A Proposed Framework of Factors Affecting Patient Adoption of mHealth in the Developing World.

Patient perceptions and experiences of mobile health (m-health) systems have been recognised as an important element to consider in the adoption of m-health based technologies. Though much research supports this, published studies on m-health use by patients appear to highlight these issues in an isolated rather than holistic manner. Consequently, there is no encompassing framework that serves as a guide for effective implementation and maximum adoption of m-health from patients’ perspective in the developing world. This review documents patient adoption issues, and uses these to develop a framework of patient adoption issues for m-health in the developing world. A structured literature search was conducted using PubMed and Scopus. For PubMed a consolidated search string combined ‘MESH’ terms and ‘All Fields’ terms for selected keywords. For Scopus an equally consolidated search string was used. The searches were restricted to articles in English during the period January 2000 - December2018 and relevant to the developing world. Duplicate articles were removed. Titles and abstracts were screened by all authors for inclusion, and studies that met the inclusion criteria were selected for full-text review. Review and data abstraction was performed by two authors. Fifty four articles reported factors that impact patient adoption. Review and data abstraction identified specific factors, initially classified under 22 categories, that promote or impede m-health adoption in the developing world. Continued iterative review and discussion reduced these to 7 primary categories, with 21 sub-categories. The review showed: great inconsistency in the approach and tools used in published studies; multiple factors impact patient adoption of m-health in the developing world; the specific factors vary from setting to setting (e.g., country, rurality, mobile device type) and by recency of findings. Successful adoption of m-health by patients in the developing world critically depends on addressing the factors identified in the proposed framework, and assessing

4 developing world. This expectation will not be met unless factors that affect adoption of m-health by patients are systematically investigated and addressed. The aim of this study was to identify those factors that enable or impede adoption of m-health by patients in the developing world, and synthesise them into a practical framework. The study findings will inform policy and help facilitate future implementation of m-health in the developing world for patients. The searches were restricted to articles in English during the period 2000-2018 inclusive and relevant to the developing world. Duplicate articles were removed. Titles and abstracts were screened by all authors, and those studies that met the inclusion criteria, based on consensus, were selected for full-text review and data abstraction. Inclusion criteria were: resources addressed mobile health or m-Health, used cell or mobile phones in the context of patients, and identified factors that facilitate or impede m-health adoption in the developing world. Resources were excluded if they addressed telemedicine, telehealth, or e-health more broadly, or were focused on the developed world, healthcare workers, or healthcare organisations. Hand searching was also performed. Final selection of resources was discussed by all three authors, and study inclusion based on consensus. Full text review and data abstraction was performed by two authors.

Results
The combined searches from PubMed (576), Scopus (326) and hand searches (40) returned 9 4 2 resources, of which 54 studies met the study inclusion and exclusion criteria ( Figure   1).
Further screening of abstracted data identified specific factors that promoted or impeded patient m-health adoption in the developing world. The findings were initially categorised under 22 thematic headings. Continued iterative review of these 22 thematic categories reduced them to seven primary categories, with 21 sub-categories (Table 1). Table 1. Final grouping of factors impacting patient adoption of m-health in the developing world, sorted by frequency of reporting in the literature.

Cost and ownership
Cost and ownership issues dominated the literature findings and related to the impact ownership of a mobile device, and included access to mobile devices and their affordability in terms of fixed (purchase) and variable (use) costs to patients. These issues were collated into four subcategories. Some issues were related to cost, affordability, and incentives. When the operating costs are not affordable [5][6][7][8][9][10]91], and patients must buy airtime (i.e., where the patient pays for the calls or messages he/she receives or makes in accessing m-healthcare from the service provider [11]), m-health is likely to fail, unless the patient is able or willing to pay [13,14]. Patients may accept the technology if the cost of owning and operating it is considered a c c e p t a b l e [13,43]. Such challenges can be reduced by introducing financial incentives to mitigate the cost burden on the patient [12]. Overall there was a lack of evidence of the cost-benefit of m-health systems which also challenges their implementation.
Other issues related to actual cell phone ownership, which was identified as a critical determinant in the adoption and uptake of m-health services [16,8,11], included owning the appropriate mobile phone with the required technology [17][18][19]. One study noted that globally, women are 21% less likely to own a mobile phone than men [33]. It was noted that ownership also influenced behaviour, with patients who received m-health services on their own phones considering it more acceptable, compared to those who shared the phone with others [21,22,77].
Another issue was the ability to keep a mobile phone charged and connected, and the associated costs. In many developing countries power was described as irregular with rural areas being most affected [35,47]. Keeping a mobile phone charged was problematic [7,49,54] and it was common to find people paying to charge their phones at street side vendors [49,60]. Likewise, phone maintenance in the event of a fault was an equally important factor that might jeopardise adoption [24, 45,47].
Sharing of mobile devices was the primary issue identified under access to mobile devices.
Many projects relied upon shared use of cell phones [14,23,50]. Although the absolute proportion of shared devices varied, for example from 21% [ 33] to 51.4% [ 15], it was recognised as a limitation to implementation. Related to cost and ownership was user characteristic issues, described below.

User characteristics
This was the next most commonly reported category and addressed the socio-cultural beliefs, perceptions, and overall setting of patients as factors that impact m-health adoption. The f our sub-categories, included the impact of socio-cultural practices and beliefs, and gender issues which were noted in many studies. ICT use in low-income countries is lower among females [8, 39,78] and a 'gatekeeper effect' was noted in several studies with women requiring permission from their parents, husband or partner to use a cellphone [4,33,44]. This was exacerbated by being ashamed to raise issues about women's diseases with their gatekeeper [4] or fear of punishment if they accessed a phone without permission [60]. Other cultural factors impacted cellphone use, with boysunlike girls -being allowed to be inquisitive and seek out information about sexual matters [44], and restricted use being enforced through fear of "inappropriate" calling with the opposite sex [60]. In Tanzania, men prevented their wives from owning mobile phones because they believed it facilitated sexual unfaithfulness [92].
Studies reported participants from adolescents to the aged, of both genders, and broad levels of education [47,38,20,60,22]. Some studies suggested that age and gender of patients should be considered when implementing m-health systems, with different age groups having preferences for certain multimedia elements [20,23], and women given less priority in male dominated communities [4,60]. Others reported that children, the elderly and the illiterate needed assistance to initiate a service request [38], or appropriate training for them to use the device [93]. Others found all age groups, genders, and education levels functioned well with m-health interventions [15,22].
Men dominated mobile phone use [8,22,49,20,78,79], although this varied by country [4,60]. Reasons included the gatekeeper effect, but also the lack of primary or higher education for women [15,49]. It was suggested that an appropriate age target for minimally educated women to use m-health would be 17 to 63 years [15,29,33] but in certain parts of the developing world older women were more likely to own and use a mobile phone for m-health than younger women [47], and in South Africa women are the dominant users [60]. Urban women found evening m-health services more convenient and rural women preferred daytime services [14,38,90].
Acceptability and perception of use, and the willingness and ability of patients to use m-Health were identified as issues impacting implementation. m-health solutions were more readily accepted and adopted by patients when they addressed a patient recognised health need [32, 4,7], were considered acceptable and useful to them [10, 37,13,22], were friendly and easy to use [34,37,45], and used appropriate multimedia modes (selected for effective communication by the target user group, whether text message, audio, video, animation, or pictures [17]). It was noted that voice accommodated those with low literacy and helped to build trust [14,36,7], while SMS messaging accommodated those with a slightly higher level of literacy [36,21,22,26]. Services that did not address patients' perceived needs impacted motivation to use the service [4].
T h e competence and readiness of healthcare workers to use technology to deliver an mhealth solution also impacted patient adoption. Patients expected healthcare workers to respond to any requests in a timely manner [36,50], and to have the requisite competencies to deliver the m-health services [49], highlighting the need for available and efficient training in the use and management of any m-health technology [27].

Language and Literacy
These were considered primary issues for successful m-health adoption [20,26,30]. The clinical benefits of conversing with a patient in their mother tongue, whether written or spoken were noted [72,73] and m-health adoption was affected when patients were not confident in communicating in a language they did not normally use or understand [5,29].
It was suggested that the National official language, which generally serves the interest of the majority, should be used in the deployment of m-health systems [16].
To participate in m-health services, patients need to be literate both in the traditional sense (able to read, write, and speak in their Mother language), but also in a broader sense (able to understand the technical needs to effectively use a mobile device, and able to understand their health issues and treatment) [ 16,32]. In poor rural areas where education levels are often lower [4] people may require the assistance of a family or community member to understand the content of a message sent to them [89]. In general m-health requires minimum literacy on the part of patients for its adoption [17], particularly when patients are appropriately trained to apply the technology [15,20,23,28,[32][33][34].

Infrastructure
The lack of, or insufficient accessibility to, digital infrastructure in the developing world was noted [1,44]. Unreliable or poor quality infrastructure [1,14,17,41,44,85] leading to mobile network fluctuations [15] or inadequate cellular signal [43], and unresolved technical issues [30] were identified impediments to m-health adoption. Technology infrastructure upgrade may be required before m-health implementation to provide dependable network infrastructure, remote accessibility, and seamless connectivity [10, 28,38,45,46]. In addition, m-Health interventions are dependent upon reliable electric power [38], although alternate innovative means such as 'pedal power' and solar power have been used to a modest degree [67,68] .
Social networks highlighting m-health services provided effective publicity and promoted implementation [47].
Collaboration and Funding m-Health system implementation and patient adoption often relies on the fusion of various independent systems and strong stakeholder collaboration [50,22]. Relevant stakeholder institutions must be willing to actively collaborate and share resources for success. This requires and an appropriate institutional setting that promotes such integration [8], where existing communities, healthcare facilities, technology infrastructure, and other service provider platforms are linked to each other in a seamless connectivity [15,22,69].
Collaboration is also necessary to identify and address patients' challenges during implementation [69]. Very clear stakeholder responsibilities are required to avoid conflict and service ambiguity. The required level of integration can be made possible when there is an existing institutional framework supporting the exchange backed by a comprehensive policy regime. The need to engage policy makers even at the stage of design through to implementation and ensuring that the system does not run in isolation to similar national or local interventions is critical to adoption.
As the government of most countries is either the sole or primary provider -or payer -of healthcare services, government facilitation and sponsorship of m-health implementations will influence adoption by patients. Government or private sponsorship (or perception of the same) is crucial for m-health adoption among patients [13]. For some patients, just involvement of government is enough to give the project some credibility.
Community ownership of m-health programmes affects patient adoption. Mbuagbaw et al.
(2014) [27] found that strong community involvement driven by advocacy during home and hospital visits, coupled with active engagement with community leaders, was an important element for patients' adoption. Advocacy both at the level of the community and the healthcare provider is crucial for the undecided user to make up her mind [48].
This system should be implemented to reflect the local contexts in which it is deployed.
There must be an effort at mobilising resources from the community to support the project internally rather than a concentration on external funding sources, if the project must succeed [27]. There must be a fusion between the community and the facility-based services for the system to reflect community context and ownership [22,32].
Availability of sustainable funding. The success of m-health systems depends on securing sustainable funding. Some of this funding will come from external sources and as such may not be reliable. For sustainability there should be mobilisation of community resources as well funding from external (government) sources, and an avoidance of over reliance on less secure external funding [27] (e.g., faith-based organisations and other NGOs).
There is a high probability of m-health adoption when there is collaboration among relevant governmental and non-governmental agencies, local community organizations, and funding agencies to reduce cost and promote system ownership [86].

Governance
Governance encompasses all of the processes that wield influence over a social system (country, organisation, village, tribe) through tools such as laws, regulations, or social norms. The patient-related m-health adoption governance issues include legal, regulatory, and ethical issues including data security aspects to maintain the privacy and confidentiality of healthcare information, records and communications [69,70]. Each of these were noted to impact patient adoption of m-health [13,22,30,36].
An enabling regulatory setting requires suitable laws, policies, and a framework that supports m-health adoption by patients. Legal and regulatory challenges to successful mhealth adoption were noted [69], requiring appropriate responses using policies, standards, and regulations [70]. The implementation of a regulatory policy must be the responsibility of all stakeholders especially the regulator and the healthcare provider [69].
Ethical Setting. Maintaining the privacy of data during collection, storage, and sharing for all patient groups was noted as critical for the adoption and sustainability of m-health systems [9, 11,21,24,30,36]. Success instilled confidence in patients [6,11,25,37] while failure had a negative impact [27]. Protecting m-health devices against unauthorised access and having effective standard operating procedures (SOPs) was also noted [38].
Some patients wanted all communications sent directly to their personal mobile devices without going through a human intermediary to guarantee confidentiality [92]. Yet where a patient does not own a mobile phone, caregivers must be contacted to make the information available to the patient; some considered this a breach of confidentiality [42,90], because mobile phone is considered a preferred medium for communicating sensitive issues [40]. Confidentiality concerns were even noted regarding asking for sociodemographic information from patients [14].

System Utility
The seventh most common category refers to how useful or beneficial an m-health solution is to patients. Three sub-categories were identified: Demonstrating clear benefit to patients, the effectiveness of the system, and evaluation and monitoring.
Demonstrating clear benefit to patients. m-Health systems were found to be more readily adopted when they demonstrated clear benefits to patients [12,27,48]. Successful adoption may be limited if there is a lack of awareness of the benefits to the general public [29]. Some authors identified that new or prospective participants may want to know if evidence exists of the benefits of m-health to patients [76,75]. Patients will adopt services that address their needs and are considered satisfactory [6]. Mobile phone functions patients viewed as beneficial included automated reminder systems, drug adherence alarms, and appointment reminders from care givers [84,85].
Effectiveness of the system. Patients must feel comfortable that an m-health system will successfully deliver what they want, and will avoid adopting an m-health system they are  [50,28]. Similarly, the use of inconsistent indicators and poor evaluation methods made cost-effective uptake of m-health in the developing world difficult to prove [28]. Additionally, adoption of m-health services is facilitated through awareness (marketing and publicity of benefits and capabilities) [14,33], and managing expectations to ensure they are realistic [25,29].

Proposed Framework
Based on the above results, it was considered that for m-health to be maximally adopted by patients in the developing world a framework in which all the above identified factors are captured must be used to guide the implementation. Below is the proposed framework.

Discussion
The study has highlighted great inconsistency of approach, tools, and indicators used in published studies that report on factors that impact patient adoption of m-health in the developing world. The study also revealed a plethora of specific factors that differed from study to study, that varied in terms of their impactfulness from setting to setting, and whose relevance was questionable given the marked change in technology over time.
After iterative review by the authors this spectrum of specific factors was reduced to 7 primary categories, with 21 sub-categories. The 7 primary categories were used to develop the proposed framework for patient adoption of m-health in the developing world.
Successful adoption of m-health by patients in the developing world will depend on assessing and addressing the factors comprising the framework before attempting to implement m-health initiatives in any specific setting.
The literature showed that certain socio-cultural practices and beliefs can serve as barriers to m-health adoption, requiring the sociocultural context and setting of a community or town to be understood and considered. In addition to sociocultural beliefs is the sometimes high and unrealistic expectations of the capabilities of an m-health system by patients, which can also cause it to fail [14]. Consequently, understanding and managing patient expectations is very important to success [10,48]. Some gender factors were identified as location specific. For example, in Ghana and Malawi, males were more likely to own and use mobile phones than females, while the opposite was true in South Africa, a pattern that has persisted over time [47,60].
Poor awareness of m-health was prevalent and impacted adoption. For example, of over 4500 adolescents in Ghana, Malawi and South Africa "only a handful had ever heard of mhealth programmes, let alone participated in them" despite using their cellphones creatively and strategically to seek healthcare [47]. Ironically, patients may not adopt mhealth if their expectations of m-health capabilities are unrealistic [25,29]. Consequently, efforts to publicise and make patients aware of the benefits and capabilities of an mhealth services are necessary [14,33].
Access to mobile devices (and accessories) is considered a precursor to successful mhealth implementation [15,81,92], but the basic cell or feature phone ("dumb" phones) still predominates in developing countries [3]. The growing tendency for m-health solutions and services to be smartphone and Internet dependent adds to the cost of ownership. This includes both the base cost for purchase of a suitable device, keeping the battery charged, but then also the cost of participating in m-health services. What is the patient's ability and willingness to afford airtime, SMS messaging, and data use to participate in any m-health services accessed? Corporate and project-based tactics employed to ameliorate these impacts have included reduced or subsidised devices and communication costs that can facilitate m-health uptake and use, but for sustained use by patients how effective and appropriate can such approaches be?
The ubiquity of cellphones was often stated or implied through reporting of high ownership figures. This is perplexing. In the developing world 12.37% of people live on less than USD$ 1.9 purchasing power parity [63]. Furthermore, poverty is associated with an increased burden of disease [26] and low educational levels which reduces income generating capacity thus increasing their likelihood to share mobile devices [8, 36,50,78].
To these people ownerships of a mobile phone or other mobile device is a luxury and the cost of ownership a stark impediment that may widen the digital divide and impede mhealth solutions and services [87]. Mobile network service coverage in the developing world differs from country to country, and even within countries urban cities have better penetration compared to rural towns and villages.
There are over 7,000 languages in the World and in Africa alone there are over 2,000 languages with more than 500 in Nigeria alone [71]. English speakers predominate as users of the www [74] and, as a result, much of the content remains in English, which poses a concern if patients are to relate to and adopt m-health content. Most people in the developing world will choose their local languages over English for information dissemination because they consider English difficult to understand [88]. The issue of language used by healthcare service providers to communicate to patients, whether written or spoken, can become a barrier to care and m-health adoption when patients are not confident in communicating in its use or understanding [5,29]. It has often suggested that the National official language, which generally served the interest of the majority, should be used in the deployment of m-health systems [16,32,17], but this may severely limit the utility of m-health solutions for rural and remote populations in the developing world.
Relevant stakeholder institutions whose services are necessary for effective m-health delivery must collaborate [50,22] based on a predefined inter-institutional framework [8] to effectively address the concerns of patients [69]. This may involve the support from government, private sector, and community leaders. Among these three, the government involvement is very crucial since it provides a sense of available funding and system credibility.
There is the need to guarantee the integrity of m-health systems by protecting patients records and communications. Protection of such information is possible in an environment of adequate legal regime and education, and the strict adherence of ethical standards [9,11,69,70].
Based on the review conducted, evidence shows that for m-health to be maximally adopted by patients in the developing world a framework in which all the above identified factors are captured must be used to guide the implementation and promote adoption. Accepting the above, and in order to achieve increased adoption of m-health by patients in the developing world, it is recommended that:

Limitations
Prior to any m-health implementation factors that may impact adoption by patients in a specific setting and for a specific m-health solution must be assessed.
The assessment must be holistic, considering all appropriate and relevant factors described in the themes and sub-categories of the 'Patient m-health Adoption Framework' It must not be assumed that factors impactful to one implementation will be relevant to implementation of a different m-health solution, or the same solution in a different setting.

Conclusion
This review shows that the success of m-health project implementation and adoption by patients in the developing world critically depends on addressing key factors identified in the 'Patient m-health Adoption Framework'. The framework will serve as the basis for informed decisions by stakeholders (policy makers, implementers, researchers, evaluators) and provide the necessary blue print for future successful m-health implementation in the developing world for patients.

Declarations
Authors' original submitted files for images

Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions MA contributed to the conception, acquisition, and interpretation of data, drafting, and revision. RES and MM contributed to the design, acquisition, and analysis of data, finalization and approval for publication. All authors read and approved the final manuscript.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.
Data Abrstraction tool for Patient.xlsx