We excluded formative research studies (ie, studies conducted before fully developing or implementing an intervention); study protocols; interventions involving computers or web-based health care (eg, websites); telehealth interventions (defined as consultation with a health worker via a mobile phone either through SMS or phone calls); and quantitative studies including randomized controlled trials, Likert scale surveys, and impact evaluations, as they did not provide in-depth reporting of qualitative factors affecting implementation. We excluded studies where mHealth was part of a larger complex intervention, studies from high-income countries, and studies that combined analysis of determinants across multiple countries where it was not possible to separate out findings from LMICs versus high-income countries.

The full search terms and strategy for the databases are detailed in Multimedia Appendix 1. Briefly, we included terms pertaining to LMICs, infectious diseases (eg, communicable disease), and mHealth terms (eg, mHealth, text message, mobile app). We did not include additional filters for qualitative versus quantitative studies.

One reviewer (JGO) screened the titles and abstracts of the search output for relevant studies. As a next step, we conducted full-text screening. Where eligibility criteria were unclear, final consensus on article eligibility was based on discussions with another member of the author team (HMY).

We extracted characteristics of the mHealth intervention including intervention setting (country, LMIC status, health care setting); intervention design; content and purpose; target disease and population; and its quantitative impact on health outcomes originally targeted by the intervention, as reported in the included qualitative study. This was to aid our interpretation of how the qualitative implementation determinants we identified may have affected targeted health outcomes. The original quantitative impact evaluations were not sourced for this review. We also extracted data on details of the qualitative implementation study setting, study population, research question, data collection method, and study size, as well as broad implementation determinants considered by the authors.

We performed a framework analysis of determinants affecting implementation based on the Tailored Implementation for Chronic Diseases (TICD) framework [25]. The TICD framework is a comprehensive checklist of determinants of clinical practice developed to inform implementation research projects that are tailored to local conditions [25,26]. The framework can be applied beyond contexts of chronic diseases, as the framework broadly focuses on health system components that determine quality of care. It identifies 7 key domains: guideline factors (clinical care guidelines or mHealth as a “guideline”); health worker factors; patient factors; professional interactions; incentives and resources; capacity for organizational change; and social, political, and legal factors. Its strength lies in its emphasis on ensuring tailoring to local conditions, which is valuable to consider for LMIC interventions, and inclusion of contextual (including political and legal) and patient factors affecting implementation. The TICD framework is a comprehensive health systems framework aligned with the systems thinking framework, which considers interactions with the broader context and patient needs in addition to the structural components of a health system [27].

The framework was used to initially categorize the reported determinants of mHealth intervention implementation and was used to structure the reported findings in the results. Additional emergent themes, such as intervention design and forgetfulness, were extracted.

There were 6 studies that reported on how lack of clarity and understanding of the intervention hindered its uptake. For example, users did not respond to mHealth intervention text message prompts because they were uncertain about how to use the personal identification number, were uncertain about how to respond to the message, or received little to no information about the background of the intervention [7,30,33,39,40,43]. Compatibility of the intervention with users’ past experiences with either the technology and its use in health care, or the intended change in health care practice, was a key facilitator that aided in implementation. Existing familiarity with the intended behavior promoted by the intervention—such as antiretroviral treatment adherence, HIV testing, or general interactions with the health care system—facilitated uptake [21,29,35,38,41]. The mHealth intervention in Babili et al’s study—WelTel, assessed for COVID-19 case and contact management—was previously implemented for HIV epidemic virtual care, which facilitated its implementation as users were familiar with the platform and its functionality [32]. Similarly, health workers commented on how an app’s alignment with existing practices of using the village clinic register increased the likelihood of engagement [29]. Existing mobile phone use or interest in the use of new technologies were additional facilitators, as users were already familiar with making appointments, making calls, or using SMS technology [8-10,32,33,36,37,42]. Trustworthiness of the source of the recommendation given by the mHealth intervention facilitated implementation, particularly when the mHealth intervention intended to aid health workers in improving their health care practice or disease management [10,29,32,35,40].

mHealth being perceived as useful by end users in improving existing health care practices facilitated the uptake and integration of the intervention. For instance, after initial use of the intervention, a perception that it might improve existing clinical practices, improve patient engagement with services, or relieve strain on the health system were key for implementation [9,29-32,35,36,39,40,43]. Health workers in Ide et al’s study perceived the app as advantageous over existing systems, as the intervention aided in conducting more accurate, error-free community case management of malaria, diarrhea, and pneumonia, which facilitated uptake [29].

Furthermore, for interventions for which the main users were health workers, attitudes toward the intervention were impacted by the perceived effect of the use of the intervention on the health worker’s reputation. Ellington et al identified that the perceived loss of trust between patients and health workers in the health worker’s ability to diagnose and treat patients due to their reliance on technology to deliver health care was a barrier to using the app [9]. In contrast, Ide et al commented on how the app facilitated perceived professionalism [29]. Twimukye et al commented on how the health worker’s use of the app improved how the patient perceived the health worker’s attention to detail and care [34].

Patients perceiving the intervention as beneficial in improving health outcomes through increased convenience, awareness, or reminders facilitated implementation of the intervention [34,38,41,43]. Adeagbo et al commented on how the app’s positive messages about HIV testing and adherence promoted users to adopt new behaviors by improving individual competency to make informed, healthy decisions concerning sexual health [41]. Language literacy was a barrier that limited engagement with 2 SMS intervention studies targeted at patients or community members [8,38].

Limited supervision and follow-up of the intervention implementation by the research team leading the intervention was noted as a barrier, as users requested feedback and confirmation of correct intervention use [30]. Lack of or limited access to training to provide necessary skills to effectively engage with the mHealth intervention was a barrier to implementation for health workers [9,36]. Access to professional training was a facilitator of implementation [10,29,40].

Several studies reported that access to resources and essential infrastructure influenced implementation. Specifically, 6 studies reported that poor telecommunications networks, problems with electricity, a lack of phone coverage, limited staffing, and a lack of equipment to implement the behavior change were barriers to implementation [10,30,33,37,40,42]. Network problems were particularly prominent in geographically remote areas. Other barriers included technology that repeatedly malfunctioned (such as periodic freezing and system crashes) and limited access to technology support systems to troubleshoot technological problems [7,29,32-34,36,38,42,43]. Access to technological resources and support in case of malfunctions facilitated implementation [33].

Financial instability, existing patient overflow, incompatible technological equipment, and length of appointment time within the clinic hindered the implementation of mHealth interventions [7,9,30,37,39]. Kaunda-Khangamwa et al reported that 90% of the health worker respondents blamed existing high workloads and drug stockouts as factors discouraging health workers to respond to SMS reminders that promoted infectious disease case management [30]. Similarly, Ellington et al’s study identified that the existing length of the appointment time was not compatible with mHealth intervention use as the time to complete a health assessment through the app took longer than the appointment duration [9]. Straw et al commented on the compatibility of the mHealth intervention with the existing organization functionality to facilitate normal workflow as a facilitator of implementation [40]. A lack of management and ownership of the intervention by health workers further contributed to a limited capacity for embedding the intervention within the health care facility [39].

One study commented that the costly nature of airing messages during the daytime and limited funding for the intervention were barriers to successful implementation [42]. Receiving national-level support on a political level including the Minister of Health or district leadership was a facilitator of implementation [9,10,32]. Babili et al commented on how the Rwandan government’s support of digitization across all governmental sectors by offering resources for implementation aided the adoption of the digital health intervention [32]. Furthermore, shared responsibility and corporate social responsibility felt among the wider community was an enabler, as the charitable community assisted in reducing costs and improving uptake [37].

Acceptance of the intervention was strongly linked to uptake of the intervention, which facilitated implementation of the intervention and its potential to improve health outcomes. Patients and health workers were influenced by the perceived benefits of the intervention in improving health outcomes, the extent of training, and previous experience with the guideline behavior or mHealth.

Perceived benefit of the intervention was influenced by ease of use and design of the intervention [9,10,30,32,34,35,37,41,42]. Where an app or SMS design engaged users in a way that improved health outcomes, patient engagement with health care services, or health care efficiency, participants were less likely to perceive the use of the mHealth technology as burdensome, facilitating successful implementation [9,29-32,34-36,38-41,43]. An intervention that provided a clear advantage compared to existing practice facilitated the perception of it being beneficial, as supported by the intervention’s design and ease of use. These findings were consistent with Davis’s Technology Acceptance Model and its application in health care, which suggests that perceived ease of use and usefulness both affect attitudes toward and use of technology [46,47]. Design considerations was an emergent factor unique to mHealth and technology-related interventions, emphasizing the value of iterative design of the mHealth intervention before implementation [9,10,30,34,35,37,41,42]. Further reviews reporting on mHealth in the context of antiretroviral therapy adherence and maternal health have identified the importance of intervention design, including tailoring SMS messages and the frequency of reminders [48,49].

Political support of the intervention ensured effective implementation; this included stakeholder buy-in and approval of the intervention, and support from health administrators on a district level [9,10,32]. The importance of understanding the need for the intervention to reduce disease burdens or improve health care services is supported by a previous review by Opoku, Stephani, and Quentin [50]. Therefore, regardless of the target user, perceived benefit of the intervention is crucial for implementation of the initiative, emphasizing the community-wide, integrated nature of mHealth interventions.

However, studies by Kaunda-Khangamwa et al and Mangam et al noted that, despite the users’ positive attitudes toward the intervention, factors such as lack of communication regarding intervention use and limited resources to implement the behavioral guideline hindered implementation [8,30]. Perceived value of the intervention and acceptance alone therefore cannot guarantee successful implementation and positive outcomes from an intervention.

Users who received communication and training on the intervention and its use before or during its implementation engaged well with the intervention and intended guideline practice [9,10,29,36]. Lack of awareness and clarity regarding how to engage with the intervention were consistently noted across studies with limited user engagement [30,33,43]. Sufficient training has been previously noted in a review as a contributing factor to mHealth intervention implementation, which is closely linked to the perceived ease of use [2]. For interventions aimed toward health workers, training was either provided by the intervention research team members or between health workers; follow-up on correct use was identified as being important for encouragement and continued intervention use [7,9,10,30,42].

Insights from this review can help shape health policies and identify key considerations when developing mHealth interventions to improve their efficacy and sustainability in improving health outcomes. A full list of recommendations as reported in each study is noted in Multimedia Appendix 2.

When interpreting findings, it is important to consider the different contexts within which mHealth interventions are implemented, as they must be tailored to the context. Several considerations are important during predevelopmental, developmental, and implementation phases of the mHealth intervention, as seen in Figure 4.

A thorough understanding of existing health and technology practices and social norms toward a disease are crucial to predict the feasibility of an mHealth intervention in a specific context. This understanding of the cultural context and structural factors—such as the broader health care workforce, capacity of existing infrastructure, and resource availability—can be achieved through formative research [42]. Communication and education about the importance of health care practices and mHealth’s ability to facilitate health can aid in the acceptance of the mHealth intervention.

There were 4 studies that performed prepilot testing and designed the intervention iteratively to maximize participant engagement with the intervention during implementation [7,8,31,42]. For example, through pilot testing, Mangam et al identified the need to incorporate interactive voice messaging in their SMS intervention, as many users were unable to read or understand the texts [8]. To complement the theoretical foundations of an intervention, studies emphasized the importance of stakeholder involvement in the iterative design process during the development of the initiative [2,8,15,16,19,20,42].

As most interventions require a change in knowledge or behavior, it is important to communicate the benefit of the mHealth intervention compared to existing practice, prior to and during implementation. Increased training will increase confident user engagement with the intervention for long-term improvement of health outcomes [10,33,36]. Supplying mobile devices and offering technological support were recommended by studies to mitigate technological barriers [9,10,36].

It is important to tailor mHealth interventions to complement existing health services and face-to-face practices to optimize the desired health outcome. Adeagbo et al commented on how the mHealth intervention alone was insufficient in completely removing the barrier of accessing HIV testing and health care services [41]. Mangam et al discussed how future mHealth-based mobile communication should complement the community’s existing familiarity with interpersonal communication, particularly as their SMS notifications of health prevention measures were not met with improvement compared to the non-mHealth, face-to-face status quo [8]. An example of complementing mHealth with non–technology-based communication is seen in the study by Bhattarai et al, who paired SMS text messaging with pamphlets [37].