Examining the effect of newborn feeding method on postnatal length of stay and healthcare utilisation among Neonates With In-utero SSRI Exposure (NeoWISE): a retrospective cohort study

Abstract

Introduction

Up to 30% of in-utero selective serotonin reuptake inhibitor (SSRI)-exposed newborns experience withdrawal and up to 60% experience behavioural impacts. Breastfeeding may offer benefits from skin-to-skin contact and transfer of SSRIs through breastmilk. Limited data are available on healthcare utilisation of these newborns beyond neonatal intensive care unit admission. We aimed to identify whether newborn feeding method affected healthcare utilisation in the first 30 days of postnatal life among Neonates With In-utero SSRI Exposure (NeoWISE).

Methods

This retrospective cohort study included newborns born in Ontario hospitals between 1 April 2012 and 31 March 2020 to beneficiaries of the Ontario Drug Benefit Program who filled at least one SSRI prenatal prescription. Administrative health and birth registry data were used. Newborn feeding methods were available from birth to hospital discharge. Outcomes were postnatal length of stay (LOS), emergency department visit(s) and hospital readmission within the first 30 days of life. Adjusted linear regression models using inverse probability of treatment weights for feeding method were conducted.

Results

The cohort included a total of 5079 NeoWISE participants, with 3321 (65.4%) exclusively breastfed. Formula-fed infants were more likely to have older mothers with an urban residence, lower income quintile, prior live birth, who smoked at delivery and to be born via caesarean delivery; these imbalances were balanced after propensity score weighting. Exclusively breastfed newborns were significantly less likely to have postnatal LOS ≥3 days (11.6% vs 16.3%) (adjusted relative risk (adjRR) 0.84, 95% CI 0.72 to 0.97), but more likely to have emergency department visit (adjRR 1.42, 95% CI 1.03 to 1.97) and hospital readmission (adjRR 1.78, 95% CI 1.21 to 2.62) compared with formula-fed newborns, primarily due to jaundice.

Conclusion

Women taking SSRI medication during pregnancy should receive anticipatory guidance about newborn feeding effectiveness, signs of jaundice and when to seek healthcare after birth. In-hospital feeding method was associated with emergency department visits and hospital readmission among NeoWISE. Breastfeeding cessation was not measured and could have affected estimated associations.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Newborns with in-utero selective serotonin reuptake inhibitor (SSRI) exposure are at risk for adverse outcomes including altered physiological and behavioural adaptation after birth.

• The association between newborn outcomes and newborn feeding method (breastfeeding vs formula feeding) has not been investigated.

WHAT THIS STUDY ADDS

• Newborns who are in-utero SSRI exposed and exclusively breastfed after birth have a reduced risk of prolonged hospital stay but a higher risk of emergency department visits and readmission to hospital as compared with formula-fed newborns.

• The primary reason for healthcare utilisation was jaundice.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This research was an important step in illuminating future research, practice and policy directions related to the potential healthcare needs of this newborn population following discharge from hospital.

Introduction

Perinatal mental health disorders are the most common complication during pregnancy and the first year after birth,¹ with reported rates in Canada and globally between 20% and 30%.^2–4 Adequate treatment of mental health is essential for maternal and fetal/newborn health.⁵ Pharmacological management is recommended for the treatment of moderate-to-severe perinatal mental health disorders when psychosocial interventions are not effective.^{1 6} Selective serotonin reuptake inhibitors (SSRIs) are the most prescribed antidepressant (AD) medication to treat mental health disorders⁷ and are recommended as a first-line pharmacotherapy for moderate-to-severe perinatal mental health disorders.^{1 6}

Although SSRI medications have been prescribed for >40 years and are generally perceived to be safe, some knowledge gaps remain regarding newborn impacts. Few researchers have investigated the effects of breastfeeding on outcomes among in-utero exposed newborns in the early postnatal period. Newborn healthcare utilisation in the first month after birth has rarely been studied in this population and often is limited to neonatal intensive care unit (NICU)/special care nursery admission. The postnatal length of hospital stay is frequently not reported, and little is known about the impact on the healthcare system after hospital discharge.

Given that pregnant women are often concerned about the potential adverse health impacts of taking medications, particularly the effects on their baby’s health,⁸ healthcare providers (HCPs) play a crucial role in the shared decision-making process by providing information about the risks and benefits of various pharmacological interventions.¹ Adverse impacts associated with in-utero SSRI exposure have been reported in up to 30% of newborns demonstrating withdrawal signs^9–11 and up to 60% demonstrating behavioural impacts.^9–15 Various signs have been reported, including, but not limited to, restless sleep, mottling, tremors, crying, poor muscle tone, jitteriness, irritability, seizures, feeding difficulties, sleep disturbances, hypoglycaemia, gastrointestinal disturbances and respiratory distress.^{9–11 16} Different terms have been used to describe this phenomenon, including neonatal abstinence syndrome, poor neonatal adaptation syndrome, neonatal withdrawal syndrome and serotonin toxicity, to name a few.¹⁷

Third trimester exposure to SSRIs has been associated with newborn withdrawal signs, which can be observed within several hours after birth and typically resolve without medical intervention in the following days and weeks.^{18 19} The exact pathophysiology of newborn withdrawal is not known; however, SSRI medications cross the placenta and are readily absorbed by the fetus. At birth, there is an abrupt discontinuation in exposure to the SSRI, precipitating withdrawal signs and behavioural adaptations in some newborns. In most instances, this transient process typically resolves within a few days to 2 weeks after birth and can be alleviated with non-pharmacological interventions, such as skin-to-skin contact, swaddling, frequent feedings and a quiet environment.¹⁸ Breastfeeding may be associated with a lower risk of withdrawal signs as compared with newborns receiving formula feeding only.¹³ It is postulated that a small amount of SSRI medication transferred into breastmilk may be beneficial in alleviating withdrawal signs.

Additional research is needed to (a) identify the health outcomes of Neonates With In-utero SSRI Exposure (NeoWISE) in the first few days and weeks after birth and (b) determine if additional support and follow-up during this timeframe would be beneficial to inform shared decision-making discussions. The overall objective of our research was to assess the impact of newborn feeding method on outcomes among NeoWISE in the first month of life. We previously reported the rates of newborns assigned a neonatal withdrawal diagnosis code among a cohort of NeoWISE born in Ontario, Canada were 1.5% in breastfed vs 2.3% in formula-fed newborns. However, there was no difference in the risk of withdrawal when relevant covariates were considered (adjusted relative risk (adjRR) 0.86, 95% CI 0.56 to 1.34).²⁰ Breastfed newborns had a reduced risk of transfer to the NICU compared with formula-fed newborns (adjRR 0.80, 95% CI 0.66 to 0.97).²⁰

In this paper, we report the length of postnatal hospital stay and healthcare utilisation (within the first 30 days of life after hospital discharge) among breastfed compared with formula-fed neonates exposed to SSRIs in utero. We hypothesised that breastfeeding newborns would have a lower risk of prolonged hospital stays, emergency department (ED) visits and hospital readmissions. This hypothesis was based on the small amount of SSRI medication transferred through breast milk, due to the larger intracellular gaps between the alveolar cells in the first few days of life^{21 22} and the benefit of close skin-to-skin contact during breastfeeding.

Methods

We conducted a retrospective cohort study including newborns born in Ontario hospitals (see online supplemental figure 1).²⁰ The inclusion criteria were newborns born between 1 April 2012 and 31 March 2020. The follow-up period was 30 days from birth/postnatal period, with the last date of follow-up occurring on 30 April 2020 (see online supplemental figure 2). We identified mothers who were beneficiaries of the Ontario Drug Benefit (ODB) program who filled one or more SSRI prescriptions during their pregnancy (ie, citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine or sertraline).²⁰ The ODB program includes prescription drug benefit information for Ontario residents who are older than 65 years of age and under 65 years who are unemployed, have disabilities or have high prescription costs relative to net household income.^{23 24}

We obtained maternal and newborn data from linked administrative health and birth registry data holdings at Institute for Clinical Evaluative Sciences (ICES). These datasets were linked using encoded identifiers and analysed at ICES. ICES is an independent non-profit research institute funded by an annual grant from the Ontario Ministry of Health and the Ministry of Long-Term Care. As a prescribed entity under Ontario’s privacy legislation, ICES is authorised to collect and use healthcare data for the purposes of health system analysis, evaluation and decision support. Secure access to these data is governed by policies and procedures that are approved by the Information and Privacy Commissioner of Ontario.²⁵

Newborn feeding data were available from birth to hospital discharge and were obtained from the Ontario Better Outcomes Registry and Network (BORN), an acquired registry at ICES.²⁶ BORN is Ontario’s prescribed maternal newborn registry and captures 99% of health records for in-hospital births.²⁶ Newborns were excluded from the cohort (a) if the mother was a non-resident of Ontario; (b) if the mother did not have an ODB drug claim during pregnancy; (c) if maternal age was <15 years of age or >49 years of age; (d) if it was a twin or higher order multiple birth and/or (e) if maternal-newborn records could not be linked.²⁰

The study exposure was newborn feeding method, defined in BORN as “the type of oral feeding given to the newborn from the time of birth to discharge from hospital or birth centre”.²⁷ Being exclusively breastfed or exclusively formula-fed from birth to hospital discharge was the exposure of interest. Exclusively breastfed was defined as receiving only breastmilk from birth until hospital discharge. Newborns who received any breastfeeding, defined as breastmilk plus a breastmilk substitute (eg, formula, pasteurised human donor milk), were excluded from the primary analysis but were included in the sensitivity analyses. Newborns with unknown feeding status and those who did not receive any feeding were excluded.²⁰

Outcomes

Postnatal length of stay (LOS) was defined as the number of days from birth until hospital discharge; data were obtained from the Canadian Institute for Health Information (CIHI) Discharge Abstract Database (DAD).²⁸ A prolonged postnatal LOS was defined as ≥3 days, given previous recommendations of a 24-hour to 48-hour observation period for this population.^{18 29} Newborn healthcare utilisation variables for this study, in the first 30 days of life, included: visit(s) to the ED and readmissions to the hospital. Data regarding ED visit(s) were obtained from the National Ambulatory Care Reporting System database and readmissions to hospital were obtained from the CIHI DAD (see online supplemental table 1).²⁰ Healthcare utilisation data included the main problem, classified according to the International Statistical Classification of Diseases, Tenth Revision (ICD-10), reported for ED visits and the primary diagnosis code (ICD-10-1) for hospital readmissions. The top 10 main problems and primary diagnosis codes were obtained for the entire population of SSRI-exposed Ontario newborns to identify the priority reasons for postnatal healthcare utilisation for the NeoWISE cohort. Codes of a similar main problem and diagnosis were then grouped together.

Analysis

We compared differences in maternal and newborn characteristics by feeding method using standardised differences, where a standardised difference of >10% indicated a meaningful imbalance between exposure groups.³⁰ Continuous variables were reported as means and SDs. The following variables were dichotomised: gestational age was categorised as <37 weeks or ≥37 weeks; type of birth was categorised as vaginal or caesarean and postnatal LOS was categorised as <3 days and ≥3 days with the latter considered a prolonged LOS. Neighbourhood income quintiles were used, with 1 being the lowest and 5 being the highest.²⁰ The top 10 reasons for ED visits and hospital readmissions were reported.

We used inverse probability of treatment weighting (IPTW) of propensity scores for type of feeding method to balance characteristics of maternal-newborn pairs based on the feeding method.³¹ The following covariates were the independent variables in our propensity score logistic models: gestational age (<37 weeks and ≥37 weeks), type of birth (vaginal and caesarean), parity (nulliparous, multiparous), maternal age, neighbourhood income quintiles (1 lowest through 5 highest), other ADs (non-SSRIs), psychotropic medications, opioid use, alcohol exposure in pregnancy, smoking at the time of birth, rurality and Elixhauser Comorbidity Index score. To estimate the treatment effect for our binary outcomes of postnatal LOS ≥3 days, any ED visit and any hospital readmission, we fit a weighted generalised linear regression model to compute a relative risk.²⁰ We conducted sensitivity analysis by (a) including only newborns with third-trimester SSRI exposure given previously reported associations of withdrawal signs with third trimester SSRI exposure³² and (b) defining exposure as any breastfeeding rather than exclusive breastfeeding to see if there was a benefit of any breastmilk in the infant. For each sensitivity analysis, the PS models were recalculated and estimated associations between feeding and study outcomes were compared with those of the primary analysis.

Finally, we calculated the relative risk for initiated feeding method in hospital and risk for ED visit and readmission to hospital within the first 7 days after hospital discharge and within 30 days of life in the NeoWISE cohort. We calculated the p value for cases involving one or more ED visit or readmission to hospital within 7 days and 30 days of life in the NeoWISE cohort.

Patient and public involvement in research

The original study direction was identified following a consultation request CC received from a nurse educator on a mother-baby unit about the care of in-utero SSRI-exposed newborns. Dr. Christina Cantin engaged interprofessional care providers, family advisors and people with lived experience in the Ontario Health East Region when conceptualising this study. A person with lived experience of SSRI medication use during pregnancy and breastfeeding was a research team member who helped to inform the relevance of study questions, interpretation of study implications and dissemination of findings.

The findings of this study were reported following the REporting of studies Conducted using Observational Routinely collected Data for non-interventional PharmacoEpidemiological research (RECORD-PE) guidelines³³ (online supplemental file 3, reporting checklist).

Results

During the study timeframe, there were 1 120 189 births of which 9926 singleton newborns were exposed to SSRIs during pregnancy. The Ontario NeoWISE cohort for this study examining newborn feeding method included 5079 newborns, with 3321 (65.4%) exclusively breastfed and 1758 (34.6%) formula-fed only (table 1). Newborns who were formula-fed compared with those who were breastfed were more likely to have older mothers who lived in an urban area and had a lower neighbourhood income quintile, had a previous live birth and reported smoking at the time of delivery. Formula-fed newborns had a slightly lower but not clinically meaningful gestational age (38.7 weeks vs 39.0 weeks) and were more likely to be born via caesarean delivery (29.9% vs 18.2%). These imbalances in maternal and newborn characteristics were balanced after propensity score weighting (table 1). In the first 30 days of life, the only newborn deaths identified were among formula-fed newborns (less than six cases; the exact number not specified due to suppression of small cells as required by ICES).

Table 1. Baseline characteristics of maternal-newborn pairs exposed to SSRI in pregnancy in the NeoWISE cohort before and after propensity weighting.

		Before propensity weighting (n=5079)			After propensity weighting* (n=5061)
Variable		Formula feeding only N=1758, n (%)	Breastfeeding only N=3321, n (%)	Standardised difference	Formula feeding only N=1754 (%)	Breastfeeding only N=3307 (%)	Standardised difference
Maternal age at delivery, mean (years) (SD)		26.9 (5.9)	26.0 (5.6)	0.155	26.2	26.2	0.007
Maternal smoking at the time of birth		781 (44.5)	1053 (31.8)	0.263	36.1	36.3	0.003
Alcohol exposure in pregnancy		37 (2.1)	76 (2.3)	0.013	2.12	2.2	0.005
Exposure 1 year prior to or during pregnancy	Other antidepressant medications	445 (25.4)	743 (22.5)	0.068	23.1	23.3	0.004
	Other psychotropics	705 (40.2)	1107 (33.5)	0.140	36.0	35.8	0.004
	Opioid	457 (26.1)	666 (20.1)	0.141	21.9	22.2	0.007
Type of birth (caesarean)		524 (29.9)	601 (18.2)	0.276	22.4	22.3	0.002
Gestational age (<37 weeks)		89 (5.1)	117 (3.5)	0.076	4.0	4.1	0.005
Previous live birth(s)		1246 (71.0)	2025 (61.2)	0.208	65.2	64.8	0.008
Nearest census-based neighbourhood income quintile*
2		418 (23.8)	793 (24.0)	0.003	23.7	23.9	0.004
3		262 (14.9)	519 (15.7)	0.021	15.4	15.4	0.001
4		147 (8.4)	370 (11.2)	0.095	10.5	10.3	0.008
5		88 (5.0)	236 (7.1)	0.089	6.6	6.4	0.007
Living in a rural area		245 (14.0)	370 (18.2)	0.116	17.1	16.8	0.008
Elixhauser Comorbidity Index score
1		84 (4.8)	118 (3.6)	0.061	3.9	3.9	0.001
2		35 (2.0)	46 (1.4)	0.047	1.7	1.7	0.000

Bold font indicates statistically significant standardised difference.

^*N=18 were missing information on neighbourhood income and were excluded from the propensity score analysis.

Exclusively breastfed as compared with formula-fed newborns had similar postnatal LOS (1.6±0.8 days vs 1.8±1.0 days, respectively). The difference in postnatal LOS was 4.8 hours. Exclusively breastfed newborns had a lower risk of a prolonged LOS (≥3 days) (11.6% vs 16.3%, RR 0.71, 95% CI 0.62 to 0.82) (table 2). Following hospital discharge, 3.7% (n=187) of the NeoWISE cohort had an ED visit and 3.2% (n=162) were readmitted to the hospital. In the unadjusted analysis, newborns who were exclusively breastfed during their hospitalisation after birth, as compared with newborns who were formula-fed, had a slight increase in risk of any ED visit (4.0% vs 3.0%, RR 1.34, 95% CI 0.98 to 1.83) and a small but statistically significant higher risk of readmission (3.8% vs 2.0%, RR 1.86, 95% CI 1.29 to 2.68) (table 2). In the adjusted models, exclusively breastfed newborns had a lower risk of a prolonged LOS (≥3 days) (adjRR 0.84, 95% CI 0.72 to 0.97). Exclusively breastfed newborns were significantly more likely to visit the ED (adjRR 1.42, 95% CI 1.03 to 1.97) and be readmitted to the hospital (adjRR 1.78, 95% CI 1.21 to 2.62) (table 2) within the first 30 days of life.

Table 2. Unadjusted and adjusted RR between feeding method and healthcare utilisation.

Analysis	Outcome	Feeding method	No outcome	Had outcome	RR (95% CI)*
			N (%)	N (%)	Unadjusted	Adjusted†
Primary cohort (n=5079)	Length of postnatal hospital stay ≥3 days‡	Formula feeding only	1471 (83.7)	287 (16.3)	1	1
		Breastfeeding only	2935 (88.4)	386 (11.6)	0.71 (0.62 to 0.82)	0.84 (0.72 to 0.97)
	Visit to emergency department§	Formula feeding only	1705 (97.0)	53 (3.0)	1	1
		Breastfeeding only	3187 (96.0)	134 (4.0)	1.34 (0.98 to 1.83)	1.42 (1.03 to 1.97)
	Readmission to hospital‡	Formula feeding only	1722 (98.0)	36 (2.0)	1	1
		Breastfeeding only	3195 (96.2)	126 (3.8)	1.86 (1.29 to 2.68)	1.78 (1.21 to 2.62)
Sensitivity: third trimester SSRI exposed only (n=2916)	Length of postnatal hospital stay ≥3 days‡	Formula feeding only	809 (81.7)	181 (18.3)	1	1
		Breastfeeding only	1685 (87.5)	241 (12.5)	0.69 (0.58 to 0.82)	0.82 (0.68 to 0.99)
	Visit to emergency department§	Formula feeding only	959 (96.9)	31 (3.1)	1	1
		Breastfeeding only	1848 (96.0)	78 (4.0)	1.29 (0.86 to 1.95)	1.36 (0.88 to 2.11)
	Readmission to hospital‡	Formula feeding only	968 (97.8)	22 (2.2)	1	1
		Breastfeeding only	1851 (96.1)	75 (3.9)	1.75 (1.10 to 2.80)	1.68 (1.01 to 2.78)
Sensitivity: any breastfeeding included (n=7420)	Length of postnatal hospital stay ≥3 days‡	Formula feeding only	1471 (83.7)	287 (16.3)	1	1
		Any breastfeeding	4769 (84.2)	893 (15.8)	0.97 (0.86 to 1.09)	0.99 (0.87 to 1.13)
	Visit to emergency department§	Formula feeding only	1705 (97.0)	53 (3.0)	1	1
		Any breastfeeding	5436 (96.0)	226 (4.0)	1.32 (0.99 to 1.78)	1.36 (1.00 to 1.83)
	Readmission to hospital‡	Formula feeding only	1722 (98.0)	36 (2.0)	1	1
		Any breastfeeding	5451 (96.3)	211 (3.7)	1.82 (1.28 to 2.58)	1.69 (1.18 to 2.43)

Bold font indicates statistically significant result.

^*RR = risk ratio

^†Propensity score weighted using IPTW with covariates adjustment for gestational age, type of birth, parity, maternal age, neighbourhood income quintiles, other antidepressants (non-SSRIs), psychotropic medications, opioid use, alcohol exposure in pregnancy, smoking at time of birth, rurality and Elixhauser Comorbidity index. Missing data (n = 18) were excluded from this analysis.

^‡Data obtained from the CIHI Discharge Abstract Database.

^§Data obtained from the National Ambulatory Care Reporting System.

CIHI, Canadian Institute for Health Information; IPTW, inverse probability of treatment weighting; RR, risk ratio; SSRI, selective serotonin reuptake inhibitor.

Sensitivity analysis

In the sensitivity analysis comparing outcomes among the NeoWISE with trimester three exposure only, exclusively breastfed newborns were less likely to have a prolonged LOS ≥3 days (12.5% vs 18.3%, RR 0.69, 95% CI 0.58 to 0.82). There were no significant differences in ED visits based on newborn feeding method. Regarding hospital readmissions, exclusively breastfed newborns were noted to have a small but statistically significantly higher rate of readmission (3.9% vs 2.2%, RR 1.75, 95% CI 1.10 to 2.80). In the models for this sensitivity analysis, the only significant findings were that exclusively breastfeeding newborns were less likely to have a prolonged LOS ≥3 days (adjRR 0.82, 95% CI 0.68 to 0.99) and a higher rate of hospital readmission (adjRR 1.68, 95% CI 1.01 to 2.78) (table 2).

In the sensitivity analysis comparing any breastfeeding versus formula feeding only, there were no differences in the risk for prolonged LOS or ED visits. Associations with hospital readmission persisted (adjRR 1.69, 95% CI 1.18 to 2.43) with newborns receiving any breastfeeding at higher risk compared with those who were formula-fed (table 2).

Reasons for healthcare utilisation

In the entire population of in-utero SSRI-exposed newborns in Ontario (n=9926), we reported the top 10 reasons for healthcare utilisation following initial postnatal hospital discharge after birth (online supplemental table 2). After combining similar main problems leading to ED visits, the top four healthcare issues were: (1) jaundice, (2) feeding problems, (3) gastrointestinal problems and (4) respiratory problems. The top four combined diagnosis codes for NeoWISE readmitted to the hospital were: (1) jaundice, (2) gestational age or birth weight-related issues, (3) neonatal withdrawal symptoms and (4) feeding problems (online supplemental table 2).

In the NeoWISE cohort, jaundice was the only significant main problem for ED visits and diagnosis code for hospital readmission (table 3). There was a greater risk of ED visits for exclusively breastfed NeoWISE as compared with formula-fed newborns within the first 7 days after discharge (2.4% vs 1.0%, RR 2.46, 95% CI 1.46 to 4.14) and within 30 days of life (2.6% vs 1.0%, RR 2.56, 95% CI 1.56 to 4.28), respectively (table 4). However, the number of ED visits was limited by small numbers of newborns. There was also a greater risk of jaundice among readmitted exclusively breastfed NeoWISE as compared with formula-fed newborns within the first 7 days of life (3.0% vs 1.3%, RR 2.30, 95% CI 1.47 to 3.61) and within 30 days of life (3.1% vs 1.3%, RR 2.37, 95% CI 1.51 to 3.71), respectively (table 5).

Table 3. Initiated feeding method in hospital and risk for visit to emergency department and readmission to hospital within 30 days of life in the NeoWISE cohort (n=5079).

	Exclusive breastfeeding, n (%) N=3321	Formula feeding only, n (%) N=1758	Relative risk (95% CI)
Within 7 days of hospital discharge
Reasons for visit to the emergency department*
Jaundice†	79 (2.4%)	17 (1.0%)	2.46 (1.46 to 4.14)
Feeding problem‡	15 (0.5%)	9 (0.5%)	0.88 (0.39 to 2.01)
Gastrointestinal problem§	≤5 (0.1%)	≤5 (0.2%)	0.40 (0.09 to 1.77)
Respiratory problem¶	≤5 (0.1%)	≤5 (0.2%)	0.40 (0.09 to 1.77)
Reasons for readmission to hospital**
Jaundice††	100 (3.0%)	23 (1.3%)	2.30 (1.47 to 3.61)
Gestational age or birth weight-related issue‡‡	≤5 (0.1%)	0 (0.0%)	NA
Neonatal withdrawal symptoms§§	≤5 (0.0%)	≤5 (0.1%)	0.53 (0.03 to 8.46)
Feeding problem¶¶	≤5 (0.1%)	≤5 (0.1%)	0.79 (0.13 to 4.75)
Other***	≤5 (0.1%)	0 (0.0%)	NA
Within 30 days of life
Reasons for visit to the emergency department*
Jaundice†	88 (2.6%)	18 (1.0%)	2.59 (1.56 to 4.28)
Feeding problem‡	29 (0.9%)	19 (1.1%)	0.81 (0.45 to 1.44)
Gastrointestinal problem§	7 (0.2%)	9 (0.5%)	0.41 (0.15 to 1.10)
Respiratory problem¶	7 (0.2%)	9 (0.5%)	0.41 (0.15 to 1.10)
Reasons for readmission to hospital**
Jaundice††	103 (3.1%)	23 (1.3%)	2.37 (1.51 to 3.71)
Gestational age or birth weight-related issue‡‡	≤5 (0.1%)	≤5 (0.1%)	1.06 (0.19 to 5.77)
Neonatal withdrawal symptoms§§	≤5 (0.0%)	≤5 (0.1%)	0.26 (0.02 to 2.92)
Feeding problem¶¶	6 (0.2%)	≤5 (0.2%)	0.79 (0.22 to 2.81)
Other***	≤5 (0.1%)	0 (0.0%)	NA

Newborns could have multiple healthcare utilisation encounters.

Bold font indicates statistically significant result.

^*Data obtained from the National Ambulatory Care Reporting System, main problem (ICD-10).

^†Jaundice (combine P590 neonatal jaundice associated with preterm delivery; P599 neonatal jaundice, unspecified; P925 neonatal jaundice from breast milk inhibitor).

^‡Feeding problem (combine P929 feeding problem of newborn, unspecified; P593 neonatal difficulty in feeding at breast; P924 overfeeding of newborn; P928 other feeding problems of newborn).

^§Gastrointestinal problem (combine P920 vomiting in newborn; P921 regurgitation and rumination in newborn).

^¶Respiratory problem (P229 respiratory distress of newborn, unspecified).

^**Data obtained from the CIHI Discharge Abstract Database, primary diagnosis code (ICD-10-1).

^††Jaundice (combine P599 neonatal jaundice, unspecified; P590 neonatal jaundice associated with preterm delivery; P593 neonatal jaundice from breast milk inhibitor).

^‡‡Gestational age or birth weight-related issue (combine P071 other low birth weight; P073 other preterm infants).

^§§Neonatal withdrawal symptoms (P961 neonatal withdrawal symptoms from maternal use of drugs of addiction).

^¶¶Feeding problem (combine P928 other feeding problems of newborn; P929 feeding problem of newborn, unspecified).

^***Other (P90 convulsions of newborn; P916 hypoxic ischaemic encephalopathy of newborn.

CIHI, Canadian Institute for Health Information; ICD-10, International Statistical Classification of Diseases, Tenth Revision; NA, not available; NeoWISE, Neonates With In-utero SSRI Exposure; SSRI, selective serotonin reuptake inhibitor.

Table 4. Cases involving one or more emergency department visits within 7 days and 30 days of life in the NeoWISE cohort (n=5079).

	Exclusive breastfeeding, n (%) N=3321	Formula feeding only, n (%) N=1758	P value
Reasons for visit to the emergency department*
Jaundice† within 7 days
One visit	68 (2.0%)	16 (0.9%)	0.002
Two or more visits	11 (0.3%)	≤5 (0.1%)
Jaundice† within 30 days
One visit	76 (2.3%)	17 (1.0%)	<0.001
Two or more visits	12 (0.4%)	≤5 (0.1%)
Feeding problem‡ within 7 days
One visit	15 (0.5%)	9 (0.5%)	0.766
Two or more visits	0 (0.0%)	0 (0.0%)
Feeding problem‡ within 30 days
One visit	29 (0.9%)	19 (1.1%)	0.467
Two or more visits	0 (0.0%)	0 (0.0%)
Gastrointestinal problem§ within 7 days
One visit	≤5 (0.1%)	≤5 (0.2%)	0.196
Two or more visits	≤5 (0.0%)	0 (0.0%)
Gastrointestinal problem§ within 30 days
One visit	6 (0.2%)	9 (0.5%)	0.09
Two or more visits	≤5 (0.0%)	0 (0.0%)
Respiratory problem¶ within 7 days
One visit	≤5 (0.1%)	≤5 (0.2%)	0.196
Two or more visits	≤5 (0.0%)	0 (0.0%)
Respiratory problem¶ within 30 days
One visit	6 (0.2%)	9 (0.5%)	0.09
Two or more visits	≤5 (0.0%)	0 (0.0%)

Bold font indicates statistically significant result.

^*Data obtained from the National Ambulatory Care Reporting System, main problem (ICD-10).

^†Jaundice (combine P590, neonatal jaundice associated with preterm delivery; P599, neonatal jaundice, unspecified; P925, neonatal jaundice from breast milk inhibitor).

^‡Feeding problem (combine P929, feeding problem of newborn, unspecified; P593, neonatal difficulty in feeding at breast; P924, overfeeding of newborn; P928, other feeding problems of newborn).

^§Gastrointestinal problem (combine P920, vomiting in newborn; P921, regurgitation and rumination in newborn).

^¶Respiratory problem (P229, respiratory distress of newborn, unspecified).

ICD-10, International Statistical Classification of Diseases, Tenth Revision; NeoWISE, Neonates With In-utero SSRI Exposure; SSRI, selective serotonin reuptake inhibitor.

Table 5. Cases involving one or more hospital readmissions within 7 days and 30 days of life in the NeoWISE cohort (n=5079).

	Exclusive Breastfeeding n (%) N=3321	Formula feeding only, n (%) N=1758	P value
Reasons for readmission to hospital*
Jaundice† within 7 days
One readmission	92 (2.8%)	23 (1.3%)	<0.001
Two or more readmissions	8 (0.2%)	0 (0.0%)
Jaundice† within 30 days
One visit	94 (2.8%)	23 (1.3%)	<0.001
Two or more visits	9 (0.3%)	0 (0.0%)
Gestational age or birth weight-related issue‡ within 7 days
One readmission	≤5 (0.1%)	0 (0.0%)	0.145
Two or more readmissions	0 (0.0%)	0 (0.0%)
Gestational age or birth weight-related issue‡ within 30 days
One visit	≤5 (0.1%)	≤5 (0.1%)	0.947
Two or more visits	0 (0.0%)	0 (0.0%)
Neonatal withdrawal symptoms§ within 7 days
One readmission	≤5 (0.0%)	≤5 (0.1%)	0.647
Two or more readmissions	0 (0.0%)	0 (0.0%)
Neonatal withdrawal symptoms§ within 30 days
One visit	≤5 (0.0%)	≤5 (0.1%)	0.243
Two or more visits	0 (0.0%)	0 (0.0%)
Feeding problem¶ within 7 days
One readmission	≤5 (0.1%)	≤5 (0.1%)	0.8
Two or more readmissions	0 (0.0%)	0 (0.0%)
Feeding problem¶ within 30 days
One visit	6 (0.2%)	≤5 (0.2%)	0.72
Two or more visits	0 (0.0%)	0 (0.0%)
Other** within 7 days
One readmission	≤5 (0.1%)	0 (0.0%)	0.303
Two or more readmissions	0 (0.0%)	0 (0.0%)
Other** within 30 days
One visit	≤5 (0.1%)	0 (0.0%)	0.303
Two or more visits	0 (0.0%)	0 (0.0%)

Bold font indicates statistically significant result.

^*Data obtained from the CIHI Discharge Abstract Database, primary diagnosis code (ICD-10-1).

^†Jaundice (combine P599 neonatal jaundice, unspecified; P590, neonatal jaundice associated with preterm delivery; P593, neonatal jaundice from breast milk inhibitor).

^‡Gestational age or birth weight-related issue (combine P071 other low birth weight; P073, other preterm infants).

^§Neonatal withdrawal symptoms (P961, neonatal withdrawal symptoms from maternal use of drugs of addiction).

^¶Feeding problem (combine P928, other feeding problems of newborn; P929, feeding problem of newborn, unspecified).

^**Other (P90, convulsions of newborn; P916, hypoxic ischaemic encephalopathy of newborn).

CIHI, Canadian Institute for Health Information; ICD-10, International Statistical Classification of Diseases, Tenth Revision; NeoWISE, Neonates With In-utero SSRI Exposure; SSRI, selective serotonin reuptake inhibitor.

Discussion

In our retrospective cohort study of Ontario NeoWISE whose mothers were dispensed an SSRI during pregnancy through the ODB program, we investigated the association between newborn hospital feeding method and postnatal hospital LOS, ED visits and readmission to the hospital. Overall, the risk of ED visits and hospital readmissions was low in both feeding groups. Exclusively breastfed newborns had a 16% lower risk for prolonged LOS after birth; however, they had a 42% higher risk for ED visits and a 78% higher risk for hospital readmission. Although numbers were small, jaundice was the most frequent reason for healthcare utilisation after hospital discharge in both feeding groups, with breastfed newborns at greater risk. This is an expected finding, as jaundice is a common reason for readmission among non-exposed newborns.³⁴ In a systematic review of risk factors associated with 31-day unplanned hospital readmission comprising a total of 28 studies published between 1 January 2000 and 30 June 2021, newborns with jaundice or hyperbilirubinaemia had the highest rates of readmission (up to 16.1%).³⁴ This contrasts with findings of a recent systematic review and meta-analysis in which in-utero SSRI exposure was reported to be protective of jaundice (OR 0.80, 95% CI 0.65 to 0.98); however, there was a greater likelihood of hypoglycaemia (OR 1.30, 95% CI 1.08 to 1.57),¹⁹ which may be related to feeding problems.

A leading clinical practice resource related to medication use in lactation identified the need for newborn monitoring following SSRI exposure, specifically sedation or irritability, not waking for feeding, poor feeding and weight gain.³⁵ SSRI medications may cause newborns to be more lethargic and thus not as able to achieve an effective latch. Ineffective transfer of milk and/or insufficient milk intake can result in elevated bilirubin levels, contributing to infant sleepiness and increasing the risk of requiring treatment for jaundice.

It remains unclear why the same trend of a higher risk for readmission to the hospital, although slightly attenuated, was observed in the any breastfeeding group compared with the exclusive breastfeeding group. Unmeasured confounding as a possible explanation would require that the unmeasured covariate(s) associated with breastfeeding would bias the risk of re-admission upwards away from the null, which seems unlikely. Our lack of information on cessation of or uptake of breastfeeding after discharge could have resulted in differential misclassification of feeding method.

More likely, insufficient milk transfer and intake resulting in elevated levels of bilirubin may explain our findings of increased risk for jaundice and the observed healthcare utilisation after hospital discharge. In a retrospective cohort study of live births in British Columbia between 1 April 1997 and 31 March 2002 (n=119 547), Oberlander et al³⁶ reported that SSRI-exposed newborns, compared with newborns of mothers who had depression but did not take medication, had a higher prevalence of jaundice (9.4% vs 7.5%) and feeding problems (3.9% vs 2.4%), respectively. These data demonstrate the underlying risk of poor feeding associated with in-utero SSRI exposure.

The higher risk of a prolonged hospital stay for formula-feeding newborns in the NeoWISE cohort may be due to extra monitoring (eg, additional temperature checks, extra blood glucose monitoring to ensure stable blood glucose levels or observation of additional feeding) to ensure newborns were ready for discharge home.³⁷ The formula-feeding group had a higher rate of caesarean birth (29.9% vs 18.2%) and being premature (<37 weeks) (5.1% vs 3.5%), which may have contributed to greater health needs resulting in a longer hospital stay. Although type of birth and gestational age were included as covariates in the propensity score analysis, there may have been an unmeasured confounding factor contributing to the findings. Respiratory distress among in-utero SSRI-exposed newborns has been reported as a statistically significant adverse outcome, ranging in prevalence from 5.7%³⁸ to 13.9%.³⁶ The risk for respiratory distress syndrome is higher for newborns born via caesarean and for premature newborns due to surfactant deficiency, immature lung anatomy and physiology.³⁹ Additionally, we noted that 6.3% (n=110) of formula-fed newborns were small for gestational age (<5%) as compared with 4.1% (n=135) of exclusively breastfeeding NeoWISE. Formula-fed NeoWISE and their mothers may have had greater health needs leading their mothers to choose to formula feed only. This is supported by the observation that a third of mothers of the formula-fed newborns (n=546, 31%) had intended to breastfeed but did not initiate breastfeeding. Although breastfeeding and/or providing expressed breast milk is strongly encouraged by HCPs, mothers may have chosen to formula feed instead.

A potential unmeasured confounding factor could be related to hospital policies such as mandatory observation periods in the NICU or rooming-in practices for the mother-baby dyad (eg, skin-to-skin contact)^40–42 that may have resulted in an indirect, positive benefit, particularly the breastfeeding NeoWISE. We previously reported that exclusively breastfed newborns had a 20% reduced risk of transfer to the NICU compared with formula-fed newborns.²⁰ Kieviet et al¹³ highlighted a potential benefit of breastfeeding in reducing the risk of poor neonatal adaptation but the precise mechanism is not known. We postulate that exclusively breastfed newborns may have benefited in the short term from a longer duration of skin-to-skin contact that occurs during breastfeeding.

It is unknown why the association between ED visits and breastfeeding among NeoWISE with third trimester SSRI exposure were no longer significant but may be due to reduced power. Further research with a larger sample is required to elucidate the clinical significance of exposure to SSRI medication in the different trimesters of pregnancy as it relates to postnatal healthcare needs.

Altogether, the increased risk observed for exclusively breastfed NeoWISE suggest that the small amount of breastmilk transferred in the first few days after birth may not be sufficient to prevent withdrawal signs or behavioural impacts associated with in-utero SSRI exposure. The observed higher rate of ED visits observed in the breastfeeding NeoWISE was likely attributable to low fluid intake, ineffective milk transfer and sleepiness, leading to dehydration and higher rates of jaundice. Although jaundice is commonly experienced by newborns, further research is required to investigate the feeding efficacy of SSRI-exposed newborns. The nature of feeding issues has not been elucidated in the literature.⁴³

From a healthcare system planning perspective, the declining trend in the postnatal LOS in Ontario among the general population of newborns⁴⁴ represents a potential concern for NeoWISE, particularly in regard to increased risk for jaundice in the first 7 days after hospital discharge. It is unknown if this group of exposed newborns would benefit from a longer hospital stay or specialised postnatal follow-up, given our observation of increased healthcare utilisation among breastfeeding NeoWISE.

According to the Canadian Pediatric Society’s position statement, the recommended monitoring after birth, discharge timing and follow-up for newborns with in-utero SSRI exposure are no different than those for unexposed newborns.¹⁸ Further research is needed to confirm this standard of care, given our observations that a subset of this population of exposed newborns required medical care beyond routine newborn follow-up with their primary care provider. It may take a few days for the withdrawal signs to fully manifest in this newborn population.²⁹ It is reassuring that there were no differences in risk for diagnosis of neonatal withdrawal symptoms after hospital discharge due to initiated newborn feeding method in hospital. That said, there may be underlying impacts of in-utero SSRI exposure on newborn feeding capability which should be investigated further.

Strengths

We intentionally restricted our study population to newborns born to pregnant women taking AD medication to account for the baseline risk of perinatal depression and anxiety itself and compared breastfed with formula-fed newborns. Including newborns with SSRI exposure in both feeding groups enabled us to examine the association of newborn feeding method and outcomes while minimising confounding by indication.

A strength of our study was the novel investigation of healthcare utilisation among NeoWISE following hospital discharge. Additional strengths of this study included the use of a province-wide cohort study design that linked multiple health administrative databases and validated registries to identify a large sample size of NeoWISE compared with other cohorts. The use of IPTW of propensity scores to balance differences in baseline characteristics related to initiated newborn feeding methods reduced confounding.³⁰

Limitations

Breastfeeding data were only available during the hospitalisation after birth. It is unknown if mothers continued using this feeding method in the following weeks. Therefore, the observed associations between ED visits and hospital readmissions may have been affected by misclassification of exposure (feeding) using the groups we assembled for this cohort. The temporality of the measurement of outcomes may be another source of bias. The length of postnatal hospital stay variable represents a few days after birth as compared with up to 30 days after birth for ED visits and hospital readmissions. The reporting of healthcare utilisation within 7 days after discharge was helpful to pinpoint the priority areas of concern.

Newborn feeding data were not recorded for 25% of the Ontario NeoWISE born within the study timeframe and were excluded from the cohort. Therefore, their postnatal LOS and healthcare utilisation are unknown, and the inclusion of their data may have altered study results if it had been available.

Healthcare utilisation was limited to ED visits and readmission to the hospital. We did not explore other types of healthcare utilisation such as routine postnatal care visits to primary care providers, appointments at walk-in clinics, well-baby drop-in visits with public health nurses and/or community health nurses, consultations with International Board Certified Lactation Consultants for breastfeeding problems or calls to Telehealth Ontario (a free 24/7 provincial hotline). These data sources may have provided additional insights into the healthcare needs of the NeoWISE cohort.

A final limitation is that the cohort was restricted to women eligible for the ODB program, as this was the only source of SSRI prescription claims data available within ICES. This limits the generalisability of research findings, as the study cohort represents <1% of the population of births in Ontario between 1 April 2012 and 31 March 2020.²⁰ Additionally, the mothers of the NeoWISE cohort were much younger than the general population of childbearing women in Ontario, which may have impacted their breastfeeding rates.²⁰

In Ontario, anxiety (5.9%–15.1%) and depression (7.2%–9.7%) were the most frequently reported mental health concerns in the study timeframe between 2012 and 2020, respectively, showing a steady upward trend with each successive year.⁴⁴ SSRI medication use in pregnancy is common with an international estimated prevalence of 3.0% (95% CI 2.3 to 3.7) across a 26-year period (1989–2015) and the highest pooled prevalence reported in North America (5.5%).⁷ Trimester of SSRI exposure was essential to differentiate as previous researchers have identified an association between the third trimester SSRI exposure and newborn withdrawal signs and behavioural impacts.^{18 19} We anticipate that the association would likely also be present among higher-income mother-baby dyads.

Future research

Future research is required to better estimate whether newborn feeding has a causal effect on healthcare utilisation among NeoWISE. A prospective cohort study design may facilitate improved data collection on maternal mental health status, medication use during pregnancy, newborn feeding patterns, neonatal behavioural/withdrawal signs and healthcare services accessed in the first month of life. More rigorous data collection would improve surveillance and reporting of adverse outcomes in NeoWISE and increase the generalisability of the findings. It is also essential to capture the decision-making process in relation to newborn feeding to assess whether changes in feeding are due to the medical needs of newborns or based on parental choice. Additionally, it is necessary to identify if ED visits after hospital discharge are medically indicated as urgent or represent a lack of access to primary care. Mixed-method studies, incorporating both quantitative and qualitative research components, would significantly contribute to our understanding of the healthcare needs of this population. The results from these research designs will provide evidence policymakers can use when determining optimal postnatal observation timeframes, follow-up healthcare services and other services required to support these maternal-newborn dyads.

Conclusions

The findings of our study provide further evidence in support of the safety of SSRI use in pregnancy as the first-line pharmacotherapeutic intervention for perinatal mood and anxiety disorders by contributing knowledge about the impact of the feeding method among NeoWISE. These new observations of a lower risk for prolonged postnatal LOS and a higher risk of jaundice leading to ED visits and hospital readmission among breastfed NeoWISE need to be confirmed in another study to ascertain if the risk is greater than the risk observed in the general population of newborns. Women taking SSRI medication during pregnancy should receive anticipatory guidance about newborn feeding effectiveness, signs of jaundice and when to seek healthcare after birth. NeoWISE may benefit from increased monitoring, regardless of their feeding method. Although serious adverse outcomes are rare, newborns require timely healthcare when needed.

The analyses, conclusions, opinions and statements expressed here are solely those of the authors and do not reflect those of the funding or data sources. No endorsement is intended, nor should any be inferred.

Data availability statement

Data are available on reasonable request.