Angiotensin-converting enzyme (ACE) inhibitors are commonly used antihypertensive medications, particularly in patients with diabetes or renal dysfunction. A recent analysis of the National Health and Nutrition Examination Survey suggested that approximately 40% of women of reproductive age using antihypertensive medications take ACE inhibitors.1 Because of this, it is also a relatively common first-trimester exposure, accounting for 10–20% of all antihypertensive exposures during this part of pregnancy.2,3
Although ACE inhibitors are clearly contraindicated in the second and third trimester as a result of a well-recognized fetopathy,4–6 the risks of first-trimester exposure are more poorly defined. A strong association between first-trimester ACE inhibitors exposure and major cardiovascular and neurologic malformations was described in an analysis of Tennessee Medicaid data,7 but other studies suggest that this association may be confounded by the indication of hypertension and associated comorbidities such as diabetes.8–11 Data on the teratogenic potential of ACE inhibitors are therefore conflicting, leading to controversy and confusion among physicians and patients regarding the risks of using these drugs in women of reproductive age. The 2013 report from the American College of Obstetricians and Gynecologists' Task Force on Hypertension in Pregnancy recommends not using ACE inhibitors in women of reproductive age “unless there is a compelling reason, such as the presence of proteinuric renal disease.”12 Resolution of this controversy with large and carefully controlled studies is needed, because evidence of teratogenicity not only informs counseling of patients who are exposed in early pregnancy, but also is a major determinant of whether these medications are appropriate to use in women who may inadvertently become pregnant. We therefore sought to examine the association between first-trimester ACE inhibitor exposure and the risk of major congenital malformations, with careful attention to confounding conditions, using a large, nationwide cohort of pregnancies linked to infants in Medicaid beneficiaries.
MATERIALS AND METHODS
Study data were drawn from the Medicaid Analytic eXtract. Medicaid is a joint state–federal health insurance program for people who have a low income. It provided coverage for approximately 40% of births in the United States annually during the study period.13 Medicaid Analytic eXtract is a database that contains the health care utilization claims for Medicaid beneficiaries including all diagnoses and procedures associated with inpatient or outpatient health care encounters. It also contains data on beneficiaries' enrollment information including demographic characteristics. Finally, it includes claims for all dispensed outpatient prescription medications.
The Partners Human Research Committee approved the use of this database for research. Using Medicaid Analytic eXtract claims from 46 states and the District of Columbia from 2000 to 2010, our group created a pregnancy cohort for pharmacoepidemiologic studies, as described by Palmsten et al.14 To accomplish this, we first identified females aged 12–55 years who delivered liveborn neonates and then linked these women with their offspring using a Medicaid identifier that is shared by families. The last menstrual period was estimated for pregnancies in the cohort using a validated algorithm based on the date of delivery and information on the length of gestation in the maternal and infant records.15 The analysis was restricted to pregnancies in which women were eligible for Medicaid from 3 months before the last menstrual period through 1 month postpartum. Pregnancies in which women had restricted benefits, private insurance, or certain capitated managed care programs were excluded because the claims for such patients may be incomplete in Medicaid Analytic eXtract. We required that infants be eligible for Medicaid for at least 3 months, unless they died, in which case a shorter eligibility period was permitted. We excluded pregnancies exposed to known teratogens during the first trimester including warfarin, antineoplastic agents, lithium, isotretinoin, misoprostol, and thalidomide or in which the infant had a chromosomal abnormality. The Medicaid Analytic eXtract pregnancy cohort has been used extensively for studies of drug safety during pregnancy.16–19
For the primary analysis, ACE inhibitor exposure was defined based on a claim for a dispensed outpatient medication from the last menstrual period to day 90 of pregnancy, corresponding to the end of the first trimester. The ACE inhibitors considered in the analysis included benazepril, captopril, enalapril, fosinopril, lisinopril, moexipril, perindopril, quinapril, ramipril, and trandolapril (Appendix 1, available online at http://links.lww.com/AOG/A895). Angiotensin-converting enzyme inhibitor monotherapy as well as combinations of ACE inhibitors and other antihypertensive medications were included. The reference group consisted of women not dispensed an ACE inhibitor during the first trimester. Women exposed to antihypertensives other than ACE inhibitors during the first trimester were excluded from the reference group, because some antihypertensives, for example, β-blockers,20 may be associated with an increased risk of malformations. Women who were dispensed ACE inhibitors in the 3 months before pregnancy but not during the first trimester were also excluded to avoid exposure misclassification in the reference group.
The primary study outcomes were 1) overall major congenital malformations, 2) cardiac malformations, and 3) central nervous system (CNS) malformations. Cardiac and CNS malformations have been specifically associated with ACE inhibitors.7 Malformations were defined based on the presence of diagnostic codes from the International Classification of Diseases, 9th Revision (Appendix 2, available online at http://links.lww.com/AOG/A895) recorded on 2 or more days in the infant inpatient or outpatient records or on 1 or more days if the infant died or underwent a corrective surgical procedure. Because conditions present in the infant are sometimes recorded in the mother's claims in Medicaid Analytic eXtract data, we also identified infant malformations in the maternal record using the same approach, taking care to exclude congenital malformations that were present in the mother. This approach has been shown to identify major congenital malformations (in a validation study of cardiac malformations) with a high positive predictive value.21
Four groups of potential confounders were selected for the analysis either because they represent known risk factors for congenital malformations or because they may represent proxies for such risk factors. These included maternal demographic characteristics, maternal medical or obstetric conditions, maternal medication exposures, and measures of health care utilization. Demographic characteristics assessed included maternal age, race and ethnicity, Medicaid eligibility type, and year of delivery. Maternal medical or obstetric conditions were assessed during the 3 months before pregnancy until the end of the first trimester and included chronic hypertension, diabetes, dyslipidemia, congestive heart failure, ischemic heart disease, renal disease, overweight or obesity, illicit drug or alcohol abuse, tobacco use, and multiple gestations. The codes used to define chronic hypertension are shown in Appendix 3, available online at http://links.lww.com/AOG/A895. The Obstetric Comorbidity Index, which is designed to summarize the burden of comorbid illness in pregnant women, was calculated for each woman in the cohort.22,23 Maternal medication exposure to other potentially teratogenic medications during the first trimester was also identified, including corticosteroids, danazol, fluconazole, methimazole, propylthiouracil, and synthetic progestins. Additionally, we ascertained exposure to insulin and noninsulin diabetes medications during the 3 months before pregnancy until the end of the first trimester as a marker for diabetes severity. Finally, we defined a number of measures of health care utilization during the 3 months before the last menstrual period, which may be markers of general comorbidity or access to health care services. These included whether the woman was hospitalized (including the number of days in the hospital), the number of emergency department and outpatient visits, the number of distinct diagnoses reported in the claims, and the number of non–ACE inhibitor prescription medications used.
We defined the absolute risk of overall, cardiac, and CNS malformations in ACE inhibitor–exposed and ACE inhibitor–unexposed pregnancies in the full cohort and calculated unadjusted relative risks (RRs) and 95% confidence intervals (CIs) for each of these outcomes. Then, because the indication of hypertension was expected to be an important confounder in this analysis, we restricted the cohort to women with a diagnosis of chronic hypertension recorded from 3 months before the last menstrual period until the end of the first trimester.
In the chronic hypertension-restricted cohort, we compared baseline characteristics in those who were exposed and unexposed to ACE inhibitors during the first trimester. We then determined the absolute risks of the malformations of interest and calculated RR and 95% CI. In the next step of the analysis, because diabetes is highly prevalent among ACE inhibitor–exposed pregnancies and is a strong risk factor for congenital malformations,24 we estimated the association between ACE inhibitors and the malformations of interest adjusting for this covariate.
Finally, to fully account for all measured confounders of the association between ACE inhibitors and malformations, we used a propensity score–based approach. The propensity score was defined using a logistic regression model that estimated the probability of being dispensed an ACE inhibitor during the first trimester based on all of the covariates specified (including diabetes), without further selection, in the hypertension-restricted cohort. We created 50 strata based on the distribution of the propensity score in the exposed pregnancies. We trimmed observations from nonoverlapping regions of the propensity score in the exposed and unexposed. The unexposed pregnancies were weighted in the outcome models based on the distribution of the exposed across the propensity score strata. Adjusted RR and 95% CI were then estimated using generalized linear models. The distribution of covariates in the trimmed population, stratified by ACE inhibitor exposure, is shown for the exposed and the unexposed (with the latter weighted to the distribution of the exposed across the propensity score strata).
A number of prespecified sensitivity analyses were performed to test the robustness of our findings. First, we performed a high dimensional propensity score analysis, which has been demonstrated to improve confounding control in some circumstances.25 The high dimensional propensity score algorithm screens all inpatient and outpatient diagnoses and procedures, in addition to claims for dispensed medications, and prioritizes 200 covariates that may be proxies for unmeasured confounders based on the strength of the association with exposure. These variables are then included in a propensity score along with all investigator-specified covariates from the main analysis. Second, to assess any potential effect of exposure misclassification, we redefined exposure based on two dispensings of an ACE inhibitor during the first trimester on the assumption that, if a woman is regularly refilling the medication, it is likely being taken as prescribed. Third, we restricted the outcomes to claims from the infant record alone to control for the possibility of maternal malformations incorrectly being attributed to the infant. Finally, we redefined hypertension using all available claims at any time before pregnancy through the end of the first trimester (and not just diagnoses recorded in the 3 months before the last menstrual period through the end of the first trimester) to increase the sensitivity with which hypertension was captured both in the exposed and unexposed.
The Medicaid Analytic eXtract pregnancy cohort includes information on only those pregnancies that result in live births. It does not capture pregnancies that result in stillbirth or spontaneous or therapeutic abortions. If the frequency of live birth is the same in those exposed and unexposed to ACE inhibitors (within the levels of the covariates included in the propensity score), RR estimates will be unbiased. However, if this is not the case and live births occur less frequently in those exposed, estimates from the main analysis may be downwardly biased. To quantify the potential effect of missing live births, we used methods previously described in detail.16,17 Briefly, we assumed that the frequency of nonlive birth (including stillbirth and spontaneous or therapeutic abortion) in nonmalformed fetuses is 20%. Then, informed by literature-based estimates of the frequency of termination for malformations, we modeled a range of live-birth probabilities in the unexposed malformed, ranging from 75% to 55% for overall malformations and cardiac malformations and 55–35% for CNS malformations (because termination frequencies for CNS malformations are higher than for other anomalies26). We then examined the effect of a 10–20% higher frequency of nonlive birth among the ACE inhibitor–exposed pregnancies on the RRs estimated in the main adjusted analysis.
The full cohort included 1,333,624 pregnancies, of which 4,107 (0.31%) were exposed to ACE inhibitors during the first trimester (Fig. 1). The hypertension-restricted cohort included 18,515 pregnancies of which 2,631 (14.2%) were dispensed an ACE inhibitor during the first trimester. Among these, the most commonly dispensed ACE inhibitors included lisinopril (n=1,437 [54.6%]), benazepril (n=504 [19.2%]), and enalapril (n=282 [10.7%]) (see Appendix 1, http://links.lww.com/AOG/A895).
There were important baseline differences in the chronic hypertension-restricted cohort between patients exposed to ACE inhibitors and those who were not (Table 1). Angiotensin-converting enzyme inhibitor–exposed women were generally older (standardized difference for maternal aged 40 years or older 0.26) and more likely to be African American (standardized difference 0.10). They had much higher prevalences of diabetes diagnosis and treatment; 32% of those exposed to ACE inhibitors carried a diagnosis of diabetes compared with approximately 10% in hypertensive women who were unexposed (standardized difference 0.55). They also had a higher prevalence of renal disease (standardized difference 0.17), ischemic heart disease (standardized difference 0.10), congestive heart failure (standardized difference 0.19), and dyslipidemia (standardized difference 0.23). The Obstetric Comorbidity scores were higher (standardized difference 0.51). Finally, all measures of health care utilization assessed were higher among the ACE inhibitor–exposed pregnancies. After balancing these characteristics using the propensity score, the prevalence of all covariates was very similar in the exposed and unexposed with absolute standardized differences of 0.03 or less across all variables included.
In the full cohort, the prevalence of overall malformations in the ACE inhibitor–exposed pregnancies was 5.9% compared with 3.3% in the unexposed (unadjusted RR 1.82; 95% CI 1.61–2.06), of cardiac malformations was 3.4% compared with 1.2% (RR 2.95, 95% CI 2.50–3.47), and of CNS malformations was 0.27% compared with 0.18% (RR 1.46, 95% CI 0.81–2.64) (Table 2). After restricting the cohort to those with a diagnosis of chronic hypertension, the association between exposure to ACE inhibitors and CNS malformations was nonsignificant (RR 1.07, 95% CI 0.51–2.27) and was markedly attenuated for overall malformations (RR 1.35, 95% CI 1.13–1.61) and cardiac malformations (RR 1.79, 95% CI 1.39–2.30). Adjustment for diabetes resulted in no significant associations: for overall malformations the adjusted RR was 0.97 (95% CI 0.79–1.19), for cardiac malformations 1.08 (95% CI 0.81–1.44), and for CNS malformations 0.68 (95% CI 0.30–1.54). In the propensity score analysis, which adjusted for all covariates, the estimates did not suggest an increase in the risk of malformations associated with first-trimester ACE inhibitor exposure: for overall malformations, the fully adjusted RR was 0.89 (95% CI 0.75–1.06), for cardiac malformations 0.95 (95% CI 0.75–1.21), and for CNS malformations 0.54 (95% CI 0.26–1.11).
Relative risk estimates across the sensitivity analyses performed were consistent with those of the main analysis for overall malformations and cardiac malformations and none of the point estimates from these analyses suggested an increase in risk associated with ACE inhibitor exposure (Table 3). However, as a result of the paucity of events, the risk estimates were less stable and the CIs were relatively wide for CNS malformations.
The analyses of the effect of missing nonlive births are shown in Fig. 2A–C. For overall malformations, the most extreme scenario considered was a probability of live birth of 55% (20% terminated or spontaneously aborted or stillborn for reasons other than malformations and 25% for malformations) and a 20% absolute decrease in the probability of live birth in the ACE inhibitor–exposed pregnancies compared with the unexposed (for both malformed and nonmalformed fetuses). Under these conditions, the estimate of RR would shift from 0.89 (as estimated in the propensity score-adjusted analysis) to 1.05. Using the same assumptions, the RR for cardiac malformations shifts from 0.95 to 1.12 in the most extreme scenario. For CNS malformations, a higher frequency of termination for malformations was modeled. The most extreme scenario considered in this setting modeled a 35% probability of live birth in the unexposed with malformations (20% terminated or spontaneously aborted or stillborn for reasons other than malformations and 45% for malformations) and a 20% absolute decrease in the probability of live birth in the ACE inhibitor–exposed pregnancies. In this scenario, the corrected RR estimate shifts substantially upward, from 0.54 to the null.
In this study based on a cohort of more than 1.3 million pregnancies, after accounting for relevant confounders, we did not observe an increase in the risk of overall malformations, cardiac malformations, or CNS malformations associated with first-trimester ACE inhibitor exposure. Although it is important for clinicians to discontinue ACE inhibitors before the second trimester to avoid the fetopathy associated with late pregnancy exposure, our data suggest that exposure early in pregnancy during the period of organogenesis does not confer an increase in the risk of malformations.
Clinically this finding is important because it suggests that this class of medication is appropriate for use in women of reproductive age who may become pregnant provided they are able to present for prenatal care before the end of the first trimester. Angiotensin-converting enzyme inhibitors are considered a first-line medication in the treatment of hypertension, particularly in patients with diabetes27 and chronic renal disease.28 However, current American College of Obstetricians and Gynecologists' recommendations suggest avoiding this antihypertensive in women of reproductive age in the absence of a compelling indication12 because of the concern of inadvertent exposure during early pregnancy. Our results suggest a critical reevaluation of this recommendation.
Our findings differ from a previous study using Tennessee Medicaid Claims, which reported ACE inhibitor exposure was associated with a 2.7-fold increase in malformations overall, a 3.7-fold increase in cardiac malformations, and a 4.4-fold increase in CNS malformations.7 However, that analysis did not adjust for the indication of hypertension. Although not traditionally considered a risk factor for malformations and although the mechanism underlying the association is not well understood, several recent studies suggest that hypertension may be an independent risk factor for malformations.8,29,30 The Tennessee Medicaid analysis may not have adequately accounted for the presence of diabetes. Furthermore, the analysis included only 209 ACE inhibitor–exposed pregnancies, resulting in very wide CIs. It is notable that our unadjusted estimates were more consistent with the findings from the Tennessee data with a significant observed increase in the risk associated with exposure. This observation indicates that the overall risk of malformation is higher in women eligible to receive ACE inhibitors compared with the general population, but the increased risk appears to be attributable to the underlying conditions of hypertension and diabetes in this population and not attributable to ACE inhibitor use. Indeed, after fully accounting for all comorbid conditions that could increase the risk of malformations, the RR estimates for ACE inhibitors were near or below the null value. This lack of risk is consistent with a more recent study, which examined 755 ACE inhibitor–exposed pregnancies from Kaiser Permanente Northern California.8 In that analysis, when the control group was specified as pregnancies to women with untreated hypertension, there was no increase in risk of overall, cardiac, or CNS malformations attributable to ACE inhibitor exposure.
It is notable that the point estimate for the relative risk of CNS malformations shifted to 0.54 after full adjustment for potential confounders. There are several potential explanations for why the point estimate might appear to be protective, even in the presence of the lack of effect of ACE inhibitor exposure on the risk of the outcome. The first is the marked instability of the RR estimate generated by the fact that there are fewer than 11 cases in the ACE-exposed group and only 41 in the reference group. Indeed, the wide CI associated with the estimate (which includes the null) and the shifts in the point estimate across the multiple sensitivity analyses demonstrate this instability. Second, an observed protective estimate could be the result of higher termination frequencies for malformations in the ACE inhibitor–exposed pregnancies than in the unexposed as shown in our sensitivity analyses examining the effect of missing live births. Finally, if treatment with ACE inhibitors is a marker for better management of chronic conditions associated with risk of malformations, such as diabetes, there may be residual confounding that bias the risk estimates downward. This said, our results are clearly not consistent with a substantial increase in the risk of CNS malformations associated with ACE inhibitor exposure.
Our study has a number of important strengths. Our data are drawn from the claims of Medicaid beneficiaries nationwide. The cohort created using these data represents one of the largest and most comprehensive pregnancy cohorts developed for the study of drug safety during pregnancy. The number of ACE inhibitor–exposed pregnancies included in the study was approximately three times larger than any prior study examining the teratogenicity of this class of medications, allowing for relatively precise estimates of risk to be made. Because the Medicaid Analytic eXtract pregnancy cohort contains complete health care utilization information on all included women from 3 months before pregnancy through delivery, there is capture of rich information regarding potential confounders including medical conditions and medication exposures. These confounders were adjusted for using advanced epidemiologic methods including propensity score stratification and high dimensional propensity score analyses. We confirmed the robustness of the findings with respect to overall malformations and cardiac malformations in multiple sensitivity analyses, although for CNS malformations, the estimates were unstable owing to a very low number of outcomes.
Our study is also subject to certain limitations inherent in its design. Our exposure definition is based on a dispensed medication during the first trimester. Although it is a reasonable assumption that filled prescriptions are taken, this cannot be empirically confirmed. To overcome this potential limitation, we performed a sensitivity analysis in which we required two dispensings of an ACE inhibitor during the first trimester, because it is likely that, if a woman refills her medication, it is being taken regularly. This analysis, which is expected to define exposure with greater specificity, yielded risk estimates that were comparable with those of the main analysis. The Medicaid Analytic eXtract data also lack certain potential confounders including measures of the severity of comorbidities such as renal disease or diabetes and patient characteristics such as body mass index. We addressed this potential limitation through the inclusion of multiple proxy variables for these conditions in the propensity score and the use of a high dimensional propensity score (which included 200 empirically defined variables in addition to prespecified confounders). Given the null associations reported, it is unlikely that residual confounding is a significant issue for our analyses. It is also important to note that as a result of small numbers of events, the risk estimates for CNS malformations are unstable. The adjusted RRs emerging for the assessment of the effect of considering only live births are thus also imprecisely estimated. Therefore, the conclusions regarding the effect of ACE inhibitors on CNS malformations must be made cautiously.
An additional potential limitation is that we define the presence of malformations based on administrative coding of the condition. However, previous work from our group has validated this approach with a chart review, finding a high positive predictive value for certain major congenital malformations defined using algorithms that rely on codes on multiple dates or corrective surgery.21 Finally, our cohort includes only live births. To address this concern, we conducted a series of sensitivity analyses exploring the effect of differential live-birth frequencies in the exposed and unexposed and the potential effect on the risk estimates from the main analysis. In the analysis of overall malformations and cardiac malformations, even under the most extreme assumptions, the corrected RR estimates were less than 1.2. Finally, although our study does not suggest an association with overall malformations, cardiac malformations, or CNS malformations in aggregate, we cannot exclude an association with specific defects.
In conclusion, our results suggest that ACE inhibitor exposure during the first trimester is not associated with an increase in the risk of congenital malformations after accounting for the underlying indication of hypertension and confounding factors such as the presence of diabetes. Our findings suggest that ACE inhibitors can be safely used in women of reproductive age, although it remains imperative to transition women off of these medications early in pregnancy to avoid the known adverse fetal effects associated with late pregnancy exposure.
1. Bateman BT, Shaw KM, Kuklina EV, Callaghan WM, Seely EW, Hernández-Diaz S. Hypertension in women of reproductive age in the United States: NHANES 1999–2008. PLoS One 2012;7:e36171.
2. Andrade SE, Raebel MA, Brown J, Lane K, Livingston J, Boudreau D, et al. Outpatient use of cardiovascular drugs during pregnancy. Pharmacoepidemiol Drug Saf 2008;17:240–7.
3. Bateman BT, Hernandez-Diaz S, Huybrechts KF, Palmsten K, Mogun H, Ecker JL, et al. Patterns of outpatient antihypertensive medication use during pregnancy in a Medicaid population. Hypertension 2012;60:913–20.
4. Barr M Jr, Cohen MM Jr ACE inhibitor fetopathy and hypocalvaria: the kidney-skull connection. Teratology 1991;44:485–95.
5. Hanssens M, Keirse MJ, Vankelecom F, Van Assche FA. Fetal and neonatal effects of treatment with angiotensin-converting enzyme inhibitors in pregnancy. Obstet Gynecol 1991;78:128–35.
6. Pryde PG, Sedman AB, Nugent CE, Barr M Jr. Angiotensin-converting enzyme inhibitor fetopathy. J Am Soc Nephrol 1993;3:1575–82.
7. Cooper W, Hernandez-Diaz S, Arbogast P, Dudley JA, Dyer S, Gideon PS, et al. Major congenital malformations after first-trimester exposure to ACE inhibitors. N Engl J Med 2006;354:2443–51.
8. Li DK, Yang C, Andrade S, Tavares V, Ferber JR. Maternal exposure to angiotensin converting enzyme inhibitors in the first trimester and risk of malformations in offspring: a retrospective cohort study. BMJ 2011;343:d5931.
9. Walfisch A, Al-Maawali A, Moretti ME, Nickel C, Koren G. Teratogenicity of angiotensin converting enzyme inhibitors or receptor blockers. J Obstet Gynaecol 2011;31:465–72.
10. Polifka JE. Is there an embryopathy associated with first-trimester exposure to angiotensin-converting enzyme inhibitors and angiotensin receptor antagonists? A critical review of the evidence. Birth Defects Res A Clin Mol Teratol 2012;94:576–98.
11. Koren G. Hypertension: ACE inhibitor use in pregnancy—setting the record straight. Nat Rev Cardiol 2012;9:7–8.
12. Executive summary: hypertension in pregnancy. American College of Obstetricians and Gynecologists. Obstet Gynecol 2013;122:1122–31.
13. Garcia G. Maternal and child health (MCH) update: states increase eligibility for children's health in 2007. Available at: http://www.nga.org/files/live/sites/NGA/files/pdf/0811MCHUPDATE.PDF;jsessionid=7B47A647247DD4E5CB9B709C8F9797AE
. Retrieved March 5, 2012.
14. Palmsten K, Huybrechts KF, Mogun H, Kowal MK, Williams PL, Michels KB, et al. Harnessing the Medicaid Analytic eXtract (MAX) to evaluate medications in pregnancy: design considerations. PLoS One 2013;8:e67405.
15. Margulis AV, Setoguchi S, Mittleman MA, Glynn RJ, Dormuth CR, Hernández-Diaz S. Algorithms to estimate the beginning of pregnancy in administrative databases. Pharmacoepidemiol Drug Saf 2013;22:16–24.
16. Huybrechts KF, Palmsten K, Avorn J, Cohen LS, Holmes LB, Franklin JM, et al. Antidepressant use in pregnancy and the risk of cardiac defects. N Engl J Med 2014;370:2397–407.
17. Bateman BT, Hernandez-Diaz S, Fischer MA, Seely EW, Ecker JL, Franklin JM, et al. Statins and congenital malformations: cohort study. BMJ 2015;350:h1035.
18. Desai RJ, Huybrechts KF, Hernandez-Diaz S, Mogun H, Patorno E, Kaltenbach K, et al. Exposure to prescription opioid analgesics in utero and risk of neonatal abstinence syndrome: population based cohort study. BMJ 2015;350:h2102.
19. Huybrechts KF, Bateman BT, Palmsten K, Desai RJ, Patorno E, Gopalakrishnan C, et al. Antidepressant use late in pregnancy and risk of persistent pulmonary hypertension of the newborn. JAMA 2015;313:2142–51.
20. Yakoob MY, Bateman BT, Ho E, Hernandez-Diaz S, Franklin JM, Goodman JE, et al. The risk of congenital malformations associated with exposure to β-blockers early in pregnancy: a meta-analysis. Hypertension 2013;62:375–81.
21. Palmsten K, Huybrechts KF, Kowal MK, Mogun H, Hernández-Diaz S. Validity of maternal and infant outcomes within nationwide Medicaid data. Pharmacoepidemiol Drug Saf 2014;23:646–55.
22. Bateman BT, Mhyre JM, Hernandez-Diaz S, Huybrechts KF, Fischer MA, Creanga AA, et al. Development of a comorbidity index for use in obstetric patients. Obstet Gynecol 2013;122:957–65.
23. Metcalfe A, Lix LM, Johnson JA, Currie G, Lyon AW, Bernier F, et al. Validation of an obstetric comorbidity index in an external population. BJOG 2015;122:1748–55.
24. Correa A, Gilboa SM, Besser LM, Botto LD, Moore CA, Hobbs CA, et al. Diabetes mellitus and birth defects. Am J Obstet Gynecol 2008;199:237.e1–9.
25. Schneeweiss S, Rassen JA, Glynn RJ, Avorn J, Mogun H, Brookhart MA. High-dimensional propensity score adjustment in studies of treatment effects using health care claims data. Epidemiology 2009;20:512–22.
26. Svensson E, Ehrenstein V, Nørgaard M, Bakketeig LS, Rothman KJ, Sørensen HT, et al. Estimating the proportion of all observed birth defects occurring in pregnancies terminated by a second-trimester abortion. Epidemiology 2014;25:866–71.
27. Chamberlain JJ, Rhinehart AS, Shaefer CF Jr, Neuman A. Diagnosis and management of diabetes: synopsis of the 2016 American Diabetes Association standards of medical care in diabetes. Ann Intern Med 2016;164:542–52.
28. James PA, Oparil S, Carter BL, Cushman WC, Dennison-Himmelfarb C, Handler J, et al. Evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA 2014;311:507–20.
29. Bateman BT, Huybrechts KF, Fischer MA, Seely EW, Ecker JL, Oberg AS, et al. Chronic hypertension in pregnancy and the risk of congenital malformations: a cohort study. Am J Obstet Gynecol 2015;212:337.e1–14.
30. Ramakrishnan A, Lee LJ, Mitchell LE, Agopian AJ. Maternal hypertension during pregnancy and the risk of congenital heart defects in offspring: a systematic review and meta-analysis. Pediatr Cardiol 2015;36:1442–51.