Introduction

Perinatal depression (PD), defined as depression occurring during pregnancy or within 12 months of delivery, is one of the most common complications of pregnancy worldwide [1]. Globally, 10–25% of women may experience PD, with women in low-resource settings being disproportionately affected [2, 3].

Untreated PD has been associated with numerous adverse outcomes for mothers and their offspring. For example, children may experience stunted postnatal growth or increased risk of later mental health difficulties, while mothers may be at higher risk for obstetric complications or premature delivery [4]. Although less frequent, infanticide or maternal suicide may also occur in severe cases [5, 6].

Treatment for PD frequently involves individual or group psychotherapy, although psychotropic medications such as antidepressants may also be used [7, 8]. Several psychological and psychosocial approaches, such as interpersonal therapy (IPT), cognitive behavioral therapy (CBT), and nondirective counselling, have been reported to be effective [9, 10]. Despite well-established, effective treatment modalities, only a small minority of women receive adequate perinatal mental health care. For example, only 10–15% of women living in high-income countries (HICs) who meet diagnostic criteria for PD receive comprehensive treatment [11]. This proportion is likely much lower in resource-constrained settings where mental health specialists are scarce and access to quality care remains limited [12].

Interventions delivered by non-specialists (also known as task sharing) have been widely recognized as an effective means of improving access to mental health care [12–14]. This strategy has been applied successfully to the provision of maternal mental health care worldwide. Non-specialist-delivered psychological interventions produced an overall reduction in common perinatal mental disorders symptomatology compared to usual perinatal care when delivered to women in low-income countries (LMICs) (SMD -0.34; 95% CI -0.53, -0.16) [15]; similar findings have also been reported in HICs [14].

While particular attention has been paid to the use of non-specialists for maternal mental health in LMICs, our understanding of their relevance and application in low-resource settings must also include rural communities. Regardless of country income status, rural communities worldwide suffer disproportionately from unemployment, poverty, and poor educational opportunities compared to urban areas [16–18]. Limited access to basic public health services, including mental health care, further compounds this urban-rural disparity [19, 20]. Non-specialists offer a potential solution to address this inequity; however, rural areas present unique challenges to the provision of mental health care, and strategies effective in urban areas (regardless of country income status) cannot be assumed to reliably apply to a rural context.

This review seeks to address the notable research gap related to utilizing non-specialists to deliver maternal mental health care in rural communities. Its goal is to enhance our understanding of the most effective and sustainable methods of mental health care delivery in this underserved and in-need population.

The specific objectives of this review are as follows:

1. To assess the effectiveness of non-specialist delivered interventions in improving outcomes related to depression including but not limited to depressive symptom severity and remission in women with perinatal depression living in rural communities;
2. To assess the effectiveness of such interventions in improving infant wellbeing and related outcomes including but not limited to physical health and cognitive development; and
3. To examine the content, processes, and elements of these interventions to aid in the design and implementation of future maternal mental health programs.

Methods

This systematic review was registered on PROSPERO on January 18^th, 2023 (CRD42023387400) and reported as per the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).

Results

Following PRISMA guidelines, the entire screening process is summarized (Fig 1). Eleven reports representing nine unique studies were eligible for inclusion in this review. One record [24] examined 12-month outcomes among the same group of participants who had received the original Philani Mentor Mother intervention [25]. Similarly, Maselko and colleagues examined outcomes 36 months later in the same cohort of women who had initially received the Thinking Healthy Peer-Delivered Program reported in another record [26].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA diagram.

NOTE: Seven additional records were identified via citation searching. Of those seven, reviewers (AA, NA) agreed that one did not pass title and abstract screening. As such, only six full-text records were sought for retrieval.

https://doi.org/10.1371/journal.pgph.0003031.g001

Study characteristics (Table 1)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study characteristics.

https://doi.org/10.1371/journal.pgph.0003031.t001

Apart from a solitary study conducted in Israel (n = 1) [27], the remaining studies were conducted in LMICs including Zimbabwe (n = 1) [28], South Africa (n = 2) [24, 25, 29], Pakistan (n = 3) [26, 30–32], and India (n = 2) [33, 34].

Five studies were RCTs [26, 28, 30–32, 34] (four of which were cluster-RCTs) and the remaining were quasi-experimental in design [24, 25, 27, 29, 33].

Five studies included adolescent participants (<18) or did not report age exclusion criteria [24, 25, 27, 28, 32, 34]. Six studies enrolled pregnant women [24–27, 30–33] and the rest enrolled participants postpartum, all within less than six months of delivery [28, 29, 34]. One study included women and their male partners as participants [28].

Three studies required all participants meet specified criteria for perinatal depression (e.g., Edinburgh Postnatal Depression Score (EPDS) ≥ 13) for enrolment [26, 30–32]. Although other studies included participants not considered to have or be at high risk for perinatal depression, they provided stratified results for a subgroup meeting such criteria, making them eligible for inclusion.

Interventions (Tables 1 and 2)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Intervention characteristics.

https://doi.org/10.1371/journal.pgph.0003031.t002

In five studies at least part, if not all, of the intervention was delivered at the participants’ homes [24–26, 29–32]. One intervention was delivered at a primary care setting [27], while others were delivered at an unspecified communal location or the homes of the community health workers [28, 33, 34].

Three studies delivered some form of the Thinking Healthy Program [26, 31, 32], a World Health Organization (WHO) endorsed intervention that has been adapted for use in numerous contexts [35]. Another three studies delivered the Philani Mentor Mother Program [24, 25, 29], and the remaining studies delivered interventions designed specifically for each program.

All interventions aimed to benefit the mother, the mother-child dyad, or both and utilized strategies such as psychoeducation, problem-solving therapy, behavioral activation, or psychosocial support. Psychoeducation, which most frequently involved providing information about perinatal depression and general maternal and child health, was the most common strategy. Some form of problem-solving therapy was also utilized as a key intervention strategy in many studies [26, 28, 31–34]. Behavioral activation and psychosocial support were particularly relevant in interventions for which peers served as the primary agent of delivery [26, 31, 34].

All interventions were delivered face-to-face. Interventions were delivered in a group format (n = 4), [28, 30, 33, 34], an individual format (n = 3) [24, 25, 29, 32], or some combination of individual and group sessions (n = 2) [26, 27, 31].

Delivery agents included community health workers (CHW) [24, 25, 28–30, 32, 33], female peers or community members [26, 31, 34], and female clinic staff (e.g., nurses, gynecologists) [27], and the extent of training received by each delivery agent varied significantly.

The number, duration, and frequency of sessions were also highly variable. Session duration ranged from 20 minutes [30] to 1 hour [28], and the number of individual intervention sessions ranged from three [27] to thirty-two [31]. Two studies did not explicitly mention the number of sessions received by participants [24, 25, 29]. The most intense frequency included multiple individual sessions within one day [33], while the longest time between sessions was every other month [31].

Study outcomes (Tables 1 and 3)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Study outcomes.

https://doi.org/10.1371/journal.pgph.0003031.t003

Measured outcomes were categorized as either maternal (e.g., disability, functioning, health-care behaviors, mother-infant interaction) or child outcomes (e.g., height, weight, cognitive development).

Eight of the nine studies included in this review identified maternal measures as primary outcomes. Six of the eight studies identified one or more depression-related outcomes as primary [24–28, 31, 33, 34], while the remaining two examined maternal help-seeking behaviors [30] and antenatal care visit attendance [29]. Only one study [32] examined child height and weight as the primary outcome.

While the outcome assessment tools and screening thresholds varied between studies, the depression-related outcomes were most frequently assessed using the EPDS [24, 25, 27–29, 33]. Other frequently used assessment tools included the WHO-Disability Assessment Score (DAS), the Global Assessment of Functioning Scale (GAF), and the WHO Motor Developmental Milestones.

Intervention type and delivery

Overall, there are too few studies included in this review to draw reliable conclusions on the effectiveness of interventions to improve maternal wellbeing based on the type of intervention and delivery agent.

Nevertheless, a broad overview of the data included herein shows that psychoeducation and problem-solving therapy were the most used intervention elements and female CHWs were the most used agent of delivery. In addition, all three delivery methods–group, individual, both–demonstrated some evidence for their effectiveness in improving maternal wellbeing, although interventions that did not report improvements in depression-related outcomes were more likely to have been conducted via group-based sessions. Finally, more frequently delivered interventions, particularly those started during pregnancy and extending to 6–12 months postpartum, appear more likely to improve maternal wellbeing. Of the four studies that did not show improvements in symptom severity or remission for participants, one had two sessions in total [30], another lasted only three days [33], and a third had a relatively short (6 weeks) postpartum follow-up period [34]. Four out of five studies reporting improvement in depression-related outcomes began during pregnancy and extended into the postpartum period [24–27, 32].

Of note, loss to follow-up is particularly common in research conducted in rural communities, and many studies employed repeated house visits or telephone calls to minimize the number of participants who did not complete all intervention sessions, particularly for interventions of longer duration. Several studies also attempted to integrate interventions into the existing rural healthcare system by employing community health workers already well known in the community.

Risk of bias assessment (Tables 4 and 5)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Risk of bias assessment using ROBINS-1 for Non-randomized studies.

https://doi.org/10.1371/journal.pgph.0003031.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Risk of bias assessment using RoB-2 for RCTs and Cluster-RCTs.

https://doi.org/10.1371/journal.pgph.0003031.t005

Concerns about the randomization process were noted for four of the six assessed RCTs due to notable sociodemographic differences between intervention and control groups at baseline. Generally, concern for domain 2 (deviations) was low, except in one study where the same CHWs delivered both the intervention and control [30]. For domain 3, five RCTs provided adequate information and reported retention rates >85%, except one trial with a 63% retention rate at 36 months [31]. Measurement concerns were noted in two trials, one relying on self-reports from CHWs [30] and another with non-blinded outcome assessors [34]. Selective reporting concerns (domain 5) were low, except in one trial without a located protocol [32].

Non-randomized controlled trials in this review had a higher risk of bias and were determined to have at least a moderate risk of overall bias.

Discussion

First and foremost, this review reveals limited evidence for the effectiveness of non-specialist delivered psychosocial interventions to significantly improve depression outcomes among women with perinatal depression living in rural communities. Only a few studies in this review reported any statistically significant differences in measured outcomes for this demographic, and the considerable overall risk of bias raises concerns about the validity and reproducibility of reported results. Furthermore, most studies did not focus exclusively on this participant population, and the lack of statistical power to analyze depression-related outcomes specifically in this demographic further reduced the body of evidence available for examination.

This overarching finding is consistent with the broader literature evaluating non-specialist interventions for common perinatal mental disorders conducted in urban and/or suburban areas. While some systematic reviews and meta-analyses on this topic have shown small, albeit promising results, the quality of evidence on which these findings are based is similarly poor. A Cochrane review reported that interventions delivered by lay health workers in LMICs might increase recovery in mothers suffering from perinatal depression but urged caution when considering this finding due to the preponderance of low-quality evidence [38]. Related studies have raised similar concerns [15, 39], and a systematic review of non-specialist delivered interventions for maternal mental health delivered in HICs noted that only half of the eligible trials presented information regarding treatment dosage and/or key supervision processes [14]. The relative lack of high-quality evidence reflected in this review and the broader literature more generally interferes with efforts to reliably improve perinatal mental health, particularly for those already suffering from significant symptoms.

Nevertheless, this review provides some evidence that interventions utilizing some combination of psychotherapeutic approaches, including problem-solving, behavioral activation, and cognitive restructuring, may be effective in treating women with perinatal depression living in rural communities. This finding is concordant with results of existing reviews and meta-analyses, which demonstrate psychotherapeutic approaches integrating interpersonal therapy or principles of CBT (e.g., cognitive restructuring, problem-solving) to be more effective in the prevention and treatment of perinatal depression than alternative approaches such as supportive counselling [14, 15, 40–42]. Similar findings are also reflected in the literature concerning adult major depressive disorder more generally; brief CBT and IPT-based psychotherapeutic approaches are more effective in the treatment of moderate to severe depression when compared to placebo [43] or usual care [44]. As such, future perinatal mental health interventions in rural communities should prioritize including such psychotherapeutic approaches in their design.

Secondly, this review indicates that non-specialist delivered psychosocial interventions do not significantly improve child outcomes, such as physical health and development, regardless of intervention type, delivery agent, or frequency. The association between perinatal depression and poor child health and development outcomes has been well established; nevertheless, few studies have been able to demonstrate that interventions for perinatal depression result in a significant improvement to child-related outcomes. While a recent meta-analysis found that non-specialist led interventions for common perinatal mental disorders were associated with better child cognitive development and improved growth indices [45], other studies have suggested otherwise. In fact, Cuijpers and colleagues found that psychological treatments for perinatal depression demonstrated no significant effects on the height and weight of offspring, results that mirror what is reported in this review [46]. As such, further exploration is needed to better understand what intervention components or approaches are necessary to reliably improve child outcomes, and future research teams should prioritize elucidating the mechanisms that underlie these complex interactions.

The notable exception to this finding, however, is the significant reduction in neonatal mortality, likely secondary to improved birth and hygiene practices, reported by Tripathy and colleagues. As such, it would be important to consider that peer-led group interventions of this nature may be of benefit in parts of the world where health literacy is low and early neonatal mortality rates are high due to unhygienic birth practices.

Third, this review also suggests that interventions delivered more frequently, over a more extended period, and initiated during pregnancy may be the most likely to significantly improve depression-related outcomes for women in rural communities. This finding must be interpreted cautiously as a review of the literature reveals little conclusive evidence as to the ideal dosage and timing of interventions for treating perinatal depression. Although a meta-analysis showed multiple-session interventions to be effective in preventing postnatal depression while those with a single session were not [47], other studies have found no association between treatment dosage and effect size [15] or have been unable to draw reliable conclusions due to poor reporting of key implementation details [14]. Previously published meta-analyses also found no association between intervention timing (e.g., during pregnancy vs postpartum) and effect sizes for depression-related outcomes [15, 48].

This review’s finding is somewhat at odds with the broader literature examining treatments for adult depression, an evaluation of which reveals strong supporting evidence for brief, low-intensity (LI) psychological interventions, particularly in the context of collaborative care and stepped-care models [49]. LI psychological interventions, abbreviated versions of evidenced-based therapies that can be delivered by workers with little to no formalized mental health training, are generally limited in terms of duration and frequency of contact (e.g., <6–10 distinct sessions) [50, 51]. Numerous systematic reviews and meta-analyses have demonstrated the effectiveness of LI interventions for treating a range of mental health conditions [38, 43, 52, 53], and their integration with stepped-care treatment models have been encouraged by the Lancet Commission for Global Mental Health and Sustainable Development as well as the WHO Mental Health Gap Action Plan [49, 54]. Stepped care models encourage the use of the least intensive, least expensive treatment first (e.g., non-specialist delivered LI psychological interventions) with the flexibility to provide more intensive care depending on individual patient needs or severity of presentation [55, 56]. As a result, this model of care ensures efficient allocation of resources and has been shown to effectively expand coverage of mental health care in resource-constrained settings such as LMICs [55, 57, 58]. While there are some concerns about the durability of LI treatment [59], stepped-care models integrating LI interventions appear to be a feasible and scalable way of providing effective depression treatment despite the limited duration and frequency of individualized sessions.

Despite the significant overlap in symptomatology between perinatal depression and depression in the general population, effective treatment for perinatal depression may differ from that of major depressive disorder due to the unique stressors and neurohormonal fluctuations characteristic of the postnatal period [60]. If, as this review suggests, longer and more intensive interventions are more effective in the treatment of perinatal depression for women living in rural communities, future implementation efforts must ensure that interventions remain both inexpensive and sustainable. This is particularly crucial in rural areas, which often have higher rates of poverty and significantly fewer mental health resources when compared to their urban or suburban counterparts [18, 61]. While this review is unable to draw conclusions as to the ideal format of delivery in rural communities due to the small number of studies included here, research suggests that both individual and group sessions are effective in the treatment of postpartum depression [15, 62]. Stepped-care models, like that discussed above, may also help ensure that women requiring more substantial, intensive care can access the care they require despite limited human resources and inadequate government funding. As such, adapting existing interventions to accommodate delivery in a group format and/or within a stepped-care system may provide an opportunity for more efficient and sustainable use of limited resources in rural communities.

In summary, this review allows us to make tentative conclusions based on a narrative synthesis of limited and low-quality data. While they provide important considerations relevant to the design and implementation of maternal mental health interventions for rural communities, the true value of this review lies in the considerable literature gap it reveals. While the evidence presented herein supports the effectiveness of the Thinking Healthy Program—a World Health Organization (WHO) endorsed intervention—in rural communities, there is a pressing need for additional research that specifically targets women with or screening positive for perinatal depression. As this review demonstrates, many studies evaluating non-specialist perinatal mental health interventions include participants with and without depression and thus fail to provide any reliable insight into the most effective intervention design for women already struggling with poor mental health.

This review’s strengths include independent screening, data extraction, and risk of bias assessment by a second reviewer, comprehensive efforts to ensure inclusiveness through expert consultation and citation searches, and strict adherence to the original protocol with minimal changes.

This review also has important limitations to consider, several of which are inherent to the individual studies and some to the methodological design of the review itself. First, this review was limited to papers available in English and did not search grey literature. Second, narrow inclusion criteria, particularly regarding the population and chosen time frame (<12 months postpartum) of the perinatal period, may have excluded studies, and some included studies were not specifically powered for the perinatal depression population. Note that the depression-related outcomes studies included in this review either 1) identified depression-related measures as secondary outcomes or 2) had broader participant inclusion criteria and thus only provided subgroup analyses for women scoring above a pre-specific threshold on self-report measures. As such, several studies included here were not powered to detect a difference between intervention and control groups within this population. There are no effect sizes or relevant p-values reported for these studies. Furthermore, few studies clearly reported the exact means of oversight employed to maintain fidelity of the intervention, information that significantly impacts researchers’ ability to accurately assess the effectiveness of each intervention and its replicability. And finally, the global heterogeneity in defining "rural" may lead to varied classifications in included studies. It is also possible that not all research conducted in rural areas is included here if study authors did not explicitly define setting.

Limitations of the studies included in this review included overall considerable risk of bias, especially in non-randomized trials (Table 3), and significant heterogeneity among studies prevented a meta-analysis due to variations in measurement tools, outcomes, follow-up duration, and intervention characteristics. Of note, control groups predominantly received standard perinatal care, which potentially exaggerated intervention effects in settings with limited healthcare access.

Despite these limitations, this review should be considered as a comprehensive overview of the relatively sparse literature examining interventions for perinatal depression in rural communities. Its rigorous methodological quality contrasts with the limited and low-quality data available, and recommendations to address this literature gap are discussed in further detail below.

Conclusion

Common perinatal mental disorders, including but not limited to perinatal depression, negatively impact mothers and their children across the world. This review presents some evidence to support the effectiveness of non-specialist delivered interventions to improve depression-related outcomes among women with or at risk for perinatal depression living in rural communities. Research suggests that women living in rural areas may be at higher risk for perinatal depression and anxiety than their urban counterparts [72], and the relative lack of high-quality studies included here demonstrates that further investment of resources to support rural maternal mental health research is urgently needed.

Finally, I’d like to thank the participants here took part in the research studies included here as without them this review wouldn’t have been possible.

Supporting information

Acknowledgments

A special thanks to everyone I consulted on this project who contributed invaluable time and effort to make this possible. Prof Simone Honikman, Prof Cindy Lee Dennis, Prof Jane Fisher, Prof Tatiana Salisbury, and Dr. Alan Stein, thank you for the time and expertise you dedicated to ensuring that this project was as comprehensive a review of the current literature as possible. And to the librarians at the London School of Hygiene and Tropical Medicine and King’s College London, thank you for taking the time to review and discuss my search strategy. I am deeply appreciative of your patience and willingness to teach.