Abstract
Objective:
To compare associations between individual antidepressants and newborn outcomes.
Design:
Retrospective cohort study
Setting:
Deliveries in a large, US medical system.
Population:
Women who received at least one antidepressant prescription 3 months prior to conception through delivery.
Methods:
Eligible women had maternal characteristics and newborn outcomes extracted from medical record data. Exposure was defined by the timing of the prescription during pregnancy.
Main outcome measures:
Newborn outcomes (any adaptation syndrome, neonatal intensive care unit (NICU) admission) were analyzed for each antidepressant and compared using standard statistics and multivariable regression compared to exposure to bupropion. Odds of outcomes based on timing of exposure were also explored.
Results:
A total of 3,694 women were analyzed. Rates of any adaptation syndrome (p<0.001), NICU admission (p<0.001), and transient tachypnea of newborn (TTN) (p=0.006) were significantly different between drugs. Infants exposed to duloxetine had the highest rates of NICU admissions (39.6%) and adaptation syndromes (15.1%). Venlafaxine-exposed infants had the highest rates of TTN (18.2%). Controlling for maternal age, race, insurance, and gestational age at delivery, early pregnancy antidepressant exposure was associated with adaptation syndrome and NICU admission for both duloxetine (adjusted odds ratio (aOR) 2.31 [95% Confidence Interval (CI) 1.11-4.80] and aOR 2.47 [95% CI 1.40-4.34], respectively) and escitalopram (aOR 1.72 [95% CI 1.09-2.70] and aOR 1.64 [95% CI 1.21-2.22], respectively). Exposure in the third trimester was associated with any adaptation syndrome for citalopram, duloxetine, escitalopram, fluoxetine, sertraline, and venlafaxine and NICU admission for bupropion, citalopram, duloxetine, escitalopram, and fluoxetine.
Conclusion:
Duloxetine and escitalopram appear to have the strongest associations with any adaptation syndrome and NICU admission whereas bupropion and sertraline tended to have among the lowest risks of these outcomes. These results can help providers and patients discuss choice of individual antidepressant drugs during pregnancy.
Funding:
National Institutes of Health (US)- UL1TR002529, T32GM842528 and the Indiana University Department of OB-GYN
Keywords: antidepressants, abstinence syndrome, pregnancy, newborn outcomes, exposure
Introduction:
Mental health is a vital component to overall health, particularly during pregnancy. Up to 20% of women experience moderate to severe depression during childbearing years.1,2 Drugs such as selective serotonin reuptake inhibitors (SSRIs) and serotonin norepinephrine reuptake inhibitors (SNRIs) are considered first-line pharmacological treatments for depression. Untreated depression in pregnancy is associated with adverse outcomes for both mother and baby including preterm delivery, preeclampsia, behavioral disturbances in baby at birth, and maternal suicidal ideations or attempts.2 Enrollment of pregnant women in clinical trials is lacking for many drugs. As such, the majority of data regarding medication use in pregnancy relies on retrospective studies looking at women who were prescribed and took certain drugs during pregnancy.
Although individual drugs are implicated in different birth defect studies, most other newborn outcome studies group antidepressant agents together. Current evidence suggests that drugs within this class, such as SSRIs or SNRIs, are generally regarded as safe for pregnancy and do not, as a class, increase risks of malformations or preterm birth more than the baseline risk for untreated depression.3 However, these SSRIs and SNRIs have unique pharmacology that may be associated with other birth outcomes including pediatric adaptation syndromes (PAS). Understanding the respective effect(s) of antidepressant agents on a variety of newborn outcomes is important for clinicians and patients.
The objective of this study was to compare associations between individual antidepressants with specific birth outcomes relative to each other. These individual drug comparisons can then be used to counsel pregnant women with depression who could benefit from drug therapy.
Methods:
This was a retrospective cohort study. It was approved by the local governing Institutional Review Board and the Regenstrief Institute Data Core approval committee as exempt. Data were obtained from electronic medical records (EMRs) through the Regenstrief Institute in Indianapolis, which serves as the honest data broker for EMRs utilized by several large health systems in Indiana. We collected data related to pregnancies delivering between January 1, 2010 and December 31, 2019 within Eskenazi Health or Indiana University Health systems. Medication orders were searched for any record of prescribed SSRI(s) and/or SNRI(s) starting from 100 days before the last menstrual period through the date of delivery. Antidepressants of interest included the SSRIs citalopram, escitalopram, fluoxetine, paroxetine, and sertraline and the SNRIs bupropion, desvenlafaxine, duloxetine, and venlafaxine. We did not include tricyclic and tetracyclic antidepressants as they are less commonly used during pregnancy for depression in our population. We defined exposure to one (or more) of the antidepressants as having a prescription written during the time period studied. As most women had multiple prescriptions written during pregnancy, we assumed that women had taken the medication(s) they were prescribed. For eligible women, the following variables were collected: maternal age at time of delivery, race, ethnicity, insurance, estimated due date, history of prior preterm birth, any other drugs prescribed during pregnancy, history of diabetes (types I, II, and gestational (GDM)), and hypertensive disorder of pregnancy. Infant outcomes collected were gestational age at birth, date of birth, birth weight, birth length, birth head circumference, still birth, diagnosis of any adaptation syndrome, neonatal intensive care unit (NICU) admission, 5 minute APGAR score, jaundice requiring treatment, diagnosis of transient tachypnea of the newborn (TTN) or respiratory distress syndrome (RDS), oxygen support, date of discharge, persistent pulmonary hypertension of the newborn (PPHN), neonatal seizures, and cardiac malformations. For the diagnosis of any adaptation syndrome, we combined any diagnosis of “neonatal abstinence syndrome” (NAS) or “pediatric adaptation syndrome” (PAS), as the hospital coding of this disorder changed to become more specific during the study time frame. Diagnoses were extracted from the electronic database discharge summaries, delivery records, and/or International Classification of Diseases (ICD) codes. Diagnoses were captured through the Regenstrief Institute’s documented Data Core process using ICD9/10 codes (Table S1).4,5 Clinical diagnoses were extracted from the delivery discharge summary written by the clinician caring for the infant. The primary outcomes were NICU admission and any adaptation syndrome. Infants are ascribed a unique medical record number at birth and all diagnoses and diagnostic codes for the infant were extracted by the Data Core through the linkage of maternal and infant medical record numbers across health systems. Approximately 1% of the records were identified and verified against the Cerner or Epic medical records of the respective health system for accuracy. These data were then de-identified for the final analyses.
The data were aggregated and analyzed using SPSS 26. The primary outcomes were the presence of any newborn adaptation syndrome, defined above, and NICU admission. Data were analyzed using chi-square and logistic regression analyses, controlling for maternal demographic variables. Because our objective was to guide patients and clinicians in their choice of antidepressants in pregnant women who needed pharmacological treatment, we did not have a control group of pregnant women who were depressed and not prescribed treatment. Thus, for our regression model, we used as a reference the drug with the lowest rates of the newborn adverse outcomes in the univariable analyses in general, bupropion. Regression results were reported in adjusted odds ratios (aOR) and 95% confidence intervals (CI). We also performed a logistic regression to explore the timing of drug exposure and associated impact on the main outcomes of any adaptation syndrome and NICU admission. For this analysis, early exposure was categorized as drug order(s) written within the 3 months before pregnancy began or in the first trimester. The first trimester exposure category was limited to women initially prescribed the antidepressant between the time of the last menstrual period and 14 weeks’ gestation, with no prior prescriptions in the system. Third trimester exposure was defined as anyone with a prescription written after 28 weeks’ gestation. All exposure time points were compared to a reference group of women prescribed a drug at some time during pregnancy but who did not have a prescription for any antidepressant drug during the relevant period. For example, for the analysis of first trimester exposure to paroxetine, the reference group would be women who took any antidepressant during pregnancy but did not have any prescriptions in the first trimester. The logistic regression models included maternal age, race, insurance status, and the gestational age at delivery.
This project was funded, in part, with support from the Indiana Clinical and Translational Sciences Institute, in part by UL1TR002529 from the National Institutes of Health, and in part by T32GM842528 (RS) from the NIH/NIGMS. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The remainder of the funding was provided by the local Obstetrics Department.
Results:
The final data included 3,694 women (Table 1). The distribution of drugs ordered included sertraline (n=1653), escitalopram (n=581), fluoxetine (n=579), bupropion (n=406), citalopram (n=385), duloxetine (n=139), venlafaxine (n=132), paroxetine (n=55), and desvenlafaxine (n=16). There were 252 women (6.8%) who were prescribed more than one antidepressant during their pregnancy, resulting in a total of 3,946 drug exposures. Typically, prescriptions were written for 30 days with varying numbers of refills. 86.6% (n=3200) of women in the study identified as white, 10.4% (n=384) as black, 0.5% (n=20) as Asian, 0.5% (n=18) identified as other, and 1.9% (n=72) did not identify their race. Hispanic ethnicity was reported in 3.4% (n=127) of women. Insurance was split between government (n=1796), commercial (n=1840), and self-pay or unknown (n=58). Of the mothers in the study, 176 (4.8%) had diabetes mellitus, 216 (5.8%) had hypertension, and 140 (3.8%) had history of preterm birth at baseline. There were 363 (9.8%) women who developed gestational diabetes and 612 (16.6%) who developed gestational hypertension. Difference in demographic characteristics and outcomes were noted for nearly all variables and outcomes by different SSRI or SNRI drug (Table 1 and Table 2). There were no differences in mean gestational age at delivery, newborn weight, length, or head circumference between drug groups (data not shown). The overall mean age of the population was 29.0 ± 5.8 years. The mean age for women taking each drug ranged from 28.3 to 30.8 years and was significantly different between drugs (p<0.001). Medication orders categorized as early exposure occurred for 43.9% of the women, exposure beginning in the first trimester occurred for 26.4% of the women, and third trimester exposure occurred for 48.7% of the women (Table 1).
Table 1.
Characteristics for women prescribed antidepressant(s) at any point throughout their pregnancy.
| Bupropion | Citalopram | Desvenlafaxine | Duloxetine | Escitalopram | Fluoxetine | Paroxetine | Sertraline | Venlafaxine | Overall cohort |
|
|---|---|---|---|---|---|---|---|---|---|---|
| Drug Frequency (women taking) | 406 | 385 | 16 | 139 | 581 | 579 | 55 | 1653 | 132 | 3694 |
| Mean maternal age (standard deviation), years | 30.0 (5.9) | 29.6 (5.6) | 29.4 (7.0) | 30.8 (5.3) | 29.3 (5.5) | 28.9 (6.0) | 28.3 (5.3) | 28.5 (5.8) | 29.5 (5.4) | 29.0 (5.8) |
| Trimester of Exposure | ||||||||||
| Early exposure | 162 (39.9%) | 190 (49.3%) | 6 (37.5%) | 61 (43.9%) | 266 (45.7%) | 260 (44.9%) | 30 (54.5%) | 742 (44.9%) | 61 (46.2) | 1621 (43.9) |
| 1st | 102 (25.1%) | 108 (28.0%) | 5 (31.2%) | 37 (26.6%) | 160 (27.5%) | 151 (26.1%) | 17 (30.9%) | 457 (27.6%) | 41 (31.1%) | 974 (26.4%) |
| 3rd | 172 (42.4%) | 178 (46.2%) | 8 (50.0%) | 59 (42.4%) | 283 (48.7%) | 275 (47.5%) | 10 (18.2%) | 883 (53.4%) | 57 (43.2%) | 1798 (48.7%) |
| Race * | ||||||||||
| Asian | 1 (0.2%) | 1 (0.3%) | 0 (0%) | 0 (0%) | 1 (0.2%) | 6 (1.0%) | 0 (0%) | 11 (0.7%) | 1 (0.8%) | 20 (0.5%) |
| Black | 30 (7.4%) | 36 (9.4%) | 1 (6.2%) | 14 (10.1%) | 45 (7.8%) | 73 (12.6%) | 5 (9.1%) | 196 (11.9%) | 6 (4.6%) | 384 (10.4%) |
| White | 373 (91.9%) | 340 (88.3%) | 15 (93.8%) | 119 (85.6%) | 528 (90.9%) | 483 (83.4%) | 49 (89.1%) | 1401 (84.8%) | 119 (90.2%) | 3200 (86.6%) |
| Other/Not Identified | 0 (0%) | 7 (1.8%) | 0 (0%) | 5 (3.6%) | 6 (1.0%) | 13 (2.2%) | 1 (1.8%) | 36 (2.2%) | 5 (3.8%) | 90 (2.4%) |
| Ethnicity * | ||||||||||
| Non-Hispanic/Latina | 399 (98.3%) | 369 (95.8%) | 15 (93.8%) | 138 (99.3%) | 569 (97.9%) | 555 (95.8%) | 53 (96.4%) | 1584 (95.8%) | 131 (99.2%) | 3567 (96.6%) |
| Hispanic/Latina | 7 (1.7%) | 16 (4.2%) | 1 (6.5%) | 1 (0.7%) | 12 (2.1%) | 24 (4.2%) | 2 (3.6%) | 69 (4.2%) | 1 (0.8%) | 127 (3.4%) |
| Hospital * | ||||||||||
| Eskenazi | 22 (5.4%) | 23 (6.0%) | 0 (0%) | 7 (5.0%) | 19 (3.3%) | 38 (6.6%) | 3 (5.4%) | 135 (8.2%) | 4 (3.0%) | 224 (6.1%) |
| IU Health | 384 (94.6%) | 362 (94.0%) | 16 (100%) | 132 (95.0%) | 562 (96.7%) | 541 (93.4%) | 52 (94.6%) | 1518 (91.8%) | 128 (97.0%) | 3470 (93.9%) |
| Insurance * | ||||||||||
| Commercial | 225 (55.4%) | 170 (44.2%) | 8 (50.0%) | 57 (41.0%) | 343 (59.0%) | 289 (49.9%) | 21 (38.2%) | 784 (47.4%) | 63 (47.7%) | 1840 (49.8%) |
| Government | 175 (43.1%) | 212 (55.1%) | 8 (50.0%) | 81 (58.3%) | 233 (40. %) | 278 (48.0%) | 33 (60.0%) | 837 (50.6%) | 68 (51.5%) | 1796 (48.6%) |
| Self-Pay/None | 6 (1.5%) | 3 (0.8%) | 0 (0%) | 1 (0.7%) | 5 (0.9%) | 12 (2.1%) | 1 (1.8%) | 32 (1.9%) | 1 (0.8%) | 58 (1.6%) |
| Maternal Comorbidities | ||||||||||
| Gestational Diabetes * | 49 (12.1%) | 34 (8.8%) | 0 (0%) | 25 (18.0%) | 46 (7.9%) | 56 (9.7%) | 5 (9.1%) | 157 (9.5%) | 16 (12.1%) | 363 (9.8%) |
| Gestational Hypertension * | 88 (21.7%) | 56 (14.6%) | 1 (6.2%) | 35 (25.2%) | 83 (14.3%) | 88 (15.2%) | 9 (16.4%) | 266 (16.1%) | 25 (18.9%) | 612 (16.6%) |
Data presented as n (% of with characteristic or outcome) in individual groups, unless otherwise indicated. 3,694 women were in the cohort, with 252 who were prescribed more than one drug, accounting for 3,946 drug prescriptions total. Early exposure was defined as drug order(s) within the 3 months before pregnancy began or in the early first trimester. First trimester exposure occurred between the time of the last menstrual period and 14 weeks’ gestation. Third trimester exposure occurred after 28 weeks’ gestation.
Statistically significant rate differences of outcome distribution between the drugs (p<0.05)
Table 2.
Pregnancy outcomes by antidepressant drug
| Bupropion N=406 |
Citalopram N=385 |
Desvenlafaxine N=16 |
Duloxetine N=139 |
Escitalopram N=581 |
Fluoxetine N=579 |
Paroxetine N=55 |
Sertraline N=1653 |
Venlafaxine N=132 |
Overall cohort N=3694 |
|
|---|---|---|---|---|---|---|---|---|---|---|
| Pregnancy Outcomes | ||||||||||
| Neonatal/Fetal Death | 3 (0.7%) | 2 (0.5%) | 0 (0%) | 0 (0%) | 1 (0.2%) | 0 (0%) | 0 (0%) | 6 (0.3%) | 0 (0%) | 11 (0.3%) |
| TTN * | 35 (8.6%) | 48 (12.5%) | 0 (0%) | 21 (15.1%) | 76 (13.0) | 83 (14.3%) | 9 (16.4%) | 172 (10.4%) | 24 (18.2%) | 437 (11.8%) |
| Newborn Intubated | 5 (1.2%) | 12 (3.1%) | 0 (0%) | 6 (4.3%) | 15 (2.6%) | 18 (3.1%) | 2 (3.6%) | 30 (1.8%) | 4 (3.0%) | 84 (2.3%) |
| Pulmonary Hypertension | 0 (0%) | 2 (0.5%) | 0 (0%) | 1 (0.7%) | 2 (0.3%) | 2 (0.4%) | 0 (0%) | 3 (0.2%) | 0 (0%) | 10 (0.3%) |
| Seizures | 1 (0.2%) | 2 (0.5%) | 0 (0%) | 0 (0%) | 3 (0.5%) | 5 (0.9%) | 1 (1.8%) | 5 (0.3%) | 0 (0%) | 15 (0.4%) |
| Jaundice | 63 (15.5%) | 55 (14.3%) | 3 (18.8%) | 21 (15.1%) | 90 (15.5%) | 91 (15.7%) | 6 (10.9%) | 228 (13.8%) | 18 (13.6%) | 548 (14.8%) |
| Cardiac Malformations | 13 (3.2%) | 19 (4.9%) | 0 (0%) | 6 (4.3%) | 26 (4.5%) | 25 (4.3%) | 4 (7.3%) | 64 (3.9%) | 3 (2.3%) | 153 (4.1%) |
| 5 Min APGAR <7 | 50 (12.3%) | 56 (14.6%) | 3 (18.8%) | 22 (15.8%) | 66 (11.4%) | 99 (17.1%) | 9 (16.4%) | 213 (12.9%) | 20 (15.2%) | 504 (13.6%) |
| Any Adaptation Syndrome * | 19 (4.7%) | 35 (9.1%) | 1 (6.2%) | 21 (15.1%) | 51 (8.8%) | 34 (5.9%) | 7 (12.7%) | 75 (4.5%) | 9 (6.8%) | 228 (6.2%) |
| NICU Admission * | 91 (22.4%) | 105 (27.3%) | 3 (18.8%) | 55 (39.6%) | 172 (29.6%) | 162 (28.0%) | 16 (29.1%) | 384 (23.2%) | 33 (25.0%) | 943 (25.5%) |
| Preterm Birth | 61 (15.0%) | 56 (14.6%) | 2 (12.5%) | 32 (23.0%) | 85 (14.6%) | 100 (17.3%) | 8 (14.6%) | 238 (14.3%) | 18 (13.6%) | 559 (15.1%) |
Data presented as n (% of with characteristic or outcome) in individual groups.
Statistically significant rate differences of outcome distribution between the drugs (p<0.05)
TTN= transient tachypnea of the newborn; NICU = neonatal intensive care unit
The drug associated with the highest rates for the most adverse newborn outcomes was duloxetine (Table 2). The lowest rates for most of the adverse newborn outcomes among drugs with at least 200 noted prescriptions was for bupropion, followed by sertraline. For univariable comparison of newborn outcomes (Table 2), NICU admission (p<0.001), TTN (p=0.006), and any adaptation syndrome (p<0.001) were significantly different between drugs, with duloxetine having the highest rates of NICU admission (39.6%) and any adaptation syndrome (15.1%). Paroxetine was also associated with a high rate of any adaptation syndrome (12.7%). Venlafaxine was associated with the highest rate of TTN (18.2%).
Given that bupropion exhibited low rates for the most newborn adverse outcomes, it was used as the reference drug for the logistic regression comparing different drugs (Table 3, Table S2). Univariable comparisons to bupropion for each newborn outcome are listed in Supplemental Table 2. Controlling for maternal age, race, insurance, and estimated gestational age at delivery, rates of any adaptation syndrome were increased with citalopram (aOR 1.81 [95% CI 1.00-3.27, did not cross 1.00]), duloxetine (aOR 3.04 [95% CI 1.54-5.98]), escitalopram (aOR 2.26 [95% CI 1.30-3.95]), and paroxetine (aOR 2.60 [95% CI 1.01-6.71]) (Table 3). Increased NICU admission rates were seen with duloxetine (aOR 2.04 [95% CI 1.29-3.23]) and escitalopram (aOR 1.68 [95% CI 1.21-2.33]). Adjusting for maternal age, race, and insurance status showed no differences in the rate of preterm birth amongst the drugs. The regression model results were very similar if gestational age at birth was removed from the model.
Table 3.
Logistic regression of individual drugs’ association with newborn outcomes
| Unadjusted ORs | Adjusted ORs | |||
|---|---|---|---|---|
| Drug | Any Adaptation Syndrome |
NICU Admission | Any Adaptation Syndrome |
NICU Admission |
| Bupropion | Reference | Reference | Reference | Reference |
| Citalopram | 0.67 (0.30-1.52) | 0.87 (0.55-1.37) | 1.81 (1.00-3.27)* | 1.41 (0.98-2.03) |
| Desvenlafaxine | 1.37 (0.64-2.93) | 1.13 (0.72-1.77) | 1.13 (0.13-9.54) | 0.98 (0.26-3.74) |
| Duloxetine | 0.91 (0.11-7.70) | 0.69 (0.19-2.58) | 3.04 (1.54-5.98)* | 2.04 (1.29-3.23)* |
| Escitalopram | 2.43 (1.07-5.23)* | 1.96 (1.17-3.31)* | 2.26 (1.30-3.95)* | 1.68 (1.21-2.33)* |
| Fluoxetine | 1.32 (0.63-2.74) | 1.26 (0.82-1.94) | 1.30 (0.72-2.34) | 1.34 (0.96-1.87) |
| Paroxetine | 0.85 (0.40-1.82) | 1.17 (0.76-1.80) | 2.60 (1.01-6.71)* | 1.64 (0.83-3.26) |
| Sertraline | 1.99 (0.70-5.65) | 1.23 (0.61-2.49) | 0.96 (0.57-1.63) | 1.08 (0.81-1.45) |
| Venlafaxine | 0.65 (0.32-1.33) | 0.91 (0.60-1.37) | 1.33 (0.58-3.06) | 1.17 (0.70-1.94) |
Adjusted logistic regression controlled for maternal age, race, insurance status, and estimated gestational age at birth.
Results reported as odd ratio (OR) (95% confidence interval).
Confidence interval does not cross 1.0, indicating a statistically significant result
NICU = neonatal intensive care unit
Table 4 displays the comparison of antidepressants associated with any adaptation syndrome and NICU admission in respect to timing of drug exposure. Compared to women who were not prescribed any antidepressant(s) during the early pregnancy period, exposure to duloxetine and escitalopram early in pregnancy had increased odds of any adaptation syndrome (aOR 2.31, [95% CI 1.11-4.80] and aOR 1.72, [95% CI 1.09-2.71], respectively) and NICU admission (aOR 2.47, [95% CI 1.40-4.34] and aOR 1.64 [95% CI 1.21-2.22], respectively). Limiting exposure to women with initial prescription(s) written in the first trimester, the only significant associations were duloxetine and escitalopram with NICU admission (aOR 2.32 [95% CI 1.13-4.77] and aOR 1.56 [95% CI 1.06 – 2.28], respectively). For third trimester exposure, citalopram, duloxetine, escitalopram, fluoxetine, sertraline, and venlafaxine were associated with increased odds of adaptation syndrome, while bupropion, citalopram, duloxetine, escitalopram, and fluoxetine were associated with increased odds of NICU admission, compared to women who did not have third trimester exposures (Table 4).
Table 4.
Logistic regression of antidepressant exposure at different stages of pregnancy associated with any adaptation syndrome and neonatal intensive care unit admission.
| Drug | Early exposure | First trimester exposure | Third trimester exposure | |||
|---|---|---|---|---|---|---|
| Adaptation syndrome |
NICU admission |
Adaptation syndrome |
NICU admission |
Adaptation syndrome |
NICU admission |
|
| Bupropion | 0.62 (0.28-1.37) | 0.76 (0.50-1.17) | 0.47 (0.17-1.30) | 0.68 (0.39-1.17) | 1.29 (0.65-2.58) | 1.51 (1.03-2.21) |
| Citalopram | 0.83 (0.43-1.58) | 1.20 (0.82-1.75) | 0.81 (0.34-1.91) | 1.09 (0.67-1.79) | 3.02 (1.82-5.04) | 1.61 (1.11-2.34) |
| Desvenlafaxine | 2.57 (0.26-25.36) | 0.61 (0.06-5.89) | 3.71 (0.33-41.62) | 0.72 (0.07-7.56) | --- * | 0.52 (0.06-4.38) |
| Duloxetine | 2.31 (1.11-4.80) | 2.47 (1.40-4.34) | 1.94 (0.72-5.25) | 2.32 (1.13-4.77) | 4.76 (2.27-9.91) | 2.71 (1.52-4.83) |
| Escitalopram | 1.72 (1.09-2.71) | 1.64 (1.21-2.22) | 1.69 (0.97-2.95) | 1.56 (1.06-2.28) | 4.20 (2.76-6.38) | 1.93 (1.44-2.59) |
| Fluoxetine | 0.92 (0.52-1.62) | 1.02 (0.74-1.43) | 0.52 (0.21-1.30) | 0.92 (0.60-1.43) | 2.08 (1.27-3.42) | 1.45 (1.07-1.98) |
| Paroxetine | 1.88 (0.62-5.67) | 0.85 (0.32-2.22) | 1.75 (0.37-8.21) | 0.86 (0.25-2.92) | 3.26 (0.65-16.23) | 1.81 (0.49-6.72) |
| Sertraline | 0.68 (0.46-1.01) | 0.93 (0.75-1.16) | 0.66 (0.40-1.07) | 0.88 (0.68-1.15) | 1.56 (1.09-2.24) | 1.15 (0.94-1.41) |
| Venlafaxine | 1.40 (0.54-3.67) | 1.03 (0.54-1.96) | 0.77 (0.18-3.32) | 1.15 (0.54-2.41) | 3.95 (1.74-8.97) | 1.53 (0.83-2.81) |
All results reported as adjusted odds ratio (aOR) (95% confidence interval) controlling for maternal race, age, insurance, and gestational age at delivery.
Not enough data for this outcome to run analysis
NICU = neonatal intensive care unit; Early exposure = prescription for drug(s) in the 3 months before pregnancy began or in the early first trimester; First trimester exposure = prescription for drug(s) between the time of the last menstrual period and 14 weeks’ gestation; Third trimester exposure = prescription for drug(s) after 28 weeks’ gestation.
All comparisons to a reference group of women who took an antidepressant at some point during pregnancy but did not have a prescription for any antidepressant during the relevant period.
Bold text added to highlight statistically significant values.
Discussion:
Main Findings
Our study found that compared to other antidepressants, duloxetine showed the strongest associations with maternal comorbidities and increased odds of many adverse neonatal outcomes. This was the case with women who received duloxetine prescriptions both early and later in pregnancy when compared to women not prescribed antidepressants during those times. Two drugs (duloxetine and escitalopram), followed by citalopram, appeared to be more consistently associated with higher rates of the adverse newborn outcomes compared to bupropion. We are unaware of other studies comparing individual SSRI and SNRI drugs in pregnancy in this manner.3 Other studies have found differences in clinical efficacy of different antidepressants during pregnancy.6
We also identified different prescribing patterns for individual SSRI and SNRI drugs by race, insurance, and hospital system. One possible explanation for this is that different providers typically have “preferred” drugs or that certain practices or systems may have protocols or formularies for first-line therapy. As the county hospital for Marion County, Eskenazi Hospital sees a high proportion of indigent patients, which may also explain the differences seen between women receiving care in that system versus Indiana University Health.
We found low rates of adverse neonatal outcomes associated with bupropion and sertraline. Although sertraline is often utilized in pregnancy3 and bupropion had fewer prescriptions in comparison (n=406 bupropion versus n=1653 sertraline), bupropion had favorable outcomes in general. Bupropion was chosen as a reasonable reference drug as it had the lowest rates of adverse outcomes in our data as well as no literature citing adverse outcomes in pregnancy.7
Because our goal was to provide information for clinicians and patients deciding which antidepressant to prescribe when medication treatment was necessary during pregnancy, we did not include a control group of women who were depressed but were not prescribed drug treatments. Thus, all women included in the final study population were evaluated and found to be beginning or continuing drug therapy for depression symptoms. Women who were diagnosed with depression but were not prescribed drug treatment would likely have a different severity of disease than women prescribed drug therapy and thus could have a different risk profile for the adverse outcomes. For this reason, we chose the drug with the lowest rates of adverse outcomes in the univariable analysis (bupropion) as the reference drug for the overall comparative analysis. We used a reference group of women who lacked order(s) for drug(s) of interest during the three time periods for the regression analysis respective to timing of exposure.
Interpretation
To our knowledge, other studies have not reported higher rates of any adaptation syndrome and NICU admission associated with mothers taking duloxetine or higher rates of TTN associated with mothers taking venlafaxine. Further investigations are warranted to replicate these findings as well as to explore potential pharmacological mechanisms that explain these results. Although cardiac malformation rate did not show statistically significant differences between drugs, the incidence rate among this group of women taking antidepressants was higher than that of the general population as published by Center for Disease Control (4.1% vs. <1%).8 Our data are consistent with recent evidence that some SSRIs are associated with higher rates of birth defects, including congenital heart defects.9
The different rates of risks observed between individual antidepressants may be attributed to drug-specific pharmacokinetic and pharmacodynamic properties. Citalopram, escitalopram, fluoxetine, paroxetine, and sertraline are selective for the 5-hydroxytryptamine (serotonin) transporter (SERT). Venlafaxine, desvenlafaxine, and duloxetine have high affinities for both SERT and the norepinephrine transporter (NET). Paroxetine, escitalopram, and duloxetine have extremely high affinities for SERT, which may contribute to the higher risk profiles of escitalopram and duloxetine.10-12 The small number of women taking paroxetine in our study may have prevented us from observing an increased risk in this population. Bupropion, on the other hand, inhibits dopamine and norepinephrine reuptake but has low affinity for SERT13, which may explain its lower risk profile observed in our study. Additionally, it appears that drugs with shorter half-lives, such as duloxetine (12 hours) or venlafaxine (5 hours) and its active metabolite desvenlafaxine (12 hours), are associated with higher risk for poor neonatal outcomes than drugs with half-lives greater than 20 hours.14 The potential association of poor neonatal clinical outcomes with shorter half-life antidepressant drugs should be explored further.
Strengths and limitations
Our study used a large electronic database representing current practice in the health system in Indiana. The large number of women receiving different antidepressant drugs allowed for analysis of less common outcomes, an important contributor to the literature. Additionally, the ability to compare clinical outcomes from women with trimester-specific exposure to women without drug exposure during that same trimester can help providers during counseling of women who need antidepressants in different periods of pregnancy.
Our study was limited by potential coding errors encountered by utilizing data from large EMR-based data sets. We verified a small proportion of obtained data and found the variables to be consistent with manual review of the EMR. We cannot rule out that some of the data may have been under-captured due to the methodology of using ICD9/10 codes. However, the rates of outcomes we obtained (preterm birth (15.1%) and NICU admission (25.5%)) were comparable to other published reports of women treated with antidepressants15, although our rate of any adaptation syndrome (6.2%) may be lower than previously reported.16 The coding for any adaptation syndrome combined diagnoses for NAS and PAS in response to recent changes in the diagnoses associated with SSRIs. Therefore, rates of this outcome could be artificially high as some of these newborns were possibly suffering from NAS associated with opioid use. However, there is also potential for under-capturing the outcome due to reliance on administrative data. Prescription records were pulled from pharmacy data and did not specify if the prescription was filled and/or taken, so it is possible that some women filled their prescription but did not take it. However, the manually-verified records supported our assumption that provider notes documented medication use. Although this does not guarantee women actually took the drug, for example due to potential social acceptability bias, we believe the assumption of defining exposure by prescriptions received is justified for these data. Although a prospective study could use technology to ensure women were taking the pills prescribed, as a retrospective study, this assumption is the closest approximation to exposure possible. Because we did not have dosage information, our findings are limited in that our associations may be influenced by women taking higher or lower doses of the prescribed antidepressant(s). We were also not able to capture complementary and alternative medicine therapies used to treat depression. Additionally, generalizability of our results may be limited as our proportion of Hispanic women was smaller than the general population. Finally, this study likely was not powered for individual drugs to find differences for rarer adverse outcomes.
Conclusion
In conclusion, individual SSRI/SNRI agents may have increased rates of adverse newborn outcomes compared to others. The risks may also differ based upon when exposure occurred during pregnancy. Compared to other antidepressants in the study, duloxetine appears to be associated with higher rates of adverse neonatal outcomes. These findings can better inform providers and patients during clinical decision making when choosing antidepressant drug therapy for depression during pregnancy to optimize both maternal and newborn outcomes.
Supplementary Material
Acknowledgements/Disclosures of interest:
This project was funded, in part, with support from the Indiana Clinical and Translational Sciences Institute funded, in part by UL1TR002529 from the National Institutes of Health, and in part by T32GM842528 (RS) from the NIH/NIGMS. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The remainder of the funding was provided by the OB/GYN Department at IUSM. The authors do not report any other disclosures of interest.
Footnotes
Details of Ethics Approval
This study was a retrospective cohort study. It was approved by the Indiana University-Purdue University Institutional Review Board and the Regenstrief Institute Data Core approval committee as exempt. It was protocol number 2006055550 and was approved June 8, 2020.
References:
- 1.Bérard A, Iessa N, Chaabane S, Muanda FT, Boukhris T, Zhao JP. The risk of major cardiac malformations associated with paroxetine use during the first trimester of pregnancy: a systematic review and meta-analysis. Br J Clin Pharmacol. 2016;81(4):589–604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Van Niel MS, Payne JL. Perinatal depression: A review. Cleve Clin J Med. 2020;87(5):273–277. [DOI] [PubMed] [Google Scholar]
- 3.Yonkers KA, Wisner KL, Stewart DE, et al. The management of depression during pregnancy: a report from the American Psychiatric Association and the American College of Obstetricians and Gynecologists. Obstet Gynecol. 2009;114(3):703–713. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Golichowski AM, McDonald CJ, Tierney WM, et al. Managing perinatal data with the Regenstrief medical record system. J Ambul Care Manage. 1992;15(4):40–53. [DOI] [PubMed] [Google Scholar]
- 5.McDonald CJ, Overhage JM, Tierney WM, et al. The Regenstrief Medical Record System: a quarter century experience. Int J Med Inform. 1999;54(3):225–253. [DOI] [PubMed] [Google Scholar]
- 6.Cipriani A, Furukawa TA, Salanti G, et al. Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis. Lancet. 2009;373(9665):746–758. [DOI] [PubMed] [Google Scholar]
- 7.Wellbutrin (bupropion hydrochloride) Prescribing Information. In: US Food and Drug Association; https://www.accessdata.fda.gov/drugsatfda_docs/label/2011/018644s043lbl.pdf. [Google Scholar]
- 8.Data and Statistics on Congenital Heart Defects. https://www.cdc.gov/ncbddd/heartdefects/data.html#:~:text=CHDs%20affect%20nearly%201%25%20of,year%20in%20the%20United%20States.&text=The%20prevalence%20(the%20number%20of,other%20types%20has%20remained%20stable. Accessed July 1, 2020.
- 9.Anderson KN, Lind JN, Simeone RM, et al. Maternal Use of Specific Antidepressant Medications During Early Pregnancy and the Risk of Selected Birth Defects. JAMA Psychiatry. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rao N. The clinical pharmacokinetics of escitalopram. Clin Pharmacokinet. 2007;46(4):281–90. [DOI] [PubMed] [Google Scholar]
- 11.Owens MJ, Morgan WN, Plott SJ, Nemeroff CB. Neurotransmitter receptor and transporter binding profile of antidepressants and their metabolites. J Pharmacol Exp Ther. 1997;283(3):1305–22. [PubMed] [Google Scholar]
- 12.Bymaster FP, Dreshfield-Ahmad LJ, Threlkeld PG, Shaw JL, Thompson L, Nelson DL, et al. Comparative affinity of duloxetine and venlafaxine for serotonin and norepinephrine transporters in vitro and in vivo, human serotonin receptor subtypes, and other neuronal receptors. Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology. 2001;25(6):871–80. [DOI] [PubMed] [Google Scholar]
- 13.Stahl SM, Pradko JF, Haight BR, Modell JG, Rockett CB, Learned-Coughlin S. A Review of the Neuropharmacology of Bupropion, a Dual Norepinephrine and Dopamine Reuptake Inhibitor. Prim Care Companion J Clin Psychiatry. 2004;6(4):159–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Harvard Women's Health Watch. Going off antidepressants [updated March 25, 2020. Available from: https://www.health.harvard.edu/diseases-and-conditions/going-off-antidepressants. Accessed February 15, 2021.
- 15.Suri R, Altshuler L, Hellemann G, Burt VK, Aquino A, Mintz J. Effects of antenatal depression and antidepressant treatment on gestational age at birth and risk of preterm birth. Am J Psych 2007;164:1206–1213. [DOI] [PubMed] [Google Scholar]
- 16.Byatt N, Deligiannidis KM, Freeman MP. Antidepressant use in pregnancy: a critical review focused on risks and controversies. Acta Psychiatr Scand. 2013;127(2):94–114. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.