Secondary Logo

Journal Logo

Clinical Science

Repeat Pregnancies Among US Women Living With HIV in the SMARTT Study: Temporal Changes in HIV Disease Status and Predictors of Preterm Birth

O'Brien, Brigid E. DO, MSa; Williams, Paige L. PhDb,c; Huo, Yanling MSc; Kacanek, Deborah ScDc; Chadwick, Ellen G. MDd; Powis, Kathleen M. MD, MPHe,f; Correia, Katharine PhDg; Haddad, Lisa B. MD, MS, MPHh; Yee, Lynn M. MD, MPHi; Chakhtoura, Nahida MDj; Dola, Chi MD, MPHk; Van Dyke, Russell B. MDa; for the Pediatric HIV/AIDS Cohort Study (PHACS)

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: November 01, 2020 - Volume 85 - Issue 3 - p 346-354
doi: 10.1097/QAI.0000000000002445



With advances in HIV treatment, the birth rate among women living with HIV (WLHIV) is increasing,1 with many experiencing multiple pregnancies.2 Factors associated with repeat pregnancies among WLHIV include younger age, lower educational attainment, receipt of public assistance, lower viral loads (VLs), higher CD4 counts, and stillbirth or abortion in the previous pregnancy.3,4 Several studies have shown improving HIV disease status over repeat pregnancies. For example, in a multicountry study of WLHIV in Latin America and the Caribbean, the proportion of women with a VL <400 copies/mL increased from 31% at their index pregnancy to 46% at the subsequent pregnancy.4 Likewise, Floridia et al5 showed that the CD4 count early in pregnancy and viral suppression at delivery improved in repeat pregnancies and was associated with greater antiretroviral treatment (ART) coverage. However, specific differences in antiretroviral (ARV) regimens in repeat pregnancies and associations with birth outcomes were not examined.

Protease inhibitor (PI) use in pregnancy is associated with an elevated risk of preterm birth, the leading cause of neonatal morbidity and mortality in the United States.6–9 In the Surveillance Monitoring for ART Toxicities (SMARTT) study, we observed a high rate of preterm birth (19%), which was associated with PI-based combination therapy in the first trimester.6 Prescribing practices for pregnant WLHIV have changed over the course of time,10 and little is known about risks associated with integrase strand transfer inhibitors (INSTIs) or newer PIs or how ART switches between pregnancies affect the risk of preterm birth.11

Our objectives were (1) to describe serial changes in CD4 counts and VLs in repeat pregnancies among WLHIV; (2) to examine associations of maternal characteristics with trends in CD4 counts and VLs over repeat pregnancies; and (3) to examine associations of PI and INSTI use in pregnancy with preterm birth.


Study Population and Inclusion Criteria

We analyzed pregnant WLHIV and their uninfected infants enrolled in the SMARTT study of the Pediatric HIV/AIDS Cohort Study, a mixed retrospective and prospective study designed to identify adverse effects of in utero ARV exposures in infants and children.6,12 For the dynamic cohort, women and their infants were enrolled between 22 weeks of gestation and 1 week after birth.6 For the static cohort, pregnant women were generally enrolled in another study during their pregnancy (the Women and Infant Transmission Study or the International Maternal Pediatric Adolescent AIDS Clinical Trials Protocol 1025 study) and were later enrolled in the SMARTT cohort. Biological mothers living with HIV with 2 or more SMARTT-enrolled live-birth deliveries through 2018 were eligible for this analysis. All women provided written informed consent for study participation. Women were included in analyses examining CD4 counts and VLs over repeat pregnancies if they had a CD4 count or VL measure in any study pregnancy. Analyses of associations between PI or INSTI use in pregnancy and preterm birth were conducted among the subset of women with repeat singleton pregnancies and documented ARV use in pregnancy and infant gestational age at birth. The Institutional Review Board at each study site and Harvard T.H. Chan School of Public Health approved the protocol.

Outcome Measures

CD4 and VL measures were obtained through medical chart abstraction. Earliest and latest CD4 counts and VL measures in a pregnancy were selected from the interval starting 14 days before conception and ending at 7 days after delivery. VL suppression was defined as ≤400 copies/mL to accommodate this higher limit of detection available in earlier years.

Gestational age was determined according to the standard obstetric practice.13 Preterm birth was defined as birth occurring before 37 completed weeks of gestation.9 Medically indicated and spontaneous preterm births were not distinguished, given the importance of all preterm births, the possibility of indication shifting, and the common occurrence of multiple etiologies.

ARV Exposure Measures and Other Covariates

PI and INSTI exposures were separately categorized as use at any time during pregnancy (vs. nonuse) and by timing of initiation (at conception; first trimester defined as 0–13 and 6/7 weeks; second trimester defined as 14–27 and 6/7 weeks; and third trimester defined as 28 weeks–3 days before delivery vs. no use during pregnancy). Within-woman changes in PI and INSTI use between consecutive pregnancies were examined. For instance, PI use in the index and the subsequent pregnancy could be classified as follows: (1) a PI-containing regimen in both pregnancies; (2) a non–PI-containing regimen in both pregnancies; or (3) discordant regimens in the 2 pregnancies. INSTI use was classified similarly.

Covariates of interest for evaluating associations with CD4 counts and VLs over repeat pregnancies included the sequence number of the pregnancy while enrolled in SMARTT (eg, first, second, third, etc.), birth cohort, and maternal sociodemographic and health information, including sexually transmitted infections (STIs), substance use during pregnancy (tobacco, alcohol, marijuana, or illicit drugs), mode of HIV acquisition (perinatal vs nonperinatal), and timing of ARV initiation. The birth cohorts were chosen to reflect changes in treatment guidelines in 2011. In evaluating associations of PI/INSTI exposures with preterm birth in repeat singleton pregnancies, maternal sociodemographic information, STIs, and substance use were considered as potential confounders of interest.

Statistical Analysis

Sociodemographic characteristics of women and median CD4 count, log10 VL, and viral suppression status for their earliest and latest measures in pregnancy were summarized for up to the first 3 SMARTT-enrolled pregnancies. Continuous log10 VL measures were analyzed descriptively. We evaluated within-woman changes in CD4 counts and log10 VL between the first 2 consecutive SMARTT pregnancies with available CD4 and VL measures in both pregnancies (defined as the index and the subsequent pregnancy, which may not correspond to the participant's first 2 SMARTT-enrolled pregnancies). Wilcoxon signed-rank tests were used to test for differences in CD4 counts and log10 VL between the index and subsequent pregnancy; these comparisons included the earliest measures in the index and subsequent pregnancy, the latest measures in the index and subsequent pregnancy, and the latest measure in the index pregnancy and the earliest measure in the subsequent pregnancy.

Generalized estimating equation (GEE) linear regression models, accounting for within-woman correlation in repeat measures through an exchangeable correlation assumption, were used to evaluate associations of maternal characteristics with earliest and latest CD4 measures across all pregnancies enrolled. A stepwise approach was used to build multivariable models using a P value of 0.15 as the criterion to select covariates noted above in the final models. Pregnancy number, birth cohort, and maternal age at delivery were retained in the final model as an a priori decision regardless of their statistical significance. We used GEE logistic regression models for the VL suppression outcome (VL ≤400 vs. > 400 copies/mL), adjusting for covariates using the same approach described above.

To evaluate associations of PI and INSTI use with preterm birth, we first described preterm birth frequency and ARV exposures (PI and INSTI) over the first 3 SMARTT-enrolled singleton pregnancies. An exact conditional logistic regression model was used to examine the association of ARV exposure with preterm birth among women receiving discordant ARV regimens (PI-containing vs. non–PI-containing regimen and INSTI-containing vs. non–INSTI-containing regimen), restricted to the index and subsequent pregnancy. GEE logistic regression models were used to evaluate associations of PI and INSTI exposure with preterm birth across all study pregnancies, adjusting for potential confounders that were selected using the stepwise approach described above. All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).


Study Population

A total of 2859 biological mothers were enrolled in SMARTT with their children (n = 3913) as of July 1, 2018. Among these, 736 women were eligible for our analysis based on having 2 or more live-birth SMARTT-enrolled pregnancies with CD4 and/or VL measured in any pregnancy. A total of 717 women who had repeat singleton pregnancies (with 1647 pregnancies) and had available information on ARV exposure and preterm birth were included in the preterm birth analysis (see Figure 1S, Supplemental Digital Content,

Of the 736 women with repeat pregnancies, 76% had 2 pregnancies, 19% 3 pregnancies, and 5% 4 or more pregnancies, with a maximum of 7 SMARTT-enrolled pregnancies. Maternal characteristics during the first 3 pregnancies are shown in Table 1. Women with a third pregnancy were more often Black, non-Hispanic, and living with a partner or spouse. Nine percent of the women had acquired HIV perinatally. Only 26% were receiving ARVs at conception of their first pregnancy compared with 47% and 50% for their second and third pregnancies, respectively. The median interpregnancy interval decreased slightly from 26.8 months between the first and second pregnancy to 22.2 months between the second and third pregnancy.

TABLE 1. - Maternal Demographic and Health Characteristics for the First 3 Study Pregnancies Among 736 Women With Repeat Pregnancies
Characteristic Study Pregnancy
First Pregnancy (N = 736) Second Pregnancy (N = 736) Third Pregnancy (N = 177)
Black race 482 (65) 482 (65) 134 (76)
Hispanic 221 (30) 221 (30) 44 (25)
Age at delivery, yrs 24.7 (21.4–29.2) 28.4 (25.0–33.0) 30.2 (26.9–34.1)
Interpregnancy interval, mo* 26.8 (13.5–48.6) 22.2 (9.9–50.3)
Year of delivery
 <2004 221 (30) 89 (12) 10 (6)
 2004–2010 339 (46) 282 (38) 62 (35)
 2011–2018 176 (24) 365 (50) 105 (59)
Annual household income ≤$20K 490 (67) 467 (63) 114 (64)
Educational attainment: <high school 251 (34) 235 (32) 67 (38)
Living with partner/spouse 374 (51) 414 (56) 111 (63)
Site region
 Puerto Rico 63 (9) 63 (9) 7 (4)
 West 128 (17) 128 (17) 33 (19)
 South 274 (37) 274 (37) 64 (36)
 Midwest 85 (12) 85 (12) 15 (8)
 Northeast 186 (25) 186 (25) 58 (33)
Substance use during pregnancy
 Tobacco 123 (17) 112 (15) 26 (15)
 Alcohol 46 (6) 53 (7) 12 (7)
 Marijuana 51 (7) 52 (7) 12 (7)
 Illicit drugs 61 (8) 62 (8) 15 (8)
Any STIs in pregnancy 305 (41) 266 (36) 69 (39)
Perinatal HIV acquisition 64 (9) 64 (9) 14 (8)
Timing of ARV initiation in pregnancy
 No ARV at any time 17 (2) 10 (1) 5 (3)
 On ARV at conception 193 (26) 343 (47) 88 (50)
 First trimester 115 (16) 143 (19) 39 (22)
 Second trimester 308 (42) 188 (26) 38 (21)
 Third trimester 62 (8) 23 (3) 3 (2)
 Unknown 41 (6) 29 (4) 4 (2)
Statistics shown are median (IQR) or count (%). Illicit drugs included marijuana/hashish, heroin, ecstasy/MDMA, cocaine/crack, methamphetamine, opium, or LSD.
Some measures were not available for the first, second, and third pregnancy, including race (50, 50, and 5), ethnicity (2, 2, and 0), household income (74, 50, and 11), education (18, 10, and 2), living arrangement (19, 12, and 2), substance use during pregnancy (37, 30, and 7), STIs (35, 21, and 6), and mode of HIV acquisition (18, 18, and 3).
*Interpregnancy interval is defined as the number of months between the delivery of one pregnancy and estimated date of conception for the subsequent pregnancy. The interpregnancy interval was only calculated on pregnancies with available gestational age.
IQR, interquartile range; LSD, Lysergic acid diethylamide; MDMA, 3,4-methylenedioxymethamphetamine; STI, sexually transmitted infection (trichomonas, gonorrhea, syphilis, chlamydia, human papillomavirus, genital warts, genital herpes, and bacterial vaginosis).

Changes in CD4 Counts Over Repeat Pregnancies

The earliest CD4 counts across the first 3 SMARTT pregnancies were relatively stable, ranging from a median of 453–509 cells/mm3 (Fig. 1A). However, there was a small increase in the median earliest CD4 measurement (median gestational age of 9–13 weeks) to the latest CD4 measurement (median gestational age of 34–35 weeks) for all pregnancies, which was most pronounced in the first pregnancy (Fig. 1A). When comparing 2 consecutive pregnancies with available CD4 data, there was a significant within-woman increase in the earliest CD4 count from the index pregnancy to the earliest CD4 count in the subsequent pregnancy {mean difference: 62.9 cells/mm3 [95% confidence interval (CI): 31.3 to 94.6]}. There were no significant within-woman differences in the latest CD4 counts between the index and subsequent pregnancy [mean difference: 22.9 cells/mm3 (95% CI: −7.4 to 53.2)] nor between the latest CD4 count in the index pregnancy and the earliest CD4 count in the subsequent pregnancy [mean difference: 16.2 cells/mm3 (95% CI: −13.8 to 46.1)].

Trends in (A) CD4 counts and (B) suppressed viral load over the first 3 SMARTT pregnancies.

Considering all pregnancies, factors independently associated with a lower CD4 count both early and late in pregnancy were birth before 2011 (vs. 2011 or later), annual household income between $20,000 and $30,000 (vs. ≤$20,000), and perinatally acquired HIV. Receiving ARVs at conception (vs. initiating ARVs in the second/third trimester) was associated with a higher earliest and latest CD4 count in pregnancy. In addition, living with a partner or spouse was associated with a higher latest CD4 count in pregnancy (Table 2).

TABLE 2. - Adjusted Associations of Covariates With CD4 Count for All Pregnancies
Covariates Earliest CD4 Count (Cells/mm3) in Pregnancy* Latest CD4 Count (Cells/mm3) in Pregnancy
Adjusted Mean Difference (95% CI) P§ Overall P Adjusted Mean Difference (95% CI) P§ Overall P
No. of pregnancy 7.17 (−23.7 to 38.1) 0.65 7.13 (−32.5 to 46.8) 0.72
Birth cohort <0.001 0.01
 2011–2018 REF REF
 <2004 −78.7 (−142 to −14.9) 0.02 −50.7 (−116 to 15.2) 0.13
 2004–2010 −92.7 (−137 to −48.1) <0.001 −73.9 (−121 to −26.7) 0.002
Age (yrs) at delivery 0.98 0.88
 ≥35 REF REF
 <25 −11.3 (−99.4 to 76.8) 0.80 15.9 (−69.6 to 101) 0.72
 25–<30 −14.1 (−98.3 to 70.0) 0.74 9.76 (−72.9 to 92.4) 0.82
 30–<35 −18.0 (−104 to 68.2) 0.68 31.9 (−65.5 to 129) 0.52
Annual household income 0.003 <0.001
 ≤$20K REF REF
 >$20–30K −66.3 (−107 to −24.9) 0.002 −68.6 (−107 to −29.8) <0.001
 >$30K 23.2 (−35.9 to 82.3) 0.44 22.1 (−40.7 to 84.8) 0.49
 Unknown 21.0 (−35.2 to 77.2) 0.46 39.1 (−17.5 to 95.7) 0.18
Perinatal HIV acquisition −141 (−228 to −55.3) 0.001 −127 (−214 to −39.7) 0.004
Living with partner/spouse 38.47 (1.11 to 75.82) 0.04
Any STIs in pregnancy −30.3 (−63.1 to 2.54) 0.07
Timing of ARV initiation <0.001 <0.001
 Second/third trimester REF REF
 No ARV −7.36 (−160 to 146) 0.92 −112 (−254 to 29.7) 0.12
 On ARV at conception 113 (71.2 to 155) <0.001 42.7 (3.76 to 81.7) 0.03
 First trimester −6.34 (−45.9 to 33.2) 0.75 −31.1 (−69.0 to 6.76) 0.11
A GEE model was fit separately for the earliest and latest CD4 count measured in repeat pregnancies, adjusted for covariates meeting model selection criteria. The adjusted mean difference represents the change in the CD4 count per additional pregnancy for sequence number of the pregnancy or comparing women with a specific characteristic to those without (or in reference level) for other covariates.
*Earliest CD4 counts in pregnancy or up to 14 days before conception or 7 days after delivery if there was none in pregnancy.
Latest CD4 counts in pregnancy or up to 7 days after delivery or 14 days before conception if there was none in pregnancy.
Adj. Diff.: estimated difference in adjusted mean CD4 counts as per additional pregnancy for the sequence number of the pregnancy or comparing women with vs. without a specific characteristic for other covariates.
§P value for testing the effect of each level of a covariate as compared to the reference level.
P value for testing the overall effect of a covariate.

Changes in VL Suppression Over Repeat Pregnancies

In the first pregnancy, only 34% of women had viral suppression at their earliest measurement (median gestational age = 13 weeks), with 81% achieving viral suppression at their latest measure in that pregnancy (median gestational age = 36 weeks) (Fig. 1B). At the beginning of their second pregnancy, 49% had viral suppression, with 82% achieving viral suppression at the latest measure in that pregnancy. The proportion with suppressed VLs was lower at the beginning of the second and third pregnancies than at the end of the previous pregnancy. At the end of the third pregnancy, 83% achieved viral suppression (Fig. 1).

There was a significant within-woman decrease in both the earliest and latest log10 VL measures between the index and subsequent pregnancy [mean difference: −0.38 log copies/mL (95% CI: −0.48 to −0.28) and −0.17 log copies/mL (95% CI: −0.23 to −0.10), respectively]. However, from late in the index pregnancy to early in the subsequent pregnancy there was a significant increase in the log10 VL [mean difference: 0.63 log copies/mL (95% CI: 0.53 to 0.72)].

Considering all pregnancies, factors independently associated with higher odds of viral suppression at the earliest measure in pregnancy were delivery year 2011 or later, receiving care at a site in the Northeast region (vs. Puerto Rico), more than a high school education, living with a partner or spouse, nonperinatally acquired HIV, not having an STI during pregnancy, and receiving ARVs at conception (vs. initiating ARV in the second/third trimester) (Table 3). Likewise, factors independently associated with higher odds of viral suppression at latest measure in pregnancy were a lower sequence number of repeat pregnancies, delivery year 2011 or later (vs. <2004), older age at delivery (≥35 years vs. <25 or 25–30 years), non-Black race, receiving care at a site in the Western US region (vs. Northeast), at least a high school education, household income >$30,000 (vs. ≤$20,000), nonperinatally acquired HIV, and receiving ARVs at conception (vs. initiating ARV in the second/third trimester) (Table 3). Women at research sites in Puerto Rico had decreased odds of viral suppression compared with those in the Northeast region.

TABLE 3. - Adjusted Associations of Covariates With a Suppressed VL for All Pregnancies
Covariates Earliest VL in Pregnancy Latest VL in Pregnancy
Adjusted OR (95% CI) P* Overall P* Adjusted OR (95% CI) P* Overall P*
No. of pregnancy 1.07 (0.91 to 1.25) 0.44 0.80 (0.65 to 0.98) 0.03
Birth cohort <0.001 <0.001
 2011–2018 REF REF
 <2004 0.40 (0.26 to 0.61) <0.001 0.26 (0.16 to 0.41) <0.001
 2004–2010 0.52 (0.40 to 0.69) <0.001 0.77 (0.53 to 1.13) 0.18
Age (yrs) at delivery 0.03 0.01
 ≥35 REF REF
 <25 0.67 (0.42 to 1.05) 0.08 0.38 (0.19 to 0.74) 0.005
 25–<30 0.87 (0.59 to 1.29) 0.49 0.36 (0.19 to 0.68) 0.002
 30–<35 1.18 (0.81 to 1.74) 0.39 0.60 (0.31 to 1.16) 0.13
Black race 0.62 (0.40 to 0.95) 0.03
Site region 0.11 <0.001
 Northeast REF REF
 Puerto Rico 0.56 (0.32 to 0.98) 0.04 0.30 (0.15 to 0.58) <0.001
 West 1.13 (0.73 to 1.74) 0.60 3.83 (1.91 to 7.66) <0.001
 South 0.77 (0.53 to 1.12) 0.17 0.93 (0.62 to 1.40) 0.72
 Midwest 0.79 (0.48 to 1.30) 0.35 1.04 (0.57 to 1.90) 0.90
Less than high school education 0.67 (0.50 to 0.90) 0.01 0.68 (0.49 to 0.96) 0.03
Annual household income 0.13 0.001
 ≤$20K REF REF
 >$20–30K 0.71 (0.49 to 1.03) 0.07 0.85 (0.56 to 1.28) 0.43
 >$30K 0.77 (0.52 to 1.14) 0.19 1.91 (1.07 to 3.40) 0.03
 Unknown 0.68 (0.40 to 1.15) 0.15 2.78 (1.42 to 5.44) 0.003
Living with partner/spouse 1.32 (1.02 to 1.71) 0.04
Perinatal HIV acquisition 0.57 (0.34 to 0.94) 0.03 0.32 (0.19 to 0.53) <0.001
Any STIs in pregnancy 0.67 (0.52 to 0.86) 0.002
Timing of ARV initiation <0.001 <0.001
 Second/third trimester REF REF
 No ARV 3.69 (1.30 to 10.48) 0.01 0.28 (0.11 to 0.74) 0.01
 On ARV at conception 7.10 (5.34 to 9.45) <0.001 1.63 (1.15 to 2.32) 0.01
 First trimester 1.79 (1.31 to 2.43) <0.001 1.30 (0.86 to 1.98) 0.22
A GEE model was fit separately for the earliest and latest HIV VL in repeat pregnancies (with suppression defined as ≤400 copies/mL), adjusted for covariates meeting model selection criteria. The adjusted OR represents the odds ratio per additional pregnancy for sequence number of the pregnancy or comparing women with a specific characteristic to those without (or in reference level) for other covariates.
*P value for testing the effect of each level of a covariate as compared to the reference level or across all levels (overall) for categorical covariates.

ARV Regimens and Preterm Birth Over Repeat Singleton Pregnancies

ARV regimens in women changed over the course of time, with an increase in the proportion of women receiving INSTI-containing regimens from their first (3%) to their second (13%) and third (17%) study pregnancies. The most common ARVs included lopinavir/ritonavir, atazanavir, darunavir, nelfinavir, raltegravir, rilpivirine, and nevirapine (see Table 1S, Supplemental Digital Content, For the first 3 singleton study pregnancies, the frequency of preterm birth was similar for the first 2 pregnancies (14% and 15%) but increased to 21% in the third pregnancy. The frequency of low birth weight infants (defined as <2500 g) was 15%, 12%, and 14% for the first 3 singleton pregnancies (see Table 2S, Supplemental Digital Content,

In considering only the first 2 consecutive study pregnancies, 55% of women were on a PI-containing ART regimen in both pregnancies (see Table 3S, Supplemental Digital Content, Comparing these 2 pregnancies, the frequency of preterm birth increased by 3.9% among women remaining on a PI-containing regimen while it decreased by 2.7% and 1.0%, respectively, among women remaining on a non–PI-containing regimen or changing to a non–PI-containing regimen. Twenty-four women received an INSTI in their index pregnancy (12 in the first trimester) and 90 in their subsequent pregnancy (61 in the first trimester) (see Table 3S, Supplemental Digital Content,

After adjusting for the sequence number of all singleton births, birth cohort, household income, and living with a partner, overall PI use during pregnancy was not significantly associated with the risk of preterm birth. However, PI initiation during the first trimester was associated with increased odds of preterm birth compared with women who did not use PIs at any time during pregnancy [odds ratio (OR): 1.97 (95% CI: 1.27 to 3.07), Fig. 2]. No such association was observed for receiving a PI regimen at conception or initiating one in the second or third trimester. Similarly, initiating an INSTI-containing regimen in the first trimester was associated with a significant increase in the odds of preterm birth [OR: 2.39 (95% CI: 1.04 to 5.46), Fig. 2] compared with not receiving an INSTI-containing regimen during pregnancy.

Adjusted associations of ARV exposures in a singleton pregnancy with preterm birth among all study pregnancies. Adjusted OR of preterm birth comparing women with and without a specific ARV use in pregnancy. A GEE model was fit separately for each ARV exposure, adjusting for number of singleton birth, birth cohort, household income, and living with partner.


Among WLHIV, CD4 counts early in pregnancy increased with repeat pregnancies. Yet, many were not virally suppressed at their first VL assessment in each pregnancy, even if they achieved viral suppression late in the previous pregnancy, confirming other studies showing viral rebound in the postpartum period.14–16 This may be due to decreased adherence, barriers to engagement in health care postpartum, and changes in treatment guidelines.14,17 The higher rates of viral suppression early in pregnancy in recent calendar years likely reflects changing guidelines. After 2011, WLHIV were more likely to continue ART for their own health after delivery, based on guideline changes in 2011 that recommended ART for all pregnant HIV-infected women regardless of the CD4 count.10,18

Identifying barriers to continued engagement in care for WLHIV is essential because postpartum viral rebound jeopardizes the health of the woman and the health of her family, creating the potential for HIV transmission to a breastfed infant and sexual partners. Several studies have identified challenges WLHIV face in the postpartum period, including balancing infant care and work, transportation, and stigma that can hinder care engagement and retention.19–22 Using a family-centered care model may optimize continuous care engagement and sustained viral suppression. A recent study of 411 WLHIV in South Africa showed that women randomized to receive integrated pediatric and maternal care after delivery were more likely to be retained in care and maintain viral suppression than those who receive standard of care with pediatric and maternal care provided separately.23 Similarly, a US study demonstrated a 16% increase in the number of women with viral suppression at 6 months postpartum with provision of services through a multidisciplinary perinatal care coordination team, including individualized care plans for WLHIV that address barriers to care and coordination of postpartum care with infant well-child care.24 Novel approaches that support linking maternal and infant treatment may enhance maternal adherence.

About 9% of the women included in this study acquired HIV perinatally. These women had lower mean CD4 counts and were less likely to be virally suppressed, both early and late in pregnancy, than women with nonperinatally acquired HIV. Previous reports from the SMARTT study25,26 and others have similarly reported that women with perinatally acquired HIV have lower baseline CD4 counts at conception and are at a greater risk for a detectable VL near delivery.27,28 These differences in the baseline CD4 count may be attributable to suboptimal ART regimens, drug resistance, and/or suboptimal adherence due to psychosocial factors.29

Black women were less likely to have viral suppression compared with non-Black women. This may be related to delayed ARV initiation30 or other contributing structural factors, such as lack of access to HIV care and less social support.31,32 Women with less than a high school education were also less likely to be virally suppressed, which may be related to poverty and poor health literacy.33,34

Preterm birth is an important clinical and public health problem, with a higher risk of preterm birth among WLHIV than the general population. In this study, the frequency of preterm birth in women's index and second study pregnancy was 14%, which increased to 21% in their third study pregnancy, compared with an overall preterm birth rate of 19% we previously reported in SMARTT.5 This increase in preterm birth in the third pregnancy may be related to increased maternal age or medical conditions, such as hypertensive disorders.35,36 Furthermore, PI use in the first trimester was significantly associated with an increased risk of preterm birth. This is consistent with our previous SMARTT analysis, which showed that PI-based combination regimens in the first trimester were associated with preterm birth.6 Surprisingly, we found that PI use at conception was not associated with an increased rate of preterm birth, perhaps, due to poor adherence before recognition of pregnancy, as evidenced by improvement in viral suppression from early to late in pregnancy. These findings raise important questions about the etiology of preterm birth among WLHIV, which remains unexplained. For example, early PI use may be associated with an increased risk of preterm birth due to an alteration of early placentation leading to greater likelihood of placentally mediated diseases, such as preeclampsia.37–39 Alternatively, a spontaneous preterm birth may be associated with a heightened inflammatory state associated with viremia and/or timing of ART regimen initiation.5 Study of women with both indicated and spontaneous preterm birth is essential to understanding these relationships. Importantly, this is the first report of an association between initiation of an INSTI in the first trimester and an increased odds of preterm birth, compared with those not using INSTIs. However, because only 96 women were exposed to INSTIs in the first trimester, future studies are needed to validate this finding and inform practice guidelines for WLHIV who are either of childbearing potential or are pregnant.

Our study had several limitations. Information on ARV side effects are not collected in SMARTT and self-reported adherence during pregnancy was not collected until 2015. ART regimen complexity, including pill burden,40 pill size, and side effects, as well as social stressors such as stigma and having multiple children can impact adherence.16,20 In addition, the number of women on discordant regimens (PI-containing and non–PI-containing regimen) with repeat pregnancies was small, limiting the power to detect differences in a preterm birth rate among women who changed regimens between pregnancies and limiting the power to detect differences based on timing of initiation. Similarly, the number of women receiving INSTIs was relatively small. It is possible that some women had additional pregnancies (with or without live births) before or between those included in this analysis, resulting in missing obstetric and ART information for these pregnancies. Thus, we did not include previous obstetric history, including preeclampsia or hypertensive disorders of pregnancy and preterm births in the multivariable models; however, these represent important risk factors for preterm birth that should be addressed in future analyses. It is particularly important to define modifiable causes of the higher preterm birth rate among WLHIV. Another limitation is that CD4 and VLs between pregnancies were not collected. Finally, the practice guidelines for ARV treatment of pregnant women have evolved over the course of time. We addressed this by controlling for year of delivery, revealing that viral suppression during pregnancy was more likely after 2010.

In conclusion, in this US cohort, most WLHIV with repeat pregnancies achieved virologic suppression during pregnancy but many had viremia early in their subsequent pregnancy. Maintaining engagement in postpartum care is critical for optimizing women's health and minimizing horizontal and vertical HIV transmission. Furthermore, HIV disease progression between pregnancies was more pronounced among women with perinatally acquired HIV, who had lower CD4 cell counts throughout pregnancy. PI-based and INSTI-based regimens initiated in the first trimester were associated with significantly increased odds of preterm birth. Given that the United States and other guidelines promote some INSTIs as part of a first line regimen in pregnancy, further studies are needed to validate our findings and identify mechanisms for the observed increase in preterm birth.


The authors thank the women and families for their participation in PHACS and the individuals and institutions involved in the conduct of PHACS. The study was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development with co-funding from the National Institute on Drug Abuse, the National Institute of Allergy and Infectious Diseases, the National Institute of Mental Health, the National Institute of Neurological Disorders and Stroke, the National Institute on Deafness and Other Communication Disorders, the National Institute of Dental and Craniofacial Research, the National Cancer Institute, the National Institute on Alcohol Abuse and Alcoholism, the Office of AIDS Research, and the National Heart, Lung, and Blood Institute through cooperative agreements with the Harvard T.H. Chan School of Public Health (HD052102) (Principal Investigator: George R Seage III; Program Director: Liz Salomon) and the Tulane University School of Medicine (HD052104) (Principal Investigator: Russell Van Dyke; Co-Principal Investigator: E.G.C.; Project Director: Patrick Davis). Data management services were provided by the Frontier Science and Technology Research Foundation (PI: Suzanne Siminski), and regulatory services and logistical support were provided by Westat, Inc. (PI: Julie Davidson). The following institutions, clinical site investigators, and staff participated in conducting PHACS SMARTT in 2018, in alphabetical order: Ann and Robert H. Lurie Children's Hospital of Chicago: E.G.C., Margaret Ann Sanders, Kathleen Malee, and Scott Hunter; Baylor College of Medicine: William Shearer, Mary Paul, Chivon McMullen-Jackson, Ruth Eser-Jose, and Lynnette Harris; Bronx Lebanon Hospital Center: Murli Purswani, Mahoobullah Mirza Baig, and Alma Villegas; Children's Diagnostic and Treatment Center: Lisa Gaye-Robinson, Jawara Dia Cooley, James Blood, and Patricia Garvie; New York University School of Medicine: William Borkowsky, Sandra Deygoo, and Jennifer Lewis; Rutgers—New Jersey Medical School: Arry Dieudonne, Linda Bettica, Juliette Johnson, and Karen Surowiec; St. Jude Children's Research Hospital: Katherine Knapp, Kim Allison, Megan Wilkins, and Jamie Russell-Bell; San Juan Hospital/Department of Pediatrics: Nicolas Rosario, Lourdes Angeli-Nieves, and Vivian Olivera; SUNY Downstate Medical Center: Stephan Kohlhoff, Ava Dennie, and Jean Kaye; Tulane University School of Medicine: Russell Van Dyke, Karen Craig, and Patricia Sirois; University of Alabama, Birmingham: Cecelia Hutto, Paige Hickman, and Dan Marullo; University of California, San Diego: Stephen A. Spector, Veronica Figueroa, Megan Loughran, and Sharon Nichols; University of Colorado, Denver: Elizabeth McFarland, Emily Barr, Christine Kwon, and Carrie Glenny; University of Florida, Center for HIV/AIDS Research, Education and Service: Mobeen Rathore, Kristi Stowers, Saniyyah Mahmoudi, Nizar Maraqa, and Rosita Almira; University of Illinois, Chicago: Karen Hayani, Lourdes Richardson, Renee Smith, and Alina Miller; University of Miami: Gwendolyn Scott, Maria Mogollon, Gabriel Fernandez, and Anai Cuadra; Keck Medicine of the University of Southern California: Toni Frederick, Mariam Davtyan, Jennifer Vinas, and Guadalupe Morales-Avendano; University of Puerto Rico School of Medicine, Medical Science Campus: Zoe M. Rodriguez, Lizmarie Torres, and Nydia Scalley.


1. Haddad LB, Wall KM, Mehta CC, et al. Trends of and factors associated with live-birth and abortion rates among HIV-positive and HIV-negative women. Am J Obstet Gynecol. 2017;216:71.e71–71.e16.
2. French CE, Cortina-Borja M, Thorne C, et al. Incidence, patterns, and predictors of repeat pregnancies among HIV-infected women in the United Kingdom and Ireland, 1990–2009. J Acquir Immune Defic Syndr. 2012;59:287–293.
3. Bryant AS, Leighty RM, Shen X, et al. Predictors of repeat pregnancy among HIV-1-infected women. J Acquir Immune Defic Syndr. 2007;44:87–92.
4. Kreitchmann R, Megazzini K, Melo VH, et al. Repeat pregnancy in women with HIV infection in Latin America and the Caribbean. AIDS Care. 2015;27:1289–1297.
5. Floridia M, Tamburrini E, Masuelli G, et al. Rate, correlates and outcomes of repeat pregnancy in HIV-infected women. HIV Med. 2017;18:440–443.
6. Watts DH, Williams PL, Kacanek D, et al. Combination antiretroviral use and preterm birth. J Infect Dis. 2013;207:612–621.
7. Powis KM, Kitch D, Ogwu A, et al. Increased risk of preterm delivery among HIV-infected women randomized to protease versus nucleoside reverse transcriptase inhibitor-based HAART during pregnancy. J Infect Dis. 2011;204:506–514.
8. Shapiro RL, Hughes MD, Ogwu A, et al. Antiretroviral regimens in pregnancy and breast-feeding in Botswana. N Engl J Med. 2010;362:2282–2294.
9. The American College of Obstetricians, Committee on Practice Bulletins-Obstetrics. Practice bulletin no. 130: prediction and prevention of preterm birth. Obstet Gynecol. 2012;120:964–973.
10. Powis KM, Huo Y, Williams PL, et al. Antiretroviral prescribing practices among pregnant women living with HIV in the United States, 2008–2017. JAMA Netw Open. 2019;2:e1917669.
11. Uthman OA, Nachega JB, Anderson J, et al. Timing of initiation of antiretroviral therapy and adverse pregnancy outcomes: a systematic review and meta-analysis. Lancet HIV. 2017;4:e21–e30.
12. Griner R, Williams PL, Read JS, et al. In utero and postnatal exposure to antiretrovirals among HIV-exposed but uninfected children in the United States. AIDS Patient Care STDS. 2011;25:385–394.
13. American College of Obstetricians and Gynecologists. Methods for estimating the due date. Committee opinion no. 700. Obstet Gynecol. 2017;129:e150–e154.
14. Swain CA, Smith LC, Nash D, et al. Postpartum loss to HIV care and HIV viral suppression among previously diagnosed HIV-infected women with a live birth in New York state. PLoS One. 2016;11:e0160775.
15. Adams JW, Brady KA, Michael YL, et al. Postpartum engagement in HIV care: an important predictor of long-term retention in care and viral suppression. Clin Infect Dis. 2015;61:1880–1887.
16. Patel K, Karalius B, Powis K, et al. Trends in post-partum viral load among women living with perinatal HIV infection in the USA: a prospective cohort study. Lancet HIV. 2020;7:e184–e192.
17. Mellins CA, Chu C, Malee K, et al. Adherence to antiretroviral treatment among pregnant and postpartum HIV-infected women. AIDS Care. 2008;20:958–968.
18. A Working Group of the Office of AIDS Research Advisory Council (OARAC). Panel on treatment of pregnant women with HIV infection and perinatal transmission. Recommendations for use of antiretroviral drugs in transmission in the United States. 2011. Available at: Accessed February 5, 2020.
19. Momplaisir FM, Storm DS, Nkwihoreze H, et al. Improving postpartum retention in care for women living with HIV in the United States. AIDS. 2018;32:133–142.
20. Boehme AK, Davies SL, Moneyham L, et al. A qualitative study on factors impacting HIV care adherence among postpartum HIV-infected women in the rural southeastern USA. AIDS Care. 2014;26:574–581.
21. Buchberg MK, Fletcher FE, Vidrine DJ, et al. A mixed-methods approach to understanding barriers to postpartum retention in care among low-income, HIV-infected women. AIDS Patient Care STDS. 2015;29:126–132.
22. Siddiqui R, Bell T, Sangi-Haghpeykar H, et al. Predictive factors for loss to postpartum follow-up among low income HIV-infected women in Texas. AIDS Patient Care STDS. 2014;28:248–253.
23. Myer L, Phillips TK, Zerbe A, et al. Integration of postpartum healthcare services for HIV-infected women and their infants in South Africa: a randomized controlled trial. PLoS Med. 2018;15:e1002547.
24. Hackett S, Badell ML, Meade CM, et al. Improved perinatal and postpartum HIV outcomes after utilization of a perinatal care coordination team. Open Forum Infect Dis. 2019;6:183.
25. Jao J, Kacanek D, Williams PL, et al. Birth weight and preterm delivery outcomes of perinatally vs nonperinatally human immunodeficiency virus-infected pregnant women in the United States: results from the PHACS SMARTT study and IMPAACT P1025 protocol. Clin Infect Dis. 2017;65:982–989.
26. Goodenough CJ, Patel K, Van Dyke RB. Is there a higher risk of mother-to-child transmission of HIV among pregnant women with perinatal HIV infection? Pediatr Infect Dis J. 2018;37:1267–1270.
27. Byrne L, Sconza R, Foster C, et al. Pregnancy incidence and outcomes in women with perinatal HIV infection. AIDS. 2017;31:1745–1754.
28. Powis KM, Slogrove AL, Okorafor I, et al. Maternal perinatal HIV infection is associated with increased infectious morbidity in HIV-exposed uninfected infants. Pediatr Infect Dis J. 2018;38:500–502.
29. Prieto LM, Fernandez McPhee C, Rojas P, et al. Pregnancy outcomes in perinatally HIV-infected young women in Madrid, Spain: 2000–2015. PLoS One. 2017;12:e0183558.
30. Meditz AL, MaWhinney S, Allshouse A, et al. Sex, race, and geographic region influence clinical outcomes following primary HIV-1 infection. J Infect Dis. 2011;203:442–451.
31. Geter A, Sutton MY, Armon C, et al. Trends of racial and ethnic disparities in virologic suppression among women in the HIV Outpatient Study, USA, 2010–2015. PLoS One. 2018;13:e0189973.
32. Edwards LV. Perceived social support and HIV/AIDS medication adherence among African American women. Qual Health Res. 2006;16:679–691.
33. Beer L, Mattson CL, Bradley H, et al. Understanding cross-sectional racial, ethnic, and gender disparities in antiretroviral use and viral suppression among HIV patients in the United States. Medicine. 2016;95:e3171.
34. Kalichman SC, Hernandez D, Kegler C, et al. Dimensions of poverty and health outcomes among people living with HIV infection: limited resources and competing needs. J Community Health. 2015;40:702–708.
35. Umesawa M, Kobashi G. Epidemiology of hypertensive disorders in pregnancy: prevalence, risk factors, predictors and prognosis. Hypertens Res. 2017;40:213–220.
36. Ye C, Ruan Y, Zou L, et al. The 2011 survey on hypertensive disorders of pregnancy (HDP) in China: prevalence, risk factors, complications, pregnancy and perinatal outcomes. PLoS One. 2014;9:e100180.
37. Sebitloane HM, Moodley J, Sartorius B. Associations between HIV, highly active anti-retroviral therapy, and hypertensive disorders of pregnancy among maternal deaths in South Africa 2011–2013. Int J Gynaecol Obstet. 2017;136:195–199.
38. Premkumar A, Dude AM, Haddad LB, et al. Combined antiretroviral therapy for HIV and the risk of hypertensive disorders of pregnancy: a systematic review. Pregnancy Hypertens. 2019;17:178–190.
39. Watts DH. Treating HIV during pregnancy: an update on safety issues. Drug Saf. 2006;29:467–490.
40. Turner BJ, Newschaffer CJ, Zhang D, et al. Antiretroviral use and pharmacy-based measurement of adherence in postpartum HIV-infected women. Med Care. 2000;38:911–925.

HIV; pregnancy; preterm birth; antiretrovirals

Supplemental Digital Content

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.