Trends in the uptake of voluntary counselling and testing for HIV in rural Tanzania in the context of the scale up of antiretroviral therapy

Objectives To describe trends in voluntary counselling and testing (VCT) use and to assess whether high-risk and infected individuals are receiving counselling and learning their HIV status in rural Tanzania. Methods During two rounds of linked serological surveys (2003–2004 and 2006–2007) with anonymous HIV testing among adults, VCT was offered to all participants. The crude and adjusted odds ratios for completing VCT in each survey were calculated to compare uptake by demographic, behavioural and clinical characteristics, stratified by sex. Repeat testing patterns were also investigated. Results The proportion of participants completing VCT increased from 10% in 2003–2004 to 17% in 2006–2007, and among HIV-infected persons from 14% to 25%. A higher proportion of men than women completed VCT in both rounds, but the difference declined over time. Socio-demographic and behavioural factors associated with VCT completion were similar across rounds, including higher adjusted odds of VCT with increasing numbers of sexual partners in the past 12 months. The proportion having ever-completed VCT reached 26% among 2006–2007 attendees, with repeat testing rates highest among those aged 35–44 years. Among 3923 participants attending both rounds, VCT completion in 2006–2007 was 17% among 3702 who were HIV negative in both rounds, 19% among 124 who were HIV infected in both rounds and 22% among 96 who seroconverted between rounds. Conclusion VCT services are attracting HIV-infected and high-risk individuals. However, 2 years after the introduction of antiretroviral therapy, the overall uptake remains low. Intensive mobilisation efforts are needed to achieve regular and universal VCT use.


Introduction
In sub-Saharan Africa, where the HIV epidemic has had its most devastating impact, the expansion of voluntary HIV testing and counselling services (VCT) is seen as a key component of public health efforts to lower HIV incidence. VCT aims to provide people with an opportunity to learn and accept their HIV status in a confidential environment, with counselling to encourage sexual behaviour change to prevent further infections (WHO 2005). The scaling up of VCT services could contribute towards reducing HIV transmission if counselling influenced the behaviour of HIV-positive individuals, including those who are sexually active and not yet ill, and HIV-negative individuals with high-risk behaviours.
VCT is also the main entry point for HIV treatment and care services (De Cock et al. 2003;Nsigaye et al. 2009). In many countries, HIV testing uptake has grown rapidly with the expansion of antiretroviral treatment (ART) -in Tanzania, the proportion ever tested increased from 15% to 32% between 200432% between and 200732% between (TACAIDS 200532% between , 2008. However, trends in VCT testing rates among HIV-infected individuals at the population level have rarely been reported, because there are very few settings in which an individual's HIV status is known prior to VCT. Furthermore, understanding the extent to which VCT already attracts recent seroconverters could provide useful information for the potential development of 'treatment as prevention' programs, which would need to identify persons with recent infection to start on ART. Re-use of this article is permitted in accordance with the Terms and Conditions set out at http://wileyonlinelibrary.com/online open#OnlineOpen_Terms We have previously reported socio-demographic, behavioural and clinical characteristics associated with VCT use among residents of a HIV cohort study in rural Tanzania 2003Tanzania -2004, shortly before free ART services (Wringe et al. 2008). The study showed that although HIV infection and certain high-risk behaviours prompted VCT use, only 10% of individuals completed VCT. VCT was offered free of charge on a fully voluntary basis and was available to study participants in their village of residence, suggesting that VCT uptake was limited by demand factors rather than availability. The purpose of this study is to describe trends in VCT uptake, lifetime VCT coverage and repeat testing patterns from 2004 to 2007 for this rural population-based HIV cohort in northern Tanzania and to investigate whether VCT completion and its determinants have changed since the introduction of free ART in 2005 through the Tanzanian government's national HIV care and treatment plan.

Study setting
The study was conducted in Kisesa ward, Mwanza region in north-west Tanzania. The ward lies 20 km east of the regional capital, Mwanza City, along the main road to Kenya. Two villages (one of which is a trading centre) lie on this main road, while the other five are rural. The population is served by a health centre on the main road and by three small dispensaries. VCT services have been available at the health centre since 2005, providing referral for those diagnosed with HIV to Mwanza city hospitals, where free care and ART including services for the prevention of mother-to-child transmission of HIV (PMTCT) for pregnant women have been available since 2005 under the national programme. In 2008, ART and PMTCT provision started at Kisesa Health Centre and local patients were transferred there from the city hospitals.
The entire population of the ward is enumerated every 6 months as part of an ongoing Demographic Surveillance System. The population of the ward grew from around 19 000 in 1994, when surveillance started, to about 28 000 in 2007 (Urassa et al. 2010). Five serological surveys have been carried out to date in 1994-1995, 1997-1998, 2000-2001, 2003-2004 and 2006-2007. During these serosurveys, adults aged 15 years and above, who constitute about 50% of the total population, provided blood samples for anonymous HIV testing and completed detailed survey questionnaires covering sexual behaviour, use of health services and attitudes towards HIV (Wambura et al. 2007). The serological and behavioural surveillance data are anonymously linked to information collected through the demographic surveillance rounds. In each serological survey, all resident adults were invited to come to a temporary clinic located at a central place in each village; since the 2000-2001 round, a separate VCT service has been provided for those opting to know their HIV results.
Voluntary counselling and testing services during the serosurvey As VCT uptake at the 2000-2001 round was <1%, this analysis uses data from the -2004and 2006rounds. In the 2003-2004and 2006ticipants were asked if they would like to use the survey VCT service to know their HIV status. Those who agreed were directed to a trained counsellor in a separate hut for pre-and post-test counselling. In 2003-2004, venous blood was collected by the counsellors and transported to the NIMR laboratory for HIV testing, and clients returned for the results 1 week later. In 2006-2007, venous blood was again collected, but rapid HIV screening tests were used -a preliminary test using Capillus, confirmed by the Determine test for those testing positive. Quality control was performed at the NIMR laboratory on a 5% subsample using ELISA tests as previously.
Individuals who underwent VCT in the serological survey in 2003-2004 and 2006-2007 were found to be HIV negative and were encouraged to repeat VCT regularly. Those who discovered they were positive in 2003-2004 were informed that treatment would become available in the near future and (with their prior agreement) subsequently traced by the VCT counsellors after the start of the ART programme and referred to the zonal hospital for HIV care and treatment. Those who were positive in 2006-2007 were referred directly to the hospital for HIV care and treatment. Bus fares and a community volunteer were provided to escort persons to the hospital if they wished (Nsigaye et al. 2009).

Reported use of VCT services
The survey questionnaire captures information on participants' previous use of HIV testing services, including the place and circumstances of HIV testing and whether the results were received and post-test counselling was given. In the 2006-2007 round, participants who reported previous VCT use were also asked about their experiences of using these services.

Risk behaviour
Survey respondents were asked about their sexual behaviour, including frequency of intercourse, number of sexual partners in the last 12 months, and condom use with spouse, regular and casual partners. Bar girls are defined as 'high risk' partners for men, and regular travellers (traders, truck drivers and fishermen) are regarded as 'high risk' partners for women -extra questions were asked to probe for these kinds of partners.

Data analysis
All survey data were double entered, and linking of interview data, HIV research tests and VCT attendance was carried out anonymously using study identification numbers. Frequencies and crude associations were obtained using cross-tabulations and Chi-square tests. To identify factors independently associated with VCT uptake, multivariable logistic regression models were constructed using a forward-fitting approach that considered sociodemographic factors first, followed by behavioural factors to identify potential confounders. HIV status was included in the models as an a priori risk factor of interest. Likelihood ratio tests were used to assess the inclusion of variables in the model. Analyses were carried out separately for men and women to assess whether patterns of VCT use differ between the sexes. Data were analysed using Stata 10.

Ethical approval
Ethical approval for all surveys was given by the Medical Research Coordinating Committee of the Tanzanian Ministry of Health and by the London School of Hygiene and Tropical Medicine.  HIV testing and received their results with post-test counselling, representing a 70% increase in VCT completion from the previous survey round.

VCT uptake at the survey rounds
Among men and women, the biggest increases in VCT completion between the two surveys were for those aged 45 and over, those in polygamous marriages, those with no schooling and those following traditional religions (Table 1). Among women only, VCT uptake increased by over 150% among residents of remote rural villages by the 2006-2007 round.
Factors associated with VCT uptake in 2003-2004 have been described in detail elsewhere (Wringe et al. 2008) and are presented in Table 2. In brief, for both sexes, factors independently and positively associated with completing VCT included higher education, being from a non-Sukuma tribe and reporting three or more sexual partners in the last 12 months. Among both sexes, the adjusted odds of VCT use were lower among those following a traditional religion, compared to Christians or Muslims, and being 45 years or above. For women only, living in a roadside village and using a condom with a high-risk partner in the last 12 months were associated with higher VCT use. For men only, the adjusted odds of VCT use were higher among HIV infected than those who were HIV negative.

Disclosure to partner and attitudes towards VCT services
During each survey, participants reporting previous VCT use were asked about receipt of results, post-test counselling and disclosure of their HIV status to their partner or others (Table 4) In 2003-2004, disclosure rates were higher among those reporting prior use of the survey VCT service compared to those using other HIV testing centres (81% of men and 89% of women vs. 52% of men and 42% of women). By 2006-2007, overall disclosure rates had slightly declined from the earlier round, with a smaller difference observed by place of VCT.
One thousand one hundred and seventy-nine participants who reported previous VCT use in [2006][2007] were additionally asked their views on service quality: over 90% of men and women reported they would recommend VCT to a friend, regardless of where they had used VCT, and <10% of men and women reported that the VCT session was embarrassing, regardless of where they had used VCT. Trust in the confidentiality of the counsellors was higher among men than women, regardless of the service used: 88% of male users of survey VCT services and 86% of other VCT users, compared with 70% of women who had used either the survey VCT service or other VCT services.

Discussion
Overall, our study has shown that VCT completion in Kisesa ward was worryingly low at 10% in 2003-2004 and 17% in 2006-2007. The increase in 2006-2007 may be due to factors such as the availability of rapid HIV tests with results given within 15 min of testing, knowledge of the availability of free ART and a national VCT campaign in early 2005. Other studies in African settings have also shown that free access to ART motivates individuals to undergo VCT, but is not sufficient to ensure high rates of testing uptake (Day et al. 2003). Compared to the Tanzanian national picture from the demographic and health surveys (DHS) where 15% men and women had ever tested in 2004 and 37% women and 26% men had ever tested in 2007 (TACAIDS 2005(TACAIDS , 2008, the proportion of ever tested in Kisesa is low: 7% of women and 12% men in 2004 rising to 24% of women and 29% of men in 2007. However, the DHS results for rural Mwanza region are much closer to the Kisesa results, which will partly reflect the later introduction of providerinitiating testing and counselling and antenatal testing in many rural areas such as Kisesa (where PMTCT only started at the end of 2008).
The proportion of participants who expressed interest in VCT in 2003-2004 was higher than in 2006-2007 (27% vs. 18%), but in terms of those completing the procedure, higher proportions were recorded in 2006-2007 compared to 2003-2004. Introduction of HIV rapid tests could partly explain this. Other possible explanations for the low completion rates in 2003-2004 by those who expressed an interest in VCT include HIV-related stigma, confidentiality concerns and indecision, which is similar to findings from other African studies (Matovu & Makumbi 2007;Bwambale et al. 2008). Encouragingly in Kisesa, by 2006Kisesa, by -2007 of those who reported prior VCT use stated that the counsellors could be trusted.
Despite variations in VCT completion rates in the two rounds, the socio-demographic characteristics of VCT users were similar. In particular, the strongest predictor of VCT uptake for both sexes in each survey round was higher levels of education, as found in certain other studies (Fabiani et al. 2007;Bwambale et al. 2008). In both rounds, older participants, those with least schooling, those Table 4 Voluntary counselling and testing (VCT) completeness and disclosure in -2004and 2006, and VCT experience in 2006  following traditional religions and women living in the most remote rural villages remained relatively underserved by the VCT services, highlighting the need for extra efforts to reach these groups. Nevertheless, it was encouraging that these inequalities in testing coverage declined over the two rounds. Odds ratios reported in this study do not directly translate into risk ratios for completion of VCT, but can be used to identify inequalities in completion of VCT. Both surveys provided good opportunities for people to access VCT as the clinics were brought to the villages, enhancing access and most of those who took up the service had never previously undergone VCT. Although the provision of VCT at 2006-2007 almost doubled the proportion of people who had ever undergone VCT in this population, a high proportion of women, older individuals and residents in remote villages have yet to use VCT. Improving access for these groups may require alternative methods for providing VCT-like door-to-door HIV testing services or regular mobile HIV testing clinics (Matovu & Makumbi 2007;Mutale et al. 2010).
In terms of high-risk behaviours, VCT use generally increased with higher numbers of sexual partners in the past year for men and women in both rounds. VCT uptake was higher among those reporting sex with a high-risk partner and condom use in the past 12 months. By adjusting for any sexual experience in the multivariable analysis, VCT completion in the group reporting high-risk sexual partners, condom use and higher number of partners, is compared to those who are sexually active but who do not report these behaviours. Consistently, higher rates of VCT use among individuals with these risk behaviours indicate risk awareness, although further research is required to assess whether the VCT counselling sessions result in subsequent reductions in reported risk behaviours among these individuals in future survey rounds.
In both surveys, infected persons of both sexes were more likely to test than the uninfected, with this being more apparent among men than women in both rounds. Nevertheless, with only 25% of HIV-positive individuals taking up VCT services during the later round after ART introduction, more progress is needed to achieve the necessary coverage to enable universal access to HIV treatment. During the 2006-2007 round, 25% (115 ⁄ 453) of HIV-infected persons who were hitherto unaware of their status were diagnosed, but this still left 58% (338 ⁄ 587) of infected people unaware of their status.
Regularly repeated VCT is crucial for any development of 'treatment as prevention' interventions that would rely on high rates of regular HIV testing to identify recent seroconverters. In our study, among the 3923 individuals who attended both surveys, the proportion of recent seroconverters who accessed VCT in both rounds (27%) was slightly higher than the proportion of repeat negatives that completed VCT at both rounds (20%).
Although this study has the advantage of being undertaken in a community-based HIV cohort study, it relies heavily on self-reporting by respondents which may lead to reporting bias because of reluctance to disclose sensitive behaviours like non-marital sexual activity. Individuals who accessed VCT were invited to participate in the serosurvey and only informed specifically about the availability of VCT during registration. This may have implications for VCT uptake if some were not prepared to make the decision and if married women felt they needed their husband's approval (Kranzer et al. 2008). Although this may have led to a reduction in VCT access, it might ensure wider participation in the serosurvey, as agreement to VCT would not be seen as implicit in survey participation.
In conclusion, this analysis shows that VCT uptake remains low, even after 2 years of free ART. Although VCT services attract HIV-infected individuals and those with high-risk behaviours, several inequalities in testing coverage persist, especially among older women, those with least education and those following traditional religions. The provision of VCT in temporary huts in villages using rapid tests could have a dramatic impact on the overall uptake of HIV testing and should be implemented in other areas in the country to improve VCT uptake.