Couples-based interventions for perinatal depression and anxiety: A global systematic review

Abstract

Background:

Globally, perinatal depression and anxiety affect approximately 26% and 12% of women, respectively, with partner support recognized as a protective factor. This systematic review aimed to synthesize the evidence on the effectiveness of couples-based interventions for perinatal depression and/or anxiety.

Methods:

We searched PsycINFO, Embase, PubMed, CENTRAL, and CINAHL from inception through December 1, 2024, identifying studies published in English that met our eligibility criteria. Two reviewers independently screened records, extracted data, and assessed risk of bias using the Joanna Briggs Institute critical appraisal tools. We conducted a narrative synthesis.

Results:

From 4733 screened records, 26 studies (23 RCTs, three quasi-experimental) met the inclusion criteria, comprising 4547 women from 14 countries. Fourteen of 26 studies (54%) demonstrated significant reductions in maternal depressive symptoms. Descriptive synthesis suggested effectiveness may depend more on intervention content than partner attendance patterns: interventions teaching specific support skills (symptom recognition, practical assistance, help-seeking facilitation) appeared effective across varying attendance levels, while relationship enhancement approaches showed inconsistent results. Six of 12 studies measuring anxiety reported significant reductions, with mixed evidence regarding the role of partner attendance. Studies reported inconsistent effects on paternal mental health and relationship quality. Risk of bias was low in 52% of RCTs.

Conclusion:

Couples-based interventions show promise for reducing perinatal depression and anxiety, though substantial heterogeneity prevents identification of optimal approaches. Descriptive patterns suggest effectiveness may depend more on intervention content than on the number of sessions partners attended, though this hypothesis requires formal testing through meta-analysis with moderator analyses. Absence of mediation analyses across all studies prevented understanding of mechanisms, and geographic concentration in high-income settings limited generalizability to LMIC contexts.

Prospero registration:

CRD42024618459.

1. Background

Perinatal mental disorders are among the most prevalent morbidities during pregnancy and up to one year postpartum (Howard et al., 2014). Globally, perinatal depression affects approximately 26% of women (Al-abri et al., 2023), while perinatal anxiety affects 12% postpartum (Feldman et al., 2025) and 15–20% antenatally (Dennis et al., 2017). This highlights the substantial burden of maternal mental health disorders during this vulnerable developmental period. Prevention and timely treatment are critical not only for protecting maternal health but also because of the cascading impacts on the family unit. Adverse perinatal mental health negatively affects newborn health through diminished mother-infant bonding, low birth weight, and increased risks for a broad range of adverse child health and nutrition outcomes including emotional, behavioral, cognitive, and physical development impairments that persist into late adolescence (Gelaye et al., 2016; Murray et al., 2010; Slomian et al., 2019; Stein et al., 2014; Carosella et al., 2024; Hakanen et al., 2019).

The perinatal period is a critical developmental transition requiring substantial adaptations for mothers, fathers, and the broader family unit (Cowan and Cowan, 2012). The transition to parenthood is associated with decreased relationship satisfaction and increased risk of psychological adjustment difficulties, including depression and anxiety. Marital satisfaction declines during the first year postpartum with medium effect sizes, and small but persistent declines continue in the subsequent 24 months (Twenge et al., 2003; Bogdan et al., 2022). These challenges are heightened by gender-specific risks, with mothers facing societal pressures to conform to idealized motherhood expectations while managing physical and hormonal changes, and fathers navigating co-parenting roles and work-family balance concerns (O’Hara and McCabe, 2013; Cameron et al., 2016).

Quality of couple relationships and psychological adjustment are correlated during the perinatal period. Relationship conflict increases perinatal depression and anxiety risk for both parents, while depression itself can further strain the couple’s relationship (Figueiredo et al., 2018; Davé et al., 2010). Evidence from high-income country settings has identified the role of partner support and couple communication skills in preventing perinatal depression and anxiety (Pilkington et al., 2015a). These important relationship factors extends beyond mental health outcomes, with male partner involvement being associated with broader maternal well-being indicators including improved women’s nutrition, reduced intimate partner violence during pregnancy, expanded social support, improved equitable decision-making, and greater female autonomy (Barker et al., 2010; Fleming et al., 2018; Santhya et al., 2008; Stern and Nyiratunga, 2017; Tokhi et al., 2018).

Despite evidence for partner support as a protective factor against maternal perinatal depression and anxiety, maternal health services remain predominantly women-focused (Taylor et al., 2018). There is a need for ‘male-friendly’ and couple-based approaches that recognize fathers as active participants of maternal mental health rather than passive bystanders (Pilkington et al., 2015a). The role of partners in perinatal mental health interventions may be through recognizing risk factors and signs of depression and anxiety, adjusting support to women’s needs, and encouraging formal help-seeking when necessary (Fonseca and Canavarro, 2017). Both expectant and new parents have advocated for greater partner involvement in preventive interventions, particularly those incorporating depression education and addressing couple relationship dynamics (Feeley et al., 2016; Wheatley et al., 2003). Furthermore, including both partners in interventions represents an approach to addressing the needs of both parents during the perinatal period (Pilkington et al., 2015a).

While prior systematic reviews have examined partner involvement in perinatal mental health interventions, significant gaps persist in our understanding of couples-based approaches for perinatal depression and anxiety. Pilkington et al. (2015a,b) reviewed 13 partner-inclusive prevention interventions for postpartum depression and anxiety, finding that nine (69%) showed significant reductions in maternal depressive symptoms (Pilkington et al., 2015a). However, their review identified minimal father involvement in most interventions and low partner attendance rates as persistent implementation barriers. Alves et al. (2018) reviewed 26 studies of partner-inclusive interventions for postpartum depression but found that only 27% reported partner attendance data, preventing assessment of whether partner participation patterns were associated with intervention efficacy (Alves et al., 2018). Park et al. (2020) conducted a meta-analysis of couples’ psychoeducational interventions, demonstrating small but significant effects on perinatal depression, but their scope was limited to psychoeducational approaches and did not examine anxiety outcomes (Park et al., 2020). Most recently, Refaeli et al. (2024) found that psychological interventions were effective in reducing maternal depression, anxiety, and stress among first-time parents, but noted that most interventions targeted mothers individually rather than couples, and their review excluded multiparous women and non-psychological intervention approaches (Refaeli et al., 2024). These reviews were constrained by one or more limitations: (1) restrictive inclusion criteria that excluded common intervention types or populations, (2) narrow focus on specific intervention modalities (e.g., psychoeducation only), (3) limited assessment of anxiety outcomes despite high comorbidity with depression, or (4) inadequate reporting of partner attendance preventing analysis of how participation patterns relate to effectiveness. To our knowledge, no prior review has comprehensively examined couples-based interventions across diverse theoretical approaches and delivery formats while assessing both depression and anxiety outcomes and systematically documenting partner participation patterns in relation to intervention effectiveness.

This systematic review addresses these gaps by: (1) synthesizing evidence on the effectiveness of couples-based interventions for preventing and treating perinatal depression and anxiety; (2) developing and applying a coding framework to systematically document intended versus achieved partner participation patterns; (3) describing intervention characteristics and partner participation patterns; and (4) assessing effects on secondary outcomes including paternal mental health, relationship quality, and infant outcomes.

2. Methods

2.1. Design

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (Park et al., 2020) and findings were synthesized narratively following synthesis without meta-analysis (SWiM) guidelines (Campbell et al., 2020). The review protocol was registered with the PROSPERO international database (protocol number: CRD42024618459.

2.2. Eligibility criteria

We included published, English language, peer-reviewed studies if they: (1) included women during the perinatal period (pregnancy through 12 months postpartum) with or without baseline depression or anxiety; note that while interventions involved both partners, only maternal mental health outcomes were required for inclusion; (2) evaluated couples-based, partner-based, or partner-inclusive interventions defined as programs that targeted both partners in prevention or treatment approaches; (3) measured outcomes related to prevention and/or treatment of perinatal depression and/or perinatal anxiety; (4) used controlled study designs including RCTs, controlled before-after studies, or interrupted time series; (5) reported measures of maternal depression and/or maternal anxiety using validated scales. We excluded studies focusing solely on postpartum blues, interventions targeting only women or their partners, observational studies without control groups, and studies in languages other than English.

2.3. Search strategy

We conducted a comprehensive search of five databases (PsycINFO, Embase, PubMed, Cochrane Central Register of Controlled Trials [CENTRAL], and Cumulative Index to Nursing and Allied Health Literature [CINAHL]) for relevant studies published from database inception to December 1, 2024. We developed a comprehensive search strategy using four key concepts: (1) perinatal period (e.g., postpartum, perinatal), (2) depression and anxiety (e.g., depression, anxiety), (3) couple and partner terms (e.g., partner, spouse), and (4) intervention (e.g., therapy, counseling). A search strategy was first developed for PubMed, vetted by academic librarians at our institutions (UCSF and London School of Tropical Hygiene and Medicine), and then adapted for the other databases (Appendix 1: Search Strategy). We also hand searched reference lists of relevant systematic reviews and conducted forward citation chaining using Web of Science.

2.4. Data extraction and coding

All included studies were first imported into Covidence (VHI), an online software for managing systematic reviews and checked for duplication. In pairs, authors (PVP, GA, ANE, LG) reviewed the studies in two stages: title and abstract screening followed by full-text screening. Disagreements were resolved through discussion and consensus; and any conflicts were resolved by a third party (AN).

Using a data extraction form piloted on a subset of studies, two pairs of reviewers (LG, PVP, GA, ANE) independently extracted the data. All included studies were systematically coded for key characteristics across five domains. Study characteristics included author, year, country, income-level classification, sample size, study design along with control type. We classified countries by income level using World Bank criteria to assess geographic representation. Intervention characteristics captured the primary approach (e.g., cognitive behavior therapy (CBT), interpersonal therapy (IPT), psychoeducation, relationship strengthening, mindfulness, creative arts), number and intensity of sessions (low: ≤8 h; moderate: 8–20 h; high: >20 h) based on the distribution of intervention duration in included studies, delivery format (individual, group, online), timing (prenatal, postnatal, or both), and whether the intervention was preventive or treatment-focused.

Given the focus on couples-based interventions, we developed a novel coding system for partner involvement or engagement in the couples-based intervention, distinguishing between dyadic design intent and actual dyadic delivery achieved. We devised the coding framework based on the qualitative data extracted from included studies. Design intent was classified as: D1 (optional partner involvement with no specific partner content), D2 (supportive partner role with limited sessions), D3 (full dyadic participation encouraged with both partners as equal participants), or D4 (combined dyadic and individual components for women and partners with structurally distinct formats). Actual dyadic delivery was coded as: F1 (partner attended <25% of the sessions), F2 (partner attended 25–75% of sessions highlighting partial attendance), F3 (partner attended most/all (>75%) sessions jointly with woman), or F4 (partner attended both joint and individual sessions as designed). Since every intervention did not have identical number of sessions and had varying intent for partner involvement, and prior reviews had typically classified studies dichotomously as partner-involved or not, we chose proportional attendance thresholds (25%, 75%) for F1-F4 categories to capture dose-response patterns and meaningfully quantify the degree of partner involvement. We acknowledge these thresholds are pragmatic choices given the lack of established cut-points in the literature. For studies reporting attendance qualitatively as ‘some’ without specific percentages, we classified these as F2 (partial attendance, 25–75%) by default, unless contextual data (e.g., study narratives indicating high partner engagement, minimal dropout) suggested near-universal attendance, in which case classification was upgraded to F3 (>75% sessions attended by the partner). For example, in the study by Milgrom and colleagues (Milgrom et al., 2005), partners were invited to 3 of 12 intervention sessions but actual partner attendance for these sessions was not reported. We used the overall program completion rate (63% completed the intervention) as a proxy to classify partner involvement as F2 (partial attendance), acknowledging this represents an estimation in the absence of session-specific partner data. Two independent reviewers from the team (PVP, GA, ANE) independently coded each study’s dyadic design intent and achieved dyadic delivery format based on published methods sections, flow diagrams, and attendance data. Inter-rater reliability was strong, with Cohen’s kappa of 0.91 for design intent (D1-D4) and 0.88 for achieved participation (F1-F4). Disagreements (n = 3) were resolved through discussion and consensus (LG).

Primary outcomes were coded for effects on perinatal depression and/or perinatal anxiety, including scales used, effect sizes where reported, timing of follow-up, and measures at various time points. Secondary outcomes captured effects on paternal mental health, relationship quality, and infant outcomes.

Due to substantial heterogeneity in intervention types, populations, outcomes, and measurement tools, meta-analysis was not feasible. We conducted a narrative synthesis following SWiM reporting guidelines (Campbell et al., 2020). Two independent reviewers extracted study characteristics, intervention characteristics (e.g., intensity, dyadic design intent, dyadic participation), and outcome data for all included studies as described in the data extraction process. We merged and reconciled the dual extractions through consensus discussion. We then grouped studies by primary outcome (depression, anxiety) and created matrices organizing studies by intervention approach, dyadic design (D1-D4), and dyadic participation level (F1-F4).

For our primary outcomes (depression and/or anxiety), we extracted between-group differences, effect sizes, and p-values at primary end-points as specified by study authors, or if not specified, at the last post-intervention follow-up assessment. Where available, we extracted and reported standardized effect sizes (Cohen’s d, η²p, or equivalent metrics) to characterize intervention magnitude. For studies not reporting standardized effect sizes, we described direction and magnitude of change using available metrics (e.g., mean differences, odds ratios). We reported sample sizes alongside all effect estimates to allow readers to assess study precision and distinguish well-powered studies with small or null effects from potentially underpowered studies with uncertain effects.

We categorized primary outcomes as “positive” (statistically significant between-group difference favoring intervention, p < 0.05) or “null” (no significant difference). We acknowledge that categorizing outcomes by statistical significance has limitations: it conflates effect detection with effect magnitude and implicitly weights studies equally regardless of sample size or precision (McKenzie and Brennan, 2024). Accordingly, statistical significance was used as a descriptive organizing heuristic rather than as a measure of evidentiary weight. Our interpretation of findings was informed by triangulation across multiple factors: direction of effect, reported effect sizes (where available), consistency across studies with similar characteristics, sample size, intervention content and delivery features, and risk of bias. This approach was necessitated by incomplete and inconsistent effect size reporting across included studies, which precluded formal meta-analysis.

For studies measuring both depression and anxiety, we synthesized each outcome separately. We also extracted the magnitude and direction of effects for secondary outcomes (paternal mental health, relationship quality, infant outcomes). In our narrative synthesis, we calculated the proportion of studies showing statistically significant improvements versus no significant differences for each outcome and synthesized findings narratively, identifying patterns and consistencies across studies while documenting sources of heterogeneity. Results are presented in summary tables and narrative text.

Risk of bias was assessed independently by two reviewers (PVP and GA or ANE) using the revised Joanna Briggs Institute (JBI) Critical Appraisal Tools for randomized and quasi-experimental studies (Barker et al., 2023, 2024). Disagreements between reviewers were resolved through discussion or consultation with a third reviewer (LG). Studies were classified as low risk of bias (≥20/26 points for RCTs; ≥14/18 points for quasi-experimental studies), moderate risk (13–19/26 points for RCTs; 10–13/18 points for quasi-experimental studies), or high risk (<13/26 points for RCTs; <10/18 points for quasi-experimental studies). We assessed risk of bias for all studies using JBI critical appraisal tools but did not apply formal certainty of evidence frameworks (e.g., GRADE) given the narrative approach and heterogeneity precluding evidence pooling. We integrated risk of bias assessments into our interpretation of findings by examining whether positive effects clustered within studies of particular methodological quality. Specifically, we calculated the proportion of low versus moderate risk-of-bias studies demonstrating positive effects for both depression and anxiety outcomes, and we identified common sources of bias (particularly attrition) associated with null findings.

3. Results

As outlined in Appendix 2: PRISMA, a total of 6074 references were imported for screening from database searches. After identifying and removing 1854 duplicates (18 manually identified and 1836 through Covidence), 4220 studies underwent title and abstract screening. Following the exclusion of 4125 records based on title and abstract review (e.g., non-perinatal populations, no partner involvement, no depression or anxiety measures, observational study designs without control group), 95 studies were assessed for full-text eligibility. During full-text assessment, 79 studies were excluded for the following reasons: ineligible study design (n = 49), full-text not found (n = 10), ineligible comparator (n = 7), ineligible intervention (n = 7), ineligible outcomes (n = 4), and ineligible patient population (n = 2). This resulted in 16 eligible studies from the database search. Forward citation searching in Web of Science yielded two additional eligible studies, and hand searching reference lists of five relevant systematic reviews identified eight additional eligible studies. In total, 26 studies met the inclusion criteria and were included in this systematic review.

4. Study characteristics

The study characteristics of the included 26 studies are presented in Table 1 in Appendix 3. The 26 studies comprised 23 RCTs and three quasi-experimental studies (Onogwu et al., 2023; Jiang et al., 2024). Twenty-two (84%) studies were conducted in high-income countries, with seven studies from the United States (Shapiro and Gottman, 2005; Feinberg and Kan, 2008; Gambrel and Piercy, 2015; Feinberg et al., 2016, 2020; McHale et al., 2022; Canfield et al., 2023), four from Australia (Buist et al., 1999; Matthey et al., 2004; Milgrom et al., 2005; Fisher et al., 2016), two each from Singapore (Shorey et al., 2017, 2023), United Kingdom (O’Mahen et al., 2022; Kennedy et al., 2023), and Hong Kong (Ngai et al., 2020; Ngai and Gao, 2022), and one each from Canada (Misri et al., 2000), Spain/France multi-country (Ortiz Collado et al., 2014), Italy (Burgio et al., 2024), Taiwan (Kuo et al., 2022), and Netherlands (Van der Meulen et al., 2023). Four studies were conducted in middle-income countries: two in Nigeria (Onogwu et al., 2023; Jiang et al., 2024) classified as lower-middle-income, one each in Iran(Alipour et al., 2020) and China (Zhao et al., 2017) classified as upper-middle-income.

The studies covered 4547 perinatal women. Women sample sizes ranged from 29 (Misri et al., 2000) to 455 (Ngai and Gao, 2022) per study, with maternal ages typically between 24 and 34.

4.1. Intervention

4.1.1. Intervention approaches

The intervention characteristics and outcomes are noted in Table 2 under Appendix 3. The couples-based interventions used diverse therapeutic approaches across the 26 studies with some overlapping strategies. Psychoeducation combined with partner support was the most prevalent strategy implemented in eight studies (Misri et al., 2000; Shorey et al. 2017, 2023; Zhao et al., 2017; Alipour et al., 2020; Ngai et al., 2020; Ngai and Gao, 2022; Burgio et al., 2024). Relationship strengthening strategies, defined as programs focusing on improving communication skills, building emotional connection, enhancing couple functioning, and coparenting dynamics, were employed in six studies (Shapiro and Gottman, 2005; Feinberg and Kan, 2008; Feinberg et al., 2016; Fisher et al., 2016; Alipour et al., 2020; Kennedy et al., 2023). Combined approaches involving multiple strategies were used in three studies (Feinberg et al., 2020; Kuo et al., 2022; Van der Meulen et al., 2023), typically integrating psychoeducation with relationship strengthening or mindfulness components. Cognitive-behavioral therapy approaches were explicitly implemented in three studies (Milgrom et al., 2005; Ngai et al., 2020; Canfield et al., 2023). Social-media based creative arts therapies, including dance, storytelling, and visual art were employed in two studies (Onogwu et al., 2023; Jiang et al., 2024). Educational and psychosocial approaches (Buist et al., 1999; Matthey et al., 2004), mindfulness-based relationship education (Gambrel and Piercy, 2015), psychosomatic interventions (Ortiz Collado et al., 2014), prenatal coparenting consultation (McHale et al., 2022), and perinatal-adapted cognitive-behavioral therapy along with mindfulness component (O’Mahen et al., 2022) were each used in single studies.

4.1.2. Intervention setting

Hospital-based settings were most common (n = 15), including antenatal clinics, maternity units, and specialized pregnancy services (Buist et al., 1999; Misri et al., 2000; Matthey et al., 2004; Shapiro and Gottman, 2005; Feinberg and Kan, 2008; Ortiz Collado et al., 2014; Gambrel and Piercy, 2015; Feinberg et al., 2016; Zhao et al., 2017; Ngai et al., 2020; Kuo et al., 2022; Ngai and Gao, 2022; O’Mahen et al., 2022; Kennedy et al., 2023; Burgio et al., 2024). Community health centers and primary care facilities hosted five studies (Milgrom et al., 2005; Fisher et al., 2016; Alipour et al., 2020; Van der Meulen et al., 2023; McHale et al., 2022). Six studies used online or digital platforms exclusively, delivering interventions via mobile apps or social media (Shorey et al. 2017, 2023; Feinberg et al., 2020; Canfield et al., 2023; Onogwu et al., 2023; Jiang et al., 2024).

4.1.3. Intervention timing and intensity

Interventions varied considerably in timing (i.e., start of the intervention). Five studies (19%) delivered interventions exclusively during pregnancy, typically starting in the second trimester (Ortiz Collado et al., 2014; Gambrel and Piercy, 2015; Zhao et al., 2017; Alipour et al., 2020; McHale et al., 2022). Six studies (23%) were postnatal-only, commencing within 4–8 weeks postpartum (Misri et al., 2000; Milgrom et al., 2005; Fisher et al., 2016; Shorey et al., 2017; Onogwu et al., 2023; Jiang et al., 2024). The majority (15 studies, 58%) spanned both periods, beginning during pregnancy and extending into the postpartum period (Buist et al., 1999; Matthey et al., 2004; Shapiro and Gottman, 2005; Feinberg and Kan, 2008; Feinberg et al., 2016, 2020; Ngai et al., 2020; Kuo et al., 2022; Ngai and Gao, 2022; O’Mahen et al., 2022; Canfield et al., 2023; Kennedy et al., 2023; Shorey et al., 2023; Van der Meulen et al., 2023; Burgio et al., 2024), recognizing the importance of continuity across the perinatal transition.

Intervention intensity varied substantially across studies, categorized into low, moderate, and high intensity levels. High-intensity interventions were implemented in 6 studies (Buist et al., 1999; Shapiro and Gottman, 2005; Feinberg et al., 2016; Ortiz Collado et al., 2014; Van der Meulen et al., 2023; Onogwu et al., 2023), typically involving 10 or more sessions or exceeding 20 h of contact time. Moderate-intensity interventions were most common, employed in 13 studies (Misri et al., 2000; Matthey et al., 2004; Milgrom et al., 2005; Feinberg and Kan, 2008; Alipour et al., 2020; Feinberg et al., 2020; Shorey et al. 2017, 2023; Zhao et al., 2017; Kuo et al., 2022; Ngai and Gao, 2022; Jiang et al., 2024), involving 6–9 sessions or 8–20 h total contact. Seven studies reported low-intensity interventions (Gambrel and Piercy, 2015; Fisher et al., 2016; Ngai et al., 2020; O’Mahen et al., 2022; Canfield et al., 2023; Kennedy et al., 2023; Burgio et al., 2024), comprising 1–5 sessions or 8 h or less of direct contact time. The intensity classification reflected both session frequency and duration, with some studies incorporating additional components such as daily home practice or continuous app access.

4.1.4. Intervention content and delivery

In terms of content, the interventions focused on enhancing relationship dynamics through communication skills training, conflict resolution strategies, and emotional regulation techniques, educating couples about perinatal mental health risks and protective factors, promoting coparenting collaboration and division of labor, and developing coping strategies for the transition to parenthood. Core session components included psychoeducation about postpartum depression and anxiety, cognitive restructuring and problem-solving skills, mindfulness practices and stress reduction techniques, partner support strategies and empathy building, attachment and bonding preparation, and relapse prevention planning. The mode of intervention delivery ranged from individual couple sessions (Misri et al., 2000; Kennedy et al., 2023), group-based classes with multiple couples(Buist et al., 1999; Feinberg and Kan, 2008; Feinberg et al., 2016; Ngai et al., 2020; Ngai and Gao, 2022), intensive weekend workshops (Shapiro and Gottman, 2005), mobile application-based self-paced modules (Shorey et al., 2017, 2023; Canfield et al., 2023), hybrid approaches combining in-person sessions with phone follow-ups (Ngai et al., 2020; Ngai and Gao, 2022; Ortiz Collado et al., 2014), to innovative social media-delivered creative arts therapies via WhatsApp platforms (Onogwu et al., 2023; Jiang et al., 2024).

4.1.5. Intervention delivery agents

Intervention delivery was facilitated by various healthcare professionals with specialized training. Mental health professionals, including clinical psychologists, psychiatrists, and licensed therapists, delivered interventions in nine studies (Matthey et al., 2004; Shapiro and Gottman, 2005; Milgrom et al., 2005; Gambrel and Piercy, 2015; Alipour et al., 2020; Kennedy et al., 2023; Burgio et al., 2024). Nursing professionals, including midwives and maternal child health nurses, were primary delivery agents in eight studies (Buist et al., 1999; Misri et al., 2000; Ortiz Collado et al., 2014; Fisher et al., 2016; Shorey et al., 2017; Ngai et al., 2020; Ngai and Gao, 2022; O’Mahen et al., 2022). Four studies used male-female facilitator teams (Feinberg and Kan, 2008; Feinberg et al., 2016; McHale et al., 2022; Zhao et al., 2017), while specialized creative arts therapists delivered interventions in two studies (Onogwu et al., 2023; Jiang et al., 2024). Digital platforms such as asynchronous support or self-administered content were used in three studies (Feinberg et al., 2020; Canfield et al., 2023; Shorey et al., 2023). Multidisciplinary teams combining different professional backgrounds were used in several studies to enhance intervention delivery expertise. None of the interventions were delivered by lay health workers or lay counselors.

4.1.6. Couples-based/partner inclusion approaches

Most interventions were designed with dyadic participation as a core component. Dyadic participation was encouraged in 19 studies (Buist et al., 1999; Misri et al., 2000; Matthey et al., 2004; Shapiro and Gottman, 2005; Feinberg and Kan, 2008; Ortiz Collado et al., 2014; Gambrel and Piercy, 2015; Feinberg et al., 2016; Shorey et al., 2017; Alipour et al., 2020; Feinberg et al., 2020; Kuo et al., 2022; Ngai and Gao, 2022; Canfield et al., 2023; Kennedy et al., 2023; Onogwu et al., 2023; Shorey et al., 2023; Van der Meulen et al., 2023; Jiang et al., 2024), where both partners were expected to attend together with curriculum specifically focused on couple-based skills and mutual engagement. Two studies within their dyadic design intent incorporated both dyadic and individual components (Ngai et al., 2020; McHale et al., 2022), featuring structured sessions that included joint couple activities alongside separate individual components for each partner. Partner involvement was limited to a supportive role in four studies (Milgrom et al., 2005; Fisher et al., 2016; O’Mahen et al., 2022; Zhao et al., 2017), where partners were invited to participate but the primary focus remained on the woman’s adjustment. Only one study employed optional partner involvement where participation was neither required nor central to the intervention design (Burgio et al., 2024).

Regarding actual dyadic delivery format, only 13 studies (50%) reported full dyadic attendance(Buist et al., 1999; Misri et al., 2000; Shapiro and Gottman, 2005; Ortiz Collado et al., 2014; Gambrel and Piercy, 2015; Feinberg et al., 2016; Alipour et al., 2020; Feinberg et al., 2020; Kuo et al., 2022; Kennedy et al., 2023; Onogwu et al., 2023; Van der Meulen et al., 2023; Jiang et al., 2024), where partners attended all sessions jointly with women. Six studies implemented dyadic plus individual delivery formats (Matthey et al., 2004; Shorey et al., 2023; Ngai et al., 2020; McHale et al., 2022; Ngai and Gao, 2022; Shorey et al. 2023, 2023), combining joint dyadic sessions with separate individual components for each dyad. Six studies had partial partner attendance (Milgrom et al., 2005; Feinberg and Kan, 2008; Fisher et al., 2016; O’Mahen et al., 2022; Canfield et al., 2023; Burgio et al., 2024), with partners participating in 25–75% of sessions. One study used woman-only delivery with minimal partner attendance (Zhao et al., 2017). Given this variation in both design intent and actual implementation of partner involvement, we examined the effectiveness of these diverse approaches on maternal mental health outcomes.

5. Outcomes

5.1. Effectiveness of couples-based interventions for perinatal depression

All 26 studies examined perinatal depression outcomes with considerable heterogeneity in effect magnitudes. Effect estimates favored the intervention in 17 studies (65%), with 14 showing statistically significant reductions in depressive symptoms. Where standardized effect sizes were reported, most studies were in the small-to-medium range (Cohen’s d = 0.20–0.56). Effect sizes were larger in two quasi-experimental studies from Nigeria (Onogwu et al., 2023; Jiang et al., 2024; η²p = 0.41–0.49).

Of the 12 studies with null findings, seven showed effects in a favorable direction, suggesting potential clinical benefit that may not have reached statistical significance due to limited power (Buist et al., 1999; Ortiz Collado et al., 2014; Fisher et al., 2016; Shorey et al., 2017; Canfield et al., 2023; Kennedy et al., 2023; Burgio et al., 2024). Four studies showed no between-group differences (Gambrel and Piercy, 2015; McHale et al., 2022; O’Mahen et al., 2022; Shorey et al., 2023), and one study showed a non-significant effect in the unfavorable direction (Van der Meulen et al., 2023). Null effects were observed in both small, potentially underpowered trials (Gambrel and Piercy, 2015; N = 33; Canfield et al., 2023, N = 30) and large, well-powered studies (Fisher et al., 2016, N = 400; Shorey et al., 2023, N = 200), suggesting true heterogeneity in treatment response rather than consistent power limitations. Among well-powered null studies reporting effect sizes, effects were consistently small with confidence intervals crossing zero (Fisher et al., 2016: AMD = −0.47, 95% CI −1.02 to 0.09; Shorey et al., 2017: MD = −0.69, 95% CI −1.66 to 0.29). One study demonstrated a non-significant trend approaching clinical meaningfulness (1.76-point EPDS reduction, p = 0.08, N = 184) (Ortiz Collado et al., 2014).

Effect magnitude varied across different levels of partner involvement. Among studies with complete dyadic delivery (either full dyadic attendance or combined dyadic plus individual delivery sessions), 8 of 13 demonstrated positive effects (Misri et al., 2000; Shapiro and Gottman, 2005; Alipour et al., 2020; Feinberg et al., 2016, 2020; Kuo et al., 2022; Onogwu et al., 2023; Jiang et al., 2024) Studies using combined joint and individual sessions showed positive effects in 3 of 6 cases (Matthey et al., 2004; Ngai et al., 2020; Ngai and Gao, 2022). Notably, positive effects also emerged from studies with partial partner attendance (Feinberg and Kan, 2008; Milgrom et al., 2005) and even minimal partner involvement (Zhao et al., 2017).

The most effective interventions included treatments for women with diagnosed depression, preventive interventions for both high-risk groups, and universal populations. The most consistent positive effects were observed in the early postpartum period, particularly around 6–8 weeks. Only two studies reported longer-term improvement at 10–12 months (Feinberg et al., 2016; Shapiro and Gottman, 2005).

5.2. Effectiveness of couples-based interventions for perinatal anxiety

Of the 12 studies that measured maternal anxiety as a distinct outcome, six studies (50%) demonstrated statistically significant reductions, though effect sizes varied considerably. Among studies reporting standardized effect sizes, two demonstrated large reductions: Kuo et al. (2022) found d = 0.76 at 36 weeks’ gestation and d = 0.62 at one week postpartum among women with high fear of childbirth (N = 106), and Milgrom et al. (2005) showed an 8-point Beck Anxiety Inventory reduction (approximately d = 0.8) at 18–30 weeks postpartum (N = 192). Moderate-to-small effects were observed in Feinberg and Kan (2008); d = 0.38, N = 169) and O’Mahen et al. (2022; d = 0.42 at 10 weeks postpartum; N = 114). Fisher et al. (2016) reduced mild-to-moderate anxiety symptoms (AOR = 0.58, 95% CI 0.35–0.97, N = 400). Additional studies reported statistically significant anxiety reductions without effect sizes: Buist et al. (1999; p = 0.036, N = 44), Alipour et al. (2020; p = 0.001, N = 60), and Canfield et al. (2023; mid-intervention only, p = 0.02, N = 30). Where effect sizes were reported for null studies, they were negligible (Feinberg et al., 2016; Van der Meulen et al., 2023), suggesting true minimal effects in well-powered samples.

Effect magnitude varied across different levels of partner involvement, with large effects observed in studies with both minimal attendance (Milgrom et al., 2005) and full attendance (Kuo et al., 2022). Null effects occurred across full-attendance (Kennedy et al., 2023), app-based (Shorey et al., 2023), and mindfulness-based couple interventions (Van der Meulen et al., 2023).

The timing of anxiety benefits varied considerably across successful studies - some from as early as one week postpartum (Kuo et al., 2022) to 18–30 weeks postpartum (Milgrom et al., 2005). Sustained anxiety benefits beyond 6 months were not observed in any study with longer-term follow-up. Milgrom et al. (2005) found no between-group differences at one-year follow-up despite achieving large short-term effects (~d = 0.8), and O’Mahen et al. (2022) reported that early trends (d = 0.42 at 10 weeks) were not sustained at 18 or 34 weeks.

5.3. Other outcomes

Paternal depression: Studies measuring paternal depression showed mixed results. Some interventions demonstrated significant benefits for fathers (Shapiro and Gottman, 2005; Feinberg et al., 2020; Ngai and Gao, 2022; Burgio et al., 2024), while others found no effects on paternal mental health despite partner participation (Feinberg and Kan, 2008; Ngai et al., 2020; McHale et al., 2022; Kennedy et al., 2023).

Relationship and family functioning: Relationship outcomes varied considerably across studies. Positive effects included improved marital satisfaction (Alipour et al., 2020; Zhao et al., 2017), enhanced coparenting quality (Feinberg et al., 2016) and dyadic adjustment (Misri et al., 2000), and reduced intimate partner violence (Feinberg et al., 2016). However, several studies found no relationship benefits (Kennedy et al., 2023; O’Mahen et al., 2022), and Feinberg (Feinberg et al., 2016) unexpectedly found decreased relationship satisfaction despite improved coparenting. A couple of studies reported gender differences in relationship satisfaction, one study (Gambrel and Piercy, 2015) found men showed significant improvements in relationship satisfaction while women showed no change, whereas another study found women experienced relationship adjustment difficulties postpartum regardless of intervention while men’s satisfaction remained stable (Ortiz Collado et al., 2014). Some couples-based interventions also improved perceived social support (Buist et al., 1999; Milgrom et al., 2005; Shorey et al., 2017), though these gains were not always sustained (Shorey et al., 2023).

Infant/child outcomes: Infant outcomes showed promise in several studies, with improvements in soothability, attention duration, and sleep patterns (Feinberg and Kan, 2008; Feinberg et al., 2016), reduced unsettled behaviors among infants with baseline sleep problems (Fisher et al., 2016), and reduced pre-term birth and higher birthweight (Ortiz Collado et al., 2014). Interventions targeting specific perinatal concerns had significant effects with a mindfulness intervention substantially reduced fear of childbirth while increasing mindfulness (Kuo et al., 2022). However, concerning findings were also noted: significantly worse parent-infant bonding in the intervention group at six months (Shorey et al., 2023), and no improvements in birth satisfaction despite intensive training (Van der Meulen et al., 2023).

None of the 26 studies included in the review conducted formal mediation analysis.

5.4. Risk of bias

Risk of bias assessment revealed that studies demonstrated generally acceptable methodological quality. For the 23 RCTs, scores ranged from 13 to 23 points out of a maximum 26 points. Twelve studies (52%) were classified as low risk of bias (20–26 points), 11 studies (48%) were classified as moderate risk of bias (13–19 points). No RCTs were rated as high risk of bias (<13 points). The three quasi-experimental studies scored 15,16, and 18 points out of a maximum 18 points, all receiving low risk of bias classifications (83%, 89%, and 100% of maximum score, respectively) (Onogwu et al., 2023; Burgio et al., 2024; Jiang et al., 2024). The most frequently identified methodological limitations were performance bias due to the inability to blind participants and intervention providers (inherent to behavioral interventions), unclear or absent outcome assessor blinding procedures, and high attrition rates in some studies. See Appendix 4: Risk of Bias.

Among studies reporting perinatal depression, significant reductions in depressive symptoms were more prevalent among low risk of bias studies: 10 of 15 (67%) demonstrated significant positive effects compared to 4 of 11 (36%) moderate-ROB studies. Null effects were distributed across both categories, with 5 low-risk of bias and 7 moderate-risk of bias studies showing no significant intervention effects. The most robust evidence, combining large samples (n > 300), low risk of bias, and significant positive effects, came from 3 RCTs: Feinberg et al., (2016) (n = 399), Ngai and Gao, 2022 (n = 455), and Zhao et al., (2017) (n = 352). Null-effect studies were often characterized by high differential attrition (Kennedy et al., 2023), high overall attrition (Shorey et al., 2023), and non-random dropout favoring participants with better baseline symptoms (Van der Meulen et al., 2023).

Positive effects on anxiety were also more prevalent among low risk of bias studies. Of the six studies demonstrating statistically significant reduction in anxiety, four were rated low risk of bias (Feinberg and Kan, 2008; Fisher et al., 2016; Alipour et al., 2020; Kuo et al., 2022) and only two were moderate risk of bias (Buist et al., 1999; Milgrom et al., 2005). Null effects occurred predominantly in moderate risk-of-bias studies (O’Mahen et al., 2022; Kennedy et al., 2023; Shorey et al., 2023; Van der Meulen et al., 2023), with two low risk-of-bias studies reporting null effects (Canfield et al., 2023; Feinberg et al., 2016), although both of these studies favored intervention. Common sources of bias in null-effect studies included high or differential attrition rates (Shorey et al., 2023; Van der Meulen et al., 2023; Kennedy et al., 2023), which may have attenuated intervention effects.

6. Discussion

This systematic review narratively examined 26 studies evaluating couples-based interventions for perinatal depression and anxiety. Studies employed diverse therapeutic approaches, from structured psychological therapies (CBT, IPT) to relationship-strengthening programs and creative arts interventions, adapted for couple delivery. The studies were conducted in 14 countries, and there was substantial heterogeneity across studies in dyadic intervention implementation, delivery formats, population characteristics, and timing that precluded meta-analysis. Fourteen studies showed significant reductions in maternal depression (14/26), with similar proportions showing benefits for maternal anxiety (6/12). These findings are consistent with previous reviews finding that partner-inclusive interventions show benefits for preventing maternal perinatal depression, though with variable effect sizes (Pilkington et al., 2015; Alves et al., 2018).

6.1. Partner participation patterns and intervention content

Our narrative synthesis of partner participation patterns reveals considerable variability in outcomes across different attendance levels. In our descriptive analysis, some interventions with minimal partner involvement (e.g., single-session partner participation) demonstrated positive effects, while certain interventions with complete dyadic attendance throughout showed null effects. This observed pattern is consistent with the hypothesis that factors beyond the partner attendance quantity, such as intervention content, implementation fidelity, outcome timing or population characteristics, may influence outcomes. However, this relationship requires confirmation through formal meta-analysis, as apparent patterns in narrative synthesis may be confounded by study quality, baseline severity, statistical power, or other design features.

Prior systematic reviews have also identified intervention content as a potentially important but underexamined factor in partner-inclusive perinatal interventions (Alves et al., 2018; Pilkington et al., 2015). Alves and colleagues specifically noted that limited reporting of partner attendance made it difficult to determine whether benefits stemmed from partner presence or intervention content (Alves et al., 2018). Our descriptive analysis provides preliminary, hypothesis-generating insights: interventions teaching specific behavioral skills, including, symptom recognition, targeted support strategies, and help-seeking facilitation, demonstrated positive outcomes across varying attendance levels (Zhao et al., 2017; Milgrom et al., 2005), while those emphasizing general relationship enhancement without concrete strategies showed inconsistent results (Kennedy et al., 2023; Gambrel and Piercy, 2015). Meta-analysis and moderator testing would be essential to quantify this relationship and control for potential confounders. This observed pattern from narrative synthesis suggests that equipping partners with focused support competencies warrants investigation as a potential mechanism. If confirmed via formal moderator testing, this pattern could have important intervention implementation implications: in settings where consistent partner attendance is challenging, interventions prioritizing skill-building content over attendance maximization may offer a viable implementation strategy. However, given the descriptive nature of our synthesis and potential for confounding, this should be considered a hypothesis for future investigation rather than a practice recommendation.

6.2. Variable effects on paternal mental health and relationship quality

Couples-based interventions produced variable effects on paternal mental health and relationship quality, with no consistent patterns across studies. This mirrors findings from Regan et al.’s (2012) cancer review of couples’ interventions which found inconsistent effects on caregiver/partner mental health and relationship quality, despite consistent benefits for patient outcomes (Regan et al., 2012). These findings suggest that supporting mothers may be insufficient to improve partner outcomes, which may require distinct intervention components addressing partners’ experiences. However, inconsistent measurement and reporting of paternal outcomes limits definitive conclusions. Future research should clarify whether interventions conceptualize partner involvement as instrumental support for maternal outcomes or as dyadic interventions intended to benefit both partners, and test whether these different approaches produce different patterns of effects.

6.3. Mechanisms and mediation

No studies conducted formal mediation analyses, despite several measuring social support, relationship quality, and communication as potential mediators. This prevents identification of mechanisms linking partner involvement to maternal outcomes and precludes evidence-based refinement of intervention components. However, this gap reflects a broader issue in that most included studies did not explicitly articulate a theoretical framework guiding the intervention or specify hypothesized mechanisms of change. Future studies should be grounded in dyadic theoretical models such as interdependency theory or communal coping frameworks and incorporate mediation analyses within their primary protocols and report effects on both partners using standardized measures. Development of reporting guidelines specific to couples-based interventions, potentially as a CONSORT extension, would standardize documentation of partner attendance, engagement quality, and reasons for non-participation.

The coding framework (D/F) for design intent and achieved participation, while novel, represents a preliminary tool for systematically documenting partner participation in couples-based interventions. While it addresses a gap identified in prior reviews regarding inconsistent reporting of partner attendance (Alves et al., 2018), the framework requires further validation. The attendance thresholds used may not optimally capture meaningful variation in partner engagement, and the categories do not capture qualitative aspects of participation such as engagement quality or skill acquisition. Future research should examine whether alternative threshold definitions yield different patterns.

6.4. Other missed opportunities

The geographic distribution of studies underscores significant inequities in research attention. With 81% of studies conducted in high-income settings and only two studies from Nigeria representing lower-middle-income countries, our findings predominantly reflect interventions developed for and tested in resource-rich contexts. This geographic bias has critical implications for global applicability. Low- and middle-income countries (LMICs) face unique challenges that may fundamentally affect intervention design and implementation: greater stigma in accessing mental health care, limited mental health infrastructure, stronger extended family involvement in childcare, different cultural conceptualizations of partner roles, and resource constraints affecting service delivery (Gelaye et al., 2016; Yargawa and Leonardi-Bee, 2015; Comrie-Thomson et al., 2015; Mitchell et al., 2023; Singla et al., 2017a). The absence of studies from South Asia, sub-Saharan Africa (except Nigeria), and Latin America – regions with the highest global burden of perinatal mental health issues underlines a major evidence gap. Future research must prioritize intervention development and testing of couples-based interventions in LMICs.

Further, none of the included studies utilized lay health workers or community health workers to deliver couples-based interventions. This represents a significant gap given the growing evidence supporting task-shifting approaches for perinatal mental health in low-resource settings (Singla et al., 2017b; Rahman et al., 2021). Lay health worker-delivered interventions have demonstrated effectiveness for maternal depression in LMICs and may offer a scalable model for couples-based approaches where mental health specialist availability is limited. The absence of such studies in our review further limits the applicability of current evidence to LMIC contexts where task-shifting is often essential for sustainable service delivery.

6.5. Limited evidence for anxiety outcomes

Yet another striking gap in the literature is the underassessment of anxiety despite its high prevalence (impacting 12–20% perinatal women) and frequent comorbidity with depression (Feldman et al., 2025; Dennis et al., 2017). Only 12 of 26 studies (46%) measured anxiety as a distinct outcome, with mixed results: six demonstrated significant reductions and six found null effects. This limited and heterogeneous evidence base precludes conclusions about the effectiveness of couples-based interventions specifically for perinatal anxiety. Several factors may contribute to the inconsistent findings: heterogeneity in anxiety measures (GAD-7, STAI, BAI, HADS, DASS-21), most interventions were designed primarily for depression rather than explicitly targeting anxiety symptoms, varying timing of assessment, and limited statistical power in some studies.

Theoretically, partners may be well-positioned to recognize and respond to observable anxiety manifestations, including intrusive thoughts about infant safety or excessive worrying about infant well-being (Feldman et al., 2025), and emerging evidence suggests that educational interventions targeting the parental dyad, particularly communication skills, and combining mental health psychoeducation with baby care skills show promise for anxiety reduction (Feldman et al., 2025). However, whether partner involvement actually improves anxiety outcomes, and through what mechanisms, remains an empirical question. Future research should prioritize systematic investigation of couples-based approaches explicitly designed for anxiety, using validated anxiety-specific measures and adequate sample sizes to detect effects.

6.6. Strength of evidence and methodological considerations

Given substantial heterogeneity in intervention approaches, populations, and outcome measurement, we assessed the overall certainty of evidence as low to moderate. Sources of uncertainty include: heterogeneity in intervention content and delivery, limiting comparability across studies; performance bias inherent to behavioral interventions (100% of studies); unclear or absent outcome assessor blinding (69% of studies); unclear allocation concealment (52% of RCTs); high or differential attrition rates (46% of studies), ranging from 15% to 37%; small sample sizes in seven studies (N < 40), limiting statistical power; variable outcome assessment timing; and concentration of evidence from high-income country settings (81%), limiting generalizability to LMIC contexts.

For perinatal depression and anxiety, positive effects did not cluster within methodologically weaker studies. Two-thirds of low-risk of bias studies (67%) demonstrated significant benefits compared to approximately one-third of moderate-risk of bias studies (36%), suggesting that methodological bias does not account for the observed positive findings. Similarly, for perinatal anxiety, four of six low-risk of bias studies (67%) showed positive effects compared to two of six moderate risk-of-bias studies (33%). The strongest evidence for depression emerged from large, well-conducted RCTs with adequate randomization, allocation concealment, and low attrition (Feinberg et al., 2016; Zhao et al., 2017; Ngai et al., 2020; Ngai and Gao, 2022). Several null-effect studies for both depression and anxiety were characterized by substantial attrition (Kennedy et al., 2023; Shorey et al., 2023; Van der Meulen et al., 2023), which may have reduced statistical power to detect true effects.

The two Nigerian quasi-experimental studies (Onogwu et al., 2023; Jiang et al., 2024) reported effect sizes (ηp² = 0.41–0.49) approximately three times larger than those observed in the randomized trials. While these studies received favorable ratings on the JBI checklist for quasi-experimental designs, the absence of randomization, use of wait-list controls, and social media-based recruitment may have contributed to inflated effect estimates. Both studies were conducted by overlapping research teams using similar methodologies (WhatsApp-delivered interventions), limiting independent replication. These findings suggest that technology-delivered couples interventions may hold promise in low-resource settings, but the effect magnitudes require confirmation in randomized designs before informing practice recommendations.

6.7. Strengths and limitations

To our knowledge, this study provides the most comprehensive synthesis of couples-based interventions for perinatal depression and anxiety to date. The review followed rigorous methodological standards including a preregistered protocol, narrative synthesis, and PRISMA compliance. A novel contribution of this review is the preliminary D/F framework for classifying dyadic participation across design intent and achieved delivery. This framework enabled systematic comparison of the extent of dyadic design and delivery, an aspect not captured in prior reviews. While this framework provides a useful heuristic to organize and analyze heterogenous interventions, this should be treated as preliminary and be validated in future reviews. However, some limitations should be considered. Attendance thresholds (25%, 75%) pragmatically chosen may not represent true inflection points for partner engagement. Second, many studies did not report partner attendance with sufficient detail to enable precise classification, requiring inferential coding based on careful reading of narrative text, tables, etc. Future research should validate these categories and ideally, papers should provide granular information on partner attendance to make the process of synthesis easier. Inconsistent reporting of partner involvement across studies highlights the need for standardized reporting guidelines, potentially a CONSORT extension for couples-based trials. The inclusion of studies from 14 countries across six continents increases geographic representation, though limited by the glaring gaps of LMICs representation. By including both RCTs and quasi-experimental designs, our study balances methodological rigor with real-world applicability where RCTs may not always be feasible. Beyond maternal outcomes, we extracted data on paternal mental health, relationship quality, and infant outcomes, revealing variable effects across family systems. Our comprehensive search strategy from database inception through 2024 aimed to capture all relevant evidence. However, it is possible that our English-language criterion may have excluded some relevant research, particularly from Latin America and Francophone Africa, and recommend that future reviews consider including non-English publications where resources permit. Finally, we considered interventions addressing both antenatal and postnatal depression in line with the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013), which expanded the postpartum onset specifier for depression to incorporate the antenatal period.

Our narrative synthesis has some limitations. Substantial heterogeneity precluded meta-analysis. Interventions varied widely in theoretical orientation, intensity (single session to 20 sessions), delivery format, and timing across the perinatal period. Without meta-analysis, we could not establish causal relationships between intervention characteristics (e.g., partner attendance level, intervention content) and outcomes, as observed patterns may reflect unmeasured confounders such as population selection, baseline severity, intervention fidelity, or cultural context. While we reported effect sizes where available and examined effect direction regardless of statistical significance, not all studies provided effect sizes, limiting our ability to fully characterize effect magnitude across all studies. Additionally, our approach of noting statistical significance alongside effect sizes does not formally weight studies by precision. Consistent with Cochrane guidance cautioning against vote counting based solely on p-values (McKenzie and Brennan, 2024), we emphasize that our conclusions are grounded in pattern recognition across heterogeneous studies, considering effect direction, magnitude, intervention characteristics, and risk of bias, rather than in the proportion of statistically significant findings per se. Further, the patterns we observed regarding partner involvement and intervention content should therefore be interpreted as hypothesis-generating rather than definitive, requiring confirmation through formal meta-analysis and moderator testing.

While 52% of RCTs demonstrated low risk of bias, performance bias was universal due to inability to blind participants in behavioral interventions, and outcome assessor blinding was often unclear. Measurement inconsistency complicated synthesis: although 18 studies used EPDS for depression, eight used different measures, and anxiety assessment varied substantially, with only 12 of 26 studies measuring it despite high comorbidity between perinatal depression and anxiety (Feldman et al., 2025; Pilkington et al., 2015).

Of the 26 included studies, 10 (38%) were identified through citation analysis rather than database searches. Couples-based perinatal mental health interventions are methodologically and terminologically diffuse. MeSH and Emtree terms do not consistently capture such intervention categories, and cross-disciplinary publication patterns across psychiatry, obstetrics, psychology, nursing, and family studies journals may contribute to inconsistent indexing. The lack of standardized terminology for couples-based perinatal interventions is a well-documented challenge (Scroggins et al., 2024), with relevant studies indexed under variable terms (e.g., “partner-inclusive”, “dyadic”, “couples therapy”, “male involvement”, “family interventions”, “relationship education”). To mitigate these structural indexing limitations, we implemented citation searching as an a priori component of the search strategy rather than as a post-hoc addition. This approach, using forward citation searching of included studies and hand-searching reference lists of relevant systematic reviews, aligns with Methodological Expectations of Cochrane Intervention Reviews (MECIR) standards and Cochrane Handbook recommendations (Briscoe et al., 2020; Higgins et al., 2024). By explicitly reporting the proportion of studies identified through various search streams, we ensure a transparent and reproducible methodology that accounts for the unique indexing complexities of the field.

7. Conclusion

Couples-based interventions can effectively reduce perinatal depression and anxiety. In our narrative synthesis, we observed that effect magnitude varied across different levels of partner involvement, with both minimal and full attendance dyadic attendance showing positive outcomes in some studies. This likely suggests that intervention content, especially teaching partners specific support skills, might be important, though formal moderator testing is needed to confirm this hypothesis and rule out confounding by study quality, population characteristics, or implementation factors. The absence of mediation analyses across all studies prevents identification of mechanisms, while variable effects on partners and concentration of research in high-income settings limit generalizability. These findings should be interpreted as hypothesis-generating given the narrative synthesis approach and substantial heterogeneity across studies. Without meta-analysis and a formal moderator testing, we cannot definitively establish causal relationships between intervention characteristics and outcomes. Future research must prioritize mechanism testing, cultural adaptation for diverse contexts, and development of standardized reporting guidelines to advance this field beyond descriptive synthesis.

Appendix 1. Search Strategy

Date of search: December 10, 2024;

Table 1: Searches used for the different databases in our study.

Search strategy for PubMed.
	(“Depression, Postpartum”[Mesh] OR ((“Depressive Disorder”[Mesh] OR “Depression”[mesh] OR depression[tiab] OR depressed[tiab] OR depressive[tiab]) AND (“Pregnancy”[Mesh] OR “Pregnancy Complications”[Mesh:NoExp] OR “Pregnant Women”[Mesh] OR pregnant[tiab] OR pregnancy[tiab] OR perinatal [tiab] OR peripartum[tiab] OR antenatal[tiab] OR postpartum[tiab] OR post-partum[tiab] OR post partum[tiab]) OR expectant[tiab]) OR antenatal depression[tiab] OR prenatal depression[tiab] OR pre natal depression[tiab] OR pre-natal depression[tiab] OR perinatal depression[tiab] OR peri natal depression[tiab] OR puerperal depression[tiab] OR postnatal depression[tiab] OR post natal depression[tiab] OR post-natal depression[tiab] OR postpartum depression[tiab] OR post-partum depression[tiab] OR maternal mental health[tiab] OR maternal depression[tiab] OR maternal depressive symptoms[tiab] OR parent depression[tiab] OR maternal depressive symptoms[tiab] OR maternal postpartum depression[tiab] OR maternal postpartum depressive symptoms [tiab] OR maternal postnatal depression[tiab] OR maternal postnatal depressive symptoms[tiab] OR maternal anxiety[tiab] OR perinatal anxiety [tiab] OR peripartum anxiety[tiab] OR antepartum anxiety[tiab] OR antenatal anxiety[tiab] OR ante-natal anxiety [tiab] OR maternal well-being [tiab] OR maternal mood disorder [tiab] OR “maternal distress” [tiab:~3] OR “maternal wellbeing” [tiab:~3] OR “maternal psychological wellbeing” [tiab:~3])	46,442
2	“cognitive behavioral therapy”[MeSH Terms] OR “couples therapy”[MeSH Terms] OR counseling[tiab] OR couples counseling [tiab] OR cognitive behavior* treatment [tiab] OR CBT[tiab] OR interpersonal therapy[tiab] OR IPT[tiab] OR behavior* activation [tiab] OR behavioral couples therapy [tiab] OR psycho-education[tiab] OR psychoeducation*[tiab] OR problem solving therapy [tiab]	155,762
3	Father* [tiab] OR Mother* [tiab] OR marital couples[tiab] OR “marriage”[MeSH Terms] OR dyadic[tiab] OR dyad [tiab] OR “partner-inclusive” [tiab:~3] OR “partner inclusive” [tiab:~3] OR partner-support [tiab] OR partner support [tiab] OR husbands[tiab] OR partners [tiab] OR couple-focused [tiab] OR couples-focused[tiab] OR couple-based [tiab] OR couples-based[tiab] OR couple*[tiab] OR partner*[tiab] OR parent* [tiab] OR husbands[tiab] OR “engaging husbands” [tiab:~4]	1,470,621
4	#1 AND #2 AND #3
		1247
		RESULTS

EMBASE.

Search Strategy for EMBASE
	‘postnatal depression’/exp OR ((‘depression’/exp OR ‘depression’:ti,ab,kw OR ‘depressed’:ti,ab,kw OR ‘depressive’:ti,ab,kw) AND (‘pregnancy’/exp OR ‘pregnancy complication’/de OR ‘pregnant woman’/exp OR ‘pregnant’:ti,ab,kw OR ‘pregnancy’:ti,ab,kw OR ‘perinatal’:ti,ab,kw OR ‘peripartum’:ti,ab,kw OR ‘antenatal’:ti,ab,kw OR ‘postpartum’:ti,ab,kw OR ‘post-partum’:ti,ab,kw OR ‘post partum’:ti,ab,kw)) OR ‘expectant’:ti,ab,kw OR ‘antenatal depression’:ti,ab,kw OR ‘prenatal depression’:ti,ab,kw OR ‘pre natal depression’:ti,ab,kw OR ‘pre-natal depression’:ti,ab,kw OR ‘perinatal depression’:ti,ab,kw OR ‘peri natal depression’:ti,ab,kw OR ‘puerperal depression’:ti,ab,kw OR ‘postnatal depression’:ti,ab,kw OR ‘post natal depression’:ti,ab,kw OR ‘post-natal depression’:ti,ab, kw OR ‘postpartum depression’:ti,ab,kw OR ‘post-partum depression’:ti,ab,kw OR ‘maternal mental health’:ti,ab,kw OR ‘maternal depression’:ti,ab,kw OR ‘parent depression’:ti,ab,kw OR ‘maternal depressive symptoms’:ti,ab,kw OR ‘maternal postpartum depression’:ti,ab,kw OR ‘maternal postpartum depressive symptoms’:ti,ab,kw OR ‘maternal postnatal depression’:ti,ab,kw OR ‘maternal postnatal depressive symptoms’:ti,ab,kw OR ‘maternal anxiety’:ti,ab,kw OR ‘perinatal anxiety’:ti,ab,kw OR ‘peripartum anxiety’:ti,ab,kw OR ‘antepartum anxiety’:ti,ab,kw OR ‘antenatal anxiety’:ti,ab,kw OR ‘ante-natal anxiety’:ti,ab,kw OR ‘maternal well-being’:ti,ab,kw OR ‘maternal mood disorder’:ti,ab,kw OR ((‘maternal’ NEAR/4 ‘distress’):ti,ab,kw) OR ((‘maternal’ NEAR/4 ‘wellbeing’):ti,ab, kw) OR ((‘maternal’ NEAR/4 ‘psychological’ NEAR/4 ‘wellbeing’):ti,ab,kw)	68,068
2	‘cognitive behavioral therapy’/exp OR ‘couple therapy’/exp OR ‘counseling’:ti,ab,kw OR ‘couples counseling’:ti,ab,kw OR ‘cognitive behavior* treatment’:ti,ab,kw OR ‘cbt’:ti,ab,kw OR ‘interpersonal therapy’ OR ‘ipt’ OR ‘behavior* activation’:ti,ab,kw OR ‘behavioral couples therapy’:ti,ab,kw OR ‘psycho-education’:ti,ab,kw OR ‘psychoeducation*’:ti,ab,kw OR ‘problem solving therapy’:ti,ab,kw	201,850
3	‘father*’:ti,ab,kw OR ‘mother*’:ti,ab,kw OR ‘marital couples’:ti,ab,kw OR ‘marriage’/exp OR ‘marriage’ OR ‘dyadic’:ti,ab,kw OR ‘dyad’:ti,ab,kw OR ‘partner-inclusive’:ti,ab,kw OR ((‘partner’ NEAR/4 ‘inclusive’):ti,ab,kw) OR ‘partner-support’:ti,ab,kw OR ‘partner support’:ti,ab,kw OR ‘partners’:ti, ab,kw OR ‘couple-focused’:ti,ab,kw OR ‘couples-focused’:ti,ab,kw OR ‘couple-based’:ti,ab,kw OR ‘couples-based’:ti,ab,kw OR ‘couple*’:ti,ab,kw OR ‘partner*’:ti,ab,kw OR ‘parent*’:ti,ab,kw OR ‘husbands’:ti,ab,kw OR ((‘engaging’ NEAR/5 ‘husbands’):ti,ab,kw)	1,880,135
4	#1 AND #2 AND #3
		1434
		RESULTS

CINAHL via EBSCOhost.

Search Strategy for CINAHL via EBSCOhost
S1	MH depression, postpartum OR MH depressive disorder OR MH Depression OR TI depression OR AB depression	207005
S2	MH Pregnancy OR MM pregnancy complications OR MH pregnant women OR TI (pregnant OR pregnancy or perinatal or peripartum or antenatal or postpartum or post-partum or post partum or expectant) OR AB (pregnant OR pregnancy or perinatal or peripartum or antenatal or postpartum or post-partum or post partum or expectant)	321768
S3	S1 AND S2	9381
S4	TI (antenatal depression OR prenatal depression OR pre natal depression OR pre-natal depression OR perinatal depression OR peri natal depression OR puerperal depression OR postnatal depression OR post natal depression OR post-natal depression OR postpartum depression OR post- partum depression OR maternal depression OR maternal depressive symptoms OR OR parent depression OR maternal depressive symptoms OR maternal postpartum depression OR maternal postpartum depressive symptoms OR maternal postnatal depression OR maternal postnatal depressive symptoms OR maternal anxiety OR perinatal anxiety OR peripartum anxiety OR antepartum anxiety OR antenatal anxiety OR ante-natal anxiety OR maternal well-being OR maternal mood disorder OR maternal distress OR “maternal wellbeing” OR “maternal psychological wellbeing” OR “paternal anxiety” OR “paternal depression” OR “maternal mental health”) OR AB (antenatal depression OR prenatal depression OR pre natal depression OR pre-natal depression OR perinatal depression OR peri natal depression OR puerperal depression OR postnatal depression OR post natal depression OR post-natal depression OR postpartum depression OR post- partum depression OR maternal depression OR maternal depressive symptoms OR OR parent depression OR maternal depressive symptoms OR maternal postpartum depression OR maternal postpartum depressive symptoms OR maternal postnatal depression OR maternal postnatal depressive symptoms OR maternal anxiety OR perinatal anxiety OR peripartum anxiety OR antepartum anxiety OR antenatal anxiety OR ante-natal anxiety OR maternal well-being OR maternal mood disorder OR maternal distress OR “maternal wellbeing” OR “maternal psychological wellbeing” OR “paternal anxiety” OR “paternal depression” OR “maternal mental health”)	11,605
S5	S3 OR S4	15802
S6	MH (cognitive behavioral therapy or cbt or cognitive behavioural therapy) OR MH couples therapy OR TI counseling OR AB counseling OR TI couples counseling OR AB couples counseling OR TI (cognitive behavior* treatment or CBT OR behavioral activation OR behavioral couples therapy OR psycho-education OR psychoeducation OR problem solving) OR AB (cognitive behavior* treatment or CBT OR behavioral activation OR behavioral couples therapy OR psycho-education OR psychoeducation OR problem solving)	61827
S7	S6 AND S5	538
S8	TI ((Father* OR Mother* OR marital couples)) OR AB ((Father* OR Mother* OR marital couples)) OR MH marriage OR TI (dyadic OR dyad*) OR AB (dyadic OR dyad*) OR TI (partner-inclusive OR partner inclusive OR partner-support OR partner support OR husbands OR partners OR couple-focused OR couple-based OR couples-focused OR couples-based OR couple* OR partner* OR parent* OR husbands OR “engaging husbands”) OR AB (partner-inclusive OR partner inclusive OR partner-support OR partner support OR husbands OR partners OR couple-focused OR couple-based OR couples-focused OR couples-based OR couple* OR partner* OR parent* OR husbands OR “engaging husbands”)	390745
S9	S7 AND S8	332

PsycINFO via ProQuest.

Search strategy for PsycINFO via ProQuest
1	(mjsub(“Postpartum Depression”) OR (mjsub(“Depressive Disorder”) OR mjsub(“Depression”) OR tiab(depression) OR tiab(depressed) OR tiab(depressive)) AND (mjsub(“Pregnancy”) OR mainsubject(“Pregnancy Complications”) OR mjsub(“Pregnant Women”) OR tiab(pregnant) OR tiab(pregnancy) OR tiab(perinatal) OR tiab (peripartum) OR tiab(antenatal) OR tiab(postpartum) OR tiab(post-partum) OR tiab(post partum) OR tiab(expectant)) OR tiab(antenatal depression) OR tiab (“prenatal depression”) OR tiab(“pre natal depression”) OR tiab(“pre-natal depression”) OR tiab(perinatal depression) OR tiab(peri natal depression) OR tiab (puerperal depression) OR tiab(postnatal depression) OR tiab(post natal depression) OR tiab(post-natal depression) OR tiab(postpartum depression) OR tiab(post-partum depression) OR tiab(maternal mental health) OR tiab(maternal depression) OR tiab(maternal depressive symptoms) OR tiab(parent depression) OR tiab (maternal depressive symptoms) OR tiab(maternal postpartum depression) OR tiab(maternal postpartum depressive symptoms) OR tiab(maternal postnatal depression) OR tiab(maternal postnatal depressive symptoms) OR tiab(maternal anxiety) OR tiab(perinatal anxiety) OR tiab(peripartum anxiety) OR tiab (antepartum anxiety) OR tiab(antenatal anxiety) OR tiab(ante-natal anxiety) OR tiab(maternal well-being) OR tiab(maternal mood disorder) OR tiab(“maternal distress”) OR tiab(“maternal wellbeing”) OR tiab(“maternal psychological wellbeing”))	42,571
2	mjsub(“cognitive behavioral therapy”) OR mjsub(“couples therapy”) OR tiab(counseling) OR tiab(couples counseling) OR tiab(cognitive behavior* treatment) OR tiab(CBT) OR tiab(interpersonal therapy) OR tiab(IPT) OR tiab(behavior* activation) OR tiab(behavioral couples therapy) OR tiab(psycho-education) OR tiab (psychoeducation*) OR tiab(problem solving therapy)	178,159
3	tiab(Father*) OR tiab(Mother*) OR tiab(marital couples) OR mjsub(“marriage”) OR tiab(dyadic) OR tiab(dyad) OR tiab(“partner-inclusive”) OR tiab(“partner inclusive”) OR tiab(partner-support) OR tiab(partner support) OR tiab(husbands) OR tiab(partners) OR tiab(couple-focused) OR tiab(couples-focused) OR tiab (couple-based) OR tiab(couples-based) OR tiab(couple*) OR tiab(partner*) OR tiab(parent*) OR tiab(husbands) OR tiab(“engaging husbands”)	578,534
4	#1 AND #2 AND #3	1679

Appendix 2. PRISMA

PRISMA flow diagram illustrating the study identification, screening, eligibility assessment, and inclusion process. Page et al., 2021.

Appendix 3

Table 1.

Characteristic of studies in the systematic review (n = 26)

First author (Year)	Country	Income-level	Women’s sample size	Couple composition	Intervention Name/Approach	Study design	Intervention category*	Sample
Buist et al. (1999)	Australia	High-income	44	Heterosexual couples	Psychoeducational antenatal classes	RCT	Preventive	Primiparous, 12–24 weeks gestation, risk score ≥8 for postpartum depression
Misri et al. (2000)	Canada	High-income	29	Heterosexual couples (married or cohabiting)	Partner-Assisted Psychoeducational Treatment for PPD	2-arm RCT; women randomised to patient-only vs. patient + partner participation	Treatment	Met the DSM-IV criteria for major depression with postpartum onset. All women were married or cohabiting, and all had an EPDS ≥12
Matthey et al. (2004)	Australia	High-income	202	Heterosexual couples	Antenatal psychosocial “empathy” session	3-arm cluster randomized controlled trial with 3 conditions: usual service (‘control’), experimental (‘empathy’), or non-specific control (‘baby-play’).	Preventive	Couples expecting their first baby, attending evening Preparation for Parenthood program, in late second or early third trimester
Shapiro and Gottman (2005)	USA	High-income	38	Heterosexual couples	Bringing Baby Home Workshop: Relationship strengthening along with education and partner support	RCT with a waitlist control group	Preventive	Either expecting a baby or had a baby born within 3 months of first interview; married; both husband and wife older than 18 years
Milgrom et al. (2005)	Australia	High-income	192	Heterosexual couples	CBT-based postnatal psychosocial interventions	RCT (4 arms) - Arm 1: Group based CBT; Arm 2: Group-based counseling; Arm 3: Individual counseling and Arm 4: Control-Routine primary care (standard of care)	Treatment	DSM-IV diagnosis of depression; 37–42 week pregnancy; infant birth-weight ≥2.5 kg; no congenital abnormality or major health problem or concurrent major psychiatric disorder
Feinberg and Kan (2008)	USA	High-income	169	Heterosexual couples	Family Foundations: Psychosocial relationship strengthening program	RCT (no treatment control)	Mixed	Couples expecting first child, living together regardless of marital status, at least 18 years of age
Ortiz Collado et al. (2014)	Spain & France	High-income	184	Heterosexual couples	Psychosomatic Antenatal Programme (Tourné approach)	Multicentre RCT (2-arm parallel)	Preventive	Low-moderate SES; women ≤20 weeks pregnant, moderate-high PPD risk, ≤2 children, no psychiatric/medical/substance issues
Gambrel and Piercy (2015)	USA	High-income	34	Heterosexual couples (n = 31), lesbian couple (n = 1), adoptive couple (n = 1)	Mindful Transition to Parenthood (MTPP): Mindfulness-based relationship education	Randomized mixed-methods clinical trial with waitlist control	Mixed	Romantically involved couples expecting first child, >18 years, pregnant women 12–34 weeks gestation
Feinberg et al. (2016)	USA	High-income	399	All heterosexual	Family Foundations (FF) (expanded with childbirth education): Coparenting-focused relationship strengthening	RCT (attn control)	Preventive	Expectant first-time couples living together, expecting, both partners ≥18 years old
Fisher et al. (2016)	Australia	High-income	400	Predominantly heterosexual couples	What Were We Thinking (WWWT): Gender-informed psychoeducation with partner support and infant management	Cluster RCT (MCHC-level randomization; parallel groups)	Preventive	Primiparous women <6 weeks postpartum, receiving care at selected clinics
Shorey et al. (2017)	Singapore	High-income	250	Heterosexual couples only;	Home-but-not-Alone mobile health psychoeducation app with focus on parenting self-efficacy	2-arm RCT: mobile app + routine care vs. routine care only	Preventive	≥21 years, any parity, smart device ownership, Singapore-based first 4 weeks postpartum, healthy newborn discharged home
Zhao et al. (2017)	China (Shanghai)	Upper-middle-income	352	Heterosexual couples (first-time parents)	Couple-Separated Psychoeducation For First Time Parents	2-arm RCT: psychoeducation vs. usual care	Preventive	High-risk pregnancy (per scoring criteria), EPDS ≥9 or PDSS ≥60, <28 weeks at enrollment, primigravida, regular prenatal care, consented
Alipour et al. (2020)	Iran	Upper-middle-income	30	Heterosexual couples	Marital Communication Skills Training (Couple-focused)	RCT (parallel groups)	Mixed	Gestational age <24 weeks, no systemic diseases/psychotropic medications, monogamous relationship, low marital satisfaction
Feinberg et al. (2020)	USA	High-income	56	Heterosexual couples	Military Family Foundations (MFF): Military-adapted coparenting psychoeducational program	Randomized controlled trial (2-arm parallel)	Preventive	Expectant first-time couples living together, expecting, both partners ≥18 years old, at least one partner in military and not assigned to deploy in next 6 months
Ngai et al. (2020)	Hong Kong	High-income	388	Heterosexual couples only; first-time parents;	Couple-Based Cognitive Behavioral Intervention (CBI)	Multisite 3-arm RCT: couple-based CBT vs. women-only CBT vs. standard care	Preventive	Couples >18 years, primiparous, uncomplicated singleton pregnancy, ≥2nd trimester, Chinese literate, Hong Kong residents
Kuo et al. (2022)	Taiwan	High-income	106	Heterosexual couples (partners invited; majority participated)	Integrated Childbirth Education + Mindfulness	2-arm, parallel, single-blind RCT	Preventive	Fear of childbirth VAS ≥7, ≥20 years, singleton pregnancy 12–24 weeks, no obstetric/medical complications, willing to attend classes, planned study hospital delivery
McHale et al. (2022)	USA	High-income	138	Unmarried, predominantly Black heterosexual dyads	Figuring It Out for the Child (FIOC):Prenatal coparenting consultation	RCT 2-arm with treatment as usual control	Preventive	≥1 parent Black-identified, first child together, unmarried, both parents consenting, mothers ≥2nd trimester
Ngai and Gao (2022)	Hong Kong & Mainland China (Guangzhou)	High-income	455	Heterosexual couples only; first-time parents; cohabiting; no psychiatric history	Couple-Based IPT	2-arm RCT: couple-based IPT vs. standard care	Preventive	Couples ≥18 years, primiparous, 12–30 weeks uncomplicated singleton pregnancy, Chinese literate, no psychiatric disorders
O’Mahen et al. (2022)	UK	High-income	114	Mostly heterosexual couples	ACORN (Adapting and testing a brief intervention to reduce maternal anxiety during pregnancy): Perinatal-adapted CBT with mindfulness	Feasibility RCT (1:1; ACORN + TAU vs TAU)	Treatment	Nulliparous pregnant individuals ≥18 years, entering second trimester, GAD-7 ≥7 (top quartile), partners invited
Canfield et al. (2023)	USA	High-income	30	Heterosexual cohabiting couples	Mothers and Babies Online Course (eMB): CBT-based psychoeducation	Feasibility RCT (1:1)	Preventive	Pregnant 13–30 weeks, PHQ-9 or GAD-7 ≥10, adult partner, cohabitating, Missouri residents with internet
Kennedy et al. (2023)	UK	High-income	60	Heterosexual couples	Tavistock First-Time Parent Study – Brief Couple-Focused Perinatal Intervention	Pilot RCT (parallel groups) with standard of care	Preventive	Married/cohabiting couples expecting first baby, no serious medical/psychiatric complications, both ≥18 years, receiving care at Royal Free Hospital London
Onogwu et al. (2023)	Nigeria	Lower-middle-income	303	Heterosexual couples	Drama Therapy + Family Counseling	Quasi-experimental; pre-test/post-test with control group; 6-month follow-up	Treatment	Couples whose female partners had given birth within 4 weeks
Shorey et al. (2023)	Singapore	High-income	400	Heterosexual, married, low-risk pregnancy, healthy newborns	Supportive Parenting App (SPA): App-based psychoeducation with peer support	2-arm RCT: SPA + standard care vs. standard care only	Preventive	Both parents ≥21 years, low-risk pregnancy >24 weeks gestation, smartphone/internet access
Van der Meulen et al. (2023)	Netherlands	High-income	141	Heterosexual couples only; married/cohabiting; Dutch-speaking	Mindfulness-Based Childbirth & Parenting (MBCP)	2-arm RCT (active comparator): MBCP vs. Enhanced Care as Usual (ECAU)	Preventive	Pregnant women with high fear of childbirth, 16–26 weeks gestational age
Burgio et al. (2024)	Italy	High-income	84	Heterosexual couples (Pregnant woman (primary), with partner participation)	Nine-session Prenatal–Postnatal Psychological Counseling for High-Risk Pregnancy	Quasi-experimental design with 2 groups (experimental and control group)	Preventive	High-risk pregnancy diagnosis, Italian literate for assessment
Jiang et al. (2024)	Nigeria	Lower-middle-income	313 families (women count not isolated; ~313 women)	Heterosexual couples	Social-media Dance/Storytelling/Visual Art Therapies	Quasi-experimental; pre-test/post-test with control group	Treatment	Families where women were within first four weeks of delivery

^*Treatment (intervention for symptomatic/diagnosed individuals); Prevention (intervention for asymptomatic individuals to prevent onset); Mixed (includes both symptomatic and asymptomatic participants or universal prevention regardless of risk status).

Table 2.

Intervention characteristics and primary outcomes (N = 26)

Study	Core content of sessions	Delivery Format	Setting	Int. timing	Dyadic design intent	Dyadic delivery format	Delivery agent	Intervention intensity	Key primary outcomes
Buist et al. (1999)	Physical preparation for parenthood Emotional coping strategies (pain, uncertainty, “good-enough” parenting) Didactic teaching with interactive group work, films, experiential exercises Postnatal debriefing and problem-solving	Group (Couples)	Hospital antenatal clinic	Antenatal (12–24 weeks gestation) plus two postpartum booster sessions (4–8 weeks and 8–10 weeks postpartum)	D3: Dyadic participation encouraged	F3: Full dyadic attendance	Midwife + psychologist/nurse teams with experience in mother-baby units	10 sessions total: 8 antenatal (weekly)– 2 postpartum boosters., Estimated session length 1.5–2 h → High intensity	Depression: No significant differences in depression scores (BDI or EPDS) between intervention and control groups at any time poin Anxiety: Trait anxiety significantly decreased in the intervention group from pre-class to 6 weeks postpartum (p = 0.036), but not in controls State anxiety showed no significant changes
Misri et al. (2000)	Seven psychoeducational visits. Both arms received same maternal content; support arm had partner present for 4/7 visits. Content: illness education, coping strategies, infant care division, communication, partner’s supportive role	Individual (woman alone) vs. dyad (woman + partner) depending on arm	2 university hospital clinics	Postnatal (≤6 weeks postpartum at enrolment)	D3: Dyadic participation encouraged	F3: Full dyadic attendance	Psychiatrist/mental-health clinician with perinatal expertise	~6 h direct contact with partner (in support arm) across 4 visits; ~10.5 h maternal contact time total - Moderate intensity	Depression: Support group showed significant positive effects on depression. While the results at Visit 6 showed trends favoring the intervention group (EPDS: control 14.6 vs support 11.4, p = 0.20), by Visit 7 there was a statistically significant difference with the support group showing substantially lower depression scores (EPDS: control 14.7 vs support 8.6, p = 0.013).
Matthey et al. (2004)	One 2-h session on postpartum concerns; gender-specific small-group discussions; partner dyadic discussion of concerns; problem-solving of hypothetical early postpartum scenarios; handout of strategies; booster mail-outs	In-person group with gender breakout and couple discussion	Hospital antenatal classes	Prenatal & Postnatal	D3: Dyadic participation encouraged.	F4: Dyadic + individual delivery	Clinical psychologist + social worker	Moderate (7 × 2 h + mail-outs)	Depression (EPDS - Edinburgh Postnatal Depression Scale) 6 weeks pp: Significant interaction effect between condition (Empathy) and self-esteem (p < 0.01). No significant differences for medium or high self-esteem women 6 months pp: No significant interaction effects; all groups improved to similar levels
Shapiro and Gottman (2005)	Communication, conflict management, relationship maintenance, shared meaning, infant development, father involvement through 18 structured couples’ exercises	In person Group (Couples)	Hospital-affiliated community program	Late pregnancy or ≤ 3 mo postpartum	D3: Dyadic participation encouraged.	F3:Full dyadic attendance	Clinical psychologist	High (12–15 h)	Postpartum Depression (SCL-90) Significant linear Time × Group effect (F(1,32) = 5.83, p = 0.022) Wives: Depression increased quadratically in control group and decreased quadratically in workshop group [initial worsening from pre-assessment to 3 months, then significant improvement from 3 months to 1 year]; significant difference at 1 year (t(32) = 2.13, p < 0.05).
Milgrom et al. (2005)	9 weekly 90-min sessions for mothers + 3 partner sessions; modules on mood, life events, problem solving; structured manuals.	Hospital outpatient clinic	Clinic	Postpartum (6–18 wk)	D2: Supportive partner role	F2: Partial attendance	Senior psychologist + co-therapists (psychologist, nurse, social worker)	Moderate (12 × 90 min)	Postpartum depression (Beck Depression Inventory-BDI) - All psychological interventions superior to routine care by ~7 points at 18–30 weeks postpartum (p = 0.005). Individual counseling best, 3–5 points better than group formats (p = 0.02). No differences between CBT vs. counseling (p = 0.97) or at 1-year follow-up. Anxiety (Beck Anxiety Inventory - BAI), Psychological interventions reduced anxiety by ~8 points vs. routine care at 18–30 weeks postpartum (p < 0.01). No differences between intervention types (CBT vs. counseling: p = 0.99; individual vs. group: p = 0.34). All active treatments equally effective. No differences at 1 year.
Feinberg and Kan (2008)	Emotional self-management; conflict management; problem solving; communication; mutual support; division of labor; fostering coparental support	Group (Couples): Multiple couples attend together in a group format.	Clinical-Hospital childbirth education departments	Prenatal & postnatal	D3: Dyadic participation encouraged.	F2: Partial attendance	Trained male-female facilitator team (childbirth educators)	Moderate (8 sessions, ~1.5–2 h each)	Maternal depression (CES-D): Significant reduction in depression (intervention effect) from pregnancy to 6 months postpartum - Effect size: 0.56 - moderate effect (B = −0.195, p < 0.01) Maternal Anxiety (Taylor Manifest Anxiety Scale): Significant reduction (B = −1.218, p < 0.05; effect size = 0.38 - small to moderate effect)
Ortiz Collado et al. (2014)	Body sensation awareness & normalization; individualized childbirth model; attachment/bonding; partner-specific goals; interactive discussion (60%), practical exercises (40%); follow-up phone calls	In person Group (Joint session for couples)	Hospital antenatal clinics	2nd trimester start → prior to delivery	D3: Dyadic participation encouraged	F3:Full dyadic attendance	Trained nurse-midwives + 1 session by intervention designer	High intensity (10 sessions × 135 min = 22.5 h contact)	Maternal depression (EPDS ≥ 12): No significant reduction in PPD risk (34.3% vs 45.5%, p = 0.26). Significant within-group mean score reduction (11.23 to 9.34, p = 0.01) but non-significant between-group difference (1.76 points, p = 0.08). Primary outcome not achieved.
Gambrel and Piercy (2015)	Week 1: Self (body scan, mindful eating, love, eye contact). Week 2: Partner (breathing, time-out, mindful hug). Week 3: Relationship (open awareness, communication, needs/dreams). Week 4: Family (loving-kindness, parenting, chores, baby cues)	Group (couples)	Holistic health centers, community spaces, birthing centers	Prenatal (12–34 wk gestation)	D3: Dyadic participation encouraged	F3:Full dyadic attendance.	Licensed Marriage and Family Therapist (first author)	4 × 2-hr sessions + 15 min daily practice (≈360 min) Low intensity	Depression/Anxiety/Stress (Combined measure - DASS-21) Not directly measured as separate outcomes Negative emotional states (combined): No effect on maternal depression/anxiety (η2partial = 0.00)
Feinberg et al. (2016)	Coparental conflict resolution & problem solving; communication; mutual support strategies; childbirth education content; group discussion; couple and group skill-building (including role-plays); viewing of video vignettes.	Group (multiple couples)	Hospital or community settings linked to maternity care	Prenatal start with follow-up postnatal	D3: Dyadic participation encouraged.	F3:Full dyadic attendance.	Male-female facilitator teams	High intensity	Depression (CES-D Scale): Combined (Mothers & Fathers): Significant reduction (b = −0.06, p < 0.05, d = 0.20) at 10 months postpartum, Combined parents: Significant reduction at 10 months postpartum (b = −0.06, p < 0.05, d = 0.20). No moderation by parent sex - mothers and fathers benefited equally Anxiety (State-Trait Anxiety Inventory):, Combined: No significant main effect (b = −0.73, ns)
Fisher et al. (2016)	(1) WWWT-trained nurse care; (2) Illustrated materials; (3) 6h couples’ seminar at 6–8 wks pp - About Babies: temperament, crying, cues, routines, settling; About Parents: expectations, role renegotiation, equitable unpaid work, communication, losses/gains, family patterns, support; active learning with discussion/practice	In person group (up to 5 couples)	Community Maternal & Child Health Centers (Victoria)	Postnatal (start 6–8 wks pp)	D2: Supportive partner role	F2: Partial attendance	MCH nurses trained in WWWT	Low intensity (~6 h seminar + brief reinforcement in routine care)	Depression (PHQ-9): Mothers: No significant reduction in continuous scores (adjusted mean difference −0.47, 95% CI −1.02 to 0.09), Depression severity categories: No significant difference for moderate/severe symptoms Anxiety (GAD-7 or Generalized Anxiety Disorder 7): Mothers: Significant reduction in mild-moderate symptoms (AOR 0.58, p < 0.05, 95% CI 0.35–0.97), Anxiety severity: Lower prevalence of scores 5–9 in intervention group
Shorey et al. (2017)	App-based delivery of parenting PDFs/videos, audios, push notifications, asynchronous chat with healthcare professionals, peer forum	Individual use + asynchronous dyadic engagement	Online (home-based via app); hospital discharge day enrollment	Postnatal start (day of discharge); 4-week intervention period	D3: Dyadic participation encouraged.	F4: Dyadic + individual delivery	Midwife (asynchronous support via app); content developed by nursing faculty and pilot-tested	Moderate: continuous access over 4 weeks; no fixed session count	Depression (Edinburgh Postnatal Depression Scale): Mothers: Mean difference of depression at 4-week followup postpartum = −0.69 (95% CI: −1.66 to 0.29), p = 0.167 - NOT significant.
Zhao et al. (2017)	6 sessions: 5 for women (mental health, complications, postpartum depression, family conflict, delivery adjustment); 1 for husbands (support, symptom recognition)	Group sessions in-person (women-only for 5 sessions; men-only for 1 session)	Hospital-based antenatal clinic	Mid-to-late pregnancy (≤28 weeks gestation start)	D2: D2: Supportive partner role [partner involvement limited to one session]	F1: Minimal (1–2 sessions)	Trained facilitators (15–18 per session); supervised by lead researcher	Moderate: 6 sessions total (~9 h contact time)	Perinatal depression (EPDS) No depression: 77.8% intervention vs 52.7% control. Minor depression: 14.4% vs 37.1%., Major depression: 7.8% vs 10.2% (p < 0.001). Intervention significantly reduced depression rates across all severity levels.
Alipour et al. (2020)	Trust/safety, pregnancy changes, communication (listening/speaking), self-awareness, problem management (thematic/personal/relational), message exchange, emotional expression, conflict/stress/anger management, mindfulness, expectations, pregnancy adaptations	In person - Group (Couples)	Public health centers	Mid-pregnancy (≤24 wks gestation)	D3:Dyadic participation encouraged	F3:Full dyadic attendance	Clinical psychologist (session lead) + health centre staff support	Moderate intensity (7 × 120 min = 14 h contact)	Depression (General Health Questionnaire): Mothers had significant reduction in depression at 3 months, 2.88 → 0.74 (Int) vs 4.40 → 3.81 (Ctrl); p = 0.001 (significant difference between groups at 3 months) Anxiety (General Health Questionnaire): Mothers also had significant reduction in anxiety at 3 months, 8.33 → 3.25 (Int) vs 7.74 → 6.77 (Ctrl); p = 0.001 (significant difference between groups at 3 months)
Feinberg et al. (2020)	Eight modules (5 prenatal, 3 postnatal): coparenting support/cohesion, expectations, communication/problem-solving, infant development, parenting efficacy, military adaptations (deployment/training)	Couple self-paced online (joint exercises within modules)	Home-based (online)	Antenatal start (approx. 24 wks gestation) and continued to postpartum	D3: Dyadic participation encouraged	F3: Full dyadic attendance	Program developers/adaptation team; asynchronous online delivery	Moderate intensity (~8 modules; contact hours variable, ~8–10 h est.)	Maternal depression (CES-D) saw significant reduction (ES = 0.51; p < 0.05). Intervention: M = 0.22 (SD = 0.18) vs Control: M = 0.43 (SD = 0.46).
Ngai et al. (2020)	3-h antenatal group session + two 30-min postnatal phone follow-ups; covered PPD psychoeducation, cognitive restructuring, problem-solving, communication skills; partner engagement in joint and individual tasks	Group (antenatal) + individual phone follow-ups at 2 and 4 weeks pp	Public hospital antenatal clinics (Hong Kong); phone follow-ups postnatally	Prenatal (2nd/3rd trimester) group sessions of 10 pairs of couples + 30-min phone follow-ups at 2 and 4 weeks pp	D4: Dyadic + individual components	F4: Dyadic + individual delivery	Trained midwife (CBT-trained); supervised by clinical psychologist	Low intensity: 3 sessions total (~4.5 h contact time)	Maternal depression (EPDS): 6 weeks: Significantly lower EPDS scores in couple-based vs control (difference 1.71, 95% CI 0.29–3.13, p = 0.01, Cohen’s d = 0.30) and vs women-alone (difference 1.46, 95% CI 0.11–2.81, p = 0.03, Cohen’s d = 0.26) 6 months: No significant differences (p = 0.97) 12 months: No significant differences (p = 0.97)
Kuo et al. (2022)	Childbirth education (physiological/psychological changes, labour/pain relief, breastfeeding, newborn/postpartum care) + simulation/practice (ice pain, cervical demo, equipment tour, birthing ball) + mindfulness (breathing, body scan, stretching, meditation, yoga/walking/eating, communication) + partner support	In person group classes	Teaching hospital (Northern Taiwan)	Prenatal (12–36 wks’ gestational age)	D3: Dyadic participation encouraged	F3: Full dyadic attendance	Experienced childbirth practitioner + certified mindfulness instructor	Moderate intensity (8 weeks × 2 h/week = 16 h total (8 h childbirth ed, 8 h mindfulness)	Depression (EPDS):, 36 weeks: MD = −3.7 (95% CI −5.1 to −2.3, p < 0.001, ES = 0.58), 1 week postpartum: MD = −3.5 (95% CI −5.2 to −1.8, p < 0.001, ES = 0.48) Anxiety (State-Trait Anxiety Inventory):, 36 weeks: MD = −7.4 (95% CI −10.3 to −4.5, p < 0.001, ES = 0.76), 1 week postpartum: MD = −6.4 (95% CI −10.7 to −2.1, p < 0.01, ES = 0.62)
McHale et al. (2022)	Coparenting awareness; communication & teamwork skills; guided problem-solving; intergenerational legacies	In person dyadic mentor–parent session	Community-based	Prenatal	D4: Dyadic + individual components	F4: Dyadic + individual delivery	Male–female community mentor teams (midwife + psych/nurse)	9 total contacts (7 prenatal + 1 booster + 1 preludes)	Depression (EPDS): Main effect of timepoint only (β = −2.41, p < 0.05). Both intervention and control groups showed similar reductions from baseline to 3 months postpartum. No intervention effect.
Ngai and Gao (2022)	3 weekly 2-h antenatal sessions + two 30-min postnatal phone follow-ups; covered role transition, conflict resolution, communication, social support; used IPT techniques (role-play, communication analysis)	Group (antenatal) + individual phone follow-ups	Public hospital antenatal clinics (Hong Kong & Guangzhou); phone follow-ups postnatally	Antenatal start (12–30 weeks GA) + postnatal follow-up (≤ 4 weeks postpartum)	D3: Dyadic participation encouraged	F4: Dyadic + individual delivery	Midwives with master’s in family counseling; trained in IPT (20 h); supervised by research team	Moderate: 5 sessions total (~7.5 h contact time)	Maternal depression (EPDS): Significant reduction in symptoms at 6 weeks (MD = 1.91, 95%CI 1.11–2.71, p < 0.001, d = 0.49) and 6 months (MD = 0.89, 95%CI 0.04–1.74, p = 0.039, d = 0.23). Depression risk (EPDS>9): 25.4% IPT vs 47.3% control at 6 weeks (p < 0.001); no significant difference at 6 months. Symptom benefits persisted but protective effect against clinical depression weakened, suggesting need for booster sessions
O’Mahen et al. (2022)	Session 1: Psychoeducation on pregnancy anxiety, self-care (activity scheduling, mindfulness/compassion) Session 2: Problem-solving pregnancy-specific worries, avoidance, partner reflective listening (partners only), distress-tolerance, loving-kindness to fetus Session 3:Communication/problem-solving with partner, coping together, relapse prevention	Group sessions in-person within/alongside routine antenatal classes	NHS maternity settings (hospital/community)	Prenatal (12 wk) → postpartum (6–34 wk gest)	D2: Supportive partner role	F2: Partial attendance	Midwife + psychological practitioner (cofacilitated)	3 sessions × 90 min; spaced 3 wks apart; homework between (Low intensity)	EPDS (depression): No intervention effect on depression, 10 weeks: Both groups similar (~10.4), no intervention effect accounting for baseline, 18 weeks: Both groups similar (~9.4), no intervention effect, 34 weeks: Both groups similar (~8.1), no intervention effect Anxiety (GAD-7): Marginal evidence of short-term anxiety reduction that wasn’t sustained. Trend toward reduction, d = 0.42 at 10 weeks (p = 0.06), 10 weeks: Treatment 6.40 vs TAU 7.58 (nonsignificant trend, p = 0.06), 18 weeks: No significant difference, 34 weeks: No significant difference
Canfield et al. (2023)	Cognitive restructuring; pleasant activities; social contact; thoughts and mood; future planning; relaxation exercises; fighting harmful and increasing helpful thoughts; coping skills; mindfulness; Missouri perinatal + COVID resources	Self-administered via Qualtrics	Virtual/remote	Antenatal (mid–late; 13–30 wks GA)	D3: Dyadic participation encouraged.	F2: Partial attendance.	Self-administered platform content	8 modules; ~1/week recommended; no participant completed all; mean logins pregnant 4.17, partner 3.44 = Low intensity	Maternal depression (EPDS): Intervention group decreased from M = 16.40 to M = 10.00 (change: −6.40); Control decreased from M = 14.13 to M = 11.36 (change: −2.77). Significant within-group improvements for intervention mothers at T2 (mid-intervention at 4 weeks) and T3 (postintervention at 8 weeks) (p < 0.05), but between-group differences not statistically significant. Maternal anxiety (GAD-7): Intervention decreased from M = 14.80 to M = 9.00; Control from M = 13.07 to M = 9.93. Significant between-group difference at T2 (MD = −4.05, 95% CI [0.82, 7.27], p = 0.02) favoring intervention.
Kennedy et al. (2023)	Explore feelings about transition to parenthood; impact of relationship on parenting; managing “triangle” dynamics; emotional health; anxiety/conflict as normal; challenging idealized views; improving communication; constructive conflict strategies	In person - couple	Hospital clinic or home visits	~30 wks GA (antenatal) → 20 wks pp	D3: Dyadic participation encouraged.	F3:Full dyadic attendance	Senior couple psychotherapists	Low intensity (~5 h total)	Maternal depression (EPDS): Significant decrease over time (27% reduction), but no difference between the groups. Depression measured using HADS (Hospital Anxiety and Depression Scale-Depression): Significant 36% increase across the study period in both groups (d = 0.37, p < 0.001) but no difference between groups, Anxiety (Hospital Anxiety and Depression Scale-Anxiety): No significant changes in either group.
Onogwu et al. (2023)	Drama therapy: Metaphor, role-play, improvisation, games, visualization, storytelling, relaxation; 4 × 45min dramas (4 repeats), group discussion; WhatsApp delivery. Family counseling: Storytelling, dramas on spousal support, discussion, 3h support session, writing exercises; WhatsApp delivery	Online (WhatsApp)	Online	Postpartum (≤4 wk); Intervention over 10 weeks and follow-up at 6m	D3: Dyadic participation encouraged	F3: Full dyadic attendance	Drama + family therapists	20 × 2 h sessions over 10 weeks per arm; high intensity	Maternal depression (EPDS): Drama alone: 18.8 → 14.1→12.0 at baseline/post/6-month (p = 0.001). Drama + spousal support: 19.6 → 10.1→5.5 (p = 0.001). Control: 19.3 → 16.2→14.2. Greatest gains with spousal support; paternal EPDS improved in both intervention arms; effects sustained at 6 months only when partners received counseling
Shorey et al. (2023)	Perinatal knowledge modules, peer support, forums, FAQs, expert advice	Individual use + optional peer chat and forums	Online via home-based app	Prenatal start (>24 weeks gestation) through 6 months postpartum	D3: Dyadic participation encouraged	F4: Dyadic + individual delivery	App-based content + trained peer volunteers (experienced mothers)	Moderate: continuous access over ~6 months; no fixed session count	Maternal depression (EPDS): No significant between-group differences at any timepoint (baseline-12 months). Both groups: ~6–8 range throughout. At 1 month: 33% scored >9 (possible depression). Control worsened 9–12 months while intervention decreased slightly (NS). Maternal anxiety (State-Trait Anxiety Inventory): No significant differences. Mothers scored worse than fathers on depression (ES = 2.58) and anxiety (ES = 6.45), suggesting need for more intensive support than app provided
Van der Meulen et al. (2023)	Mindfulness meditation, childbirth & parenting education, home practice	In person group-based, 3-h weekly sessions for 9 weeks	Midwifery care settings	Prenatal start (16–26 weeks gestation); ends shortly after birth	D3: Dyadic participation encouraged.	F3: Full dyadic attendance	Midwifery clinics and academic centers	High: 9 weekly 3-h sessions + daily home practice (30 min/day)	Depression (EPDS): No sig. differences between MBCP and ECAU groups at any timepoint. Both groups started with mean scores ~8–10 at baseline. Scores remained relatively stable (6–8 range) across all follow-ups. No significant time × condition interaction effect Maternal anxiety (Anxiety (DASS-21 Combined Score - Depression, Anxiety, Stress Scale): No sig differences between groups. Baseline: MBCP 35.72 (SD 17.61) vs ECAU 40.25 (SD 21.03). Significant decrease over time in BOTH groups but no superior impact of MBCP over ECAU
Burgio et al. (2024)	Three phases × 3 sessions: emotional support, meaning-making, coping skills, prenatal attachment, parenting adaptation, birth trauma support, distress management, COVID-19 coping	In person + phone- Group sessions (Multiple couples per group)	High-risk pregnancy clinic	Antenatal start (~20–24 wks GA) through 3 mo postpartum	D1: Optional partner involvement	F2: Partial attendance	Psychologists	Low intensity (9 × 50 min = 7.5 h total)	Maternal depression (Beck Depression Inventory): Significant time effect across groups (F(2,82) = 25.4, p < 0.001, η2p = 0.237), but no significant between-group differences. Mean BDI scores decreased over time in both groups.
Jiang et al. (2024)	Dance: view/comment videos, counseling. Storytelling: share postpartum narratives, discuss morals. Visual Art: create/share drawings reflecting feeling	Asynchronous & synchronous group interactions	Online (social media; WhatsApp)	Postpartum (≤4 wk start), symptom-reduction focus	D3:Dyadic participation encouraged	F3: Full dyadic attendance	Dance/art/, story therapists	Moderate (~40 h over 10 wks)	Maternal depression (EPDS): Art therapy and storytelling were most effective. At follow-up (6m), treatment groups maintained lower scores (Art: 8.4, Dance: 8.1, Storytelling: 7.2) vs control (15.1), p = 0.03, η2p = 0.492 (Very large effect sizes)

Note: For Dyadic design intent: D0 = Woman only; D1 = Partner optional; D2 = Partner supportive role; D3 = Full couple design; D4 = Combined joint and separate sessions.

For Dyadic delivery format: F0 = No partner participation; F1 = Minimal (1–2 sessions); F2 = Partial (25–75% attendance); F3 = Full joint attendance; F4 = Combined joint and separate delivery.

Appendix 4. Risk of Bias Assessment

Risk of bias analysis for all studies.

Study	Study design	Total points	Max points	Percentage	Risk Level
Buist et al., 1999	RCT	16	26	62%	Moderate
Misri et al., 2000	RCT	18	26	69%	Moderate
Matthey et al., 2004	RCT	21	26	81%	Low
Shapiro and Gottman, 2005	RCT	18	26	69%	Moderate
Milgrom et al., 2005	RCT	13	26	50%	Moderate
Feinberg and Kan, 2008	RCT	20	26	77%	Low
Ortiz Collado et al., 2014	RCT	23	26	88%	Low
Gambrel and Piercy, 2015	RCT	18	26	69%	Moderate
Fisher et al., 2016	RCT	22	26	85%	Low
Feinberg et al., 2016	RCT	22	26	85%	Low
Shorey et al., 2017	RCT	22	26	85%	Low
Zhao et al., 2017	RCT	21	26	81%	Low
Alipour et al., 2020	RCT	21	26	81%	Low
Feinberg et al., 2020	RCT	16	26	62%	Moderate
Ngai et al., 2020	RCT	21	26	81%	Low
Kuo et al., 2022	RCT	20	26	77%	Low
McHale et al., 2022	RCT	19	26	73%	Moderate
Ngai and Gao, 2022	RCT	22	26	85%	Low
O’Mahen 2022	RCT	19	26	73%	Moderate
Canfield et al., 2023	RCT	20	26	77%	Low
Kennedy et al., 2023	RCT	19	26	73%	Moderate
Shorey et al., 2023	RCT	19	26	73%	Moderate
Van der Meulen et al., 2023	RCT	19	26	73%	Moderate
Burgio et al., 2024	Quasi-experimental	16	18	89%	Low
Onugwu 2023	Quasi-Experimental	14	18	78%	Low
Jiang et al., 2024	Quasi-Experimental	18	18	100%	Low

Risk of bias by study design.

Study Design	Low Risk	Moderate Risk	High Risk	Total
RCT	12	11	0	23
Quasi-Experimental	3	0	0	3
Total	15	11	0	26

Detailed risk of bias analysis by each study.

	Buist et al., 1999	Misri et al., 2000	Matthey et al., 2004	Shapiro and Gottman, 2005	Milgrom et al., 2005	Feinberg and Kan, 2008	Ortiz Collado et al., 2014
1. Was true randomization used for assignment of participants to treatment groups	1	1	2	1	0	2	2
2. Was allocation to treatment groups concealed	1	1	1	1	1	1	2
3. Were treatment groups similar at the baseline	2	2	2	2	1	2	2
4. Were participants blind to treatment assignment	0	0	2	0	0	0	2
5. Were those delivering the treatment blind to treatment assignment?	0	0	0	0	0	0	0
6. Were treatment groups treated identically other than the intervention of interest?	2	2	2	2	0	2	2
7. Were outcome assessors blind to treatment assignment?	1	0	2	1	1	1	2
8. Were outcomes measured in the same way for treatment groups?	2	2	2	2	2	2	2
9. Were outcomes measured in a reliable way?	2	2	2	2	2	2	2
10. Was follow up complete and if not, were differences between groups in terms of their follow up adequately described and analysed?	0	2	2	1	1	2	1
11. Were participants analysed in the groups to which they were randomized?	1	2	0	2	2	2	2
12. Was appropriate statistical analysis used?	2	2	2	2	2	2	2
13. Was the trial design appropriate and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for in the conduct and analysis of the trial?	2	2	2	2	1	2	2
Total Risk of Bias	16	18	21	18	13	20	23
	Moderate	Moderate	Low	Moderate	Moderate	Low	Low

	Gambrel and Piercy, 2015	Feinberg et al., 2016	Fisher et al., 2016	Shorey et al., 2017	Zhao et al., 2017	Alipour et al., 2020
1. Was true randomization used for assignment of participants to treatment groups	1	2	2	2	2	2
2. Was allocation to treatment groups concealed	0	2	2	2	1	2
3. Were treatment groups similar at the baseline	2	2	2	2	2	2
4. Were participants blind to treatment assignment	0	0	0	0	0	0
5. Were those delivering the treatment blind to treatment assignment?	0	0	0	0	0	0
6. Were treatment groups treated identically other than the intervention of interest?	2	2	2	2	2	2
7. Were outcome assessors blind to treatment assignment?	1	2	2	2	2	2
8. Were outcomes measured in the same way for treatment groups?	2	2	2	2	2	2
9. Were outcomes measured in a reliable way	2	2	2	2	2	2
10. Was follow up complete and if not, were differences between groups in terms of their follow up adequately described and analysed?	2	2	2	2	2	2
11. Were participants analysed in the groups to which they were randomized?	2	2	2	2	2	1
12. Was appropriate statistical analysis used?	2	2	2	2	2	2
13. Was the trial design appropriate and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for in the conduct and analysis of the trial?	2	2	2	2	2	2
Total Risk of Bias	18	22	22	22	21	21
	Moderate	Low	Low	Low	Low	Low

	Feinberg et al., 2020	Ngai et al., 2020	Kuo et al., 2022	McHale et al., 2022	Ngai and Gao, 2022	O’ Mahen et al., 2022
1. Was true randomization used for assignment of participants to treatment groups	2	2	2	2	2	2
2. Was allocation to treatment groups concealed	1	2	2	1	2	2
3. Were treatment groups similar at the baseline	2	2	2	2	2	0
4. Were participants blind to treatment assignment	0	0	0	0	0	0
5. Were those delivering the treatment blind to treatment assignment?	NA	0	0	0	0	0
6. Were treatment groups treated identically other than the intervention of interest?	0	1	0	2	2	2
7. Were outcome assessors blind to treatment assignment?	1	2	2	1	2	1
8. Were outcomes measured in the same way for treatment groups?	2	2	2	2	2	2
9. Were outcomes measured in a reliable way?	2	2	2	2	2	2
10. Was follow up complete and if not, were differences between groups in terms of their follow up adequately described and analysed?	0	2	2	1	2	2
11. Were participants analysed in the groups to which they were randomized?	2	2	2	2	2	2
12. Was appropriate statistical analysis used?	2	2	2	2	2	2
13. Was the trial design appropriate and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for in the conduct and analysis of the trial?	2	2	2	2	2	2
Total Risk of Bias	16	21	20	19	22	19
	Moderate	Low	Low	Moderate	Low	Moderate

	Canfield et al., 2023	Kennedy et al., 2023	Shorey et al., 2023	Van der Meulen et al., 2023
1. Was true randomization used for assignment of participants to treatment groups?	2	2	2	2
2. Was allocation to treatment groups concealed?	1	2	2	1
3. Were treatment groups similar at the baseline?	2	2	2	2
4. Were participants blind to treatment assignment?	0	0	0	0
5. Were those delivering the treatment blind to treatment assignment?	NA	0	0	0
6. Were treatment groups treated identically other than the intervention of interest?	2	2	2	2
7. Were outcome assessors blind to treatment assignment?	1	1	1	2
8. Were outcomes measured in the same way for treatment groups?	2	2	2	2
9. Were outcomes measured in a reliable way?	2	2	2	2
10. Was follow up complete and if not, were differences between groups in terms of their follow up adequately described and analysed?	2	0	0	0
11. Were participants analysed in the groups to which they were randomized?	2	2	2	2
12. Was appropriate statistical analysis used?	2	2	2	2
13. Was the trial design appropriate and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for in the conduct and analysis of the trial?	2	2	2	2
Total Risk of Bias	20	19	19	19
	Low	Moderate	Moderate	Moderate

Risk of bias for quasi-experimental studies.

	Onugwo 2023	Jiang et al., 2024	Burgio et al., 2024
1. Is it clear in the study what is the “cause” and what is the “effect” (i.e. there is no confusion about which variable comes first)?	2	2	2
2. Was there a control group?	2	2	2
3. Were participants included in any comparisons similar?	0	2	1
4. Were the participants included in any comparisons receiving similar treatment/care, other than the exposure or intervention of interest?	1	2	2
5. Were there multiple measurements of the outcome, both pre and post the intervention/exposure?	2	2	2
6. Were the outcomes of participants included in any comparisons measured in the same way?	2	2	2
7. Were outcomes measured in a reliable way?	2	2	2
8. Was follow-up complete and if not, were differences between groups in terms of their follow-up adequately described and analysed?	2	2	1
9. Was appropriate statistical analysis used?	2	2	2
Total Risk of Bias	15	18	16
	Low	Low	Low