From a reproductive health-based and rights-based perspective, all women should have access to methods for avoiding unintended pregnancies . Additionally, HIV-positive women have particular needs for contraception to avoid unintended pregnancy: to preserve their own health , and to eliminate the risk of transmitting HIV to an infant. In the Glion Call to Action , a four-element strategy to reduce mother-to-child transmission (MTCT) of HIV was articulated: primary prevention of HIV infection, prevention of unintended pregnancies among HIV-positive women, prevention of HIV transmission from an infected mother to child, and provision of care and support for HIV-infected mothers, their infants, partners, and families. Although the Glion Call states that all four elements are essential for meeting United Nations General Assembly Special Session (UNGASS) goals, issues around the second element [The second preventing MTCT (PMTCT) element is described as follows. Unintended pregnancies among HIV-infected women should be prevented. Reproductive health and family planning services should be improved and made widely available to all women, including those with HIV infection, to provide support and appropriate services to avoid unintended pregnancies. Increased availability of counseling and testing services would enable women to find out their serostatus and then obtain essential care and support services, including reproductive health and family planning services, and make informed decisions about their future reproductive lives.] remain poorly understood, particularly regarding fertility preferences among HIV-positive women.
There is a dearth of research using nationally representative data on fertility preferences and contraceptive needs among HIV-positive women; only one such study  has appeared in the peer-reviewed literature. There is no consensus about whether HIV-positive women who know their status have different fertility desires. Some studies indicate that desire to limit fertility and unmet need for contraception are prevalent among HIV-positive women aware of their status; however, these findings are based on nonrepresentative studies of special populations [5,6]. Some longitudinal studies also suggest that HIV-positive women have greater desires to limit their fertility, and that this desire may increase over time ; however, these results might be confounded by repeated questioning about fertility intentions, which could encourage socially desirable responses. Other studies [8,9] find that there is no practical difference in fertility desires according to serostatus.
The purpose of this study is to describe reproductive intentions and need for contraception at the population level among women living with HIV. Using nationally representative data from women of age 15–49 years in four high-prevalence countries, this study assesses the contribution of knowledge of one's own HIV-positive serostatus – as proxied by the receipt of HIV test results in the past year – to women's fertility desires. We then examine the association between knowledge of one's own serostatus and contraceptive use, with particular focus on condoms, given their utility in preventing pregnancy and HIV transmission and their compatibility with antiretroviral regimens. Although we focus our analytical effort on four high-prevalence countries in sub-Saharan Africa, we expect the findings to be relevant throughout sub-Saharan Africa.
Data and methods
Data are from the Zambia 2007, Swaziland 2006–2007, Zimbabwe 2005 and Lesotho 2004 Demographic and Health Surveys [DHS; funded by the United States Agency for International Development (USAID)], which included HIV testing with test results linked to the main body of demographic and health data. These countries were selected because risk of MTCT is high in these settings: HIV prevalence among women of age 15–49 years is 16% in Zambia, 31% in Swaziland, 21% in Zimbabwe and 26% in Lesotho. Additionally, the data from these surveys are sufficient to support the analysis: women's response rates to the survey exceeded 90% for all countries, and women's response to the HIV testing component of the surveys exceeded 75% for each country [Zambia (77%), Swaziland (83%), Zimbabwe (76%) and Lesotho (81%)]. Data were collected after the Glion Call for all but one country (Lesotho), increasing the likelihood that effects of PMTCT-related programmatic and policy changes in the wake of the Call may be observed in the data. HIV-related programmatic rollout in these countries at the time of each survey was variable, as indicated by the percentage of women ever tested for HIV: 40–41% in Zambia and Swaziland, 26% in Zimbabwe and 15% in Lesotho.
Data were collected from representative probability samples selected using stratified two-stage cluster designs. All respondents eligible for HIV testing were asked to give informed consent to be anonymously tested for HIV. Samples for HIV testing were obtained by collecting blood drops from a sterile fingerstick onto a filter paper card. Additional information on sampling and response rates for each survey is available in the final report for each country [10–13]; information about the standard HIV testing protocols in the DHS, including laboratory protocols and quality control, is available in a separate publication .
The analysis uses bivariate and multivariate methods (chi-square and logistic regression) to describe the association of a proxy variable for knowledge of one's own HIV-positive serostatus with three dependent variables: women's fertility desires, need for contraception, and contraceptive method choice. Nonrespondents to the survey's HIV test were excluded from analysis. Questions on fertility preferences and contraception were asked independently of HIV-related questions or testing.
Definition of variables used in analysis
Fertility desires and knowledge of HIV status
The variable reflecting fertility desires has three categories for the bivariate analysis: wanting another child, wanting no more children (including sterilized women), and a residual category composed of ‘don't know’ responses and declarations of infecundity. These data are derived from the following survey question: ‘Now I have some questions about the future. Would you like to have (a/another) child, or would you prefer not to have any (more) children?’ The study sample for the bivariate analysis is composed of all women of age 15–49 years. Multivariate analysis is restricted to women with a declared fertility preference, resulting in a dichotomous dependent variable: women who want a/another child (0) and women who want no more children (1).
The key independent variable for all three analyses is a proxy for knowledge of one's own serostatus. To create the variable, HIV biomarker data collected in the survey were combined with women's responses to questions about their HIV testing history, that is, whether they received HIV test results within the past year. The response categories in the bivariate analyses are HIV negative, HIV positive without HIV test results in the past year, and HIV positive with HIV test results in the past year. Because the primary focus of the bivariate analysis is on HIV-positive women, HIV-negative women are not disaggregated according to their testing history; however, this step is taken for the multivariate analyses.
A core set of independent variables is also incorporated: age, education, residence, asset-based household wealth quintiles , exposure to the risk of pregnancy, and number of living children.
Need for contraception
Need for contraception is defined by three categories for the bivariate analysis: no need, unmet need, and met need for contraception (current contraceptive users). This variable is based on a standard DHS recode variable composed of numerous pieces of information collected in the DHS (Unmet need for spacing includes pregnant women whose pregnancy was mistimed, amenorrhoeic women who are not using family planning and whose last birth was mistimed or whose last birth was unwanted but now say they want more children, and fecund women who are neither pregnant nor amenorrhoeic, who are not using any method of family planning, and say they want to wait 2 or more years for their next birth. Also included in unmet need for spacing are fecund women who are not using any method of family planning and say they are unsure whether they want another child or who want another child but are unsure when to have the birth. Unmet need for limiting refers to pregnant women whose pregnancy was unwanted; amenorrhoeic women who are not using family planning, whose last child was unwanted, and who do not want any more children; and fecund women who are neither pregnant nor amenorrhoeic, who are not using any method of family planning, and who want no more children. Using for spacing is defined as women who are using some method of family planning and say they want to have another child or are undecided whether to have another. Using for limiting is defined as women who are using and who want no more children. But the specific methods used are not taken into account here). The multivariate analysis is restricted to women who have a need for contraception, producing a dichotomous dependent variable: contraceptive users (0) and women with unmet need for contraception (1).
In addition to the proxy variable for knowledge of one's own serostatus and the core set of independent variables, we include a continuous composite variable reflecting the degree to which a respondent has difficulty in accessing healthcare for herself. This variable is based on information about women's experiences with eight obstacles to accessing healthcare: each item was dichotomized as being a big problem (1) or not (0), and the items were summed for each respondent; higher scores indicate greater difficulty in accessing healthcare services. A variable representing a region of residence was initially included but later removed for lack of statistical significance and failure to improve the models.
Contraceptive method choice
For the bivariate analysis of type of contraceptive method used among women currently using a method, the dependent variable has the following categories: condom users, other modern method users, and traditional/folkloric method users. Information on contraceptive use is self-reported; it is not possible to determine whether the methods are used consistently or correctly. This variable is cross-tabulated with the proxy variable for knowledge of HIV-positive serostatus to assess the association between knowledge of one's own serostatus and contraceptive method choice.
Women with higher risk sexual behaviors may be more likely to use condoms and also more likely to be HIV positive; condom use would then reflect higher risk sexual behavior rather than prevention-oriented behavior motivated by knowledge of one's own serostatus. To clarify the relationship between knowledge of one's own serostatus and use of condom as a contraceptive method, we analyze the odds of using a condom versus any other method among contracepting HIV-positive women. The independent variable of interest reflects HIV testing history: (0) never received HIV test results, (1) ever received test results but not in the past year, and (2) received test results in the past year. We control for number of lifetime partners, as well as the core set of independent variables. This analysis was run for all countries except Lesotho, where data were insufficient for analysis.
The primary caveat of this analysis involves the proxy variable for knowledge of one's own serostatus. Although we expect this to be a reasonable proxy variable, receipt of test results in the past year is not 100% specific for knowledge of one's own serostatus: there will be some women who seroconverted as after they obtained their test result within the year prior to the survey. Further, not receiving test results in the past year is not 100% specific for ignorance of one's own serostatus: there will be women who received an HIV-positive test result prior to the 1-year reference period who do know their status but did not obtain a more recent test result in the past year. The larger the proportion of HIV-positive women who know their status from a test prior to 1 year before the survey (which presumably correlates with the national level of voluntary counseling and testing programming coverage prior to 1 year before the survey), the more the behaviors of that category of women can be expected to resemble those of women who received test results in the past year.
Another limitation is that this study does not address men and their fertility preferences in the context of being HIV positive themselves or of having an HIV-positive partner. Planned extensions of the present analyses are to look at corresponding data from men and from couples disaggregated by concordancy of serostatus.
Table 1 shows the percentage distributions of HIV-negative women and HIV-positive women who did and did not receive a recent HIV test result according to their fertility preferences; in all four countries, the chi-square tests are significant (P < 0.002). For all countries, women with an HIV-positive test result in the past year are less likely than other women to want to have a/another child. The differences among women's desires for more children according to serostatus are largest in countries where HIV testing coverage is greatest (Swaziland and Zimbabwe). HIV-positive women with a recent HIV test result were more likely than other women to report wanting no more children in Zambia, Swaziland and Zimbabwe. For these countries, HIV-positive women with no test result in the past year exhibit fertility desires that fall between those of HIV-negative women and HIV-positive women with a recent test result, suggesting that some proportion of women in this category received HIV-positive test results prior to the 1-year reference period. In Lesotho, HIV-positive women as a group are more likely than HIV-negative women to report a desire to limit births.
To clarify these bivariate associations, we control for confounding factors using a multivariate approach. The logistic regression results presented in Fig. 1 and Table 2 show that for all countries except Lesotho, both HIV-positive and HIV-negative women with recent test results were more likely than HIV-negative women without recent test results to report that they wanted no (more) children. This suggests that women who receive HIV test results are a select population. However, the desire to limit births is considerably higher among recently tested HIV-positive women than among recently tested HIV-negative women: in Zambia and Zimbabwe, recently tested HIV-negative women have 20–30% higher odds of desiring to limit their births compared with the reference group; recently tested HIV-positive women are about twice as likely as the reference group to desire to limit their fertility. In Swaziland, recently tested HIV-negative women have 43% higher odds of desiring to limit their births compared with the reference group, whereas recently tested HIV-positive women are 2.5 times as likely to want to limit their fertility. These results suggest an effect of knowledge of one's own serostatus on the formation of future fertility desires. The failure to find the expected associations in Lesotho may be due to the lack of PMTCT-related information and HIV testing service coverage available in Lesotho at the time of the survey.
Need for contraception
Table 3 shows the percentage distributions of HIV-negative women and HIV-positive women who did and did not receive a recent HIV test result according to need for contraception; in all countries, the chi-square tests are significant at a P value of less than 0.000. For all countries except Lesotho, HIV-positive women who received a test result in the past year are more likely than HIV-negative women to have a demand for contraception. In Zambia, recently tested HIV-positive women are more likely than other women to use contraception; they are also more likely to have an unmet need for contraception, although the difference compared with HIV-negative women is small. In both Swaziland and Lesotho, levels of current use of contraception are about the same for both categories of HIV-positive women and are considerably higher than among HIV-negative women, again suggesting that some proportion of HIV-positive women without a recent test result did receive HIV-positive test results prior to the 1-year reference period used in this analysis. In Zimbabwe, there is no practical difference in contraceptive use amongst the three groups of women; however, there is a significantly higher level of unmet need for contraception among HIV-positive women with a recent test result (19%) as compared with other women (8%). In Lesotho, HIV-positive women with a recent test result are the least likely to have an unmet need for contraception.
Corresponding with the analysis of desire to limit births, these results indicate that HIV-positive women, particularly those who likely know their serostatus, have greater need for and use of contraception. However, the correlation of unmet need with the serostatus variable is inconsistent across countries, suggesting that country-specific variations in programmatic rollout and efficacy may be playing a role.
Addressing whether knowledge of one's own serostatus is associated with unmet need for contraception among women with demand for contraception, we conducted logistic regressions with the study sample stratified by HIV serostatus, as well as unstratified, to control for unobserved variables associated with serostatus. Results were similar for all models, so only the results for the unstratified analysis are presented (Table 4). Full results are presented in the table; only the results of the independent variable of interest are discussed in the text. In Swaziland and Lesotho, the proxy variable for knowledge of one's own serostatus seems to better reflect access to health services: both HIV-negative and HIV-positive women who have obtained an HIV test in the past year are significantly less likely to have an unmet contraceptive need compared with the reference category. In Zimbabwe, however, recently tested HIV-positive women have twice the odds of having an unmet need for contraception compared with HIV-negative women without a recent test result. In Zambia, recently tested HIV-negative women are most likely to have an unmet need for contraception. These results correspond to the heterogeneous findings on unmet need from the bivariate analysis.
Contraceptive method choice
Table 5 shows the percentage distributions of HIV-negative women and HIV-positive women who did and did not receive a recent HIV test result and who are using a contraceptive method according to type of method used. In all four countries except for Lesotho, chi-square tests are significant (P < 0.002); Lesotho did not have enough contracepting HIV-positive women with a recent test result for analysis.
Condom use as a proportion of total contraceptive use in these countries is variable: Swaziland has the highest proportion of contraceptive users who are using a condom at 36%, whereas Zimbabwe has the lowest at 5%, and Zambia and Lesotho are in the middle with 17 and 20%, respectively. In terms of condom use according to serostatus, recently tested HIV-positive women are significantly more likely to be using a condom compared with HIV-negative women. In Zambia, condom use is twice as high among recently tested HIV-positive women as compared with HIV-negative women; in Zimbabwe, it is over three times as high as compared with HIV-negative women. This can be compared with Swaziland, where overall levels of condom use are higher; thus, condom use is only 37% higher among recently tested HIV-positive women (44% use condoms) as compared with HIV-negative women (32% use condoms). Zambia is remarkable for its high levels of traditional method use (18%).
We conducted a multivariate analysis of condom use among HIV-positive women to discern the effect of knowledge of one's own serostatus on condom use, effectively controlling for unobserved variables contributing to HIV infection (Table 6). Recently tested HIV-positive women remain significantly more likely to use condoms: in Swaziland, recently tested HIV-positive women were 77% more likely than HIV-positive women who never received test results to use condoms (P = 0.006); in Zimbabwe, women with a recent test result were more than three times as likely to choose condoms compared with those who had never received a test result (P = 0.005). In Zambia, women with a recent test result were 87% more likely to choose condoms than women with no test result (P < 0.10). These results, particularly those from Zimbabwe and Swaziland, clearly demonstrate the effect of knowledge of serostatus on choice of contraceptive method.
Results from the multivariate analyses indicate that women who know they are HIV positive (as determined by our proxy variable of receipt of HIV test results in the past year) are more likely than other women to desire to limit their childbearing except in Lesotho. Women who know they are HIV positive are also more likely than other women to be using contraception and have significantly greater unmet need only in Zimbabwe, suggesting that special attention should be given to the contraceptive needs of HIV-positive women living in political or economic crisis. In Swaziland and Lesotho, recently tested HIV-positive women are significantly less likely than the reference category to have unmet need for contraception, suggesting that integrated HIV testing and family planning services may already be functional in these contexts. Finally, women who know they are HIV positive are more likely to choose condoms as their method of contraception in all countries with sufficient data for analysis. In Swaziland, levels of condom use among both infected and uninfected contracepting women were high, suggesting that the idea that ‘everyone is at risk’ has been more widely accepted by the Swazi population than anywhere else.
The findings on fertility preferences among HIV-positive women who likely know their status correspond well with recent research from smaller scale studies but less well with earlier research on fertility preferences. This, in addition to our results from Lesotho, suggests that as PMTCT programming has rolled out in several high-prevalence countries, women are learning more about MTCT and basing their decision-making on more complete information.
Taken together, these findings provide evidence to support the argument for women's right to simultaneous universal access to HIV testing services and appropriate reproductive health/family planning services, so that all women may make fully informed decisions about when and whether to have children . The self-reported fertility preferences of HIV-positive women who likely know their status and the actions they take to implement those preferences in terms of contraceptive use, particularly condoms, are bellwethers of the progress that could be made in preventing vertical transmission if all women were aware of their HIV serostatus. We expect the relevance of these findings to grow over time: with increasing survival of HIV-infected patients, the lifetime need for contraceptive information and services will also increase.
Ensuring that women's reproductive rights are universally upheld would significantly reduce levels of unintended pregnancy among HIV-positive women, in turn reducing incidence of MTCT [16,17] without stigmatizing HIV-positive women or implying that HIV-positive women should be targeted for fertility control. The propensity for contracepting HIV-positive women to choose condoms demonstrates an additional HIV-preventive benefit of ensuring universal access to both HIV testing and reproductive health/family planning services. Preventing unwanted pregnancy among HIV-positive women is also a cost-effective PMTCT strategy: reducing unintended pregnancy by 16% would yield a reduction in infant cases equivalent to that achieved through the use of nevirapine-based prophylactic strategies . Avoiding unwanted pregnancies allows women living with HIV to maximally preserve their own health  and eliminates reproductive health risks associated with pregnancy and childbirth in the general population .
Preventing unwanted pregnancy among HIV-positive women is an intervention located at the intersection of reproductive health and HIV concerns. It can serve as the basis of a joint, rights-based agenda to eliminate vertical transmission of HIV. Joining reproductive rights-based perspectives with epidemiological perspectives to reduce MTCT is already implicitly a part of the larger development agenda; both universal access to HIV/AIDS treatment for all who need it and universal access to reproductive health have been added as targets to the United Nations Millennium Development Goals . However, the confluence of joint interests and actions must be made more explicit; both reproductive health/family planning and HIV constituencies must make concerted efforts towards integration at all levels of program planning, coordination, and implementation.
Women function within a gendered ecological system in which the ability to negotiate fertility-related outcomes is often contingent upon the cooperation of others in the immediate environment. Ensuring universal provision of services by itself will not provide a complete solution to the problem of unintended pregnancy, but it is an essential element of a rights-based approach to eliminating transmission of HIV from mother to child.
The authors thank Bernard Barrère for his support of this research, and Ani Hyslop, Amber Peterman and Rebecca Thornton for their insightful review comments. The authors also extend their appreciation to two anonymous reviewers for their helpful review comments. This analysis was conceptualized by Johnson, Rutstein and Akwara; Johnson conducted the data analysis with key inputs from Rutstein and wrote the manuscript with inputs from all coauthors. Akwara and Bernstein provided critical policy and programmatic-related inputs as well as critiques for improving the analysis. This analysis was partially supported under the USAID-funded DHS program.
Conflicts of interest: None.
1. United Nations. Report of the International Conference on Population and Development: Programme of Action
; 5–13 September 1994; Cairo. New York, New York: United Nations; 1995.
2. Lieve VP, Shafer LA, Mayanja B, Whitworth JA, Grosskurth H. Effect of pregnancy on HIV disease progression and survival among women in rural Uganda. Trop Med Int Health 2007; 12:920–928.
3. United Nations. Glion Call to action on family planning and HIV/AIDS in women and children
. New York, New York: United Nations: 3–5 May 2004.
4. Adair T. Unmet need
for contraception among HIV-positive women in Lesotho and implications for mother to child transmission. J Biosoc Sci 2009; 41:269–278.
5. McCarraher D, Cuthbertson C, Kung'u D, Otterness C, Johnson L, Magiri G. Sexual behavior, fertility desires and unmet need
for family planning among home-based care clients and caregivers in Kenya. AIDS Care 2008; 20:1057–1065.
6. Mutiso SM, Kinuthia J, Qureshi Z. Contraceptive use among HIV infected women attending Comprehensive Care Centre. East Afr Med J 2008; 85:171–177.
7. Taulo F, Berry M, Tsui A, Makanani B, Kafulafula G, Li Q, et al
. Fertility intentions of HIV-1 infected and uninfected women in Malawi: a longitudinal study. AIDS Behav 2009; 13(Suppl 1):20–27.
8. Cooper D, Moodley J, Zweigenthal V, Bekker LG, Shah I, Myer L. Fertility intentions and reproductive healthcare needs of people living with HIV in Cape Town South Africa: implications for integrating reproductive health and HIV care services. AIDS Behav 2009; 13(Suppl 1):38–46.
9. Rutenberg N, Gillespie D, Flanagan E, et al.Examining the evidence: fertility desires and family planning needs of people with HIV
. In: Conference on Linking Reproductive Health and Family Planning with HIV/AIDS Programs in Africa
; 9–10 October 2006; Addis Ababa, Ethiopia.
10. Central Statistical Office (CSO), Ministry of Health (MOH), Tropical Diseases Research Centre (TDRC), University of Zambia, and Macro International Inc. Zambia Demographic and Health Survey 2007. Calverton, Maryland: CSO and Macro International Inc.; 2009.
11. Central Statistical Office (CSO) (Swaziland) and Macro International Inc. Swaziland Demographic and Health Survey 2006–07. Mbabane, Swaziland: Central Statistical Office and Macro International Inc.; 2008.
12. Central Statistical Office (CSO) (Zimbabwe) and Macro International Inc. Zimbabwe Demographic and Health Survey 2005–06. Calverton, Maryland: CSO and Macro International Inc.; 2007.
13. Ministry of Health and Social Welfare (MOHSW) (Lesotho), Bureau of Statistics (BOS) (Lesotho), and ORC Macro. Lesotho Demographic and Health Survey 2004. Calverton, Maryland: MOH, BOS, and ORC Macro; 2005.
14. Macro International, Inc. HIV prevalence estimates from the Demographic and Health Surveys. Calverton, Maryland: Macro International, Inc.; 2008. http://www.measuredhs.com/pubs/pdf/OD51/OD51.pdf
15. Rutstein SO, Johnson K. The DHS wealth index and DHS comparative reports [No. 6]. Calverton, Maryland: ORC Macro; 2004.
16. Reynolds HW, Janowitz B, Homan R, Johnson L. The value of contraception to prevent perinatal HIV transmission. Sex Transm Dis 2006; 33:350–356.
17. Reynolds HW, Janowitz B, Wilcher R, Cates W. Contraception to prevent HIV-positive births: current contribution and potential cost savings in PEPFAR countries. Sex Transm Infect 2008; 84(Suppl 2):ii49–ii53.
18. Sweat MD, O'Reilly KR, Schmid GP, Denison J, De Zoysa I. Cost-effectiveness of nevirapine to prevent mother-to-child HIV transmission in eight African countries. AIDS 2004; 18:1661–1671.
19. Filippi V, Ronsmans C, Campbell OM, Graham WJ, Mills A, Borghi J, et al
. Maternal health in poor countries: the broader context and a call for action. Lancet 2006; 368:1535–1541.
20. United Nations. Report of the Secretary General on the work of the Organization
. General Assembly, sixty-second Session. Suppl. No. 1 (A/62/1); 2007.