Abstract
Objective:
To compare the effect of pre-conception switching from nevirapine (NVP) and efavirenz (EFV) to dolutegravir (DTG) based ART on adverse birth and maternal outcomes.
Design:
Two target trial emulations using the Tsepamo Birth Outcomes Surveillance Study in Botswana.
Methods:
Among individuals who initiated NVP or EFV, we compared those who switched to DTG with those who did not switch prior to pregnancy. We estimated adjusted risk ratios (RRs) and 95% CIs for any adverse birth outcome (stillbirth, neonatal death, preterm delivery, or small-for-gestational-age), any severe adverse birth outcome (stillbirth, neonatal death, very preterm delivery, or very small-for-gestational-age), low (<50kg) and high (≥80kg) early pregnancy weight, and maternal hypertension (SBP≥140 or DBP≥90 mm/Hg).
Results:
Comparing NVP switchers (n=1,054) with non-switchers (n=3,163), RRs (95% CIs) were 0.81 (0.74, 0.89) for any adverse birth outcome, 0.81 (0.68, 0.96) for any severe adverse birth outcome, 0.74 (0.57, 0.96) for early pregnancy weight <50kg, 1.35 (1.15, 1.59) for early pregnancy weight ≥80kg, and 0.93 (0.82, 1.06) for maternal hypertension. Comparing EFV switchers (n=1,538) with non-switchers (n=7,227), RRs (95% CIs) were 0.99 (0.91, 1.07) for any adverse birth outcome, 1.05 (0.91, 1.22) for any severe adverse birth outcome, 0.93 (0.78, 1.12) for early pregnancy weight <50kg, 1.22 (1.06, 1.39) for early pregnancy weight ≥80kg, and 1.29 (1.14, 1.45) for maternal hypertension.
Conclusions:
Pre-conception switch from NVP may reduce the risk of adverse maternal and infant outcomes. Switch from EFV had no clear associations with adverse infant outcomes but may increase the risk of maternal hypertension.
Keywords: Antiretroviral therapy, Dolutegravir, Nevirapine, Efavirenz, Adverse Birth Outcomes, Target Trial Emulation
Introduction
In utero exposure to ART has been shown to increase the risk of adverse birth outcomes such as stillbirth, preterm delivery, and small-for-gestational-age (SGA) in randomized clinical trials,1,2 and a growing body of evidence suggests that these risks vary across ART regimens.2-5 For example, exposure to zidovudine (ZDV), lamivudine (3TC), and nevirapine (NVP) from conception has been associated with an increased risk of adverse birth outcomes compared to tenofovir (TDF), emtricitabine (FTC), and efavirenz (EFV).4,5 Other regimens, such as those containing ritonavir-boosted lopinavir (LPV-R) in combination with TDF/FTC or ZDV/3TC, have also been associated with an increased risk of adverse birth outcomes compared with EFV/FTC/TDF.2,4 Exposure to dolutegravir (DTG) and FTC/TDF has been associated with similar risks of adverse birth outcomes as EFV/FTC/TDF6,7 when initiated during pregnancy, and these regimens have been associated with risks of adverse outcomes that are closest to those observed among individuals without HIV.7 ART regimens may also differentially impact maternal outcomes. For example, DTG has been associated with more weight gain,8 more gestational weight gain,9,10 and a potential increased risk of maternal hypertension11,12, compared with EFV.
The mechanism by which most antiretrovirals or specific combinations of antiretrovirals drive adverse birth outcomes remains unknown,13 and despite strong evidence that in-utero exposure to different ART regimens may have differential effects on adverse birth outcomes, it is unknown whether switching ART regimens prior to conception can alter the risk of adverse birth outcomes. The most likely possibility is that specific antiretroviral effects at the level of the placenta directly impact birth outcomes,14-19 in which case a pre-conception switch to a safer ART regimen would be expected to improve outcomes. However, if chronic prior exposure were more important, or if results from prior observational studies were affected by residual confounding, then switching regimens pre-conception would be expected to have little impact on outcomes.
In May 2016, Botswana became the first country in the world to recommend use of DTG as first-line ART for all adults, including individuals of childbearing potential and pregnant individuals,20 and programs around the world continue to follow suit. This early switch from legacy regimens such as NVP (and later EFV) to DTG-based ART allows us to evaluate the impact of pre-conception ART switching on adverse birth outcomes in Botswana, and to provide the first data for countries making a similar switch. Because this question cannot be feasibly studied in a traditional randomized clinical trial as it would require enrolling individuals of childbearing potential and following them for many years, we used nationwide surveillance data from the Tsepamo Study to emulate (hypothetical) target trials21 of pre-conception switch from NVP-based and EFV-based ART to DTG-based ART on adverse birth and maternal outcomes.
Methods
The Tsepamo Study
Tsepamo is an ongoing birth outcomes surveillance study in Botswana which has operated at government delivery hospitals since August 2014.4 Data are abstracted from the maternity obstetric record at the time of discharge from the maternity ward from all livebirths and stillbirths. Tsepamo included 8 hospitals (~45% of all births in Botswana) from August 2014 to July 2018; 18 hospitals (~72% of births) from July to December 2018; and 16 hospitals (~68% of births) from January 2019 onward. Tsepamo captures >99% of births that take place at included sites.4,22 In Botswana, approximately 95% of births occur at a hospital.23 For this analysis, we used data from deliveries on or before August 15, 2023.
The maternity obstetric record captures information on demographics, medical history, medications prescribed during pregnancy, HIV (time of diagnosis, ART regimens, CD4 counts and HIV viral loads), and clinical information recorded during pregnancy. Gestational age is documented by midwives at delivery based on the estimated date of delivery (EDD). EDD is calculated at the first ANC visit using the reported last menstrual period (LMP) and confirmed by ultrasound when available. If LMP is unknown or suspected to be incorrect, fundal height measurements are used to estimate gestational age.
In Botswana, the recommended first-line three-drug ART regimen changed from a NVP-based regimen (NVP/ZDV/3TC or NVP/TDF/FTC) to an EFV-based regimen (EFV/TDF/FTC) in 201224, and from EFV/TDF/FTC to a DTG-based regimen (DTG/TDF/FTC or DTG/TDF/3TC) in 2016. Following this change, the national ART program began switching individuals from legacy regimens to DTG-based ART. However, the timing of switch differed by district and by antiretroviral supplies, allowing for period when deliveries occurred among both those who had and had not switched ART.
Protocol of the target trials
We emulated separate (hypothetical) target trials of pre-conception switch from NVP- or EFV-based ART to DTG-based ART following the 2016 guideline change in Botswana (Table 1 and Appendix Figure 1 and 2). The target trials enrolled individuals of childbearing potential with a known date of HIV diagnosis, who initiated an NVP- or EFV-based ART regimen within 5 years of HIV diagnosis (between January 1, 2002 and December 31, 2013 for NVP; between January 1, 2009 and December 31, 2015 for EFV); remained on their legacy ART regimen on January 1, 2016; and were not pregnant on January 1, 2016. Individuals who switched to different backbones (e.g., from NVP/ZDV/3TC to NVP/TDF/FTC) before 2016 were still eligible for the target trial. In each target trial, eligible individuals were randomly assigned on January 1, 2016 to one of two treatment strategies: (1) Switch to a DTG-based ART regimen prior to pregnancy; or (2) Do not switch from their legacy ART regimen prior to pregnancy. Both strategies implicitly require a singleton pregnancy and delivery of a livebirth or stillbirth infant at or after 24 weeks gestation. Both groups were then followed longitudinally for maternal and infant outcomes recorded through the time of discharge from the maternity ward after delivery. To allow adequate follow-up time, pregnancies for which the estimated LMP was on or after September 27, 2022 (46 weeks prior to August 15, 2023) were excluded. After delivery, individuals could re-enroll in either target trial if they continued to meet the eligibility criteria.
Table 1.
Protocol of a target trial of pre-conception switch from nevirapine-based ART (NVP) to dolutegravir-based ART (DTG) and from efavirenz-based ART (EFV) to DTG and emulation using the Tsepamo Study
| Protocol Element | Target Trial of NVP to DTG switch |
Target Trial of EFV to DTG switch |
Emulation Using Tsepamo Data |
|---|---|---|---|
| Eligibility | -Childbearing potential -Known date of HIV diagnosis -Initiate NVP from 2002-2013, within 5 years of HIV diagnosis -Remain on NVP on January 1, 2016 -Not pregnant on January 1, 2016 |
-Childbearing potential -Known date of HIV diagnosis -Initiate EFV from 2009-2015, within 5 years of HIV diagnosis -Remain on EFV on January 1, 2016 -Not pregnant on January 1, 2016 |
Same. |
| Treatment strategies | (1) Switch to DTG prior to pregnancy (2) Do not switch from NVP prior to pregnancy Both strategies implicitly require a singleton pregnancy from 2016-2023 and delivery of a livebirth or stillbirth infant at or after 24 weeks gestation. |
(1) Switch to DTG prior to pregnancy (2) Do not switch from EFV prior to pregnancy Both strategies implicitly require a singleton pregnancy from 2016-2023 and delivery of a livebirth or stillbirth infant at or after 24 weeks gestation. |
Same. |
| Randomized assignment | Individuals will be randomly assigned to either strategy at enrollment, and will be aware of the strategy they have been assigned to. | Individuals will be randomly assigned to either strategy at enrollment, and will be aware of the strategy they have been assigned to. | Individuals are assumed to be randomly assigned at enrollment within levels of the measured covariates. |
| Follow-up period | Start: January 1, 2016 End: Discharge from the maternity ward after delivery |
Start: January 1, 2016 End: Discharge from the maternity ward after delivery |
Same |
| Outcomes | Individual and composite adverse birth outcomes | Individual and composite adverse birth outcomes | Same |
| Causal contrasts of interest | -Intention-to-treat effect* | -Intention-to-treat effect* | Observational analogue to the intention-to-treat effect |
| Analysis plan | -Intention-to-treat analysis (log-binomial models to estimate risk ratios) | -Intention-to-treat analysis (log-binomial models to estimate risk ratios) | Same, plus adjustment for baseline covariates necessary to achieve conditional exchangeability |
Among those who become pregnant and deliver a livebirth or stillbirth infant at or after 24 weeks gestation.
Maternal outcomes included low (<50kg) and high (≥80kg) early pregnancy weight (defined using the earliest weight measured prior to 24 weeks gestation), hypertension in pregnancy (any systolic blood pressure ≥140 mm/Hg or diastolic blood pressure ≥90 mm/Hg), and early pregnancy weight as a continuous variable. Adverse birth outcomes included stillbirth (fetal death ≥24 weeks gestation); preterm delivery (<37 weeks’ gestation); very preterm delivery (<32 weeks’ gestation); SGA (<10th percentile of birth weight using WHO norms); and very SGA (<3rd percentile of birth weight using WHO norms)25,26. The combined endpoint of any adverse birth outcome included stillbirth, neonatal death, preterm delivery, or SGA. The combined endpoint of any severe adverse birth outcome included stillbirth, neonatal death, very preterm delivery, or very SGA.
In each target trial, we fit log-binomial regression models for each outcome of interest to estimate the intention-to-treat effect of switching to a DTG-based ART regimen prior to pregnancy versus not switching prior to pregnancy. For the continuous outcome of early pregnancy weight, we fit a linear regression model.
Emulation of the target trials using Tsepamo Data
We emulated the target trial using The Tsepamo Data. We include the following baseline covariates in our regression models: time from HIV diagnosis to NVP or EFV initiation, year of NVP or EFV initiation, age on January 1, 2016, health district of first antenatal care visit, occupation, education, and parity. Individuals with multiple pregnancies could be included more than once if the individual continued to meet the eligibility criteria.
Sensitivity analyses for confounding
We performed two primary sensitivity analyses in both target trials. First, we adjusted our estimates for year of conception and maternal age at first antenatal care visit. Because switching to DTG-based ART was more likely in later years, we hypothesized that switchers might be older at conception and pregnant in later calendar years. Second, we restricted the analysis to individuals who conceived during an era when both treatment strategies were common (2018-2021 for the NVP target trial and 2020-2022 for the EFV target trial). This analysis was conducted to minimize potential confounding due to temporal trends in outcomes and due to unknown or unmeasured factors that could have informed decisions to switch treatment. Both of these analyses involve adjusting for (or restricting based on) post-baseline variables and rely on the assumption that such variables (age and calendar year at conception) are not on the causal pathway.
For the NVP target trial, we conducted two additional sensitivity analyses. We excluded individuals on ART containing ZDV at conception (<1% of individuals in the EFV analysis were on ART containing ZDV). First line NVP-based ART in Botswana included both ZDV and TDF backbones, whereas first line DTG-based ART included TDF backbones only. This analysis isolated the effect of switching from NVP from a potential effect of switching from ZDV. To control for potential confounding by CD4 cell count, we conducted an analysis using a subset of the data (including deliveries on or before August 22, 2021) that matched each switcher to one non-switcher, by age (+/− 2.5 years) and urban versus rural health district. We used an online medical record available at some healthcare facilities in Botswana to identify CD4 cell counts for this subset and additionally adjusted this analysis for CD4 cell count.
Finally, to determine whether there may have been time trends in adverse outcomes in Botswana during the study period, we used data from all singleton deliveries to individuals without HIV in Tsepamo and calculated the risk of each adverse birth and maternal outcome by calendar year.
Results for NVP Target Trial
Of 48,416 individuals living with HIV in Tsepamo, 4,531 started NVP-based ART in 2002-2013 within 5 years of HIV diagnosis and remained on an NVP-based ART regimen while not pregnant on January 1, 2016. Of these individuals, 1,133 (25%) switched from NVP to DTG-based ART prior to pregnancy, of whom 1,054 had a singleton pregnancy before September 27, 2022. Of the eligible individuals, 3,244 (72%) did not switch from NVP-based ART prior to pregnancy, of whom 3,163 had a singleton pregnancy (Figure 1). Switchers and non-switchers had similar baseline characteristics, including year of NVP initiation, time from HIV diagnosis to NVP initiation, occupation, education, parity, and health district. Compared with non-switchers, switchers were younger at baseline (in 2016) but were older in pregnancy and became pregnant in later calendar years. The median (IQR) maternal age at first antenatal visit was 37 (32, 40) years among switchers and 36 (32, 39) years among non-switchers (Table 2). Most switchers had pregnancies beginning in 2020-2022, whereas most non-switchers had pregnancies beginning in 2019 or earlier (Appendix Table 1). The most common backbones prior to pregnancy were TDF/3TC (89%) and TDF/FTC (8%) for switchers and ZDV/3TC (60%) and TDF/FTC (39%) for non-switchers (Appendix Table 2).
Figure 1.
Study inclusion and eligibility criteria, The Tsepamo Study
Table 2.
Characteristics by pre-conception ART regimen switch status, The Tsepamo Study
| Characteristic | Switch from NVP to DTG n=1,054 Number (%) |
Do not switch from NVP n=3,163 Number (%) |
Switch from EFV to DTG n=1,538 Number (%) |
Do not switch from EFV n=7,227 Number (%) |
|---|---|---|---|---|
| Year of NVP/EFV initiation | ||||
| 2002-2008 | 498 (47.25) | 1,492 (47.17) | - | - |
| 2009-2013 | 556 (52.75) | 1,671 (52.83) | - | - |
| 2009-2012 | - | - | 459 (29.84) | 2246 (31.08) |
| 2013-2015 | - | - | 1079 (70.16) | 4981 (68.92) |
| Time from HIV diagnosis to NVP/EFV initiation | ||||
| 1 year or less | 873 (82.83) | 2,560 (80.94) | 1277 (83.03) | 6052 (83.74) |
| More than 1 year | 181 (17.17) | 603 (19.06) | 261 (16.97) | 1174 (16.26) |
| Median (IQR) number of days | 0 (0, 274) | 0 (0, 334) | 0 (0, 208) | 0 (0, 246) |
| Age on Jan 1 2016 | ||||
| Less than 25 years | 208 (19.73) | 297 (9.39) | 496 (32.25) | 1720 (23.80) |
| 25-30 years | 219 (20.78) | 510 (16.12) | 494 (32.12) | 2121 (29.35) |
| 30 years or older | 613 (58.16) | 2,306 (72.91) | 508 (33.03) | 3204 (44.33) |
| Unknown | 14 (1.33) | 50 (1.58) | 40 (2.60) | 182 (2.52) |
| Median (IQR) (years) | 31.3 (27.1, 34.6) | 33.3 (29.8, 36.4) | 28 (24, 32) | 29 (25, 33) |
| Occupation | ||||
| Salaried | 408 (38.71) | 1,281 (40.50) | 536 (34.85) | 2496 (34.54) |
| Student, housewife, none, unknown | 646 (61.29) | 1,882 (59.50) | 1002 (65.15) | 4731 (65.46) |
| Education | ||||
| None or primary or unknown | 140 (13.28) | 555 (17.55) | 199 (12.94) | 906 (12.54) |
| Secondary or more | 914 (86.72) | 2,608 (82.45) | 1339 (87.06) | 6321 (87.46) |
| Parity | ||||
| 0 | 121 (11.48) | 263 (8.31) | 102 (6.63) | 449 (6.21) |
| 1 or more | 931 (88.33) | 2,897 (91.59) | 1435 (93.30) | 6767 (93.63) |
| Unknown | 2 (0.19) | 3 (0.09) | 1 (0.07) | 11 (0.15) |
| Health district of first antenatal visit | ||||
| Urban (Gaborone or Francistown) | 304 (28.84) | 1,001 (31.65) | 405 (26.33) | 2017 (27.91) |
| Rural (all others or unknown) | 750 (71.16) | 2,162 (68.35) | 1133 (73.67) | 5210 (72.09) |
| Year of Conception (LMP) a | ||||
| 2016-2017 | 12 (1.14) | 1,568 (49.57) | - | - |
| 2018-2019 | 144 (13.66) | 1,418 (44.83) | - | - |
| 2020-2023 | 898 (85.20) | 177 (5.60) | - | - |
| 2016-2019 | - | - | 163 (10.60) | 5643 (78.08) |
| 2020-2021 | - | - | 810 (52.67) | 1421 (19.66) |
| 2022-2023 | - | - | 565 (36.74) | 163 (2.26) |
| Age at first antenatal visit a | ||||
| Less than 32 years | 241 (22.87) | 671 (21.21) | 577 (37.52) | 3214 (44.47) |
| 32-37 years | 255 (24.19) | 1105 (34.94) | 485 (31.53) | 2260 (31.27) |
| 37 years or older | 558 (52.94) | 1385 (43.79) | 475 (30.88) | 1750 (24.21) |
| Unknown | 0 (0.0) | 2 (0.06) | 1 (0.07) | 3 (0.04) |
| Median (IQR) Age (years) | 37 (32, 40) | 36 (32, 39) | 34 (29, 37) | 32 (28, 36) |
Not a “baseline” variable. Included as adjustment variable in sensitivity analyses only.
NVP, Nevirapine; DTG, dolutegravir; EFV, efavirenz
Note n denotes the number of deliveries included in the analysis (individuals may be included more than once if they have more than one delivery)
Comparing switchers with non-switchers, adjusted risk ratios (95% CIs) were 0.81 (0.74, 0.89) for any adverse birth outcome, 0.81 (0.68, 0.96) for any severe adverse birth outcome, 0.74 (0.57, 0.96) for early pregnancy weight <50kg, 1.35 (1.15, 1.59) for early pregnancy weight ≥80kg, and 0.93 (0.82, 1.06) for any hypertension in pregnancy. The adjusted mean difference (95% CI) in first weight in pregnancy was 3.0 (1.8, 4.2) kg, comparing switchers with non-switchers. The adverse birth outcome results were in part driven by a decreased risk of SGA (adjusted risk ratio: 0.79; 95% CI: 0.69, 0.91) and very SGA (adjusted risk ratio: 0.76; 95% CI: 0.61, 0.96) (Figure 2). Results were not appreciably different when we adjusted for year of conception and age at first ANC, required conception in 2018-2021, excluded those on ZDV backbones (Figure 2, Appendix Table 3, Appendix Table 4), and adjusted for CD4 cell count (Appendix Figure 3, Appendix Table 5), though confidence intervals were wider than in the primary analysis.
Figure 2.
Adjusted* risk ratios for adverse birth and maternal outcomes comparing switching from nevirapine (NVP) to dolutegravir with not switching from NVP prior to pregnancy in primary and sensitivity analyses, The Tsepamo Study
*Risk ratios are adjusted for time from HIV diagnosis to NVP initiation (1 year or less; more than 1 year), year of NVP initiation (2002-2009; 2009-2013), age on January 1, 2016 (<25 years; 25-30 years; 30 years or older or unknown), health district of first antenatal care visit (urban: Gaborone or Francistown; rural: all others or unknown), occupation (salaried; other or unknown), education (none, primary, or unknown; secondary or more), and parity (0 or unknown; 1 or more). Maternal weight analyses are restricted to those with a weight measurement prior to 24 weeks gestation. Maternal hypertension analyses are restricted to those with a systolic and diastolic blood pressure measurement at some point in pregnancy. These results are also shown in Appendix Tables 3 and 4.
Results for EFV Target Trial
Of 48,416 individuals living with HIV in Tsepamo, 9,187 started EFV-based ART in 2009-2015 within 5 years of HIV diagnosis and remained on EFV-based ART regimen while not pregnant on January 1, 2016. Of these individuals, 1,655 (18%) switched from EFV to DTG-based ART prior to pregnancy, of whom 1,538 had a singleton pregnancy before September 27, 2022. Of the eligible individuals, 7,381 (80%) did not switch from EFV-based ART prior to pregnancy, of whom 7,227 had a singleton pregnancy (Figure 1). Switchers and non-switchers had similar baseline characteristics, including year of EFV initiation, time from HIV diagnosis to EFV initiation, occupation, education, parity, and health district. Compared with non-switchers, switchers were younger at baseline (in 2016) but were older in pregnancy and became pregnant in later calendar years. The median (IQR) maternal age at first antenatal visit was 34 (29, 37) years among switchers and 32 (28, 36) years among non-switchers (Table 2). Most switchers had pregnancies beginning in 2020-2023, whereas most non-switchers had pregnancies beginning in 2019 or earlier (Appendix Table 6). The most common backbones prior to pregnancy were TDF/3TC (91%) and TDF/FTC (7%) for switchers and TDF/FTC (98%) for non-switchers (Appendix Table 7).
Comparing switchers with non-switchers, adjusted risk ratios (95% CIs) were 0.99 (0.91, 1.07) for any adverse birth outcome, 1.05 (0.91, 1.22) for any severe adverse birth outcome, 0.93 (0.78, 1.12) for early pregnancy weight <50kg, 1.22 (1.06, 1.39) for early pregnancy weight ≥80kg, and 1.29 (1.14, 1.45) for any hypertension in pregnancy. The adjusted mean difference (95% CI) in first weight in pregnancy was 1.9 (0.9, 2.9) kg, comparing switchers with non-switchers. The adverse birth outcome result was in part driven by a decreased risk of very preterm birth (adjusted risk ratio: 0.74; 95% CI: 0.55, 1.01) but an increased risk of SGA (adjusted risk ratio: 1.10, 95% CI: 0.98, 1.23) (Figure 3). Results were not always consistent in our sensitivity analyses that adjusted for year of conception and age at first ANC and that required conception in 2018-2021, notably for stillbirth and risk of early pregnancy weight ≥80kg (Figure 3, Appendix Table 8, and Appendix Table 9).
Figure 3.
Adjusted* risk ratios for adverse birth and maternal outcomes comparing switching from efavirenz (EFV) to dolutegravir with not switching from EFV prior to pregnancy in primary and sensitivity analyses, The Tsepamo Study
*Risk ratios are adjusted for time from HIV diagnosis to EFV initiation (1 year or less; more than 1 year), year of EFV initiation (2009-2012; 2013-2015), age on January 1, 2016 (<25 years; 25-30 years; 30 years or older or unknown), health district of first antenatal care visit (urban: Gaborone or Francistown; rural: all others or unknown), occupation (salaried; other or unknown), education (none, primary, or unknown; secondary or more), and parity (0 or unknown; 1 or more). Maternal weight analyses are restricted to those with a weight measurement prior to 24 weeks gestation. Maternal hypertension analyses are restricted to those with a systolic and diastolic blood pressure measurement at some point in pregnancy. These results are also shown in Appendix Tables 8 and 9.
Results for deliveries among individuals without HIV
Among individuals without HIV in Tsepamo who conceived in 2016-2022 and had a singleton delivery, we did not observe any appreciable trends in the risk of the composite outcomes of any adverse birth outcome and any severe adverse birth outcome. However, coinciding with the COVID-19 epidemic, the risk of stillbirth was higher in 2020-2022 (ranging from 2.26% in 2021 to 2.52% in 2022) compared with 2016-2019 (ranging from 1.92% in 2018 to 2.11% in 2017); the risk of preterm delivery and very preterm delivery was higher in 2022 compared with 2016-2021 (risk of preterm delivery was 14.61% in 2022 compared with 13.68%-15.02% in 2016-2021; risk of very preterm delivery was 3.42% in 2022 compared with 2.79%-3.18% in 2016-2021); and the risk of SGA and very SGA was higher in 2021-2022 compared with 2016-2020 (risk of SGA was 15.77%-15.94% in 2021-2022 compared to 14.24%-14.87% in 2016-2020; risk of very SGA was 5.76%-5.92% in 2021-2022 compared to 5.26%-5.58% in 2016-2020) (Appendix Table 10). We did not observe any appreciable trends in the risk of early pregnancy weight <50kg or any hypertension in pregnancy. However, the risk of early pregnancy weight ≥80kg was higher in 2020-2022 (ranging from 18.69% in 2020 to 19.23% in 2021) compared with 2016-2019 (ranging from 15.11% in 2016 to 16.81% in 2017).
Discussion
Leveraging data from the largest birth outcomes surveillance study in Africa and the programmatic switch to DTG-based ART that occurred in Botswana starting in 2016, we emulated two target trials of pre-conception switch to DTG-based ART. Our findings suggest that pre-conception switch from NVP-based ART may reduce the risk of SGA and low maternal weight in early pregnancy, whereas pre-conception switch from EFV-based ART was possibly associated with reduced risk of very preterm birth but an increased risk of maternal hypertension in pregnancy. Taken together, these findings suggest that different antiretroviral combinations can have specific effects on maternal and infant outcomes, and that generally favorable trends were identified for infant outcomes when switching to DTG from either NVP or EFV prior to conception.
Switching from NVP was associated with a lower risk of SGA, potentially through effects at the level of the placenta, which is consistent with prior studies.4,5 On the other hand, we found no difference in the risk of stillbirth after pre-conception switch, despite very high stillbirth risk reported in prior studies.4,27 Although this may suggest chronic prior NVP exposure could have a lingering effect on stillbirths, there are important caveats. First, the population included in our analysis was older (median age 36-37 years) compared to previous analyses, and the overall risk of stillbirth was high (>3.5%); it remains possible that pre-conception switch could impact stillbirth risk in younger populations with lower baseline risk. Second, switching from NVP was more likely among those who delivered in later calendar years, and we also observed noticeable time trends in the risk of stillbirth in the HIV unexposed group; this could have masked a protective effect of pre-conception switching, even though results appeared robust in sensitivity analyses adjusting for calendar year.
Switching from EFV was associated with an increased risk of maternal hypertension, which is consistent with prior studies of pre-conception initiation of DTG-based ART.11 This finding could be explained in part by increased early pregnancy weight among switchers, though differences in early pregnancy weight were inconsistent across our analyses. Switching from EFV was also associated with a decrease in very preterm delivery (though CIs were wide), which has not been suggested by prior studies.6,7 However, prior studies focused on ART initiation in pregnancy rather than pre-conception switch. We also observed a small increased risk of SGA after pre-conception switch from EFV, but this finding could be explained by confounding due to temporal trends in SGA (since switching was more likely in later calendar years). The increased risk of stillbirth after pre-conception switch from EFV observed in our primary analysis was not observed in our sensitivity analyses and is likely explained by temporal trends in stillbirth.
A strength of our study was the target trial approach which helped to reduce (but not eliminate) sources of confounding. We also combined this with several sensitivity analyses which largely supported our findings, with a few exceptions. There were also limitations to our analysis, which primarily relate to residual confounding. Given the strong association between ART regimen and calendar year, a causal interpretation of our findings relies on the assumption that there are no time-trends in the outcomes of interest or that we were successfully able to adjust for these trends. In our analyses of pregnancies in individuals without HIV, we observed higher risks of several adverse birth outcomes in later calendar years compared with earlier calendar years, coinciding with the COVID-19 epidemic. While we adjusted for baseline calendar year in all analyses and calendar year in pregnancy in sensitivity analyses, any residual confounding by calendar year would mask a protective effect (or overestimate a harmful effect) of switching to DTG-based ART (because switching was more common in later years). Further, confounding by baseline maternal weight is possible since measurements of weight prior to pregnancy were not available. However, because of the programmatic switch to DTG in Botswana, it is unlikely that providers used weight to inform decisions about switching.
A causal interpretation of our findings also relies on the assumptions of no selection bias and no measurement error. Both the target trials and our emulations are susceptible to selection bias because individuals were excluded from the analysis based on events that occurred after baseline.27,28 Individuals who do not become pregnant, become pregnant but experience a pregnancy loss <24 weeks gestation, do not have a birth outcome at ≥24 weeks gestation at one of the sites included in our study, or deliver twins or triplets were excluded from our analysis. We also excluded individuals who switched to a non-DTG regimen prior to pregnancy (~3% in the NVP-DTG target trial and ~1% in the EFV-DTG target trial). Measurement error for preterm delivery and SGA is possible since gestational age was typically calculated using LMP, but it is unlikely that this error would be differential with respect to ART regimen. Finally, because we did not have data on multiple births from the same mother, we were not able to adjust our analysis for clustering, which could have led to an underestimation of the variance.
HIV treatment programs around the world are switching individuals from legacy ART regimens to DTG-based ART, including individuals of childbearing potential. We evaluated the impact of pre-conception switch to DTG-based ART on adverse birth and maternal outcomes, using a target trial emulation framework. We found that pre-conception switching from NVP-based ART was associated with reduced risk of low maternal weight and SGA, and that pre-conception switching from EFV-based ART was associated with reduced risk of very preterm delivery but increased maternal hypertension. Further, we observed an increase in the risk of several adverse birth outcomes over time, coinciding with the calendar years of the COVID-19 epidemic. The emulated trial design provided a powerful platform for analyzing clinical questions that could not be feasibly randomized and holds promise for the longitudinal analysis of programmatic surveillance datasets such as Tsepamo.
Supplementary Material
Acknowledgements:
EC Caniglia, R Zash, and R Shapiro designed and conceptualized the study. E Caniglia analyzed the data and drafted the manuscript. All authors contributed to drafting, reviewing, and revising the manuscript.
Funding:
This work was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development (Grants K01 HD100222, R01 HD080471, K23 HD088230, and P01HD107670), and the National Institute of Allergy and Infectious Diseases (K24 AI131924).
Footnotes
Conflicts of interest: The authors report no conflicts of interest.
References
- 1.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. The Journal of infectious diseases 2011; 204(4): 506–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fowler MG, Qin M, Fiscus SA, et al. Benefits and Risks of Antiretroviral Therapy for Perinatal HIV Prevention. The New England journal of medicine 2016; 375(18): 1726–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Zash R, Souda S, Chen JY, et al. Reassuring Birth Outcomes With Tenofovir/Emtricitabine/Efavirenz Used for Prevention of Mother-to-Child Transmission of HIV in Botswana. Journal of acquired immune deficiency syndromes (1999) 2016; 71(4): 428–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zash R, Jacobson DL, Diseko M, et al. Comparative Safety of Antiretroviral Treatment Regimens in Pregnancy. JAMA pediatrics 2017; 171(10): e172222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Caniglia EC, Zash R, Jacobson DL, et al. Emulating a target trial of antiretroviral therapy regimens started before conception and risk of adverse birth outcomes. AIDS (London, England) 2018; 32(1): 113–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lockman S, Brummel SS, Ziemba L, et al. Efficacy and safety of dolutegravir with emtricitabine and tenofovir alafenamide fumarate or tenofovir disoproxil fumarate, and efavirenz, emtricitabine, and tenofovir disoproxil fumarate HIV antiretroviral therapy regimens started in pregnancy (IMPAACT 2010/VESTED): a multicentre, open-label, randomised, controlled, phase 3 trial. Lancet (London, England) 2021; 397(10281): 1276–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zash R, Jacobson DL, Diseko M, et al. Comparative safety of dolutegravir-based or efavirenz-based antiretroviral treatment started during pregnancy in Botswana: an observational study. The Lancet Global health 2018; 6(7): e804–e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kouanfack C, Mpoudi-Etame M, Omgba Bassega P, et al. Dolutegravir-Based or Low-Dose Efavirenz-Based Regimen for the Treatment of HIV-1. The New England journal of medicine 2019; 381(9): 816–26. [DOI] [PubMed] [Google Scholar]
- 9.Caniglia EC, Shapiro R, Diseko M, et al. Weight gain during pregnancy among women initiating dolutegravir in Botswana. EClinicalMedicine 2020; 29-30: 100615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Venter WDF, Moorhouse M, Sokhela S, et al. Dolutegravir plus Two Different Prodrugs of Tenofovir to Treat HIV. The New England journal of medicine 2019; 381(9): 803–15. [DOI] [PubMed] [Google Scholar]
- 11.Jacobson DL, Diseko M, Mabuta J, et al. Hypertension in Pregnant Persons by HIV Status and by DTG vs EFV Use in Botswana. Conference on Retroviruses and Opportunistic Infections. Denver, Colorado; 2024. [Google Scholar]
- 12.Hoffman R, Brummel S, Pinilla M, et al. Hypertension in a Randomized Trial of DTG vs EFV-Based ART in Pregnant and Postpartum Women. Conference on Retroviruses and Opportunistic Infections. Denver, Colorado; 2024. [Google Scholar]
- 13.Papp E, Mohammadi H, Loutfy MR, et al. HIV protease inhibitor use during pregnancy is associated with decreased progesterone levels, suggesting a potential mechanism contributing to fetal growth restriction. The Journal of infectious diseases 2015; 211(1): 10–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dos Reis HLB, Boldrini NAT, Rangel AFR, Barros VF, Merçon de Vargas PR, Miranda AE. Placental growth disorders and perinatal adverse outcomes in Brazilian HIV-infected pregnant women. PloS one 2020; 15(4): e0231938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kojovic D, Ghoneim RH, Serghides L, Piquette-Miller M. Role of HIV and Antiretroviral Therapy on the Expression of Placental Transporters in Women with HIV. Aaps j 2020; 22(6): 138. [DOI] [PubMed] [Google Scholar]
- 16.Ikumi NM, Malaba TR, Pillay K, et al. Differential impact of antiretroviral therapy initiated before or during pregnancy on placenta pathology in HIV-positive women. AIDS (London, England) 2021; 35(5): 717–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dunk CE, Serghides L. Protease inhibitor-based antiretroviral therapy in pregnancy: effects on hormones, placenta, and decidua. The lancet HIV 2022; 9(2): e120–e9. [DOI] [PubMed] [Google Scholar]
- 18.Yampolsky M, Shlakhter O, Deng D, et al. Exploring the impact of HIV infection and antiretroviral therapy on placenta morphology. Placenta 2021; 104: 102–9. [DOI] [PubMed] [Google Scholar]
- 19.Shapiro RL, Souda S, Parekh N, et al. High prevalence of hypertension and placental insufficiency, but no in utero HIV transmission, among women on HAART with stillbirths in Botswana. PloS one 2012; 7(2): e31580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Botswana Ministry of Health. Handbook of the Botswana 2016 Integrated HIV Clinical Care Guidelines. 2016. http://apps.who.int/medicinedocs/documents/s22413en/s22413en.pdf (accessed Sep 13 2019).
- 21.Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American journal of epidemiology 2016; 183(8): 758–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zash R, Holmes L, Diseko M, et al. Neural-Tube Defects and Antiretroviral Treatment Regimens in Botswana. The New England journal of medicine 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Organization WH. Botswana: WHO statistical profile. 2015. http://www.who.int/gho/countries/bwa.pdf?ua=1&ua=1 (accessed 21 October 2019).
- 24.Botswana National HIV & AIDS Treatment Guidelines. 2012 Version. 1 April 2012. Edition. http://www.moh.gov.bw (accessed 8 April 2016).
- 25.Villar J, Cheikh Ismail L, Victora CG, et al. International standards for newborn weight, length, and head circumference by gestational age and sex: the Newborn Cross-Sectional Study of the INTERGROWTH-21st Project. Lancet (London, England) 2014; 384(9946): 857–68. [DOI] [PubMed] [Google Scholar]
- 26.Villar J, Giuliani F, Fenton TR, Ohuma EO, Ismail LC, Kennedy SH. INTERGROWTH-21st very preterm size at birth reference charts. Lancet (London, England) 2016; 387(10021): 844–5. [DOI] [PubMed] [Google Scholar]
- 27.Caniglia EC, Zash R, Jacobson DL, et al. Emulating a target trial of antiretroviral therapy regimens started before conception and risk of adverse birth outcomes. AIDS (London, England) 2018; 32(1): 113–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Caniglia EC, Zash R, Diseko M, et al. How much could anemia-related interventions reduce the HIV disparity in adverse birth outcomes? American journal of epidemiology 2025; 194(1): 122–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.