Scaling Mhealth in Africa: Lessons From The Implementation of The MomConnect Program

Background: Mobile health programs have strengthened health systems in Low- and Middle-Income Countries (LMICs) to achieve health-related goals. MomConnect, a mobile health program in South Africa targeted at improving antenatal and maternal health, has scaled rapidly since its creation in 2014. This study explores the barriers and facilitators to the implementation and scaling of the MomConnect program and the applicable lessons for the scaling of mhealth programs in the region. Methods: We conducted a qualitative study with key project partners and leaders who worked on the MomConnect project. Interviewees were initially identied through a literature review, publications, and evaluations of the project. Interviewees included individuals serving in implementation oversight, champions, partners, funders and frontline implementer roles. The Consolidated Framework for Implementation Research (CFIR) informed the a priori codes for directed content analysis. In total, 15 key stakeholders were interviewed. Interviewees were asked to identify any barriers or facilitators to the implementation of MomConnect and how they would overcome those barriers and strengthen the facilitators. Results: This qualitative study identied multiple barriers and facilitators to implementation within our domain of CFIR: characteristics of the intervention (complexity, trialability, evidence strength & quality, cost, design quality & packaging, adaptability), inner setting (available resources, compatibility, implementation climate, access to knowledge & information), outer setting (cosmopolitanism, external policy & incentives) and process (planning, external change agents, champions, formally appointed internal implementation leaders). Overarching thematic areas spanning the barriers and facilitators included: (1) strategic partnership and coordination across multiple sectors, (2) innovation costs and funding, (3) operationalization of the innovation to local and national settings and (4) mhealth policy and legislation frameworks. Conclusion: The barriers and facilitators identied under the CFIR domains can be used to build knowledge on how to strengthen mhealth programs in Africa. The continued success of the MomConnect program will require overcoming identied barriers and capitalizing on known facilitators. These ndings can serve as a foundation for the effective design and scale of mhealth interventions in the region.

healthcare providers were able to receive feedback from users and improved their services, whereas without MomConnect the healthcare providers would have to invest in and conduct their own feedback surveys (11). In addition, 2 years after MomConnect was launched, NurseConnect was launched in January 2016. This program set out a complimentary program for nurses to provide information they need to successfully run MomConnect. NurseConnect provided SMS text messages and access to a helpdesk, and a mobile website that included information and resources for nurses (12). Delivering the project digitally through SMS text messages was more cost-effective which was of great bene t (13). The MomConnect project was easily scalable because women are just as likely as men to own a mobile phone, and if they did not own a mobile phone, they could use a shared mobile with family members. The project often served as one of the most reliable registries for pregnant mothers. It provided vital health information to new and expectant mothers and families through mobile devices (14).
Several studies have evaluated the MomConnect program (10,13,(15)(16)(17)(18)(19). However, there has been limited acknowledgement of the root causes of the barriers in implementing and scaling MomConnect and limited perspectives provided from those who were involved in implementing and scaling the project.
We were therefore interested in understanding how the MomConnect program was implemented and what lessons could be drawn and adapted for other mHealth projects in the region.
The objectives of this study was to provide an assessment of the barriers and facilitators to implementing and scaling of MomConnect from the perspectives of key implementers and partners, and how the barriers and facilitators identi ed could inform the implementation and scale of other mhealth programs.

Study purpose and sampling strategy
The purpose of this study was to identify barriers and facilitators to the implementation of the MomConnect program. We did so by interviewing key informants (leadership and partners of the MomConnect project). Interviews were conducted by two of the authors. Key informants were initially identi ed through a literature review and publications related to MomConnect. We chose to interview current and past practitioners involved in the start, leadership and or evaluation of the MomConnect project. Using snowball sampling, identi ed key informants recommended additional interviewees. The total number of stakeholders interviewed consisted of 15 key stakeholders, all of whom were selected based on their expertise and experience with the implementation of mHealth interventions, speci cally in relation to MomConnect.

Data Collection
Interviews were approximately 25-57 minutes in length and were conducted online using Skype or by phone utilizing a semi-structured questionnaire [see Additional File 1 for interview questionnaire]. The interviews were recorded to be used later to transcribe the interviews. Each interviewee was asked to describe their role in the intervention. Additionally, they were asked about their knowledge on the current implementation of MomConnect and whether the intervention was di cult or easy to implement. The stakeholders were then asked to identify what they believed were the major barriers to the implementation of MomConnect, and subsequently what would be the most effective way to overcome the barriers mentioned. Lastly, they were asked to denote the major facilitators in the implementation of MomConnect, and what will be the most effective way to strengthen the facilitators mentioned. All 15 interview transcripts were transcribed manually and prepared for next phase of analysis.

Data Analysis
The Consolidated Framework for Implementation Research (CFIR) guided our data analyses and is represented in Figure 1 (20). The interview transcripts were analyzed through manual coding, utilizing ATLAS.ti a qualitative research analysis tool (21). The o cial 15 transcripts of interviews were uploaded into ATLAS.ti, where each transcript was manually assessed to identify the barriers and facilitators mentioned by key stakeholders. The CFIR framework is a compilation of constructs that are used to evaluate effective implementation (20). This framework consists of 5 domains and 39 constructs as represented in Figure 1. Each barrier and facilitator was coded respectively using this framework. The analyses and coding of the transcripts were done independently by authors 2 and 3. Any discrepancies were discussed with author 1. The data was organized by barrier and facilitator names, along with the coded constructs, domains, and key quotes from the interviews. Based on the data extracted, the frequency of key stakeholders who mentioned certain barriers and facilitators were organized accordingly, as represented in Table 1. The data was then assessed for pertinent themes.

Ethics approval
We obtained Research Ethics Board approval and consent from each participant was received prior to conducting these interviews. Throughout the process, participants were given the opportunity to withdraw voluntarily, at any time, and to pass on questions they were uncertain about or did not want to answer.

Results
From the data extraction and analysis there were 22 barriers and 23 facilitators identi ed, as represented in Table 2. These barriers and facilitators were grouped into four themes, as represented in Table 3. These themes included partnership relations; innovation costs and funding; operationalization of the innovation; and policy & legislation.

Champions
Eleven key informants emphasized that strong political will from the government was a critical factor taking the MomConnect project to scale: "There was very strong and very senior level government buy in for the Mom Connect project… I have never seen with any health project before" (Int. 12). Such leadership from the government was unique for a mHealth project and was key in scaling among South Africa. Additionally, seven key informants expressed how stakeholders effectively championed the implementation of MomConnect through collaboration, determination, and support from a multidisciplinary team: "what made the project so successful was also the collaboration -and was the partners that really did come together, and each had a really unique perspective, and a really focused area of work within the project." (Int. 4). Four key informants however expressed the inadequate support from premier principals, operational managers, and CEOs who were involved in launching MomConnect. With such inconsistency in support across the different levels of government and stakeholders there was an impact on how MomConnect was implemented and regulated.

Cosmopolitanism
Five key informants expressed that the management of stakeholders was inadequate in respects to the level of consistency in collaboration and participation in the implementation of the project. They mentioned with the implementation of MomConnect it is di cult to manage the politics and partnerships that exist among the diverse stakeholders. Additionally, there were many stakeholders who had their own objectives which limited program sustainability as they were not all consistently involved: "challenges for the sustainability of the program because there are a number of different partners involved and all of that is different pieces in the puzzle and if one of those pieces isn't there, the whole thing doesn't work. So, ensuring sustainability across the system I think is easier said than done" (Int. 4).

Formally Appointed Internal Implementation Leaders
One key informant identi ed in the implementation of MomConnect there was a gap in technical leadership. Acquiring a NDOH technical lead who would understand the technology stacks, was hard because it was expensive. Without a technical lead there was a limited overview on the technology and understanding how the different sets of technology worked to build the applications used in MomConnect.

External Change Agents
Four key informants noted partner relations during the implementation of MomConnect was de ned by the different partners and leaders working together and supporting each other: "Also a very strong leadership particularly in the minister; strong top-down director coupled with simple tech, and relationship amongst the funders, and with HIPPA and NDoH made the project scale quickly" (Int. 2). Strong partner relations and leadership among donors, Health Insurance Portability and Accountability Act (HIPPA), the minister and NDOH encouraged the project's successful ability to scale.

Access to Knowledge and Information
Four key informants stated that project practitioners such as nurses had inadequate access to information that the users had access to through MomConnect. Thus, without adequate access to the information users had, nurses felt less inclined to push registration and departments were not effectively communicating with each other. A different key informant expressed that communication among project stakeholders was a challenge as they could not coordinate or maintain regulations.

Cost
Five key informants expressed that service providers and technology partners greatly supported the program by subsidizing SMS's and messaging to provide a cost-effective adaptation of messaging which bene tted scaling measures. However, nine key informants noted that high cost of data was a large factor that determined accessibility and participation in the program. The same informants mentioned that there is a base cost and linear pricing for SMS costs that limited scaling efforts: "So trying to keep the cost down of the consumable elements, like the recurring elements; like SMS and USSD segment, keeping cost down on that was not easy and I think there were different factors that played there that made it more di cult to really sustain it" (Int. 9). Additionally, two key informants expressed there were uncontrollable factors such as market forces that control the costs of SMS, data and access to mobile network signal.

Available Resources
Two key informants noted that many funders were able to establish their commitment to the program by providing the initial funds. This showed how committed the funders were to the project and mHealth in its early phases. Unfortunately, ve key informants expressed that there was inadequate training on the use of program technology and program procedure among the project practitioners. They relayed that the level of training was dependent on each facility, it was challenging to keep training consistent and up to date with the high-job turnover rates.

Operationalization of the innovation Design Quality and Packaging
Ten key informants mentioned that the program greatly bene tted from having technology that is both easy and accessible: "Everybody has a phone, the software penetration in South Africa is over 100%" (Int. 13). With the participants' existing familiarity and accessibility with the technology it was easier to implement and scale the program. Three key informants expressed the success of having two-way communication, as it helped assess complaints effectively with direct communication to all levels (clinic, national, district, provincials) and a report that outlined how they xed the problem. Additionally, MAMA served as an effective guideline for MomConnect, as two key informants noted the MAMA model was adapted to better t the South African context for MomConnect.

Adaptability
Three informants described the content translation to be easy to access and appropriate which encouraged MomConnect users to participate. The project was also commended by ve key informants for the engagement with end-users, as they received a lot of positive feedback that helped effectively scale MomConnect. In respect to language translation, three key informants noted some inconsistencies with translation into the eleven o cial languages largely due to the cost and ability to manage such a need. They expressed concern about the loss of meaning in translation as many languages were more so spoken than written.

Evidence Strength and Quality
Uniquely two key informants expressed that the program was ready from the beginning to be successfully implemented directly without a pilot phase. The program bene tted from having a strong vision and political support to drive the direct implementation with province wide scaling. Five key informants however expressed that without proper development of a baseline, MomConnect's ability to effectively scale was greatly impacted. The key informants mentioned that the project required baseline evidence to support the government in understanding the program's impact. Additionally, the same informants acknowledged the lack of evidence in MomConnect's implementation was a signi cant factor that impacted the timeline of the project as they needed to provide proof of e cacy.

External Change Agents
Three key informants discussed how stakeholders' often placed the program implementation under pressure on conducting the program within a certain time period, thus MomConnect was impacted in terms of quality and adoption of the program: "I think the reason for the pushback from some of the facilities which was due to the time crunch, the Minister of Health wanted and set a date for Mom Connect to be launched. I don't know how he came to that date speci cally, but it meant there was very little time for any individual facility level engagement around the Mom Connect project" (Int. 12). The pressure of time from the Minister of Health, in uenced the uptake and engagement of facilities and end users.

Trialability
Two key informants suggested the program was not able to test the intervention and obtain a better understanding of the impact or e cacy of MomConnect before implementation because the program went nationally immediately. As a result, this posed some challenges when scaling and implementing.

Compatibility
Two key informants mentioned that MomConnect was not fully compatible with the health system, as a large number of facilities in South Africa were unable to adopt MomConnect because they felt already stretched for resources, time, or staff it was too much work to add onto their workload. Additionally, two key informants mentioned some participants experienced the inability to register privately or postnatally with MomConnect, due to the switch between sectors they have missed information offered by MomConnect thus far. Although it is now possible for mothers to register for MomConnect postnatally. [1] Engaging Three key informants mentioned that the effective marketing materials and engagement from minister advanced the project promotion and awareness for the MomConnect platform in South Africa. This effective promotion and support from the NDoH among the target population encouraged MomConnect scaling and engaged more participants.

Implementation Climate
Two key informants believed the high literacy rates in South Africa greatly bene tted the implementation of the intervention. These high literacy rates helped adopt the program into the country. However, two key informants expressed that there was pushback from ground-level project implementors of MomConnect. As many of the workers felt the extra tasks in the project to be an added burden.

Policy & legislation External Policy and Incentives
One key informant highlighted that in the implementation of MomConnect they put the South African Normative Standards Framework into practice. Having such legislative standards made the project run smoother and surpass any technical challenges than a project where this framework was not upheld.
[1] This information was obtained from personal communication with one of the key informants.

Discussion
This qualitative study identi ed multiple barriers and facilitators to implementing the MomConnect project. The major emerging themes were those related to intervention characteristics of the MomConnect application, the scaling process, the inner setting and the outer setting of the MomConnect program.
These themes included: partnership relations; innovation costs and funding; operationalization of the innovation; and policy & legislation.

Partnership relations
The theme of partnership relations was identi ed to be important in respects to the implementation of MomConnect as it impacted the leadership and the overall implementation of the program. The program was greatly supported by the Minister of Health and there was a great amount of political will. The importance of the NDOH involvement and the political support as recognized by two other studies, helped to bring together an interdisciplinary group of partners and effectively develop the program (18,19). Similarly, the importance of political support from the Minister of Health was iterated to be critical in engaging stakeholders and management of the program (19). One important factor this study found that impacted partnership relations was the gap in technical leadership, there was no one to manage and overlook the technical partners. Without this technical lead there was limited understanding and overview between partners, departments, and technology applications. However, one of the critiques suggested by key informants was there was a great amount of burden placed on the ground-level staff as they had additional tasks in an already strict time frame. Additionally, it was also reported that these focal people were often not given the tools to navigate the innovation. Another study also recognized the burden many workers felt when they were introduced to the program, with registrations being more of a timeconsuming task for the staff (16).
One feature of the innovation that many key informants appreciated was the highly effective helpdesk. This feature provided the participants with direct and immediate feedback from the physicians or nurses. Studies have noted that with the complaints on the helpdesk have helped identify when systemic improvements within the drug system should be made (22). Likewise another mhealth project, WelTel, a texting intervention in Kenya adopted this feature and expressed how the patients felt someone cared and provided direct care (23). WelTel allows individuals to have an integrative experience with medical professionals virtually through the two-way texting platform, which greatly increased treatment adherence and was cost-effective (24). Overall, this improved the culture of care between the patients and medical staff as it became easier to access care.

Innovation costs and funding
Many key informants expressed that the high costs of data for the intervention was a limitation in scaling the program. South Africa has one of the highest rates of data. Although the costs of data were covered for the participants, the rates were still very expensive although it was supplemented by telecoms companies. The cost of USSD was discussed in another study to be covered completely and this meant there was no cost to the end-user, which encouraged women to register because it was free of charge (19). Similarly, mHealth programs are attributed to be successful due to their relatively cost-e cient use of technology. A study highlighted that the low-and middle-income countries including South Africa achieved high mobile penetration levels of 60%, thus this allowed mhealth programs to utilize existing technologies to save on costs (25). However, many mHealth innovations conclude that the nancial sustainability of the mHealth interventions to scale and mobilize funders could be further strengthened with concrete evidence. Funding for large-scaled project implementation with the government needs to be realistic because ICT projects cost more than what is initially planned (26). Achieving the 'Golidocks zone' is ideal, providing the technology requirements, scope for future scaling and its affordability, while also matching with the project's requirements for implementation (26). The Mobile for Reproductive Health (m4RH) project was an SMS based health information services. It was noted that the SMS data collection method utilized in the project was fast and cost-effective to learn more about the users and functionality of the platform (27). This greatly supported future scaling of the innovation and securing future funding (28).

Operationalization of the innovation
One major strength in the implementation of MomConnect program was the adaptations made with respect to the South African context, as this truly increased the accessibility of the program to the respective participants. One of the early-on decisions made when implementing MomConnect was to utilize a lot of existing technology to better scale and implement MomConnect. Another study similarly established that MomConnect made the early decision for universal coverage by designing the program to utilize basic technology that was easily accessible, such decisions strengthened MomConnect's ability to scale and implement (19). MomConnect utilized these pre-existing technologies and effectively adapted the MAMA project to allow for an easier implementation in the South African context. The users were able to adapt and accept the technology used because it was known, Brinkel and others noted in their study user motivation and technology acceptance were highly important and a basic requirement to implement a program effectively (29). As user attitude, acceptance and comprehensibility of mobile health in healthcare is important and sensitive (29). A study on the effectiveness of mHealth in medication adherence solutions showcased that about 40% of patients who were enrolled in sub-Saharan African antiretroviral therapy programs discontinued their program within 2 years (30). Patients in sub-Saharan Africa had limiting conditions that led to their poor drug adherence, this included bad weather, transportation, health conditions preventing them from leaving home, language barriers, etc. However, with the mobile technology SIMpill, patients receive alerts when they have forgotten to take their medication as prescribed and further alerts family members or friends if patients continue to neglect mediations (30). This mHealth program showed a signi cant success rate in treatment of 94% from participating in the system (31). Thus mHealth can provide services that can effectively adapt to the needs of the local community or population which increases the services sustainability and ability to scale. This study's results demonstrated how the positive feedback provided by the mothers was a large bene t in facilitating interest and scaling the program in South Africa. The women acknowledged that the messages were appropriate and felt personalized which made it feel like the physicians truly cared, this encouraged further scaling and implementation. Additionally, an evaluation study highlighted South Africa had high literacy rates among adults this made texting a common activity on mobile phones and was appropriate way for the mHealth program to be scaled (11).

Policy & legislation
Lastly, identifying the political policies and legislations upheld by MomConnect demonstrated how the project was effectively scaled and implemented among all these regions in South Africa. The project put the South African Health Normative Standards Framework (SAHNSF) into practice which provided a structure on how the project should be managed and scaled in South Africa. This made it easier to surpass any technical hurdles that could have come up if the framework were not utilized from the beginning of the project. In a study conducted on optimising health information systems in South Africa, SAHNSF was emphasized to be imperative in following standards and keeping interoperability among the different national health information systems (17). One study depicts the importance of de ning standards and protocols, having a framework allows the program to be implemented in a swift operational manner (32). In addition, with the application of mHealth projects that need to be implemented on large and national scales, it is important that national e-health strategies and frameworks are in place that can create the enabling environment for the implementation and scale of such projects as was the case for South Africa vis a viz the e health strategy and the National Health Normative Standards Framework for Inter-operability in eHealth (33).

Scaling mHealth programs
The use of policies or legislations to implement mHealth innovations are important to ensure it is swift and easy from the beginning of implementation as it provides a better understanding of any of the implementation hurdles that may come up. In the case of MomConnect the South African Health Normative Standards Framework (SAHNSF) was utilized. This framework proved to be effective as it helped prevent any technical hurdles, thus maintaining the use of this framework will be bene cial for the program overall to maintain the ability to effectively scale and implement the program further.
One critical factor in MomConnect that was revealed from this study is the role of champions in the program as these partnerships impacted the implementation and scaling of MomConnect. Having a strong top-down leadership was greatly important to help the program successfully scale in a timely manner. Many studies determined how impactful a government commitment from the beginning of a program can help scale the program nationally (34). MomConnect was one of the very few programs that was donor-funded program and adopted by the government (34). Stakeholders noted that MomConnect was unique in respects to the persistent political push from the government, which helped the program scale effectively. Though the top-down leadership helped scale the innovation quickly it also limited the amount of time available to establish bottom-up support for the program. As such ground-level workers such as nurses had criticisms about the facilities and how they were being conducted given the lack of support and resources they were given to run the program. In addition, the rapid scaling helped the program gain attention, but it limited the amount of time available to conduct a pilot or baseline study. Stakeholders note that the program was under pressure due to timeline restrictions by NDOH stakeholders. The baseline study is important in evaluating MomConnect's implementation and scaling opportunities. It determines whether the program can sustain funding and provide proof of return for the program's government stakeholders. One study, expressed their evaluation of the MomConnect program was limited due to the rapid nationwide scaling efforts, an adequate baseline data was not collected, this made it more di cult to showcase the bene ts of the program in comparison to other health interventions (10).
There have been calls for more evidence on the impact of mHealth interventions on health outcomes (6).
MomConnect was able to subside a lot of the traditional research steps because they had the attention of the national government. However, stakeholders have mentioned that impact evaluations (RCTs) on MomConnect are about to kick off, which will support the evidence base data on the program's impact. It is important for health programs to envision a bottom-up approach, targeting local needs and creating a more sustainable program with the knowledge and evidence. Conducting evaluation studies or pilots, collecting evidence and support from project stakeholders and local organizations can help a project scale more effectively and attract adequate funds from diverse donors. As indicated in another review, there is generally a paucity of data on the impact of mhealth on patient outcomes in LMICs (35).
Language translation was a key factor in the MomConnect program. Although the program was able to translate the content into the 11 o cial languages of South Africa, there were issues with the quality of the translation. Similarly, another study that assessed the experiences of users' from different language and cultural groups, found out the quality of the translation in the Xhosa language was di cult to understand and not all languages were available in every community (36). Because the women were unable to understand the translations offered in Xhosa they were forced to request the service in English which was not as convenient to understand, limiting engagement (36). mHealth projects have been proven to be successful when they are adapted to the local context and language, thus this is an important component of mHealth project implementation (37).
Another compelling theme based on this study was the high cost of data and the reliance on donors for appropriate funding. The high cost of data is an issue that has persisted among South Africa due to the monopoly telecommunication companies hold on market prices for data (38). One study reveals in South Africa's past efforts to revamp the telecommunications sector, they sold their stakes in state-owned networks to companies such as Telkom, who exercised their monopoly power by raising prices and ultimately betrayed the government (39). Although there was a signi cant adoption of the program it was not successfully implemented in all regions of South Africa, as one study states about 10% of the population had no access to cellphones (40). In a recent systematic review, high cell phone rates were highlighted as one of the impediments to transitioning to scale of mhealth projects in Africa (41). Thus, one recommendation to help reduce the dependence on telecom companies and donors to provide support and funding assistance, is utilizing mobile applications such as WhatsApp where there are no costs for SMS and is more consistent. This could help with scaling the project to other regions and registering more participants. Additionally, with the signi cant adoption rates of mobile phones with smartphone application use there is a higher chance of participants registering and keeping up with the program.

Strengths & Limitations
Our study showcases the viability of using the CFIR framework to identify common barriers and facilitators in the implementation of mHealth innovations. Its ability to utilize a standardized language to analyze the various barriers and facilitators mentioned by key informants. This framework was used as a part of the data analysis tools [see Additional File 2 Qualitative Study Checklist]. A potential limitation was our selection of a determinate framework (CFIR) as they have been criticized for their inadequacy in addressing causal mechanisms or how change takes place (42). However, the use of the CFIR allowed the findings to be placed in the context of the wider implementation research literature (43,44).
We also acknowledge that our stakeholder sample only involved stakeholders who were partners and implementers of the MomConnect program and not users of the innovation. However, we would like to recognize that there are existing studies that have examined the user perspectives and ability to use the program (35,45). One study noted that the users in MomConnect were consistently positive about the impact of the messages and their online access.

Conclusion
This qualitative study demonstrates that CFIR is a useful tool that can be used to guide the formative evaluation processes of mhealth programs. The barriers and facilitators identi ed under the CIFR domains can be used to build knowledge on how to adapt mhealth programs to national and local settings. Key areas of importance for mhealth programs as revealed by this study include (1)