Background
Sub-Saharan Africa continues to have the highest rates of new HIV infections [
1] as well as unintended pregnancies [
2]. As a region, sub-Saharan Africa has the lowest level of contraception prevalence [
3] with an estimated 25 % unmet need for family planning among women aged 15 and 49 [
4]. Eliminating unmet need for effective family planning among all women has great personal, familial and societal benefits. This includes reduced maternal and child mortality and better educational and economic gender equity [
5‐
7]. In addition, the World Health Organization mentions reducing the unmet need for family planning among women living with HIV as a highly cost-effective strategy that is essential for the elimination of MTCT [
7,
8].
Improving family planning use requires the availability of contraceptive methods that allow women to align their choice of method with their reproductive intentions. Although short-acting methods such as condoms, oral contraceptives (OCs) and injectables are the most common methods in sub-Saharan Africa, long-acting reversible (LARC) and permanent methods are more effective under “typical use” conditions, as they eliminate the need for adherence and have higher continuation rates [
9,
10]. Hence, unintended pregnancy could be high in Botswana, where the available methods are mostly short-acting, requiring daily or quarterly adherence. The current use of modern contraception among women 15–49 in Botswana has increased from 16 % in 1984 to 51 % in 2007. The most commonly used contraceptives are condoms, oral and injectable contraceptives; use of LARC methods such as the intrauterine device and implants is negligible [
11]. The male condom comprised 93.3 % of current use in 2006 [
12], and the condom’s predominance has been attributed to a national multimedia Condom Social Marketing Program implemented by the Botswana Population Services in 1994 and 1995 in response to the HIV epidemic [
11]. In Botswana, where 30.4 % of women of reproductive age are HIV-infected [
13] and antiretroviral treatment (ART) coverage is high [
14,
15], access to a variety of contraceptive options is critical because HIV-infected women may have particular contraceptive needs and may change their childbearing plans in response to learning their HIV serostatus [
16,
17].
We assessed pregnancy intention, contraceptive use, and future childbearing desires among pregnant and recently post-partum HIV-positive and HIV-negative women in Botswana. Our objectives were to: 1) assess the frequency of unintended pregnancy among women according to knowledge of HIV serostatus prior to becoming pregnant; 2) determine whether prior knowledge of HIV serostatus is an independent predictor of unintended pregnancy; 3) determine among women with unintended pregnancies, whether prior knowledge of HIV serostatus is associated with contraceptive use; and 4) compare future childbearing desires by actual HIV serostatus at study enrolment. There are currently no published data on these topics from Botswana and little data from southern Africa in this era of widespread ART use.
Methods
Study population
This was a planned analysis of baseline data collected in a prospective observational cohort study of child health and neurodevelopment (“Tshipidi” study: a Setswana word meaning “a journey of a thousand miles begins with one step”). This study was funded by the National Institute of Mental Health at the National Institutes of Health, R01MH087344. In total, 475 HIV-uninfected and 474 HIV-infected pregnant and recently postpartum women aged 18 years and older and their infants were enrolled between 2010 and 2012 at 28 antenatal clinics and 5 maternity wards in two locations in Botswana: the capital city Gaborone and in the large village of Mochudi, which is 45 km North of Gaborone. As part of the main cohort study, mothers and infants were followed for two years to evaluate the effect of maternal HIV status on child health and neurodevelopmental outcomes. Participants were enrolled antepartum (n = 805 (84.8 %)) or within seven days of delivery (n = 144 (15.2 %)). The current analysis utilizes data from structured questionnaires administered privately at enrollment by trained study nurses at the time of enrollment.
Outcome assessment
To measure pregnancy intendedness, the primary outcome, all women were asked “Were you trying to become pregnant when you conceived the baby?” This yielded a combined category of unintended pregnancy, which included mistimed and unwanted pregnancies, irrespective of whether contraception was being used. Women who reported that they were not trying to become pregnant at the time of conception were also asked whether they were using contraception and the specific contraceptive method type was documented. We considered women as having experienced contraceptive failure or misuse (including poor adherence), if they reported that the current pregnancy was unintended and that they were using contraception prior to the pregnancy. Women were considered as having an unmet need for family planning if they reported that the current pregnancy was unintended and that they were not using contraception prior to the pregnancy.
Exposure assessment
Women’s self-reported knowledge of HIV status prior to becoming pregnant was recorded as “known HIV-infected”, “known HIV-uninfected” or “unknown HIV status.” HIV status at the time of study enrolment was also determined by confirmatory testing on all participants.
Statistical analysis
We performed basic descriptive analyses, including a comparison of study enrolment characteristics stratified by maternal knowledge of HIV serostatus prior to becoming pregnant, and intendedness of pregnancy. We used logistic regression to estimate univariable and multivariable-adjusted odds ratios (OR) and 95 % confidence intervals (CI) for pregnancy intendedness. We considered several covariates that likely temporally preceded the current pregnancy and birth and might also be associated with knowledge of HIV status prior to becoming pregnant and pregnancy intendedness. Therefore, we included all of the following variables in the multivariable-adjusted model as potential confounders: enrollment site, age, relationship status, educational level, employment status, income, household size, household assets, and lifetime number of pregnancies and history of a child dying after birth but before five years of age. All statistical analyses were conducted using SAS software version 9.3 (SAS Institute, Cary, NC).
In secondary analyses, we used the same process to assess the relationship between a) maternal knowledge of HIV serostatus prior to becoming pregnant and contraceptive use and b) HIV status at enrollment and future childbearing desires in multivariate-adjusted analyses.
Ethics
Ethics approval for the Tshipidi study was obtained from the Office of Human Research Administration (OHRA) at the Harvard School of Public Health and the Health Research Development Committee at the Botswana Ministry of Health. All women provided written informed consent prior to study participation. Only unique study identifiers were used in the database.
Discussion
This is the first published data on unintended pregnancy, contraceptive use and HIV serostatus in Botswana, a country where 30.4 % of women of reproductive age are HIV-infected [
13]. We found that both contraceptive failure (or misuse) and unmet need for family planning may have contributed to the high levels of unintended pregnancies among both HIV-infected and HIV-uninfected women, similar to findings from other recent studies in sub-Saharan Africa [
10,
18]. Overall, 44 % of women from two areas of Botswana reported that their pregnancy was unintended. The Botswana 2013 Global AIDS Response Report estimated levels of unplanned pregnancy at 50 % [
19]. Factors associated with unintended pregnancy in our study were similar to those found in other studies in sub-Saharan Africa [
10,
20].
Reported contraceptive use prior to unintended pregnancy was high (64 %), but the methods used were exclusively short-acting requiring daily (OC), quarterly adherence (DMPA), or with every act of sexual intercourse (male condom). The vast majority of women with an unintended pregnancy (81 %), regardless of HIV serostatus, reported relying on the male condom for contraception. Our findings suggest that reliance on condoms for pregnancy prevention is not an effective strategy. This is supported by recent studies in areas of high HIV prevalence, which have shown that more effective methods of contraception are substituted for male condom-only use [
20‐
22]. Women and couples should be informed that while male condoms are the only contraceptive method that can reduce the risk of sexually transmitted infections including HIV and that their correct and consistent use is imperative in that regard, they may have a high failure rate for pregnancy prevention. This is likely due to low levels of correct and consistent use, with a typical-use contraceptive failure rate for the male condom of about 21 % within the first year [
23]. Ideally in the context of high levels of HIV infection, a dual method approach (i.e., combining condoms, male or female, with a highly effective contraceptive method) should be promoted [
4].
Our data highlight the urgent need for women’s access to effective contraceptive methods that align with pregnancy intentions and reduce the potential for incorrect or inconsistent use. LARCs such as the intrauterine device (IUD) and the contraceptive implant reduce or eliminate the need for daily or per-act of intercourse adherence. The IUD is not frequently provided in Botswana’s public health sector, with its use among women aged 12–49 having peaked at 4.1 % in the late 1980’s and declining since to about 0.8 % in 2007, due to now disproved safety concerns about IUD use in the context of HIV [
24]. Sterilization has never been widely practiced in Botswana (utilized among <2 % of women aged 12–49 from 1985 to 2007) [
24]. Fortunately, the Botswana Sexual and Reproductive Health Department has recently prioritized increasing the supply of and demand for the LARC methods (personal communication, Sexual and Reproductive Health Division, Botswana Ministry of Health).
Further objectives of our study were to examine heterogeneity in unintended pregnancy, contraceptive use and future childbearing desires by HIV serostatus. We found that 26 % of HIV-infected women reported not knowing their HIV status prior to conception. While it is possible that some of these women did in fact know that they were HIV-infected but were not comfortable disclosing this at the time of study enrollment, this may also suggest missed opportunities for HIV diagnosis prior to pregnancy, and thus missed opportunities for pregnancy planning and safer conception or contraceptive counseling for HIV-infected women. Sixty-nine women (15 %) who reported known HIV-uninfected status before pregnancy were diagnosed with HIV during their pregnancy, possibly reflecting a high risk of seroconversion in pregnancy than has been previously reported in Botswana [
25] or elsewhere (some of these women may also have known that they were HIV-infected pre-conception but not disclosed this at study enrollment, as noted above). These findings support the need for continuing HIV-prevention efforts before and during pregnancy.
Levels of unintended pregnancy were high irrespective of knowledge of HIV serostatus prior to becoming pregnant. The prevalence of unintended pregnancy was significantly higher among women who reported not knowing their HIV status (52 %) compared to those who knew they were HIV-uninfected (38 %) even after controlling for other factors. Considering that HIV-infected women generally maintain regular contact with health services and may have strong motivations to prevent pregnancy, it is discouraging that unmet needs for effective family planning were high. That women of unknown HIV status report the highest levels of unintended pregnancy is not surprising given that these women may be the least engaged with health services of any kind.
Levels of contraceptive use prior to the unintended pregnancy and method type did not differ by knowledge of HIV serostatus prior to becoming pregnant. This contrasts with recent data from Zambia [
26] where contraceptive use at time of conception was higher in HIV-infected women than their HIV uninfected counterparts. Other studies from sub-Saharan Africa have shown that HIV-infected women may have poorer access to contraception than HIV-uninfected women, potentially due to barriers accessing family planning services or fear of stigma from providers [
27‐
29]. Previous studies from Southern Africa have found that safe and effective long-acting methods are sometimes not recommended by providers or accessible to women living with HIV due to limited knowledge and skills of the health care workers providing HIV services; lack of operational guidelines; and poorly integrated reproductive health/family planning and HIV services [
22,
30‐
32]. Historically, HIV and family planning services have been provided separately in Botswana. There are current efforts by the Botswana Ministry of Health to integrate and link strategies across health services [
33]. Our data suggest that in Botswana the challenges to obtaining effective contraception may apply relatively equally to both HIV-uninfected and HIV-infected women.
Most participants (61 %) reported not wanting additional children in the future, and this was strongly and independently associated with being HIV-infected, though it is important to note that a large proportion of HIV uninfected women also reported not wanting more children. Unfortunately, we did not collect data on post-partum contraceptive counseling or use, and thus cannot comment on how women’s post-partum contraceptive use will or will not allow them to meet their stated future childbearing desires. However, the methods most widely available and promoted in Botswana at present (male condoms and OCs) are unlikely to enable many of these women to achieve their stated desires. Long-acting reversible and permanent methods are considered the most appropriate for women who do not want more children [
34,
35].
It is important to note that while this study was focused on unintended pregnancy, 51 % of pregnancies were intended among women who knew that they were HIV-infected at the time of conception. When asked about future childbearing plans, 23 % of HIV-infected women reported wanting (or not knowing if they wanted) more children in the future. Research indicates that health care providers seldom discuss family planning with their HIV-infected clients [
36]. Nevertheless, accumulating data, including these data from Botswana, indicate that HIV-infected women continue to seek and achieve pregnancy following diagnosis with HIV infection [
17,
36‐
38]. Supportive, informed patient–provider communication about fertility intentions and safer conception is essential in this context.
This study has limitations. We assessed HIV status at the time of conception by self-report and did not have data on the timing of the most recent HIV test. Our pregnancy intention measure was a simplification of commonly used measures, and did not distinguish unwanted versus mistimed pregnancies, limiting direct comparison with some other studies. Reporting of pregnancy intention may have been subject to reporting bias, especially if HIV-infected women are differentially counseled to delay or avoid childbearing and over-report unintended pregnancy, and the reporting of condom use by known HIV positive women may be subject to social desirability bias. Furthermore, in our analysis of factors associated with pregnancy intention there may be unmeasured confounders that would bias our estimates of independent effect, which must be considered when interpreting the results. Finally, we did not collect information on desire for contraception prior to conception or post -partum.
Competing interests
The authors declare that they have no competing interests.
Author contributions
GM led conduct of the overall study, conceived of this analysis, and led the writing of the manuscript. KW conducted the main analysis. CM helped direct additional analyses and contributed to manuscript writing including reference review and selection. SM assisted with manuscript writing and review. GA helped lead the overall study, and assisted with reviewing the manuscript. MD, MS, JDM, and KM interviewed and cared for study participants, and reviewed the manuscript, JL assisted with data preparation, analysis and interpretation, JM contributed to overall study oversight and the manuscript. BK and SL designed and led the overall study and assisted with writing and review of the manuscript. All authors read and approved the final manuscript.