advanced search

<< Back To Home

CSU 113/2010: HIV STATUS IN DISCORDANT COUPLES

Monday, 8th of November 2010

CSU 113/2010: HIV STATUS IN DISCORDANT COUPLES

IN SUB-SAHARAN AFRICA

From the authors’ discussion:

‘Our meta-analysis has shown that women and men are equally likely to be an index partner in an HIV discordant couple. These results have important implications for prevention strategies. Although most social marketing aimed at reducing extramarital relationships is targeted at men, efforts should also take into account the role of women. Similarly, both men and women in relationships should be informed about the need for condom use when HIV status is unknown. Finally, although the man's role in infecting the female partner has been the dominant focus in prevention strategies, the emphasis should be revised in the context of stable couples, since uninfected men and women seem to have an equal chance of having a stable partner who has HIV.

‘Our findings might seem counterintuitive to the large amount of evidence showing male sexual behaviours and risk taking.22 However, we believe that this evidence partly reflects a research bias, because substantially fewer studies have examined sexual risk taking in women in stable relationships. Our study should not be interpreted as assigning blame towards either sex; instead we hope that this study will stimulate a more gender-balanced approach in the orientation of behavioural research and prevention interventions.’

Full text, with graphics, is at

http://www.thelancet.com/journals/laninf/article/PIIS1473-3099(10)70189-4/fulltext

Good reading.

BD

 

outline goes here

The Lancet Infectious Diseases, Volume 10, Issue 11, Pages 770 - 777, November 2010

HIV status in discordant couples in sub-Saharan Africa: a systematic review and meta-analysis

Original Text

Oghenowede Eyawo MPH a, Damien de Walque PhD b, Nathan Ford PhD c, Gloria Gakii MA d e, Richard T Lester MD e f, Dr Edward J Mills PhD g h

Background

Most couples affected by HIV/AIDS in sub-Saharan Africa live in discordant relationships. Men are thought to be the index case in most relationships, and most social marketing and awareness campaigns are focused on men. We investigated serodiscordance in stable relationships to establish the gender balance of index-case infections.

Methods

We did a systematic review, random-effects meta-analysis, and meta-regression of published and unpublished studies enrolling discordant couples and assessed the proportion of men and women that were index cases. We repeated the analysis with data from demographic and health surveys (DHS) from the 14 countries that have documented the HIV status of couples. Our primary outcome was the total number of HIV discordant couples, including the proportion of HIV-positive women.

Findings

We included data from 27 cohorts of 13 061 couples and DHS data from 14 countries of 1145 couples. The proportion of HIV-positive women in stable heterosexual serodiscordant relationships was 47% (95% CI 43—52), which shows that women are as likely as men to be the index partner in a discordant couple. DHS data (46%, 41—51) and our sensitivity analysis (47%, 43—52) showed similar findings. Meta-regression showed that urban versus rural residence (odds ratio 0·31, 95% CI 0·22—0·39), latitude (β coefficient 0·02, 0·023—0·034), gender equality (β coefficient −0·42, −0·56 to −0·27), HIV prevalence (β coefficient −0·037, −0·04 to −0·030), and older age (β coefficient 0·20, 0·08—0·32) were associated with the proportion of female index cases.

Interpretation

Our study shows the need to focus on both sexes in HIV prevention strategies, such as promotion of condom use and mitigation of risk behaviours.

Funding

None.

The HIV/AIDS pandemic has had an unprecedented effect on mortality and morbidity in sub-Saharan Africa. Several studies1—7 have shown that heterosexual transmission is the predominant method of HIV infection in adults, and major preventive efforts have focused on interventions to reduce sexual transmission of HIV (eg, increasing access to both male and female condoms, encouraging fidelity in people in stable relationships, improving gender equality, and promoting abstinence and delayed sexual initiation).8 Prevention programmes have focused on both most-at-risk populations9 and HIV-positive individuals.10

Many individuals in stable relationships are infected,11 which has led to growing interest in the role of concurrent relationships in fuelling the epidemic.12—14 Data from the first demographic and health surveys (DHS) to include results from HIV tests suggest that in at least two-thirds of couples in whom at least one of the partners is HIV positive, only one person is infected.15 The proportion of serodiscordance resulting from infection before couples got together versus infections introduced into the stable relationships through outside partners is unknown. However, people in HIV serodiscordant relationships, in which one partner is HIV positive and the other is HIV negative, are at an especially high risk of becoming infected.16 The yearly risk of infection for a partner of a person with HIV is about 10%,16, 17 and up to 95% of new HIV infections in Rwanda and Zambia are in couples living together.18 Targeting of strategies at reducing overlapping relationships is therefore being promoted as a way to decrease the number of new infections entering a relationship and subsequently to decrease the overall number of new HIV infections.19

In HIV discordant heterosexual couples in Africa, men are generally regarded as the source of HIV in the relationship, and are commonly referred to as an index case (in serodiscordant couples the index case is the person infected and the partner is not).20, 21 The expectation that men rather than women are the index cases has been widely promoted by evidence of low condom use by men, a greater burden of sexually transmitted infections, male dominance in sex-related negotiations, greater number of sexual partners (including polygamous marriages), more frequent alcohol misuse, and greater likelihood of transactional (when a client exchanges money or gifts for sex) or intergenerational sex.22, 23 Much of the social marketing aimed at reduction of concurrency has therefore been informed by a perspective of addressing male domination and women's empowerment.24—28 However, the relative HIV burden within heterosexual discordant partnerships in sub-Saharan Africa is poorly understood.15, 29, 30 We did a systematic review and meta-analysis of serodiscordance in stable relationships to establish the gender balance of index-case infections.

Search strategy and selection criteria

Our primary analysis examined published and unpublished studies. A secondary analysis used DHS data from 14 countries in sub-Saharan Africa (DHS assessments are done roughly every 5 years). We included research from the African continent that investigated HIV discordance and included a measure of the distribution of HIV infection within stable partnerships. We defined relationships as married or unmarried couples who are in a self-identified stable habitual relationship. We included cohort studies and DHS data that assessed household HIV status.15, 31 We excluded studies done outside Africa, presented only qualitative data, or presented quantitative data in a way that did not allow the proportional gender attribution of discordance to be calculated. We excluded data for polygamous couples. When applicable, we captured reporting on status of widowhood, divorces, and separation.

Two investigators (OE, EJM) searched, independently and in duplicate, the following databases from inception until Dec 1, 2009: Medline via PubMed, Embase, CinAhl, Amed, Toxnet, Cochrane Central, and E-Psyche. We also searched the accessible websites of all International AIDS Society conferences (up to July, 2009) and all Conference on Retroviruses and Opportunistic Infections meetings (up to February, 2009). For the primary search, we used one or more combinations of the following keywords for our searches: “HIV”, “discord* (discordant/discordance/discordancy)”, “sero-discord* (discordant/discordance/discordancy)”, and “Africa”. We supplemented our searches by examining the bibliographies of published and relevant papers. Additionally, we included data from two programmatic cohorts in Uganda. DdW obtained findings from DHS data by analysing the survey data available from MEASURE DHS.

Two investigators (OE, EJM) scanned all abstracts and full-text articles when available. We achieved final inclusion of abstracts and full-text studies through consensus. When agreement was unattainable between both reviewers, we achieved consensus through third-party arbitration, including verification with the study authors when appropriate.

Data extraction

We abstracted data from the eligible studies independently and in duplicate with a standardised and tested extraction form. Our primary outcome was the total number of HIV discordant couples including the proportion of HIV-positive women. We also sought data for study type, country or setting (rural vs urban), year of study, study population (age by sex), proportion formally married or widowed, sample size, study follow-up period, and whether the cohort was followed up for HIV transmission risk research. Finally, we assessed the following factors as determinants of methodological quality: whether clinically eligible patients received highly active antiretroviral therapy (HAART); whether index or reference clients were included according to a-priori inclusion criteria; whether viral load was assessed; and whether sample size was based on a-priori assumptions of differences in effect size or arbitrary availability. We did not apply these criteria to DHS data because these data use a standardised approach to sampling and anonymous HIV testing.

Statistical analysis

We calculated the φ statistic as a measure of interobserver agreement independent of chance on the inclusion of studies.32 Our primary effect measure used the proportion of couples with HIV-positive women and the 95% CI for the proportions. To pool proportions we first stabilised the variances of the raw proportions (r/n) with a Freeman-Tukey type arcsine square-root transformation and applied a random-effects model. Although several methods of pooling proportions exist,33 the Freeman-Tukey method works well with both fixed-effects and random-effects meta-analysis.33, 34 We used a DerSimonian-Laird random-effects analysis. We report the τ2 value as a measure of heterogeneity, because this measure is less affected by the number of studies than is the more commonly used I2.35, 36 As pooled proportions yield high rates of heterogeneity, we applied a random-effects method-of-moments meta-regression to explain heterogeneity with our covariates that were decided a priori. We examined the effect of our regression variables on proportion heterogeneity by calculating the Q value for residual heterogeneity.37 We calculated Clopper-Pearson CIs of proportion of couples with HIV-positive female partners. We did a sensitivity analysis of the pooled cohort study proportion to ensure that we did not enter duplicate populations by excluding possible duplicate and unpublished studies.

We applied a random-effects unrestricted maximum likelihood meta-regression analysis on the cohort studies to establish whether population demographics might affect the rate of the gender-specific magnitude of representation in the studies. Generally, we applied one covariate for every seven datapoints.38 We examined the following variables, according to the data available for the closest year of the included study: gender gap score by country (as determined by most recent Global Gender Gap Report by the World Economic Forum) because gender inequality has been associated with risk taking;39 urban or rural populations (with urban defined as a city with more than 500 000 inhabitants) because we expected sexual networks to differ; latitude of the cohort (a widely accessible location indicator) to explore region-specific trends; human development index as measured by UNDP Human Development Reports because this score provides a ranking of availability of social services and education;40 year-specific HIV country prevalence (as determined by UNAIDS) because increased HIV prevalence might affect sexual risk-taking;39 and mean difference between mean age of men and of women, because older age might increase the likelihood of remarriage. We assessed the difference between published and unpublished data with a Z test.

We used StatsDirect (version 2.7.6) and STATA (version 10.0) for all analysis. p values are two-sided, with a p value less than 0·05 regarded as statistically significant.

Role of the funding source

There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Our initial search identified 1614 abstracts (figure 1), 1555 of which were from conference proceedings. 569 abstracts were for studies done outside Africa, and 632 were either unrelated to our research question or did not adequately report our primary outcome. 58 journal abstracts and 355 conference abstracts passed initial screening. After review of full-text articles and abstract papers, we reached agreement on the inclusion of 24 studies: 19 full-text journal articles and five conference abstracts (φ=0·91). An additional article was included from review of article bibliographies.41 Additionally, we included unpublished data from our own two cohorts from The AIDS Support Organization (TASO), Uganda. Overall, our analysis included 24 distinct cohorts in seven countries and three multicountry cohorts of 13 061 couples (table 1). As a secondary analysis, we included data from 14 DHS assessments, comprising 1145 discordant couples across the 14 countries.

 

Figure 1 Full-size image (31K) Download to PowerPoint

Flow diagram of study selection process

DHS=demographic and health survey.

Table 1Table image  

Characteristics of included studies showing the proportion of seropositive women

Most studies were cohort studies of stable heterosexual partnerships done in eastern and southern Africa (table 1). Sample sizes differed greatly (median 110 couples, IQR 55—429), as did population settings (rural vs urban; table 1). Most of the conference abstracts did not report study setting and were excluded from the corresponding regression analysis. Specific characteristics of populations were poorly reported in most studies, with no study reporting the proportion of couples that included widowed individuals or the proportion in a stable but unmarried relationship. No study identified couples in which the index partner was already infected versus those infected during their marriage. Table 2 shows the methodological characteristics of the included cohorts. Average follow-up was 27·3 months (range 0·25—84). Table 3 presents the findings from the included DHS data.

Table 2Table image  

Methodological quality assessment

Table 3Table image  

Findings from demographic and health surveys by country and year

Data from 27 studies were included in our primary meta-analysis, which assessed the overall proportion of female HIV-index cases in heterosexual serodiscordant relationships. From the pooled overall DerSimonian-Laird analysis, the proportion of HIV-positive women in stable heterosexual serodiscordant relationships was 47% (95% CI 43—52; figure 2), with no significant difference between the number of couples in which women or men were the index cases. We identified three studies7, 20, 46 that were reports from the same cohort, but reported differing sample sizes over different periods. However, in a sensitivity analysis in which we excluded the two most recent studies7, 46 the results were not changed (pooled proportion 47%, 95% CI 43—52). Furthermore, exclusion of the unpublished data from TASO Uganda did not affect the findings (pooled proportion 47%, 42—52; figure 2 and webappendix). We identified a similar proportion of female partners who were seropositive when we pooled the DHS data (46%, 41—51; figure 3).

 

Figure 2 Full-size image (39K) Download to PowerPoint

Meta-analysis of the proportion of HIV-positive women in heterosexual serodiscordant partnerships in 27 cohorts

*Unpublished (data available from authors).

 

Figure 3 Full-size image (17K) Download to PowerPoint

Meta-analysis of the proportion of HIV-positive women in heterosexual serodiscordant partnerships in 14 DHS country sets

DHS=demographic and health survey.

We found large heterogeneity among studies (τ2= 0·18). Findings from our meta-regression (table 4) showed that urban versus rural residence, latitude, gender equality, HIV prevalence, and age were associated with the effect size and accounted for a substantial amount of the observed heterogeneity. When we assessed DHS data, we noted that both gender equality and decreased HIV prevalence were associated with the effect size (table 4).

Table 4Table image  

Meta-regression outcomes by covariate

Our meta-analysis has shown that women and men are equally likely to be an index partner in an HIV discordant couple. These results have important implications for prevention strategies. Although most social marketing aimed at reducing extramarital relationships is targeted at men, efforts should also take into account the role of women. Similarly, both men and women in relationships should be informed about the need for condom use when HIV status is unknown. Finally, although the man's role in infecting the female partner has been the dominant focus in prevention strategies, the emphasis should be revised in the context of stable couples, since uninfected men and women seem to have an equal chance of having a stable partner who has HIV.

Our findings might seem counterintuitive to the large amount of evidence showing male sexual behaviours and risk taking.22 However, we believe that this evidence partly reflects a research bias, because substantially fewer studies have examined sexual risk taking in women in stable relationships. Our study should not be interpreted as assigning blame towards either sex; instead we hope that this study will stimulate a more gender-balanced approach in the orientation of behavioural research and prevention interventions.

Our analysis has several strengths and limitations. We identified 27 distinct cohorts of serodiscordant couples across Africa, but, although we searched extensively for both published and unpublished data, we know of many more unpublished cohorts within routine programmes, including cohorts in which discordant couples might not be aware of their HIV status.19 We cannot formally explore this limitation because publication bias assessments, such as funnel plots, do not apply to proportions. We note, however, that data from unpublished cohorts in Uganda showed similar event rates to our pooled analysis. Our sensitivity analyses showed that our results were not affected by cohorts reporting from the same districts at previous periods or by unpublished data (webappendix). We supplemented our searches with DHS data and recorded consistent findings. DHS data are nationally representative and are collected in a way that ensures HIV-positive individuals are included.31

We detected a high level of heterogeneity, which was consistent with pooling proportions.62 Tests for heterogeneity might over-represent heterogeneity when pooling proportions because of a larger emphasis on variability between studies.63, 64 We applied the τ2 statistic, which is less affected by the number of pooled studies than are other common measures.36 Furthermore, we applied a meta-regression technique to explain heterogeneity in our analysis and attempted to reduce the effect of heterogeneity through covariates determined a priori. Our cohort could have been systematically biased by the fact that participants in most studies were followed up for the purpose of research, and participation could favour one HIV-positive sex over another for reasons such as stigma. However, such bias is unlikely since similar estimates are provided by the DHS data and the programmatic data from Uganda. Additionally, emerging evidence from HIV prevention trials in eastern Africa shows that recruitment of men seems to be more feasible than recruitment of women.65

We noted that studies poorly reported periods of follow-up. The proportions of index cases who were men and women could have changed over time. We examined this possibility in a regression analysis, but did not identify an effect over time. We also recognise that the self-identification of stable habitual relationships might vary according to setting or study, with some settings considering concurrent partnerships more acceptable than others. Finally, an important limitation is that only four studies reported the number of couples in which one or both of the members had been previously married or were infected before their present stable relationship.

Previously married and widowed people have a greater prevalence of HIV/AIDS than do non-widowed people, and some cultures encourage marriage after the loss of a partner whereas others do not. Similarly, some cultures promote HIV/AIDS testing for widowers, whereas others do not.66 A previous analysis examining HIV status at the time of marriage compared with that at the time of DHS data collection noted that female discordancy increased an average of three times since the point of marriage.67 In the only study examining HIV prevalence in widowers (done in Kenya), HIV seroprevalence was 54% for men and 60% for women.66 In a study from rural Tanzania,68 44·8% of men and 33·9% of women had been divorced at least once; and in a rural study from Malawi,69 34·4% of women had been divorced at least once and 9·8% widowed at least once. Beegle and de Walque31 examined these findings within the same DHS datasets. They excluded most cases of infections before the couples' present union by limiting data to couples in which the woman has been in only one union for 10 years or more. In five countries, the number of HIV-positive couples who had been in the same union for at least 10 years is too small for meaningful analysis. In the other nine countries the proportion of discordant female couples is decreased, but not substantially. The proportion of discordant couples in which the woman is infected in Côte d'Ivoire, Cameroon, and Kenya exceeds 30% of HIV-positive couples, whereas the proportion is 20—30% in Burkina Faso, Malawi, and Tanzania, and 10—20% in Zimbabwe, Rwanda, and Lesotho.31

Our results are consistent with a previous study by one of us (DdW) that used population-based DHS data from five different countries and showed that women represented an average of 30—50% of all discordant couple index cases.15 Findings from a follow-up study that examined marriage as a potential risk factor for HIV infection in 13 African countries showed that previous marriage (average of 16·9% of all HIV-infected people across 13 countries) or remarriage (average of 14·5% of all HIV-infected married people in 13 countries), whether attributable to divorce, separation, or widowhood, were significantly associated with HIV infection risks.70 This finding was confirmed by an analysis of 33 countries in people marrying in their late 20s in Africa.71 In our regression analysis, older age was associated with increased risk of the women being the index case, with each year in this study increasing the likelihood by 20%. This finding indicates the need to target medical care, HIV testing before marriage, and social marketing that is aimed at previously married men and women.

Our regression analysis showed several possible explanations for increased or decreased infection rates in women. Significant predictors included urban dwelling, high latitude (ie, eastern Africa), gender status, low national HIV prevalence, and older female age. This finding is consistent with our understanding of possible risk factors for HIV infection in women, and generally points towards settings such as eastern Uganda, where women enjoy fairly high levels of gender equality compared with countries in southern Africa, but where remarriage and polygamy are common.

A possible alternative explanation for our findings could be that women are at a greater risk of infection by non-sexual transmission mechanisms than are men, possibly because of nosocomial infection during pregnancy. However, despite controversial claims that non-sexual transmission methods are common,72, 73 studies show that although nosocomial infection is possible, it is not sufficient to create widespread infections.67, 74, 75

In conclusion, our analysis provides a basis for discussions and action points that can guide HIV/AIDS programming. In particular, programmes focusing on serodiscordant couples should be planned to target men and women equally. In this way, the expected outcomes of reducing transmission within this very large group might be attainable.

Contributors

EJM, NF, DdW, RTL, and GG were responsible for the study concept. OE, EJM, DdW, and NF acquired and analysed the data. All authors drafted and revised the report, and approved the final version.

Conflicts of interest

We declare that we have no conflicts of interest.

Supplementary webappendix

PDF (247K)

1 Piot P, Bartos M, Ghys PD, Walker N, Schwartländer B. The global impact of HIV/AIDS. Nature 2001; 410: 968-973. CrossRef | PubMed

2 UNAIDS. 2008 report on the global AIDS epidemic. Geneva: The United Nations Joint Programme on HIV/AIDS (UNAIDS), 2008.

3 UNAIDS. Country response: sub-Saharan Africa. http://www.unaids.org/en/CountryResponses/Regions/SubSaharanAfrica.asp. (accessed July 3, 2009).

4 Allen S, Lindan C, Serufilira A, et al. Human immunodeficiency virus infection in urban Rwanda. Demographic and behavioral correlates in a representative sample of childbearing women. JAMA 1991; 266: 1657-1663. PubMed

5 Grésenguet G, Séhonou J, Bassirou B, et al. Voluntary HIV counseling and testing: experience among the sexually active population in Bangui, Central African Republic. J Acquir Immune Defic Syndr 2002; 31: 106-114. PubMed

6 Kennedy CA, Skurnick JH, Foley M, Louria DB. Gender differences in HIV-related psychological distress in heterosexual couples. AIDS Care 1995; 7 (suppl 1): S33-S38. PubMed

7 Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group. N Engl J Med 2000; 342: 921-929. CrossRef | PubMed

8 Murphy EM, Greene ME, Mihailovic A, Olupot-Olupot P. Was the “ABC” approach (abstinence, being faithful, using condoms) responsible for Uganda's decline in HIV?. PLoS Med 2006; 3: e379. CrossRef | PubMed

9 Global Fund for AIDS, Tuberculosis & Malaria. Round 9 (MALAYSIA). http://wwwmacorgmy/globalfund/formdoc. (accessed Dec 23, 2009).

10 Bunnell R, Mermin J, De Cock KM. HIV prevention for a threatened continent: implementing positive prevention in Africa. JAMA 2006; 296: 855-858. CrossRef | PubMed

11 Malamba SS, Mermin JH, Bunnell R, et al. Couples at risk: HIV-1 concordance and discordance among sexual partners receiving voluntary counseling and testing in Uganda. J Acquir Immune Defic Syndr 2005; 39: 576-580. PubMed

12 Halperin DT, Epstein H. Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for prevention. Lancet 2004; 364: 4-6. CrossRef | PubMed

13 Mah TL, Halperin DT. Concurrent sexual partnerships and the HIV epidemics in Africa: evidence to move forward. AIDS Behav 2010; 14: 11-16. CrossRef | PubMed

14 Shelton JD. Why multiple sexual partners?. Lancet 2009; 374: 367-369. Full Text | PDF(716KB) | CrossRef | PubMed

15 de Walque D. Sero-discordant couples in five African countries: implications for prevention strategies. Popul Dev Rev 2007; 33: 501-523. PubMed

16 Hugonnet S, Mosha F, Todd J, et al. Incidence of HIV infection in stable sexual partnerships: a retrospective cohort study of 1802 couples in Mwanza Region, Tanzania. J Acquir Immune Defic Syndr 2002; 30: 73-80. CrossRef | PubMed

17 Quinn TC, Wawer MJ, Sewankambo N, et alfor the Rakai Project Study Group. Viral load and heterosexual transmission of human immunodeficiency virus type 1. N Engl J Med 2000; 342: 921-929. CrossRef | PubMed

18 Dunkle KL, Stephenson R, Karita E, et al. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. Lancet 2008; 371: 2183-2191. Summary | Full Text | PDF(178KB) | CrossRef | PubMed

19 Mermin J, Musinguzi J, Opio A, et al. Risk factors for recent HIV infection in Uganda. JAMA 2008; 300: 540-549. CrossRef | PubMed

20 Serwadda D, Gray RH, Wawer MJ, et al. The social dynamics of HIV transmission as reflected through discordant couples in rural Uganda. AIDS 1995; 9: 745-750. CrossRef | PubMed

21 UNAIDS, UNFPA, UNIFEM. In: Women and HIV/AIDS: confronting the crisis. Geneva and New-York: UNAIDS, UNFPA and UNIFEM, 2004: 7-16.

22 Nattrass N. AIDS, gender and access to antiretroviral treatment in South Africa. Cape Town: Centre for Social Science Research, University of Cape Town, 2006. Report number: working paper 06/178.

23 Dunkle KL, Jewkes RK, Brown HC, Gray GE, McIntryre JA, Harlow SD. Gender-based violence, relationship power, and risk of HIV infection in women attending antenatal clinics in South Africa. Lancet 2004; 363: 1415-1421. Summary | Full Text | PDF(117KB) | CrossRef | PubMed

24 Piot P, Bartos M, Larson H, Zewdie D, Mane P. Coming to terms with complexity: a call to action for HIV prevention. Lancet 2008; 372: 845-859. Summary | Full Text | PDF(635KB) | CrossRef | PubMed

25 Preventing HIV/AIDS. Comprehensive condom programming: a strategic response to HIV and AIDS. http://www.unfpa.org/hiv/programming.htm. (accessed Dec 23, 2009).

26 UNFPA, WHO. Condom programming for HIV/AIDS prevention. A manual for service providers. http://www.unfpa.org/publications/detail.cfm?ID=234&filterListType=3. (accessed Dec 23, 2009).

27 Keating J, Meekers D, Adewuyi A. Assessing effects of a media campaign on HIV/AIDS awareness and prevention in Nigeria: results from the VISION Project. BMC Public Health 2006; 6: 123. CrossRef | PubMed

28 Van Rossem R, Meekers D. The reach and impact of social marketing and reproductive health communication campaigns in Zambia. BMC Public Health 2007; 7: 352. CrossRef | PubMed

29 Feldblum PJ. Results from prospective studies of HIV-discordant couples. AIDS 1991; 5: 1265-1266. CrossRef | PubMed

30 Guthrie BL, de Bruyn G, Farquhar C. HIV-1-discordant couples in sub-Saharan Africa: explanations and implications for high rates of discordancy. Curr HIV Res 2007; 5: 416-429. CrossRef | PubMed

31 Beegle K, de Walque D. Demographic and socioeconomic patterns of HIV/AIDS prevalence in Africa. Washington: World Bank, Policy Research Working Paper 5076. 2009.

32 Meade MO, Guyatt GH, Cook RJ, et al. Agreement between alternative classifications of acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163: 490-493. PubMed

33 Newcombe R. Two sided confidence intervals for the single proportion: a comparative evaluation of seven methods. Stat Med 1998; 17: 857-872. CrossRef | PubMed

34 Newcombe R. Interval estimation for the difference between independent proportions. Stat Med 1998; 17: 873-890. CrossRef | PubMed

35 Walters DE. The need for statistical rigour when pooling data from a variety of sources. Hum Reprod 2000; 15: 1205-1206. CrossRef | PubMed

36 Rucker G, Schwarzer G, Carpenter JR, Schumacher M. Undue reliance on I(2) in assessing heterogeneity may mislead. BMC Med Res Methodol 2008; 8: 79. CrossRef | PubMed

37 Borenstein M, Hedges L, Higgins J, Rothstein H. In: Meta-regression, in Introduction to meta-analysis. Chichester: Wiley, 2009: 190-202.

38 Vittinghoff E, McCulloch CF. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007; 165: 710-718. CrossRef | PubMed

39 World Economic Forum. Global Gender Gap Report 2009. http://www.weforum.org/en/media/publications/index.htm. (accessed Dec 23, 2009).

40 Human Development Reports. Human development report 2009. http://hdr.undp.org/en/statistics/. (accessed Dec 23, 2009).

41 Freeman EE, Glynn JR. Factors affecting HIV concordancy in married couples in four African cities. AIDS 2004; 18: 1715-1721. CrossRef | PubMed

42 Allen S, Tice J, Van de Perre P, et al. Effect of serotesting with counselling on condom use and seroconversion among HIV discordant couples in Africa. BMJ 1992; 304: 1605-1609. CrossRef | PubMed

43 Bunnell RE, Nassozi J, Marum E, et al. Living with discordance: knowledge, challenges, and prevention strategies of HIV-discordant couples in Uganda. AIDS Care 2005; 17: 999-1012. CrossRef | PubMed

44 Carpenter LM, Kamali A, Ruberantwari A, Malamba SS, Whitworth JA. Rates of HIV-1 transmission within marriage in rural Uganda in relation to the HIV sero-status of the partners. AIDS 1999; 13: 1083-1089. CrossRef | PubMed

45 Fideli US, Allen SA, Musonda R, et al. Virologic and immunologic determinants of heterosexual transmission of human immunodeficiency virus type 1 in Africa. AIDS Res Hum Retroviruses 2001; 17: 901-910. PubMed

46 Gray RH, Wawer MJ, Brookmeyer R, et al. Probability of HIV-1 transmission per coital act in monogamous, heterosexual, HIV-1-discordant couples in Rakai, Uganda. Lancet 2001; 357: 1149-1153. Summary | Full Text | PDF(90KB) | CrossRef | PubMed

47 Hira SK, Feldblum PJ, Kamanga J, Mukelabai G, Weir SS, Thomas JC. Condom and nonoxynol-9 use and the incidence of HIV infection in serodiscordant couples in Zambia. Int J STD AIDS 1997; 8: 243-250. CrossRef | PubMed

48 Kamenga M, Ryder RW, Jingu M, et al. Evidence of marked sexual behavior change associated with low HIV-1 seroconversion in 149 married couples with discordant HIV-1 serostatus: experience at an HIV counselling center in Zaire. AIDS 1991; 5: 61-67. CrossRef | PubMed

49 Kempf MC, Allen S, Zulu I, et al. Enrollment and retention of HIV discordant couples in Lusaka, Zambia. J Acquir Immune Defic Syndr 2008; 47: 116-125. CrossRef | PubMed

50 Lingappa JR, Kahle E, Mugo N, et al. Characteristics of HIV-1 discordant couples enrolled in a trial of HSV-2 suppression to reduce HIV-1 transmission: the partners study. PLoS One 2009; 4: e5272. CrossRef | PubMed

51 Lurie MN, Williams BG, Zuma K, et al. Who infects whom? HIV-1 concordance and discordance among migrant and non-migrant couples in South Africa. AIDS 2003; 17: 2245-2252. CrossRef | PubMed

52 Roth DL, Stewart KE, Clay OJ, et al. Sexual practices of HIV discordant and concordant couples in Rwanda: effects of a testing and counselling programme for men. Int J STD AIDS 2001; 12: 181-188. CrossRef | PubMed

53 Ryder RW, Ndilu M, Hassig SE, et al. Heterosexual transmission of HIV-1 among employees and their spouses at two large businesses in Zaire. AIDS 1990; 4: 725-732. CrossRef | PubMed

54 Senkoro KP, Boerma JT, Klokke AH, et al. HIV incidence and HIV-associated mortality in a cohort of factory workers and their spouses in Tanzania, 1991 through 1996. J Acquir Immune Defic Syndr 2000; 23: 194-202. PubMed

55 Stephenson R, Barker J, Cramer R, et al. The demographic profile of sero-discordant couples enrolled in clinical research in Rwanda and Zambia. AIDS Care 2008; 20: 395-405. CrossRef | PubMed

56 Tang J, Shao W, Yoo YJ, et al. Human leukocyte antigen class I genotypes in relation to heterosexual HIV type 1 transmission within discordant couples. J Immunol 2008; 181: 2626-2635. PubMed

57 Akani C, Erhabor O, Ejele OA, Opurum H, Nwauche C, and HIV/AIDS Study Group. HIV sero-discordance among Nigerian couples. Challenges and controversies; XVI International AIDS Conference; Toronto, Canada; Aug 13—18, 2006. TUPE0454 (abstr).

58 Bandezi VN, De Bruyn G, Dladla S, McIntyre J, Gray G. Couple HIV counselling and testing characteristics in Soweto, South Africa: a strategy to involve men in HIV prevention? XVI International AIDS Conference; Toronto, Canada; Aug 13—18, 2006. CDC1131 (abstr).

59 Dladla AN, Mwamburi D, Lurie MN. Protecting the partner: sexual behaviours of partners who are HIV positive in discordant relationships, in rural KwaZulu-Natal, South Africa. XIV International AIDS Conference; Barcelona, Spain; July 7—12, 2002. B10509 (abstr).

60 Omilabu SA, Badaru SOS, Ajayi GO. Seroprevalence pattern of spouses attending prenatal clinical laboratory in LUTH, Lagos, Nigeria. XV International AIDS Conference; Bangkok, Thailand; July 11—16, 2004. B10509 (abstr).

61 Vwalika C, Rwanda Zambia HIV Research Group Z.E.H.R.P. Enrollment of HIV discordant couples for HIV prevention trials. 3rd Conference on HIV Pathogenesis and Treatment; Rio de Janeiro, Brazil; July 24—27, 2005. WePe13.13P09 (abstr).

62 Borenstien M, Hedges L, Higgins JP, Rothstein H. In: Introduction to meta-analysis. Chichester: John Wiley & Sons, 2008: 312.

63 Mills EJ, Nachega JB, Bangsberg DR, et al. Adherence to HAART: a systematic review of developed and developing nation patient-reported barriers and facilitators. PLoS Med 2006; 3: e438. CrossRef | PubMed

64 Mills EJ, Nachega JB, Buchan I, et al. Adherence to antiretroviral therapy in sub-Saharan Africa and North America: a meta-analysis. JAMA 2006; 296: 679-690. CrossRef | PubMed

65 Medical News Today. African AIDS vaccine conference addresses future trials in Africa, lower participation rates among women. http://www.medicalnewstoday.com/articles/174439.php. (accessed Sept 29, 2010).

66 Hawken M, Melis R, Temmerman M, et al. International Conference on AIDS; Bangkok, Thailand; July 11—16, 2004. WePpD2075 (abstr).

67 Mishra V, Vaessen M, Bignami-Van Assche S, Hong R. Why do so many HIV discordant couples in sub-Saharan Africa have female partners infected, not male partners? HIV Implementers Meeting; Kigali. http://hivimplementers.com/2007/agenda/pdf/U3/U3-Mishra%20Abstract%201717.ppt.pdf. (accessed Sept 29, 2010).

68 Boerma JT, Urassa M, Nnko S, et al. Sociodemographic context of the AIDS epidemic in a rural area in Tanzania with a focus on people's mobility and marriage. Sex Transm Infect 2002; 78: i97-i105. CrossRef | PubMed

69 Boileau C, Clark S, Bignami-Van Assche S, et al. Sexual and marital trajectories and HIV infection among evermarried women in rural Malawi. Sex Transm Infect 2009; 85: i27-i33. CrossRef | PubMed

70 de Walque D, Kline R. The association between remarriage and HIV infection. Evidence from national HIV surveys in Africa. Washington: World Bank, Policy Research Working Paper 5118, 2009.

71 Bongaarts J. Late marriage and the HIV epidemic in sub-Saharan Africa. Popul Stud (Camb) 2007; 61: 73-83. CrossRef | PubMed

72 Gisselquist D. Double standards in research ethics, health-care safety, and scientific rigour allowed Africa's HIV/AIDS epidemic disasters. Int J STD AIDS 2009; 20: 839-845. CrossRef | PubMed

73 Gisselquist D, Rothenberg R, Potterat J, Drucker E. HIV infections in sub-Saharan Africa not explained by sexual or vertical transmission. Int J STD AIDS 2002; 13: 657-666. CrossRef | PubMed

74 Schmid GP, Buvé A, Mugyenyi P, et al. Transmission of HIV-1 infection in sub-Saharan Africa and effect of elimination of unsafe injections. Lancet 2004; 363: 482-488. Summary | Full Text | PDF(115KB) | CrossRef | PubMed

75 de Walque D. Do unsafe tetanus toxoid injections play a significant role in the transmission of HIV/AIDS? Evidence from seven African countries. Sex Transm Infect 2008; 84: 122-125. CrossRef | PubMed

a Faculty of Health Sciences, Simon Fraser University, Vancouver, BC, Canada

b Development Research Group, The World Bank, Washington, DC, USA

c Centre for Infectious Disease Epidemiology and Research, University of Cape Town, Cape Town, South Africa

d Pumwani Sex-worker Cohort, Faculty of Medicine, University of Nairobi, Nairobi, Kenya

e Department of Microbiology, Faculty of Medicine, University of Nairobi, Nairobi, Kenya

f Department of Medicine, University of British Columbia, Vancouver, BC, Canada

g Faculty of Health Sciences, University of Ottawa, Ottawa, ON, Canada

h The AIDS Support Organisation (TASO), Kampala, Uganda

Correspondence to: Dr Edward J Mills, Faculty of Health Sciences, 43 Templeton, Ottawa, ON K1N 6N5, Canada