Attention-deficit/hyperactivity disorder (ADHD) diagnoses and stimulant prescriptions have markedly increased over the past two decades in women of childbearing age, resulting in increased stimulant exposure during pregnancy.1–3 Despite increasing use, there are limited safety data on stimulant use in pregnancy. Although discontinuing treatment may be a viable option for some, women with severe symptoms may benefit from continued use of stimulants during pregnancy because severe, untreated ADHD can result in anxiety and aggressive behaviors that can disrupt family relationships and newborn care.4 Thus, understanding the safety of ADHD medications is critical for appropriate patient counseling regarding stimulant use during pregnancy.
The safety of first-line ADHD medications (eg, amphetamine–dextroamphetamine and methylphenidate) is of concern, because these stimulants cause vasoconstriction, which may impair placental perfusion.5 Impaired uteroplacental perfusion has been linked to obstetric complications including preeclampsia, placental abruption, and fetal growth restriction6 as well as preterm birth.7
Our objective was to evaluate whether stimulant use in pregnancy is associated with increased risks of preeclampsia, placental abruption, small-for-gestational-age (SGA) neonates, and preterm delivery. We also aimed to determine whether stimulant compared with nonstimulant treatments for ADHD have different safety profiles with respect to placental complications and to define the etiologically relevant exposure period for any observed increase in risk. We hypothesized that use of stimulants may adversely affect placentation (early pregnancy) or placental function (later pregnancy) leading to different manifestations of ischemic placental disease including impaired fetal growth, preeclampsia, placental abruption, and preterm birth.
MATERIALS AND METHODS
We designed a population-based cohort study within the large, diverse population of low-income pregnant women covered by Medicaid (public insurance) in the United States. We linked pregnant women and their liveborn offspring within the Medicaid Analytic eXtract database from 2000 to 2010. Medicaid Analytic eXtract data are managed by the Research Data Assistance Center at the University of Minnesota. This mother–neonate cohort has been described in detail previously8 and has been extensively used to study drug safety during pregnancy.9–14 The cohort includes deliveries in mothers ages 12–55 years from 46 states and Washington, DC. For this study, we included women who were continuously enrolled in Medicaid from 3 months before the first day of their last menstrual period (LMP) until 1 month after delivery without supplemental insurance or restricted benefits. Neonates were also required to be enrolled or have a claim in the month after birth, unless they died. We excluded pregnancies in which the neonate was diagnosed with a major congenital anomaly (3.4%), because we were interested in whether stimulants increase the risk of adverse pregnancy outcomes related to placental complications in pregnancies without congenital malformations.
Our early exposure window of interest was defined as the first 20 weeks of pregnancy (LMP to LMP+140 days). We chose this window because we hypothesized that exposure during this period could affect placental implantation and early development leading to the placental complications of interest. We focused on the most common stimulants, amphetamine–dextroamphetamine and methylphenidate. They have similar psychostimulant effects and side effects including insomnia and anorexia; however, the effects of amphetamine–dextroamphetamine are longer lasting.15 Atomoxetine is a nonstimulant ADHD medication used in cases in which other stimulants are ineffective or need to be avoided as a result of side effects or potential for abuse.16 We included it in our study as a negative control exposure to evaluate the possibility of residual confounding by indication. We defined monotherapy based on at least one filled prescription from LMP to LMP+140 days with no prescription for any other ADHD medication from LMP–90 days to LMP+140 days. Our reference group for the early exposure window consisted of women without a prescription for any ADHD medication (amphetamine, amphetamine–dextroamphetamine, dextroamphetamine, methamphetamine, lisdexamfetamine, methylphenidate, dexmethylphenidate, pemoline, atomoxetine, guanfacine, and clonidine) during the 3 months before pregnancy or LMP+140 days.
Because very few women newly initiated stimulant treatment late in pregnancy, we compared women who continued stimulant (amphetamine–dextroamphetamine or methylphenidate) treatment after LMP+140 with those who discontinued to evaluate whether the etiologically relevant risk period extends into the second half of pregnancy. The number of women exposed to atomoxetine in late pregnancy was too small to allow for a similar analysis. We defined continuation of stimulant monotherapy as filling at least one prescription for the same medication in the second half of pregnancy. We assessed late exposures from 141 until 245 days (greater than 20–35 weeks) of gestation to avoid a differential opportunity for exposure in preterm compared with term births.
Preeclampsia was defined by an inpatient diagnosis from 141 days after LMP to 30 days after delivery. A prior validation study showed the positive predictive value (PPV) of this outcome definition to be 95%.17 Other outcome definitions were validated in a subset of study participants who received obstetric care at Brigham and Women's Hospital or Massachusetts General Hospital (He M, Cottral JA, Bartels DD, Dejene SZ, Mogun H, Huybrechts KF, Hernandez-Diaz S, Bateman BT. Validation of algorithms to identify perinatal outcomes in large claims databases . Pharmacoepidemiology Drug Saf 2017; 26(S2): 3-636.). Placental abruption was defined by at least one diagnosis during the delivery hospitalization (PPV 92%). Small for gestational age was defined by a diagnosis in maternal or neonatal claims from delivery to 30 days after delivery (PPV 84%). Preterm birth was defined by diagnosis and procedure codes that were present in maternal or infant claims from delivery to 60 days after delivery (PPV 78%). The International Classification of Diseases, 9th Revision (ICD-9) codes used to identify the outcomes are presented in Appendix 1, available online at http://links.lww.com/AOG/B33.
We considered risk factors for ischemic placental disease and preterm birth including demographic characteristics (age, race, geographic region, year, multiparity), maternal and pregnancy characteristics (multifetal gestation, alcohol use, tobacco use, other drug abuse or dependence, obesity or overweight), and certain chronic conditions (inflammatory, cardiovascular, renal) as potential confounders. Additionally, we adjusted for indications for stimulants and proxies for indication severity, other psychiatric and pain conditions, proxies for health care utilization intensity, and cotreatment with psychiatric and pain medication. Confounders were defined based on ICD-9 codes or prescriptions dispensed from LMP–90 days to LMP+140 days. Health care utilization variables were assessed during the period before pregnancy (LMP–90 to LMP–1) to capture the use resulting from pre-existing health conditions, which may serve as a proxy for general health before pregnancy. The following potential indications for psychostimulant treatment were defined as one or more relevant ICD-9 codes from LMP–90 days to delivery to enhance sensitivity: ADHD, migraine, bipolar, chronic fatigue syndrome, narcolepsy, and epilepsy. We assumed that these chronic conditions temporally preceded the use of stimulants, although they may have been captured after the prescription claim was identified.
We estimated risk ratios (RRs) and 95% CIs using log-binomial regression. We controlled for potential confounding using propensity scores. Propensity score models predict the likelihood of exposure based on the measured confounders. This enables the comparison of participants who had a similar probability of receiving the treatment. Thus, propensity score methods allow us to mimic a randomized trial in which comparable women are randomly assigned to treatment or control, assuming there are no unmeasured confounders. The propensity score model included exposed compared with a reference as the dependent variable and confounders as independent variables: demographic characteristics; pregnancy characteristics; and chronic inflammatory, cardiovascular, and renal conditions (24 covariates); potential indications and proxies for the indication severity and health care utilization (35 covariates; see Appendix 3, available online at http://links.lww.com/AOG/B33, for a complete list).
After we estimated the propensity scores using logistic regression, we trimmed the nonoverlapping regions of the exposed and reference group distributions and then used a fine stratification approach, which has been shown to control for confounding better or more efficiently than matching or coarse stratification when exposures are rare.18 Exposed women were stratified into 50 equally sized strata. Then reference participants in each stratum were weighted according to the distribution of the exposed women.
We conducted sensitivity analyses to explore the robustness of our findings. To address the possibility that women with only one prescription may not truly be exposed in pregnancy, we modified the definition of exposed to require at least two filled prescriptions in the first 20 weeks of gestation. Next, we redefined exposure as having a drug supply available from 8 to 18 weeks of gestation, the period during which extensive remodeling of the uterine spiral arteries occurs, which is a process noted to be incomplete in ischemic placental disease.19 This time period may be particularly relevant if the mechanism whereby stimulants affect these outcomes interferes with early placental development. We also stratified the exposed according to days' supply available during the first 140 days of gestation, grouped as less than 30, 30–60, 61–90, and greater than 90 days; this was collapsed to less than 30, 30–60, and greater than 60 days for atomoxetine as a result of limited sample size.
The study was approved by the institutional review board at Brigham and Women's Hospital, and the need for informed consent was waived.
The primary reference group was pregnant women who were not dispensed any ADHD medication from 90 days before LMP to 140 days after LMP (n=1,461,493). The exposed groups included women exposed to monotherapy of amphetamine–dextroamphetamine (n=3,331), methylphenidate (n=1,515), and atomoxetine (n=453) during the first 140 days of gestation (Appendix 2, available online at http://links.lww.com/AOG/B33). Women using ADHD medication monotherapy in pregnancy were on average younger, more likely to be white, and had more risk factors for placental-mediated pregnancy complications. Women who continued to use stimulant monotherapy after 20 weeks of gestation were more similar to women who discontinued use in the first 20 weeks of gestation, although they were older, more likely to be multiparous, and more likely to have a diagnosis of ADHD (Table 1). After propensity score weighting, comparison groups were similar with respect to the covariates of interest (Appendices 3–6, available online at http://links.lww.com/AOG/B33).
Among unexposed women, the risks of the outcomes were 3.7% for preeclampsia, 1.4% for placental abruption, 2.9% for SGA, and 11.2% for preterm birth. In unadjusted analyses, women using amphetamine–dextroamphetamine monotherapy in the first half of pregnancy (irrespective of their exposure in the second half of pregnancy) had an increased risk of all placental-mediated complications examined. Women using methylphenidate had increased risks for all outcomes examined except for placental abruption. However, after adjusting for potential confounders, the associations were attenuated and suggested no effect for most exposure–outcome contrasts except for stimulant exposure as a class and preeclampsia (adjusted RR 1.29, 95% CI 1.11–1.49) and amphetamine–dextroamphetamine and preeclampsia (1.33, 1.12–1.58). Atomoxetine use was not associated with an increased risk for the outcomes in crude or adjusted analyses (Fig. 1).
Continuing to use stimulant monotherapy (amphetamine–dextroamphetamine or methylphenidate) in the second half of pregnancy compared with discontinuation in the first half of pregnancy was also associated with an increased risk of preterm birth after adjusting for confounders (adjusted RR 1.30, 95% CI 1.10–1.95; Table 2).
The positive associations identified in the primary analyses remained similar or slightly strengthened for women filling two or more prescriptions during the first half of pregnancy. Redefining exposure as using stimulants in a narrower time window surrounding placentation (8–18 weeks of gestation) also produced consistent findings. Amphetamine–dextroamphetamine use was associated with preeclampsia, placental abruption, and preterm birth and methylphenidate use with preeclampsia in these sensitivity analyses. Atomoxetine remained unassociated with the outcomes studied (Fig. 2). When we stratified the exposed women based on days supply of medication in the first 140 days after LMP, the association between amphetamine–dextroamphetamine and preeclampsia was apparent for women who had at least a 30-day supply (adjusted RR 1.03, 95% CI, 0.71–1.50 for less than 30 days, 1.33 [1.10–1.74] for 30–60 days, and 1.38 [0.98–1.95] for 61–90 days, and 1.50 [1.11–2.02] for greater than 90 days); similar findings were not observed for placental abruption or preterm birth, in which only greater than 90 days of exposure to amphetamine–dextroamphetamine was associated with an increased risk of preterm birth. There was no evidence of a clear dose–response relationship for methylphenidate and either preeclampsia or SGA; only exposure greater than 60 days was associated with increased risk of preeclampsia. There were no trends corresponding to dose of atomoxetine for any of the outcomes examined (Fig. 3).
In a cohort of almost 1.5 million pregnancies, we identified more than 5,000 pregnancies exposed to ADHD monotherapy. Stimulant monotherapy in the first half of pregnancy remained associated with a 1.3-fold increased risk of preeclampsia after controlling for confounding by indication and other risk factors. Continued stimulant exposure late in pregnancy was associated with an increased risk of preterm birth (1.3-fold compared with discontinuation). The absolute increases in risk are small and may not justify additional surveillance of pregnancies exposed to psychostimulant therapy. Atomoxetine, a nonstimulant ADHD medication, was not associated with the adverse pregnancy outcomes examined.
We would expect stronger associations for amphetamine–dextroamphetamine and methylphenidate than for atomoxetine monotherapy. Amphetamine-type drugs (amphetamine, dextroamphetamine, and methylphenidate) increase levels of circulating epinephrine through activation of the sympathetic nervous system producing vasoconstriction.20 Alternatively, atomoxetine acts as a selective norepinephrine transporter inhibitor primarily in the prefrontal cortex and effects on vasoconstriction are limited.21 Thus, atomoxetine may be an ideal negative control exposure as a result of the similar indication for use and pharmacologic effects but limited potential for side effects on the maternal vasculature. Experiments in sheep demonstrated that amphetamines cause vasoconstriction and decreased uteroplacental perfusion.5 Furthermore, if stimulants cause inadequate placental blood flow, exposure during the second half of pregnancy may be particularly relevant for growth restriction because most fetal weight gain occurs in late pregnancy.
Small studies that lacked control for confounders suggested an association between stimulant abuse and therapeutic use for weight control during pregnancy and lower birth weight (including SGA) and premature birth.22,23 A cohort study also identified an association between late pregnancy therapeutic methamphetamine use and fetal growth restriction.24 A recent cohort study compared women who used ADHD medication and women with ADHD diagnosis without medication with women with no ADHD diagnosis or medication in the general population of pregnancies in Denmark from 1997 to 2008.25 They reported that methylphenidate or atomoxetine use in pregnancy (n=186) was not associated with reduced birth weight or gestational age at birth. However, they reported that women with a hospital-based diagnosis of ADHD who did not use medication (n=275) were at an approximately twofold increased risk of preterm birth. This study highlighted the potential for cofounding by indication (ie, ADHD) and the need for larger studies of amphetamine–dextroamphetamine and methylphenidate, which are the most commonly used stimulants in the United States.
Residual confounding is always a concern in observational studies. As a result of prior research implicating depression and antidepressant use in the risk of preeclampsia and preterm birth,9,26 adjusting for depression and associated treatment was of key concern. Although we defined potential indications, smoking, and overweight or obesity from ICD-9 codes, these may have been under recorded. However, because we controlled for many psychiatric and physical conditions and comedication use, which may be correlated with ADHD, smoking, and obesity, we likely controlled for these to a large extent by proxy. The negative findings for atomoxetine provide further reassurance in this regard.
Outcomes were defined by a set of validated algorithm-based definitions designed to maximize specificity while preserving sensitivity and should lead to unbiased estimates of the relative risks. Small for gestational age appeared to be underrecorded in these data, because the rate of 2.9% in the unexposed women was lower than the expected rate of approximately 10%. It has been shown that, if specificity is high and both sensitivity and specificity are nondifferential, RRs are unbiased.27 However, the lower PPV for some outcomes (eg, SGA, preterm birth) may have biased results toward the null.
The modest effect estimates for early pregnancy exposure in this study may be a result of frequent discontinuation of stimulants early in the first half of pregnancy. Indeed, requiring two prescriptions or a day’s supply that overlapped the period from 8 to 18 weeks of gestation resulted in somewhat stronger associations. Additional risk was also observed for women who continued to use stimulants in the second half of pregnancy compared with discontinuation during the first half of pregnancy. Continuation of use reflects both a higher cumulative exposure and exposure at a different time compared with discontinuation. It therefore remains unclear whether later exposure or longer duration of treatment explains the increased risk of preterm birth, primarily associated with continuation of treatment in late pregnancy.
In conclusion, stimulant use was associated with increased risks of preeclampsia and preterm birth. Although the point estimates were elevated, associations for placental abruption and SGA were not statistically significant. However, the absolute increases in risks are small and, thus, women with significant ADHD should not be counseled to suspend their ADHD treatment based on these findings.
1. Renoux C, Shin JY, Dell'Aniello S, Fergusson E, Suissa S. Prescribing trends of attention-deficit hyperactivity disorder (ADHD) medications in UK primary care, 1995–2015. Br J Clin Pharmacol 2016;82:858–68.
2. Haervig KB, Mortensen LH, Hansen AV, Strandberg-Larsen K. Use of ADHD medication during pregnancy from 1999 to 2010: a Danish register-based study. Pharmacoepidemiol Drug Saf 2014;23:526–33.
3. Louik C, Kerr S, Kelley KE, Mitchell AA. Increasing use of ADHD medications in pregnancy. Pharmacoepidemiol Drug Saf 2015;24:218–20.
4. Besag FM. ADHD treatment and pregnancy. Drug Saf 2014;37:397–408.
5. Stek AM, Baker RS, Fisher BK, Lang U, Clark KE. Fetal responses to maternal and fetal methamphetamine administration in sheep. Am J Obstet Gynecol 1995;173:1592–8.
6. Ananth CV, Vintzileos AM. Medically indicated preterm birth: recognizing the importance of the problem. Clin Perinatol 2008;35:53–67, viii.
7. Nijman TA, van Vliet EO, Benders MJ, Mol BW, Franx A, Nikkels PG, et al. Placental histology in spontaneous and indicated preterm birth: a case control study. Placenta 2016;48:56–62.
8. 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.
9. Palmsten K, Huybrechts KF, Michels KB, Williams PL, Mogun H, Setoguchi S, et al. Antidepressant use and risk for preeclampsia. Epidemiology 2013;24:682–91.
10. 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.
11. 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.
12. Huybrechts KF, Hernández-Díaz S, Patorno E, Desai RJ, Mogun H, Dejene SZ, et al. Antipsychotic use in pregnancy and the risk for congenital malformations. JAMA Psychiatry 2016;73:938–46.
13. Bateman BT, Patorno E, Desai RJ, Seely EW, Mogun H, Dejene SZ, et al. Angiotensin-converting enzyme inhibitors and the risk of congenital malformations. Obstet Gynecol 2017;129:174–84.
14. Patorno E, Huybrechts KF, Bateman BT, Cohen JM, Desai RJ, Mogun H, et al. Lithium use in pregnancy and the risk of cardiac malformations. N Engl J Med 2017;376:2245–54.
15. Pliszka SR, Browne RG, Olvera RL, Wynne SK. A double-blind, placebo-controlled study of Adderall and methylphenidate in the treatment of attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 2000;39:619–26.
16. Subcommittee on Attention-Deficit/Hyperactivity Disorder; Steering Committee on Quality Improvement and Management, Wolraich M, Brown L, Brown RT, DuPaul G, Earls M, et al. ADHD: clinical practice guideline for the diagnosis, evaluation, and treatment of attention-deficit/hyperactivity disorder in children and adolescents. Pediatrics 2011;128:1007–22.
17. Palmsten K, Huybrechts KF, Kowal MK, Mogun H, Hernández-Díaz S. Validity of maternal and infant outcomes within nationwide Medicaid data. Pharmacoepidemiol Drug Saf 2014;23:646–55.
18. Desai RJ, Rothman KJ, Bateman BT, Hernandez-Diaz S, Huybrechts KF. A propensity-score-based fine stratification approach for confounding adjustment when exposure is infrequent. Epidemiology 2017;28:249–57.
19. Pijnenborg R, Bland JM, Robertson WB, Brosens I. Uteroplacental arterial changes related to interstitial trophoblast migration in early human pregnancy. Placenta 1983;4:397–413.
20. Broadley KJ. The vascular effects of trace amines and amphetamines. Pharmacol Ther 2010;125:363–75.
21. Bymaster FP, Katner JS, Nelson DL, Hemrick-Luecke SK, Threlkeld PG, Heiligenstein JH, et al. Atomoxetine increases extracellular levels of norepinephrine and dopamine in prefrontal cortex of rat: a potential mechanism for efficacy in attention deficit/hyperactivity disorder. Neuropsychopharmacology 2002;27:699–711.
22. Debooy VD, Seshia MM, Tenenbein M, Casiro OG. Intravenous pentazocine and methylphenidate abuse during pregnancy. Maternal lifestyle and infant outcome. Am J Dis Child 1993;147:1062–5.
23. Naeye RL. Maternal use of dextroamphetamine and growth of the fetus. Pharmacology 1983;26:117–20.
24. Smith LM, LaGasse LL, Derauf C, Grant P, Shah R, Arria A, et al. The infant development, environment, and lifestyle study: effects of prenatal methamphetamine exposure, polydrug exposure, and poverty on intrauterine growth. Pediatrics 2006;118:1149–56.
25. Bro SP, Kjaersgaard MI, Parner ET, Sorensen MJ, Olsen J, Bech BH, et al. Adverse pregnancy outcomes after exposure to methylphenidate or atomoxetine during pregnancy. Clin Epidemiol 2015;7:139–47.
26. Tak CR, Job KM, Schoen-Gentry K, Campbell SC, Carroll P, Costantine M, et al. The impact of exposure to antidepressant medications during pregnancy on neonatal outcomes: a review of retrospective database cohort studies. Eur J Clin Pharmacol 2017;73:1055–69.
27. Copeland KT, Checkoway H, McMichael AJ, Holbrook RH. Bias due to misclassification in the estimation of relative risk. Am J Epidemiol 1977;105:488–95.
Supplemental Digital Content
© 2017 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.