Women Living with Human Immunodeficiency Virus with Intended and Unintended Pregnancy: Characteristics and Patient Outcomes

Introduction

In the United States, ∼5,000 women of reproductive age were diagnosed with human immunodeficiency virus (HIV) ¹ in 2019, and an estimated 5,000 women living with HIV (WLWH) give birth each year. ² In the setting of more effective and well-tolerated antiretroviral therapy (ART) resulting in perinatal transmission rates of <1%, the proportion of women who desire pregnancy is similar in WLWH and those without HIV. ^{3 –5}

Unintended pregnancies are pregnancies that are mistimed, unplanned, or unwanted at the time of conception. ⁶ In 2011, 45% of all pregnancies in the United States were unintended. ⁷ There are fewer data on whether pregnancies in WLWH in the United States are intended or not intended. Rahangdale et al. reported 23% “unplanned,” 58% “ambivalent,” and 19% “planned” pregnancies among WLWH in 12 urban medical centers in the United States, ⁸ while Sutton et al. reported that 85.3% of WLWH reported a history of one or more unplanned pregnancies after HIV diagnosis. ⁹ The Canadian HIV Women's Sexual and Reproductive Health Cohort Study found that 61% of 492 pregnancies after HIV diagnosis were unintended. ¹⁰

Studies have demonstrated higher rates of adverse pregnancy outcomes and lower rates of involvement of the father of the baby (FOB) among women with unintended pregnancy (WWUP), ^{11 –14} as well as differences in frequency and type of contraception use both before and after pregnancy. ^{15 –18} Pregnancy intendedness in WLWH in particular is important, as it has been linked to improved maternal HIV outcomes. ^19,20

This study aimed to examine pregnancy intendedness, associated maternal variables, and maternal outcomes, including viral load suppression, substance use, partner disclosure, linkage to care, and contraception choice postpartum, in a population of pregnant WLWH followed in a dedicated HIV and Pregnancy clinic in a single large urban U.S. setting.

Materials and Methods

We conducted a retrospective analysis of WLWH receiving care through the Johns Hopkins HIV and Pregnancy program from 2003 to 2014. We identified 340 women who registered for prenatal care and underwent an initial nursing intake visit; only the first pregnancy during the time period was included. Eighty-five women (25.0%) had unknown pregnancy intendedness and were excluded in subsequent analyses comparing women with intended pregnancy (WWIP) and WWUP. Women with a new HIV diagnosis (N = 47) were excluded from baseline viral load and CD4 data, but included in all other analyses.

Thirteen (5.1%) women did not deliver at Johns Hopkins Hospital, 6 (2.6%) women had a spontaneous or missed abortion, 3 (1.1%) had a fetal death, 2 (0.7%) had an elective abortion, and 11 (4.3%) women were lost to follow-up before delivery. The preceding 35 women were excluded from analyses pertaining to maternal or neonatal outcomes, including delivery HIV laboratories, but were included in all other analyses.

Data were abstracted from the maternal electronic medical record (EMR) into a database: the nursing intake form is integrated into the EMR. Intendedness was assessed first as a “yes/no” question within the standardized nursing intake form; if this field was not completed, a text search within the EMR was performed for the terms “planned” and “intended.” Intendedness was classified as “unknown” if neither of these strategies identified pregnancy intendedness.

Our retrospective method of data collection did not allow us to determine why pregnancy intendedness was not documented in these patients. This cohort was similar demographically and in their medical history to the studied cohort of women with known pregnancy status, although they had a higher gravity and were more likely to report a history of illicit substance use (data not shown).

The database contained 171 variables in addition to intendedness, including demographics, HIV variables (including risk behavior, CD4 count and percentage nondetectable viral load [NDVL] at pregnancy entry and delivery, and disclosure patterns), pregnancy outcomes, and selected psychosocial variables. Demographics, baseline characteristics, and psychosocial variables (including being “pleased to be pregnant”) were abstracted from the nursing intake, with the exception of HIV disclosure status and HIV status of the father of the baby, which was documented in the EMR during provider and social work encounters during the pregnancy.

HIV laboratory data were abstracted from the EMR: we defined NDVL as <50 copies/mL through 2012, and <20 copies/mL after 2012 per assay thresholds during those times. Missing demographic and psychosocial data, either due to failure of provider questioning, patient nonresponse, or lack of documentation, were excluded from analysis. Substance use during pregnancy was abstracted from urine toxicology reports when feasible (cannabinoids, opiates, and cocaine) or from patient self-report (tobacco, alcohol). Birth outcomes, including infant HIV status, were abstracted from the maternal chart alone, given IRB approval for maternal chart access only.

Contraception was categorized as “none” (including no plan for contraception or natural family planning), “condoms,” “prescription” (including hormonal pills, patch, and ring), “DMPA” (medroxyprogesterone acetate injection), “LARC” (long-acting reversible contraception, including hormonal and nonhormonal intrauterine device and hormonal implant), and “sterilization” (including tubal ligation and hysterectomy): if patients reported their intent to use multiple methods, the choice was categorized as more effective method.

Demographic data were described using frequencies, means, ranges, and standard deviations. One-way analysis of variance (ANOVA) testing was used to compare differences in intendedness across time. Univariate analysis was performed using chi-square tests for categorical data and two-tailed Student's t-test for continuous data. If intendedness was found to be a significant predictor of an outcome in univariate analysis, we used linear (CD4 count at pregnancy entry and delivery) and logistic (NDVL at pregnancy entry and delivery, tobacco use during pregnancy, and attendance of postpartum visit) forward stepwise regression models to determine whether pregnancy intendedness remained an important predictor for each outcome.

Prediction models included baseline covariates that were associated with each outcome in unadjusted models, and were limited to continuous outcomes and those binary outcomes with sufficient sample sizes (number of observed outcomes) to support these procedures. Analyses were conducted using IBM SPSS Statistics version 28.0.1.1 (USA). Two-tailed p-value <0.05 was considered significant. The study was approved by the Johns Hopkins School of Medicine IRB.

Results

Discussion

This study examined variables associated with intended versus unintended pregnancies in an urban cohort of pregnant women living with HIV followed at a single academic medical center. This study confirms prior data from the United States documenting high levels of unintended pregnancy both in WLWH and in the general population: less than one-third of women in our study reported intending to get pregnant. However, approximately three-quarters of women in this study said that they were pleased to be pregnant. Pregnancy intendedness was a positive predictor of NDVL at pregnancy entry.

Studies on predictors of pregnancy intendedness in WLWH in high-resource countries have been relatively few but have demonstrated several similar findings. Recent U.S. and Canadian studies observed that WLWH with intended pregnancies were more likely to be married, to be older, to be an immigrant, and to be employed. ^10,20,21 Similar patterns are also seen in population-wide studies of pregnancy intendedness in U.S. women: rates of unintended pregnancy were higher in women who were younger, unmarried, had less educational attainment, and Black or Hispanic. ⁷

Unintended pregnancy was associated with a higher frequency of tobacco, alcohol, opiate, and cocaine use during pregnancy. The association of unintended pregnancy with substance use disorder has been shown in studies in the general population ^{21 –23} and in WLWH. ²⁰ Both drug and alcohol use are associated with adverse pregnancy outcomes independent of HIV status. ²⁴ Ideally, substance use should be identified before pregnancy to provide prepregnancy evaluation and counseling, as well as treatment during pregnancy. Failure to screen for unintended pregnancy may result in missed opportunities to provide more robust resources.

Pregnancy intendedness was significantly associated with HIV disclosure to the FOB, a novel observation. While our patient population overall had higher disclosure rates than those reported in previous studies, ^25,26 our data suggest that WWUP may benefit from increased support surrounding disclosure counseling, especially during pregnancy. This is further supported by the findings of Brittain et al. who noted that disclosure was associated with higher depression scores in South African women with unintended but not intended pregnancy. ²⁷ Appropriate support during disclosure is particularly vital as nondisclosure of HIV status during pregnancy may be associated with less prenatal care, ART nonadherence, and lower likelihood of postpartum visit attendance. ^24,34

WWIP in our cohort were more likely to have NDVL throughout pregnancy, similar to prior reports. In the United States, Zahedi-Spung et al. observed that women with planned pregnancy were more likely to report daily ART use during pregnancy, ¹⁹ and Dude et al. found that WWIP were more likely to have viral suppression at delivery. ²⁰ Brittain et al. reported persistently elevated viral loads in women reporting unplanned pregnancy in prospective study of 459 WLWH living in South Africa. ²⁸ While a lower percentage of WWUP in our cohort had NDVL at delivery compared with the more recent cohort studied in Dude et al, our findings were comparable with those from a similar time period. ²⁹

As pregnancy intendedness was an important predictor of having NDVL at pregnancy entry, WWIP likely benefit from optimization of ART regimen before pregnancy if their HIV status was known. The integration of HIV care into obstetrical care, peer support groups, and phone calls/text message reminders has been used within our program, and has been a promising intervention to increase patient retention rates in other studies. ^30,31 Control of viral replication throughout pregnancy is essential in preventing mother-to-child transmission and optimizing maternal health, ^32,33 and WWUP may require more intensive intervention to close the initial gaps observed at pregnancy entry.

In contrast to the findings of Dude et al., pregnancy intendedness was not an important predictor of NDVL at delivery in our multivariate prediction model. However, it was associated with this outcome, which may be confounded by social risk factors that impact viral control: multisite studies would be helpful in better characterizing the adjusted association between pregnancy intendedness and viral suppression.

All women have the potential to benefit from prepregnancy counseling to be in optimal health before conception and to time pregnancies appropriately. While we did not assess contraception use before the pregnancies studied, we did note differences in contraception uptake and type between WWUP and WWIP in the postpartum period. In the general U.S. population, Guzzo et al. observed that WWUP were more likely to use highly effective methods such as LARC and sterilization, ¹⁸ and concluded that the risk of repeat unintended pregnancy may be mediated by differences in long-term contraception use or continuation rather than initial postpartum contraception choices.

We similarly found that WWUP were more likely to undergo sterilization at the time of discharge, and that WWIP were more likely to receive condoms. It is important to note that the majority of all women, including WWIP and WWUP, reported being pleased to be pregnant, consistent with prior data showing that the majority of WLWH report a desire to become pregnant in the future. Therefore, while WWUP received highly effective nonreversible contraception, assessing for current and future pregnancy intendedness is essential to determine whether highly effective reversible methods, such as LARC, would be more appropriate. ³⁴

Our study benefits from a large sample size with documentation of pregnancy intendedness. All women were followed in a single academic center with a small group of experienced HIV obstetrical providers. The use of a standardized prenatal intake assessment allowed for minimal reporter bias and for consistency.

We are the first to examine HIV disclosure at the time of pregnancy and delivery, and differences in postpartum contraception between intended and unintended pregnancy in WLWH. Although the rarity of disclosure and contraception outcomes within this cohort did not support multivariate analysis, our findings provide strong preliminary data and guidance for further research with larger cohort studies.

However, our study has several limitations. We used a simple “yes/no” question obtained during nursing intake to document intendedness, although more nuanced evaluations of pregnancy intendedness exist. ³⁵ WLWH may experience stigma during pregnancy, which may lead to decreased report of pregnancy intendedness due to social desirability bias. Our dataset spans from 2003 to 2014, including a period of time before the use of integrase inhibitors and universal ART use, and may not reflect more recent patient outcomes in our clinic. Our EMR access was limited to the maternal chart, and our analysis of birth outcomes was consequently limited. Finally, our findings at an urban clinic serving primarily low-income women may not be generalizable to demographically dissimilar populations.

Conclusions

This study adds to the limited literature from the United States and higher resource countries examining pregnancy intendedness in WLWH, and, to our knowledge, is the first study to demonstrate gaps in HIV disclosure during pregnancy between WWIP and WWUP. Our results highlight the need for assessment of women's desired pregnancy timing throughout HIV care to allow for appropriate management of substance use disorders, support in HIV status disclosure, attention to ART adherence, and effective contraception provision for those who do wish to conceive. For women who become pregnant, identification of those pregnancies that were not intended should trigger more intensive resources and support throughout pregnancy and in the postpartum period to optimize maternal and infant outcomes.