Epidemiology and Social
HIV incidence during a cluster-randomized trial of two strategies providing voluntary counselling and testing at the workplace, Zimbabwe
Corbett, Elizabeth La,b; Makamure, Beautyb; Cheung, Yin Buna; Dauya, Ethelb; Matambo, Ronnieb; Bandason, Tsitsib; Munyati, Shungu Sb,c; Mason, Peter Rb,d; Butterworth, Anthony Ea,b; Hayes, Richard Ja
From the aDepartment of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK
bBiomedical Research and Training Institute, Harare
cNational Institute of Health Research, Harare
dDepartment of Medical and Laboratory Sciences, University of Zimbabwe, Harare, Zimbabwe.
Received 22 June, 2006
Revised 27 August, 2006
Accepted 27 September, 2006
Correspondence to Dr Liz Corbett, Biomedical Research and Training Institute, National Institute of Health Research, Josiah Tongogara Avenue, PO Box CY 1753 Causeway, Harare, Zimbabwe. E-mail: email@example.com
Objective: To investigate HIV incidence during a trial of two voluntary counselling and testing (VCT) strategies. Counselling may promote beneficial behavioural change, although knowledge of negative status does not appear to contribute further benefit.
Design: The parent cluster-randomized trial demonstrated much greater uptake of VCT when counselling and rapid testing were available on-site (intensive VCT) than through pre-paid vouchers to an external provider (standard VCT). Anonymous HIV tests had been requested from all employees at enrolment and after 2 years intervention.
Methods: The study setting was 22 businesses in Harare, Zimbabwe. Participants were 3146 HIV-negative individuals remaining in employment at the end of intervention, of whom 2966 (94.3%) consented to repeat testing. VCT linked to basic HIV care was provided and the main outcome measures were HIV incidence under each study arm, as a retrospective secondary analysis.
Results: Mean VCT uptake in this cohort was 70.7 and 5.2%, respectively, in the intensive and standard arms. Crude HIV incidence was 1.21 per 100 person-years, with non-significantly higher rates in the intensive VCT arm [mean site incidence 1.37 and 0.95 per 100 person-years, respectively; adjusted rate ratio 1.49 (95% confidence interval 0.79–2.80).
Conclusions: Highly acceptable VCT did not reduce HIV incidence in this predominantly male cohort. HIV incidence was highest in the high uptake VCT arm, lending support to a US trial in which rapid testing appeared to have adverse behavioural consequences in some HIV-negative clients. Careful comparison of outcomes under different counselling and testing strategies is needed to maximize HIV prevention from global scale-up of VCT.
Antiretroviral roll-out in sub-Saharan Africa has been accompanied by massive scale-up of voluntary counselling and testing (VCT) capacity and activity during the last few years [1–3]. Despite wide implementation, surprisingly little is known about the behavioural impact of VCT in Africa, or how to maximize its potential for HIV prevention. Knowledge of serostatus, with counselling, can in theory reduce transmission from HIV-positive clients, and HIV incidence among HIV-negative clients [4–10]. Brief counselling with personalized risk reduction can reduce the risk of sexually transmitted infections (STIs) , and reduced numbers of sexual partners and increased condom use following knowledge of serostatus have been consistently demonstrated for HIV-positive clients and discordant couples [4–10]. However, there is much less evidence for beneficial behavioural change from knowledge of negative serostatus [5–10].
We have previously reported high VCT uptake in a workplace intervention providing on-site counselling and testing with same-day results to employees of 11 businesses in Harare, Zimbabwe, evaluated in a cluster-randomized trial . This intensive VCT strategy achieved a 51% VCT uptake at the total workforce level compared to a 4% VCT uptake in 11 other businesses randomly allocated an alternative (standard) strategy providing pre-paid vouchers for external free-standing VCT centres . During the trial, which was linked to provision of basic HIV care at the workplace and primarily concerned with VCT uptake and subsequent health outcomes, blood for anonymous HIV testing was requested twice from all employees; once at enrolment and once at the end of a 2-year intervention period at each business.
The main aim of this retrospective secondary analysis was to investigate whether delivery of counselling and testing to a high proportion of HIV-negative employees reduced HIV incidence in the intensive VCT arm. The design of the parent trial allows assessment of the combined effect of both counselling and knowledge of serostatus.
Twenty-two businesses were recruited, categorized into three strata according to absenteeism rates, and randomly allocated to either intensive or standard VCT, with randomization stratified by absenteeism category . All employees expected to remain employed for at least 3 months were invited for interview and offered VCT according to the allocated strategy. Intensive VCT was modelled on UNAIDS recommendations , except that couple-counselling and testing was not possible unless both partners were employed by the same company. Participants had pre-test counselling, risk assessment, testing, results and post-test counselling with risk reduction planning on the same day. Follow-up counselling and repeat VCT were available. Participants at businesses randomized to standard VCT had pre-test counselling and risk assessment and were then given a pre-paid voucher to a chain of free-standing VCT providers using similar counselling and testing methods. A 2-week appointment was given to discuss their results and risk reduction plan. Employees who had not used their voucher by 2 weeks were actively followed-up and given up to two further reminders and follow-up appointments before being considered not to have completed VCT .
VCT was linked to the same package of basic workplace HIV care provided for 2 years at each business, with the intervention finishing in July 2004 . Free condoms were available throughout the intervention. New employees were enrolled and offered VCT. Counselling was conducted by experienced nurse-counsellors who underwent in-house training, and used checklists to standardize their approach. Sessions were periodically observed by a supervisor, who also conducted exit interviews. Debriefing meetings and refresher training were held every 2 weeks and 6 months respectively.
Demography and anonymous HIV tests
All employees, whether accepting VCT or not, were asked to complete a questionnaire and provide venous or finger-prick blood or oral mucosal transudate (OMT) for anonymous HIV testing, with written informed consent, at both enrolment and the finish of the intervention. Results of anonymous HIV tests were not made available to participants or clinic staff. The option of OMT was only provided at the exit survey, and limited to participants who declined to give blood.
On-site VCT was carried out by trained nurse-counsellors using parallel Determine and Unigold with either venous or finger-prick blood. Anonymous HIV-testing used Determine (Abbott, Wiesbaden, Germany) for blood and Vironostika II plus O (bioMerieux, Inc., Durham, North Carolina, USA) for OMT. One in 10 specimens were retested for quality assurance [Unigold; (Trinity Biotech, Dunblane, Scotland) for serum specimens; Vironostika II plus O for dried blood spots and OraQuick (OraSure Technologies, Inc., Bethlehem, Pennsylvania, USA) for OMT]. All specimens were stored at −20°C.
Anonymous HIV specimens were allocated a laboratory number recorded on two forms: one with the study identity number (linking data) and one on which results were recorded (results data). These were separated before tests were run and stored apart. Serial HIV results were merged with other data into one record per person using the linking file and a written computer program that immediately deleted all personal identifiers other than laboratory numbers, to preserve complete confidentiality. Stored enrolment and end-of-intervention specimens from otherwise unidentified individuals with apparent seroconversion were checked for data errors and retested with Determine, Unigold and Vironostika II plus O for serum and dried blood spots, and Vironostika II plus O and OraQuick for OMT. HIV-positivity at baseline and follow-up was defined by reactivity to at least two of the three tests for serum and dried blood spots, and to both of the tests for OMT.
Written informed consent was obtained from all participants, including consent for secondary analysis of HIV results. The trial was approved by the Ethics Committees of the London School of Hygiene and Tropical Medicine and the Medical Research Council of Zimbabwe (MRCZ), Harare.
HIV incidence was not an outcome of the VCT trial, and so no power studies were performed before analysis. A cohort was constructed from all initially HIV-negative participants who provided specimens at the end of intervention. Time-at-risk was defined as the interval between the two tests. Unadjusted HIV incidence was calculated for each site per 100 person-years follow-up (PYFU). Multivariable Poisson regression analysis was used to control for the stratified randomization and other potential confounders, and to predict the expected number of events. The ratio of observed and expected numbers of events [standardized incidence ratio (SIR)] for each site was then calculated as the outcome measure .
Based on cluster-randomized trial methodology, the data were reduced to 22 records [13,14]. Wilcoxon rank sum and t-tests were used to compare the incidence and SIR between the two arms of the trial. Relative risk estimates were calculated as the outcome value in the on-site arm divided by the outcome value in the off-site arm, and 95% confidence intervals (CI) obtained by the delta method. Individual level analyses and comparison of characteristics of participants used robust standard errors that took clustering at workplace level into account .
The study profile is given in Fig. 1. There were no significant differences between randomization arms, with 95.4 and 93.2% of initially HIV-negative employees in the standard and intensive VCT arms, respectively, who were still in employment consenting to anonymous HIV testing at the end of intervention. Loss to follow-up before the second HIV test was 32.2 and 30.3%, respectively, in the intensive and standard arms, mainly because of leaving employment. The 77 available and initially HIV-negative employees who declined the second HIV test were significantly more likely to have a non-manual job (48.1 versus 20.6% of participants, P adjusted for clustering by site < 0.001) but otherwise were similar to participants: 24 had accepted VCT (all from the intensive arm, and with negative results).
HIV incidence was analysed among 2966 initially HIV-negative employees. There were no significant differences in baseline characteristics between the two randomization arms, which are shown in Table 1. VCT uptake during the previous 2 years had been significantly different, with mean uptake per site of 70.7% in the 11 businesses randomized to intensive VCT and 5.2% at the 11 businesses randomized to standard VCT (P < 0.001). At the standard VCT businesses a mean of 17.1% of cohort participants accepted pretest counselling and risk assessment but did not subsequently attend for testing (incomplete VCT). Mean VCT uptake in the current cohort was higher than at the workforce level (which was 51.1 and 4.3% respectively ) in both arms, reflecting higher VCT uptake by employees who consented to anonymous HIV testing at enrolment than in non-participants.
HIV incidence rates
Sixty-one seroconverters were identified during 5022 person-years follow-up (crude HIV incidence 1.21 per 100 PYFU; 95% CI, 0.92–1.64). HIV incidence was higher in the intensive VCT arm (mean per-site HIV incidence 1.37 per 100 PYFU) than in the standard VCT arm (mean per-site HIV incidence 0.95 per 100 PYFU), but the difference was not significant (unadjusted relative risk 1.44; 95% CI, 0.77–2.71) as shown in Table 2. Multivariate adjustment for potential confounders did not substantially affect the results, with the adjusted relative risk for seroconversion being 1.49 (95% CI, 0.79–2.80) for intensive compared to standard VCT provision (Table 2). Figure 2 shows a scatter-plot of seroconversion rates by uptake of VCT among cohort participants for the 11 businesses in each randomization arm, illustrating the major difference in acceptability of VCT with no apparent relationship to HIV incidence. Repeating the analysis having excluding the site with the highest HIV incidence in each arm (so removing the potential outlier from the intensive VCT arm) reduced the difference between arms to an unadjusted rate ratio of 1.30 (95% CI, 0.75–2.22) and an adjusted rate ratio of 1.28 (95% CI, 0.76–2.19).
Individual level analysis of risk factors for seroconversion is shown in Table 3. On univariate analysis the seroconversion rates varied significantly by marital status. Individuals who had accepted VCT had higher HIV incidence than those who did not, but not significantly so (incidence rate ratio 1.34; 95% CI, 0.88–2.06). HIV incidence did not vary significantly by specimen type (data not shown), which were predominantly venous blood (78% of specimens: no significant differences between randomization arms). On multivariate analysis, age less than 25 years (incidence rate ratio compared to 35 to 44 year olds, 2.71; 95% CI, 1.23–5.94) and being widowed (incidence rate ratio 7.59; 95% CI, 2.34–24.65) were significantly associated with seroconversion.
This study showed no significant difference in HIV incidence among HIV-negative employees under two randomly allocated, workplace-based VCT strategies, despite a major difference in acceptability . HIV incidence in the 11 businesses of the intensive VCT arm, in which mean uptake of VCT by the current cohort participants was 70.7%, was non-significantly higher than in the 11 businesses of the standard VCT arm, where mean uptake of VCT was only 5.2%. The apparent lack of a benefit is consistent with our previous finding that high-risk behaviour in the first 3 months after VCT was common and not significantly different from that preceding VCT on routine follow-up of HIV-negative VCT clients in the intensive arm of this study .
The study design, with near equal participation in baseline and follow-up HIV testing under the two randomization arms and a biological outcome measure, provides an unusually rigorous assessment of the impact of intensive promotion of VCT, and is the first randomized trial to report HIV incidence under different VCT strategies in Africa. In addition to gaining knowledge of serostatus, VCT acceptors in the current study received counselling with personalized risk reduction planning, an intervention shown to reduce the incidence of STIs compared to didactic health information in an American randomized trial . As such the results are particularly disappointing. They are, however, in agreement with a large body of research, mostly from the United States, indicating that HIV testing has little effect on risk-taking or incidence of STIs among individual HIV-negative clients . Behavioural change following knowledge of serostatus, in some cases supported by reduced incidence of STIs, appears to be pronounced among HIV-positive individuals  and discordant couples [5,8,9,17,18]. Policies in the United States have changed in response to these findings, from targeting the uninfected to greater emphasis on early identification and treatment of HIV-positive individuals .
In Africa, where the need for effective HIV prevention is greatest, there have been relatively few studies investigating the outcome of VCT among HIV-negative individuals [7,8,20–22], apart from those demonstrating beneficial impact in discordant couples [8,9,17,18]. A randomized trial comparing VCT and health information without testing reported greater behavioural change among those allocated VCT . Results were not presented separately by serostatus, although it was noted that the greatest change was in HIV-positive men and discordant couples. Sustained behavioural change has been reported in studies delivering health information, education and counselling at the population or cohort level [23–26]. By contrast, observational studies in Africa have reported no difference in subsequent behaviour and biological outcomes when HIV-negative VCT acceptors are compared to non-acceptors, although such comparisons may be confounded by factors affecting VCT acceptance [20–22].
This study has a number of limitations. Importantly, we cannot comment on the overall HIV prevention impact of the intensive VCT arm because onward transmission from HIV-positive employees could not be investigated with the current study design. Based on results of other studies, however, behavioural change among HIV-positive clients identified through VCT is likely to have made the most substantial contribution towards HIV prevention [4–10]. Behavioural data were collected only during pre-VCT risk assessment, and so were not available for all participants. HIV incidence was not a planned outcome of the VCT trial (which was instead focused on HIV-related morbidity) and, at these relatively low HIV incidence rates, only major differences between arms would have reached significance. Our participants were predominantly male, and couple counselling was not part of this intervention. Several factors may have diluted the difference between the two VCT strategies, including non-acceptance of VCT in almost one-third of the intensive arm. VCT may have occurred outside the intervention, particularly in the standard arm, but is unlikely to have been a frequent event as out-of-study VCT manifesting as individuals presenting for HIV care was very uncommon. Health information, provided to all employees, and pre-test counselling, provided to participants who took pre-paid vouchers but did not complete VCT in the standard arm, may also have generally reduced HIV incidence. HIV incidence appears to be in decline in Zimbabwe [23,27,28], and our HIV incidence estimates are substantially lower than those reported from similar types of businesses in Harare during the early to mid-1990s .
The widespread implementation of VCT in Africa has many benefits  but is currently being delivered with maximized acceptability, not HIV prevention outcome, as the prime concern. Rapid HIV-testing with same day results is being strongly promoted [1–3]. This is the second randomized trial of outcomes in HIV-negative clients after VCT with rapid-testing: an American individually randomized trial demonstrated better acceptability but a significantly higher STI incidence in some sub-groups when VCT with rapid testing was compared with delayed receipt of HIV results . These and our results clearly demonstrate that the acceptability and behavioural impact of VCT are distinct outcomes, at least for HIV negative clients, that may vary independently under different approaches to providing VCT. Whereas the impact of rapid HIV-testing on acceptability is well described [1–3], we have little understanding of the behavioural consequences for HIV-negative clients . Although the contribution from VCT towards HIV prevention comes primarily from identification of HIV-positive clients [4–10], the majority of clients in most locations will be HIV-negative. With global scale-up the need for research to inform and maximize the benefits and minimize any potential harm from VCT is all the more imperative.
We thank the employees and businesses for their participation. None of the authors have any conflict of interest to declare. E.L.C. had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Sponsorship: The study was funded by The Wellcome Trust.
1. Joint United Nations Programme on HIVAIDS. Intensifying HIV prevention: UNAIDS position paper
. Geneva, Switzerland: UNAIDS; 2005.
2. World Health Organization. Progress on global access to HIV antiretroviral therapy: an update on ‘3 by 5’: June 2005.
Geneva, Switzerland: WHO; 2005.
3. Joint United Nations Programme on HIVAIDS. The impact of Voluntary Counselling and Testing: a global review of the benefits and challenges
. Geneva, Switzerland: UNAIDS; 2001.
4. De Cock KM, Marum EL, Mbori-Ngacha D. A serostatus-based approach to HIV/AIDS prevention in Africa. Lancet 2003; 362:1847–1849.
5. Weinhardt LS, Carey MP, Johnson BT, Bickham NL. Effects of HIV counseling and testing on sexual risk behavior: a meta-analytic review of published research, 1985–1997. Am J Public Health 1999; 89:1397–1405.
6. Elwy AR, Hart GJ, Hawkes S, Petticrew M. Effectiveness of interventions to prevent sexually transmitted infections and human immunodeficiency virus in heterosexual men. Arch Intern Med 2002; 162:1818–1830.
7. Merson MH, Dayton JM, O'Reilly K. Effectiveness of HIV prevention interventions in developing countries. AIDS 2000; 14:S68–S84.
8. The Voluntary HIV-1 Counseling and Testing Efficacy Study Group. Efficacy of voluntary HIV-1 counselling and testing in individuals and couples in Kenya, Tanzania, and Trinidad: a randomised trial. The Voluntary HIV-1 Counseling and Testing Efficacy Study Group
9. Painter TM. Voluntary counseling and testing for couples: a high-leverage intervention for HIV/AIDS prevention in sub-Saharan Africa. Soc Sci Med 2001; 53:1397–1411.
10. Marks G, Crepaz N, Senterfitt JW, Janssen RS. Meta-analysis of high-risk sexual behavior in persons aware and unaware they are infected with HIV in the United States: implications for HIV prevention programs. J Acquir Immune Defic Syndr 2005; 39:446–453.
11. Kamb ML, Fishbein M, Douglas JM Jr, Rhodes F, Rogers J, Bolan G, et al
. Efficacy of risk-reduction counseling to prevent human immunodeficiency virus and sexually transmitted diseases: a randomized controlled trial. Project RESPECT Study Group. JAMA 1998; 280:1161–1167.
12. Corbett EL, Dauya E, Matambo R, Cheung Y-B, Makamure B, Bassett MT, et al
. Uptake of workplace HIV counseling and testing: a cluster-randomised trial in Zimbabwe. PLoS Med 2006; 3:e238.
13. Bennett S, Parpai T, Hayes RJ, Cousens S. Methods for the analysis of incidence rates in cluster randomized trials. Int J Epidemiol 2001; 31:839–846.
14. Hayes RJ, Bennett S. Simple sample size calculation for cluster-randomized trials. Int J Epidemiol 1999; 28:319–326.
15. Binder DA. On the variances of asymptotically normal estimators from complex surveys. Int Stat Rev 1983; 51:279–292.
16. Matambo R, Dauya E, Mutswanga J, Makanza E, Chandiwana S, Mason PR, et al
. Voluntary counseling and testing by nurse counselors: what is the role of routine repeated testing after a negative result? Clin Infect Dis 2006; 42:569–571.
17. Allen S, Serufilira A, Bogaerts J, Van de PP, Nsengumuremyi F, Lindan C, et al
. Confidential HIV testing and condom promotion in Africa. Impact on HIV and gonorrhea rates. JAMA 1992; 268:3338–3343.
18. Allen S, Tice J, Van de PP, Serufilira A, Hudes E, Nsengumuremyi F, et al
. Effect of serotesting with counselling on condom use and seroconversion among HIV discordant couples in Africa. BMJ 1992; 304:1605–1609.
19. Janssen RS, Holtgrave DR, Valdiserri RO, Shepherd M, Gayle HD, De Cock KM. The Serostatus Approach to Fighting the HIV Epidemic: prevention strategies for infected individuals. Am J Public Health 2001; 91:1019–1024.
20. Kipp W, Kabagambe G, Konde-Lule J. Low impact of a community-wide HIV testing and counseling program on sexual behavior in rural Uganda. AIDS Educ Prev 2001; 13:279–289.
21. Matovu JK, Gray RH, Makumbi F, Wawer MJ, Serwadda D, Kigozi G, et al
. Voluntary HIV counseling and testing acceptance, sexual risk behavior and HIV incidence in Rakai, Uganda. AIDS 2005; 19:503–511.
22. Forsyth AD, Coates TJ, Grinstead OA, Sangiwa G, Balmer D, Kamenga MC, et al
. HIV infection and pregnancy status among adults attending voluntary counseling and testing in 2 developing countries. Am J Public Health 2002; 92:1795–1800.
23. Gregson S, Garnett GP, Nyamukapa CA, Hallett TB, Lewis JJ, Mason PR, et al
. HIV decline associated with behavioural change in Eastern Zimbabwe. Science 2006; 311:664–666.
24. Ng'weshemi JZ, Boerma JT, Pool R, Barongo L, Senkoro K, Maswe M, et al
. Changes in male sexual behaviour in response to the AIDS epidemic: evidence from a cohort study in urban Tanzania. AIDS 1996; 10:1415–1420.
25. Kaul R, Kimani J, Nagelkerke NJ, Fonck K, Keli F, MacDonald KS, et al
. Reduced HIV risk-taking and low HIV incidence after enrollment and risk-reduction counseling in a sexually transmitted disease prevention trial in Nairobi, Kenya. J Acquir Immune Defic Syndr 2002; 30:69–72.
26. Quigley MA, Kamali A, Kinsman J, Kamulegeya I, Nakiyingi-Miiro J, Kiwuwa S, et al
. The impact of attending a behavioural intervention on HIV incidence in Masaka, Uganda. AIDS 2004; 18:2055–2063.
27. Mahomva A, Greby S, Dube S, Mugurungi O, Hargrove J, Rosen D, et al
. HIV prevalence and trends from data in Zimbabwe, 1997–2004. Sex Transm Infect 2006; 82(Suppl 1):i42–i47.
28. Hallett TB, Aberle-Grasse J, Bello G, Boulos LM, Cayemittes MP, Cheluget B, et al
. Declines in HIV prevalence can be associated with changing sexual behaviour in Uganda, urban Kenya, Zimbabwe, and urban Haiti. Sex Transm Infect 2006; 82(Suppl 1):i1–i8.
29. Mbizvo MT, Machekano R, McFarland W, Ray S, Bassett M, Latif A, et al
. HIV seroincidence and correlates of seroconversion in a cohort of male factory workers in Harare, Zimbabwe. AIDS 1996; 10:895–901.
30. Metcalf CA, Douglas JM, Malotte CK, Cross H, Dillon BA, Paul SM, et al
. Relative efficacy of prevention counseling with rapid and standard HIV testing: a randomized, controlled trial (RESPECT-2). Sex Transm Dis 2005; 32:130–138.
HIV; HIV incidence; voluntary counseling and testing; HIV prevention; cluster-randomized trial
© 2007 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.