Key Informants and the Technical Expert Panel gave input on the KQs, definition of the perinatal period, and list of prioritized outcomes.

KQ 1: What are the effectiveness and comparative effectiveness and harms of nonpharmacologic treatments for mental health conditions in perinatal individuals?

1. Depressive disorders
2. Anxiety disorders
3. Bipolar disorder
4. Post-traumatic stress disorder
5. Obsessive-compulsive disorder

KQ 2: What are the comparative effectiveness and harms of nonpharmacologic treatments compared with pharmacologic treatment alone for mental health conditions in perinatal individuals?

1. Depressive disorders
2. Anxiety disorders
3. Bipolar disorder
4. Post-traumatic stress disorder
5. Obsessive-compulsive disorder

Findings in Relation to What is Already Known

Nonpharmacologic Treatments for Depressive Disorders

CBT, IPT, exercise and behavioral activation each effectively treated depressive symptoms for individuals with depressive disorders when compared with TAU. Similarly, CBT and IPT reduced anxiety symptoms among individuals with depressive disorders compared with TAU. CBT and IPT were also associated with increased rates of remission of depressive symptoms for individuals with depressive disorders compared with TAU. Our findings regarding the effectiveness of CBT compared with TAU for treating psychological symptoms of individuals with perinatal depressive disorders are consistent with recent systematic reviews, which concluded that CBT was more effective in treating depressive and anxiety symptoms among individuals with depressive disorders. However, one of the systematic reviews classified behavioral activation and problem-solving therapies as CBT interventions, which may affect the interpretation of their findings in regard to specific interventions. The current review builds on existing reviews in this area by establishing that effect sizes do not differ between studies using diagnostic or screening tools and between the mode of delivery or setting of CBT interventions, and providing greater detail about the evidence and gaps for multiple mental health conditions. Our findings regarding the effectiveness of IPT interventions for perinatal individuals with depressive disorders, confirm and provide additional evidence to an existing review on this topic. A previous systematic review concluded that IPT was likely associated with reduced depressive and anxiety symptoms at the end of treatment; however, this review did not conduct meta-analyses. Our meta-analysis of studies comparing IPT with TAU, lend empirical support to prior findings and also demonstrate the effect sizes do not differ by the use of diagnostic or screening tools, or by the mode of delivery; indicating IPT can be delivered using multiple modalities and maintain effectiveness. A previous systematic review of exercise interventions for perinatal individuals with depressive disorders concluded that exercise effectively reduced depressive symptoms compared with TAU. Our findings are consistent with the prior review's findings. Our review also implemented stricter inclusion criteria for defining a diagnosis of depression. Behavioral activation is a common CBT technique, which has been found to be an effective treatment for depression in its own right. However, the effectiveness of behavioral activation interventions for perinatal depressive disorders had yet to be explored. Our review found behavioral activation effectively reduced depressive symptoms when compared with TAU. Compared with CBT, behavioral activation could potentially be a more parsimonious, transportable, and cost-effective treatment for perinatal depressive disorders. This review found remission rates of depressive symptoms were higher among those who received acupuncture compared with non-specific or sham comparison groups. Acupuncture aims to restore the flow of energy through the body by inserting and stimulating needles. A narrative review has suggested that acupuncture is an effective treatment for perinatal depressive disorders. The findings of this review support these claims; however the strength of evidence was rated low. Bright-light therapy is another complementary therapy that has been tested for perinatal mental health disorders, however we identified insufficient evidence to draw conclusions due to a lack of consistency and precision among included. Larger, more rigorous trials testing complementary therapies are needed.

Nonpharmacologic Treatments for Anxiety Disorders

We identified two RCTs testing nonpharmacologic interventions for anxiety disorders. One study tested a multicomponent (problem-solving therapy, self-care, and psychoeducation) intervention compared to TAU for individuals with anxiety disorders and found no significant differences in depressive symptoms, quality of life or generalized anxiety symptoms (as measured by the Generalized Anxiety Disorders Scale [GAD-7]) between the groups. However, they did find a small, statistically significant difference in pregnancy-related anxiety scores at the end of treatment in favor of the multicomponent intervention. Similarly, another study testing CBT compared to TAU found no differences in depressive or anxiety symptoms between the groups. Due to the heterogeneity in interventions, we were unable to draw conclusions. Previous reviews have also highlighted the paucity of RCTs in this area. Given that one study found a significant effect of a multicomponent intervention in reducing anxiety on a scale developed for pregnant populations but did not find a significant effect on a scale developed for the general population, it possible that anxiety scales created for the general population are not sensitive enough or validated for use in the perinatal population.

Nonpharmacologic Treatments for Depressive and/or Anxiety Disorders

We identified four studies comparing the effectiveness of nonpharmacologic treatments for perinatal individuals with combined depressive and anxiety disorders. Three studies compared CBT with TAU. Previous reviews have grouped studies of participants with both depressive and anxiety disorders and studies of participants with depressive or anxiety disorders together, without delineating between these distinct patient groups. Three studies compared CBT with TAU for individuals with depressive and anxiety disorders. CBT was associated with reduced anxiety symptoms, but not reduced depressive symptoms, at the end of treatment compared with TAU. These findings provide limited evidence that CBT could be an effective transdiagnostic (i.e., addressing both depressive and anxiety symptoms) intervention for perinatal depressive and anxiety disorders. One study compared exercise with TAU for perinatal depressive and anxiety disorders, however this evidence was not insufficient for drawing evidence-based conclusions, Three studies compared CBT with TAU for individuals with depressive or anxiety disorders, however there was insufficient evidence available to draw conclusions.

Nonpharmacologic Treatments for PTSD

A previous systematic review included two studies testing nonpharmacologic treatments for perinatal individuals with PTSD, however, these studies were analyzed alongside interventions for individuals with other anxiety and stress-related conditions and the authors did not delineate outcomes by mental health conditions. We identified a similar paucity of studies testing nonpharmacologic treatments for PTSD, with insufficient evidence available for drawing conclusions regarding the effectiveness of nonpharmacologic treatments for PTSD.

Nonpharmacologic Treatments for OCD

We identified one study testing the effectiveness of CBT with TAU for perinatal individuals with OCD. No previous reviews have been conducted in this area, perhaps reflecting the lack of primary research.

Nonpharmacologic Treatments for Bipolar Disorder

We did not identify any studies assessing the comparative effectiveness of nonpharmacologic treatments for individuals with bipolar disorder, nor have there been any prior reviews on this topic.

Nonpharmacologic Compared with Pharmacologic Treatment for Perinatal Mental Health Conditions

Previous reviews have assessed the comparative effectiveness of pharmacologic treatments for perinatal mental health disorders, however there is limited research comparing pharmacologic with nonpharmacologic treatments. We identified two RCTs comparing the effectiveness of selective serotonin reuptake inhibitors (SSRIs) to nonpharmacologic treatments for depressive disorders. Both studies found that both CBT and SSRIs reduced depressive symptoms at the end of treatment. One study found reported that the reduction in depressive symptoms was greater among individuals who received CBT compared to SSRIs alone and CBT combined with SSRIs. Further research is needed to confirm these findings.

Implications for Clinical Practice and Research

Implications for Clinical and Policy Decision Making

The summary of findings table displays the evidence for the effectiveness of nonpharmacologic treatment for perinatal mental health conditions. These findings may lend support to clinical practice guidelines and consensus statements for the use of nonpharmacologic interventions during the perinatal period. Our findings lend the strongest empirical support to CBT and IPT treatments, however this may primarily reflect the large number of RCTs for these treatments. There is evidence that effectiveness of CBT or IPT may not vary by the mode of delivery (i.e. group, individual, or self-guided) or by the delivery setting (in the patient's home, in the clinic, or remotely), indicating there may be flexibility in terms of how such interventions can be delivered while maintaining effectiveness. CBT was an effective treatment for individuals with depressive disorders and combined anxiety and depressive disorders, but not for individuals with depressive or anxiety disorders. Most interventions in this review that aimed to treat combined depressive and anxiety disorders described how the intervention was adapted to be transdiagnostic (addressing both disorders), however such adaptations were not noted in studies testing CBT for depressive or anxiety disorders. These findings suggest that for interventions to be effective across diagnostic groups they must be specifically adapted to be transdiagnostic.

Given the effectiveness of multiple interventions (CBT, IPT, exercise, behavioral activation, and acupuncture) there is a need for a well-trained work force across multiple specialties, included by not limited to obstetrics and gynecology, psychiatry, nursing, and psychology.

The finding of this review suggest that technology could be used to increase access to nonpharmacological treatments during the perinatal period. For example, many rural and underserved urban areas have limited numbers of providers of mental health care who patients can feasibly access. Treatments that could be delivered via telephone or online may help increase access to mental health treatment, although some barriers including access to high-speed internet may remain.

However, the RCT evidence base does not adequately address the comparison between CBT and IPT or the effect or comparative effectiveness of other nonpharmacologic interventions. Evidence for other nonpharmacologic interventions including behavioral activation, bright-light therapy, non-directive counseling and acupuncture is more limited than the evidence supporting CBT and IPT. Future, rigorous trials comparing nonpharmacologic and pharmacologic interventions for perinatal mental health disorders are needed to inform clinical and policy decision making.

Implications for Research

We identified multiple areas for further research. Firstly, although it is estimated that approximately 20 percent of perinatal individuals meet criteria for anxiety disorder, we identified only a limited number of studies testing the comparative effectiveness of nonpharmacologic interventions for this population. Given the high prevalence of anxiety disorders in the perinatal period, and some limited evidence that CBT or IPT is effective for treating anxiety symptoms in individuals with anxiety and depressive disorders, it is crucial that future research addresses this gap in the evidence.

We identified several studies comparing the effectiveness of nonpharmacologic interventions for perinatal individuals with PTSD. However, the heterogeneity in interventions and inconsistent reporting of outcome measures made it challenging to draw evidence-based conclusions. Further, rigorous research is needed on this topic.

Although they are rarer clinical diagnoses, there are similar gaps in the research for perinatal individuals experiencing bipolar disorder or obsessive-compulsive disorder. Future research is needed to confirm whether nonpharmacologic therapies are safe alternative or adjunctive therapies to pharmacologic treatment for these conditions.

The feasibility and effectiveness of different nonpharmacologic treatments may vary by factors including the perinatal period, loss of the infant or infant death, or breast feeding. Both the stressors affecting mothers and the feasibility of implementing different treatments may be different in different peripartum stages and may require different nonpharmacologic approaches. While there was insufficient evidence available in this review, future research is needed to understand which intervention may work best or not work for various socio-demographic groups during the perinatal period. Furthermore, there are significant barriers to accessing mental health care during the perinatal period, there is a need to understand how to implement and increase access to nonpharmacological treatments in different intersectional groups, including by age, race, ethnicity, gender, sexual orientation, geographical location, and partner status.

The effectiveness of nonpharmacologic interventions may also vary by the component "ingredients". For example, studies may evaluate the effectiveness of CBT but implement different components (e.g. cognitive restructuring or behavioral activation). Conversely, studies may be testing different interventions which share similar components (e.g. keeping a diary or setting goals). Future research is needed to test which interventions components and what combination of components common to nonpharmacological interventions are likely to be most effective for the different perinatal disorders. By identifying which components are most effective, it might be possible to develop brief interventions while maintaining effectiveness.

Studies included in this review addressed a wide range of outcome domains and measures. There is a need to adopt a core set of outcomes and measures, and standardized approaches to reporting outcomes. We are aware that efforts are underway. However, it will be essential that once a core outcome set is developed, researchers implement investigation and reporting of these outcomes. It is important to note that this is a growing area of research and numerous trials testing the effectiveness of nonpharmacologic treatments for perinatal mental health conditions are currently underway. As of April 2, 2024, there are 21 active studies registered on ClincialTrials.gov testing nonpharmacologic treatments for perinatal mental health conditions.

This review does not provide cost information.

Conclusions

Although we identified a large number of studies, we are able to make only a few specific conclusions for prioritized outcomes in this review. There is evidence that CBT, IPT, exercise, and behavioral activation are effective treatments for treating depressive symptoms in individuals with perinatal depressive disorders. CBT, IPT and specific-acupuncture were associated with increased remission rates for depressive disorders. However, conclusions regarding remission rates were rated low SoE. There was evidence that CBT and IPT are effective in treating anxiety in perinatal individuals with depressive and anxiety disorders. Additionally, IPT but not CBT was found to be effective in reducing anxiety symptoms in individuals with combined depressive and anxiety disorders. RCTs of nonpharmacologic treatments for PTSD are heterogeneous and did not adequately report results. Thus, we were not able to draw conclusions related to the nonpharmacologic treatment of PTSD. Similarly, RCT evidence on nonpharmacologic treatments for OCD is sparse. There is also only very sparse RCT evidence comparing nonpharmacologic to pharmacologic interventions. While findings are promising regarding the potential effectiveness of nonpharmacologic treatments, we identified numerous areas with no or sparse existing evidence. To guide clinical and policy decision-making, future research is needed to evaluate the comparative effectiveness of lesser studied nonpharmacologic interventions and lesser studied perinatal mental health disorders.

Applicability

There are a number of factors that may affect the applicability of our findings. We aggregated findings from interventions conducted in a range of settings, which varied in terms of delivery, study population, and which outcome measures.

For KQ 1, 18/71 studies (25%) were conducted in the United States, which has important implications for what is considered treatment as usual. For most studies in this review, participants in the TAU had access to locally available services for individuals during the perinatal period. Locally available services can vary greatly between countries and health care providers. While some studies gave clear descriptions of what usual care would likely consist of, many did not. Furthermore, in most cases, participants in the TAU groups were not precluded from accessing other nonpharmacologic or pharmacologic treatment from their primary health care or obstetric care providers, which may have an impact on the effect sizes reported in this review. One would expect that studies with more "active" TAU, with more services available compared with other studies, would yield smaller differences with CBT and other active interventions.

The race and ethnicity of participants included in this review may differ from the general population in important ways, affecting the generalizability of these findings. Across studies, the largest proportion of participants (65.2%) were White, almost one-quarter (26.0%) were Black, 4.0% were Asian, 16.5% were Hispanic, and 7.5% were from other racial and ethnic groups. The effectiveness of nonpharmacologic treatments may vary by ethno-cultural background and by socioeconomic factors, including discrimination and minoritization, for which race categorization may be a partial proxy. Participants from different ethno-cultural backgrounds may hold differing beliefs regarding the nature and treatment of mental health conditions, which may result in different responses to nonpharmacologic therapies.

We aggregated studies that tested nonpharmacologic treatments delivered in a range of settings, including in a clinic, at the participants home, or remotely. It is possible that the effectiveness of treatments may vary greatly depending on the mode of delivery or setting. For interventions and comparisons with a more limited evidence base it was not possible to explore whether the study setting or mode of delivery affected their effectiveness. However, where sufficient data was available for studies comparing CBT with TAU and IPT with TAU among individuals with depressive disorders, we found no differences in effect sizes by intervention setting or mode of delivery.

Included studies used either diagnostic tools or screening tools to identify perinatal individuals with mental health conditions. Most studies (n = 44, 62%) used a diagnostic tool or criteria (such as Diagnostic and Statistical Manual of Mental Disorders [DSM] or International Classification of Diseases [ICD] criteria) to confirm the presence of a mental health condition. However, many studies relied on screening tools only. We restricted studies that used screening tools only to identify mental health disorders to those using a validated cut-off. However, many studies employed vastly different cut-offs and there is little consensus regarding what cut-off is likely to identify clinical levels of common mental health conditions for perinatal individuals on screening tools. Where different cut-offs were suggested by prior studies, we chose to use the most inclusive which may have implications for the applicability of our findings. For example, it is possible that interventions may be successful in treating perinatal individuals with less-severe depressive symptoms who would not be diagnosed with a depressive disorder, but may be less effective at treating those with clinically confirmed mental health conditions. Where sufficient data were available, we did conduct subgroup analyses to explore whether effect sizes differed between studies using diagnostic and screening tools to identify mental health conditions. For studies comparing CBT with TAU and IPT with TAU, we found small differences in summary effect sizes between these two sets of studies, but these indirect comparisons across studies do not rule out that real differences may exist.

Studies in this review employed a wide variety of outcome measures at different time-points. Where studies measured sufficiently similar domains, we calculated standardized mean differences (SMD) for outcomes at the end of treatment. This allowed us to draw comparisons across studies, however some measures may be more relevant for perinatal individuals than others and important nuance may be lost by pooling effect sizes across different measures.

Strengths and Limitations

Strengths and Limitations of the Review Process

We followed contemporary standards for SR, including engagement with multiple types of stakeholders in defining and refining both KQs and careful adherence to current SR standards for protocol publication and registration, literature searching, screening, data extraction, risk of bias assessment, qualitative synthesis, quantitative synthesis, and SoE assessment.

One of the biggest challenges in conducting this review was the heterogeneity of study populations, interventions, comparisons, and outcomes used by the eligible RCTs. To manage the heterogeneity in study populations we, in consultation with our domain experts and prior literature, determined cut-offs for identifying perinatal mental health conditions on validated screening tools a priori. However, it is important to note that there is very little consensus regarding the use of cut-offs to identify clinical mental health disorders and the cut-off used by this study may not be applicable to all perinatal populations.

To manage the heterogeneity in interventions and comparisons we implemented an intervention taxonomy which was applied by an expert in complex interventions and reviewed by clinical experts in our team. Our intervention taxonomy grouped interventions into broad categories, which allowed us to draw conclusions regarding the comparative effectiveness between interventions and comparisons groups. Decisions regarding how to classify and intervention could be challenging, especially when the details of the intervention were not well reported. We further managed the heterogeneity in interventions and comparisons by restricting full data extraction, risk of bias assessment, and analysis to comparisons with three or more studies, however this is an important limitation. For example, one limitation of the decision to restrict full data extraction to comparisons with three or more studies is that multi-component interventions were less likely to be fully extracted and analyzed, limiting the conclusions we can draw about these types of interventions. The key findings from such interventions are listed in the evidence map in Appendix Table C-1.

Studies reported wide range of outcomes and used multiple measures to address the same domains. We only combined measures which clearly assessed the same domain and calculated standardized mean differences. However, it is important to note that some outcome measures may be more appropriate for perinatal populations than others. Most studies in this review included outcomes measuring the benefits of nonpharmacologic treatments and few reported on the potential harms. Therefore, we are unable to draw conclusions regarding the potential harms of nonpharmacologic treatments during the perinatal period. Furthermore, effect-sizes are in this review are reported as standardized mean differences, which may not equate to a clinically important difference.

Strengths and Limitations of the Evidence Base

For several areas of this review, including those summarized in the evidence map, evidence was limited or entirely absent. The majority of studies in this review compared CBT with TAU among various perinatal mental health conditions.

We found very few studies testing the comparative effectiveness of nonpharmacologic treatments for individuals with anxiety PTSD or OCD. We did not identify any studies testing nonpharmacologic treatments for individuals with bipolar disorder and very few studies compared nonpharmacologic with pharmacologic treatments.

Studies included in this review differed in how they named, defined, and reported nonpharmacologic treatments, making it challenging to determine if interventions could be meaningfully combined or compared.

Other than conducting subgroup analyses for type of diagnostic/screening tool, delivery methods, and settings of CBT and IPT studies, very little data was available for exploring potential modifiers of treatment effect.

Many studies included in this review were assessing the potential benefits of nonpharmacologic treatments. We identified few RCTs reporting data on any potential harms or adverse effects associated with nonpharmacologic treatments during the perinatal period. We were unable to expand the scope of the review to look for harms among observational studies. This is an important area for future research.

Studies were inconsistent in full reporting of outcome data. Some studies only reported effect sizes, without reporting on averages, at the end of treatment, only reported positive results in tables and reported negative results in the text, or were missing key outcome data. We only included studies in meta-analysis where full outcome data were available.