Search strategy

A comprehensive search was conducted of OVID MEDLINE, OVID EMBASE, EBSCO CINAHL, and SCOPUS according to a search strategy created with assistance from an academic health sciences librarian (MS). The search was executed on January 30^th, 2025. The results were excluded if the publication date was restricted between January 1^st, 2003, to January 30^th, 2025, since the majority of mHealth interventions were within the last 2 decades. The search strings used in each database are provided in S1 Appendix Search Strings.

Eligibility criteria

To be eligible for the review, the studies needed to be primary research articles published between January 1^st, 2003, and January 30^th, 2025. Studies needed to report on maternal support utilizing messaging platforms as the intervention, addressing a rural population anywhere globally, with participants of any age. Studies were included with no limitations on language. The eligibility criteria are presented in Table 1.

Download:

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

Table 1. Selection, inclusion and exclusion criteria for the preliminary analyses of title and abstract screening and full-text review.

https://doi.org/10.1371/journal.pone.0336168.t001

Study selection

Screening for this review was completed using Covidence, a systematic review management tool, through a three-step process. First, pilot testing of the eligibility criteria was completed to ensure consensus between the team members, after which all article titles and abstracts were screened independently and in duplicate by two reviewers (SR and AO). All studies captured by the search had English titles and abstracts. Conflicts were resolved through discussion between the reviewers. In the next step, the full texts of each study were located and imported into Covidence, following which the two reviewers completed full-text screening independently. Conflicts were resolved through group discussions.

Data extraction

Following a pilot test of the data extraction templates, data extraction for the included articles was conducted independently by two reviewers. The templates captured the following items: authors, year of publication, study title, country, study design, population, study objectives, evidence of impact, barriers and facilitators, the respective excerpts, and Consolidated Framework for Implementation Research (CFIR) constructs. Conflicts in data extraction were resolved through group discussion. Conflicts in identifying evidence of impact included cases where one reviewer found a different instance where impact was demonstrated than the other reviewer. Conflicts for barriers and facilitators included cases where a reviewer either identified different barriers or facilitators from the other reviewer or labelled the barrier or facilitator differently.

A study characteristics table was created to compile all descriptive information. Data related to the barriers and facilitators were compiled into two consensus documents that consisted of the name of the barrier or facilitator that was extracted from each study, an excerpt that demonstrated the barrier or facilitator, along with the country and author for the study and the CFIR construct and domain that aligned with the barrier or facilitator. These documents were used to create collapsed tables for the barriers and facilitators, respectively. One team member grouped barriers and facilitators with the corresponding studies that had identified them, including excerpts that demonstrated the barrier or facilitator in the study.

Data synthesis

In this review, “impact” was defined as any significant change in maternal health outcomes resulting from the use of messaging platforms. This included both measurable health outcomes (e.g., increased ANC visits, improved knowledge) and qualitative outcomes (e.g., satisfaction or perceived benefits). When studies presented mixed findings, all reported impacts were included in the analysis. For qualitative studies, impact was assessed based on participants’ narratives or themes indicating changes in attitudes, behaviours, or experiences linked to the intervention.

The CFIR framework was selected for its comprehensive structure, enabling the analysis of individual, organizational, and system-level factors influencing implementation. Unlike frameworks that focus more narrowly on behaviour change or outcome evaluation, CFIR is better suited for capturing the contextual barriers and facilitators present for maternal mHealth messaging interventions. Two team members used the CFIR framework to categorize each barrier and facilitator identified through data extraction. A pilot test was conducted with the first three barriers and facilitators, followed by a consensus meeting to review the work, during which any inconsistencies were clarified to ensure they could be mitigated. After this, the remaining barriers and facilitators were organized using the framework, with any conflicts being resolved through group discussion. The final consensus meeting was utilized to produce a table of the results with excerpts from the studies. A thematic analysis, which is the process of identifying, organizing and interpreting themes, was then conducted by two reviewers on the barriers and facilitators [17,18]. The themes emerged through an inductive process guided by a phenomenological lens, focusing on lived experiences with messaging interventions from pregnant women, healthcare workers and implementers. The two reviewers were deeply immersed in the data, reviewing quotes and reported outcomes, conducting open coding of extracted findings, and refining themes through discussions. By engaging with the findings across studies through this process, recurring patterns and salient concepts were identified. Outcome reporting (positive, negative, or neutral) was noted to support a balanced interpretation of the intervention’s effects across included studies.

Quality assessment

The Mixed-Methods Appraisal Tool (MMAT) was used by two reviewers independently to assess the quality of the eligible studies [19]. Discrepancies between the reviewers were resolved through group discussions. The MMAT was selected because it allows for the appraisal of all three study types within a single tool, ensuring consistency and comparability across all studies. It evaluates studies based on clear criteria, including the coherence between research questions, data collection methods, sources of evidence, and interpretations. The MMAT assessment is included in S2 Table MMAT.

Types of Study/Methods

The studies included in this systematic review consisted of qualitative (n = 3) [20–22], quantitative (n = 4) [23–26], and mixed methods (n = 8) [27–34]. Details are shown in S3 Table Study Characteristics.

In the qualitative studies (n = 3), data collection methods included in-depth interviews (n = 2) [21,22], document reviews (n = 1) [20], field observations (n = 1) [20], and focus group discussions (n = 2) [20,21]. In the quantitative studies (n = 4), methods included randomized controlled trials (n = 2) [25,26], a randomized parallel, two-arm standard of care-controlled feasibility trial (n = 1) [23], and a non-randomized intervention study with self-reported demographic information (n = 1) [24].

Language

The majority of the articles captured by the search were in English (n = 429), with the search across all databases capturing only two studies in other languages that were specifically in French (n = 2). The two French studies were screened and were found not to meet the eligibility criteria during the title and abstract screening [35,36].

Year of publication

The included articles were published as early as 2014 and as recently as 2023. The bulk of the articles used in our study were published between 2018 and 2020 (n = 8), with the most articles being published in 2018 (n = 5) [20,24,25,27,29]. The fewest articles were published in 2014 (n = 1) [28] and 2020 (n = 1) [34].

Participants

The most frequently engaged target population across all studies in this review was pregnant women (n = 9) [20,22,24–27,29,33,34]. Other participants included those in maternal and child health contexts. For instance, mothers, including lactating mothers and postpartum women who had received the intervention, were recruited across several studies (n = 5) [20,22,29,31,33]. Health workers, community leaders, and community health workers, who are healthcare providers that receive lower levels of education than certified healthcare providers and live in the community they serve, were also significant participants (n = 3) [21,30,33]. Household members, both male and female, and caregivers were included in some studies (n = 2) [28,32]. Additionally, married women, regardless of their pregnancy status, were included (n = 1) [30]. Specific demographic groups, such as rural pregnant women who were positive for depression (n = 1) and women initiating breastfeeding within 24 hours of giving birth on antiretroviral treatment (n = 1), were also targeted [23,29]. All participants were from rural areas, ensuring a focused examination of rural maternal health dynamics.

Countries

A total of 10 countries were included across all studies, including the United States [27,29], India [25,28], South Africa [21,23], Kenya [22,26,32], Nigeria [37], Vietnam [20], Cambodia [31], Uganda [24], Ghana [33], and Samoa [34]. The most frequently cited settings were Africa and Asia. The most frequently studied countries were the United States (n = 2), India (n = 2), South Africa (n = 2), and Kenya (n = 3) [21–23,25–29,32]. Fig 2 shows the barriers and facilitators emerging from each country.

Download:

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

Fig 2. Barriers & Facilitators to mHealth Messaging Platforms in Rural Populations by Country.

https://doi.org/10.1371/journal.pone.0336168.g002

The barrier and facilitators labels have been shortened for the purposes of the figure. (A) The barriers corresponding with the country are noted in red. (B) The facilitators corresponding with the country are noted in green.

MMAT

The MMAT was used to organize and assess methodological quality according to the tool’s established criteria. S2 Table MMAT provides the results for all 15 studies. No articles were removed after the MMAT assessment.

Evidence of impact

The evidence from the 15 studies demonstrates the supportive nature of messaging interventions in improving maternal and child health outcomes. Participants in numerous studies expressed high satisfaction with the interventions, reporting that they found the messages helpful and convenient [20,22,27–29,31]. Additionally, messaging interventions were associated with increased knowledge of maternal and infant health topics, including awareness of the importance of iron and folic acid supplementation during pregnancy and knowledge about low-birth-weight babies [28]. They also improved communication between healthcare providers and pregnant women, leading to enhanced access to perinatal depression treatments and improved maternal health outcomes [29]. Messaging interventions were found to support maternal and neonatal health during pregnancy and childbirth, reduce home deliveries in favour of health facility deliveries, and increase uptake of antenatal care [21,24,26,28,29]. However, some studies found a lack of significant impact on exclusive breastfeeding rates and the number of follow-up antenatal care visits [23,34]. Despite these variations, the collective evidence suggests that messaging interventions hold promise in positively influencing maternal and infant health outcomes.

It is important to note that several interventions included complementary resources beyond text messaging, such as weekly CHW follow-ups, in-person counselling, transport support systems, non-digital materials, and voice calls [21–23,25,27,30,33]. These additions likely enhanced engagement and impact.

Barriers and facilitators

A total of 26 barriers and facilitators to the implementation of maternal health messaging interventions were identified across all eligible studies and were categorized using the Consolidated Framework for Implementation Research (CFIR). The detailed extraction of barriers and facilitators, along with their respective excerpts from eligible studies, is included in S4 Table Collapsed Barriers and Facilitators with CFIR. Fig 3 depicts the domains and key constructs with the corresponding barriers and facilitators.

Download:

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

Fig 3. Domains and Constructs captured by Eligible Studies.

https://doi.org/10.1371/journal.pone.0336168.g003

There may be more constructs captured for each domain, however the most critical constructs have been included in the figure. (A) The barriers corresponding with the constructs are noted in red. (B) The facilitators corresponding with the constructs are noted in green.

Innovation domain

The innovation domain refers to aspects related to the intervention being implemented [38]. The relevant constructs under this domain included innovation adaptability, innovation complexity, and innovation design.

Five out of 15 studies described barriers and facilitators related to the design of the messaging platforms and how it could impact how they valued the information provided [20–22,28,31]. The design of the intervention was noted as a barrier when there were limited word counts restricting detailed health content and messages sent too frequently or at inconvenient times, leading to fatigue and dismissal [28].

One study cited limited adaptability as a barrier, specifically with language being inadequately adapted for Spanish-speaking participants, who had lower participation rates due to the intervention not supporting their native language [27].

Complexity also affected engagement in two studies, particularly when managing intervention tools like multiple phones [27,31]. This logistical issue reduced engagement, as participants sometimes forgot to check messages.

In terms of facilitators for the design of the innovation, clear and relevant messages, with timing and content tailored to pregnancy stages, were appreciated by participants, which helped improve engagement and usefulness [20,22,31].

Outer setting domain

The outer setting domain refers to external influences on implementation [38]. Constructs relevant under this domain included local attitudes, local conditions, partnerships & connections. In four out of 15 studies, barriers associated with existing perceptions, along with the social and physical environment of the region, affected participants’ willingness and ability to engage with the intervention [21,22,27,30].

In two studies, strong community beliefs led to a preference for traditional birth attendants (TBAs), who are individuals with no formal medical training that assist women during childbirth using traditional practices, over hospital care [21,22]. This reduced engagement with messaging interventions, as TBAs were seen as equally safe as medical care, and the messaging intervention was used only when traditional methods failed [21,22].

Infrastructural conditions in the intervention’s environment affected the effectiveness in another study, where the unavailability of transport workers and limited access to healthcare facilities in rural Nigeria led to many requests for rides being unmet due to a lack of available vehicles at critical times [30].

In another study, buy-in from partners limited its reach and impact as there was resistance from clinical partners, who were hesitant to refer patients to messaging interventions [27].

Inner setting domain

The inner setting refers to aspects of the setting in which the intervention is being implemented [38]. Relevant constructs under this domain included structural characteristics, relational connections, communications, compatibility, and available resources.

The availability of resources, including funding and infrastructure, was referenced as a barrier or facilitator in five out of 15 studies [21–23,32,33]. Specifically, in four studies, limited resources posed significant barriers, including poor network connectivity, lack of electricity, limited phone storage, and financial constraints preventing access to phones, airtime, or data [21,23,32,33].

Conversely, in two studies, participants lacked access to knowledge and information on how the intervention works or how to actually use the intervention [28,34]. Participants with low digital literacy struggled to use mobile phones effectively, limiting their ability to benefit from text-based messaging interventions [28].

Additionally, lack of compatibility was noted in one study where the messaging platform did not fit well within existing healthcare workflows, leading to inconsistent implementation by midwives and healthcare providers [34]. However, the interventions’ compatibility with the existing workforce was found to be a facilitator in a separate study where the messaging platform reduced the workload of CHWs by streamlining communication [31].

Other facilitators included communication of information and support to participants, which was cited in two out of 15 studies. Participants appreciated having CHWs available between appointments and receiving follow-up calls from midwives, as both forms of direct support provided personalized care and increased their sense of support during pregnancy [29,33].

Available resources were identified as a facilitator in one study, linking women to reliable transport providers, which improved access to healthcare services despite infrastructural challenges [22].

Individuals domain

The individual’s domain focuses on the traits of the people involved in implementation (CFIR). It included the construct motivation, which was referenced in one out of 15 studies. In the study, there were negative experiences with healthcare providers, such as a perceived lack of concern for patient issues, which discouraged some individuals and impacted their motivation to engage with the intervention [22].

Implementation process domain

The implementation process domain included the strategies taken to implement the intervention, and included the constructs assessing needs and tailoring strategies, which were present as facilitators in 2 studies. One study found that evaluating participant needs and offering alternative enrollment methods, such as paper registration instead of SMS, increased participation by removing cost barriers [34]. Another study reported that tailoring messages to participants’ pregnancy stages improved the relevance and impact of the information [27]. The constructs reflected in eligible studies are reflected in Table 2.

Download:

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

Table 2. Frequency Table of Cited Consolidated Framework for Implementation Research (CFIR) Constructs.

https://doi.org/10.1371/journal.pone.0336168.t002

Implications of specific CFIR domains and constructs

The presence of constructs like “Infrastructure and Resource Availability” and “Innovation Cost” highlights the critical importance of addressing logistical and economic challenges in rural settings. The lack of affordable access to airtime and phones, combined with limited network coverage, demonstrates how infrastructure can fundamentally limit the success of messaging interventions. The inclusion of “Innovation Design” as a facilitator in some studies demonstrates that when the intervention is tailored to user needs (e.g., clear messaging, easy access), it can enhance engagement and outcomes. On the other hand, domains and constructs like “Financing” or “External Pressure” were not represented, suggesting that issues like funding or external policy pressures were either insignificant in the reviewed studies or not explored. The absence of these constructs could indicate that the interventions were focused more on immediate logistical challenges rather than larger systemic factors, which might be areas for further research. The domains and constructs that are not underrepresented or not represented at all may not be as critical to the mHealth messaging interventions and thus were not captured.

Emergent Themes

Five overarching themes from the analysis of the results were identified, including: 1. Intervention Accessibility and Content (encompassing the direct delivery of clear, timely text messages and educational components that empower maternal decision-making, while also highlighting barriers such as managing multiple phones, financial constraints, and design limitations related to message length and local language availability), 2. Infrastructure and Resource Availability (reflecting facilitators like linking women with reliable transport providers, yet noting barriers such as poor network connectivity, unavailability of transport workers, and limited phone memory that hinder access to care), 3. Cultural, Religious, and Social Preferences (indicating that in some settings, a preference for traditional birth attendants or healers over digital support can limit intervention uptake), 4. Equitable Engagement and Tailoring (underscoring the benefits of personalized messaging, consistent advice, follow-up calls, and alternative registration methods, while also revealing challenges like lower participation among Spanish-speaking participants and issues arising from shared phone usage), and 5. Healthcare Provider Attitudes (highlighting the importance of clinical partner buy-in and supportive healthcare worker interactions, with barriers including resistance to change and blurred boundaries between professional and personal communication).

Strengths and Limitations

The study provides a comprehensive analysis of maternal mHealth messaging interventions by encompassing a broad range of studies from different regions. The search was not restricted to any countries or languages, so transferable patterns in rural communities are offered globally. However, the search terms, publication dates and number of databases limited the results of this study, as is common with systematic and scoping reviews. If the words identified in the search string were not used or a study was not included in our selected databases, the article would not be captured. Varying definitions of rurality across the studies could also affect comparability of barriers and facilitators across rural communities. We considered articles in all languages; however, the search captured only two articles in languages other than English, which might result in the sample not being representative of all populations. Insights from secondary sources, including reviews, gray literature and program evaluations, were not considered as the inclusion criteria were limited to only primary resources.

Furthermore, choosing the CFIR framework might be seen as a limitation because this framework has been critiqued for not adequately explaining causal mechanisms or the process of change. Despite this, the CFIR framework provided a structured approach to identify and organize barriers and facilitators in mHealth messaging interventions for maternal support, helping to clarify how contextual and design factors shaped implementation in rural settings. However, while there was representation in all 5 domains, only 16 out of 67 constructs were explored, reflecting the limited emphasis on implementation science considerations within existing mHealth messaging intervention studies. Using the CFIR was still valuable as it highlighted areas that were often overlooked, such as individual characteristics, organizational readiness, and strategic planning, which are likely to play an important role in intervention uptake and sustainability. Additionally, it has been effectively used in other similar systematic reviews and helped us situate the findings within the broader scope of implementation research [56–59].