Article Text
Abstract
Objective Selective serotonin reuptake inhibitor (SSRI) use is common in pregnancy. It is associated with delayed neonatal adaptation. Most previous studies have not adjusted for the severity of maternal mental health disorders or examined the impact of SSRI type and dosage. We examined whether treatment with SSRIs in late pregnancy (after 20 weeks) is associated with delayed neonatal adaptation independent of maternal depression and anxiety.
Design, setting and patients Retrospective population-based birth cohort of 280 090 term infants born at 15 Kaiser Permanente Northern California hospitals, 2011–2019. Individual-level pharmacy, maternal, pregnancy and neonatal data were obtained from electronic medical records.
Exposure Dispensed maternal SSRI prescription after 20 weeks of pregnancy.
Main outcome measures Delayed neonatal adaptation defined as a 5 min Apgar score ≤5, resuscitation at birth or admission to a neonatal intensive care unit for respiratory support. Secondary outcomes included each individual component of the primary outcome and more severe neonatal outcomes (pulmonary hypertension, hypoxic-ischaemic encephalopathy and seizures).
Results 7573 (2.7%) infants were exposed to SSRIs in late pregnancy. Delayed neonatal adaptation occurred in 11.2% of exposed vs 4.4% of unexposed infants (relative risk 2.52 (95% CI 2.36 to 2.70)). After multivariable adjustment, there was an association between SSRI exposure and delayed neonatal adaptation (adjusted OR 2.14 (95% CI 1.96 to 2.32)). This association was dose dependent. Escitalopram and fluoxetine were associated with the highest risk of delayed neonatal adaptation.
Conclusions Infants exposed to SSRIs have increased risks of delayed adaptation in a type and dose-dependent relationship, pointing toward a causal relationship.
- Intensive Care Units, Neonatal
- Mental health
- Neonatology
- Epidemiology
- Resuscitation
Data availability statement
Data are available upon reasonable request. The datasets generated for this study are stored at the KPNC Division of Research. Deidentified data can be provided upon reasonable request to the corresponding author, and with permission from the KPNC Institutional Review Board.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
In the USA, 4–6% of mothers are prescribed selective serotonin reuptake inhibitors (SSRIs) during pregnancy. SSRI use during pregnancy is associated with neonatal adaptation syndrome. Neonatal adaptation syndrome encompasses a wide range of different clinical presentations.
WHAT THIS STUDY ADDS
Low-dose sertraline exposure is associated with a doubling in the odds of delayed neonatal adaptation independent of maternal mental health diagnoses. The odds of delayed adaptation increase with increasing doses of SSRI and are higher for neonates on other types of SSRIs than sertraline.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study quantifies the additional risks of delayed adaptation in neonates exposed to SSRIs at various doses and types. It provides data for parents and providers to discuss SSRI treatment during pregnancy and the management of infants after birth.
Introduction
Depression and anxiety affect one in five pregnant individuals,1 and a quarter of them are treated with selective serotonin reuptake inhibitors (SSRIs).2 The frequency of SSRI treatment during pregnancy increased by one-third in the last decade in the USA.2
In making decisions regarding treatment for prenatal depression and anxiety, clinicians and women must consider the benefits of treatment as well as potential adverse effects. Unfortunately, the adverse effects of SSRI exposure in utero are poorly quantified.3 Some studies have shown in utero SSRI exposure to be associated with delayed neonatal adaptation4 5 manifested by lower Apgar scores,6 7 respiratory distress8–10 and persistent pulmonary hypertension of the newborn11 12 leading to higher neonatal intensive care unit (NICU) admission rates.4 13 Transient, self-limited symptoms such as abnormal tone,14–17 jitteriness and irritability18 have also been described. The risk of more severe neurological conditions such as hypoxic-ischaemic encephalopathy (HIE)19 and seizures20 21 is unclear.
Past studies are limited by small sample size,5 10 16 19 ascertainment bias of both exposure and outcome of interest,4 5 17 19 reliance on administrative databases8 9 and the inability to differentiate the adverse effects of SSRIs themselves from the impact of maternal anxiety and depression on neonatal outcomes.9 10 22 23 Data on the relationship between the type and dose of SSRI treatment and adverse outcome are also needed. We conducted a population-based cohort study of mother–infant dyads, to evaluate the relationship between maternal SSRI treatment and delayed neonatal adaptation, adjusting for maternal depression and anxiety, and evaluating the impact of SSRI dosage and type.
Methods
Study population and setting
This cohort study included infants born at ≥37 weeks’ gestational age at 15 Kaiser Permanente Northern California (KPNC) hospitals between 1 November 2011 and 31 July 2019, whose mothers received prenatal care in KPNC. We excluded infants with congenital anomalies or genetic abnormalities. KPNC is an integrated healthcare system serving over 4.6 million members, representing approximately 40% of the insured population in Northern California. The sociodemographic distribution of the KPNC membership is broadly similar to the Northern California population, though the extremes of the income distribution are under-represented.24
Exposure
The primary exposure was any maternal SSRI dispensing by a KPNC pharmacy during late pregnancy, that is, after 20 weeks’ gestation.25 We identified all maternal dispensing for SSRIs (ie, sertraline, fluoxetine, citalopram, escitalopram, paroxetine, dapoxetine, vortioxetine and fluvoxamine) during pregnancy. KPNC pharmacies provide more than 96% of prescription medications to their members.26 For each prescription, we collected dispensation date, daily dosage and number of pills dispensed. We converted SSRI prescriptions into sertraline equivalent dosage using previously published conversion factors.27 We categorised SSRI doses into four categories with clinically meaningful cut-offs at 50, 100 and 150 mg/day of sertraline equivalent.
Covariates
We ascertained maternal inpatient and outpatient International Classification of Diseases (ICD) diagnoses of depression and anxiety during pregnancy (online supplemental file 1) from existing KPNC databases.28 We classified the severity of depression symptoms using each mother’s maximal Patient Health Questionnaire–9 (PHQ-9) score as follows: none or mild (0–9), moderate (10–14), moderate to severe (15–19), severe (20–27).29 We extracted demographic and clinical data from existing databases30: birth year; maternal age, self-reported race/ethnicity and parity; state-subsidised insurance, quartile of neighbourhood deprivation index; maternal type 2 diabetes, chronic hypertension and obesity (pre-pregnancy body mass index ≥30); mode of delivery, infant sex, birth weight and gestational age. In KPNC, universal urine drug screening is a component of the prenatal evaluation; we defined a positive pregnancy toxicology screen as the presence of methadone, cocaine, amphetamine or opiates and separately assessed exposure to cannabinoids based on urine screen. Finally, we examined maternal dispensing for non-SSRI antidepressants and antipsychotics at any time during pregnancy (online supplemental file 2).
Supplemental material
Outcomes
Delayed neonatal adaptation, the primary composite outcome, was defined as the presence of at least one of the following: 5 min Apgar score ≤5; positive pressure ventilation or intubation during delivery room resuscitation; or admission to the NICU with respiratory distress requiring invasive or non-invasive ventilation (including high-flow nasal cannula ≥2 L/min) during the birth hospitalisation. Secondary outcomes included each individual component of the primary composite outcome and more severe neonatal outcomes: persistent pulmonary hypertension of the newborn (ICD-9 747.83; ICD-10 P29.3 or treatment with nitric oxide), culture-proven neonatal sepsis, HIE (ICD-9 768.5–768.7 or 768.9; ICD-10 P21.0 or P91.6; or diagnosis of HIE based on chart abstraction), seizures (ie, electrographic seizures or focal-clonic events31) based on chart review of all infants with a seizure diagnosis (ICD-9 779.0, ICD-10 P90) or who received an antiepileptic medicine before day 3 after birth, treatment with therapeutic hypothermia, length of hospital stay greater than 90th percentile for mode of delivery (57 hours for vaginal delivery; 94 hours for caesarean delivery) and transfer to another hospital for higher level of care.
Statistical analysis
We calculated unadjusted relative risks and 95% CIs for our primary and secondary outcomes, comparing infants exposed and unexposed to SSRIs using Χ2. We used multiple logistic regression to evaluate the adjusted association between SSRI exposure and our primary and secondary outcomes. Our final model included clinically relevant covariates, which could not be mediators or colliders, associated with both exposure to SSRI and our primary composite outcome, with an unadjusted p value of <0.10 (online supplemental file 3). Marginal adjusted risk differences and 95% CIs for our primary and secondary outcomes were calculated based on the same multiple logistic regression models.32 All regression models used robust variance estimates using the clustered sandwich estimator, which accounts for the clustering of births within hospitals. Secondary analyses evaluated the associations between type and doses of SSRI using the same multiple logistic regression model described above. The rates of delayed neonatal adaptation vary by mode of delivery, infant sex and gestational age. In our main analysis, we did not adjust for those variables as they could be in the pathway between SSRI exposure and adverse outcome. However, we performed a stratified sensitivity analysis to assess their differential effect. We further performed two sensitivity analyses: one including each substance use from the toxicology screen as individual covariates and the other one removing maternal mental health diagnoses from the covariate list. Analyses were performed using Stata, V.17 (StataCorp, College Station, Texas, USA).
Results
Of 280 682 infants born at ≥37 weeks’ gestation, we excluded 592 with congenital or genetic abnormalities. Among 280 090 infants in the final study cohort, 7573 (2.7%) were exposed to SSRIs in late pregnancy. Maternal characteristics associated with SSRI treatment are described in table 1. Depression or anxiety during pregnancy was reported in 12% of the population, and an additional 5% had a maximum PHQ-9 score ≥10, indicating the presence of depressive symptoms.
Delayed neonatal adaptation occurred in 11.2% of exposed infants and 4.4% of the unexposed (relative risk 2.52 (95% CI 2.36 to 2.70)). After adjustment using multivariable logistic regression (table 2), SSRI exposure remained associated with delayed adaptation (adjusted OR (aOR) 2.14 (95% CI 1.96 to 2.32)). After accounting for SSRI exposure, maternal anxiety remained weakly associated with delayed neonatal adaptation (aOR 1.38 (95% CI 1.29 to 1.48)), but neither a maternal diagnosis of depression nor depression severity scores were associated with delayed neonatal adaptation. In sensitivity analysis, stratifying by mode of delivery, gestational age and infant sex, exposed infants in all strata had a 3.7–6.6% increased absolute risk of delayed neonatal adaptation compared with unexposed infants (online supplemental file 4). On sensitivity analysis, including each substance used from the toxicology screen as individual covariates or removing maternal mental health diagnoses from the covariate list did not change the results (online supplemental file 5).
Exposure to SSRIs was associated with all individual components of the composite primary outcome (table 3). We did not find a statistically significant association between SSRI exposure and pulmonary hypertension nor treatment with nitric oxide. While there was an association between SSRI exposure and HIE on univariable analysis, this association was not significant after adjusting for confounders (aOR 1.45 (95% CI 0.86 to 2.44)). Similarly, there was an association between SSRI exposure and treatment with therapeutic hypothermia (aOR 1.55 (95% CI 1.00 to 2.38)) of borderline statistical significance. There was no association between SSRI exposure and neonatal seizures (table 3). Neonates exposed to SSRI had longer hospital stays and were more likely to be transferred to a higher level of care facility (aOR 1.44 (95% CI 1.15 to 1.81); table 3).
Types and doses of SSRI prescriptions are described in table 4. Sertraline was the most common SSRI prescribed (70.2% of all infants). While ranges of sertraline equivalent exposure were similar for all SSRI types, the distribution varied among types. There was a dose–response relationship between sertraline equivalent daily dose and delayed neonatal adaptation (test for trend p<0.001; table 5). All SSRI types were associated with increased odds of delayed neonatal adaptation. When stratified by dose, all other types of SSRI were associated with increased odds of delayed neonatal adaptation compared with sertraline (table 5). In the subgroup of 865 mothers who discontinued SSRI between 20 and 30 weeks of gestation, the risk of delayed neonatal adaptation was not increased compared with unexposed mothers (OR 0.99 (95% CI 0.68 to 1.44)).
Discussion
In this large population-based cohort with detailed maternal, pregnancy and neonatal information, exposure to SSRIs in late pregnancy was associated with a doubling in the odds of delayed neonatal adaptation after adjusting for maternal anxiety and depression. The strength of the association between SSRI exposure and delayed neonatal adaptation was dose and type dependent.
Mothers exposed to SSRIs in our cohort not only had higher rates of mental health diagnoses but higher rates of obesity, diabetes, hypertension and advanced age, which are all risk factors for delayed neonatal adaptation. We also found that maternal anxiety, but not depression, was independently associated with higher odds of delayed neonatal adaptation. The pathophysiology underlying the role of maternal anxiety on neonatal outcomes is unclear and deserves further investigation.
Neonatal adaptation syndrome was first described with fluoxetine exposure in 1996.33 Its definition varies among studies4 34 and can include delayed neonatal adaptation, respiratory distress and transient neurological abnormalities such as jitteriness, tremors and irritability.4 5 In this study, we focused on delayed neonatal adaptation as an increased need for neonatal respiratory support in the first minutes, hours or days after birth. We did not examine transient neurological signs because they do not require intervention and are not reliably documented for all infants. In contrast with some previous studies,19 20 we did not find an association between SSRI exposure and HIE or seizures. The absence of association between SSRI exposure and seizures is consistent with a recent Swedish study.21 In contrast, a meta-analysis reported a fivefold increased risk of seizures with SSRI exposure. However, our study used a more stringent seizure definition based on recent guidelines,31 requiring documentation of electrographic seizures or focal clonic or tonic movements, thus avoiding misclassification of non-epileptic movements as seizures.
There were borderline increased odds of treatment with therapeutic hypothermia. Similarly, a case–control study reported infants treated with therapeutic hypothermia were five times more likely to be exposed to SSRIs.19 The increased risk of therapeutic hypothermia may be due to the overlap in symptomatology between delayed neonatal adaptation caused by SSRI exposure and mild HIE, leading to treatment with therapeutic hypothermia.35 While the symptoms infants exposed to SSRIs may exhibit can mirror mild HIE, they are not due to disrupted blood and oxygen supply and may not benefit from therapeutic hypothermia.
There was a trend toward an increased risk of pulmonary hypertension that did not reach statistical significance. After adjusting for other risk factors, more than 3000 infants would have to be exposed to SSRI for one additional case of pulmonary hypertension. Two meta-analyses reported an association between late SSRI exposure and pulmonary hypertension,4 36 but did not adjust for maternal characteristics. A previous large study11 that similarly adjusted their analysis for maternal depression and anxiety also found no association. These findings highlight the need to adjust for maternal characteristics when assessing causality.
Adverse neonatal effects from prenatal SSRI exposure must be balanced against the benefits of maternal mental health treatment to both the mother and the infant.37 Among pregnant individuals, suicide resulting from undertreated mental health disorders is a leading cause of death.38 Untreated depression and anxiety are associated with low birth weight, prematurity and need for NICU admission.39 SSRIs are considered an effective treatment for maternal depression and anxiety during pregnancy, although efficacy has not been evaluated with a randomised controlled trial. Assuming causality, treatment of 22 pregnant individuals with SSRIs would result in one additional case of delayed neonatal adaptation, treatment of 40 individuals would incur one additional NICU admission and treatment of more than 500 individuals would result in one additional case of mechanical ventilation, pulmonary hypertension or HIE.
The dose–response relationship between SSRI exposure and delayed neonatal adaptation points toward a causal relationship; however, the mechanism is unclear. Placental transfer of SSRIs is substantial, suggesting a possible direct toxic effect on the fetal lungs and central nervous system.40 Increased levels of 5-hydroxytryptamine in the fetus could alter respiratory maturation or adaptation to the extrauterine environment, leading to neonatal respiratory distress.5 However, plasma levels of SSRIs in affected newborns do not correlate with symptom severity.5 40 In our study, while all SSRI types were associated with delayed adaptation, citalopram, fluoxetine and escitalopram were associated with higher risks than sertraline. When exposure was discontinued before 30 weeks, there was no increased risk of delayed neonatal adaptation, suggesting a time-dependent mechanism. Further studies are needed to understand the mechanism behind this differential safety profile.
Strengths of this study include the large population, high-quality assessment of SSRI exposure and maternal mental health diagnoses, and ability to adjust for confounders. One limitation of our study is that mothers may have received SSRI prescriptions outside of KPNC or may not have taken the dispensed medication; this non-differential misclassification of exposure would bias our results toward the null. Although we adjusted for PHQ-9 scores, some residual confounding by the severity of depression or anxiety may persist. It is unlikely that large-scale studies will have more detailed information regarding depression severity. We did not evaluate the effect of sudden discontinuation versus continuous SSRI exposure after birth via breast milk. While this may affect the timing or duration of neurological symptoms, it would not affect delayed neonatal adaptation at birth. Finally, the study was limited to short-term neonatal outcome of SSRI exposure and it is important to evaluate any long-term neurological sequelae.
Conclusion
In utero exposure to SSRIs during late pregnancy was associated with increased odds of delayed neonatal adaptation. This association was independent of maternal mental health and was dose and type dependent. SSRI exposure was not associated with increased odds of pulmonary hypertension, HIE or seizures. These data may guide parents and physicians in the shared decision-making balancing the benefits and risks of treatment with various doses and types of SSRI during pregnancy. It may also guide paediatricians and neonatologists caring for an increasing number of exposed infants.
Data availability statement
Data are available upon reasonable request. The datasets generated for this study are stored at the KPNC Division of Research. Deidentified data can be provided upon reasonable request to the corresponding author, and with permission from the KPNC Institutional Review Board.
Ethics statements
Patient consent for publication
Ethics approval
The University of California San Francisco and the Kaiser Foundation Research Institute Institutional Review Boards approved this study (IRB #1045-1276201).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors M-CC, YWW, HF and AWS had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design—M-CC, LAA, TBN, YWW and MWK. Acquisition, analysis or interpretation of data—M-CC, LAA, AS, HF, TBN, AWS, YWW and MWK. Drafting of the manuscript—M-CC, YWW and MWK. Critical revision of the manuscript for important intellectual content—M-CC, LAA, AS, HF, TBN, AWS, YWW and MWK. Statistical analysis—M-CC, AWS and TBN. Obtained funding—M-CC, YWW and LAA. Administrative, technical or material support—HF. Supervision—MWK, TBN and YWW.
Funding This study was funded by a Thrasher Early Career Award, the UCSF Newborn Brain Research Innovation award, grants R01HD099216 (PI: YWW), R01HD101483 (PI: LAA) and K23HD109684 (PI: M-CC) from the NICHD.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Linked Articles
- Highlights from this issue