Abstract
Background
Several studies have assessed the association between antiretroviral (ARV) therapy use during pregnancy and small for gestational age (SGA), but the evidence remains incompletely elucidated.
Methods
We linked data from Tennessee Medicaid files and vital records to evaluate pregnancies among human immunodeficiency virus (HIV)-infected women who delivered between 1994 and 2009.Maternal HIV status was defined based on diagnosis codes, ARV prescriptions, and laboratory codes for CD4 count or HIV RNA assays. ARV use was identified from pharmacy claims. Risk of SGA (defined as birthweight below the 10th percentile for gestational age) and preterm birth were evaluated using logistic regression models.
Results
477 HIV- infected pregnant women contributing 604 singleton pregnancies were identified; 156 (26%) delivered SGA infants. ARV use during pregnancy was not associated with SGA (adjusted odds ratio [aOR] = 0.93; 95% confidence interval [CI], 0.56-1.56) or preterm birth (aOR = 0.74; 95% CI, 0.42-1.32). Exposure to a protease inhibitor (PI) during the first trimester was associated with a lower risk of SGA (OR, 0.54; 95% CI, 0.29-1.01) compared to non-exposure to a PI throughout pregnancy.
Conclusions
We observed no evidence of an association between ARV exposure during pregnancy and SGA delivery in this Medicaid cohort of HIV-infected women.
Keywords: HIV, antiretroviral, pregnancy, small-for-gestational age, Medicaid
The use of antiretroviral (ARV) therapy during pregnancy is the standard of care for human immunodeficiency virus (HIV)-infected pregnant women.1, 2 Some epidemiologic studies have indicated that maternal ARV use during pregnancy may be associated with significant increases in adverse pregnancy outcomes,3-10 including small for gestational age (SGA),7, 10 but the evidence remains inconclusive as other studies did not observe such associations.5, 11-15 The conflicting evidence regarding SGA may be due to differences in the definition of SGA or ARV comparison groups between studies; also, comparison of SGA risk by ARV drug class has not been adequately explored. Our study used a variety of exposure classifications, including classification by drug class, to evaluate the effect of ARV use during pregnancy on the risk of SGA in the newborn (using the standard definition of birth weight below the 10th percentile for gestational age) among HIV-infected pregnant women in the US.
Methods
Study cohort
From a source population of females aged 10-55 years who were enrolled in Tennessee Medicaid (TennCare) from January 1, 1994 through December 31, 2009, we identified a cohort of pregnant women with evidence of HIV infection, based on an algorithm that required codes for HIV diagnosis, HIV related laboratory testing (CD4 count or HIV viral load), or prescription dispensing for ARVs.16 Only HIV-infected pregnant women for whom Medicaid and vital records databases could be linked were included. Linkage of mothers to their infants was achieved by using an already established linkage algorithm for women enrolled in TennCare as described elsewhere.17
We restricted our study cohort to women who were continuously enrolled in Medicaid with drug coverage from at least 30 days before their last menstrual period (LMP) through the date of delivery or fetal death. This restriction was done in order to identify ARV prescriptions that were dispensed prior to the LMP and had a supply that extended into pregnancy in order to ensure complete follow-up of women and complete ascertainment of exposure information throughout pregnancy. We also required infants to have been enrolled in Medicaid within the first 30 days of life and for at least 90 days in order to ascertain adverse fetal outcomes. Since for some women the first diagnosis of HIV infection may occur only during pregnancy or after delivery, we further restricted the cohort to pregnancies that had evidence of HIV infection at any time from up to one year before the LMP through to the delivery date [Figure 1]. All singleton pregnancies that occurred during the study period and were recorded in TennCare were included in our final cohort. Our analysis was restricted to pregnancies for which there was a live birth and a recorded LMP on the birth certificate.
Figure 1. Cohort structure.
HIV = human immunodeficiency virus; LMP = last menstrual period; ARV = antiretroviral
aThese are pregnancies that were either (1) not enrolled with drug coverage from LMP-30 through delivery (N=364) or (2) had an HIV claim only prior to LMP-365 (N = 639) or (3) had an HIV claim only prior to LMP-365 and after delivery (N=26)
bIncludes women with the following indication(s) for HIV: (a) one (or more) inpatient diagnoses (prior to the birth hospitalization), or (b) two (or more) outpatient diagnoses separated by at least 30 days, or (c) two (or more) prescriptions for ARVs on different dates, or (d) any combination of two indications (one or more of each of the following indications – birth hospitalization diagnosis plus outpatient diagnosis, or birth hospitalization diagnosis plus prescription, or birth hospitalization diagnosis plus procedure, or outpatient diagnosis plus prescription, or outpatient diagnosis plus procedure, or prescription plus procedure)
Validation of HIV infection
To validate our definition of HIV-infected women from claims data, we reviewed a random sample of medical charts for the women identified as HIV-infected who were not exposed to any ARVs throughout pregnancy. We then calculated the positive predictive value of our definition based on these chart reviews to assess how well our algorithm for HIV correctly classified infected women.
Antiretroviral exposure
We identified prescriptions for ARVs from the women's Medicaid pharmacy claims. A woman was considered exposed to a specific ARV drug or ARV drug class during pregnancy if she had at least one prescription dispensed for that ARV from 30 days prior to the LMP through delivery; trimester exposures were also based on at least one prescription dispensed during the specific trimester. Exposure was classified in three different ways: (1) exposure to any ARV during pregnancy; (2) exposure to ARV treatment containing a protease inhibitor (PI), or a non-nucleoside reverse-transcriptase inhibitor (NNRTI), or a nucleoside reverse-transcriptase inhibitor (NRTI), categorized hierarchically for the entire pregnancy (regardless of any treatment switch during pregnancy); and (3) first exposure to a PI-containing treatment in the first, second, or third trimester, or no exposure to a PI-containing treatment during pregnancy.
Classification of small for gestational age (SGA)
Information on gestational age at birth and birthweight was obtained from the birth certificate files. The gestational age recorded in birth certificates is typically calculated from an algorithm that uses information on LMP (based on maternal recall), ultrasound and clinical assessment; a study that used data from the same population (TennCare, 1985-2002) and data system as the one used for our study showed a concordance of 94% between the date of the LMP in birth certificates and medical records;18 We determined the percentile for birthweight for each infant based on the gestational age using sex-specific references.19 SGA infants were defined as those with a birthweight below the 10th percentile for their gestational age.
Definition of study covariates
Information on potential risk factors for SGA (maternal age, race, education and parity; trimester in which prenatal care began; and alcohol, tobacco and illicit drug use during pregnancy) was obtained from birth certificate files; clinical information came from Medicaid inpatient and outpatient diagnosis claims, and information on non-ARV medications used during pregnancy was obtained from Medicaid pharmacy claims.
Statistical analysis
We compared the characteristics of women who were exposed versus those unexposed to ARVs during pregnancy using χ2 and Fisher's exact tests for categorical variables and the Wilcoxon rank sum test for continuous variables. Univariable logistic regressions were performed to identify potential risk factors for SGA. All variables with a 2-sided P-value <0.10 were included in the multivariable logistic regression models to estimate the effect of ARV use on the risk of SGA. To assess whether there were temporal changes in the effect of ARVs on SGA resulting from the evolving clinical management of HIV during pregnancy over the course of the study period (including the temporal availability of ARVs), we performed analyses stratified by four birth cohorts; the cutoffs for the birth cohorts were determined from observed patterns of ARV use in the Medicaid population nationwide.20 We found no significant differences in the effect estimates by birth cohort and therefore all study years were included in our analyses. Unadjusted and adjusted regression models were fit using generalized estimating equations (GEE) to account for clustering of multiple pregnancies by a woman during the study period. All final models were adjusted for the year of delivery and HIV-related maternal illnesses in order to minimize the potential impact of confounding by temporal trends in ARV use and maternal HIV disease stage. Similar logistic regression models were used in a secondary analysis to estimate the association of ARV use with the risk of preterm delivery (gestational age <37 completed weeks).
We also conducted a sensitivity analysis using an alternative definition of SGA to make a distinction between SGA infants who were preterm to those who were full term since the risk of morbidity and mortality may differ between these two groups. For the preterm infants we retained the same definition of SGA (birthweight below the 10th percentile for gestational age) while for full term infants SGA was redefined as birthweight < 2,500g. Statistical significance was defined as two-sided p-values<0.05. All analyses were performed using SAS (version 9.2; SAS Institute, Cary, NC).
Results
From 1994 to 2009, 3228 women in TennCare had evidence of HIV and at least one pregnancy; 2408 of them were enrolled in Medicaid with drug coverage from 30 days prior to the LMP through delivery and 2284 women (3,793 pregnancies) had their infants enrolled within the first 30 days of life and for at least 3 months. After excluding 1944 pregnancies in which women had their first recorded HIV diagnosis only after delivery and restricting our cohort to mothers with an indication of HIV or ARV from one year before the LMP through delivery, 820 pregnancies remained. A further exclusion of 25 multiple gestation pregnancies (3 of which included fetal deaths), and an additional 5 singleton fetal deaths left 790 pregnancies in our study cohort contributed by 623 women [Figure 1].
Among the 790 pregnancies, 151 (19.1%) were unexposed to ARVs from one year before the LMP through delivery. We reviewed medical records for a random sample of these 151 ARV unexposed pregnancies. Out of 31 medical charts reviewed, there were 14 from women with only 1 HIV claim in the dataset; we confirmed the HIV positive status for 3 women (positive predictive value: 21%); the other 11 pregnancies were HIV negative. Of 17 medical records with 2 or more HIV claims in the dataset, we confirmed the HIV positive status for 16 women (positive predictive value: 94%), and the one HIV negative woman had positive enzyme immunoassay (EIA) but no HIV RNA detected in plasma. Therefore, we excluded women with only one HIV claim from our cohort of HIV positive women, leaving 597 women contributing 764 singleton pregnancies that resulted in live births. Finally, we excluded pregnancies with missing information on the LMP date (N = 157) and gestational age (N = 3) since our primary outcome of interest was SGA.
The characteristics of the final cohort of 477 HIV- infected pregnant women contributing 604 pregnancies are shown on the Table1. A total of 93 pregnancies (15%) were unexposed to ARVs throughout pregnancy. Compared to ARV exposed pregnancies, those unexposed had a higher proportion of women with late (third trimester) or absent prenatal care (24% vs. 6%); overall prevalence of absent or late access to prenatal care was approximately 9% before 2002, and remained at about 8% by 2009 [Figure 2].There was a slight decrease over time in the prevalence of ARV unexposed pregnancies (19%, before 2002; 16%, 2002 – 2004; 13%, 2005 – 2007; 12%, 2008 – 2009). Among the 511 (85%) pregnancies exposed to ARVs, 222 (37%) were exposed to a PI-containing treatment, 78 (13%) to NNRTI-containing treatment, 109 (18%) to NRTI dual or triple therapy, and 102 (17%) to NRTI monotherapy [Table 2].
Table 1. Characteristics of HIV-Infected Pregnant Women Enrolled in TennCare between 1994 and 2009 (N = 604 Pregnancies).
| Characteristic | Women with ARV prescriptions during pregnancy (N = 511) | Women without ARV prescriptions during pregnancy (N = 93) |
|---|---|---|
| Year of delivery | ||
| 1994 – 1998 | 124 (24) | 30 (32) |
| 1999 – 2002 | 167 (33) | 31 (33) |
| 2003 – 2006 | 130 (25) | 20 (22) |
| 2007 – 2009 | 90 (18) | 12 (13) |
| Age, median (range) | 26 (14 – 43) | 25 (17 – 40) |
| Race, no. (%) | ||
| Black | 412 (81) | 77 (83) |
| White | 99 (19) | 16 (17) |
| Years of Education ≤ 12 years, no. (%) | 427 (84) | 80 (86) |
| Parity, mean (25th – 75th) | 2 (1 – 3) | 2 (1 – 4) |
| Trimester prenatal care began, no. (%) | ||
| 1st | 304 (61) | 46 (52) |
| 2nd | 162 (32) | 22 (25) |
| 3rd | 22 (4) | 7 (8) |
| None | 11 (2) | 14 (16) |
| Missing | 12 | 4 |
| Number of prenatal visits among those with at least one visit, median (range) | 10 (1 – 33) | 10 (2 – 18) |
| Hypertension diagnosis, no. (%) | 56 (11) | 12 (13) |
| Mental health disorder diagnosis, no. (%) | 57 (11) | 16 (17) |
| Substance abuse diagnosis, no. (%) | 22 (4) | 3 (3) |
| Other chronic health diagnosesa, no. (%) | 56 (11) | 13 (14) |
| Maternal Illnesses, no (%) | ||
| Possibly HIV-relatedb | 6 (1) | 1 (1) |
| Other c | 237 (46) | 13 (14) |
| None | 268 (52) | 79 (85) |
| AIDS diagnosis, no. (%) | 279 (54) | 16 (17) |
| Non-ARV medication use during pregnancy d, no (%) | 209 (41) | 15 (16) |
| Alcohol use, no. (%) | ||
| Yes | 22 (7) | 4 (6) |
| No | 306 (93) | 58 (94) |
| Tobacco use, no. (%) | ||
| Yes | 124 (24) | 27 (29) |
| No | 387 (76) | 66 (71) |
| Illicit drug use, no. (%) | ||
| Yes | 34 (10) | 9 (15) |
| No | 295 (90) | 53 (85) |
HIV = human immunodeficiency virus; ARV = antiretroviral
epilepsy, sickle cell disease, asthma, renal disease, neoplastic disease, cardiac disease, cystic fibrosis, alcohol abuse, cerebrovascular disease, obesity, migraine, sexually transmitted infection, or hepatitis
pneumocystis pneumonia, tuberculosis, kaposis sarcoma, cytomegalovirus infection, toxoplasmosis, HIV encephalopathy, cryptosporidiosis, isosporiasis, histoplamosis, coccidiomycosis, or lymphoma.
weight loss, mycobacterial infection, nephropathy, cardiomyopathy, or diarrhea.
No ARV fills – Anti-hypertensives, 3 (3%); Mental health disorder, 5 (5%); Anti-infectives, 3 (3%); Asthma, 3 (3%); ARV fills – Anti-hypertensives: 38 (7%); Mental health disorder: 88 (17%); Anti-infectives: 76 (15%); Asthma: 69 (13%).
Note: Information on alcohol and illicit drug use was missing for >30% pregnancies, 214 missing observations for alcohol and 213 missing observations for illicit drug use.
Figure 2. Trends over time - Late access to prenatal care and SGA among HIV-infected pregnant women in Tennessee Medicaid: 1994 – 2009.
SGA: small-for-gestational age (defined as birthweight weight below the 10th percentile for the gestational age).
The number of pregnancies for each birth cohort is as follows: <2002 = 154; 2002 – 2004 = 198; 2005 – 2007 = 150; 2008 – 2009 = 102
Note: Late access to prenatal care was defined as starting prenatal care in the 3rd trimester or not having any prenatal care throughout pregnancy. A slight increase in SGA births among those with late access to prenatal care over time was observed (<2002, 5.8%; 2002 – 2004, 7.8%; 2005 – 2007, 9.7%; 2008 – 2009, 12%)
Table 2. Pregnancy Outcomes by Type of ARV Prescriptions Received During Pregnancy (N=604 Pregnancies).
| Outcome, no. (%) | ARV Prescriptions containing a: | |||
|---|---|---|---|---|
|
| ||||
| PI a (n = 222) | NNRTI, b no PI (n = 78) | NRTI monotherapy (n = 102) | NRTI dual/triple therapy (n = 109) | |
|
| ||||
| Birth weight | ||||
| <1500 | 6 (3) | 1 (1) | 1 (1) | 2 (2) |
| 1500 – 2499 | 43 (19) | 25 (32) | 12 (12) | 24 (22) |
| 2500 – 4000 | 171 (77) | 52 (67) | 86 (84) | 80 (73) |
| >4000 | 2 (1) | 0 (0) | 3 (3) | 3 (3) |
|
| ||||
| Gestational age at delivery c | ||||
| <32 weeks | 9 (4) | 3 (4) | 0 (0) | 3 (3) |
| 32 – 36 weeks | 41 (18) | 14 (18) | 12 (12) | 21 (19) |
| =>37 weeks | 172 (77) | 61 (78) | 90 (88) | 85 (78) |
|
| ||||
| SGA d | 55 (25) | 28 (36) | 23 (23) | 27 (25) |
|
| ||||
| SGA by alternative definition e | 29 (13) | 20 (26) | 8 (8) | 16 (15) |
|
| ||||
| C-section delivery | 132 (59) | 49 (63) | 21 (21) | 46 (42) |
|
| ||||
| NICU admission | 25 (11) | 6 (8) | 9 (9) | 11 (10) |
ARV = antiretroviral; NICU = neonatal intensive care unit; SGA = small-for gestational age; NRTI = nucleoside reverse transcriptase inhibitor; NNRTI = non-nucleoside reverse transcriptase inhibitor; PI = protease inhibitor
NRTI + PI (n = 191); NRTI + NNRTI + PI (n = 29; ; note these are possibly women who switched from a NNRTI-based regimen to an PI-based regimen or vice versa); PI only (2)
NRTI + NNRTI (n = 78)
Gestational age at delivery based on completed weeks.
SGA defined as birthweight weight below the 10th percentile for the gestational age.
SGA defined as birth weight below the 10th percentile for gestational age for preterm births, and birth weight <2500 g for full term births (37 weeks or later).
There were 156 (26%) deliveries of SGA infants. The prevalence, which remained above 20% throughout the study years [Figure 2], was 26% for ARV-exposed and 25% for ARV-unexposed pregnancies; those exposed to NNRTI-containing treatment had significantly higher prevalence of SGA births (36%) compared to all other groups [Table 2]. Maternal characteristics associated with SGA were prescription of medications other than ARVs during pregnancy, smoking, alcohol and drug use. After adjusting for these variables, and for the year of delivery and HIV-related maternal illnesses, we found no association between any ARV exposure during pregnancy and the risk of SGA (aOR, 0.93; 95% CI, 0.56-1.56) [Table 3]. Similarly, for ARV exposure by drug class, we observed no statistically significant associations between use of ARV regimens containing a PI, a NNRTI, or only NRTIs (one, two or three NRTIs) and the risk of SGA. However, pregnancies that were exposed to ARV treatment containing a PI in the first trimester compared to those with no PI throughout pregnancy had marginally lower risk of SGA (aOR, 0.54; 95% CI, 0.29-1.01).
Table 3. Risk of SGA Comparing Different ARV Prescriptions, Tennessee Medicaid, 1994 - 2009.
| Exposure Group | N and Percent with prescription | Crude Models OR (95% CI) | P | Adjusted Models 1a OR (95% CI) | P | Adjusted Models 2b OR (95% CI) | P |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Any ARV vs. No ARV | 511 (85) | 1.10 (0.68, 1.78) | 0.70 | 1.04 (0.63, 1.72) | 0.87 | 0.93 (0.56, 1.56) | 0.79 |
|
| |||||||
| Prescriptions containing: | |||||||
| No ARV | 93 (15) | Ref | --- | ref | --- | ref | --- |
| PIc | 222 (37) | 0.99 (0.58, 1.69) | 0.98 | 0.88 (0.50, 1.54) | 0.65 | 0.74 (0.42, 1.32) | 0.31 |
| NNRTId | 78 (13) | 1.60 (0.86, 2.97) | 0.14 | 1.47 (0.78, 2.79) | 0.23 | 1.21 (0.63, 2.33) | 0.57 |
| NRTIdual/triple therapy | 109 (18) | 1.08 (0.59, 1.96) | 0.80 | 0.99 (0.53, 1.83) | 0.97 | 0.88 (0.47, 1.66) | 0.69 |
| NRTI monotherapy | 102 (17) | 0.99 (0.52, 1.88) | 0.97 | 1.14 (0.58, 2.26) | 0.69 | 1.19 (0.55, 2.57) | 0.65 |
|
| |||||||
| First PI-containing prescription in: | |||||||
| 1st Trimester | 71 (12) | 0.67 (0.36, 1.24) | 0.20 | 0.60 (0.33, 1.12) | 0.11 | 0.54 (0.29, 1.01) | 0.05 |
| 2nd Trimester | 96 (16) | 1.05 (0.64, 1.70) | 0.86 | 0.97 (0.57, 1.65) | 0.91 | 0.86 (0.49, 1.50) | 0.59 |
| 3rd Trimester | 55 (9) | 0.88 (0.46, 1.68) | 0.70 | 0.74 (0.37, 1.47) | 0.39 | 0.74 (0.37, 1.46) | 0.38 |
| No PI | 382 (63) | ref | --- | ref | --- | ref | --- |
SGA = small-for gestational age; ARV = antiretroviral; NRTI = nucleoside reverse transcriptase inhibitor; NNRTI = non-nucleoside reverse transcriptase inhibitor; PI = protease inhibitor; OR = odds ratio; CI = confidence interval
Adjusted for non-antiretroviral medication use during pregnancy (binary), smoking (binary) and alcohol use (binary).
Adjusted for all variables included in Model 1 plus birth year cohort (categorical – 4 levels) and HIV-related maternal illnesses during pregnancy (categorical - 3 levels)
NRTI + PI (n = 191); NRTI + NNRTI + PI (n = 29; note these are possibly women who switched from a NNRTI-based regimen to an PI-based regimen or vice versa); PI only (2)
NRTI + NNRTI (n = 78)
Note: Drug use was significant in the univariable analysis, but was highly correlated with alcohol use in the multivariable model; therefore we didn't adjust for it in the multivariable model.
The prevalence of SGA was sensitive to the definition used. When we used an alternative definition of SGA for full term births (defined as birthweight < 2,500g) the prevalence of SGA decreased to 14%. However, results for the association between ARV exposure and SGA using this alternative definition were similar to those observed using the primary definition of SGA. For our secondary analysis assessing the association between ARV exposure and the risk of preterm delivery, we observed an overall preterm delivery prevalence of 21%; 20% among ARV-exposed pregnancies and 29% among unexposed [Table 2]. The adjusted OR for the association between any ARV exposure during pregnancy and the risk of preterm birth was 0.74 (95% CI, 0.42-1.32).
Discussion
In this cohort of 477 HIV-infected women who were enrolled in Tennessee Medicaid and had a total of 604 pregnancies between 1994 and 2009, we found no association between ARV exposure and the risk of both SGA and preterm delivery. However, women who were exposed to a PI-containing treatment in the first trimester had a marginally significant lower risk of SGA compared to those who were unexposed to a PI throughout pregnancy.
The prevalence of SGA infants was high in our study population (26%) compared to the US general population (11%),21 and to studies conducted in France (4%),22 in Botswana (18%),10 and in other US populations (16% and 7%, respectively).14, 23 These differences are most likely due to the different definitions of SGA and the reference populations used. We used US sex-specific references to define SGA as infants below the 10th percentile for their gestational age at birth. This resulted in 71 full term infants (gestational age ≥ 37 weeks) with a birthweight > 2500g being classified as SGA, comprising 46% of all SGA infants in our cohort. Our observed slight increase in the prevalence of SGA infants during the study period may be a reflection of changes in clinical management decisions over time (such as performing an indicated preterm delivery, induction, or cesarean delivery) based on the presence of fetal growth restriction and subsequent SGA.
Unlike previous studies, our main reference group was HIV-infected women who were unexposed to ARVs throughout pregnancy. Overall, we observed a similar prevalence of SGA for ARV exposed women compared to those who were unexposed. The main differences were among the specific drug classes, with the highest SGA prevalence among women who were exposed to NNRTIs.
The high proportion of HIV-infected pregnant women who were not exposed to ARVs throughout pregnancy (15%) is a notable finding of our study. This could be a result of women entering prenatal care late (3rd trimester) or not receiving prenatal care throughout pregnancy as seen in the Table 1. To verify that we did not miss some ARV information in the pharmacy claims data, we searched for any mention of ARV use in the random sample of the 31 medical charts reviewed and found that while indeed some of the women did not receive any ARV during pregnancy, administration of prophylactic therapy during labor and delivery to avoid maternal-fetal transmission was recorded. The chart reviews and medical profiles further revealed a high proportion of mental health disorders, drug abuse, incarceration, and other social problems in this population which could act as barriers to treatment. We note that the high occurrence of social problems that we observed from the chart reviews could have been accentuated by our study design since we restricted the cohort to women eligible for Medicaid before LMP. That is, to be eligible, these women had to be either multiparous or have disabilities. Our methods may overestimate ARV exposure during pregnancy if HIV-infected women without Medicaid coverage during their pregnancy have worse access to ARV medications. On the other hand, our methods may underestimate ARV exposure during pregnancy if those on Medicaid before pregnancy are the most disadvantaged populations and have worse access (including poor adherence) to prenatal care and ARVs. This would imply that if ARVs increased the risk of SGA as shown in a few studies,7, 10 our results would be biased towards the null as the exposure group in our analysis would be misclassified by including those with poor adherence as being exposed; poor ARV adherence during pregnancy has been shown to be high among adolescents, substance users, and women with mental illnesses.24-27 In addition, changes in Medicaid eligibility and enrollment over time may have also led to different populations in the context of HIV severity and obstetric interventions, a phenomenon we accounted for in all our analyses by adjusting for calendar year of birth.
Our study had several limitations. First, our exposure measure only considered when prescriptions were dispensed and not necessarily when the women took the medications which could lead to exposure misclassification. Second, we were unable to assess the effect of specific combination regimens, which is the standard of care for HIV infected pregnant women in resource-rich nations and often the main exposure of interest in studies assessing the effect of ARV use on birth outcomes. This is because we could not ascertain that prescriptions for individual ARVs were taken concomitantly, but nonetheless our analysis using any ARV exposure provides a valid statistical test for the null hypothesis that no combination therapy has an effect on SGA. Third, since we only used a proxy for HIV severity and could not directly adjust for the women's CD4 cell counts and viral loads, we cannot rule out potential confounding by indication. Furthermore, we did not have information on the HIV status of the infant at birth which has been shown to be associated with SGA and preterm delivery.28, 29 However, both maternal HIV severity and fetal HIV acquisition might be intermediates on the pathway between ARV exposure and SGA, in which case adjusting for them in the analyses would require advanced methods and more granular information of the intermediates.30 Also, greater than 30% pregnancies had missing information on illicit drug and alcohol use, both of which are major risk factors for SGA and may be important confounders of the association between exposure to ARVs and SGA. Furthermore, while our sample size was small to detect smaller yet clinically significant increases in risk of SGA, we did have the power to detect an increase in SGA from 25% among ARV-unexposed to 40% (OR = 2.0) in women who were exposed to ARVs during pregnancy. Finally, since women may only become eligible for Medicaid as a result of their pregnancy, and many women only find out they are HIV-infected when they become pregnant, our eligibility criteria requiring continuous enrolment in Medicaid from 3 months prior to LMP through delivery oversampled multiparous women or those with disabilities. In addition, excluding pregnancies with missing information on the LMP date and gestational age may have resulted in a biased sample; however, we would expect this to be the case only if those excluded were systematically different from those included in ways that correlate with both ARV exposure and unmeasured risk factors for SGA.
A major practical strength of our study is the use of a valid, existing database over a long period of time (1994-2009). Our cohort represents a population that is disproportionately affected by HIV infection compared to women of other races/ethnicities31 and therefore might be closer to the real population of ARV users or those in need of ARVs. The characteristics of the cohort minimize the possibility of missing data since Medicaid enrollees are unlikely to have other resources to pay for their medical care and thus virtually all treatment prescriptions, lab measurements, and health outcomes are captured in the data files. Generalizability to populations in developing countries and also to populations not in the low income category in developed countries, however, will have to be made with caution. The conflicting results from previous studies7, 10, 14, 22, 23 possibly reflect geographic differences in the populations studied including different baseline risks, availability of obstetrical care, concomitant morbidities and treatments, and confounding by indication for ARV use. Our findings of no effect of ARV use on the risk of SGA are concordant with results from studies conducted in the U.S.14, 23
In conclusion, we found no association between ARV exposure and both SGA and preterm delivery. Our results showed evidence of a high proportion of HIV-infected pregnant women who did not use any ARVs throughout pregnancy despite 77% of them initiating prenatal care in the first or second trimester. This has important implications for appropriate perinatal and maternal HIV treatment during pregnancy, suggesting greater prioritization of obstetric and gynecologic services in Medicaid recipients of childbearing age.
Acknowledgments
We gratefully acknowledge the contribution of our research coordinator, Shannon Stratton, together with our study nurses, Patricia Gideon, RN; Leanne Balmer, MSN, RN; and Michelle DeRanieri, RN, MSN, as well as the nurses who work at the individual hospitals where the women in our study delivered their babies. We are also grateful to the Tennessee Department of Health and the TennCare Bureau for permitting the use of this data, as well as the institutional review boards at the Harvard School of Public Health and Vanderbilt University.
Source of Funding: This work was supported by the Eunice Kennedy Shriver National Institute of Child and Human Development (NICHD) [R01HD056940-01].
Footnotes
Conflicts of Interest: SHD has consulted for Novartis, AstraZeneca and GSK. All other authors do not have any commercial or other association that might pose a conflict of interest.
Meetings where the information has been presented: None
References
- 1.World Health Organization. Antiretroviral drugs for treating pregnant women and preventing HIV infection in infants. 2010 http://www.who.int/hiv/pub/mtct/antiretroviral2010/en/ [PubMed]
- 2.Panel on treatment of HIV-infected pregnant women and prevention of perinatal transmission. Recommendations for use of antiretroviral drugs in pregnant HIV-1 infected women for maternal health and interventions to reduce perinatal HIV transmission in the United States. [Accessed January 10, 2014]; http://aidsinfo.nih.gov/contentfiles/lvguidelines/perinatalgl.pdf.
- 3.Lorenzi P, Spicher VM, Laubereau B, Hirschel B, Kind C, Rudin C, et al. Antiretroviral therapies in pregnancy: Maternal, fetal and neonatal effects. Swiss HIV cohort study, the Swiss collaborative HIV and pregnancy study, and the Swiss neonatal HIV study. AIDS. 1998 Dec 24;12(18):F241–7. doi: 10.1097/00002030-199818000-00002. [DOI] [PubMed] [Google Scholar]
- 4.European Collaborative Study, Swiss Mother and Child HIV Cohort Study. Combination antiretroviral therapy and duration of pregnancy. AIDS. 2000 Dec 22;14(18):2913–20. doi: 10.1097/00002030-200012220-00013. [DOI] [PubMed] [Google Scholar]
- 5.European Collaborative Study. Exposure to antiretroviral therapy in utero or early life: The health of uninfected children born to HIV-infected women. J Acquir Immune Defic Syndr. 2003 Apr 1;32(4):380–7. doi: 10.1097/00126334-200304010-00006. [DOI] [PubMed] [Google Scholar]
- 6.Thorne C, Patel D, Newell ML. Increased risk of adverse pregnancy outcomes in HIV-infected women treated with highly active antiretroviral therapy in Europe. AIDS. 2004 Nov 19;18(17):2337–9. doi: 10.1097/00002030-200411190-00019. [DOI] [PubMed] [Google Scholar]
- 7.Townsend CL, Cortina-Borja M, Peckham CS, Tookey PA. Antiretroviral therapy and premature delivery in diagnosed HIV-infected women in the United Kingdom and Ireland. AIDS. 2007 May 11;21(8):1019–26. doi: 10.1097/QAD.0b013e328133884b. [DOI] [PubMed] [Google Scholar]
- 8.Machado ES, Hofer CB, Costa TT, Nogueira SA, Oliveira RH, Abreu TF, et al. Pregnancy outcome in women infected with HIV-1 receiving combination antiretroviral therapy before versus after conception. Sex Transm Infect. 2009 Apr;85(2):82–7. doi: 10.1136/sti.2008.032300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Powis KM, Kitch D, Ogwu A, Hughes MD, Lockman S, Leidner J, 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 Aug 15;204(4):506–14. doi: 10.1093/infdis/jir307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chen JY, Ribaudo HJ, Souda S, Parekh N, Ogwu A, Lockman S, et al. Highly active antiretroviral therapy and adverse birth outcomes among HIV-infected women in Botswana. J Infect Dis. 2012 Dec 1;206(11):1695–705. doi: 10.1093/infdis/jis553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tuomala RE, Shapiro DE, Mofenson LM, Bryson Y, Culnane M, Hughes MD, et al. Antiretroviral therapy during pregnancy and the risk of an adverse outcome. N Engl J Med. 2002 Jun 13;346(24):1863–70. doi: 10.1056/NEJMoa991159. [DOI] [PubMed] [Google Scholar]
- 12.Tuomala RE, Watts DH, Li D, Vajaranant M, Pitt J, Hammill H, et al. Improved obstetric outcomes and few maternal toxicities are associated with antiretroviral therapy, including highly active antiretroviral therapy during pregnancy. J Acquir Immune Defic Syndr. 2005 Apr 1;38(4):449–73. doi: 10.1097/01.qai.0000139398.38236.4d. [DOI] [PubMed] [Google Scholar]
- 13.Morris AB, Dobles AR, Cu-Uvin S, Zorrilla C, Anderson J, Harwell JI, et al. Protease inhibitor use in 233 pregnancies. J Acquir Immune Defic Syndr. 2005 Sep 1;40(1):30–3. doi: 10.1097/01.qai.0000174651.40782.95. [DOI] [PubMed] [Google Scholar]
- 14.Cotter AM, Garcia AG, Duthely ML, Luke B, O'Sullivan MJ. Is antiretroviral therapy during pregnancy associated with an increased risk of preterm delivery, low birth weight, or stillbirth? J Infect Dis. 2006 May 1;193(9):1195–201. doi: 10.1086/503045. [DOI] [PubMed] [Google Scholar]
- 15.Patel K, Shapiro DE, Brogly SB, Livingston EG, Stek AM, Bardeguez AD, et al. Prenatal protease inhibitor use and risk of preterm birth among HIV-infected women initiating antiretroviral drugs during pregnancy. J Infect Dis. 2010 Apr 1;201(7):1035–44. doi: 10.1086/651232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Phiri K, Hernandez-Diaz S, Dugan KB, Williams PL, Dudley JA, Jules A, et al. First trimester exposure to antiretroviral therapy and risk of birth defects. Pediatr Infect Dis J. 2014 Jul;33(7):741–6. doi: 10.1097/INF.0000000000000251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Piper JM, Ray WA, Griffin MR, Fought R, Daughtery JR, Mitchel E., Jr Methodological issues in evaluating expanded Medicaid coverage for pregnant women. Am J Epidemiol. 1990 Sep;132(3):561–71. doi: 10.1093/oxfordjournals.aje.a115692. [DOI] [PubMed] [Google Scholar]
- 18.Cooper WO, Hernandez-Diaz S, Gideon P, Dyer SM, Hall K, Dudley J, et al. Positive predictive value of computerized records for major congenital malformations. Pharmacoepidemiol Drug Saf. 2008 May;17(5):455–60. doi: 10.1002/pds.1534. [DOI] [PubMed] [Google Scholar]
- 19.Oken E, Kleinman KP, Rich-Edwards J, Gillman MW. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr. 2003 Jul 8;3:6. doi: 10.1186/1471-2431-3-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Phiri K, Fischer MA, Mogun H, Williams PL, Palmsten K, Seage GR, III, et al. Trends in antiretroviral drug use during pregnancy among HIV-infected women on Medicaid: 2000 - 2007. AIDS patient care and STDs. 2014 doi: 10.1089/apc.2013.0165. forthcoming. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.National vital statistics system. [Accessed December 11, 2012];Birth Data. http://www.cdc.gov/nchs/births.htm.
- 22.Briand N, Mandelbrot L, Le Chenadec J, Tubiana R, Teglas JP, Faye A, et al. No relation between in-utero exposure to HAART and intrauterine growth retardation. AIDS. 2009 Jun 19;23(10):1235–43. doi: 10.1097/QAD.0b013e32832be0df. [DOI] [PubMed] [Google Scholar]
- 23.Watts DH, Williams PL, Kacanek D, Griner R, Rich K, Hazra R, et al. Combination antiretroviral use and preterm birth. J Infect Dis. 2013 Feb 15;207(4):612–21. doi: 10.1093/infdis/jis728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bardeguez AD, Lindsey JC, Shannon M, Tuomala RE, Cohn SE, Smith E, et al. Adherence to antiretrovirals among US women during and after pregnancy. J Acquir Immune Defic Syndr. 2008 Aug 1;48(4):408–17. doi: 10.1097/QAI.0b013e31817bbe80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kapetanovic S, Christensen S, Karim R, Lin F, Mack WJ, Operskalski E, et al. Correlates of perinatal depression in HIV-infected women. AIDS Patient Care STDS. 2009 Feb;23(2):101–8. doi: 10.1089/apc.2008.0125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kreitchmann R, Harris DR, Kakehasi F, Haberer JE, Cahn P, Losso M, et al. Antiretroviral adherence during pregnancy and postpartum in Latin America. AIDS Patient Care STDS. 2012 Aug;26(8):486–95. doi: 10.1089/apc.2012.0013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nachega JB, Uthman OA, Anderson J, Peltzer K, Wampold S, Cotton MF, et al. Adherence to antiretroviral therapy during and after pregnancy in low-income, middle-income, and high-income countries: A systematic review and meta-analysis. AIDS. 2012 Oct 23;26(16):2039–52. doi: 10.1097/QAD.0b013e328359590f. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Dreyfuss ML, Msamanga GI, Spiegelman D, Hunter DJ, Urassa EJ, Hertzmark E, et al. Determinants of low birth weight among HIV-infected pregnant women in Tanzania. Am J Clin Nutr. 2001 Dec;74(6):814–26. doi: 10.1093/ajcn/74.6.814. [DOI] [PubMed] [Google Scholar]
- 29.Ramokolo V, Lombard C, Fadnes LT, Doherty T, Jackson DJ, Goga AE, et al. HIV infection, viral load, low birth weight, and nevirapine are independent influences on growth velocity in HIV-exposed South African infants. J Nutr. 2014 Jan;144(1):42–8. doi: 10.3945/jn.113.178616. [DOI] [PubMed] [Google Scholar]
- 30.VanderWeele T, Vansteelandt S. Conceptual issues concerning mediation, interventions and composition. Statistics and Its Interface. 2009;2:457–468. [Google Scholar]
- 31.Centers for disease control and prevention. HIV/AIDS. HIV among women fact sheet. [Accessed July 11, 2014]; http://www.cdc.gov/hiv/risk/gender/women/facts/