Introduction
The lifetime risk for depression is greater in women than men ( ) and is especially high during female reproductive events including the perimenstrual, peripartum, and perimenopausal phases. The increased depression prevalence in women during reproductive transitions is hypothesized to be due to psychological, social, and physiological changes including the unique temporal hormonal fluctuations associated with each transition. Treatments available for women suffering from depression during the reproductive life cycle are similar to those for the general population and include psychotherapy, pharmacotherapy, and neuromodulation. Limited data are available on treatment-resistant depression (TRD) during the peripartum and perimenopausal periods.
In this chapter, we will review the epidemiology, risk factors, symptomatology, and distinct treatment considerations for unipolar depression occurring during the peripartum and perimenopausal periods. We will also discuss the treatment of peripartum bipolar illness including data on the relative safety profiles of mood stabilizers and antipsychotics in pregnancy and lactation.
Unipolar peripartum depression
Diagnosis of unipolar peripartum depression
The Diagnostic and Statistical Manual Fifth Edition (DSM-5) defines peripartum depression as a major depressive episode with peripartum onset for patients whose symptoms arise during pregnancy or within 4 weeks of delivery ( ). Despite this narrow definition of the postpartum period, many clinicians and the World Health Organization (WHO) extend the definition of the postpartum period to up to 1 year after childbirth ( ). Peripartum depression is frequently distinguished by the presence of anxiety with excessive concern for the child’s health and fear of harming the child. A 2017 study isolated peripartum depressive symptom patterns that differed in severity, time of onset, and type of symptoms, and categorized these distinct presentations into the following five subtypes: severe anxious depression, moderate anxious depression, anxious anhedonia, pure anhedonia, and resolved depression ( ).
Epidemiology of unipolar peripartum depression
Peripartum depression is one of the most common complications of childbirth and is a serious public health problem. In 2018, the Centers for Disease Control and Prevention (CDC) estimated the overall prevalence of postpartum depression in the United States (U.S.) to be 13%, with a range of 9.7%–23.5% by state ( ). According to a 2019 CDC analysis, the rate of depression diagnosis at delivery increased sevenfold from 2000 to 2015 ( ). Among women with peripartum depression, 11.5% develop symptoms antenatally, 66.5% experience symptoms within 6 weeks of delivery, and 22% exhibit symptoms within 12 months of delivery ( ).
Despite the high prevalence of peripartum depression, it is frequently underdiagnosed and undertreated with resultant deleterious consequences to mothers and their offspring. Of women with antenatal depression, only 49.9% are identified in clinical settings, 13.6% receive treatment, 8.6% receive adequate treatment, and 4.8% achieve remission. Of women with postpartum depression, only 30.8% are identified in clinical settings, 15.8% receive treatment, 6.3% receive adequate treatment, and 3.2% achieve remission ( ).
The U.S. Preventive Services Task Force and the American Psychiatric Association (APA) recommend screening for depression in pregnant and postpartum women ( ; ). The APA recommends screening with a validated tool twice during pregnancy, once in early pregnancy and again later in the pregnancy, as well as postpartum during pediatric visits during the first 6 months as recommended by the American Academy of Pediatrics (AAP) ( ; ). The Edinburgh Postnatal Depression Scale (EPDS) and the Patient Health Questionnaire 9 (PHQ-9) are two self-rating screening tools that are validated to identify peripartum depression ( ); however, overreliance on peripartum depression screening tools may lead to misdiagnosis if screening is not followed by thorough psychiatric diagnostic evaluation. In women with an EPDS score > 11/30, 31.4% had major depressive disorder (MDD), 13.1% had bipolar disorder, 60.8% had an anxiety disorder ( ).
Untreated or undertreated antenatal depression can impair a mother’s ability to function and care for herself resulting in inadequate weight gain and is associated with nicotine, alcohol, and substance use ( ). Antenatal depression is also associated with increased risk of miscarriage ( ), pregnancy-induced hypertension, preeclampsia, preterm or operative delivery, fetal growth restriction ( ), low birth weight ( ), maternal–infant attachment difficulties ( ), and impaired lactation ( ). Additionally, offspring of depressed mothers are at increased risk of neurodevelopmental delay ( ) as well as emotional and behavioral problems, including depression ( ), hyperactivity, and conduct problems ( ). Tragically, maternal suicide is the leading cause of direct maternal mortality in the first postpartum year ( ).
Risk factors associated with unipolar peripartum depression
Numerous risk factors are associated with peripartum depression. Biological risk factors include personal or family history of psychiatric illness and smoking, alcohol or substance use ( ). Psychosocial risk factors include poor social support, stressful life events, adverse childhood experiences, domestic violence, and lifetime history of abuse ( ). Demographic and economic factors include young or advanced maternal age, African American or Latina race, single status, unemployment, and low education level or socioeconomic status ( ). Pregnancy-related factors include multiparity, unintended or unwanted pregnancy, present or past pregnancy complications, and pregnancy loss ( ).
Overview of treatment of unipolar peripartum depression
The treatment of depression in pregnant and postpartum women involves use of similar evidence-based psychological, pharmacotherapeutic, and neuromodulation interventions as in nonperipartum women, with some special considerations. When developing an individualized treatment plan for a peripartum patient, the provider should consider factors including gestational age (i.e., early, middle, or late gestation), plans for lactation or lactation status, psychiatric diagnoses, the severity of current and past symptoms, a suicidal and homicidal risk assessment, past psychiatric history particularly related to past pregnancies, efficacy of prior and current psychiatric treatments to determine level of treatment-resistance, medical comorbidities, family psychiatric history, patient preferences for treatment, and ability to adhere to the recommended treatment plan.
Psychotherapeutic treatment of unipolar peripartum depression
An APA taskforce recommends either psychotherapy or antidepressants as first-line treatment for mild-to-moderate peripartum depression ( ). Psychotherapy should also be considered as an adjunct to antidepressants in moderate-to-severe depression and in cases where women decline to take antidepressants. Both cognitive–behavioral therapy and interpersonal psychotherapy are evidence-based, effective forms of psychotherapy for peripartum depression ( ; ). For a comprehensive review of the evidence base for psychological interventions in peripartum depression, we refer the reader to Stuart and Koleva ( ).
Pharmacologic treatment of unipolar peripartum depression
Evaluating the safety and efficacy of antidepressant use for peripartum depression
Safety is an important consideration when choosing among psychopharmacotherapies for peripartum women. Past restrictions on randomized controlled trials (RCTs) of psychopharmacotherapies in pregnancy limited available data on safety and efficacy of even commonly used antidepressants. Most published pharmacotherapy studies are observational in design and therefore limited in accurate assessments of reproductive drug safety due to the potential for selection bias, recall bias, confounding, and exposure or outcome misclassification. While there are inconsistent data and quality of the literature, when evaluating the risk of antidepressant use during the peripartum period, it is crucial to ensure that the study adequately controlled for confounders, used a representative sample, and had results that have been replicated.
Antidepressants are indicated for moderate-to-severe peripartum depression, with selective serotonin reuptake inhibitors (SSRIs) typically considered first-line therapy. While there are no RCTs examining efficacy of antidepressants in antenatal depression, eight RCTs ( ; ; ; ; ; ; ; ) and four open-label studies ( ; ; ; ) have assessed use of SSRIs in treatment of postpartum depression. A meta-analysis of three studies comparing treatment with SSRIs with placebo for treatment of postpartum depression found that patients randomized to SSRIs were more likely to show response or remission of depressive symptoms at follow-up ( ). A systematic review reported that SSRIs, nortriptyline, and psychotherapy are effective for short-term treatment of postpartum depression, but there is not enough evidence to demonstrate clear superiority of either ( ). Despite limited evidence, there is consensus supporting the use of antidepressants in both antenatal and postpartum depression.
There are no RCTs evaluating use of selective norepinephrine reuptake inhibitors (SNRIs), bupropion, mirtazapine, trazodone, nefazodone, tricyclic antidepressants (TCAs) (except nortriptyline), or monoamine oxidase inhibitors (MAOIs). Open-label studies support efficacy for venlafaxine ( ), desvenlafaxine ( ), bupropion ( ), and nefazodone ( ) in treatment of postpartum depression.
General considerations for the antenatal use of antidepressants
When possible, any changes to a medication regimen should be made prior to conception to ensure symptom stability and to minimize exposure to the fetus. Abrupt discontinuation of antidepressants in pregnant women with a history of depression is associated with high risk of relapse without known benefit ( ). Women with history of recurrent unipolar MDD or history of symptom recurrence associated with past reduction or discontinuation of pharmacotherapy should not discontinue their antidepressant during pregnancy. All psychotropic medications cross the placenta. When considering the use of antidepressants during pregnancy, it is preferable to prescribe those with a well-studied reproductive safety profile, often older antidepressants which have been studied over time, and to use a single antidepressant at the lowest effective dose rather than polypharmacy.
Treatment in pregnancy is complicated by physiologic and pharmacokinetic changes, including changes in gastrointestinal absorption, hepatic and renal blood flow, changes in glomerular filtration rate, and changes in both phase 1 (hepatic cytochrome P450) and phase 2 (uridine diphosphate glucuronosyltransferase) hepatic enzyme activities ( ). Changes in maternal drug pharmacokinetics in conjunction with placental transfer and fetal drug metabolism affect fetal psychotropic drug exposure ( ). Pregnancy-associated changes in absorption, distribution, metabolism, and elimination may lower psychotropic drug levels and possibly decrease treatment effects, particularly in late pregnancy. Furthermore, increased plasma volume and changes in protein binding may increase volume of distribution for lipophilic drugs ( ). Dose increases are often clinically necessary for citalopram, escitalopram, clomipramine and imipramine after 20 weeks’ gestation. In addition, antenatal dose requirements for fluoxetine, fluvoxamine, nortriptyline, paroxetine, and sertraline often increase in the third trimester ( ). Patients should be educated on the likely need for dose increases as gestation progresses, and symptoms should be monitored closely for changes in severity that would indicate need such increases.
While there is inadequate evidence to recommend routine therapeutic drug monitoring (TDM) of antidepressants other than TCAs during the peripartum period, measurement of serum drug levels when using TCAs to monitor efficacy and prevent toxicity is clinically necessary ( ). During pregnancy, we recommend monthly monitoring of TCA trough levels and frequent psychiatric symptom assessment. In the postpartum period, we recommend weekly monitoring of depressive symptoms and a review of systems to monitor for TCA side effects. The TCA dose often must slowly be decreased towards the preconception dose during the 2–6 weeks after delivery to reduce risk of TCA toxicity. We recommend checking a TCA blood level whenever adverse effects emerge in the early postpartum or at week six when nortriptyline level/dose ratios have been reported to peak.
When considering use of antidepressants during pregnancy, potential risk of under- or untreated depressive symptoms in both the mother and fetus must be weighed against the potential risk of antidepressant exposure for the following types of adverse events: (1) fetal organ malformation, (2) poor pregnancy and birth outcomes including spontaneous abortion, stillbirth, preterm birth, low birth weight and postpartum hemorrhage, (3) neonatal complications including poor neonatal adaptation syndrome and persistent pulmonary hypertension of the newborn, and (4) long-term neurobehavioral sequelae in the offspring. Regarding potential risks of antenatal antidepressant use, most studies have evaluated the safety of SSRIs and SNRIs, while less data are available on other antidepressants such as TCAs and MAOIs. For women with unipolar TRD who often do not achieve remission with nonpharmacotherapeutic interventions alone, the overall small absolute risks of antidepressant use during gestation often outweigh the risks of untreated depression. All women should be educated on the potential risks of pharmacotherapy during the peripartum period. The discussion of the risk/benefits/alternatives to treatment, the patient’s capacity to make the treatment decision, and ideally the partner’s agreement to treatment should be documented in the patient’s chart.
Risk of teratogenesis with antenatal antidepressant use
Birth defects affect 3% of infants born in the U.S. each year ( ). Any consideration of antidepressant effects on rate of teratogenesis must consider baseline rates. Most studies indicate that antidepressants are not associated with overall major congenital malformations ( ; ). While some studies have observed a correlation between antidepressant exposure and specific birth defects, (e.g., anencephaly, craniosynostosis, and omphalocele), only a small increase in the absolute risk (AR) has been observed ( ). Some studies have noted the potential for a small increase in AR for cardiovascular defects ( ), specifically septal defects, in antidepressant-exposed infants, particularly paroxetine; however, many other studies, after controlling for underlying maternal depressive illness and genetics, have found no association between antenatal antidepressant use and heart defects ( ; ; ).
Pregnancy and birth complications associated with antenatal antidepressant use
Risk of pregnancy loss
Miscarriage is a relatively common outcome of pregnancy, with most studies reporting 12%–15% loss among recognized pregnancies by 20 weeks’ gestation ( ). Across multiple studies, antenatal antidepressant use may contribute a small to nonexistent risk for spontaneous abortion after adjusting for potential confounders ( ; ). Antenatal antidepressant use does not appear to be associated with elevated risk of stillbirth, neonatal death or postneonatal death after controlling for potential confounders ( ).
Risk of preterm birth
There are mixed data regarding association between antenatal antidepressant use and preterm birth (PTB), defined as delivery at < 37 weeks of gestation. The PTB rate in U.S. infants is 10% ( ). A large national registry study which adjusted for potential confounding factors found that risk of PTB was lower in neonates with antenatal exposure to SSRIs than in unexposed neonates (4.7% vs 5.4%) ( ). A meta-analysis reported that depressed women with antenatal SSRI use had increased risk of PTB compared with controls (adjusted odds ratio (aOR) 1.24, 95% confidence interval (CI) 1.09–1.41) ( ). Control groups included both women with depression but without SSRI exposure and women without depression or SSRI exposure. In a subgroup analysis, risk of PTB remained significant when comparing depressed women with SSRI use with depressed women treated with psychotherapy alone (6.8% vs 5.8%; OR 1.17, 95% CI 1.10–1.25). Other studies indicate that antenatal antidepressant exposure may be associated with PTB ( ; ) but that the reduction in gestational age at birth (approximately 3 days shorter) may not be clinically significant.
Risk of low birth weight
Approximately 8.28% of neonates in the U.S. are born at a low birth weight (LBW), defined as < 2500 g ( ). There are mixed data regarding an association between antenatal SSRI exposure and LBW. A meta-analysis found that antenatal SSRI exposure was associated with LBW (relative risk (RR) 1.48, 95% CI 1.22–1.79) ( ); however, there was significant heterogeneity across studies and few studies adequately controlled for confounding factors. A larger meta-analysis observed that neonates exposed to antenatal antidepressants (mostly SSRIs) had lower birth weights than unexposed newborns ( ). However, the mean difference between exposed and unexposed babies was noted to be small (74 g) and unlikely to be clinically significant. Furthermore, when the control group was limited to neonates without antenatal antidepressant exposure born to depressed mothers, there was no longer a significant association between antidepressant exposure and lower birth weight. In summary, if there is an association, the AR increase for LBW appears to be small with antidepressant use, and studies have found that untreated antenatal depression is associated with LBW ( ).
Risk for postpartum hemorrhage
Postpartum hemorrhage (PPH), classically defined as an estimated blood loss ≥ 500 mL after vaginal birth or ≥ 1000 mL after cesarean delivery, has been estimated to occur in 1%–3% of U.S. deliveries ( ). Some studies have associated serotonergic antidepressant use with a small increased AR for PPH, hypothesized to be related to serotonin-induced inhibition of platelet aggregation ( ). A large U.S. Medicaid database study found an increased risk for PPH in women who used serotonergic antidepressants close to delivery, after adjustment for maternal illness (adjusted relative risk (aRR) 1.42, 95% CI 1.27–1.57) ( ). A significant study limitation was the inability to control for other potential confounders including alcohol, drug, and tobacco use or the use of over-the-counter medications. Other large studies have documented no increase in risk of PPH with antenatal serotonergic antidepressant use ( ).
Neonatal complications associated with antenatal antidepressant use
Poor neonatal adaptation syndrome
Antenatal exposure to antidepressants has been associated with poor neonatal adaptation syndrome (PNAS) in up to 30% of neonates ( ). PNAS is characterized by typically mild physical and neurobehavioral symptoms including abnormal irritability, sleep disturbances, lethargy, jitteriness, mild respiratory distress, hypothermia, feeding disturbances, altered muscle tone, tremor and rarely seizures ( ). PNAS is hypothesized to be due to a side effect of antidepressants persisting in the infant (antidepressant toxicity) or to antidepressant withdrawal in the neonate (discontinuation syndrome) ( ). PNAS is a transient condition, usually developing hours after birth and self-resolving within days to 2 weeks ( ). One study observed persistent PNAS signs in newborns at 1 month of age, but they were subtle and not clinically significant ( ). PNAS symptoms are typically managed with supportive care, rarely requiring neonatal intensive care unit (NICU) admission. One large national registry study found that 13.7% of infants with antenatal exposure to SSRIs were admitted to the NICU compared with 8.2% of unexposed babies in the general population (aOR 1.5, 95% CI 1.4–1.5) ( ); however, this study concluded that the AR for severe disease was low. It is not recommended to discontinue or lower the antidepressant dose before delivery because this approach does not reduce the risk of PNAS ( ) but may increase the relapse risk of maternal depression.
Risk for persistent pulmonary hypertension of the newborn
Some studies have identified an association between late gestation antidepressant use with persistent pulmonary hypertension of the newborn (PPHN) ( ; ), a serious but rare condition, characterized by failure of the neonatal cardiopulmonary transition to ensue, that occurs in 1.9 per 1000 live births in the general population ( ). Newborns with PPHN typically present within 24 h after birth with signs of respiratory distress and cyanosis that can progress to severe respiratory failure requiring intubation and mechanical ventilation ( ). Risk factors include increased maternal age, smoking, group B streptococcus infection, diabetes mellitus, hypertension, substance abuse, meconium aspiration, cesarean delivery, and prematurity and postmaturity ( ). Among other known causes of PPHN, the pathophysiology of PPHN may involve serotonin-induced damage to the neonatal lungs from antenatal antidepressant exposure characterized by vasoconstriction of pulmonary vessels, leading to increased vascular resistance ( ).
A 2014 meta-analysis reported an associated between late gestation SSRI use and increased risk of PPHN (OR 2.50, 95% CI 1.32–4.73) ( ); however, there was a moderate degree of heterogeneity across the studies. Furthermore, it is important to consider that since the frequency of PPHN in the general population is low, the AR increase related to late pregnancy antidepressant exposure must also be low. This meta-analysis estimated that the AR difference between late pregnancy SSRI exposure and no exposure in the development of PPHN was approximately 3 per 1000 infants and that approximately 350 women would have to be treated with SSRIs in late pregnancy to result in one additional case of PPHN. A study limitation included inadequate control for factors associated with PPHN that are also common in depressed women, including obesity, cesarean delivery, and PTB. Finally, a large observational study found that antenatal exposure to SSRIs was associated with a much lower risk of PPHN or no association at all ( ). In summary, there appears to be either a nonexistent or small increase in the AR for PPHN from 1.9 per 1000 in the general population to approximately 3 per 1000 with late gestation SSRI use.
Risk for psychopathology in offspring
Antenatal antidepressant exposure does not appear to be associated with increased risk for attention deficit hyperactivity disorder (ADHD) or autism in the offspring, based upon several studies which adjusted for potential confounders ( ; ). Antenatal antidepressant use may be associated with increased risk of depression in the offspring. A large national registry study which controlled for potential confounding factors found that the cumulative incidence of depression was greater in offspring (with antenatal SSRI exposure) by age 15, compared with offspring of women with psychiatric disorders without SSRI use (8% vs 2%) ( ). Additionally, depression occurred more in offspring with antenatal SSRI exposure compared with offspring of women who used SSRIs prior but not during pregnancy (8% vs 3%). The strength of these conclusions, however, are limited due to potential residual confounding in the study by maternal postpartum depression, which has been shown to be linked with an increased risk of depression in offspring by 16 years of age ( ).
Antenatal use of atypical antidepressants in unipolar depression
Bupropion
There is inconsistent evidence regarding whether antenatal exposure to bupropion is associated with teratogenesis. While a few studies have reported an association between antenatal bupropion exposure and congenital heart defects, the type of defect differs across studies. A retrospective case-control study, which adjusted for potential confounding factors, concluded that there was a greater likelihood of first trimester bupropion exposure in neonates with left ventricular outflow tract obstruction (LVOTO) compared to newborns without malformations (OR 2.6, 95% CI 1.2–5.7) ( ); however, the AR was small (2.1 per 1000 births). Additionally, bupropion was not associated with other cardiac malformations, such as ventricular septal defects (VSDs). More recently, a large registry case-control study, which did not control for potential confounding from maternal mental illness, found that there was a greater likelihood of first trimester bupropion exposure in infants with VSDs than neonates without malformations (aOR 2.5, 95% CI 1.3–5.0); however, bupropion was not associated with other congenital heart defects, such as LVOTO defects ( ). The studies are limited in that neither controlled for confounding by indication (using bupropion for depression or smoking), due to the small number of infants exposed to bupropion antenatally.
Several other studies have reported that bupropion is not a major teratogen. A large retrospective cohort study observed that first trimester bupropion exposure was not associated with an increased risk for overall congenital malformations or cardiac malformations ( ). A separate large U.S. Medicaid database study evaluating first trimester bupropion exposure found no increased risk for any congenital cardiac malformations, VSDs, or right ventricular outflow tract obstruction (RVOTO) after adjustment for maternal illness ( ).
Bupropion is often a treatment choice for women with peripartum depression and comorbid ADHD or nicotine dependence. One multistate insurance claims database study found that fetal exposure to bupropion, particularly in the second trimester, was strongly associated with an increased risk for ADHD ( ). While this study adjusted for some confounding factors (i.e., sex of the child, parental psychiatric diagnoses, and peripartum complications), it did not control adequately for tobacco, alcohol, or drug use, and there may have been confounding by undiagnosed parental ADHD. Future research is needed to evaluate the potential risk for ADHD in offspring exposed prenatally to bupropion.
Mirtazapine
Mirtazapine is not often used as a first-line therapy for antepartum depression as there is limited information available concerning its safety profile during pregnancy. The available data suggest that mirtazapine is not a major teratogen. In a systematic review of six observational studies, mirtazapine was not associated with increased incidence of major congenital malformations ( ). Unlike other antidepressants, mirtazapine has antiemetic properties that have used in the treatment of hyperemesis gravidarum ( ), which is frequently associated with severe anxiety and depression. As such, mirtazapine may be a particularly favorable choice for pregnant women suffering from severe nausea and comorbid depression or anxiety. For women who have failed traditional SSRI or SNRI pharmacotherapies, mirtazapine may additionally be a beneficial choice.
Trazodone and nefazodone
Trazodone and nefazodone are not used as first-line therapy for antepartum depression or insomnia as they do not have well-studied reproductive safety profiles. Limited data on the antenatal use of trazodone and nefazodone are reassuring regarding the risk of teratogenesis and major pregnancy complications. One small prospective observational study of found that first trimester exposure to trazodone or nefazodone ( n = 147) was not associated with major malformations, miscarriages, or stillbirths ( ), but larger studies powered to find rare adverse effects are not available in the literature.
Antenatal use of TCAs in unipolar depression
TCAs are generally not used as first- or second-line therapy for antepartum depression, as they can be poorly tolerated due to their anticholinergic side effect profile. In addition, data on the risks of antenatal exposure to TCAs are limited and includes smaller sample sizes compared to those of SSRIs. TCAs should be reserved for use in women with TRD who have not responded to a SSRI or SNRI. Desipramine and nortriptyline are favored TCAs as they are less anticholinergic, and thereby less likely to worsen orthostatic hypotension and constipation during pregnancy. Most studies have found that exposure to antenatal TCA use is not associated with congenital malformations, including cardiac malformations ( ; ). Although TCAs as a class are not thought to be teratogenic, some data suggest that antenatal exposure to clomipramine may be associated with a modestly elevated risk of total malformations (OR 1.36, 95% CI 1.07–1.72), cardiovascular defects as a group (OR 1.63, 95% CI 1.12–2.36), and VSDs and/or atrial septal defects (aOR 1.84, 95% CI 1.13–2.97) ( ). A main limitation is that the analysis did not control for maternal depression, illicit drug use, or other nonantidepressant medication, and the results may have been due to confounding by indication.
Antenatal use of MAOIs in unipolar depression
MAOIs are not used as first- or second-line therapy for antepartum depression because of the potential for interactions with foods and other drugs as well as vasoconstrictive effects. In addition, data on the risks of antenatal exposure to MAOIs are primarily limited to case reports ( ; ). One study that evaluated fetal exposure to MAOIs with other antidepressants did not identify adverse infant outcomes ( ). MAOIs should generally be avoided during pregnancy due to insufficient data regarding their reproductive safety profile and restrictions associated with their use. MAOIs may be considered for use in women with TRD who have not responded to other agents.
Antidepressant use during lactation
Antidepressant use should not discourage women from breastfeeding, as the overwhelming benefits of breastfeeding usually outweigh the low risks of antidepressant use in women suffering from postpartum depression. Furthermore, postpartum depression is associated with decreased lactation duration and increased breastfeeding difficulties ( ). Data about safety of infant exposure to antidepressants through breast milk are derived from small observational studies, as no RCTs have been performed. The relative infant dose (RID) is an estimate of infant drug exposure from breast milk derived from dividing the dose provided to the infant via breast milk (mg/kg/day) by the mother’s weight-adjusted dose (mg/kg/day). For most drugs, a RID from breast milk of < 10% of the weight-adjusted maternal dose is satisfactory while breastfeeding a healthy infant. Caution is required for drugs with a RID from breast milk of 10%–25% ( ).
Most antidepressants are transferred to breast milk in low concentrations, with few reaching 10% of the maternal weight-adjusted dose ( ). Paroxetine, sertraline, bupropion, duloxetine, fluvoxamine, and mirtazapine have RIDs < 3%, while citalopram, escitalopram, venlafaxine, and desvenlafaxine produce RIDs in breast milk closer to 10%. In contrast, fluoxetine has a RID slightly above the 10% guideline. Several TCAs may be used while breastfeeding because infant exposure has been found to be low and serious side effects in breastfed infants have not been observed, except for doxepin ( ). Nortriptyline is generally the preferred TCA for use with lactation due to its favorable safety profile. MAOIs are generally not used in breastfed infants as insufficient data are available about their use ( ). Limited data are available regarding the safety of serotonin modulators (trazodone, nefazodone, and vilazodone) in breastfed infants.
Nonspecific adverse signs to watch for in breastfed infants exposed to antidepressants include sedation, sleep disturbance, irritability, and poor feeding ( ). Strategies to limit infant exposure such as advising mothers to take antidepressants immediately after breastfeeding or discarding breast milk produced at the time of highest drug concentration are not recommended, as they can potentially make breastfeeding more stressful and are not evidence-based.
Estradiol and progestin as treatments for unipolar postpartum depression
A small double-blind RCT found that women with severe postpartum depression treated with transdermal estradiol showed a greater and faster improvement in symptoms compared to those who received placebo patches ( ). One large RCT was terminated prematurely after the investigators found that an 8-week course of treatment with transdermal estradiol for postpartum depression did not result in significantly increased serum estradiol concentrations ( ). Further evidence from large RCTs is needed before estradiol may be recommended as a therapeutic intervention for postpartum depression due to its associated risks for venous thromboembolism and impaired lactation ( ).
There is less evidence examining the effects of progesterone as treatment for postpartum depression. Some RCTs found that administration of synthetic progestogens, including intramuscular injections of norethisterone enanthate ( ) and depot medroxyprogesterone acetate ( ), was associated with higher risk of developing postpartum depressive symptoms compared with those taking placebo ( ) or with an intrauterine device ( ). In contrast, a retrospective review concluded that use of synthetic progestogens did not appear to predispose postpartum women to depressive symptoms ( ). A Cochrane review reported that synthetic progestogens should not be used to prevent postpartum depression and that they should be used cautiously in postpartum women ( ).
Neuroactive steroids for the treatment of postpartum depression
In 2019, the U.S. Food and Drug Administration (FDA) approved the first medication with an indication for unipolar postpartum depression in adult women. Brexanolone (ZULRESSO™) is a soluble, intravenous preparation of synthetic allopregnanolone, an endogenous neuroactive steroid and is administered as a postpartum 60-h peripheral IV infusion. Brexanolone is a potent, selective, positive allosteric modulator (PAM) of extrasynaptic and synaptic GABA A receptors. A recent series of open-label ( ) and placebo-controlled RCTs ( ; ) of brexanolone demonstrated rapid reduction of postpartum depression symptoms.
Three studies were completed in adult women, ≤ 6 months postpartum, with a diagnosis of peripartum depression with moderate-to-severe symptoms as rated by the 17-item Hamilton Rating Scale for Depression (HAM-D 17 ). Participants were randomized to receive placebo or brexanolone iv 90 μg/kg/h or 60 μg/kg/h for 60 h. The primary outcome measure was the least-squares mean change in HAM-D 17 total score from baseline to hour 60 and safety, tolerability and depression severity were assessed through day 30. Across a range of disease severities, at hour 60, there were significantly larger mean reductions from baseline in HAM-D total scores with brexanolone iv 90 μg/kg/h (− 17.0; P < 0.001) and brexanolone iv 60 μg/kg/h (− 19.1; P < 0.001) vs placebo (− 12.8). Significant differences from placebo were observed at hour 24 (both dose groups P = 0.001) and maintained through day 30 ( P ≤ 0.021) ( ).
In clinical studies, brexanolone caused sedation and somnolence that required dose interruption or reduction in some patients during the infusion (5% of brexanolone-treated patients compared to 0% of placebo-treated patients). Some patients were also reported to have loss of consciousness or altered state of consciousness during the brexanolone injection infusion (4% of the brexanolone-treated patients compared with 0% of the placebo-treated patients). Time to full recovery from loss or altered state of consciousness, after dose interruption, ranged from 15 to 60 min ( ). Brexanolone has a black box warning for excessive sedation and sudden loss of consciousness and is available only through a restricted program under a risk evaluation and mitigation strategy (REMS) to reduce the risk of serious adverse events. According to the package insert, brexanolone appears compatible with breastfeeding as the reported maximum RID during infusion is between 1% and 2% of the maternal weight adjusted dosage.
Zuranolone is an investigational product under development for the treatment of unipolar peripartum depression. Zuranolone is a potent and selective extrasynaptic and synaptic GABA A -R PAM with a pharmacology distinct from benzodiazepines and has oral bioavailability ( ; ). Zuranolone was recently tested in an outpatient randomized, double-blind, placebo-controlled phase 3 study in severe unipolar peripartum depression ( NCT02978326 ) and results were presented at scientific conferences ( ; ). Participants included 151 adult women with severe peripartum depression who received 2 weeks of either zuranolone 30 mg or placebo and then were followed for 4 weeks. Zuranolone achieved the primary endpoint of a significant reduction in LS mean HAM-D 17 total score vs placebo (− 17.8 vs − 13.6, P = 0.0029). Significant differences favoring zuranolone vs placebo were observed at day 3 ( P = 0.0255) and were sustained through day 45 ( P = 0.0027). HAM-D response (72% vs 48%, P = 0.0050) and remission rates (45% vs 23%, P = 0.0122) were significantly greater in the zuranolone group compared to the placebo group at day 15, and these significant improvements were maintained through day 45 (response P = 0.0220; remission P = 0.0102). The most common (≥ 5%) adverse events in the zuranolone group were somnolence, headache, dizziness, upper respiratory tract infection, diarrhea, and sedation. There were no reports of loss of consciousness or syncope in either group. Oral zuranolone is currently being evaluated in an additional randomized, double-blind placebo-controlled trial in severe unipolar postpartum depression ( NCT04442503 ).
Additionally, in development for the treatment of unipolar postpartum depression is the 3β-methylated synthetic analog of allopregnanolone, ganaxolone. Ganaxolone is an extrasynaptic and synaptic GABA A -R PAM, similar to allopregnanolone, but unlike allopregnanolone, in binding studies, ganaxolone is reported to show no affinity for estrogen or progesterone receptors ( ). It has been tested in both IV and oral forms ( NCT03460756 ; NCT03228394 ) but at the time of print, published results were not available.
Neuromodulation
Antenatal electroconvulsive therapy
There are no antenatal electroconvulsive therapy (ECT) clinical trials in the literature. Data are derived from case series and retrospective database reviews, mainly of unipolar peripartum depression with fewer reports for bipolar disorder. ECT has been approved for use in all three trimesters ( ). Recent evidence suggests that ECT in pregnancy is low risk to both the mother and fetus and should be used when clinically warranted. Indications for ECT include severe depression, high suicide risk, catatonia, medication-resistant illness, psychotic agitation, and severe physical decline ( ; ). The risks, adverse reactions, length of treatment, and response in pregnancy are like that observed in ECT in a nonpregnant population ( ). Fetal death, major congenital malformations, and PTB appear to occur at rates like the psychiatric population not receiving ECT ( ).
Older studies of ECT in pregnancy should be interpreted cautiously, as the techniques utilized were markedly different from the procedure today ( ). Maternal and fetal adverse events have been variable, and it is difficult to establish a causal relationship. Some adverse events occurred weeks after ECT, and some complications occurred in the context of medical comorbidities ( ). Two seminal reviews of ECT in pregnancy, which include a total of 339 cases, show a 10% rate of maternal and fetal complications, most of which are transient and nonlife threatening. The most frequent complications were vaginal bleeding, uterine contractions, abdominal pain, and fetal arrhythmias. There was one report of fetal death from status epilepticus indirectly attributed to ECT as the patient had comorbid medical conditions ( ).
In a recent review including 169 women with unipolar depression, bipolar depression, and psychotic depression throughout pregnancy, there was a 4.7% stillborn/neonatal mortality rate, and a 4.1% rate of major congenital malformations. The most common adverse obstetric outcome was preterm delivery (11%), while reduced fetal heart rate was the most common adverse event reported during ECT (8.8%) ( ). A retrospective chart review of 33 antenatal women admitted inpatient and treated with ECT showed a complete response to treatment in 84.2% of patients with MDD, and a partial response to treatment in 15.8%. For women with bipolar disorder, 91.7% of the patients had a complete response and 8.3% had a partial response, rates similar to those in nonpregnant depressed women ( ).
An obstetrician should be readily available in cases where there is an existing increased risk for preterm labor or vaginal bleeding. After week 24 of pregnancy, the woman should be positioned in pelvic tilt to decrease the risk of supine hypotension due to uterine pressure on the vena cava. Aspiration precautions should be taken, and succinylcholine can be used to reduce risk of decreased blood supply to the fetus in the case of a prolonged seizure ( ). The most common anesthetics cross the placenta, but are not known to be teratogenic ( ). There is no evidence to support a link between ECT and teratogenicity ( ).
Long-term effects of ECT in pregnancy are not well studied. In many cases the well-being of the mother in need of treatment will outweigh any potential risks ( ). Recent case reports of pregnant women who received ECT for unipolar depression, bipolar depression, and primary psychotic disorder show normal postnatal development measured from 9 months of age to 3.5 years ( ; ; ).
Postpartum ECT
There are no RCTs studying the use of ECT for postpartum women ( ). ECT is considered the treatment of choice for severe and refractory postpartum psychosis and postpartum depression ( ). Case reports and case series have described positive treatment outcomes with women with treatment refractory postpartum illness who received ECT ( ). ECT may be beneficial for women whose illness severity necessitates rapid resolution of symptoms and need not be reserved solely for treatment refractory cases. It should also be considered in cases of catatonia, or when women with postpartum psychosis or postpartum depression refuse to eat ( ). ECT has been proposed as a first-line treatment for postpartum psychosis ( ; ). ECT should also be considered for treatment of postpartum depression with psychotic features, given the relatively longer median duration of illness compared with postpartum mania ( ).
Transcranial magnetic stimulation
Data on transcranial magnetic stimulation (TMS) for peripartum depression is minimal, with few adverse events reported in the existing literature. The first study of antenatal TMS was an open-label pilot study of low frequency, right-sided TMS of the dorsolateral prefrontal cortex (DLPFC) in 10 women with MDD. The mean HAM-D 17 total score decreased by 60%, with 70% achieving response and 30% achieving remission. There were no reports of fetal arrhythmia, uterine contractions, PTB, NICU admissions, prolonged hospital stays or other adverse events ( ). In an open-label trial on high-frequency left-sided TMS of the DLPFC in 30 antenatal depressed women, 41% had a > 50% reduction in the HAM-D 17 score and 21% achieved remission (HAM-D score < 8). There were no reports of complications in the infants ( ). Case reports published to date have also been reassuring. Babies born to mothers who received TMS in the second or third trimester were born full-term and healthy ( ). There is one report of an infant born prematurely at 36 weeks to a mother who received TMS starting at 31 weeks’ gestation as well as venlafaxine. The baby was irritable for 1 week but otherwise healthy ( ). Pelvic tilt should be employed in women at 24 weeks of pregnancy and beyond to prevent vena cava syndrome and decrease the risk of supine hypotension ( ).
Vagus nerve stimulation
Vagus nerve stimulation (VNS) was approved for TRD in 2005 ( ). There is one case report of VNS for antenatal depression. The woman with unipolar depression and medical comorbidities in the report had been receiving VNS for about 3 years prior to her pregnancy and continued to receive treatment at her regular parameters, and pharmacotherapy with citalopram and bupropion. No adverse peripartum events were reported ( ).
Clinical summary for the treatment of unipolar peripartum depression
Peripartum depression is common, yet frequently underdiagnosed and undertreated. Untreated and undertreated unipolar peripartum depression is associated with adverse health outcomes for mothers and their offspring. The management of unipolar depression in the peripartum period follows the same treatment algorithm as that of the general population, along with unique considerations of dosing and the potential risk of medication use during pregnancy and lactation. Current best practice is to conduct an individualized risk–benefit discussion with the patient regarding medication use that considers her illness course and psychosocial factors, and weighs the benefit and potential risk of medication exposure vs the risk, to both mother and child, of exposure to maternal depression.
Current effective psychotherapies and pharmacotherapies for unipolar peripartum depression are available but underutilized. Psychotherapies are recommended as monotherapy for mild unipolar peripartum depression. Antidepressants are effective and indicated for moderate-to-severe or recurrent unipolar peripartum depression. When using pharmacotherapy, it is recommended to select medications based on individual and family history plus consideration of the reproductive safety data, and to use the lowest effective dose and fewest number of agents to achieve euthymia. Most evidence is reassuring regarding the safety of taking antidepressants (SSRIs, SNRIs, TCAs, bupropion, and mirtazapine) during pregnancy and lactation. While there are limited data for MAOIs, they may be considered in women with TRD who have not responded to other agents. Brexanolone is the only FDA-approved medication for postpartum depression and has rapid antidepressant effects. Neuromodulation may be useful in TRD, as well as for women who wish to avoid pharmacotherapy. ECT appears to be relatively safe and effective in pregnancy. Most complications observed are mild and limited, but more research is needed on this topic. There are little data safety and efficacy on TMS and VNS in pregnancy. These modalities may present a safe and efficacious treatment alternative and thus warrant future research.
Peripartum bipolar depression
Introduction
Women with bipolar disorder are at risk for symptom exacerbation across reproductive events including prior to menstruation ( ), during the peripartum period ( ), and during perimenopause ( ). Antenatal bipolar illness is associated with a 66% increase in risk for postpartum episode ( ). A personal history of postpartum depression, postpartum psychosis, or a family history of a first-degree relative with bipolar disorder increases the risk for peripartum illness onset. Women with bipolar disorder are 50% more likely to develop postpartum depression than women with MDD ( ). Women with bipolar I disorder report an approximately 50% risk of a peripartum major affective episode per pregnancy/postpartum period, while women with bipolar II disorder report an approximately 40% risk of an affective episode ( ). Peripartum mood episodes in bipolar disorder tend to be depressive ( ; ). Women with bipolar disorder are also at an increased risk for postpartum psychosis ( ).
Bipolar disorder during pregnancy is associated with adverse outcomes in both treated and untreated women ( ). There is an increased risk of preterm labor, cesarean section, infants who are small for gestational age, gestational hypertension, antepartum hemorrhage, and placenta previa. Infants of women with untreated bipolar disorder may also be at increased risk of microcephaly and neonatal hypoglycemia ( ). Additionally, untreated peripartum mood episodes are associated with intrauterine growth retardation, elevated levels of fetal stress hormones, and adverse neurodevelopmental outcomes ( ). Critically, women with bipolar disorder are at an increased risk for suicide and infanticide ( ). Despite the substantial risks of untreated symptoms, bipolar disorder in the peripartum period is underdiagnosed and undertreated ( ).
When determining appropriate treatment of bipolar disorder in pregnancy, it is important to keep in mind the limitations of the available evidence base. Most available data derive from observational cohort studies or medical record databases. RCTs examining mood stabilizers and second-generation antipsychotics (SGAs) in pregnancy and large prospective studies that follow women over time across the peripartum period are lacking.
Treatment overview of peripartum bipolar depression
Medication discontinuation has been associated with a twofold increase in the risk of recurrence (37% vs 85%), shorter time to first recurrence, and longer duration of recurrence. Symptoms occur most frequently in the first trimester of pregnancy, and depressive or mixed episodes are most likely. Rapid discontinuation further increases this risk. Gradual discontinuation of medication over a period > 14 days, was associated with a 20-week delay in reaching a 50% recurrence rate as compared to abrupt discontinuations of < 14 days ( ).
The benefits of pharmacologic treatment of bipolar disorder in pregnancy and lactation often outweigh risks. Preconception counseling should be offered at least 3 months prior to considering pregnancy or immediately for women who have become pregnant ( ). The general guidelines used for the acute treatment of bipolar depression in the general population should be applied to pregnant women, with special attention to safety and dosing considerations ( ). The medication chosen for the acute treatment should ideally also be effective in the prevention of postpartum mood episodes of both polarities ( ). A single medication at a higher dose is preferable to polypharmacy. Medication switches should be minimized during pregnancy to limit exposure of the fetus to multiple medications. The medication selection process should be guided by a history of efficacy, prior exposure during pregnancy, and available reproductive safety. If a euthymic woman presents for preconception planning, a baseline therapeutic drug level should be obtained to use as a benchmark for TDM during pregnancy. The preconception period may also be used to establish the lowest effective dose of medication, in order to minimize potential exposure to the fetus while avoiding increasing the mother’s risk of relapse. A comprehensive relapse prevention plan that includes prophylactic medication, minimization of sleep deprivation, and provision of psychosocial support should be established.
Pharmacotherapy for peripartum bipolar depression
Lithium
Lithium can be safely used in pregnancy with close TDM. Abrupt discontinuation of lithium in pregnancy has been associated with high relapse rates ( ). Women with mild and infrequent mood episodes may taper lithium prior to conception in lieu of an alternative agent. Women with a history of more severe episodes, who are not at high risk for relapse in the short term, should taper lithium prior to conception, and restart after organogenesis. Women with severe or frequent episodes of illness should continue lithium throughout pregnancy. Some women taking lithium and another agent may rely on monotherapy with the other agent for the first trimester, and restart lithium in the second trimester to minimize possible teratogenicity from lithium. When lithium is continued, it should be maintained at the lowest therapeutic range throughout pregnancy.
Dosing and therapeutic drug monitoring
The pharmacodynamics of lithium are influenced by weight, renal function, age, co-administered medications, pregnancy, and lactation ( ). Renal lithium clearance increases by 30%–50% in pregnancy, causing a substantial dip in serum concentrations, and an increased risk of maternal relapse if the dose is not adjusted ( ). Lithium levels may drop 24% in the first trimester, 36% in the second trimester, and 21% in the third trimester ( ). Pregnancy-related complications can also alter maternal lithium concentration. Fluid loss due to hyperemesis gravidarum, or normal labor and delivery, can decrease intravascular volume and increase lithium levels. Preeclampsia may reduce renal clearance and increase lithium levels. Lithium dose must often therefore be increased to maintain effective concentrations. It is critical to establish the effective serum lithium concentration prior to pregnancy. When this is not available, the dose should be titrated to the lower end of the therapeutic range and monitored ( ).
Lithium level should be monitored monthly or minimally at every trimester in a euthymic woman on a stable maintenance dose ( ). Women who initiate or reinitiate lithium in pregnancy, or women with medical comorbidities affecting lithium absorption or clearance, may require weekly or bi-weekly lithium levels in the last month of pregnancy ( ). A sustained release formulation may be preferable as it may produce more stable lithium levels ( ). A level should be checked when the woman presents for delivery. Adequate hydration should be ensured, and nephrotoxins should be avoided ( ; ). Vascular volume rapidly decreases at delivery, and lithium clearance abruptly decreases to prepregnancy levels ( ). A level should be drawn immediately after delivery and for any clinical worsening or change in psychiatric or medical status. Preconception dosing can be resumed immediately following delivery once the woman is medically stable ( ). If lithium was initiated in pregnancy, levels should be checked at the end of the first week postpartum, and repeated weekly to achieve a therapeutic level while preventing toxicity ( ).
Teratogenicity
Data regarding the risk of Ebstein’s anomaly and overall congenital heart defects are limited and conflicting ( ). Older retrospective studies and case reports suggested that first trimester lithium use was associated with a rate of Ebstein’s anomaly of 1:1000, as compared to a background risk of 1:20,000 ( ; ). The largest study of in utero lithium exposure to date concluded that the baseline risk of cardiac malformations without lithium exposure was 1.15%, derived from studying 1,322,955 nonexposed infants. Whereas the rate of cardiac malformations in infants exposed to lithium in utero was 2.41%, RR 1.65. The AR of cardiac malformations is 1.90% with lithium exposure. This effect is more modest than in older studies and was found to be dose related. The risk of RVOTO defects was 0.60 per 100 live births among infants exposed to lithium and 0.18 per 100 among unexposed infants ( ).
In a recent systematic review and meta-analysis analyzing over 1.3 million pregnancies, the risk of adverse outcomes associated with lithium exposure at any time in pregnancy was low but increased with first-trimester exposure or exposure to higher doses. Mothers with serum levels of 0.64 mEq/L and doses of 600 mg/day had reactive newborns without increased risk of cardiac malformations. Lithium exposure in pregnancy was associated with any congenital anomaly (OR = 1.81, NNH = 33) and cardiac anomalies (OR = 1.86, NNH = 71). The risk of cardiac anomalies was only significantly increased with first-trimester exposure ( ). Women with first-trimester lithium exposure should receive fetal echocardiography and level II ultrasound at 16–18 weeks’ gestation ( ).
Adverse obstetrical outcomes
An increased risk of PTB of infants has been reported among women treated with lithium in pregnancy ( ). However, this is confounded by the twofold increased risk of preterm delivery in women with bipolar disorder compared to women without a history of mental illness ( ). A large cohort study of 10,575 women on lithium monotherapy or anticonvulsant monotherapy did not find an association with use of these drugs and PTB or placenta-mediated complications when adjusted for confounders. Moreover, women who remained on the medications throughout pregnancy were not at increased risk of ischemic placental disease or PTB compared to women who used mood stabilizers exclusively in the first half of pregnancy ( ).
Postpartum neonatal adverse effects
Fetal plasma lithium concentrations are almost equal to maternal plasma concentrations due to equilibration of ions across the placenta. Toxicity can occur in neonates exposed to high maternal blood levels with higher concentrations at delivery associated with more initial complications ( ). Predominant symptoms include lethargy, hypotonicity, and poor suck reflexes, and can persist beyond 7 days ( ). Neonatal nephrogenic diabetes insipidus ( ), cardiac arrhythmias ( ), floppy baby syndrome ( ), hypoglycemia, polyhydramnios, and reversible changes in thyroid function ( ) have also been reported. Women treated with lithium in pregnancy should thus be considered obstetrical high-risk patients and should receive care from a maternal-fetal medicine specialist.
Neurodevelopmental outcomes
There have been no adverse neurodevelopment outcomes identified in the limited data available. Two small retrospective cohort studies showed no adverse effects on cognitive development at ages 5 and 15 after exposure to lithium in utero ( ; ).
Use of anticonvulsants in peripartum bipolar disorder
Most of the reproductive safety data on the use of anticonvulsants as mood stabilizers in pregnancy and postpartum is derived from epilepsy registry data ( ). This presents potential confounders, and more studies of women taking mood stabilizers for bipolar disorder are needed.
Valproic acid
Valproic acid (VPA) should be avoided in all women of childbearing age who are not surgically sterile due to valproate’s association with serious neural tube defects (NTDs), reduced IQ, autism, and neurodevelopmental delays in children ( ). When clinically appropriate, VPA should be discontinued at least 6 months prior to planning for conception to allow time for taper, initiation of an alternate agent, and adequate monitoring for euthymia. However, if the decision is made that VPA is the only appropriate medication for peripartum mood stabilization, the risks/benefits/alternatives to treatment should be discussed thoroughly with the mother and her partner and documented.
For the very rare instances when VPA is continued during pregnancy, a decrease in VPA level has been observed towards the end of pregnancy due to increased VPA clearance ( ). Several authors have noted that while total VPA level declines, free VPA level remains unchanged or increased. Free and total plasma levels of VPA should thus be measured ( ). Baseline VPA levels should be obtained preconception to identify optimal serum concentration for mood stabilization. Levels should be checked at least monthly to maintain this concentration ( ). VPA concentration decreases sharply in the immediate postpartum period. The dose should be tapered rapidly to avoid toxicity and to maintain preconception VPA levels ( ).
Teratogenicity
VPA has been shown consistently to have the highest rate of congenital malformations (8.2%–13.2%) among the antiepileptic drugs (AEDs) ( ). Most malformations are structural abnormalities, particularly NTDs, with a 1%–3.8% risk with antenatal VPA use ( ; ). Other abnormalities associated with antenatal VPA exposure include cardiovascular ( ), craniofacial ( ), and limb ( ) defects. A conglomerate of abnormalities known as “fetal valproate syndrome,” characterized by distinct facial dysmorphism, congenital anomalies, and developmental delays, has been reported ( ).
Adverse neonatal outcomes
Acute neonatal risks include hepatotoxicity ( ), coagulopathies ( ), and neonatal hypoglycemia ( ). Withdrawal symptoms have also been reported ( ).
Neurodevelopmental outcomes
Prenatal VPA exposure may also negatively impact neurocognitive development ( ; ). In a network meta-analysis of 933 children exposed to AEDs, VPA was the only AED with a significant association with cognitive delay including an increased odds of language delay. It was also associated with greater odds of developing autism ( ). These findings are consistent with a smaller meta-analysis in which in utero VPA exposure was associated with significant reductions in mean full scale IQ, verbal IQ and performance IQ ( ).
Lamotrigine
Continuation of lamotrigine in pregnancy is associated with a decreased risk of both manic, hypomanic, and depressive episodes. A comparison among pregnant women with bipolar disorder who remained on lamotrigine vs women who discontinued showed a 30% relapse rate with continued lamotrigine use vs 100% relapse rate with discontinuation ( ).
Lamotrigine undergoes hepatic glucuronic acid conjugation to an inactive metabolite. Pregnancy-related sex steroid changes increase phase 2 glucuronidation, which increases lamotrigine clearance ( ). Pregnancy related changes in plasma protein binding, absorption, and transplacental transfer can also affect lamotrigine level. Lamotrigine clearance increases substantially at pregnancy onset and continues to increase progressively through the third trimester, potentially to greater than 330% from preconception ( ). Women with epilepsy, require an average dose increase of 250% to sustain therapeutic drug levels across pregnancy. A preconception level should serve as a guide for adjusting the dose throughout pregnancy when available ( ). Lamotrigine elimination rate rapidly declines within days of delivery, and plasma concentrations increase ( ). Women requiring a higher dose in pregnancy should taper their dose to the prepregnancy dose postpartum to avoid toxicity ( ).
Teratogenicity
Lamotrigine has a favorable reproductive profile and has not been associated with an increased risk of congenital malformations. While early studies signaled a possible increased risk of cleft palate or cleft lip deformity with first trimester lamotrigine exposure ( ), more recent studies have not found this risk, or demonstrate only mild elevation above baseline risk ( ; ; ). The risk of major congenital malformations (MCM) with lamotrigine monotherapy exposure in women with epilepsy (2.0%–2.9%) appears to be consistent with the rate major congenital malformations found in the general population (3%–5%) ( ; ; ). Similarly, a meta-analysis of 21 studies did not identify an association between lamotrigine exposure and increased risk of MCM ( ). There may be a potential dose-related difference in risk of cardiac malformations with utero lamotrigine exposure when dosed < 300 mg, vs > 300 mg ( ). However, this finding is inconsistent across studies ( ).
Adverse obstetrical outcomes
Antenatal lamotrigine exposure has not been associated with pregnancy complications or adverse infant outcomes. A systematic review and meta-analysis of 21 studies found no increased risk of miscarriages, stillbirths, preterm deliveries, and small for gestational age neonates after prenatal lamotrigine exposure ( ).
Neurodevelopmental outcomes
Antenatal lamotrigine exposure has not been associated with reduced IQ, or verbal, nonverbal, or spatial abilities, or increased need for educational intervention ( ; ; ).
Carbamazepine
Carbamazepine (CBZ) is converted hepatically to an active metabolite, which then undergoes glucuronidation, conjugation, and hydroxylation to inactive compounds. Greater CYP3A4 activity in women due to progesterone effects may accelerate metabolism and reduce plasma levels of CBZ ( ). However, the data on CBZ metabolism in pregnancy are conflicting. Some studies report declining total concentrations of CBZ during the second and third trimester ( ; ), but others have not found a significant change in CBZ plasma clearance ( ; ). A rapid taper to the prepregnancy CBZ dose following delivery is advised ( ).
Teratogenicity
Antenatal CBZ exposure may increase the rate of congenital anomalies. A meta-analysis of 1255 exposures found that CBZ during pregnancy increased the rate of congenital anomalies, most commonly NTDs, cleft palate, as well as cardiovascular and urinary tract anomalies. CBZ was more teratogenic in combination with other drugs than when used as monotherapy ( ). Antenatal CBZ exposure has also been associated with a fetal CBZ syndrome manifested by facial dysmorphia and fingernail hypoplasia ( ).
Adverse neonatal outcomes
CBZ has been associated with neonatal transient hepatic toxicity and hyperbilirubinemia ( ). Antenatal CBZ exposure has been linked to reduced fetal head growth, restricted uterine growth, and LBW in some studies ( ; ). Other studies have not found these outcomes to differ significantly from controls ( ).
Neurodevelopmental outcomes
It is unclear whether fetal CBZ exposure increases the risk of developmental delay ( ; ). In one study of children exposed to CBZ in utero, there was no effect on overall IQ or nonverbal and spatial abilities. However, IQ sub-score was lower in verbal ability compared to controls. This effect was not dose related. There was no association between CBZ exposure and increased educational intervention ( ). A meta-analysis found that CBZ in pregnancy was not associated with reduced full-scale IQ and verbal IQ but was associated with lower performance IQ ( ). Other studies report similar findings ( ; ).
Other anticonvulsants
There are limited data on the relative safety of other anticonvulsants used as mood stabilizers, such as oxcarbazepine and topiramate. Since they are not first line agents in bipolar disorder, they should be avoided in pregnancy.
Use of antipsychotics in peripartum bipolar disorder
Antipsychotics undergo hepatic metabolism via the CYP450 system. CYP3A4 and CYP2D6 are induced in pregnancy. Quetiapine is primarily metabolized by CYP3A4, while aripiprazole is primarily metabolized by CYP2D6 into an active metabolite, which is metabolized by CYP3A4 ( ). A reduction in serum quetiapine and aripiprazole levels of > 50% may occur in pregnancy ( ). Perphenazine (CYP2D6), ziprasidone (CYP3A4), haloperidol and risperidone (CYP2D6 and CY3A4) serum levels could also decline in pregnancy, even with intramuscular administration. In contrast, olanzapine and clozapine are primarily metabolized via CYP1A2, which has decreased activity during the second and third trimesters. Little or no change in levels of these medications has been observed. Serum antipsychotic drug concentrations return to prepregnancy levels within the first 4 weeks postpartum ( ).
A study of placental passage of antipsychotics found that olanzapine, haloperidol, risperidone and quetiapine demonstrate incomplete placental passage. Of these antipsychotics, olanzapine had the highest placental passage (mean 72.1%), followed by haloperidol (mean 65.5%), risperidone (mean 49.2%), then quetiapine (mean 23.8%) ( ).
Teratogenicity
There are considerable data on the reproductive safety of first-generation antipsychotics (FGAs). No clear increased risk of congenital malformations with the use of FGAs in pregnancy has been identified ( ; ). High-potency FGAs are preferred over low-potency, as some data suggest a possible increased risk of congenital malformations with prenatal exposure to low-potency antipsychotics ( ). Data support the use of haloperidol in pregnancy, including a prospective study of 188 women exposed to haloperidol who showed no increased rate of congenital anomalies compared to the control group ( ). Large studies have not detected a difference in the rate of malformations in infants exposed to FGAs vs SGAs ( ; ).
Despite the widespread availability and use of SGAs, reliable reproductive safety data for these agents are sparse. Safety data are derived from case studies, manufacturer reports, and more recently a select number of larger cohort studies ( ). While studies are lacking, the available data largely support the use of SGAs and suggests that the teratogenic risk is not greatly raised above baseline ( ; ; ). More reproductive safety data are needed, especially for newer drugs like lurasidone, iloperidone, and brexpiprazole ( ). Risperidone may be a possible exception and may confer a small increased risk of congenital malformations ( ; ).
The National Pregnancy Registry for Atypical Antipsychotics, in an ongoing prospective cohort study, has not found a difference in rate of congenital malformations in infants exposed to SGAs compared to infants of women with psychiatric illness treated with other psychotropic medication ( ). However, some studies have suggested an increased risk of congenital malformations with SGA exposure. Among 130 infants exposed to FGAs and SGAs, there was a 6% risk of congenital malformations, particularly cardiac, compared to the 3.1% baseline rate in Australia ( ). A large prospective cohort study found that the odds of a major malformation, particularly cardiac, in women exposed to SGAs was 5.1%, nearly a twofold increase compared to unexposed controls ( ). However, given the varied methodology among the studies and the potential for detection bias when reporting cardiac anomalies, these results should be interpreted cautiously. Additionally, these studies used healthy comparison groups, introducing the significant confounding variable of psychiatric illness. These studies do not distinguish between agents, doses, or diagnosis of the pregnant women using the drugs.
Adverse obstetrical and neonatal outcomes
Both FGAs and SGAs in pregnancy have been associated with LBW ( ; ). A study in Sweden found an excess of PTB and LBW in 570 infants exposed to SGAs and FGAs in the first trimester, but no certain effect on intrauterine growth ( ). Underlying pathology or unidentified confounders may explain the excess risk. The authors concluded that in a risk–benefit analysis, these findings should not prevent necessary treatment to women ( ).
A large-scale systematic review and meta-analysis found a significant association between antenatal antipsychotic exposure and increased risk of multiple adverse obstetric and neonatal outcomes, including neonates small for gestational age (OR 2.44) and preterm delivery (OR 1.86) ( ). However, most studies included had limited adjustment for potential confounding variables, and no study adjusted for type or severity of underlying mental illness. The authors concluded that women who take antenatal antipsychotics are at higher risk for adverse obstetric and neonatal outcomes, but they could not conclude that antipsychotics cause this increased morbidity in pregnancy ( ). Antenatal use of SGAs has also been associated with delivering large for gestational age infants ( ). Pregnant women on SGAs should be closely monitored and screened for gestational diabetes mellitus ( ). It is not clear what other potential neonatal complications of SGAs may be if any. FGAs have been associated with transient extrapyramidal side effects in the neonate ( ).
Neurodevelopmental outcomes
There are no studies that examine neurodevelopmental effects of antipsychotic use in pregnant women with mood disorders. Available data come from observational studies of women with schizophrenia. One case-controlled prospective study suggests that FGAs may cause short-term developmental delays in cognitive, motor, social-emotional, and adaptive behavior, but not in language, bodyweight, or height. Differences were identified in offspring at 2 and 6 months of age, but not at 12 months. The authors note that the women in the exposed group were independently at greater risk for adverse outcomes ( ). Case reports are consistent with the above study ( ; ; ).
Postpartum psychosis
Postpartum psychosis (PPP), also termed puerperal psychosis, is a psychiatric emergency that occurs in 1–2/1000 child-bearing women, usually within the first 2 weeks postpartum ( ). PPP is generally considered to be a bipolar spectrum illness, as clinical presentation, illness course, and family history tend to align with bipolar disorder ( ; ). Initial symptoms are commonly insomnia, mood fluctuations, and obsessive concerns regarding the newborn. These may be followed by more severe symptoms including delusions, hallucinations, and disorganized behavior. Primary mood disturbance may be mania, depression, or a mixed state ( ). Untreated mothers with PPP have an estimated 4% rate of infanticide ( ).
Lithium is effective in the prevention of PPP ( ; ). Based on a study of women with first-onset psychosis or mania in the postpartum period, Berjink et al. proposed a four-step treatment algorithm resulting in high remission rates in patients with first-episode PPP. The treatment involves the subsequent addition of benzodiazepines, antipsychotics, and lithium at set time points in the treatment course. Women who do not respond to pharmacotherapy proceed to ECT. The study included women with bipolar mania with psychotic features, bipolar mixed episodes, unipolar depression with psychotic features, and psychosis without prominent affective symptoms. Using this sequential treatment, 98.4% of patients achieved complete remission within the first three steps, and 79.9% of patients were in sustained remission at 9 months postpartum. Patients treated with lithium had a significantly lower rate of relapse compared with those treated with antipsychotic monotherapy ( ).
Pharmacotherapy use during lactation
Breastfeeding poses a challenge in the treatment of women with bipolar disorder. All psychotropic medications studied to date can enter human milk ( ; ). When considering relative safety in lactation, the degree of exposure and the implications for infant health outcomes must be considered. The severity of the maternal disorder and functional impairment must be weighed against the therapeutic and side effects of medications, and the available scientific data regarding relative safety of psychotropic drugs in breastfeeding. Obtaining sufficient uninterrupted sleep to reduce a woman’s risk of mania or depression should be factored into her decision regarding breastfeeding. Supplementation with formula or donor breastmilk is advisable in many cases.
Lithium
Breastfeeding with lithium use is controversial as safety data are limited ( ). Data suggest that lithium exposure through breast milk may confer less risk than previously considered ( ; ). In most cases, lithium transferred through breast milk is much less than the complete placental transfer of lithium ( ). The concentration of lithium in the serum of breastfed infants ranges from 30% to 40% of maternal levels, with a RID range of 12%–30%, although findings vary greatly between studies ( ; ). Premature or LBW infants are more vulnerable to dehydration and therefore lithium toxicity. In a study of 10 mother–infant dyads, maternal serum, breast milk, and infant serum daily through concentrations of lithium averaged 0.76, 0.35, and 0.16 mEq/L, respectively, each lithium level approximately one-half the preceding level. No serious adverse events were observed. Four infants had minor and transient elevations of thyroid-stimulating hormone, blood urea nitrogen, and creatinine, which were not always associated with high infant serum lithium levels ( ). Newmark et al. conducted a systematic review of lithium in breastfeeding, where adverse infant effects were reported in 9.4%. Abnormal lab values were reported in an additional 9.4% of the infants. Complications included cyanosis, lethargy, electrolyte abnormalities, and gross irregular twitching. Among the adverse effects reported, it is difficult to differentiate poor outcomes from other factors affecting infant health, concomitant medications, and gestational lithium exposure ( ). The longer-term significance of lithium exposure and its associated abnormalities is unclear.
The AAP and LactMed, an NIH supported database, state that lithium can be continued during lactation with careful monitoring of the infant, including careful monitoring of the maternal serum concentrations and possibly of infant serum concentrations ( ; ). However, others consider lithium exposure through breast milk as possibly hazardous given the relatively high exposure and the theoretical risks, and recommend formula feeding ( ). Guidance on whether to breastfeed while taking lithium must be personalized to the mother–infant dyad.
Valproic acid
The infant/maternal ratio of serum drug concentration is lower in VPA compared to other mood stabilizers. The RID of VPA is low (0.68%) and is compatible with breastfeeding. Recent studies show serum levels in infants exposed through breastmilk in the range of 0.7%–6.6 μg/mL, corresponding to a mean of 2.1% of the maternal serum level. One adverse event, anemia and thrombocytopenia purpura, was reported in one of nine infants in recent studies ( ).
Carbamazepine
The RID of CBZ (4.3%) is low and compatible with breastfeeding. The ranges for milk/plasma ratio, infant serum levels of CBZ, and infant/maternal ratio of serum CBZ concentrations were 0.0–1.4, 0.0–2.6, and 0.0–0.7, respectively ( ). Due to confounders, it is unclear whether the cases of adverse events were due to CBZ exposure through breastmilk or another cause.
Oxcarbazepine
The literature is limited to two case reports on use of oxcarbazepine during lactation. The RID was 1.5%–1.7%, and infant/maternal ratio of serum oxcarbazepine concentration was < 0.5–1.0. Oxcarbazepine use during lactation was not associated with adverse events in these studies ( ; ).
Lamotrigine
Considerable amounts of lamotrigine are transmitted to the infant through breastmilk ( ; ). Lamotrigine RIDs range from 2.0% to 21.1%. However, lamotrigine use during lactation has rarely been associated with adverse effects and is thus considered low risk ( ; ). In a systematic review of 122 women taking lamotrigine during lactation, adverse events were reported in only 3.27% of infants. Of note, most women in the sample had epilepsy, and fewer women had bipolar disorder ( ). Maternal dose has been shown to be a significant predictor of lamotrigine concentration in breastmilk, whereas time after maternal dose has not ( ).
Atypical antipsychotics
Data suggest that levels of SGAs transmitted in breast milk are low, with RIDs generally < 3% of maternal dose ( ). Olanzapine, risperidone, and quetiapine have the most data available to inform safety in lactation with RIDs < 10%, reflecting acceptable infant exposure ( ). Data are limited for other SGAs such as aripiprazole and lurasidone. Adverse events are mild and infrequent ( ). Breastfeeding can be considered with concurrent infant monitoring. Lactation safety data for FGAs are minimal and derived mainly from case reports and case series. Data for haloperidol and chlorpromazine are largely reassuring, with one case report of drowsiness and lethargy with high-dose chlorpromazine found in breast milk ( ).
Clinical summary for the treatment of peripartum bipolar depression
The peripartum period is a time of increased risk for both first-onset and recurrent mood episodes in women with bipolar disorder. Untreated bipolar illness presents known risks to the fetus, so maintaining euthymia is paramount. Pharmacotherapy plays a key role in the treatment of acute antenatal symptoms, as well as in the prevention of relapse. Safety data on the use of psychotropics in pregnancy and lactation in women with bipolar disorder are sparse. However, the available data, derived from observational studies, are largely positive for lamotrigine and antipsychotics, and the teratogenicity associated with lithium in more recent studies is less than older data suggested. Medication choice should consider the available safety data as well as the mother’s treatment history and desire to breastfeed. Given the limited data available, the known benefits of breastfeeding must be weighed against potential unknown effects of medication, and treatment recommendations may vary based on the unique needs of the mother–infant dyad.
Choosing an agent that is protective for both depression and mania is preferable in order to avoid exposure to polypharmacy. Ideally, the lowest effective dose should be determined during preconception. Pregnancy associated metabolic changes must also be considered, with careful TDM and dose adjustments when warranted. ECT also plays an important role the treatment of peripartum bipolar disorder. It should be used in treatment-resistant bipolar depression and considered for depression with psychotic features, catatonia, and cases which warrant rapid resolution of symptoms. It is also an effective treatment for PPP.
Depression during the transition to menopause
Introduction
The menopausal transition occurs at the end of the female reproductive life cycle: in the perimenopausal period, menstrual irregularity occurs as variability in primarily follicle stimulating hormone (FSH) increases, leading to concurrent variability in estradiol levels ( ). The postmenopausal period begins after 1 year of amenorrhea, which retroactively identifies the final menstrual period (FMP) ( ). FSH and estradiol continue to fluctuate for some time through the postmenopausal period; per the Stages of Reproductive Aging Workshop (STRAW) + 10, an update to the landmark study seeking to establish scientific consensus on the menopausal transition, it may require up to 8 years after the FMP before hormones stabilize ( ; ). Despite a period of scientific debate on the question, longitudinal studies have established that the perimenopausal period is one of increased vulnerability to clinical depression in women ( ; ; ). Women without a prior episode of depression are also at increased risk ( ; ), although prior episodes are both a risk factor for reoccurrence and higher symptom burden ( ). Women who have demonstrated sensitivity to hormone fluctuations in the past through prior diagnoses of peripartum depression or premenstrual dysphoric disorder (PMDD) are also at increased risk of perimenopausal depression ( ). This lends credence to the current theories regarding the pathophysiology of perimenopausal mood changes as relating to sensitivity in fluctuations of neuroactive hormones rather than their absolute values in the steady state ( ). Adverse effects on mood are only one of several symptom clusters observed in the perimenopausal period; others include vasomotor symptoms, cognitive symptoms, sleep disturbance, pain symptoms, tension, urinary incontinence, and changes in libido ( ; ). Comprehensive treatment should target multiple symptom clusters whenever possible.
Treatment approaches to depression during the menopausal transition
Pharmacotherapy
Perimenopausal depression is commonly treated according to standard clinical practice of treating MDD: SSRIs, SNRIs, TCAs, and MAOIs are all used in clinical practice ( ). However, research on the specific efficacy of these agents in the perimenopausal period has been limited. Open-label trials have found favorable outcomes for mirtazapine, escitalopram, citalopram, duloxetine, venlafaxine, and vortioxetine, but there are methodological limitations ( ; ; ).
The most robust evidence available is for desvenlafaxine, and it has been shown to be significantly effective vs placebo on depressive symptoms during the peri- and postmenopausal periods per evidence from double-blind, randomized studies as well as large pooled analyses examining doses ranging from 50 mg per day to 200 mg per day ( ; ). However, a randomized, placebo-controlled comparative study failed to demonstrate superiority of desvenlafaxine over escitalopram ( ). Similarly, venlafaxine extended release did not vary in efficacy from fluoxetine with respect to menopausal status or sex ( ), despite some earlier evidence suggesting diminished efficacy of SSRIs in older and postmenopausal women ( ; ).
A similar clinical controversy concerns demonstration of superior efficacy of SSRIs vs TCAs in premenopausal women ( ; ), findings that bring into focus broader questions about sex, hormones, and efficacy of antidepressants that warrant further study before clinical recommendations can be made.
With regards to intravenous ketamine for TRD, data available to date are analyses from one randomized, placebo-controlled study, which demonstrated no change in efficacy or tolerability related to sex or menopausal status ( ).
Concurrent treatment of vasomotor symptoms
Vasomotor symptoms, including acute onset of fever-like sensation of heat commonly referred to as “hot flashes,” are experienced by most women in the perimenopausal period ( ), and for some women, these symptoms are distressing enough to significantly interfere with quality of life ( ) as well as exacerbate depressive symptoms ( ; ; ). Although there are many hormonal treatments for vasomotor symptoms, research has shown that the antidepressants venlafaxine, desvenlafaxine, escitalopram, citalopram, fluoxetine, and sertraline are also effective in managing these symptoms, although none are FDA approved for that indication ( ; ). The only antidepressant currently FDA approved for treatment of vasomotor symptoms is paroxetine 7.5 mg per day, marketed as Brisdelle ( ).
Gabapentin, pregabalin, and clonidine have also been studied as potential treatments for vasomotor symptoms of perimenopause. A meta-analysis of four studies of clonidine vs placebo found decreases in frequency of vasomotor symptoms ( ). Gabapentin was efficacious compared to placebo in reducing severity and frequency of hot flashes according to a recent systematic review and meta-analysis; however, the same study did not demonstrate conclusive evidence for use of pregabalin largely due to methodological limitations ( ). Neither treatment carries an FDA approval for the indication.
Concurrent treatment of concentration difficulties
In addition to the appropriate identification and treatment of underlying depression in perimenopause, there may be a role for use of psychostimulants in the treatment of concentration and cognition difficulties in perimenopausal women. One double-blind, placebo-controlled crossover trial of atomoxetine for nondepressed perimenopausal women with cognitive symptoms but without a history of ADHD found significant improvement compared to placebo in the total score of the Brown Attention Deficit Disorder Scale as well as multiple subscales ( ).
Concurrent treatment of insomnia
The etiology of insomnia in depressed perimenopausal women is multidimensional, and a thorough review of pharmacotherapy for sleep is beyond the current scope of this chapter. However, sleep difficulties in perimenopausal women are sometimes related to vasomotor symptoms ( ), and treating those symptoms may lead to symptomatic improvement in insomnia.
Hormone replacement therapy
The evidence base for use of hormone replacement therapy (HRT) for perimenopausal depression is fraught with scientific controversy and complexity. Low-dose testosterone augmentation has been hypothesized to be a potential treatment for TRD in women, but a placebo-controlled study including both premenopausal and postmenopausal subjects failed to demonstrate efficacy ( ). The hypothesis that prophylactic HRT could be used to prevent the onset of depression throughout menopause has largely not been supported by the evidence ( ), however, one recent randomized, placebo-controlled trial of estrogen patch and oral progesterone found that this combination was significantly more effective than placebo in preventing development of clinically significant depressive symptoms during perimenopause in previously euthymic women ( ). The same study aligned with prior evidence that depression in the postmenopausal period has not been shown to be responsive to HRT ( ; ). While there is some contrary evidence that depression in the perimenopausal period may be successfully treated with HRT, there are significant risks associated with the initiation of HRT in older women, notably the increased risk of venous thromboembolism ( ). Any trial of estrogen therapy for perimenopausal depression that has failed treatment with antidepressants should occur within a collaborative effort between a gynecologist and treating psychiatrist.
Clinical summary for the treatment of depression during the transition to menopause
The depression risk is increased in the menopausal transition, and its treatment is complicated by exacerbating factors such as vasomotor symptoms, concentration difficulties, and insomnia. Whenever clinically relevant, therapies that address multiple symptom domains should be utilized, such as the antidepressants with demonstrated efficacy for depression as well as vasomotor symptoms: desvenlafaxine, venlafaxine, and escitalopram. Psychostimulants should be considered as they may be effective even for women without a previous diagnosis of ADHD. Refractory cases warrant referral to an obstetrician-gynecologist for the initiation of collaborative care due to the controversial and complex evidence surrounding the use of estrogen replacement therapy in perimenopausal TRD. Finally, although a search yielded no studies of ECT within the specific population of women with perimenopausal depression, ECT has been demonstrated to be a safe and effective treatment option for TRD that should be considered in cases that fail to respond to pharmacotherapy ( ).