Jordan-Harder B: Knowledge about reproductive health, sexual behaviour and attitudes of 12 to 15 year olds in school youth in Mbeya Region/Tanzania; Lwihula GK, Kwesigabo G: Assessment of knowledge and changes in behaviour in relation to reproductive health among groups receiving interventions in Mbeya Region/Tanzania. Articles are available from the authors]. Among the indicators measured (see Table 2) were onset of sexual activity, level of knowledge about HIV/AIDS, STDs and other areas of reproductive health and ability to openly discuss reproductive health matters. The sample sizes were 500 in 1995, and 1317 for the intervention and 1085 for the non-intervention groups in 1999.
Between October 1988 and December 2000, a total of 23 308 pregnant women aged 15 to 40 years were included in the antenatal clinic-based sentinel surveillance. Of these 9333 were enrolled in the urban stratum, 6534 n the rural stratum, 5098 in the roadside stratum and 2343 in the border stratum.
The median age of the unstratified sample and the urban, rural and roadside strata was 23 years [interquartile range (IQR), 20–28 years]; the median age for the border stratum was 22 years (IQR, 19–27 years). In Rukwa region 2030 pregnant women aged 15 to 40 years were tested between 1991 and 1999, with data missing for 1992 and 1996. The median age was 23 years (IQR, 20–28 years).
The overall age-adjusted HIV-1 prevalence rate increased significantly from 1988 until 1994/1995 for all strata and for pooled data (nine sites) (Table 1). The border and roadside strata showed the most prominent increase of prevalence reaching as high as 36.1% in the border stratum. This development was followed by a decline for all strata, which was only significant for the pooled data and the border stratum.
More pronounced changes were found for 15- to 24-year-old women. The age-specific prevalence rates increased until 1994/95 and showed a significant downward trend from 1994/95 to 2000 for all strata and pooled data. The prevalence ratios for 1988/1994 were significantly higher than those for 2000/1995. For Rukwa an increase of the prevalence rate was estimated until 1994. This trend continued until 1999.
In Mbeya, the highest relative increase was estimated for the rural stratum, in which the prevalence reached 15.6% (P = 0.03) (Fig. 1). The highest absolute prevalence occurred in the border stratum with 36.1% (95% CI, 27.9–44.3%). The prevalence ratio 1988/1994 varied between 7.9 (95% CI, 1.88–33.18) for the rural and 2.47 (95% CI, 1.49–4.09) for the border stratum. The prevalence of the pooled data declined significantly to 14.6% in 2000 (P = 0.001). Furthermore, for the border stratum a significant downward trend was estimated with the prevalence reaching 19.4% (P < 0.05). For the roadside stratum the prevalence was estimated to be 12.5% (P = 0.03) after it had reached 26.7% (95% CI; 21.2, 32.3) in 1994. The lowest prevalence was found for the rural stratum with 8.8% (P < 0.05). After a significant decrease until 1999 the prevalence increased again slightly for the urban stratum. Therefore the downward trend was no longer significant for the urban stratum.
The prevalence ratio decreased to below 1 for all strata with the exception of the urban stratum and Rukwa (Fig. 2). In Rukwa a continuous increase was observed accompanied by an increase of risk. Comparing the relative risk of 15- to 24 year-old women in Mbeya and Rukwa in 1999, women living in Mbeya were found to be at a lower risk with RR, 0.52 (95% CI, 0.37–0.74).
Factors influencing spread of HIV-1
A proxy indicator with which to measure condom use and behaviour change is the number of condoms sold. Condom sales at 2278 selling points in urban and rural areas alike have increased continuously since 1995 by 10% per year and, are the highest in the country after the city of Dar-es-Salaam. A total of 2 194 000 condoms were sold during 2000, roughly one condom per capita. In Rukwa region, however, only 125 568 condoms were sold, or 0.11 per capita in the year 2000 .
As a rough estimate, at least 5% of the sexually active population of Tanzania contracts a STD every year and needs treatment. In 1999 84% of this estimated proportion suffering from a STD received professional treatment in health facilities in Mbeya region . Syphilis prevalence among pregnant women attending antenatal clinics declined significantly from 14.8% in 1989 to 4.1% in 2000 (P = 0.0001).
Studies conducted among primary school pupils 12 to 15 years old revealed a significantly higher level of knowledge about issues related to HIV/AIDS, STDs and other areas of reproductive health in 1999 in comparison with 1995 (Table 2). This difference was also found between the intervention and non-intervention groups in 1999. Communication about topics connected with sexual reproductive health has improved with 50.9% of pupils answering that they talk often about these issues compared with 35.5% of pupils in the non-intervention group (P = 0.0001).
In 1999 a total of 21.2% of pupils enrolled in the intervention group and 23% of those in the control group were sexually active. Comparing the control and intervention groups, first sexual intercourse was significantly delayed from 10.6 to 13.2 years of age (P = 0.003) for this sub-sample of sexually active pupils. The mean age of first sexual engagement for boys was 10.8 and 13.1 years and for girls 10.1 and 12.4 years, respectively.
The antenatal clinic-based sentinel surveillance data revealed a significant trend of declining HIV-1 prevalence for 15- to 24-year-old women since 1994 for all strata. A similar age-specific decline has been observed in some areas of Uganda, Zambia and the Kagera region in Tanzania. [18–20]. A diversity of patterns in the spread of the infection existed for the different geographic areas.
The age-specific prevalence rate among women aged 15–24 years in the border stratum was already 16.2% in 1989. At the same time the prevalence of this group for the urban stratum, where the highest risk is to be expected, was 13.4%. In the rural and roadside strata, rates of only 9.4 and 8.4%, respectively, were recorded.
The prevalence peaked for all strata in 1994/95, with a very high level of 36.1% recorded for the border stratum. Different to the other strata, the prevalence trend until 1995 was more comparable with that of pregnant women surveyed in Malawi . Besides bordering Malawi the town hosting the site is situated at Lake Malawi and people work in rice fields. This situation might contribute to a high prevalence of schistosomiasis haematobium infection leading to erosion and ulceration of the epithelium of the genital tract associated with an increased risk of HIV-1 transmission [22–24].
The prevalence rate for women in the urban stratum, who were expected to be at the highest risk as described in other countries [5,25], reached a rate of 20.6% in 1995. A lower prevalence rate was identified for women of the rural stratum, which only increased to 15.6%, whereas the prevalence in women belonging to the roadside stratum, close to busy locations with bars and guesthouses, increased to 26% in 1994 but dropped thereafter to the level of the women of the rural stratum.
The declining trend in HIV-1 prevalence in young women between 15 and 24 years of age was most pronounced for the women attending the roadside site, with a 54% reduction, followed by declines at the rural and border sites by 43 and 46%, respectively. Women in the urban centre, however, showed the lowest reduction with 20%. These findings differed from those in Zambia, where the decline was most pronounced in the urban area .
These different patterns occurred although the prevention efforts of the regional programme reached all places with a similar intensity. Populations might be exposed to different risk but also respond differently to prevention activities for reasons not yet clearly understood. A high prevalence of tropical parasitic diseases might also contribute to observed differences in HIV-1 prevalence.
Incidence and mortality determine the HIV-1 prevalence.
A number of questions arise, such as whether the described decline of prevalence can be interpreted as a reduction in incidence connected to behavioural change; whether the implemented control programme has contributed to the change; and whether the estimated prevalence represents that of this age group in the general population.
Prevalence in women aged 15–24 years probably reflects incidence, because the infection must have occurred recently and mortality remains relatively low. A study carried out in neighbouring Uganda revealed no significant difference in HIV-1 prevalence estimates for women participating in sentinel surveillance and a population-based survey when comparing women in the 15–19 and 20–24 years age groups . Similar studies in Kagera region of Tanzania and in Zambia have confirmed this finding [20,26].
Selection bias due to low attendance rates associated with the poor economic situation of women or difficulties in reaching such a service can be discounted for the data presented. In Tanzania 98% of all pregnant women attend antenatal clinics . Antenatal service is offered in all health facilities free of charge and good access is provided equally in urban and rural areas.
The assumption that fertility in young women between the ages of 15 and 24 years is either little or not reduced because they are more likely to have contracted an HIV-1 infection recently  is supported by the data on fertility rates in Tanzania. The age-specific fertility rate (per 1000 women) of women aged 15 to 24 years remained stable or even increased (15–19 years, 1988, 106; 1999; 138; 20-24 years, 1988, 280; 1999, 268) . For the data presented, the selection bias due to a reduced fertility rate is therefore minimized for the sub-sample of young women aged 15–24 years.
The results of a mathematical model used by Kilian et al. comparing age-specific trends of ANC-based data and simulating behavioural change versus no changes strongly support the assumption that the monitored decline of prevalence among women in the young age group reflects change in incidence based on behavioural change . Young people tend to adopt changes in behaviour more rapidly than older persons. Consequently the incidence of HIV-1 infection in younger people is lower if preventive programmes are effective .
Indicators monitored continuously to judge the effectiveness of implemented activities revealed a change of behaviour taking place in parallel with the decline of prevalence. Condom sales interpreted as being similar to use of condoms in an income-poor country have increased in all areas of the region, whereas those reported for all other regions of the country were much lower. The knowledge, behaviour and attitude of adolescents have changed significantly in directions recognized as being the most effective in reducing the incidence of HIV-1 infection . The extensive treatment of persons suffering from STDs might have contributed to a gradual reduction of persons with STDs as an important factor facilitating HIV-1 transmission and contributing to reduced incidence [30,31].
As a result of the activities associated with the programme, services of good quality were accessible and awareness of the importance of treatment was created. Condoms were made available in urban and rural areas alike, so that the population could use them after having been convinced of the importance of protected sex. School- and community-based activities might have contributed to the changes observed in adolescents.
These are all of course only indirect indicators hinting at processes, which are necessary for behavioural change and reduction of incidence. In Uganda and the former Zaire the combination of strategies used has also been shown to be effective in reducing the incidence [29,32].
Based on the results of a multi-centre study of factors determining HIV-1 prevalence a few strategies are recommended to successfully prevent the further spread of the epidemic [6,18]. These are mainly effective management of other treatable STDs, interventions to change behaviour of adolescents at an early age to delay onset of sexual activities, interventions among sex workers and condom promotion. All these strategies are implemented in Mbeya.
The probability that the implemented programme has contributed to the mentioned changes is supported by the trend of the HIV-1 prevalence in 15- to 24-year-old women in the neighbouring Rukwa region. HIV-1 prevalence has increased significantly between 1991 and 1999. No specific factors are known that could explain this difference. Male circumcision is not common in either of the two regions. No civil unrest with the known consequences existed. However, unlike Mbeya, only few and sporadic control activities were implemented in Rukwa region.
Taken together, the change of different indirect indicators for behavioural change and the reduction of a biological factor relevant for HIV transmission, an influence on HIV incidence appears plausible. These changes were partly possible as a consequence of the activities implemented by the regional programme. The data are important but not yet sufficient to fully understand the development of the epidemic in the different strata. Further information is needed to describe more precisely the change of behaviour of the population of the different strata and possible underlying social and environmental determinants of risk to adjust preventive efforts accordingly.
The authors would like to thank their colleagues working in the Mbeya Referral Hospital's HIV laboratory and at the regional office of the Ministry of Health for their support in this work.
1. UNAIDS in collaboration with Wellcome Trust Centre for the Epidemiology of Infectious Diseases: Trends in HIV Incidence and Prevalence.
Geneva, Switzerland: UNAIDS; 1999.
2. Chin J. Public health surveillance of AIDS and HIV infections.Bull WHO
3. Fylkesnes K, Musonda R M, Sichone M, Ndhlovu Z, Tembo, F. Declining HIV prevalence and risk behaviours in Zambia: evidence from surveillance and population-based surveys.AIDS
4. Kigadye R, Klokke A, Nicoll A. Nyamuryekungè KM, Borgdorff M, Barongo L, et al
. Sentinel surveillance for HIV-1 among pregnant women in a developing country: 3 years’ experience and comparison with a population serosurvey.AIDS
5. Fylkesnes K, Ndhlovu Z, Kasumba K, Musonda RM, Sichone M. Studying dynamics of the HIV epidemic: population-based data compared with sentinel surveillance in Zambia.AIDS
6. Buve A, Carael M, Hayes RJ, Auvert B, Ferry B, Robinson NJ, et al
. Multicentre study on factors determining differences in rate of spread of HIV in four African cities: summary and conclusionsAIDS
7. Ferry B, Carael M, Buve A, Auvert B, Laoutou L, Kanhonou L, et al
. Comparison of key parameters of sexual behaviour in four African urban populations with different levels of HIV infection.AIDS
8. Buve A, Weiss HA, Laga M, van Dyck E, Musonda R, Zekeng L, et al
. The epidemiology of trichomonas in women in four African cities.AIDS
9. Fleming DT, Wasserheit JN. From epidemiological synergy to public health policy and practice: the contribution of other sexually transmitted diseases to sexual transmission of HIV infection.Sex Transm Infect
10. Jordan-Harder B, Koshuma Y, Pervilhac C, Vogel U. Hope for Tanzania: Lessons Learned from a Decade of Comprehensive AIDS Control in Mbeya Region
, Part I; Eschborn: Deutsche Gesellschaft für Technische Zusammenarbeit; 2000. Dar es Salaam: November 2000.
11. Bureau of Statistics, Ministry of Finance, Economic Affairs and Planning. Preliminary report, National Population Census 1988.
Dar es Salaam: Government Printer; 1988.
12. World Health Organisation/global Programm on AIDS: Unlinked Anonymous Screening for the Public-Health Surveillance of HIV Infection - Proposed International Guidelines.
Geneva: WHO; June 1989 (GPA/SFI/89.3).
13. Heptonstall J, Gill ON: The legal and ethical basis of unlinked anonymous HIV seroprevalence monitoring.Comun Dis Rep
14. World Health Organisation/Global Programme on AIDS: Recommendation for the selection and use of HIV-antibody tests.Weekly Epidemiol Rec
15. Hartung J, Elpelt B, Kloesner K-H. Statistik: Lehr-und Handbuch der angewandten Statistik,
6.Aufl., Muenchen, Wien, Oldenbourg: Carl Hanser Verlaq; 1987.
16. Population Service International. Salama Condoms Sales Performance, Coverage per Zones, per Capita 1999 and Year 2000 Sales Targets
. Dar es Salaam; Population Service International; 2000.
17. NACP reports Ministry of Health: National AIDS Control Programme HIV/AIDS/STD Surveillance
, Report No 14, 1999, Dar es Salaam: Government Printer; 2000
18. Asiimwe Okinoror G, Opio A, Musinguzi J. Change in sexual behaviour and decline in HIV infection among young pregnant women in urban Uganda.AIDS
19. Kwesigabo G, Killewo J, Godoy C, Urassa W, Mbena E, Mhalu F, et al
. Decline in the prevalence of HIV-1 infection in young women in the Kagera region of Tanzania.J Acquir Immune Defic Syndr Hum Retrovirol
20. Fylkesnes K, Musonda R, Sichone M, Ndhlovu Z, Tembo F, Monze M, et al
. Declining HIV prevalence and risk behaviours in Zambia: evidence from surveillance and population-based surveys.AIDS
21. Taha T, Dallabetta G, Hoover D. Trends of HIV-1 and sexually transmitted diseases among pregnant and postpartum women in urban Malawi.AIDS
22. Harms G, Feldmeier H. Review: HIV infection and tropical parasitic diseases-deleterious interactions in both directions?Trop Med Int Health
23. Poggensee G, Kiwelu I, Weger V, Goeppner D, Diedrich T, Krantz I, et al
. Female genital schistosomiasis of the lower reproductive tract: prevalence and schistosome-related morbidity.J Infect Dis 2000;181
24. Mabey D. Interaction between HIV infection and other sexually transmitted diseases.Trop Med Int Health
25. Fabiani M, Accorsi S, Lukwiya M. Trend in HIV-1 prevalence in an antenatal clinic in North Uganda and adjusted rates for the general female population.AIDS
26. Kwesigabo G, Killewo J, Urassa W, Mbena E, Mhalu F, Lugalia JLP, et al
. Monitoring of HIV-1 infection prevalence and trends in the general population using pregnant women as a sentinel population: 9 Years experience from the Kagera Region of Tanzania.J Acquir Immune Defic Syndr
27. National Bureau of Statistics and Macro International Inc. 2000. Tanzania Reproductive and Child Health Survey 1999.
Dar es Salaam: November 2000.
28. Zaba B, Gregson S. Measuring the impact of HIV on fertility in Africa.AIDS
29. Kilian A, Gregson S, Ndyanabangi B, Walusaga K, Kipp W, Sahlmueller G, et al
. Reduction in risk behaviour provide the most consistent explanation for declining HIV-1 prevalence in Uganda.AIDS
30. Grosskurth H, Mosha F, Todd J. Impact of improved treatment of sexually transmitted diseases on HIV infection in Tanzania: results of a randomized, controlled trial.Lancet
31. Mayaud P, Mosha F, Todd J. Improved treatment services significantly reduce the prevalence of sexually transmitted diseases in rural Tanzania: results of a randomized controlled trial.AIDS
32. Laga M, Alary M, Nzila N. Condom promotion, sexually transmitted disease treatment, and declining incidence of HIV-1 infection in female Zairian sex workers.Lancet