In sub-Saharan Africa, HIV continues to be one of the primary health issues of our time, especially among women and children. In Malawi, for example, HIV prevalence among antenatal women is 19.8%.1 This burden of disease has an enormous impact on all aspects of life in Malawi, and despite growing access to highly active antiretroviral therapy (HAART),2 HIV-related stigma and mortality remain pervasive.
Most interventions for prevention of mother-to-child transmission (pMTCT) of HIV have focused on the prevention of secondary transmission by means of peripartum antiretroviral therapy (ART).3 Modelers have determined, however, that a strategy focusing on contraception rather than on single-dose ART could avert 28.6% more HIV-positive births at the same level of expenditure.4 In line with these findings, the World Health Organization (WHO) recently listed “preventing unintended pregnancies among HIV-infected women” as one of its key strategies for decreasing the proportion of infants infected with HIV.5 This strategy involves improving access to family planning (FP) programs, providing FP clients with access to HIV voluntary counseling and testing (VCT) services, and supporting HIV-infected women in making informed choices about their reproductive lives.5
The need to combine FP and pMTCT strategies in sub-Saharan Africa is all too apparent. In Malawi, for example, only 26% of all women report the use of any contraceptive method6 and the fertility rate is 6.0, with a calculated pregnancy incidence of 16.7 per 100 person-years.6,7 In sub-Saharan Africa, as many as 50% of pregnancies among HIV-infected women are considered unintended.8,9 Although a growing number of short-course ART programs for pMTCT are available, access to those services is not universal. Even with ART support, approximately 15% of exposed infants are infected at 4 to 6 weeks, and as many as 20% are infected after 1 year of breast-feeding.10,11 In Malawi, HIV infection is responsible for approximately 9% of all pediatric deaths in children younger than 5 years of age.12
There are efforts in developing countries to address issues of primary pMTCT directly, including access to FP counseling and services for HIV-infected women.13-16 To inform such efforts, the purpose of this prospective observational study was to assess the effect of learning one's positive HIV status on intentions to have future children and on contraceptive use among HIV-infected women in Malawi. This study also sought to assess differences in pregnancy incidence after receipt of positive HIV test results according to stated pregnancy intentions. As a secondary objective, we sought to examine differences in each of these relations by women's perceived health, number of children, disclosure of HIV status, recruitment site, and marital status.
Study Design, Setting, and Population
This prospective cohort study was conducted at 3 clinical sites within Kamuzu Central Hospital in Lilongwe, Malawi: an FP clinic, a sexually transmitted disease (STD) clinic, and a VCT center. The Institutional Review Boards (IRBs) in Malawi and at the University of North Carolina at Chapel Hill approved the protocol and consent forms. All women of unknown HIV status presenting to these clinics from December 2003 through January 2005 were invited to participate. After providing informed consent, participants completed a questionnaire about demographics, pregnancy intentions, contraceptive use, perceived health, and reproductive history. After completing the questionnaire, participants received pregnancy testing, VCT, and FP counseling. Women found to be HIV-positive and not pregnant were offered enrollment. Those who enrolled were followed for 1 year, with visits 1 week and 1 and 3 months after VCT and quarterly thereafter.
Data and Specimen Collection
At each visit, participants received a physical examination and urine pregnancy test, and they answered the same questionnaire administered at screening, with additional questions about disclosure of HIV results. HIV testing conducted at screening consisted of 2 parallel rapid tests (Determine; Abbott Laboratories, Abbott Park, IL and Unigold; Trinity Biotech, Wicklow, Ireland), with Western blot testing to resolve discordant rapid test results. CD4 cell testing was conducted at screening for all HIV-infected participants and was repeated at the 6-month and 12-month visits. Women with CD4 counts ≤200 cells/mm3 and women found on the physical examination to have stage 3 or 4 HIV disease according to the WHO classification were referred immediately to the free ART clinic in the adjoining building. All pregnant women with CD4 counts >200 cells/mm3 who were not eligible for ART at the conclusion of the study were referred to the HIV outpatient clinic for semiannual evaluations.
We examined 3 main relations: (1) the effect of learning one's HIV-positive status on pregnancy intentions, (2) the effect of learning one's HIV-positive status on self-reported contraceptive use, and (3) the association between pregnancy intentions and pregnancy incidence. We also evaluated differences in each of these relations by participants' perceived health, number of living children, disclosure of HIV-positive status to partners, recruitment site, and marital status.
Pregnancy intentions at each visit were dichotomized according to participants' responses to the question, “Would you like to have another child, or would you not like to have any more children?” Contraceptive use at each visit was considered in general, for condom use in particular, and for “dual contraception” (ie, use of 2 contraceptive methods, including condoms) based on participants' response to the question, “Which methods of family planning are you currently using?” Perceived health status at each visit was dichotomized as excellent/good versus poor/fair according to participants' responses to the question, “How would you rate your current health status: poor, fair, good, or excellent?” The number of living children was dichotomized (0 live children vs. ≥1 live child) based on participants' responses to the question at the time of enrollment, “How many children do you have living?” We also explored the number of living children as a 4-category nominal variable (0 live children, 1 live child, 2 live children, or ≥3 live children). Disclosure to partners was dichotomized based on participants' response to the question, “Since your last visit, with whom have you shared your HIV result?” Marital status was dichotomized as married versus unmarried.
We used generalized estimating equations (GEEs) with a logistic link function to examine the effect of knowing one's HIV-positive status on dichotomized pregnancy intentions. In a first stage of modeling, we used knowledge of HIV status (0 at baseline and 1 thereafter) as the main predictor variable, and we included a linear spline term starting at the week 1 follow-up visit to model the trend in pregnancy intentions over time. We ran 6 separate additional models with interaction terms to check for effect measure modification by (1) perceived health status, (2) existence or absence of live children, (3) number of live children (0, 1, 2, or ≥3), (4) disclosure to partner, (5) recruitment site, and (6) marital status.
We also used GEE models to examine the effect of knowing one's HIV-positive status on self-reported contraceptive use. In the first of these models, the dependent variable was any contraceptive use. In the second, the dependent variable was condom use or dual use. In both models, we used knowledge of HIV status as the main predictor variable and we included piecewise linear spline terms (one starting at the week 1 visit and another starting at 90 days after screening) to model trends in contraceptive/condom use over time. For the outcome of any contraceptive use, we ran 6 additional models to assess effect measure modification by the same covariates listed in the previous paragraph.
We used Poisson regression to examine the association between pregnancy intentions 1 week after VCT and pregnancy incidence during follow-up, using interaction terms and likelihood ratio tests to assess separately each of the 6 potential effect measure modifiers listed previously. We also assessed each of these covariates as predictors of pregnancy incidence in separate Poisson regression models.
In longitudinal analyses, women contributed to an analysis until they were found to be pregnant or lost to follow-up, and they were censored thereafter. For cross-sectional comparisons of baseline data across clinic settings and by HIV status, we used χ2 tests for categoric data and t tests for continuous data. We conducted all statistical analyses with SAS 8.2 (SAS Institute, Cary, NC).
A total of 1709 women of unknown HIV status presented to the study sites during the study period. Of these women, 816 (47.7%) consented to participate in the screening phase of the study; 261 (32.0%) of these participants tested HIV-positive. Twenty-one percent (106 of 515 women) at the FP clinic, 43% (53 of 122 women) at the STD clinic, and 57% (102 of 179 women) at the VCT center had positive HIV test results (P < 0.05 for each pairwise comparison). Of the 261 positive women, 227 (87.0%) consented to continue their participation in the longitudinal study: 46 (20.3%) from the STD clinic, 96 (42.3%) from the FP clinic, and 85 (37.4%) from the VCT center. Among the 227 enrolled women, 17 (7.5%) died (median CD4 count = 128 cells/mm3) of AIDS-related causes and 10 (4.4%) were lost to follow-up. There were no demographic or immunologic differences between those lost to follow-up and those who completed the study (data not shown). Of the 67 surviving women with CD4 counts <200 cells/mm3, 53 (79.1%) successfully initiated ART during follow-up.
Baseline descriptions of the 227 enrolled HIV-positive women are presented in Table 1. Most of the women were in their twenties (median age = 26 years), married (72%), and of lower socioeconomic status (only 27% had electricity), and almost half (43%) were illiterate. The median CD4 count at screening was 295 cells/mm3. Of the 227 women enrolled, 33% had CD4 counts <200 cells/mm3. Thirty-eight percent were using some form of contraceptive at the HIV testing visit. Women attending the FP clinic were significantly more likely to be married than were women presenting to the VCT center (P < 0.0001) and STD clinic (P = 0.0001) sites. Additionally, women presenting to the FP clinic were significantly more likely to report contraceptive use than were women presenting to the VCT center and STD clinic sites (P < 0.0001 for both comparisons: FP clinic compared with VCT center, FP clinic compared with STD clinic) and had significantly more living children (P = 0.0001 comparing FP clinic with VCT center and P = 0.001 comparing FP clinic with STD clinic). Women presenting to the VCT center were significantly more likely to have electricity in their homes than were women attending the FP clinic (P = 0.02) or STD clinic (P = 0.007). The mean CD4 cell count was highest at the FP clinic (P < 0.0001 in comparison to the VCT center and P = 0.02 in comparison to the STD clinic). As expected, perceived health status was associated with CD4 cell count (data not shown); those with CD4 counts >200 cells/mm3 were significantly more likely to report “good” or “excellent” health than were women with CD4 counts ≤200 cells/mm3 (P = 0.0002).
Longitudinal Analysis: Effect of Knowing HIV Status on Pregnancy Intentions
As shown in Figure 1, the percentage of HIV-positive women stating a desire for a future child declined from 33% before receipt of HIV test results to 15% 1 week later (P < 0.0001 comparing pregnancy intentions before and after test). The effect of knowing one's HIV-positive status remained constant over the 1-year follow-up period; the proportion of women reporting a desire for a future child ranged only from 15% to 19% across follow-up visits (P = 0.98 for a test of the null hypothesis that the rate of change from week 1 to the end of follow-up was 0).
The general trend of significantly reduced pregnancy intentions 1 week after receipt of HIV-positive test results did not differ by perceived health status, number of live children, disclosure to partner, recruitment site, or marital status (Fig. 2). In each subgroup, the decrease in pregnancy intentions was sustained throughout follow-up (P > 0.3 for within-subgroup tests of the null hypothesis that the rate of change from week 1 to the end of follow-up was 0; data not shown).
Longitudinal Analysis: Effect of Knowing HIV Status on Contraceptive Use
Use of a contraceptive increased from 38% before receipt of positive HIV results to 52% 1 week later (P < 0.0001; Fig. 3). Contraceptive use then held relatively constant through the 3-month visit (P = 0.46 for a test of the null hypothesis that the rate of change was 0 between week 1 and the 3-month visit) and then decreased over the remainder of follow-up (P = 0.06 for a test of the null hypothesis that the rate of change was 0 between the 3-month and 12-month visits), declining to 46% at the last visit.
Condom use was reported by 3.5% of women before receipt of positive HIV test results and increased to 7.5% 1 week later (P = 0.0014; see Fig. 3). The percentage of women reporting condom use then increased to 16% by the 3-month visit (P = 0.0023) and decreased thereafter (P < 0.0001), declining to only 4% at the 1-year follow-up visit (see Fig. 3). The percentage of women using a dual method was 0.4% (1 of 227 women) before and 1.32% (3 of 227 women) 1 week after HIV testing.
The general trend of significantly increased contraceptive use 1 week after receipt of HIV-positive test results was observed in each subgroup, with the exception of those with no living children, those who were unmarried, and those who enrolled at the VCT center (Fig. 4). Knowledge of HIV status had a significantly greater effect on contraceptive use among those who disclosed their status to their partner compared with those who did not disclose (P = 0.03), among those in the FP clinic compared with those at the other 2 sites (P < 0.0001), and among those who were married compared with those who were not (P = 0.03). In all subgroups, contraceptive use held constant or increased slightly from 1 week through 3 months and then remained stable or decreased over the remainder of the study period (data not shown).
Longitudinal Analysis: Association Between Pregnancy Intentions and Pregnancy Incidence
During the follow-up period, 29 (12.8%) of the 227 women became pregnant, for an overall pregnancy incidence of 14.5 pregnancies per 100 person-years (Table 2). The pregnancy incidence (26.6 pregnancies per 100 person-years) among women who said they wanted a future child 1 week after receipt of HIV-positive test results was 2.2 (95% confidence interval [CI]: 1.01 to 4.87) times as high as the incidence rate (12.0 pregnancies per 100 person-years) among women who did not report a desire for a future child. Nevertheless, 62.1% (18 of 29) of the pregnancies occurred among women who reported not wanting a future child at week 1 (data not shown) and were considered unintended.
We did not detect any statistically significant associations between pregnancy incidence and screening CD4 cell counts, perceived health at week 1, marital status, disclosure of HIV-positive status to partner, number of living children, or recruitment site (see Table 2). Further, the association between pregnancy intentions and pregnancy incidence was not modified by any of these covariates (P > 0.4 for each likelihood ratio test comparing interaction-term models with a model containing week 1 pregnancy intentions as the only predictor).
From 10% to 50% of all pregnancies, including those among HIV-infected women, are considered unintended,8,9 but a woman's fertility intentions may not be closely correlated with her subsequent reproduction.17-21 Factors affecting pregnancy intentions and fertility include women's educational and economic opportunities and status, their decision-making relations with their partners, their and their partners' access to FP counseling and services, the survival of previous children, pressures from the community and extended family, urban versus rural residence, religion, HIV status, and the overall psychosocial and physiologic imperative to bear children.6,9,22,23
The percentage of women in this clinic-based cohort who reported a desire for future children declined significantly and quickly after the women were armed with knowledge of their HIV-positive status, and these attitudes persisted for the full year of follow-up. Furthermore, the pregnancy incidence rate among women who expressed a desire for no future children was less than half of that among women who desired future children. Our findings suggest that decreased fertility among HIV-positive women has a significant behavioral component, adding to the physiologic mechanisms proposed in previous reports of decreased fertility among HIV-positive women.24-29
There were several limitations to this study. This cohort of HIV-infected women acted as its own control group by providing data before HIV testing; we did not follow a cohort of HIV-negative women to compare changes in outcomes. It is possible that a similar clinical and counseling intervention would lead to reduced pregnancy intentions and increased contraceptive use among such a control group; however, we would expect the effect to be less pronounced. We also note that our 48% recruitment rate may have resulted in some bias, but we do not have data on demographics and HIV status among the nonparticipants, so we cannot predict the extent of any such bias. It is also possible that the time frame implied by the question “Would you like to have another child?” was interpreted differently across women, potentially resulting in ambiguity of the pregnancy intentions outcome. Finally, because we included an FP clinic as one of our recruitment settings, the increased uptake of contraceptive use after VCT may have been influenced by prior intent. We note, however, that most women (64%) from this clinic were already using contraceptives at the baseline visit.
Previous studies in South Africa, Côte d'Ivoire, and Rwanda30-32 have shown that knowledge of HIV-positive results had little effect on pregnancy intentions, especially among women with no children or with fewer than the perceived optimal number of offspring. In contrast, we found that receipt of HIV-positive test results led to a significant reduction in pregnancy intentions, regardless of the number of existing children. Consistent with other studies,33,34 we found a greater desire for future children among women with fewer existing children and among women with a positive health perception. We found that the intent for future children was predictive of pregnancy incidence but that the number of existing children or health perception was not. We believe this finding was the result of the supportive reproductive counseling and the easy access to FP and HIV clinical services that accompanied each follow-up visit.
The estimated overall pregnancy incidence in Malawi is 16.7 per 100 person-years, which is higher than the 12.0 per 100 person-years we found among the HIV-infected women who did not state a desire for future children after receipt of HIV results but lower than the 26.6 per 100 person-years we found among the 15% of infected women who stated a desire for future children. It is possible that new knowledge of one's HIV-positive status may have accelerated the pregnancy rate among a few of these HIV-positive women.
The increased use of a contraceptive method after receipt of a positive HIV test result was not as long-lasting as the stated desire to have or not to have a future child, likely because of the fertility factors stated previously35 and the side effects of the more popular and available hormonal methods.36 There are clear needs for, and advantages to, easy and institutionalized access to FP counseling and services around the time of HIV testing. Each woman who tests HIV-positive should receive informational counseling about the effects of pregnancy and childbirth on maternal and infant mortality and on the positive and negative social and personal implications of their reproductive choices. This model has been shown to work in pilot projects in Zambia,17 where FP uptake doubled, and Kenya, where post-HIV result contraceptive uptake reached 72%.37 Additionally, barriers to FP services should be reduced; optimally, FP services could be integrated into HIV care.
In this cohort of HIV-infected and mostly married women, it was not surprising that so few reported condom use, because condom use is generally low in Malawi.6 Condoms serve contraception and HIV prevention functions, and their use relies on participation by the male partner, underscoring the importance of male involvement in FP and HIV counseling. Not surprisingly, we found that women who disclosed their HIV-positive status to their partner had a significant increase in contraceptive uptake compared with women who did not disclose their results. This finding suggests the need for stronger proactive partner notification systems that include couples counseling.
In summary, after Malawian women have been notified of their HIV infection and have received FP counseling, they are less likely to desire future pregnancies and more likely to use a contraceptive method. If they do not desire future pregnancies, they are less likely to become pregnant within a year of receiving their positive HIV test results. Based on these findings, we recommend the implementation of HIV testing and counseling services in clinical settings such as FP, antenatal, and STD clinics. In these settings, we suggest FP counseling and easy access to contraceptive services for all HIV-positive women at the time HIV test results are received. This integration of VCT, FP, and HIV care could lead to an increase in primary pMTCT of HIV.
The authors acknowledge and thank all the administrative and clinical staff of the Kamuzu Central Hospital STD and FP Clinics as well as Sam Phiri and the staff of the Lighthouse HIV Clinic and counseling and testing center for their cooperation and assistance in this study. They also thank all the staff of the University of North Carolina Project for their accurate, caring, and excellent work in the implementation of this study as well as the study subjects who have provided us with data that we hope eventually translates into improved services for the HIV-infected women of Malawi. The authors thank the Bill and Melinda Gates Institute for Population and Reproductive Health for their generous funding.
1. Bello GA, Chipeta J, Aberle-Grasse J. Assessment of trends in biological and behavioral surveillance data: is there any evidence of declining HIV prevalence or incidence in Malawi? Sex Transm Infect
. 2006;82(Suppl 1):i9-i13.
2. Harries AD, Schouten EJ, Libamba E. Scaling up antiretroviral treatment in resource-poor settings. Lancet
3. Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet
4. Reynolds HW, Janowitz B, Homan R, et al. The value of contraception to prevent perinatal HIV transmission. Sex Transm Dis
5. WHO. Strategic approaches to the prevention of HIV infection in infants: a report of a WHO meeting. Morges, Switzerland, March 20-22, 2002; Section 4.2:10.
6. Malawi Demographic and Health Survey 2004
. Calverton, MD: National Statistical Office (Malawi) and ORC Macro; 2005.
7. van den Broek NR, White SA, Ntonya C, et al. Reproductive health in rural Malawi: a population-based survey. Br J Obstet Gynaecol
8. United Nations Family Planning Association. United Nations Family Planning Association 2000 Annual Report
. New York: United Nations Population Fund, 2001.
9. Desgrées-du-Lou A, Msellati P, Viho I, et al. Contraceptive use, protected sexual intercourse and incidence of pregnancies among African HIV-infected women. DITRAME ANRS 049 Project, Abidjan 1995-2000. Int J STD AIDS
10. Biggar RJ, Miotti PG, Taha TE, et al. Perinatal intervention trial in Africa: effect of a birth canal cleansing intervention to prevent HIV transmission. Lancet
11. Taha TE, Brown ER, Hoffman IF, et al. A phase III clinical trial of antibiotics to reduce chorioamnionitis-related perinatal HIV-1 transmission. AIDS
12. Walker N, Schwartländer B, Bryce J. Meeting international goals in child survival and HIV/AIDS. Lancet
13. Peck R, Fitzgerald DW, Liautaud B, et al. The feasibility, demand, and effect of integrating primary care services with HIV voluntary counseling and testing: evaluation of a 15-year experience in Haiti, 1985-2000. J Acquir Immune Defic Syndr
14. Maharaj P, Cleland J. Integration of sexual and reproductive health services in KwaZulu-Natal, South Africa. Health Policy Plan
15. Shelton JD, Fuchs N. Opportunities and pitfalls in integration of family planning and HIV prevention efforts in developing countries. Public Health Rep
16. Dehne KL, Snow R, O'Reilly KR. Integration of prevention and care of sexually transmitted infections with family planning services: what is the evidence for public health benefits? Bull World Health Organ
17. Allen S. Why is fertility an issue for HIV-infected and at-risk women? J Acquir Immune Defic Syndr
. 2005;38(Suppl 1):S1-S2.
18. Hirschman C. Why fertility changes. Annu Rev Sociol
19. Caldwell J, Caldwell P. The cultural context of high fertility in sub-Saharan Africa. Popul Dev Rev
20. Bankole A, Westoff CF. The consistency and validity of reproductive attitudes: evidence from Morocco. J Biosoc Sci
21. Williams L, Abma J. Birth wantedness reports: a look forward and a look back. Soc Biol
22. Allen S. International data. J Acquir Immune Defic Syndr
. 2005;38(Suppl 1):S7-S8.
23. Yacouba N, Meda N, Valériane L, et al. Sexual and reproductive life of women informed of their HIV seropositivity: a prospective cohort study in Burkina Faso. J Acquir Immune Defic Syndr
24. Zaba B, Gregson S. Measuring the impact of HIV on fertility in Africa. AIDS
. 1998;12(Suppl 1):S41-S50.
25. Carpenter LM, Nakiyingi JS, Ruberantwari A, et al. Estimates of the impact of HIV-1 infection on fertility in a rural Ugandan population cohort. Health Transit Rev
. 1997;7(Suppl 2):S113-S126.
26. Gray RH, Wawer MJ, Serwadda D, et al. Population-based study of fertility in women with HIV-1 infection in Uganda. Lancet
27. Fabiani M, Nattabi B, Ayella EO, et al. Differences in fertility by HIV serostatus and adjusted HIV prevalence data from an antenatal clinic in northern Uganda. Trop Med Int Health
28. Lindsay MK, Grant J, Peterson HB, et al. The impact of knowledge of HIV serostatus on contraceptive choice and repeat pregnancy. Obstet Gynecol
29. Glynn JR, Buve A, Carael M, et al. Decreased fertility among HIV-1-infected women attending antenatal clinics in 3 African cities. J Acquir Immune Defic Syndr
30. Allen S, Serufilira A, Gruber V, et al. Pregnancy and contraception use among urban Rwandan women after HIV testing and counseling. Am J Public Health
31. Aka-Dago-Akribi H, Desgrées Du Loû A, Msellati P, et al. Issues surrounding reproductive choice for women living with HIV in Abidjan, Côte d'Ivoire. Reprod Health Matters
32. Cooper D, Harries J, Myer L, et al. ‘Life is still going on’: reproductive intentions among HIV-positive women and men in South Africa. Soc Sci Med
33. Loko M, Toure S, Dakoury-Dogbo N, et al. Decreasing incidence of pregnancy by decreasing CD4 cell count in HIV-infected women in Côte d'Ivoire: a 7-year cohort study. AIDS
34. Nebié Y, Meda N, Leroy V, et al. Sexual and reproductive life of women informed of their HIV seropositivity: a prospective cohort study in Burkina Faso. J Acquir Immune Defic Syndr
35. Shah I. Contraceptive use patterns in countries with different levels of HIV epidemic. J Acquir Immune Defic Syndr
. 2005;38(Suppl 1):S5-S6.
36. The Practice Committee of the American Society for Reproductive Medicine. Hormonal contraception: recent advances and controversies. Fertil Steril
. 2006;86(Suppl 1):S229-S235.
37. Balkus J, Bosire R, John-Stewart G, et al. High uptake of postpartum hormonal contraception among HIV-1-seropositive women in Kenya. Sex Transm Dis