Background. Untreated maternal mental health disorders can have devastating sequelae for the mother and child. For women who are currently or planning to become pregnant or are breastfeeding, a critical question is whether the benefits of treating psychiatric illness with pharmacologic interventions outweigh the harms for mother and child.
Methods. We conducted a systematic review to assess the benefits and harms of pharmacological interventions compared with placebo, no treatment, or other pharmacological interventions for pregnant and postpartum women with mental health disorders. We searched four databases and other sources for evidence available from inception to December 11, 2018, dually screened the results, and analyzed eligible studies. We included studies of pregnant, postpartum, or reproductive-age women with a new or preexisting diagnosis of a mental health disorder treated with pharmacotherapy; we excluded psychotherapy. Eligible comparators included women with the disorder but no pharmacotherapy or women who discontinued the pharmacotherapy before pregnancy.
Results. A total of 138 studies (142 articles) met eligibility criteria. Brexanolone probably improves depressive symptoms at 30 days (least square mean difference in the Hamilton Rating Scale for Depression, -2.6; p=0.02; N=209) when compared with placebo. Sertraline may improve response (calculated relative risk [RR], 4.00; 95% CI, 0.98 to 16.31; N=36), remission (calculated RR, 4.22; 95% CI, 0.98 to 18.12; N=36), and depressive symptoms (p-values range from 0.01 to 0.05) when compared with placebo. Discontinuing use of mood stabilizers during pregnancy may increase recurrence (adjusted hazard ratio [AHR], 2.2; 95% CI, 1.2 to 4.2; N=89) and reduce time to recurrence (AHR, 12.1; 95% CI, 1.6 to 91; N=26) for bipolar disorder when compared with continued use. Brexanolone may increase the risk of sedation or somnolence leading to dose interruption or reduction when compared with placebo (5% vs. 0%). Observational studies suggest some associations between specific antidepressants and adverse maternal outcomes such as postpartum hemorrhage, preeclampsia, and spontaneous abortion and child outcomes such as respiratory issues, low Apgar scores, persistent pulmonary hypertension of the newborn, depression in children, and autism spectrum disorder; between quetiapine or olanzapine and gestational diabetes; and between benzodiazepine and neonatal intensive care admissions. We found insufficient evidence on benefits and harms from comparative effectiveness studies, with one exception: one study suggested a higher risk of overall congenital anomalies (adjusted RR [ARR], 1.85; 95% CI, 1.23 to 2.78; N=2,608) and cardiac anomalies (ARR, 2.25; 95% CI, 1.17 to 4.34; N=2,608) for lithium compared with lamotrigine.
Conclusions. Few studies have been conducted in pregnant and postpartum women on the benefits of pharmacotherapy; many studies report on harms but are of low quality. The limited evidence available is consistent with some benefit, and some studies suggest increased adverse events. However, because these studies could not rule out underlying disease severity as the cause of the association, the causal link between the exposure and adverse events is unclear. Patients and clinicians need to make an informed, collaborative decision on treatment choices.
- Few studies have been conducted in pregnant and postpartum women on the benefits of pharmacotherapy; many studies report on harms but are of low quality.
- Brexanolone probably improves depressive symptoms; it may increase the risk of sedation or somnolence, leading to dose interruption or reduction.
- Sertraline may improve response, remission, and depression and anxiety symptoms.
- Mood stabilizers may reduce recurrence and increase time to recurrence.
- Associations may exist between antidepressants
- Lithium is more likely to be associated with overall congenital and cardiac anomalies than lamotrigine.
- The paucity of evidence does not mean that pharmacotherapy is not beneficial, nor that harms do not exist; rather, it underscores the absence of high-quality research.
Background and Purpose
Untreated mental health disorders in perinatal (pregnant and postpartum, including breastfeeding) women can have devasting sequelae. Pregnancy-associated suicide kills more women than hemorrhage or preeclampsia. Depressive symptoms are associated with reduced safety for the child, increased harsh punishment, impaired development of infant emotional regulation and attachment, and greater risk of psychiatric disease in the child. Treatment choices for mental health disorders include pharmacotherapy, psychotherapy, and other approaches (e.g., yoga, mindfulness, self-care, nutritional or herbal supplements). For women who are currently or planning to become pregnant, a critical question is whether the benefits for mother and fetus of treating psychiatric illness with pharmacologic interventions outweigh the harms; a systematic review will help clarify the balance of benefits and harms.
We employed methods consistent with those outlined in the Agency for Healthcare Quality and Research Evidence-based Practice Center Program Methods Guidance, and we describe these in the full report. Our searches covered publication dates from inception to December 11, 2018.
A total of 138 studies (142 articles) met eligibility criteria. Most studies were observational in design and had high risk of bias.
Key Question (KQ) 1: Benefits of Pharmacological Treatments Versus No Treatment or Placebo for Pregnant and Postpartum Women With Anxiety, Depression, Bipolar Disorder, or Schizophrenia. Substantial evidence exists on the effectiveness of medications in the general population, but evidence in pregnant and postpartum women is sparse (9 trials and 6 observational studies). When evidence was available, we found low to moderate strength of evidence of benefit. Specifically, for depression, we found moderate strength of evidence that brexanolone is associated with improved depressive symptoms shortly after infusion (60 hours) and at 30 days after treatment; low strength of evidence from two trials that sertraline achieves response, remission, and improvements in depressive symptoms. For bipolar disorders, we found low strength of evidence from two cohort studies that mood stabilizers prevent recurrence and increase time to recurrence.
KQ 2: Comparative Benefits of Pharmacological Treatments for Pregnant and Postpartum Women With Anxiety, Depression, Bipolar Disorder, or Schizophrenia. For depression and bipolar disorder, we found insufficient evidence to judge the comparative effectiveness of a very limited number of outcomes and interventions from 1 trial and 3 observational studies. For anxiety and schizophrenia, we found no evidence on comparative effectiveness for anxiety or schizophrenia.
KQ 3: Harms of Pharmacological Treatments Versus No Treatment or Placebo for Pregnant and Postpartum Women With Mental Health Disorders. We found 5 trials and 56 observational studies. We found low strength of evidence for several outcomes. Specifically, for maternal harms, several antidepressants may be associated with a higher risk of postpartum hemorrhage; SNRIs and TCAs may be associated with an increased risk of preeclampsia; SNRIs may be associated with spontaneous abortion; quetiapine and olanzapine may be associated with an increased risk of gestational diabetes; and brexanolone may be associated with the risk of sedation or somnolence leading to dose interruption or reduction. For child harms, we found an association between benzodiazepine and neonatal intensive care unit (NICU) admissions; between selective serotonin reuptake inhibitors and respiratory issues, low Apgar scores, persistent pulmonary hypertension of the newborn, and depression in children; and between citalopram and autism spectrum disorder. Signals of harms that we identified above may be partially or wholly attributable to residual confounding. Importantly, we found insufficient evidence on congenital anomalies and cardiac defects.
KQ 4: Comparative Harms of Pharmacological Treatments for Pregnant and Postpartum Women With Mental Health Disorders. We found 1 trial and 44 observational studies. Evidence from one study suggests that the association between exposure to lithium and overall congenital anomalies and cardiac anomalies may be greater than the association between lamotrigine and the same outcomes (low strength of evidence). The evidence is insufficient for all other comparisons and outcomes.
We identified few randomized controlled trials of pharmacotherapy for mental health disorders during pregnancy or lactation; therefore we relied on observational studies for the bulk of this review. A significant constraint to interpreting the evidence is the widespread risk of confounding. Underlying mental health disorders result in the use of psychotropic medications. They may also result in some of the harms investigated in this review regardless of exposure to medications. Studies varied greatly in the extent to which they were able to address underlying severity of mental health disorders. The majority were unable to address confounding, often because of a lack of the necessary variables in registry datasets. A small subset of studies attempted various measures to address confounding (e.g., propensity score adjustment, stratification by number of disorders). In many instances, controls for confounding reduced the effect size and, in some instances, reversed the direction of effect. Most studies were unable to identify dose and duration of exposure. For the benefits question (KQ 1), eligible studies had comparator arms of women with the same disorder as in the treatment arm. For the harms question (KQ 3), however, we were more inclusive and included studies with comparator arms comprising women with prior exposure to the drug, even if the disorder status was not specified. As a result, our KQ 1 evidence base controls for underlying severity as a confounder better than the KQ 3 evidence base.
We elected to restrict the evidence to women with mental health disorders as a means of reducing the potential for confounding in the evidence base. However, this criterion excluded studies of well-conducted negative controls that might bolster the evidence on the association between the exposure and the outcome. Also, this criterion resulted in the exclusion of studies reporting on relevant outcomes for exposures to the intervention for other clinical conditions. Studies of multiple drug exposures presented results for each exposure but did not always present results separately for the women with multiple drug exposures. In these studies with overlapping arms, we were not able to attribute the effect of the intervention to a single drug. As a result, we excluded these studies. The exclusion of studies with overlapping arms also restricted the comprehensiveness of our review. These limitations of the evidence and of our review criteria mean that the signals of harms that we identified above may be partially or wholly attributable to residual confounding. We may also have missed eligible studies because of our restriction to English language studies.
Implications and Conclusions
The central decisional dilemma facing pregnant and postpartum women with mental health disorders and their healthcare providers is how to balance benefits and harms of psychotropic drugs for both themselves and their children. One such critical trade-off is whether improved symptoms in the mother outweigh the harms from potential congenital anomalies in the fetus. Given long-standing restrictions on including pregnant and lactating women in clinical trials,1 few clinical trials have evaluated the effectiveness of pharmacotherapy. By contrast, evidence is voluminous but of low quality on the harms of pharmacotherapy. Our findings indicate the need for clear communication to patients on four primary points: (1) evidence exists that medication works in the general populations, (2) few studies have measured effectiveness in pregnant women, (3) the limited evidence available is consistent with some benefit, and (4) some studies do suggest some increased adverse events, many of which are rare or transient. However, because these studies could not rule out underlying disease severity as the cause of the association, the causal link between the exposure and adverse events is unclear. The patient and her provider are uniquely qualified to determine whether the mother’s medical need for treatment exceeds any potential harms.
- Task Force on Research Specific to Pregnant Women and Lactating Women. Report to Secretary, Health and Human Services, Congress. Rockville, MD: National Institutes of Health; 2018. Accessed on May 20, 2020.