Explaining the impact of mHealth on maternal and child health care in low- and middle-income countries: A theory-driven scoping review

Introduction Despite the technology in maternal and child health, the contextual factors and mechanisms by which interventional outcomes are generated have not been subjected to extensive review. In this study, we sought to identify context, mechanisms and outcome elements from implementation and evaluation studies of mHealth interventions to formulate theories or models explicating how mHealth interventions work (or not) both for health care providers and for pregnant women and new mothers. Method An electronic search of six online databases (Medline, Pubmed, Google Scholar, Scopus, Academic Search Premier and Health Systems Evidence) was performed. Using appropriate MeSH terms and selection procedure, 32 articles were considered for analysis. A theory-driven approach, narrative synthesis, was applied to synthesise the data. Thematic content analysis was used to delineate the elements of the intervention, including its context, actors, mechanism and outcomes. Retroduction was applied to link these elements using a realist evaluation heuristic to form generative theories. Results Mechanisms that promote the implementation of mHealth by community health workers/health care providers include motivation, perceived skill and knowledge improvement, improved self-efficacy, improved confidence, improved relationship between community health workers and clients, perceived support of community health workers, perceived ease of use and usefulness of mHealth, For pregnant women and new mothers, mechanisms that trigger the uptake of mHealth and use of maternal and child health services included: perceived service satisfaction, perceived knowledge acquisition, support and confidence, improved


Introduction
The potential for mobile health (mHealth) to enhance healthcare utilisation, promote affordability and support accountability in low-and middle-income countries (LMICs) is supported by the near-universal availability of mobile phones, with increasing coverage in many developing countries [1,2]. There is increasing attention in the use of information and communication technologies (ICT) such as mobile phones to improve the provision and quality of healthcare services. In this context, mHealth offers a personalised and interactive tool aimed at promoting healthcare access and awareness [3,4]. mHealth has the potential to strengthen the public sector for optimal management of chronic conditions and improvement of maternal and child health (MCH) services [5][6][7]. In addition to promoting health education among patients and reducing waiting times and cost of healthcare service delivery, mHealth enhances patient support, providing a system for emergency response and monitoring [6].
While outcomes-based evaluation of mHealth interventions can offer insight into their performance, little is known about how and why these interventions work (or not). There remains a need to improve the access to primary healthcare services among pregnant women and mothers and hence the need to understand how different contextual factors, such as culture, influence the uptake of mHealth services. This theory-driven scoping review sought to respond to this need by exploring and conceptualising contextual elements and mechanisms that interact to explain the observed effects of mHealth intervention on the uptake of MCH services in LMICs. It aims to build a plausible theoretical model using realist evaluation principles to explain how, why, for whom, and under which circumstances, mHealth 5 supports MCH services in LMICs [19].

Methodological approach
Our review was informed by the realist understanding of generative causality as conceptualised by Pawson and Tilley [20]. They proposed the formula M (resource) + C→M (reasoning) = O, to express the relationship between context, mechanism and outcomes to explicate how an intervention "causes" a behaviour. According to this formula "an outcome (O) is a product of mechanism (M) in a specific context (C)" [20]. Following this generative causality approach, theories or models can be formulated, tested, confirmed and/or modified using a context-mechanism-outcome configuration (CMOc) [21]. Some implementation scientists have suggested modifications to the CMOc heuristic to improve the explanatory power of the generative causality principle [22,23]. Marchal et al. [24] and Mukumbang et al. [25] proposed adding "intervention" (I) and "actors" (A) components to the configuration based on the fact that an intervention (I) can only work when adopted by actors (A). Based on this modification, the generate understanding postulates that "an outcome (O) is produced by a mechanism (M) activated in context (C) through an actor (A) when an intervention (I) is executed" [19,26]. Developing the models in this study was achieved by formulating Intervention-Context-Actors-Mechanism-Outcome (ICAMO) configurations (Table 1).

Study design
The scoping review was informed by the "York framework" proposed by Arksey and O'Malley [29], which include the following five stages:  (Figure 1).

Figure 1 PRISMA diagram illustrating the study selection process
From 813 records identified in the database searches, 747 articles were excluded after duplicates removal and abstract and title screening. Of the remaining articles, 14 systematic reviews were also excluded. Fifty-two (n=52) full-text articles were screened for potential inclusion, and twenty (n=20) records excluded for various reasons (Figure 1), yielding 32 eligible articles. were not common before that time.

Stage 4: Information charting
The data was charted using an ICAMO framework as described above (additional file1).

Stage 5: Summary and finding of review reporting
The narrative synthesis (NS) model proposed by Popay et al [30] informed the process of collating, summarising and reporting results. The NS framework proposes a theory-driven approach to data synthesis and is compatible with the philosophical assumptions guiding theory formulation in realist evaluation [31]. NS relies on the application of various methods of inference making through the use of words and text [30]. To this end, NS is applied in reviews addressing a number of questions, with research evidence in the context of studies that strives to inform policy and practice [30].
Four interrelated steps are involved in the conduction of an NS: (i) Theory development of how the intervention works: (ii) development of a preliminary synthesis of results of included studies; (iii) exploring associations in the data; (iv) assessment of the vigorous of the synthesis.
Step 1. Theory development of how, why and for whom the intervention works According to Arial et al. [32], a thinking framework viewed as a way of understanding and continuously testing and revising our understanding of how the intervention could improve people's health [33], is required as a first step. This thinking framework guides the process of operationalizing the mechanisms into theories or models at the end of the synthesis process. Figure 2 shows a tentative  Step

Development of preliminary synthesis of results
We applied a deductive thematic analysis, to extract the data [34,35], based on the concepts outlined in the ICAMO heuristic tool [36], and used an inductive approach to code constructs within each of the concepts of the framework (see additional file 1). We identified relevant aspects of the intervention (I), context factors (C), mechanisms (M) and outcomes (O) related to the delivery of the mHealth programme for CHWs/HCPs and pregnant women and new mothers separately.

Step 3. Exploring association in the data
The realist evaluation approach [21,37] informed the process of constructing the explanatory model. We aplied retroductive inferencing to explore and link the elements of the ICAMO heuristic tool. Retroductive inferencing is a mechanismfocused analytical approach used to reconstruct the basic conditions of a phenomena, based on the available data (abductive reasoning). Counterfactual thinking was applied to argue toward transfactual conditions -the existence of powers, potentials and liabilities which cause observations [34]. We then mapped possible explanations based on the data through the process of configurational mapping -a process of organising and representing knowledge by linking and specifying the relationship (s) between concepts.

Step 4. Assessment of the vigorous of the synthesis
To assess the robustness of the synthesis, three different steps were used: First, we applied the TAPUPAS criteria (Table 2), an appraisal tool developed by Pawson et al. [38], to appraise the selected article for relevance. Overall, included studies met the cited criteria and were considered relevant for our review.

Table 2 TAPUPAS criteria
Secondly, a quality assessment was performed for each article using a research evidence appraisal tool [39] (additional file 2). Eight of the 32 articles were of high quality, and 24 were classified as having good or moderate quality. We concluded that results from these studies could provide the relevant and credible information towards developing theories.
Finally, two of the study authors (EMK and FCM) applied judgmental rationality -the ability to evaluate different positions as being better or worse -to map ICAMO elements using Vensim® software. This was achieved through discursive and iterative consultation among the researchers.

Results
Of the total 32 studies retrieved from different geographic group including Sub-Sahara Africa twenty-one, Asia Pacific ten and Latin America one was included in this review (addition file 2). In accordance with the initial framework, we presented the findings at users and healthcare providers level of users. Figure 3 shows an explanatory model of how and why CHWs/HCPs implement mHealth intervention appropriately (or/not).

Figure 3: CHWs/HCPs (A) configuration mapping of mHealth ICAMO
The first aspect relevant to the delivery of the mHealth intervention is that it offers a communication platform [40][41][42][43]  The second relevant aspect to the delivery of the mHealth intervention relates to its ability to offer a data collection, data security and management platform within the system (I) [17,[45][46][47][48]. The importance of data collection, data security and management is influenced by the organisation of the health system, CHWs' training, supervision, support and mobilisation, availability of CHWs, and availability of resources (C) [45,46]. Having a functional data collection, data security and management platform improves the knowledge acquisition, confidence and selfefficacy (M+) [45][46][47]  which also improves confidence of CHWs/HCPs (M+) and hence, improvement of MCH services delivery (O+) [51,52].
Importantly, the mHealth intervention offers a health education platform (I) to CHWs/HCPs [4,45,[53][54][55][56][57]. A relevant health education platform is determined by three groups of contextual factors, including experience with technology and level of education; training, supervision, support, and mobilisation; and availability of resources (C). Having a reliable health education platform enhances the communication among CHWs/HCPs, which improves their knowledge acquisition and confidence (M+) [4] in providing ANC and PNC services, data collection and increased quality of collaboration with community members (O+).  The mHealth intervention also offers a mobile phone consultation with HCPs service (I) [17,43]. The mobile phone consultation with HCPs is influenced by socio-cultural practices norms, political clout, health literacy, awareness of intervention, lack of trust in technology and face-to-face preference, access to a working mobile phone, technical aspects of mobile phone services, community buy-in, and socio-economic status (C). The mobile phone consultation with HCPs motivates pregnant women and new mothers (M+) to enhance their use of ANC/PNC services (O+), improves facility delivery and emergency obstetric care (O+) and increases the use of iron tablets and immunization (O+) [17,43].

Discussion
The present study aim to explore how, for whom, and under which conditions, decision support and guidelines, and health education. Our results are consistent with those found in a study done by White et al. [70] indicating that mHealth tools such as smartphones and tablets can be used successfully to enhance the quality of data collection and, compliance with treatment protocols among patients. The authors emphasised that even though the acceptability of mHealth is sky-scraping, their application is not common. Seven mechanisms explaining how the implementation of the mHealth intervention are achieved by CHWs/HCPs were established, and included: motivation, perceived skill improvement and knowledge (improved self-efficacy, improved confidence, improved relationship between CHWs and client, perceived ease of use and usefulness, and knowledge gained) gives CHWS confidence.
Abejirinde et al. [48], identified usability and empowerment as important mechanisms to explain the adoption of mHealth. The authors explained that empowerment of health workers explained the competence of CHWs and that mHealth empowered CHWs to adopt and use mHealth in contexts where it aligns to their needs, workload, training, and skills [48]. The perceived useful and ease of use of mHealth encouraged and empowered HCPs with skills and confidence Gagnon et al. [71] also found perceived usefulness related to design and technical concerns, cost, time, privacy, ease of use, and security issues, risk-benefit assessment experience with the technology, and contact with others (viz management, colleagues, and patients).
Our study also identified seven mechanisms related to the adoption of mHealth by pregnant women and new mothers: perceived service satisfaction, knowledge acquisition, support and confidence, information overload, improved self-efficacy, encouragement, empowerment and motivation. Azhar and Dhillon [72] also identified perceived usefulness and ease of use, behavioural intent, self-efficacy, social-influence, attitude and perceived privacy threat as factors that influenced the successful use of mHealth applications for self-care. A systematic review done by Aker et al. [73], found that users' perceived platform quality, perceived services satisfaction, perceived quality interaction and outcome were found to influence users' uptake of mHealth for health care services utilisation. Aker et al. [73] identified other dimensions, such as system reliability, privacy, availability, adaptability, efficiency, assurance, responsiveness, functional and emotional 16 benefits to influence the uptake of the mHealth intervention.

How our model compares to relevant frameworks
The use of theoretical frameworks in mHealth evaluation has been found beneficial to formulate best practices in the field [74]. The Fogg Behaviour Model (FBM) [75] is a psychological model which proposes that for a targeted behaviour to occur, they must be the presence of the following at the same time: Ample motivation, ability and an active trigger [75]. Fogg explained that users with high motivation and abilities are likely to perform the target behaviour. When users have a low ability for instance due to technology requirements, they may find it difficult to complete the task. Users may have a high motivation and ability, but if there is not a trigger (a trigger can take many forms such text messages, an announcement etc.) users will not achieve their target behaviour. When a trigger is identified and associated with target behaviour then users are motivated and able to perform the behaviour.
The FBM is somewhat in line with the realist notion of generative causality, our realist theory identified motivation as a relevant mechanism for the succeffully implementation of mHealth programme by CHWs/HCPs and improved the uptake of MCH services by pregnant women and new mothers.
Another theory, the behaviour change support system (BCSS), has been used to measure how outcomes are generated and focuses on whether more traditional information systems can be used to persuade, influence, nudge, and coercion individuals into adopting a behaviour [76]. BCSS uses a technology that provides learning automated behaviour change, using a persuasive application to change human attitude or behaviour through the power of software designs [74]. This is similar to our ICAMO configuration, which explained that intervention modalities influenced actors' or users' conception by improving their self-efficacy, confidence and knowledge acquisition, satisfaction, motivation, encouragement and empowerment. Well-designed messages and a health education and communication platform are designed to give users persuasive information that could influence the way they would react towards MCH services utilisation.
Our ICAMO based theory aligns with the Fit between Individual, Task and Technology (FITT) framework developed by Goodhue and Thompson [77] to explain the degree to which a technology functionality matches task requirements and individuals in performing the portfolio of their work or abilities. FITT is influenced by technology characteristics, individual abilities, and task requirements on performance and users' evaluation of technology [77]. These categories are outlined in our ICAMO configuration which shows that users perceived ease of use -the ability of using the technology, and satisfaction with the technology can produce the desired positive behaviour change. FITT therefore, highlights the influence of mechanisms such as motivation, which is similar with that found in the model developed in this study that explains mHealth adoption provided a claire knowledge on the role of motivation in the delivery of MCH services.
Another theory from the field of information systems is the Technology Acceptance Model (TAM) developed by David [78] TAM seeks to explain users' adoption or rejection of information technology. David focused on two theoretical constructs, including perceived ease of use and usefulness, which are theorised to be basic determinants of systems use. David showed that attitude impacted the effect of perceived usefulness on intention to use [78]. Whether users adopt or do not adopt an application depend on the extent to which they believe the application will assist their job performance (perceived usefulness). If potential users believed that the application is useful (perceived usefulness), they may at the same time believe that the system is easy or not easy to use, which makes the performance of benefit of usage out weighted by the effort of using the application. TAM is in line with the ICAMO configuration model developed in this study, which identified perceived ease of use and usefulness among some other mechanisms that influence behaviour change (delivery and the uptake of MCH services).

Strengths and limitations
The present review provides important insights to understanding how mHealth programmes work, for whom and under which conditions, and motivates the importance of mHealth for addressing MCH service issues in LMICs. Understanding

Consent for publication
Not applicable

Availability of data and material
The dataset(s) supporting the conclusions of this article is (are) included within the article (and its additional file(s)).

Concepts Definition/descriptions
Intervention (I) Refers to the characteristics of the various mHealth interventions such as type of technology and mode of delivery. In this case mHealth modality was defined as the use of mobile phones an use of text, audio, images, short messaging services (SMS), voice SMS, applications accessible v radio service.
Context (C) Describes the conditions required for a programme mechanism to activate or not. Context c circumstances that facilitate or constrain mechanisms, including pre-existing individual, organisa cultural conditions, external to the interventions [27]. In this case context is categorised as: which comprises of the broad external environment in which the intervention is situated, includ economic, social, technological, legal, and infrastructural environment [2] b) Organisational/hea includes resources, policies and structures directly related to the unique health facility settin mHealth technology is introduced [2].

Actors (A)
Includes the individuals, groups, and institutions that play a role in the implementation an intervention [28] -In this study actors include pregnant women, mothers, children, healthcare and community health workers (CHWs).

Mechanism (M)
A mechanism refers to the causal forces, powers, processes or interactions that generate beha realist evaluation terms, mechanisms include the choices, perception, reasoning and decisions t as a result of the resources provided by the programme.

Outcomes (O)
Defined as the product of mechanisms activated within a specific context. Outcomes are the unanticipated (emergent) consequences of the intervention [21].