Reproductive decisions structure adult experience and these decisions are especially complex for women with HIV. Such women are more likely than uninfected women to avoid pregnancy, through both contraception and sexual abstinence , yet more than 5% conceive annually .
For women with HIV considering pregnancy, highly active antiretroviral therapy (HAART) has provided hope that they may survive to see their children grow . The proportion of pregnant women on HAART has risen , while antiretroviral therapy (ART) has minimized vertical transmission , leading fewer women to elect abortion [6–8]. But studies of reproductive choice among women with HIV largely antedate HAART [9–16]. Women with HIV have been reported to experience adverse pregnancy outcomes more often, but these results may apply predominantly to injecting drug users, those in developing countries, or those not on ART .
To provide more contemporary insight, we reviewed pregnancies among HIV infected and similar uninfected women enrolled in the Women's Interagency HIV Study (WIHS), comparing pregnancy rates and outcomes and assessing the impact of HAART.
Epidemiologic and enrollment features of the WIHS, a multicenter prospective cohort study of the natural history of HIV infection and related health states among women with HIV and seronegative control women, have been described . The WIHS opened in 1994 at six sites. Written informed consent was obtained after local human subjects committees’ approval. HIV-infected and uninfected women were recruited from similar sources and frequency matched on demographics and key risk factors, including drug use and number of sexual partners. Recruitment targets were adjusted monthly when imbalances arose.
Trained interviewers obtained demographic and medical information at baseline and every 6 months. Women living with a partner were classed as married. Because many received prenatal care and delivery services outside study centers, pregnancy confirmation, outcomes, and complications could not be verified. Self-reported information about birth weight and gestational age were unreliable and were not recorded.
Pregnancy outcomes and dates were self-reported between 1 October 1994 and 31 March 2002. Pregnancies without a reported outcome (n = 10) were excluded. Outcomes were defined by the participant guided by a trained interviewer according to predetermined definitions: miscarriage for pregnancy lost before 20 weeks'/5 months’ gestation, stillbirth for a child born dead after that time, live birth for a baby born alive, abortion for a pregnancy terminated by an elective procedure, and ectopic for any tubal pregnancy. Pregnancy rates were determined from the number of women at risk for pregnancy.
Women were considered not at risk for pregnancy and so ineligible for analysis from the time they first reported hysterectomy, bilateral oophorectomy, or tubal sterilization, or reached 50 years of age. Women were also ineligible if they reported abstinence from vaginal intercourse for 12 months. Women reporting contraceptive use were included in calculations of pregnancy rates because pregnancies occurred in this group.
Recreational drug use, smoking, and alcohol use were determined from self-reported information at the visit prior to outcome. For HIV seropositive women, indices of disease severity (CD4 lymphocyte count, HIV RNA level, and history of AIDS) obtained at the prior visit were assessed as determinants of pregnancy outcome.
ART was summarized by whether participants reported no therapy, combination or monotherapy, or HAART. The definition of HAART was guided by the DHHS/Kaiser Panel as: (i) two or more nucleoside reverse transcriptase inhibitors (NRTI) in combination with at least one protease inhibitor (PI) or one NNRTI; (ii) one NRTI in combination with at least one protease inhibitor (PI) and at least one non-nucleoside reverse transcriptase inhibitors (NNRTI); (iii) a regimen containing ritonavir and saquinavir in combination with one NRTI and no NNRTI; and (iv) a regimen containing abacavir or tenofovir and consisting of three or more NRTI and no PI or NNRTI, but not combinations of zidovudine and stavudine with either a PI or NNRTI. Periods corresponding to treatment eras [19,20] were used to examine population effects of HAART: 1994–1996 (pre-HAART), 1997–1998 (early HAART), and 1999–2002 (later HAART). When analyzing baseline characteristics, baseline therapy was used. In other analyses, therapy use was as reported at the visit prior to a pregnancy outcome.
Pregnancy rates were the number of pregnancies annually according to time at risk. Each person contributed time from entry to last time at risk for pregnancy or last follow-up. Direct adjustment using the population age structure observed during 1996 was used to standardize yearly rates by age. Poisson regression was used to test whether rates differed by HIV serostatus or calendar time.
The frequency of pregnancy outcomes in HIV seropositive and seronegative women was compared by chi-square test. Logistic regression models assessed determinants of conception, miscarriage, and induced abortion. Characteristics significant at P < 0.10 in univariate analysis were placed in multivariate logistic regression models. Final parsimonious models, used to assess the effect of HAART on pregnancy outcome, incorporated other characteristics statistically (< 0.05) associated with outcome to account for confounding. All estimates and 95% confidence intervals (CI) were obtained using the SAS GENMOD procedure (SAS Institute, Cary, North Carolina, USA) with generalized estimating equations to adjust for the correlation of repeated observations.
Of 2059 HIV seroprevalent women, 1409 were at risk for pregnancy, with 1271 having at least two visits and contributing 4713 person-years at risk for conception. Of 569 enrolled seronegative women, 409 were at risk for pregnancy, with 377 contributing 1524 person-years.
A total of 232 HIV seropositive and 119 seronegative women reported 585 pregnancies (352 seropositive, 233 seronegative), for pregnancy rates of 7.4 and 15.2 per 100 person-years, respectively [odds ratio (OR), 0.53; 95% CI, 0.44–0.65]. Fig. 1 presents age-adjusted conception rates over time. Disparity persisted even after adjusting for age and population (OR, 0.57; 95% CI, 0.44–0.73), and after introduction of HAART. Except for a decline in pregnancy rates in 2000–2001 among HIV seropositive women, no significant changes in conception rates were observed. Grouping conception rates according to HAART era did not reveal significant trends.
Of 585 pregnancies, 409 (70.4%) occurred among women reporting contraceptive use at the visit before conception. Of these, 139 were in HIV seronegative women, while 270 were seropositive, representing 60% and 77% of pregnancies in these groups. Only 51 (9.4% of all pregnancies) occurred in women reporting hormonal contraception at the prior visit, 23 in HIV seronegative women and 28 in seropositive women.
Table 1 shows baseline demographic and behavioral characteristics for women who conceived versus those who did not. Women who conceived were younger; subsequent comparisons adjusted for age. Compared to seropositive women who did not conceive, those who did were more likely to be non-Hispanic black and unmarried, with fewer than four prior births and a baseline history of abortion or miscarriage.
Table 2 shows enrollment HIV characteristics for seropositive women. Compared to those not conceiving, women who conceived more commonly had heterosexual exposure as their only HIV transmission risk and had been taking combination ART at enrollment. Injecting drug users were less likely to conceive. Women who conceived had higher CD4 lymphocyte counts and lower HIV RNA levels.
Baseline determinants of conception that remained significantly associated with conception in multivariate analysis for HIV seropositive women included being younger (OR, 1.20; 95% CI, 1.16–1.23) or unmarried (OR, 1.59; 95% CI, 1.02–2.44), and having had a prior abortion (OR, 1.79; 95% CI, 1.25–2.63) or lower HIV RNA level (OR, 1.30; 95% CI, 1.10–1.54 for each log decrease). Baseline ART use was associated with lower likelihood of conception, but although the OR for HAART use was lower than for mono- or combination therapy (0.34 versus 0.69), this achieved significance only for women using non-HAART therapies (95% CI, 0.03–4.28 for HAART, 0.49–0.98 for mono- or combination therapy). For HIV seronegative women, significant correlates of conception in multivariate analysis included younger age (OR, 1.22; 95% CI, 1.17–1.28) and no prior abortion (OR, 1.80; 95% CI, 1.25–2.60) but not marital status. For both groups, women with fewer than four children were more likely to conceive.
The overall distribution of pregnancy outcomes did not differ significantly by HIV serostatus (Table 3). Specifically, HIV serostatus was not a significant correlate of miscarriage (OR, 1.33; 95% CI, 0.79–2.22 for HIV seropositive versus seronegative women). To assess the effects of HIV parameters and ART on outcomes, we restricted analysis to seropositive women. Among these, significant correlates of miscarriage from multivariate analysis included prior miscarriage (OR, 1.94; 95% CI, 1.01–3.72 compared to women with no prior miscarriage) and marijuana use persisting across two visits prior to pregnancy (OR, 6.64; 95% CI, 1.80–24.5 compared to women with no or inconsistent use). ART at the visit prior to pregnancy appeared protective against miscarriage (for monotherapy or non-HAART combinations OR, 0.37; 95% CI, 0.18–0.73; and for HAART OR, 0.37; 95% CI, 0.15–0.94). This negative correlation with miscarriage was not observed when therapy was analyzed by era.
Elective abortion was not significantly correlated with serostatus (for HIV seropositive OR, 0.86; 95% CI, 0.57–1.30, compared to seronegative women). In multivariate analysis, being unmarried was a strong predictor of abortion (OR, 6.93; 95% CI, 2.54–18.9 compared to married women). Women using ART were less likely to abort (OR, 0.38; 95% CI, 0.20–0.75 for monotherapy or non-HAART combinations; OR, 0.33; 95% CI, 0.17–0.67 for HAART, compared to women on no therapy). Annual income above $12 000 was associated with abortion, but only for women reporting prior abortion (OR, 7.00; 95% CI, 2.28–21.4 for women with incomes of $12 000–$24 000; OR, 9.07; 95% CI, 2.79–29.4 for women with incomes above $24 000 with a history of prior abortion). Abortion was less likely during the HAART era (OR, 0.68; 95% CI, 0.35–1.33 during the early HAART era; OR, 0.46; 95% CI, 0.23–0.90 during the later HAART era, compared with before HAART). In this same model, but not in the model accounting for individual antiretroviral use, having children increased the likelihood of abortion (OR, 3.44; 95% CI, 1.27–9.29).
HIV seropositive women were less likely to conceive than their seronegative counterparts. Nevertheless, 7% reported conception annually. Predictors of conception included being younger or unmarried and having a prior abortion or lower HIV RNA level. Women on ART at baseline were less likely and those with higher CD4 lymphocyte counts and lower HIV RNA levels were more likely to conceive. Ethnicity, study site, drug use, and correlates of socioeconomic status did not predict conception. Preconception and contraceptive counseling, while important for all HIV seropositive women, should be directed especially to younger, healthier single HIV-infected women with children.
The abortion rate among HIV-infected women in our study fell after HAART introduction and was indistinguishable from that in at-risk control women. This decline occurred while conception rates remained stable. This suggests that seropositive women who conceived were more likely to continue their pregnancies after HAART rather than that HAART led women to conceive who would not otherwise have done so.
Despite lower conception rates, specific rates of miscarriage, abortion, ectopic pregnancy, and stillbirth among seropositive women who conceived were not higher than those of their seronegative counterparts. We lacked consistent access to delivery records and so cannot determine whether prematurity or low birth weight may be increased among HIV infected women, and even our large study may lack statistical power to detect differences in uncommon outcomes. Nevertheless, our findings should reassure seropositive women considering pregnancy.
Miscarriage was associated only with prior miscarriage and marijuana use, while ART appeared to be protective. However the likelihood of miscarriage among women on HAART was similar to that among women on monotherapy or non-HAART combinations, as well as before and after the adoption of HAART in our cohort. Whether this is because women stopped HAART at conception, because the antiretroviral effect on miscarriage is independent of virologic control, or because of other factors remains to be determined.
These results are in contrast to prior findings. A meta-analysis found that women with HIV are at increased risk for spontaneous abortion, stillbirth, perinatal and infant mortality, intrauterine growth retardation and low birthweight, and preterm delivery, but studies reviewed largely antedated antiretroviral use and especially HAART . Any effect of maternal HIV infection in these studies may be dominated by the concomitant adverse effects of smoking and injecting drug use . Further study with more detailed data including medical records abstraction for confirmation of pregnancy outcome and complications is planned for WIHS.
This study has several limitations. Information about pregnancy outcomes and complications was limited. We were not able to control the counseling women received about pregnancy options. The median age of our cohort was relatively advanced; results may not be generalizable to younger HIV-infected women. We excluded sterilized participants, and we cannot assess the impact of HIV diagnosis or HAART use on decisions to undergo sterilization. We did not assess religiosity or family or social support other than marital status; strength in these areas may increase a woman's willingness to consider conception while minimizing the likelihood of termination.
Some 90% of pregnancies in women with HIV are unplanned , and reports from the WIHS and elsewhere indicate that contraception in these women is suboptimal [1,12]. In our study, 77% of pregnancies occurred in women who reported using contraception. Pregnancy rates were lower among women using hormonal contraception; barrier methods such as condoms may be inadequate. Further, 36% of conceptions in our study ended in abortion, a rate indistinguishable from that in at-risk HIV seronegative women and within the 3–47% range reported prior to the introduction of HAART [6,15]. Rather than being a response to HIV infection itself, a high abortion rate may reflect the social dislocation that places women at risk for HIV. This may be further reflected in two correlates of elective abortion: being unmarried, and not using ART.
Women with an annual income above $12 000 were more likely than more indigent women to elect abortion, but only if they reported a prior abortion. Access to abortion may depend on both financial means and having a known provider. Our finding that abortion was less common in the later HAART era suggests that women who conceive are increasingly willing to carry pregnancies to term. Clinicians providing care to women with HIV must provide not just counseling about vertical transmission and referral to quality prenatal care but also effective contraception and access to safe abortion.
1. Wilson TE, Massad LS, Riester KA, Barkan S, Richardson J, Young M, et al
. Sexual, contraceptive, and drug use behaviors of women with HIV and those at high risk for infection: results from the Women's Interagency HIV Study. AIDS
2. Chu SY, Hanson DL, Jones JL. Pregnancy rates among women infected with human immunodeficiency virus. Obstet Gynecol
3. Anastos K, Kalish LA, Hessol N, Weiser B, Melnick S, Burns D, et al
. The relative value of CD4 cell count and quantitative HIV-1 RNA in predicting survival in HIV-1 infected women: results of the Women's Interagency HIV Study. AIDS
4. Minkoff H, Ahdieh L, Watts H, Greenblatt RM, Schmidt J, Schneider M, et al
. The relationship of pregnancy to the use of highly active antiretroviral therapy. Am J Obstet Gynecol
5. Watts DH. Management of human immunodeficiency virus infection in pregnancy. New Engl J Med
6. Hankins C, Tran T, Lapointe N. Sexual behavior and pregnancy outcome in HIV-infected women. Canadian Women's HIV Study Group. J Acquire Immune Defic Syndr Hum Retrovirol
7. Bongain A, Berrebi A, Marin-Barjoan E, Dunais B, Thene M, Pradier C, Gillet JY. Changing trends in pregnancy outcome among HIV-infected women between 1985 and 1997 in 529 southern French university hospitals. Eur J Obstet Gynecol Reprod Biol
8. van Benthem BHB, de Vincenzi I, Delmas MC, Larsen C, van den Hoek A, Prins M. Pregnancies before and after HIV diagnosis in a European cohort of HIV-infected women. AIDS
9. Selwyn PA, Carter RJ, Schoenbaum EE, Robertson VJ, Slein RS, Rogers MF. Knowledge of HIV antibody status and decisions to continue or terminate pregnancy among intravenous drug users. JAMA
10. Johnstone FD, Brettle RP, MacCallum LR, Mok J, Peutherer JF, Burns S. Women's knowledge of their HIV antibody state: its effect on their decision whether to continue the pregnancy. BMJ
11. Sunderland A, Minkoff HL, Handte J, Moroso G, Landesman S. The impact of human immunodeficiency virus serostatus on reproductive decisions of women. Obstet Gynecol
12. Kline A, Strickler J, Kempf J. Factors associated with pregnancy and pregnancy resolution in HIV seropositive women. Soc Sci Med
13. Stephenson JM, Griffioen A. The effect of HIV diagnosis on reproductive experience. AIDS
14. Thackway SV, Furner V, Mijch A, Cooper DA, Holland D, Martinez P, et al
. Fertility and reproductive choice in women with HIV-1 infection. AIDS
15. Smits AK, Goergen CA, Delaney JA, Williamson C, Mundy LM, Fraser VJ. Contraceptive use and pregnancy decision making among women with HIV. AIDS Patient Care STDs
16. Bedimo AL, Bessinger R, Kissinger P. Reproductive choices among HIV-positive women. Soc Sci Med
17. Brocklehurst P, French R. The association between maternal HIV infection and perinatal outcome: a systematic review of the literature and meta-analysis. Br J Obstet Gynecol
18. Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al
. The Women's Interagency HIV Study. Epidemiology
19. Cook JA, Cohen MH, Grey D, Kirstein L, Burke J, Anastos K, et al
. Use of highly active antiretroviral therapy in a cohort of HIV-seropositive women. Am J Public Health
20. Gange SJ, Barron Y, Greenblatt RM, Anastos K, Minkoff H, Young M, et al
. Effectiveness of highly active antiretroviral therapy among HIV-1 infected women. J Epidemiol Community Health
21. Johnstone FD, Hamilton BA. The effect of human immunodeficiency virus infection and drug use on birth characteristics. Obstet Gynecol
22. Lindsay MK, Grant J, Peterson HB, Willis S, Nelson P, Klein L. The impact of knowledge of human immunodeficiency virus serostatus on contraceptive choice and repeat pregnancy. Obstet Gynecol