Prenatal selective serotonin reuptake inhibitors and therapeutic hypothermia for suspected hypoxic ischemic encephalopathy

Abstract

Objective

To evaluate the association between prenatal selective serotonin reuptake inhibitor (SSRI) exposure and postnatal therapeutic hypothermia for suspected hypoxic ischemic encephalopathy.

Study Design

Matched case-control study of singleton deliveries at a tertiary hospital from 2010–2016. Cases were infants treated with therapeutic hypothermia for suspected hypoxic ischemic encephalopathy. Controls were non-case infants, matched on gestational age, maternal age, obstetric provider group, and hospital shift.

Result

Prenatal SSRI exposure occurred in 18.4% of cases compared to 4.1% of controls (aOR: 5.9, 95% CI: 1.8 – 19.7). Among all cases, 36.8% had evidence of hypoxic ischemic encephalopathy on postnatal MRI. In addition, 28.6% of SSRI-exposed cases and 38.7% of SSRI-unexposed cases had MRI confirmation of hypoxic ischemic encephalopathy, respectively.

Conclusion

Future research to disentangle signs of SSRI-exposure from true hypoxic ischemic encephalopathy may facilitate targeting therapeutic hypothermia stewardship toward infants more likely to benefit.

Introduction

Depression during pregnancy is common, with an estimated prevalence of 7 – 13%.(1) Approximately 8 – 9% of pregnant women are exposed to antidepressants, with selective serotonin reuptake inhibitors (SSRIs) being the most frequently prescribed antidepressants.(2–4) SSRIs readily cross the placenta, thereby exposing the developing fetus.(5) Some studies have shown an increased risk of adverse perinatal outcomes associated with prenatal exposure to SSRIs.(3,6–8) Specifically, use of SSRIs during pregnancy has been associated with miscarriage and low Apgar scores, independent of maternal depression.(8,9) Furthermore, infants prenatally exposed to SSRIs may exhibit a constellation of signs that constitute an abnormal neurologic examination, including abnormal muscle tone and irritability.(6,10,11) In addition, a recent study showed that prenatal SSRI exposure was associated with changes to fetal brain development, specifically in regions of the brain important for emotional processing.(12) Low Apgar scores and subsequent abnormal neurologic examinations raise clinical suspicion for hypoxic ischemic encephalopathy (HIE) of the newborn, a major cause of neurologic morbidity and mortality.(13)

Multiple randomized, controlled trials have demonstrated that therapeutic hypothermia improves neurodevelopmental outcomes in infants with HIE.(14,15,24–27,16–23) Decreasing the core body and brain temperature is thought to mitigate further hypoxic injury by preventing apoptotic pathways and reducing oxidative damage.(28) Yet, therapeutic hypothermia is not entirely benign. It increases the risk of subcutaneous fat necrosis, often involves exposure to narcotics to maintain comfort during the cooling process, and physically separates infants from their parents for 72 hours or more, which may affect breastfeeding and bonding. In addition, previous studies have shown other adverse effects including an increase in sinus bradycardia (RR: 7.42, 95% CI: 2.52, 21.87) and thrombocytopenia (RR: 1.47, 95% CI: 1.07, 2.03).(14,29,30) Given that both infants who were prenatally exposed to SSRIs and infants with HIE may present with low Apgar scores and abnormal neurologic examinations, it can be difficult for providers to differentiate between the two. Hence, infants prenatally exposed to SSRIs may be treated with therapeutic hypothermia even if their signs are not due to HIE, a diagnosis that can only be confirmed days later. Using therapeutic hypothermia in the absence of HIE exposes infants to the risks of therapeutic hypothermia without benefit. To determine whether infants exposed to prenatal SSRIs had higher odds of receiving treatment with therapeutic hypothermia than unexposed infants, we compared the prevalence of prenatal SSRI exposure among infants who were and were not treated with therapeutic hypothermia.

Patients and methods

Study design

We performed a matched case-control study of in-born singleton deliveries at a tertiary care hospital, Beth Israel Deaconess Medical Center, in Boston, MA, from December 2010 through December 2016. This center delivers approximately 4,800 infants a year of which 900 are high-risk newborns admitted to the neonatal intensive care unit. The exposure of interest was prenatal exposure to SSRIs, defined as either present or absent based on maternal and newborn medical records. To determine whether infants were prenatally-exposed to SSRIs, two investigators (DN and EN) independently reviewed each maternal chart, blinded as to whether the infants received therapeutic hypothermia. Cases (n = 38) were infants treated with therapeutic hypothermia for concern for HIE and were identified from the electronic medical record, which included an indicator for therapeutic hypothermia blanket use. At our institution, to receive therapeutic hypothermia, infants must be born at 36 weeks of gestation or greater and weigh at least 2000 grams. In addition, infants are supposed to meet the following set of criteria: (1) evidence of fetal or neonatal distress, as defined by history of a perinatal event (i.e. placental abruption, cord prolapse, or severe fetal heart rate abnormality), pH ≤ 7.0 or base deficit ≥ 16 mmol/L on cord blood gas or postnatal blood gas obtained within one hour after birth, 10-minute Apgar score of ≤ 5, or assisted ventilation initiated at birth and continued for at least 10 minutes, and (2) abnormal cerebral function by physical neurologic exam or amplitude integrated electroencephalography (aEEG). Importantly, if the neurologic examination is normal, the infant is not eligible for therapeutic hypothermia. Controls (n = 148) were the next four non-case deliveries within six months of the case and who matched to cases by obstetric provider group, hospital shift, completed weeks of gestation, and maternal age ± three years. A sample of convenience was used. Matched variables were determined a priori and based on prior knowledge of the hospital and clinical practice. We consulted a neonatologist and epidemiologist (HB and MH) to determine variables to match on. We matched on obstetric provider group because provider groups serve distinct patient populations and the availability of documentation for prenatal SSRI exposure varies by obstetric practice. Hospital shifts were defined as weekday (7:00 AM to 7:00 PM, Monday through Friday) and off hours (nights, weekends, and holidays). Among infants who were treated with therapeutic hypothermia, we examined whether there were differences in meeting standard therapeutic hypothermia initiation criteria. Additionally, to understand whether infants prenatally exposed to SSRIs who received therapeutic hypothermia were less likely to have confirmatory evidence of HIE on brain MRI three days after birth, two investigators (DN and HB) reviewed each of the MRI results as read by the clinical pediatric neuroradiologists which included assessment of diffusion-weighted images.

Statistical Analyses

We compared categorical variables using a Chi-square or Fisher’s exact test and continuous variables using the Wilcoxon rank sum test. Conditional logistic regression was used to estimate the odds ratio and 95% confidence interval (CI) for SSRI exposure among infants treated with therapeutic hypothermia compared to controls. We considered several potential confounding variables, including maternal race/ethnicity, parity, and insurance. Variables that changed the odds ratio by more than 10% were retained in the adjusted models. We considered p-values < 0.05 to be statistically significant and all tests were two sided. Data were analyzed using SAS 9.4 (SAS Institute, Cary, NC, USA). The institutional review board at Beth Israel Deaconess Medical Center approved this study. The protocol was approved with a Waiver of Informed Consent and Authorization.

Results

We identified 38 infants treated with postnatal therapeutic hypothermia for concern for HIE and matched them to 148 controls from December 2010 through December 2016. We were able to identify four controls for all but three cases due to difficulty matching on maternal age. Specifically, we could identify only two controls for one case and three controls for two cases. The cases and controls were similar with respect to the matching factors – maternal age, gestational age, delivery shift, and obstetric provider group, as well as race/ethnicity and insurance status. Demographic characteristics of cases and controls are presented in Table 1.

Table 1.

Maternal characteristics among infants treated with therapeutic hypothermia for suspected hypoxic ischemic encephalopathy and matched controls

Characteristic	Cases n = 38	Controls n = 148
Maternal agea (years)	33.2 (29.8 – 36.0)	31.2 (27.8 – 33.4)
Nulliparous	22 (57.9)	77 (52.0)
Race/ethnicity
Non-Hispanic white	18 (47.4)	70 (47.3)
Non-Hispanic black	3 (7.9)	15 (10.1)
Asian	6 (15.8)	25 (16.9)
Hispanic	1 (2.6)	9 (6.1)
Other/unknown/more than one	10 (26.3)	29 (19.6)
Shift of deliverya
Weekday	12 (31.6)	45 (30.4)
Night/weekend/holiday	26 (68.4)	103 (69.6)
Provider groupa
Hospital faculty practice	7 (18.4)	28 (18.9)
Resident practice	1 (2.6)	3 (2.0)
Community health center	6 (15.8)	28 (18.9)
Private practice	24 (63.2)	89 (60.1)
Insurance
Private	22 (57.9)	89 (60.1)
Public	12 (31.6)	48 (32.4)
Other	4 (10.5)	11 (7.4)

Data presented as median (interquartile range) or n (%)

^aMatching factors

Cases were less likely to have a non-operative vaginal delivery (34.2%) compared to controls (69.6%), and, among controls, this was the most common mode of delivery. The most common mode of delivery among cases was unscheduled cesarean delivery (Table 2).

Table 2.

Perinatal and infant characteristics among infants treated with therapeutic hypothermia for suspected hypoxic ischemic encephalopathy and matched controls

Characteristic	Cases n=38	Controls n=148	P
Gestational agea (weeks)	39.0 (38.0 – 40.0)	39.0 (38.0 – 40.0)	0.87
Mode of delivery			<0.001
Vaginal	13 (34.2)	103 (69.6)
Operative vaginal	5 (13.2)	9 (6.1)
Scheduled cesarean	5 (13.2)	15 (10.1)
Unscheduled cesarean	15 (39.5)	21 (14.2)
Apgar score – 1 min	2 (2 – 3)	9 (8 – 9)	<0.001
Apgar score – 5 min	5 (3 – 7)	9 (9 – 9)	<0.001
Birthweight (g)	3230 (2970 – 3605)	3232 (2925– 3582.5)	0.95
Prenatal SSRI exposure	7 (18.4)	6 (4.1)	0.006

Data presented as median (interquartile range) or n (%)

^aMatching factor

There were 13 infants prenatally exposed to SSRIs in this study population. Cases were more likely to be exposed to prenatal SSRIs (18.4%) than controls (4.1%; p = 0.006). The median Apgar scores at 1 and 5 minutes were significantly lower among cases compared to controls (p < 0.001), but birthweight was not significantly different between the two groups (p = 0.95). The crude odds ratio for the association between prenatal SSRI exposure and therapeutic hypothermia was 5.3 (95% CI: 1.7 – 17.0). After adjusting for race/ethnicity, the association persisted, demonstrating that cases were nearly six times as likely to have been prenatally exposed to SSRIs than controls (adjusted odds ratio: 5.9 (95% CI: 1.8 –19.7)). Adding parity and insurance to the model did not appreciably alter the effect estimate.

Among infants treated with therapeutic hypothermia, 86.8% met criteria for receiving therapeutic hypothermia (Table 3). All of the cases met the gestational age and birthweight criteria, 92.1% had evidence of fetal or neonatal distress, and 94.7% had evidence of moderate to severe neonatal encephalopathy. There were some differences in meeting standard therapeutic hypothermia initiation criteria based on prenatal SSRI exposure; although, none met statistical significance. Specifically, 14.3% of cases prenatally exposed to SSRIs had a base deficit > 16 compared with 38.7% of unexposed cases. Additionally, 100% of cases prenatally exposed to SSRIs required mechanical ventilation compared with 58.1% of unexposed cases. Notably, there was abnormal cerebral function by amplitude-integrated electroencephalography in just 57.1% of exposed infants compared with 83.9% of unexposed infants. Similarly, only 28.6% of cases prenatally exposed to SSRIs had MRI confirmation of HIE compared with 38.7% of unexposed cases.

Table 3.

Therapeutic hypothermia initiation criteria among infants treated with therapeutic hypothermia

	Total n = 38	SSRI n = 7	No SSRI n = 31	P
Fetal/neonatal distress	35 (92.1)	7 (100.0)	28 (90.3)	1.0
Perinatal event	32 (84.2)	6 (85.7)	26 (83.9)	1.0
pH ≤ 7.0	7 (18.4)	2 (28.6)	5 (16.1)	0.59
Infant blood gas, base deficit ≥ 16 mmol/L	13 (34.2)	1 (14.3)	12 (38.7)	0.39
Apgar score – 10 min ≤ 5	10 (26.3)	1 (14.3)	9 (29.0)	0.65
Mechanical ventilation	25 (65.8)	7 (100.0)	18 (58.1)	0.07
Moderate to severe encephalopathy	36 (94.7)	7 (100.0)	29 (93.5)	1.0
Physical exam	26 (68.4)	5 (71.4)	21 (67.7)	1.0
Abnormal cerebral function/aEEG	30 (78.9)	4 (57.1)	26 (83.9)	0.15
Met all criteria	33 (86.8)	7 (100.0)	26 (83.9)	0.56

Data presented as n (%)

One unexposed infant did not have an EEG, represents cerebral function monitor or EEG reading.

aEEG: amplitude-integrated electroencephalography

Discussion

Infants treated with therapeutic hypothermia due to concern for HIE at birth were significantly more likely to have been prenatally exposed to SSRIs than matched, healthy controls. In other words, prenatal exposure to SSRIs was associated with a significant increase in the odds of receiving therapeutic hypothermia for clinical suspicion for HIE.

Infants prenatally exposed to SSRIs have an increased risk of low Apgar scores and abnormal neurologic examinations after birth, which are clinical presentations that are similar to infants suspected of having HIE.(6,8) We suspect that the association between prenatal exposure to SSRIs and treatment with therapeutic hypothermia is due to difficulty in distinguishing between the clinical presentations of an infant diagnosed with HIE and an infant with prenatal SSRI exposure without HIE.(6,8)

Furthermore, while the neurologic examinations of some infants who present with signs concerning for HIE normalize within the first few hours after birth, infants exposed to SSRIs can have abnormal neurologic examinations that persist for days.(6) Since a normal neurologic examination excludes an infant from receiving therapeutic hypothermia, infants prenatally exposed to SSRIs may be more likely to be treated because their examinations are less likely to normalize during a period of observation for potential therapeutic hypothermia initiation. This could lead to treatment with therapeutic hypothermia even in the absence of HIE.

Alternatively, it is possible that SSRIs, or the conditions they treat, may place infants at higher risk of HIE. However, given that fewer of the cases prenatally exposed to SSRIs had evidence of HIE on MRI, we hypothesize that infants prenatally exposed to SSRIs more often are misclassified as having HIE compared with unexposed infants. Our data suggest that prenatal SSRI exposure may be a risk factor for unnecessary treatment for HIE. However, given that in both exposed and unexposed infants, more than half of all treated infants had no evidence of HIE on post-natal MRI, more work to correctly classify all infants with suspected HIE is needed.

We are unaware of other studies evaluating the association between prenatal exposure to SSRIs and therapeutic hypothermia; however, there are some data on the effects of SSRIs and neurologic outcomes of infants. A prospective, observational study examined the effects of infants prenatally exposed to SSRIs and the quality of infants’ general movements, a method used to assess neurological function. The study showed that prenatal exposure to SSRIs had an adverse effect on the neurological function of the infant, which was seen not only in the first few days after birth, but also at three to four months of age; the association persisted after adjusting for maternal depression and anxiety, maternal education, paternal depression and anxiety, gestational age, birthweight, and sex.(6) There also have been studies demonstrating that infants prenatally exposed to SSRIs are at increased risk of low Apgar scores. Specifically, a retrospective cohort study found that use of SSRIs during pregnancy was associated with an increased risk of low Apgar scores, independent of maternal depression.(8) In addition, Bakaysa et.al., conducted a retrospective cohort study and concluded that use of SSRIs during pregnancy was associated with an increased risk of Apgar scores < 7 at 1 and 5 minutes.(11)

Treatment of infants with therapeutic hypothermia for HIE has evolved over the last decade. After several trials with varying inclusion criteria, individual neonatal units developed their own inclusion criteria.(14) Our study should be interpreted within the context of the temporal change in practice of initiating therapeutic hypothermia among infants who meet some, but not all of the criteria (as seen in our study). Specifically, at our institution, the guideline states: “This is a general guideline, and it does not represent a professional standard of care governing providers’ obligations to patients. Care is revised to meet individual patient needs.”(31) This trend is consistent with some other neonatal units.(32) In mild or borderline cases where patients may meet most but not all criteria, providers may initiate therapeutic hypothermia because of concern that missing the opportunity could be detrimental to the long-term neurodevelopmental of the infant. Although adverse effects associated with therapeutic hypothermia are uncommon, it is important to note as changes in guidelines occur, that previous studies have identified potential adverse effects, such as bradycardia and thrombocytopenia.(14,29,30)

Our study has several potential limitations, including possible misclassification of exposure, lack of generalizability, imprecision, confounding by indication, provider bias, and small sample size. Prenatal exposure to SSRIs may have been misclassified if women did not disclose their medication use, or it was not documented in the medical record. Additionally, we were not able to assess duration and timing of exposure. This would most likely be non-differential based on outcome status and bias the odds ratio towards the null. With respect to generalizability, we conducted our study at a single hospital. Thus, our findings may be less relevant to other settings as there may be treatment variation by hospital. The number of infants treated with therapeutic hypothermia was relatively small (n = 38), resulting in wide confidence intervals. The association we observed may be due to depression itself, not its treatment. Nonetheless, SSRI exposure may serve as a marker of potentially unnecessary therapeutic hypothermia exposure in a population that may be particularly vulnerable to the psychological consequences of maternal-infant separation. In addition, it is possible that providers could have learned about prenatal SSRI exposure in the delivery room and thus become more cognizant of potential adverse outcomes, such as HIE. However, it is more likely that exposed infants would require resuscitation at birth, due to SSRIs, raising concern for HIE and requiring NICU providers to be called to the delivery room. Furthermore, relying on MRI for HIE diagnosis has limitations due to the complexities of interpretation and differences in readings among pediatric neuroradiologists. Thus, there may misclassification of HIE diagnoses. Lastly, we acknowledge the small sample size of 38 cases, of which 7 were exposed to SSRIs and 31 were unexposed. The results of this study should be considered within the bounds of this limitation and be interpreted with caution as providers vary treatment practices.

While our study has limitations, it also has several strengths. We ascertained prenatal SSRI exposure from the prenatal medical record, which predated the outcome of treating the infant with therapeutic hypothermia for concern for HIE. This avoided the recall bias that often affects case-control studies. Furthermore, investigators who reviewed the medical records were blinded as to whether the infant was treated with therapeutic hypothermia. Lastly, because controls were the next four non-case deliveries within six months of the case, we minimized the effect of temporal trends in SSRI use and/or therapeutic hypothermia initiation.

Therapeutic hypothermia is not a risk-free treatment for infants exhibiting signs of abnormal neurologic examinations and low Apgar scores. It is important for neonatologists and pediatric providers to weigh the risks and benefits of therapeutic hypothermia with the goal of enhancing infant outcomes by taking into account prenatal SSRI exposure among infants that exhibit signs of HIE. Our study alone should not change practice. Infants who meet trial guidelines for therapeutic hypothermia initiation for concern for HIE should be treated regardless of SSRI exposure. However, therapeutic hypothermia stewardship is reasonable in cases of potential mild HIE where the benefits of therapeutic hypothermia are unproven. In these cases, considering SSRI-exposure as a potential cause of a persistent abnormal neurologic examination may help providers to avoid initiation of therapeutic hypothermia and the risks associated with the treatment. Further, our findings may support SSRI stewardship in pregnancy. The absolute risk of receiving therapeutic hypothermia is extremely low (<2 per 1,000 births in our study). However, given this risk and the small increased risk of congenital heart disease and neurologic sequalae associated with prenatal SSRI exposure, women who can safely come off SSRIs before and during pregnancy may choose to consider doing so while balancing the risks of untreated depression.(6,7)

Prenatal exposure to SSRIs was significantly associated with a nearly six-fold increase in the odds of postnatal treatment of therapeutic hypothermia for concern for HIE. Our findings provide preliminary evidence to perhaps support SSRI stewardship among women who can safely come off SSRIs before and during pregnancy. Additionally, for infants who do not meet therapeutic hypothermia inclusion criteria, but for whom there is some suspicion for mild HIE, taking into consideration SSRI-associated neurologic examination findings may help inform the provider’s decision to avoid the risks associated with therapeutic hypothermia.