HIV Infection and Pregnancy Status Among Adults Attending Voluntary Counseling and Testing in 2 Developing Countries

Abstract

Objectives. This study investigated the impact of HIV voluntary counseling and testing (VCT) on reproduction planning among 1634 adults in 2 sub-Saharan countries.

Methods. Data were obtained from a multisite randomized controlled trial.

Results. At 6 months post-VCT, the women more likely to be pregnant were younger (odds ratio [OR] = 2.5; 95% confidence interval [CI] = 1.0, 6.5), not using contraceptives (OR = 0.1; 95% CI = 0.1, 0.3), and HIV infected (OR = 3.0; 95% CI = 1.3, 7.0). An interaction emerged linking pregnancy intention at baseline and HIV serostatus with pregnancy at follow-up (OR = 0.1; 95% CI = .0, 0.4) Partner pregnancy rates did not differ by HIV serostatus among men.

Conclusions. HIV diagnosis may influence reproduction planning for women but not for men.

The effect of an HIV diagnosis on reproduction planning in developing countries is not well understood. Clinical symptoms of disease,1,2 coexisting sexually transmitted infections,3 and HIV-induced amenorrhea4 are associated with reduced fertility. HIV infection may also result in pregnancy complications such as decreased birthweight, prematurity, and elevated risk of preterm delivery.5

Psychosocial factors also affect reproduction planning and fertility among individuals infected with HIV, both in developed and developing countries. Pregnancy after HIV diagnosis has been associated with relationship stability,6,7 comorbid substance abuse,8,9 reduced sexual activity,10 and decreased contraception use.4,11,12 In addition, childbearing after knowledge of HIV infection has been associated with nulliparity,1,3,13 lack of information about vertical transmission of HIV,14,15 perceived absence of clinical illness,16 and the wish to avoid stigma, abandonment, and other repercussions of being perceived to be infected or unable to bear offspring.13,17,18

Missing from many of these studies is consideration of the effect of intention to bear offspring on reproduction following HIV diagnosis. In developed countries, HIV diagnosis appears to lead many women to avoid pregnancy or to undergo pregnancy termination,6,7,9 decisions that may be attributed in part to availability of and access to voluntary counseling and testing (VCT), modern contraception, and safe and legal abortion services.3,4 By contrast, studies in several African countries have suggested that HIV diagnosis has little effect on subsequent childbearing.2,11,19 Yet, 1 limitation of these studies is that none have directly examined the effect of intention to reproduce on reproduction following HIV diagnosis. One exception, a study of HIV-infected and uninfected Rwandan women, showed that both groups expressed equally the desire for additional offspring after HIV testing,17 suggesting that knowledge of HIV infection and intention to reproduce following diagnosis may interact to influence subsequent reproduction.

METHODS

We examined the association of intention to bear children and knowledge of HIV serostatus with subsequent reproductive behavior among men and women receiving HIV voluntary counseling and testing in Kenya and Tanzania. In addition to providing descriptive analyses, we tested the direct and interactive effects of HIV serostatus and intention to reproduce on pregnancy status 6 months post-VCT. Assuming that the desire to avoid vertical transmission would outweigh contextual factors favoring reproduction among infected individuals,7,18,20,21 we hypothesized that participants diagnosed with HIV infection who also reported intention to reproduce before learning their serostatus would be less likely than uninfected participants to be pregnant or to have a pregnant partner 6 months post-VCT.

All participants included in this analysis were drawn from the Voluntary HIV Counseling and Testing Efficacy Study, a randomized controlled trial evaluating the efficacy of VCT in changing sexual risk behavior among adults in Nairobi, Kenya, and Dar es Salaam, Tanzania.22 The target study population included individuals and couples seeking HIV-related services in an urban public hospital in Dar es Salaam and a free-standing clinic in a low-income suburban settlement in Nairobi. Site-specific recruitment and intervention materials were developed.23

The study involved a longitudinal design with assessments at baseline (time 1), 6 months post-VCT (time 2), and 12-month follow-up (time 3). Participants were randomized to 1 of 2 conditions: an ideographic VCT intervention designed to facilitate HIV risk behavior change, or a comparison condition in which participants received general health information relevant to HIV (HI). At time 2, participants in the comparison group received the VCT intervention, such that by time 3 all participants had received VCT when they returned for assessment. Given that the focus of this study was pregnancy status within 6 months of HIV serotesting, only data corresponding to pre- (baseline) and post-VCT (follow-up) assessments were used for this study. Only participants unaware of their HIV serostatus at screening were eligible to enroll in the study. A detailed description of the VCT protocol is available elsewhere.22

Intervention Protocol

After baseline assessment of demographic, psychosocial, and sexual behavior variables, all participants were randomly assigned to receive either the VCT or the HI intervention. The VCT intervention was developed according to guidelines from the Centers for Disease Control and Prevention24 and the World Health Organization Global Programme on AIDS. The intervention included a personalized risk assessment and development of an individualized risk reduction plan. Couples were counseled together, but members of each dyad received test results individually first. Testing for HIV was conducted with a commercial enzyme-linked immunosorbent assay (ELISA), and all reactive tests were confirmed with a second ELISA. Discrepancies between first and second tests were resolved by Western blot antibody testing. Participants were scheduled to return 2 weeks after testing to receive results and posttest counseling. In the HIV comparison condition, participants watched a brief, site-specific videotape that included a condom use demonstration; any questions about HIV transmission were addressed by a health educator or nurse. No HIV antibody testing was conducted with this group at baseline. Following both interventions, participants received 25 condoms and information for their correct use; returning to the site for additional condoms was encouraged.

Data Analysis

Multiple logistic regression was used to assess associations of demographic variables, intention to bear children, and HIV serostatus with the pregnancy status of female participants and male participants’ female partners as reported by them at 6 months post-VCT (time 2). Model fitting procedures included examination of full and reduced models via log likelihood estimates to provide tests of improvement in fit. For women and men, predictors of pregnancy or partner pregnancy at 6 months included main effects for age, married or cohabiting relationship status, number of children, modern contraception use (e.g., condoms, oral contraception, intrauterine device), HIV serostatus, and intention to become pregnant (or to impregnate a sexual partner) at baseline. In addition, participants indicated their awareness of vertical transmission risk by responding correctly that mothers can transmit HIV to their infants, even through breastfeeding. HIV disease progression was estimated from selfreported symptoms associated with HIV infection, including night sweats, fever, and swollen lymph nodes. Group assignment was entered as a covariate to control for possible effects of participation in VCT at different periods in the design. Likewise, 2 variables were entered to control for frequency of sexual activity and recruitment site. Finally, an interaction term was tested to examine the multiplicative effect of pregnancy intention at baseline and HIV serostatus on pregnancy or partner pregnancy at follow-up. Variables theorized to be relevant predictors of pregnancy status were included in multivariate models. Analyses were conducted separately by sex.

RESULTS

Of 2942 eligible participants, 17% were sexually abstinent at baseline (n = 494), 17% did not return for initial HIV antibody results (n = 494), 5% were pregnant (n = 158), 3% were men 60 years or older (n = 73), 1% were women 40 years or older (n = 26), and 1% reported only same-sex partners (n = 24). One percent of women reported a history of tubal ligation at baseline (n = 39). After we excluded these participants, data for 1634 individuals remained for secondary data analyses. Table 1 ▶ provides descriptive data for the total sample and for men and women separately. Participants’ mean age was 27.4 (standard deviation [SD] = 6.6, range = 16–57 years); 89% were aged 20–39 years (n = 1441). Nearly 50% of participants were women (n = 774), 54% (n = 862) were married or cohabiting, and 57% (n = 925) reported at having least 1 child. Sixty-five percent (n = 1062) were Catholic or Protestant. More than 45% of participants (n = 747) reported use of modern contraceptives, including oral contraceptives, intrauterine devices, and condoms. Before receiving HIV testing results, 23% (n = 374) were planning a pregnancy within 2 months, and most (75%) of these participants were aware of the potential for vertical transmission of HIV to offspring (n = 1193). Antibody testing revealed that 19% of the sample (n = 306) was infected with HIV, and that prevalence among women (27%) was more than twice that among men (11%). Twenty-eight percent (n = 461) endorsed 2 or more symptoms often associated with HIV infection, including night sweats, fever, and swollen lymph nodes. Half of participants were recruited from Kenya.

TABLE 1.

—Summary of Key Variables at Baseline (n = 1634)

Variable	Sample % (n)	Men % (n)	Women % (n)
Age, y
16–19	7 (117)	6 (47)	9 (70)
20–29	60 (975)	56 (478)	64 (497)
30–39	29 (466)	30 (259)	27 (207)
≥ 40	6 (75)	9 (75)	. . .
Female	47 (774)	. . .	. . .
Married/cohabiting	54 (862)	48 (407)	60 (455)
No. of children
None	43 (685)	52 (438)	32 (247)
1–2	37 (600)	31 (264)	44 (336)
≥ 3	20 (325)	17 (143)	24 (182)
Religion
Muslim	30 (481)	35 (294)	24 (187)
Catholic	34 (552)	32 (274)	36 (278)
Protestant	31 (510)	28 (234)	36 (276)
Other	1 (19)	1 (7)	2 (12)
None	4 (61)	5 (43)	2 (18)
Contraception use	46 (747)	47 (406)	44 (341)
Planning a pregnancy	23 (374)	19 (157)	28 (217)
Aware of vertical transmission of HIV	75 (1193)	70 (586)	80 (607)
HIV infected	19 (306)	11 (98)	27 (208)
Reported HIV-related symptoms (≥ 2)	28 (461)	28 (236)	29 (225)
Recruitment site
Kenya	51 (828)	50 (427)	52 (401)
Tanzania	49 (805)	50 (432)	48 (373)

Note. Totals may not sum to n = 1634 owing to missing observations.

Table 2 ▶ presents univariate findings for women and men. We found higher follow-up pregnancy rates among women who were younger, had fewer than 3 children, were not using modern contraceptives, and were planning a pregnancy at baseline. No univariate differences were found for relationship status, religion, or knowledge of the risk for vertical transmission of HIV. Also, as Table 2 ▶ shows, we found higher partner pregnancy rates among men who were married or cohabiting, not using modern contraceptives, planning a pregnancy at baseline, and aware of the risk for vertical transmission of HIV. We found no significant univariate differences in pregnancy rate by HIV serostatus, self-reported HIV symptoms, or recruitment site for men or women.

TABLE 2.

—Univariate Predictors of Pregnancy Status at Follow-up

	Women (n = 808)			Men (n = 826)
Variable (Reference)	Pregnant, % (n)	Not Pregnant, % (n)	OR (95% CI)	Pregnant, % (n)	Not Pregnant, % (n)	OR (95% CI)
Age, y ( < 30)	88 (52)	72 (418)	2.9 (1.3, 6.5)*	68 (42)	60 (405)	1.4 (0.8, 2.5)
Married/cohabiting	67 (39)	58 (329)	1.5 (0.9, 2.7)	66 (40)	46 (307)	2.3 (1.3, 3.9)*
No. of children ( ≥ 3)	12 (7)	26 (151)	0.4 (0.2, 0.9)*	12 (7)	18 (120)	0.6 (0.3, 1.4)
Religion (Christian)	71 (42)	72 (419)	1.0 (0.5, 1.7)	68 (42)	59 (399)	1.5 (0.8, 2.5)
Contraceptive use	17 (10)	56 (326)	0.2 (0.1, 0.3)*	31 (19)	56 (381)	0.3 (0.2, 0.6)*
Planned pregnancy	41 (24)	27 (155)	1.9 (1.1, 3.2)*	35 (21)	18 (116)	2.5 (1.4, 4.5)*
Aware of vertical transmission	85 (50)	80 (454)	1.4 (0.7, 2.9)	87 (54)	68 (450)	3.2 (1.5, 6.8)*
HIV infected	27 (16)	25 (142)	1.1 (0.6, 2.1)	13 (8)	11 (75)	1.2 (0.5, 2.6)
No. of HIV symptoms ( ≥ 2)	36 (21)	35 (201)	1.0 (0.6, 1.8)	34 (21)	31 (208)	1.2 (0.7, 2.0)
Recruitment site (Kenya)	41 (24)	53 (309)	0.8 (0.5, 1.4)	47 (29)	50 (338)	0.9 (0.5, 1.5)

Note. Reference = reference group for odds ratios; OR = odds ratio; CI = confidence interval.

*P < .05.

Table 3 ▶ presents results of the multiple logistic regression model for women, which were significant (P < .001) and accounted for 18% of the variability in pregnancy status at follow-up. Women more likely to be pregnant 6 months after learning their serostatus were younger, not using modern contraceptives, and HIV infected. Presence of HIVassociated symptoms was not associated with follow-up pregnancy rates. There was a significant interaction between pregnancy planning and HIV serostatus, indicating that women who had received a recent HIV diagnosis and had expressed the intention to become pregnant before HIV testing were less likely to be pregnant at follow-up.

TABLE 3.

—Multiple Logistic Regression of Predictors of Time 2 Pregnancy for Women (n = 639)

Variable (Reference)	OR (95% CI)
Age, y (< 30)	2.5 (1.0, 6.5)*
Married	1.1 (0.5, 2.1)
No. of children (≥ 3)	0.8 (0.3, 2.1)
Contraceptive use	0.1 (0.1, 0.3)**
Aware of vertical transmission	1.7 (0.8, 3.8)
Planned pregnancy	1.7 (0.8, 3.5)
HIV infected	3.0 (1.3, 7.0)**
No. of HIV-related symptoms (≥ 2)	0.9 (0.4, 1.6)
Intention × infected	0.1 (0.0, 0.4)**
Frequency of sexual activity	0.9 (0.4, 1.6)
Treatment group	1.6 (0.8, 3.2)

Note. Reference = reference group for comparisons; OR = odds ratio; CI = confidence interval.

*P < .10; **P < .05.

Table 4 ▶ presents percentages, by pregnancy planning intention and HIV serostatus, of women reporting a pregnancy 6 months after HIV diagnosis. Post hoc comparisons confirmed that pregnancy rates for uninfected women who were not planning a pregnancy (6.7%) were not different from rates for infected women who were planning a pregnancy at baseline (5.8%; odds ratio [OR] = 0.4; 95% confidence interval [CI] = 0.11, 1.52). Women who were planning a pregnancy at baseline and were newly diagnosed with HIV infection reported fewer pregnancies (5.8%) than did infected women who were not planning a pregnancy (14.4%; OR = 0.3; 95% CI = 0.10, 0.92) and uninfected women planning a pregnancy (20.6%; OR = 0.1; 95% CI = 0.03, 0.57). Additional post hoc analyses confirmed that infected women reported more HIVassociated symptoms than did uninfected women (OR = 1.6; 95% CI = 1.2, 2.3) and that HIV-infected women who were planning a pregnancy at baseline reported fewer symptoms associated with HIV disease than did infected women who were not planning a pregnancy (χ²₁ = 4.4; P < .05, 1-sided Fisher exact test).

TABLE 4.

—Percentage of Women Reporting a Pregnancy Following HIV Diagnosis, by Pregnancy Planning Intention and HIV Serostatus

	Planning a Pregnancy
HIV serostatus	Yes	No
Infected	5.8	14.4
Uninfected	20.6	6.7

Table 5 ▶ presents results of the multiple logistic regression model for men, which were significant (P < .001) and accounted for 14% of the variability in partner pregnancy status at follow-up. We found that partner pregnancies were more likely among men who were younger, married or cohabiting, not using modern contraception, and aware of the risks of vertical transmission of HIV. These findings were independent of the negligible main effects for number of children, frequency of sexual activity, presence of HIV-associated symptoms, HIV serostatus, and planning a partner pregnancy at baseline.

TABLE 5.

—Multiple Logistic Regression of Predictors of Partner Pregnancy for Men (n = 586)

Variable (Reference)	OR (95% CI)
Age, y (< 30)	2.1 (1.0, 4.3)*
Married	2.0 (1.0, 4.1)*
No. of children (≥ 3)	0.7 (0.2, 1.7)
Contraceptive use	0.2 (0.1–0.5)*
Aware of vertical transmission	4.2 (1.7, 10.1)*
Planned pregnancy	1.2 (0.6, 2.4)
HIV infected	1.6 (0.5, 5.1)
No. of HIV-related symptoms (≥ 2)	1.1 (0.6, 2.1)
Intention × infected	0.5 (0.1, 3.4)
Frequency of sexual activity	0.7 (0.3, 1.3)
Treatment group	1.9 (1.0, 3.5)*

Note. Reference = reference group for comparisons; OR = odds ratio; CI = confidence interval.

*P < .05.

DISCUSSION

We found that women who were younger, not using contraception, aware of the risk of vertical transmission, and HIV infected were more likely than those with the opposite characteristics to be pregnant at time 2. Although pregnancy planning showed no main effect on pregnancy status for women, fewer of those who tested positive for HIV infection than of those who tested negative reported a pregnancy at 6 months post-VCT. Taken together, these findings suggest that diagnosis with HIV may have precipitated a reduction in fertility among healthier women while accelerating fertility among less healthy women, independent of the effects of relationship status, number of children, and frequency of sexual activity. Men who were younger, married or cohabiting, not using contraception, and aware of risks of vertical transmission were more likely to report that their partners were pregnant at 6 months post-VCT, regardless of their HIV serostatus. The latter findings suggest that men’s awareness of their HIV serostatus and baseline intention to reproduce were unrelated to partner pregnancies reported 6 months later.

Despite consistency of these findings with the literature with respect to the impact of age and relationship status on reproduction, important differences remained. For example, several studies have documented that knowledge and use of contraceptives is low in many sub-Saharan countries,25 and yet nearly half of participants in this study reported using modern contraceptives. These data were derived from a randomized controlled trial rather than a population-based survey; selfselection bias may account for the differences in observed contraceptive use. Likewise, findings from this investigation contrast with those of earlier studies suggesting that fertility is unaffected or reduced following HIV diagnosis.2,4,7,26 An important difference is that the present study examined intention to reproduce and revealed that serostatus and baseline intentions interact to influence postdiagnosis pregnancy status. Finally, these data depart from the literature in that women infected with HIV were found in multivariate analyses to be 3 times more likely to be pregnant at 6 months post-VCT than were uninfected women. This difference may be attributed to the relatively short interval between HIV serotesting and postintervention assessment; findings from studies conducting 12- to-24-month follow-up assessments suggest that this pattern of fertility may change with HIV disease progression.3,17

We found evidence of a significant interaction of HIV infection and intention to bear children with pregnancy status at follow-up for women, such that infected women planning a pregnancy had lower pregnancy rates than infected women not planning a pregnancy at baseline. One explanation for this finding relates to women’s self-perceived health status. Post hoc findings suggested that infected women with fewer HIV-associated symptoms altered their reproduction planning and reduced their fertility accordingly, perhaps because they felt they still had time to reproduce or because they wished to attend to their own health before proceeding with reproduction. By contrast, infected women reporting more HIV-associated symptoms may have felt greater pressure to reproduce while they were still capable of doing so. Unfortunately, the study data were not sensitive enough to permit a more thorough examination of changes between baseline and follow-up. Even so, the data shed new light on the impact of decisionmaking among HIV infected individuals in developing countries.

It is not altogether surprising to find that serostatus had no direct effect on pregnancy status for men, given that similar findings have been reported elsewhere.27–29 There is considerable evidence that, even in the face of the AIDS epidemic, men are highly resistant to adopting condom use,30 attending VCT interventions,17 or enacting strategies for reducing the risk of sexual transmission of HIV to female partners.31 Qualitative and quantitative research indicates that variables affecting the decision to continue to engage in unprotected sexual intercourse for both men and women include cultural beliefs that may conflict with HIV-preventive behavior,21 social norms that relegate condom use to sex with nonintimates,31 perceived deleterious effects of condom use on sexual health,18,32 and threats to masculinity conferred by condom use.33 Moreover, social and economic inequalities place decisionmaking, particularly that regarding sexual health, overwhelmingly in the domain of men, many of whom may not feel compelled to participate in VCT, alter their sexual risk behavior, or even disclose their serostatus to their primary partners. In an extreme example of the latter, Baingana et al.37 observed that 88% of Ugandan women whose husbands were receiving treatment for AIDS were unaware that their husbands were even infected with HIV. These and other findings underscore the observation that the success of HIV prevention programs in developing countries depends considerably on the involvement of male partners.32

One challenging ethical issue raised by these results pertains to reproductive counseling of HIV infected persons in developing countries. The Joint United Nations Programme on HIV/AIDS advocates early intervention with women of childbearing age known to be infected with HIV, with the goal of permitting them to make informed decisions about becoming pregnant or avoiding pregnancy.34 We agree with this position, adding that VCT intervention may serve as 1 of a few critical sources of emotional and resource support to infected women who elect to avoid pregnancy in the face of considerable social and economic pressure to bear offspring. In addition, counseling following HIV diagnosis may represent a unique opportunity to address alternatives to childbearing such as adoption, thereby limiting the risk of vertical transmission. Although the implications for men are less clear, univariate analysis indicated that men who were planning a partner pregnancy at baseline were 2.5 times more likely than those who were not to report a pregnant partner at follow-up, independent of their serostatus. Thus, we advocate that VCT interventions treat infected men’s intentions to bear children as important markers of future sexual risk that may result in both sexual and vertical transmission of HIV. Providing men with sufficient information and support to make informed decisions about future partner pregnancies, as well as facilitating prevention dialogues with partners, may assist in reducing the risk of transmission.

This study has several limitations. First, several potential confounding variables, such as sexually transmitted infections and partner serostatus, were not included in the statistical analyses. Although the latter may be taken to represent a significant omission, a considerable proportion of men and women in developing countries engage in high-risk sexual behaviors with partners known to be infected with HIV,35 lending support to the observation that modeling partner serostatus in HIV prevention research may be less important than previously believed.36 Several authors have noted the considerable social and economic imbalance favoring men’s preferences in sexual decisionmaking in many subSaharan countries.18,37,38 These sex inequities inevitably restrict women’s options with respect to HIV-preventive sexual behavior.

A second limitation of this investigation is that it did not include a sufficiently sensitive measure to demonstrate changes in contraceptive practices following HIV testing, which might have clarified the differential pregnancy rates observed among infected women. Similarly, a third limitation is that the measure of HIV-associated disease relied on participant self-report rather than biomedical indicators of disease. Although self-perceived health status appears to be important in reproduction planning, demonstrating that fecundity was not diminished secondary to HIV disease would have strengthened our findings. Finally, given the exploratory nature of this analysis, the findings are best interpreted as signposts for future inquiry.

In summary, evidence from this investigation indicates that addressing intention to bear children, and its interaction with HIV serostatus, may represent an important counseling issue to be included in VCT protocols. Powerful sociocultural and economic norms that give priority to childbirth, even in the face of high risk for vertical transmission, will likely continue to compete with efforts to reduce the spread of HIV in developing countries. Future VCT intervention research would do well to tailor protocols to appeal to men and increase their participation, incorporate family planning issues in VCT programming (including alternatives to childbearing), and enhance VCT messages to appeal to participants’ desire to ensure the well-being of their offspring.39 Ultimately, adults infected with HIV will require considerable support to make informed decisions around family planning.