UBC Faculty Research and Publications

Operationalizing mHealth to improve patient care: a qualitative implementation science evaluation of… Bardosh, Kevin L; Murray, Melanie; Khaemba, Antony M; Smillie, Kirsten; Lester, Richard Dec 6, 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12992_2017_Article_311.pdf [ 719.62kB ]
JSON: 52383-1.0361568.json
JSON-LD: 52383-1.0361568-ld.json
RDF/XML (Pretty): 52383-1.0361568-rdf.xml
RDF/JSON: 52383-1.0361568-rdf.json
Turtle: 52383-1.0361568-turtle.txt
N-Triples: 52383-1.0361568-rdf-ntriples.txt
Original Record: 52383-1.0361568-source.json
Full Text

Full Text

RESEARCH Open AccessOperationalizing mHealthinterests, determining appropriate financing pathways, maintaining network growth, and “packaging” the intervention forologicalplatformschnologicalBardosh et al. Globalization and Health  (2017) 13:87 DOI 10.1186/s12992-017-0311-zFlorida, 2055 Mowry Road, Gainesville, FL 32610, USAFull list of author information is available at the end of the article* Correspondence: bardosh_kevin@hotmail.com; kevinbardosh@ufl.edu1Department of Anthropology & Emerging Pathogens Institute, University ofmoving from mHealth pilots to scale is a difficult, context-specific process that couples social and techninnovation. Fostering new organizational partnerships and ways of learning are paramount, as mHealthstraddle the world of research, industry and public health. Partnerships need to avoid the perils of the tefix, and engage the structural barriers that mediate people’s health and access to services.Keywords: Mobile health, mHealth, WelTel, Sms, HIV, Tb, Maternal and child health, ANC, Kenya, Canadaimpact and relevance.Conclusions: Our comparative case study, of a unique transnational mobile health research network, revealed thatpatient care: a qualitative implementationscience evaluation of the WelTel textingintervention in Canada and KenyaKevin Louis Bardosh1* , Melanie Murray2,3,4, Antony M. Khaemba5, Kirsten Smillie2 and Richard Lester2AbstractBackground: Mobile health (mHealth) applications have proliferated across the globe with much enthusiasm,although few have reached scale and shown public health impact. In this study, we explored how differentcontextual factors influenced the implementation, effectiveness and potential for scale-up of WelTel, an easy-to-use and evidence-based mHealth intervention. WelTel uses two-way SMS communication to improvepatient adherence to medication and engagement in care, and has been developed and tested in Canadaand Kenya.Methods: We used a comparative qualitative case study design, which drew on 32 key informant interviews,conducted in 2016, with stakeholders involved in six WelTel projects. Our research was guided by theConsolidated Framework for Implementation Research (CFIR), a meta-theoretical framework, and our analysisrelied on a modified approach to grounded theory, which allowed us to compare findings across theseprojects.Results: We found that WelTel had positive influences on the “culture of care” at local clinics and hospitals inCanada and Kenya, many of which stretched beyond the immediate patient-client relationship to influencewider organizational systems. However, these were mediated by clinician norms and practices, the availabilityof local champion staff, the receptivity and capacity of local management, and the particular characteristics ofthe technology platform, including the ability for adaptation and co-design. We also found that scale-up wasinfluenced by different forms of data and evidence, which played important roles in legitimization and partnershipbuilding. Even with robust research evidence, scale-up was viewed as a precarious and uncertain process, embeddedwithin the wider politics and financing of Canadian and Kenyan health systems. Challenges included juggling different© The Author(s). 2017 Open Access This articInternational License (http://creativecommonsreproduction in any medium, provided you gthe Creative Commons license, and indicate if(http://creativecommons.org/publicdomain/zeto improvele is distributed under the terms of the Creative Commons Attribution 4.0.org/licenses/by/4.0/), which permits unrestricted use, distribution, andive appropriate credit to the original author(s) and the source, provide a link tochanges were made. The Creative Commons Public Domain Dedication waiverro/1.0/) applies to the data made available in this article, unless otherwise stated.Bardosh et al. Globalization and Health  (2017) 13:87 Page 2 of 15BackgroundKeep a watch also on the faults of the patients, whichoften make them lie about the taking of thingsprescribed. For through not taking disagreeable drinks,purgative or other, they sometimes die.– Hippocrates, DecorumSince the days of Hippocrates, doctors have prescribedmedicines that patients have just as quickly refused totake. Historical examples aside, non-adherence is amajor global health problem, responsible for increased(and un-necessary) morbidity and mortality across a vastnumber of acute and chronic diseases around the globe[1]. An estimated 50% of patients today do not takemedication as prescribed, costing health systems billionsof dollars annually [1, 2].Just as the outcomes of non-adherence are complex,so too are its causes; concentrating on the “lying” patienthas all-too-often obscured this fact. HIV is an instructiveexample; over 90% of worldwide infections occur inresource-limited settings and an estimated 20% of pa-tients are lost to follow-up within 12 months of initiat-ing therapy [3]. A plethora of processes are involved andinclude (among others): socio-economic factors, health-care system characteristics, patient social networks, cul-tural models of health and disease, personal and psycho-logical factors, clinic factors and drug regimencharacteristics [4].While the adherence challenge defies simplistic explana-tions, efforts to address it have largely focused on trying tochange patient and healthcare provider behavior. Oster-berg and Blaschke [2] divided these into four broad cat-egories: patient education, dosing schedule changes,changes in staff practices and improved communicationbetween clinicians and patients. Approaches range from:changes in care routines, more information, counseling,family therapy, self-monitoring, reminders and reinforce-ments [5]. Behavioral research on these interventions hashighlighted multifaceted challenges associated with facili-tating change; there is no “magic bullet.”Within this context, the mobile phone technologyrevolution may offer a unique opportunity to build onthese experiences through greater connectivity and data.This has generated significant interest from industry,government and public health practitioners. Mobilephone subscriptions have reached more than 6 billion,and 80% of new subscribers are from low or middle-income countries [6]. More people have access to a mo-bile phone than to adequate sanitation or clean runningwater.The most common mobile health (mHealth) approachis a variety of daily and weekly one- or two-way SMS(Short Message Service) communication interventionsthat encourage patients to take their medication ([7, 8]).Other related innovations include digital technologies,such as electronic drug monitors, video observed ther-apy (VOT) and smartphone adherence apps.MHealth has attracted much attention, but the fieldhas been likened to the “Wild West.” Despite the ubi-quity of cheap cell-phones, medical applications have yetto be widely adopted in the health policy domain or in-tegrated into health systems beyond the realm of enthu-siasm [9, 10]. Evaluations that go beyond the level ofefficacy to explore issues of effectiveness and scale-uphave also been few and far between ([11, 12]). As Tom-linson et al. [12] noted: “The current wave of mHealthinterventions are the equivalent of black boxes. Eachsmall entrepreneur or researcher includes whatever bellsand whistles that their funding allows in an attempt todemonstrate efficacy.” Moving from efficacy to effective-ness trials, as the transition to scale process demands,requires considering how interventions, once devel-oped, diffuse in the “real-world.” This includes under-standing how social, cultural, technological andpolitical factors and processes, operating at differentlevels, influence and mediate the “bells and whistles”that have been developed. In global health, this has be-come increasingly known as an “implementation sci-ence” approach [13, 14], and is also a central focus formedical anthropologists.In this paper, we explore the implementation of a two-way SMS intervention (WelTel) in both the GlobalNorth (Canada) and Global South (Kenya). The WelTelmodel is one of the first examples demonstrating impactfrom mHealth on medication adherence to Antiretro-viral Therapy (ART) and HIV viral suppression, througha randomized controlled trial in Kenya [15]. Since then,it has been expanded to include further work on HIV[16], tuberculosis [17], Maternal and Child Health [18],and Asthma [19]. This has included working with HIV-positive drug users in Canada [20] and remote pastoral-ists in northern Kenya. The WelTel InternationalmHealth Society, a not-for-profit organization, wasfounded to scale-up the WelTel service in Africa (http://www.weltel.org).The core of the technology is a simple “ask, don’t tell”approach to patient care developed through consulta-tions, pilot testing and formative qualitative research[21, 22]. Patients receive a weekly text message that asks,“Are you okay?” (or “Mambo?” in Kiswahili). The mes-sage is sent every Monday and patients have 48 h to re-spond that they are either “well” (i.e. “Sawa” inKiswahili) or if they have an issue or problem to discuss(i.e. “Shida” in Kiswahili). Another text is sent on Wed-nesday to remind clients that have not responded, andfollow-up phone calls are initiated to contact the patientif no contact is made (see Fig. 1 in English). A keyfeature of the WelTel service is that it is easy-to-use andmanage.Here we report the results of a qualitative study ex-ploring how the WelTel intervention was perceived, dif-fused, adopted and used by different health systemactors in Canada and Kenya. Our aim was to provide aunique, comparative (Global North and South) perspec-tive on the “real-world” complexity (or “messiness”) ofimplementation, and on the challenges and potential forscale-up, of an established mHealth application.MethodsProject portfolio: The WelTel networkThis study sought to comparatively explore enabling fac-tors and challenges associated with implementationacross a number of related but different projects. Intotal, we conducted interviews with stakeholders in-volved in six ongoing WelTel projects in Canada andKenya: WelTel eAsthma, WelTel Kenya-2 Grand Chal-lenges Canada (GCC), Cedar Project, WelTel Oak Tree,WelTel Retain and WelTel LTB1 (see Fig. 2).The primary focus was on the most developed pro-jects: WelTel in Kenya’s Northern Arid Lands (WelTelKenya2 GCC) for HIV and Maternal, Neonatal andBardosh et al. Globalization and Health  (2017) 13:87 Page 3 of 15Child Health (MNCH), and to a lesser degree, on twoHIV projects in British Columbia, Canada (Oak Treeand Cedar). The Kenyan project was primarily aimed atFig. 1 The core WelTel texting intervention for HIV/AIDSbeginning to expand to other sites.3) Cedar Project built on existing HIV action researchfor indigenous people in BC, Canada [20]. Theproject has an ongoing cohort of 200 HIV andHepatitis C vulnerable patients. The WelTelintervention was implemented to better connectyoung indigenous people who use drugs to Cedarcase managers in a community-based setting. Itincluded the weekly SMS texting service, andprovision of free phones to vulnerable patients.4) WelTel Oak Tree was a clinical effectiveness studyexploring WelTel’s impact on clinical outcomesamong vulnerable HIV-positive patients on ART.The project was run out of Oak Tree clinic at BCWomen’s Hospital in Canada, and built on a suc-cessful pilot project [21, 22]. The study enrolled85 HIV positive participants who were consideredto be vulnerable and at risk for loss to engagement.It included the weekly SMS service, and provision ofa free phone with text-message plan where required.In addition to clinical outcomes, this study collecteddetailed data on health care provider utilization andcost.5) WelTel Retain was a National Institutes of Health(NIH) funded RCT at Kibera Community HealthCentre in Nairobi, Kenya. The study exploredwhether the WelTel intervention improves retentionImarisha project, a major health sector NGO inKenya funded by USAID. The project aimed toscale-up in the Northern Arid Lands (NALs), aremote pastoralist region with poor healthindicators. At the time of research, WelTel wasbeing used for nearly 700 HIV patients and some1600 pregnant women in Isiolo District Hospital(IDH), a referral hospital in Isiolo County home toroughly 200,000 people. Activities were justscaling-up and finding ways to integrate the servicewithin the local health system, while Oak Tree andCedar were still very much focused on generating evi-dence and proof-of-concept. We also explored the rela-tionships between the Kenyan and Canadian projects.Projects are described throughout the text; here we pro-vide a short summary:1) WelTel eAsthma was a small-scale RandomizedControlled trial (RCT) with patients with severeAsthma in BC, Canada. The trial included weeklytexts to patients and a web-based platform thatprovides access to patient action plans.2) WelTel Kenya2 Grand Challenges Canada was a$2 million dollar “transition to scale” investment inpartnership with Amref Health Africa/ Aphia plusin care of HIV-infected individuals (n = 700) whohave not yet started ARTs [16, 17].Bardosh et al. Globalization and Health  (2017) 13:87 Page 4 of 156) WelTel LTB1 was a two-year RCT study toinvestigate the impact of WelTel on Latent Tubercu-losis Infection (LTBI) treatment completion rates.The project was implemented in collaboration withthe BC-CDC and implemented at two clinics inCanada [16, 17].Conceptual frameworkResearch was informed by the Consolidated Frameworkfor Implementation Research (CFIR), a meta-theoreticalframework particularly well-suited to a comparative,cross-project evaluation. The CFIR includes five do-mains (intervention characteristics, outer setting, innersetting, characteristics of the individuals involved, andFig. 2 WelTel intervention projectsthe process of implementation), which are divided intoover 30 different constructs, or “sub-domains” (see [14]).Key informant interviewsBased on the CFIR, our study methodology involved acomparative case study design. To guide this, we devel-oped three key informant interview guides, for healthadministers/managers, researchers and clinicians to beused across the various projects. These were divided intofive sections: impressions before implementation, im-pressions during the early stages of implementation, theintervention-health system interface, the functionality ofthe technology platform and scaling-up.We conducted 32 key informant interviews in BritishColumbia, Canada (11), and in Isiolo and Nairobi, Kenya(21), between February and April 2016 (see Table 1). Wepurposively selected our informants to cover a range ofperspectives. Interviews lasted between 45 min to one-and-a-half hours. All interviews were conducted in pri-vate, and data collection included manual notes. Consentforms were signed for formal interviews, although wesupplemented these with a more ethnographic approach,generating data through casual conversations at IsioloDistrict Hospital (IDH), with other stakeholders andwith the WelTel team in Canada.Semi-structured interviews were done with nine re-searchers involved in current projects. In Canada, inter-views also included two clinic staff responsible formanaging the platform. A total of eight WelTel staff inKenya were interviewed. Research at Isiolo District Hos-pital (IDH) included 10 different staff members at theAntenatal clinic (ANC) and HIV clinic. We also inter-viewed health managers and government officials. Thefocus on researchers in Canada and WelTel staff andclinic staff in Kenya (Table 1) reflected different levels ofknowledge engagement. In Canada, most of the re-searchers we interviewed were intimately involved in theimplementation of the pilot projects, whereas in Kenyathis was the responsibility of WelTel staff working withthe local clinics.Data analysisWe used a modified approach to grounded theory fordata analysis. This involved open coding, preformedmanually on data collection notes by a trained qualita-tive researcher (KB), in order to generate a key list ofTable 1 Qualitative Interviewees by Category and CountryKey Informants Canada Kenya TotalResearcher 9 1 10Weltel staff 8 8Clinic staff 2 10 12Clinic manager 1 1Government official 1 111 21 32Bardosh et al. Globalization and Health  (2017) 13:87 Page 5 of 15codes. A field-note diary was also kept, for brainstorm-ing and reflection. This included case-based and analyt-ical memos. This process facilitated the exploration ofrelationships and connections between different themesand subthemes, generating our analytical interpretations.Importantly, analysis was validated through a follow-upworkshop in Kenya with IDH and WelTel project staffin July 2016, and through providing drafts of this articleto a sub-group of key informants in Canada, as a formof member checking.Ethical approvalThe study was approved by the University of British Co-lumbia’s Clinical Research Ethics Board (H16–00189),and Amref ’s Ethics and Scientific Review Committee(AMREF-ESRC P161/2015).ResultsBased on our qualitative data analysis, we divided this paperinto five sections. First, we discuss the provider-patient rela-tionship and how the Weltel intervention was perceived toinfluence this. Second, we explore how the interventioninteracted with service provision and organization at theclinic level. Third, we go on to discuss the socio-technicaldimensions of WelTel and how the technology itself wasperceived and used. Fourth, we present data on the import-ant role of evidence and data in generating support and le-gitimacy for the intervention from different stakeholders,and how this process influenced implementation. Lastly, weturn to the issue of scaling-up and discuss the importantrole of networks, policy and politics.Improving the “culture of care”Our interviews focused heavily on how health sector ac-tors perceived the utility and benefits of using WelTel.We found that these extended far beyond the immediateSMS communication to influence the “culture of care”at local clinics, including services, standards,organization, management and accountability. In linewith previous studies [22], the intervention was widelyperceived as a “tool” that empowered patients by betterconnecting them to medical staff between appointmentsand increasing their access to medical expertise and out-reach. In both Canada and Kenya, SMS is widely used,making it easy to connect through text [23], althoughaccess was believed to be lower in Kenya among pastor-alist communities and women.By allowing patients to seek immediate feedback onquestions and problems, WelTel facilitated a “sense thatsomeone cares”, and helped patients direct their owncare.1 This included addressing questions related to ap-pointments, medical issues, medication side effects, andsocial issues. Importantly, our interviews also stressedthe ways in which it facilitated greater access to care. InKenya, one informant summed this up with the catchysaying: “the more you talk, the more friendly you be-come” (Kuongea ni Kuongeza Urafiki, in Swahili). Simi-lar euphemisms were used in Canada. Complex issuescould be triaged on the phone, connecting patients tomultiple healthcare providers, while also allowing pa-tients with chronic conditions, like HIV, to build stron-ger rapport with their providers. In Kenya, and to alesser degree in Canada, one of the most emphasizedbenefits was the ability for patients to circumvent longwait times for immediate health needs.“The flexibility [of the technology] is good. You canbe texting a patient. They have a problem. You goand find that [specialist] in the clinic. You call andgive the phone right to the health worker. They getimmediate feedback! And if they need to be seen,they come right away.”(Interview, Clinician, Kenya)Texting in Kenya was largely restricted to respondingto the weekly check-ins, with occasional auxiliary textsto “encourage” patients to follow medical recommenda-tions. In comparison, patient-clinic interactions weremuch more dynamic, common and personalized inCanada due to the explicit “patient-centric” approach ofthe clinics involved. In the Canadian HIV projects, pa-tients were reported to regularly text about the “joys andchallenges of life” (Interview, Clinician, Canada) whilestaff could also use the service to check-in with patientswho were going through difficult periods, either clinic-ally or emotionally.Versatility of the SMS technology was described as a“lifeline” that allowed patients to engage on their ownterms, when and how they needed assistance. Theycould be very active on the platform, and then remain si-lent until a future need arose. The benefits of SMS alsoallowed some more vulnerable patients in Canada to“open-up” about complex issues, due to the “emotionalsafety”, as one nursing staff described it, of the SMStechnology. These benefits were widely discussed as aform of “patient empowerment.”Improved bidirectional communication was viewed asimproving the culture of care at clinics in multiple ways.This included impacts on staff motivation, performance,teamwork, job satisfaction, work routines and relationsbetween staff and managers. Staff and managementnoted how the positive impact on patients facilitated animproved sense of work satisfaction, with positive sec-ondary effects on performance. As one HIV counselor inKenya mentioned:“It helps with our sense of teamwork and satisfaction ofclients… Psychologically, you feel better connected toBardosh et al. Globalization and Health  (2017) 13:87 Page 6 of 15patients. And it really helps to motivate staff to ‘pull uptheir socks’ and care for the patients more!”In Canada, HIV staff discussed the improvement inworkflow for dealing with complex patient issues, whichpreviously required keeping track of multiple text-basedand phone conversations. As an organizational tool, theintervention was noted as helping to save time for clini-cians and outreach workers in Canada by systematizingpatient communications:“We were texting before [with patients] but you couldeasily forget about someone’s [text]…[the] Weltel[computer-based platform] creates regularity. It’s anorganizational tool that is much easier to use thanscrolling down on my [cell phone]. The platform helpsus never forget.”(Interview, Clinician, Canada)In the resource-limited context of northern Kenya,noteworthy changes in patient care, with important sec-ondary influences on hospital management, were em-phasized. This included assisting with timely emergencymedical outreach. One widely discussed example in-cluded how the texting service helped link the hospitalto a remote pastoralist woman in childbirth, who washaving life-threatening complications. An ambulancewas mobilized and the woman delivered safely at thehospital and recovered. WelTel also helped justify finan-cial investments for other forms of patient care. This in-cluded helping rejuvenate HIV psychosocial supportgroups, with support from an NGO-partner. Further-more, the greater communication with HIV clients droveclinicians to better appreciate the full scope of HIV-TBco-infected patients at Isiolo District Hospital. This ledto the purchasing an expensive GeneXpert machine tofacilitate more prompt TB diagnosis at the hospital.Perhaps most importantly from the patient perspec-tive, the service allowed patients to report malpracticesor administrative errors, facilitating changes in clinicguidelines and protocols. This is noteworthy, given thathealth worker performance is often low in many publichospitals in Africa and there are often few avenues forpatients to provide feedback and complaints to manage-ment [24]. Phone communication, so Kenyan staff felt,could improve clinician accountability. One example in-cluded a clinical officer intern at IDH who prescribedthe wrong HIV drugs to a patient; the patient contactedthe hospital through the SMS service to flag the error. Ifnot for the service, the error would have gone unnoticed(according to multiple interviews). In response, the hos-pital revisited its intern work-related policies.Clinic staff also highlighted the ways in which theintervention raised awareness regarding individualpatient circumstances. In Kenya, this included situationsof extreme poverty, food insecurity issues and lack of ac-cess to drugs for opportunistic infections that had animpact on health outcomes. Kenyan staff questionedtheir ability to address these issues, and commented onthe importance of clearly explaining the limitations ofthe intervention. As one staff member discussed:“How can you get [the patient] to appreciate thelimitations [of WelTel]? For example, [the drug]dapsone is free but not always in stock…the patientwill complain, “This person offered to respond to my[problem], but now they are not willing!” We tell themwe will try our best, but we are not always able toaddress everything.”This social context also influenced enrollment andutilization patterns. Data from Isiolo District Hospitalshowed enrollment to be roughly 50% of the HIV patientpopulation, despite the intention to enroll all patientswho had access to a mobile phone. Staff noted a numberof important disincentives for enrollment, related mostlyto larger structural barriers to care (illiteracy, stigmaand, for some remote pastoralist groups, lack of phoneownership). Despite the service being, for the most part,free, WelTel staff estimated that only 20% of HIV pa-tients enrolled in the intervention immediately; 80% en-rolled only after being encouraged by other patients.Unlike in Kenya, HIV staff in Canada reported that mostpatients enrolled without any problem and that one of themajor strengths was helping to connect patients to socialservices when they needed it. If they did not have food,they could be referred to a food bank or an emergencyfood grant, for instance. But in Kenya, where social servicesare few and far between and dependent on NGO grants,addressing these issues was more complex. With help fromsocial workers and IDH, the WelTel intervention did pro-vide some assistance to patients with healthcare coststhrough a waiver system for the most vulnerable and poor.But IDH clinicians recommended that future operationsinclude greater linkage between the WelTel model of careand broader developmental goals to assist with broader so-cial determinants of health. However such recommenda-tions, in some ways, ran counter to a core objective of theintervention: to be easily implementable and scalableacross large geographical areas and diverse health contexts,with minimal direct investment in human capacity.Despite these differences, the ways in which patientsbenefited from the service in Canada and Kenya hadsimilarities. In the original RCT trial in Kenya, Shida re-sponses were recorded at 2% of all participants by 6-months [15]. Preliminary data showed a similar numberof patients communicating with clinic staff on a weeklybasis. One Canadian nurse stated:Bardosh et al. Globalization and Health  (2017) 13:87 Page 7 of 15“At the beginning, some thought WelTel would bemagic bullet for those that we had not connected with.It would be the missing piece, and for a handful, itwas...For others, they didn’t use it. The funny thing isthat you could not have predicted beforehand whowould have benefited most! People use it verydifferently depending on the needs they are having atthe time.”In this sense, the WelTel experience echoed broaderfindings on behavior change interventions in publichealth: impact was focused, incremental and highlyvariable depending on the needs and wants of particularend-users. Accurate figures on scaling-up and long-termpublic health impact would have to account for thesenuanced utilization patterns, and how they change over-time.Organizing servicesHow the WelTel service improved the culture of care,and to what extent and with what effect, was influencedby the ways in which the service was organized at theclinic level. This mediated the “implementation climate”described by Damschroder et al. [14], including the “ab-sorptive capacity for change and receptivity” of a healthinnovation.The ways in which the service was delivered andintegrated with the health system had significant im-plications for cost, quality of care and sustainability.In Canada, higher clinic capacity allowed the interven-tion to remain the prerogative of clinic staff. Thisallowed the Oak Tree HIV Clinic to continue WelTelwith roughly 40 patients after research funding cameto an end in 2015, as part of routine services. In con-trast, researchers and staff in Kenya emphasized thechallenges involved in integrating the service withclinic work routines. To address this, for example, theWelTel Retain Randomized Control Trial (RCT) inNairobi paid a higher than average salary to study staffin the Kibera slum clinic to ensure the integrity ofresearch protocol and quality results. But this type ofincentive was not done at Isiolo District Hospital(IDH), where the main objective was to explore scalingup within existing systems.Achieving this goal was, not surprisingly, difficult.Kenyan staff at IDH highlighted how a combination ofhuman resource shortages, low morale, work culturenorms and low technology skills meant that the originalintention of the project (funded by Grand ChallengesCanada) shifted; the WelTel mHealth Society needed tohire two local “expert patients” to enroll and triage pa-tients. This is a common theme in global health projects,responsible for the large number of so-called “vertical”interventions [25].We found that various geographic, cultural, social,economic and political factors of the Northern AridLands (NALs) were invoked in different ways to explainthis. As a key informant in Kenya discussed it:“It’s about attitude – WelTel requires consistency andwork discipline…This is very hard for [clinic] workerswith ‘fluid schedules.’ You need someone with basictechnology skills; someone educated about the benefitsof technology and innovation. But this is just notprioritized in the staff mentality…the region is veryremote.”At IDH more broadly, the running of both the ANCand HIV clinics depended heavily on one or two focalnurses who were overburdened with responsibilities. Amanager summed this up:“At [the HIV clinic] if the in-charge nurse is not there,things will not work well. You have a situation whereyou depend on one person to run everything. We wit-nessed this very much – you find that when this personleaves, everything drops-off.”Another pervasive narrative was that staff lacked thetechnology skills to manage the service effectively with-out direct oversight support. Participants commented atlength on the lack of “IT literacy” of hospital staff, espe-cially in reference to another technology innovation in-troduced by Amref: IQ-CARE, an ambitious project toinitiate a complete electronic records system to replacethe current paper-based system. This initiative ran intomultiple bottlenecks and was, for the most part, laterabandoned.While staff liked WelTel, there were a number whoexpressed reluctance to manage it if external supportwas withdrawn. Some explanations were not related atall to technology skills, or a widely discussed sense ofbeing over-worked; rather, they were rather to socialnorms. One repeated reason was that nurses did notwant to be “accused” of overusing airtime, or using thephone credit for personal use. In a resource-poor clinic,small material benefits were viewed as adding-up, withthe potential to generate social conflicts [26]. All of thisrelated to the internal legitimacy of the intervention, stillconsidered an “external project” and not a core routineactivity of IDH, overseen directly by management.However by relying on WelTel staff, many IDH clini-cians did not fully appreciate how WelTel worked at thetime of research, despite it being implemented nearly ayear prior. This contrasted with the Canadian HIV pro-jects. Cedar, for example, had planned their own one-week systematic training for all staff members, which in-cluded inviting Oak Tree staff to attend (and provideBardosh et al. Globalization and Health  (2017) 13:87 Page 8 of 15mentorship), role playing to familiarize staff with the ser-vice, and group consensus on all triage protocols; two-weeks of piloting had then followed. While resource in-tensive, this enhanced training could be valuable in theKenyan context.In this sense, all projects emphasized the need for a“focal person” or “local champion” to manage and over-see the service, given the multitude of other administra-tive and work-related responsibilities. This appears be amajor limiting constraint to the effective scalability ofmHealth interventions. The differences in capacity ofthe ANC clinic at IDH in Kenya (over-crowded; longwait times; lower levels of care) and the HIV clinic(greater capacity and infrastructure; with direct donorsupport) required different responses. WelTel mHealthSociety staff at IDH included an “expert patient” (HIV-positive individual) at the HIV clinic and a more edu-cated Information Technology (IT) diploma holder atANC, both of who were paid very modest (and different)monthly salaries. Responsibilities included giving healtheducation talks, retrieving files, enrolling patients, man-aging the tablet platform, making follow-up calls andtriaging patients. As found in other studies, challengesof motivation and capacity ensued [27]. These two paidvolunteers were overseen and supported by WelTelmHealth Society staff in Isiolo (a project coordinator,IT/data analyst and administrator working from theAmref office in Isiolo). These individuals provided directday-to-day management to ensure that activities wentaccording to plan. They also planned scale-up, whichbegan in mid-2016 in neighboring Marsabit and Sam-buru counties, just as research was being conducted.One major question for scaling to these sites revolvedaround whether or not to rely solely on existing clinicstaff to implement WelTel or to pay auxiliary staff,which was felt to have major implications for cost, sus-tainability and effectiveness.While the context in Canada was very different, clinicstaff also emphasized a dependence on one focal personto ensure the service ran smoothly. The LTB1 studyfound that one administrator was instrumental in assist-ing with enrollment and follow-up. The Canadian HIVprojects stressed the importance of having experiencedoutreach nurses play this role. At Cedar, patients endear-ingly called the main outreach worker managing theplatform, the “phone lady.” Interviewees stressed the im-portance of linking “trust and relationships” with “thecold technology” in order to facilitate improved care.You needed a “human touch” to the intervention, as oneinformant termed it. While focal staff originally thoughtWelTel was going to overburden their normal activities,they found that once accustomed, it consumed modestamounts of time and streamlined existing forms of pa-tient communication, as discussed above. One focalnurse in Canada estimated that she spent approximately2 to 3 h per week managing the platform, calls andfollow-ups.Tailoring technologySo far, we have discussed how the WelTel intervention,a relatively easy-to-use and evidence-based mHealth ap-plication, influenced quality of care and organizationalcultures. Our data showed that both of these domainswere influenced, in different ways, by the technical char-acteristics of the WelTel platform. This included issuesof functionality, end-user friendliness, complexity andadaptability. A major question, echoed throughout ourinterviews, regarded the degree to which the interven-tion could be tailored and customized to specific groupsof patients and in ways that met the wants and needs ofclinic staff and management.The experience of WelTel showed that considerationsfor scaling mHealth should be located in the challengesof effectively designing and refining the platform, orproduct, which can take much longer than sometimesappreciated. Technical adaptation dated back to theearly Kenyan RCT trial (2005–2007) where nurses hadto manually text patients weekly using paper registers.Multiple iterations of the software and platform weresubsequently developed, with requests from each projectand partnership to improve functionality. The most sig-nificant system innovation was the software redevelop-ment that shifted the WelTel platform from facility-based (WelTel V1.0) to a centralized-server system(WelTel V2.0). WelTel V1.0 had to be downloaded ontoindividual computers at each clinic. During the KenyanCDC Foundation study (2011–2012) in seven clinicsaround Nairobi, the initial version had become difficultand expensive to maintain, having to be physically ser-viced and dependent on the city’s irregular power supply.Medical superintendents were reluctant to have patientdata posted online since it would run against existingMinistry of Health policies. In response, the WelTelteam designed a centralized-server system, housed on aprivate server at the Amref office in Nairobi. Thisallowed for security, consistent electricity and high-speed internet connection. Redevelopment also usednew and more efficient programing language, allowingflexible upgrades.WelTel mHealth society staff spoke about this periodas a lesson in “back-scaling.” The initial intention ofscaling-up from the CDC pilot had to be reconsidered,as substantial technical re-investment was needed. Thisinformed the current “off-line” software platform atIsiolo District Hospital (IDH), which allows patients tobe enrolled and followed-up on tablets without immedi-ate access to the internet; enrolment is uploaded ontothe server when connection is available. This was seenBardosh et al. Globalization and Health  (2017) 13:87 Page 9 of 15as making WelTel ideally suited for deployment in re-mote regions, like Northern Kenya, compared to manyother similar systems being developed in Africa.Interviews in Kenya also highlighted three othertechnology-related issues relevant to scale-up. First, therewere different views regarding the ease to which the low-cost tablets piloted at IDH could be scaled-up. Some clinicstaff emphasized that it would be better to utilize existingclinic phones (provided by NGOs like Amref), while othersstressed that tablets were superior and training could easilybe provided. A second related issue was the high number ofweekly non-respondents that required to be called back –approximately 30% of the 700 clients at the HIV clinic andnearly 80% of the roughly 1600 ANC clients did not regu-larly respond to the weekly SMS message (some had alreadydelivered). Calling back each client that did not respond tothe follow-up weekly check-in text required a larger thanexpected budget for airtime, and the protocol was changed(similar results were reported from the Canadian projects).Only those with self-identified problems were then calledback each week. A major challenge was phone connectivityand the inability of the offline system to organize and man-age communications. Patient phones were frequently offdue to lack of battery or being out of the service area.Tracking which numbers had been successfully contactedcould not be done on the tablets, and required paper notes.Clinic staff emphasized the need to increase functionality ofthe tablet-based system to better manage patient communi-cation, as had been developed in Canada.All of this related to the perpetual need for tailoring andcustomization, for problem solving and co-development.However, there was a simultaneous recognition of theneed to balance evidence-based changes with operationalfeasibility and streamlined functionality. Trade-offsneeded to be considered in each adaptation (whether pro-grammatic or platform-related), especially those thatmoved away from the initial “simplicity” of the RCT Ken-yan trial. The push for further customization of the plat-form was highlighted in nearly all interviews. As oneCanadian researcher put it, “We had a lot of input into thenew software. But already we want to add new features!”This ranged from customized messages, greater function-ality to manage the platform, inclusion of other healthconditions and, critically, the ability to communicate testresults by text. While some requests were unrealistic, orwould play little role in changing health outcomes, otherswere actively taken-up in an instance of “co-design” withend-users, coupling research, clinic input and software de-velopment. One of the best examples included implemen-tation in the ANC in Kenya, where Isiolo hospitalmanagers wanted to include more versatile messaging:“We started with the Mambo program but then weswitched…we wanted messages for all antenatalmeetings. But [WelTel] said this was too hard. So wefound a middle line.”Changes included reminder texts during different timesof pregnancy, check-ins during the delivery period andthen childhood immunization reminders.One last area, important to scaling, related to small“glitches” that required IT support. Although WelTelV2.0 had run continuously for over a year at time of re-search, periodic issues did arise. Texts could sometimes“hang” for days in the network, influenced by the speedof the phone company, while messages could occasion-ally disappear and phone numbers get scrambled.Legitimization: The role of evidence and dataAs we explored the ways in which WelTel projects gen-erated funding and enrolled partners and collaborators,it became clear that data and evidence facilitated legit-imacy to network across different global and local scales,and that this was key to scale-up. In effect, data andmetrics functioned as a form of “organizational capital.”Interviews stressed that the research data behind Wel-Tel was “robust” and “some of the best out there.” Thesignificance of the initial Kenyan RCT trial (published aseditors choice in the Lancet; cited more than 600 times)was often referenced. The WelTel Kenya1 RCT remainsone of the only trials to show an improvement in bio-logical markers for HIV using SMS, and it continues toreceive positive reference [8]. This research focus led toobtaining multiple research grants, approximately $3million stretched over 10 years, to adapt and evaluateWelTel in a range of contexts. A strong research trackrecord of over 20 high-quality, peer-reviewed academicpapers (proof-of-concept, health behavior, cost-effectiveness analysis, and RCT design) provided support.Academic research helped obtain media attention, inScientific American and CBC’s The National. It also elic-ited comment from the National Institutes of Health(NIH) Director, Dr. Francis Collins, during an mHealthSummit:“Clinical trials on mHealth ought to be the best way todetermine what is actually working and what isn’t…there is no substitute…WelTel demonstrated anenormously important effect.”As a new field subject to crowding and competition,the emphasis on being “evidence-based” was seen as away to distinguish WelTel from other mHealth applica-tions. As one key WelTel staff member stated:“In the early 2000s, SMS was seen as a ‘no-brainer’area to get into; so everyone got into it. Now it is achaotic environment. Everyone is doing their ownBardosh et al. Globalization and Health  (2017) 13:87 Page 10 of 15thing. But everyone is still in the pilot phase! Everyone.It is like all these different groups are creating theirown word processors! …But people tend to overthinkthings. They want to have so many features, or havelots of information sent to the patient…[but] much ofthis stuff isn’t evidence-based…”WelTel could be “evidence-based” because it wasamenable to clearly defined metrics: recruitment and ad-herence rates to therapies and appointments, text andphone records of patient-client interaction, clinic/hos-pital utilization and biological markers, such as CD4count and viral loads for HIV. All current projects gen-erated data aimed at this sort of evaluation.However, moving from the controlled setting of awell-funded research study, often overseen by academicand NGO-staff, into the public health infrastructure inKenya generated uncertainties with data and evidencethat required negotiation. We found that quality of data,capacity for analysis and challenges in attribution wereraised, with implications for data interpretation. For in-stance, a major issue was how to interpret data on “de-faulters” in Kenya (someone who stops taking their HIVmedication), which is important for evaluating the HIVclinic data. There are multiple reasons someone woulddefault, or be counted as a defaulter. One major uncer-tainty related to what was believed to be a high numberof patients that regularly switched clinics to avoid beingseen or noticed. Defaulting rates some quarters were ashigh as 70% pre-WelTel, and there were clear difficultieswith using the existing hospital-based data system foranalysis due to errors and gaps in data.Data and evidence, however, did not only play an im-portant role at the global and national level as a form oforganizational capital; they were also instrumental ingenerating legitimacy and buy-in from local clinic staffand management, albeit in a different way. A number ofmanagement staff in Kenya expressed that they were ini-tially ambivalent or skeptical of WelTel until “seeing thedata” from the first few months of activity at IDH:“At first, some thought this was just another NGOcoming to benefit itself… but then we did acomparative analysis of fourth visits to the ANC aftera few months and saw a big benefit…I took this dataand did a presentation at county-level to convincethem. Everyone was impressed, especially managementlevel.”The example of the antenatal clinic data at IsioloHospital revealed how new forms of health data canplay important political roles, which is well understoodby local managers and staff. Improving MNCH indica-tors in Kenya is a major policy prerogative of thegovernment and donors, especially in the NALs region.In 2013, the president put into place a new policy (TheBeyond Zero Campaign) for free maternal care.Funding to the counties is based on a reimbursementsystem, and payments from central government areoften delayed for months. If the hospital could keepbetter track of pregnant mothers and their deliverydates, so the reasoning went, funding could be morereadily obtained from the central level. This was widelynoted as one of the major motivating factors for IDH topush changes to the WelTel texting system in theantenatal clinic.Interestingly, access to data also played a de-stabilizingrole, whereby uncertainties of data security policies andlaws generated barriers to scaling and integration inclinic routines. Patient confidentiality issues had in-formed the design of the WelTel service; phone num-bers were not traceable to the clinic nor was HIVmentioned during the weekly routine texts (this wasnoted as very important for safety and mental well-being). For the Canadian projects, confidentiality wasconsidered a “Mount Everest issue” that was hard tounderstand within the shifting legal regulations overseenby the health authority (Interview, Researcher, Canada).While it was easy to implement the WelTel interventionin a research context due to the clear protocols for in-formed consent, this was much less clear in routine care– a barrier that had prevented Oak Tree from fullyimplementing the service beyond ex-research partici-pants when their research funding ran out.Scaling-up: Networks and policy contextsConcerns about evidence and data were not the only fac-tors shaping policy development – broader social, polit-ical and institutional drivers were considered equally, ifnot more, important. In our interviews, it was clear thatsuccessful scale-up would have to include the forming ofnew networks and linkages and the testing of differentfinancing arrangements. The challenges involved arewhat make successfully breaking-out of the pilot stageso difficult in the mHealth field.A bit of institutional history is informative. Partnershipswere clearly an important mobilizing force for WelTel,grounded in a unique linkage of academic research, soft-ware development, clinical medicine and service delivery.This appears to be a common aspect of new mHealth plat-forms, especially transnational collaboration that spansboth the Global North and Global South. The originalRCT trial emerged from an international Manitoba-Nairobi network of HIV researchers, and took more than3 years to finish. It also provided a different form of legit-imacy, or “social capital”: sufficient depth of experienceworking in public health in Africa. As one Kenyan inform-ant commented:Bardosh et al. Globalization and Health  (2017) 13:87 Page 11 of 15“He [WelTel’s founder] has legitimacy. He has lived inKenya, conducted clinical research in Kenya. Heknows the country. He has friends. He is not seen assomeone running in and running out just to benefithimself.”This experience facilitated strong networking capabil-ities, which explained the institutional history of WelTel.Moving from Kenya to the University of British Columbia(UBC) in Canada and the BC Centers for Disease Control(BC-CDC), WelTel’s founder was able to leverage theemerging interest in mHealth from Canadian researchersand funders. The Canadian projects were described as atype of “reverse innovation”, moving the Kenyan experi-ence to vulnerable populations in Canada.This also facilitated networking at the global stage, in-cluding participation on a number of advisory commit-tees and contributions to international guidelines forHIV, TB and mHealth. This budding organizational visi-bility provided some unexpected opportunities – for in-stance, an Ethiopian postdoctoral student in Germanyobtaining pilot financing from the Ethiopian Ministry ofHealth to trial WelTel in Gondar, Ethiopia. Other emer-ging opportunities included collaborations in South Af-rica, Zimbabwe, Uganda, Rwanda and the United States.Implementing partners provided the necessary skills tofacilitate service delivery and navigate local contextualterrains. One of the most significant collaboration in-volved links with Amref Health Africa, part of a largehealth consortium in Kenya (APHIA plus Imarisha)funded by USAID. Amref was described as a “legitimiz-ing partner”, since the NGO has high visibility in Kenyaand especially in NALs (where it has worked on MNCHand HIV). Amref was key in facilitating access to gov-ernment leaders and clinics, and played a major role inobtaining the Grand Challenges Canada financing, pro-viding significant in-kind contributions. Developing thiscollaboration took time, even after the grant started, asdid setting-up human resources and partner networks inIsiolo county.The benefits of partnering with established organiza-tions were also noted in the Canadian context. Both of theCanadian HIV-projects, at Cedar and Oak Tree, werewell-recognized organizations with established researchexperience, and with previous Canadian Institutes ofHealth Research (CIHR) grants. This had been importantin their negotiation with a national telecom company toenable an affordable phone plan for vulnerable HIV pa-tients, allowing unlimited text messaging and some calltime without the typical conditions and terms.However, the overarching challenge for WelTel andtheir partners was: how to move beyond the “research”stage? What did this mean, and how could it be done? In-terviewees stressed the importance of forming newnetworks of partners especially with the business commu-nity and with focal Ministry of Health staff as a key path-way forward. But many of the key stakeholders, part of theWelTel partner network, tended to balance multiple re-sponsibilities and had limited time for sustained engage-ment. One Canadian clinician-researcher highlighted thechallenge of wearing multiple hats (balancing clinical re-sponsibilities, research studies, searching for future fund-ing, and networking with provincial authorities) whiletrying to bring the WelTel program into mainstream clin-ical care:“I am a clinician, I have 30% time for research. Isimply don’t have the time to chase up the heads ofhospitals [to negotiate scale-up].”Others stressed the importance of having a full-timebusiness director in order to seek out opportunities forgrowth in the private sector. Without core funding, thiswas difficult to arrange, and highlighted the different in-stitutional capacity needs of a clinical-research networkand a software company – as an organization, WelTelwas, in effect, at a stage of development where it wasbalancing both.The organization was described, after all, as a “socialbusiness.” Discussions centered on different financingarrangements, and how health impact in Africa could beapproached sustainably while maintaining organizationalgrowth. An ideal scenario, discussed by some, would befor clinics, hospitals and district and national health au-thorities to pay for the service, or at least co-finance it.In this scenario, adoption would be promoted based onan intrinsic value, which did not require augmentingstaff salaries or supportive infrastructure (as had beendone at Isiolo District Hospital). The incentives forhealth providers would be to help streamline work rou-tines, avoid crises with patients, to receive feedback andrecognition, and assist with work satisfaction. As one re-searcher mentioned in reference to Kenya:“WelTel can’t solve the human resource problem inAfrica. The service helps with efficiency; it builds andreinforces existing capacity. It is not for us to plug hugegaps in capacity…Ideally, we want the service to beintegrated into the budget lines of local healthauthorities.”However in order to achieve “buy-in” from local healthauthorities, cost data was needed.“The health system is not going to throw money at you!You need to prove that [WelTel] works. You need touse cost analysis to make the case for it and yourhealth [care provider] utilization data has to showBardosh et al. Globalization and Health  (2017) 13:87 Page 12 of 15that it’s not too expensive. It’s about showing a “bangfor your buck”…really, it’s about budgets.”(Interview, Clinician, Canada)Cost data was repeatedly cited as the most importantfactor by county and hospital management in Isiolo, andby the Canadian teams, as fundamental to any discus-sions about sustainability. Maintenance costs were esti-mated to be very low – airtime for the HIV and ANCclinics, at Isiolo District Hospital in Kenya, were esti-mated at only $10 and $20 per month respectively, androughly $300/month for the weekly SMS messages. Inthe Canadian HIV projects, discussions focused on themost vulnerable patients, who take up a disproportion-ate amount of inpatient time and resources. One re-searcher estimated that:“The most vulnerable [HIV-positive] patients [inCanada] will take up 30 or 40% of our time andresources but they are about 10% or so of our clients…so addressing prevention issues in this population withWelTel should have major cost-savings.”A cost analysis done by WelTel, which has been pre-sented at major academic conferences, showed substantialsavings (in the millions of dollars) for ART donors in Af-rica, extrapolated from the original Kenyan RCT findings.The asthma group at UBC in Canada noted how the dis-ease affected over 3 million Canadians (with 500,000 hav-ing chronic obstructive pulmonary disorder) and cost theCanadian health system nearly $9 billion per year. Hence,researchers were quick to discuss scalability in terms ofpotential savings to the health system if low adherencerates could be addressed. However, interviews also re-vealed that this was an area that was still under-developed, and required more concerted work by healtheconomists and business leaders to show cost-savingsfrom mHealth interventions.Aside from the need for cost data, there were differentviews about the willingness and ability for hospitals andclinics, in both Africa and Canada, to finance WelTel dir-ectly. The CDC demonstration study in Kenya had aimedto have the system sustained and financed by eight hospi-tals in Nairobi. However, in practice, nurses had to be fi-nancially compensated for their extra time, and the clinicmanagement (although very supportive of the service) didnot commit the necessary resources.“These places did not want to provide airtime. Thatwas it…the evidence was there; they loved us. But youfind a major disconnect between the clinic staff whosaw the day-to-day benefits and the managers that di-rected funds…We were pitching ideas but they weren’ttaking them.” (Interview, Kenya)Certainly, there were challenges in engaging manage-ment of African health systems to support health tech-nology innovation when basic infrastructure and staffsalaries were difficult to maintain. In Isiolo, user feescould have been used to budget for the (modest) cost ofairtime to support WelTel. However, with the devolutionof local government in the new 2010 constitution allfunds were decentralization and managed by the countyoffice. To better address these issues, a “WelTel sustain-ability committee” had been organized in Isiolo in 2015.Staff and airtime from the county budget were tenta-tively promised, but major political norms and a lack oftechnical capacity problematized their realization. Asone interviewee discussed:“There are many problems with this idea [of havingthe county pay for WelTel services]. The major one isa perception problem: politicians like roads – thingsyou can see…There is already a perception in countygovernment that health receives too muchmoney…[and secondly], county officials do notprioritize technology…There is a gap in appreciation oftechnology.”Discussions at the county level in Isiolo revealed thathealth sector funding is taken-up overwhelmingly withstaff salaries, with the limited field budgets directed toper diems, fuel and drug costs. Government officialsnoted that they were most interested in technology man-agement platforms that would allow them to track leak-ages in procurement, revenue collection and “ghostworkers.” Going back to the example of IDH, decision-makers were most interested in how technology couldimprove existing systems and not necessarily focused onpatient-centric approaches. Additionally, there was apervasive perception, set by historic-political precedence,for commodities (i.e. drugs) and other supportive inno-vations to come from central government or donors.Donors, such as PEPFAR, provide most HIV drugs inKenya, and other clinical services, like psychosocialsupport, are seldom paid for by clinics and hospitals,especially in poorer regions like NALs. Rather, they areprovided through grants to NGOs like Amref, which areintermittent and dependent on shifting donor priorities.Generating major donor or government support wouldrequire, as one informant stated, “high-level commit-ments and a system ready to deploy massively [at scale]”(Interview, Isiolo).Hence discussions around financing led to concernsabout the need to “package” the WelTel service for pol-icy uptake. Ambitions were aimed at scaling WelTelacross Africa as part of HIV, TB and ANC services, andin some other form of commercialization for the NorthAmerican market. Various business consultants hadBardosh et al. Globalization and Health  (2017) 13:87 Page 13 of 15been brought in, often on a gratis basis, to explore this.A free licensing model for resource-limited settings sup-ported by technical support contracts had been adopted,which would model Socially Responsible Licensing (SRL)principles. However as a new field, mHealth applicationsand supportive policies were noted to be still nascentand fragmented, and willing-to-pay rather uncertain. Asa WelTel-associated researcher discussed:“People expect mHealth to be based on tiny budgets,or that it should all be provided for free, open source.Everyone wants to do their own thing. But ARVs arenot open source! It’s an innovation that needs to beprovided, like a business…like a social business.”The WelTel software was provided for a licensing feeto each of the individual research projects, but some cli-nicians stressed that this expense would be hard to jus-tify year-after-year at the facility-level:“It would be better if it was something like MicrosoftWord where you buy it and you can just use it…youcan even modify it with upgrades as you go along.”(Interview, Clinician, Canada)Interviewees also highlighted some challenges in “sell-ing” the service to policymakers. In Canada, perhaps oneof the most challenging questions was related to theability to convince policymakers that vulnerable HIV-positive clients should be given a cell-phone (which hadbeen the case with the two HIV projects):“You can clearly see the impact [of the WelTelintervention and phone provision] but how do youconvince people to give money to give a drug user aphone? It is a hard sell!”Some believed that engaging in auxiliary technology fea-tures (such as the provision of phones) could distract fromthe aims of rapid scalability. To others, however, it was es-sential to target the most vulnerable patients. Hence, therewas still a sense that the organization was wresting withthe proverbial question: what type of service should Wel-Tel be offering and to whom? There was also an ongoingdebate about how to link public health impact with busi-ness interests and organizational growth.DiscussionWhile the mobile phone revolution has enormous po-tential for public health, both in terms of improving pa-tient adherence to therapy and outcomes, and widerhealth system strengthening, current evidence of suc-cessful scale-up and impact is still in its infancy ([9, 11,28]). The promise of investing in mHealth may offer aunique opportunity in developing countries due to theminimal infrastructure requirements comparable to e-health [29] but for this to be realized, systemic healthsystem weaknesses need to be understood, navigatedand negotiated. Furthermore, there needs to be a delib-erate effort by funders and governments, in both theGlobal South and Global North, to combine the enthusi-asm for scaling with the rigors of evidence-based prac-tice – less scaling is viewed more as market growth byindustry than as an opportunity to benefit public health[12]. This makes understanding real-world program-matic challenges, and the individual institutional histor-ies of particular mHealth innovations – how networksform, evolve and change – all the more urgent.In a recent systematic review of mHealth in Africa,Chib et al. [9] distilled lessons learned from 44 projects,most all of which were small-scale. They found that ap-propriate project design, stakeholder participation, inte-gration with the health care system and the subsequentuse of appropriate technology and resources were keyfactors in successful implementation. In contrast, theynoted how lack of funds, unknown cost-effectivenessand lack of evidence inhibit scalability. A second review,Gagnon et al. [30] analyzed 33 mHealth studies on indi-vidual, organizational and contextual factors influencinghealthcare provider adoption. They highlighted the im-portance of usefulness, ease of use, design, cost, time,and a variety of related factors.In this paper, we have provided a nuanced and contex-tualized perspective on these issues, using the Consoli-dated Framework for Implementation Research (CFIR)as a guiding conceptual lens to explore the implementa-tion process for WelTel in both resource-limited (Kenya)and resource-rich (Canada) settings. Through this quali-tative and comparative approach, we have discussedvarious aspects of the complexity of implementation andhow they are linked at multiple levels. As the WelTelservice and other initiatives move forward, implementa-tion of mHealth, as a disruptive and positive changetechnology could benefit substantially from further in-depth case study evaluations.However, the benefits of unpacking the textured socialfactors integral to moving mHealth innovations forwardshould not end with knowledge generation and academicpublishing. Closing the “implementation gap” in publichealth [31] requires coupling research on operational is-sues with pathways for “actionable intelligence” – feed-back loops that can, in the maze of factors that canderail or enhance an implementation pathway andorganizational network, help guide stakeholders towardsnew ways of problem solving.As with many global health innovations, one import-ant challenge for the mHealth community going for-ward, including for WelTel, is to maintain flexibility andBardosh et al. Globalization and Health  (2017) 13:87 Page 14 of 15adaptability in the technology platform and service deliv-ery pathway when moving from proof-of-concept towider scale-up [31]. This is the classic transplantationchallenge in public health – when scale-up occurs, thecareful attention to detail that facilitated the initial suc-cess becomes marginalized [25]. The organizational cul-ture of the project is challenged with the complexities ofgrowth; for example, each implementing clinic will stillrequire a local champion to ensure effective implemen-tation. Hence, harnessing mHealth to impactpopulation-based health outcomes will require thinkingcarefully about how interventions interact with, and in-fluence, their social milieu. As we have shown in thispaper, this includes diverse stakeholder interests, clinicwork routines, local organizational cultures and broaderhealth systems context.ConclusionRealizing the health benefits of the cell phone revolutionremains an area for concerted work. The danger is thatan over-emphasis on technological innovation, on the“bells and whistles”, as Tomlinson et al. [12] called them,can obscure the coupling of technical, social and polit-ical dynamics essential to effective implementation andlong-term scale-up. The drive towards “the next bestthing” could mean that funders and innovators abandonmHealth platforms, or do not provide sufficient path-ways to growth, just as they are beginning to showpromise in terms of scalability. Even now, some viewmHealth as “old-news”, while others remain skeptical orunsure of how, and in what ways, mHealth platformscan effectively move into the commercial market, espe-cially for vulnerable populations, whether in Africa orNorth America. This is all to say that research networksthat have developed still nascent but promising prod-ucts, like WelTel, are placed in a tricky position. As wehave endeavored to show in this paper, constantinnovation is required not only in technology but also innetwork building, institutional growth and partnershipdevelopment.Endnotes1Where quotes are not qualified by participant cat-egory, it is because the statement was used by multipleparticipants and represents a widely shared perspective.AcknowledgementsWe would like to thank the many research participants and otherstakeholders who contributed to this research, both in Canada and KenyaWe would also like to thank Shaina Pennington for administrative support atUBC.FundingThe study was funded by a Canadian Institutes of Health Research (CIHR)Foundation Grant to UBC.Availability of data and materialsData sharing is not applicable to this article as no datasets were generatedor analysed during the current study.Authors’ contributionsKB was responsible for the study conceptualization and design, datacollection, analysis, interpretation, and writing of the manuscript. Otherauthors (MM, AK, KS, RL) contributed to editing the final manuscript,assistance with overall interpretation and in help facilitating the studyinterviews. All authors read and approved the final manuscript.Ethics approval and consent to participateThe study was approved by the University of British Columbia’s ClinicalResearch Ethics Board (H16–00189), and Amref’s Ethics and Scientific ReviewCommittee (AMREF-ESRC P161/2015).Consent for publicationNot applicableCompeting interestsThe technology platform (WelTel/SMS) has been developed by a non-profitorganization, WelTel mHealth Society, and a private company, WelTel Incor-porated. Dr. Richard Lester has financial as well as professional interests inboth organizations. MM was the study PI on the WelTel Oak Tree Pilot Study.AK was the coordinator of the WelTel Kenya2 Grand Challenges Canada pro-ject in Northern Kenya.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Department of Anthropology & Emerging Pathogens Institute, University ofFlorida, 2055 Mowry Road, Gainesville, FL 32610, USA. 2Department ofMedicine, Division of Infectious Diseases, University of British Columbia,E600B – 4500 Oak Street, Vancouver, BC, Canada. 3Women’s Health ResearchInstitute, British Columbia Women’s Hospital, Vancouver, BC, Canada. 4OakTree Clinic, BC Women’s Hospital, Vancouver, BC, Canada. 5WelTelInternational mHealth Society, PO Box 50197 00100, Nairobi, Kenya.Received: 14 November 2016 Accepted: 8 November 2017References1. Sabaté E, editor. Adherence to long-term therapies: evidence for action.Geneva, Switzerland: World Health Organization; 2003.2. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353(5):487–97.3. Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up tothree years on treatment in sub-Saharan Africa, 2007–2009: systematicreview. Tropical Med Int Health. 2010;15(s1):1–15.4. Castro A. Adherence to antiretroviral therapy: merging the clinical andsocial course of AIDS. PLoS Med. 2005;2(12):e338. doi:10.1371/journal.pmed.0020338.5. Haynes RB, McKibbon KA, Kanani R. Systematic review of randomised trialsof interventions to assist patients to follow prescriptions for medications.Lancet. 1996;348(9024):383–6.6. ICD (2012) Information and communications for development: maximizingmobile. Available: http://www.worldbank.org/ict/IC4D2012. Accessed 10June 2016.7. Aranda-Jan CB, Mohutsiwa-Dibe N, Loukanova S. Systematic review on whatworks, what does not work and why of implementation of mobile health(mHealth) projects in Africa. BMC Public Health. 2014;14(1):1.8. Mills EJ, Lester R, Thorlund K, Lorenzi M, Muldoon K, Kanters S, et al.Interventions to promote adherence to antiretroviral therapy in Africa: anetwork meta-analysis. Lancet HIV. 2014;1(3):e104–11.9. Chib A, van Velthoven MH, Car J. mHealth adoption in low-resourceenvironments: a review of the use of mobile healthcare in developingcountries. J Health Commun. 2015;20(1):4–34.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services Submit your next manuscript to BioMed Central and we will help you at every step:Bardosh et al. Globalization and Health  (2017) 13:87 Page 15 of 1510. Leon N, Schneider H, Daviaud E. Applying a framework for assessing thehealth system challenges to scaling up mHealth in South Africa. BMC MedInform Decis Mak. 2012;12(1):1.11. Gurman TA, Rubin SE, Roess AA. Effectiveness of mHealth behavior changecommunication interventions in developing countries: a systematic reviewof the literature. J Health Commun. 2012;17(sup1):82–104.12. Tomlinson M, Rotheram-Borus MJ, Swartz L, Tsai AC. Scaling up mHealth:where is the evidence? PLoS Med. 2013;10(2):e1001382.13. Allotey P, Reidpath DD, Ghalib H, Pagnoni F, Skelly WC. Efficacious, effective,and embedded interventions: implementation research in infectious diseasecontrol. BMC Public Health. 2008;8(1):1.14. Damschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC.Fostering implementation of health services research findings into practice:a consolidated framework for advancing implementation science.Implement Sci. 2009;4(1):50.15. Lester RT, Ritvo P, Mills EJ, Kariri A, Karanja S, Chung MH, et al. Effects of amobile phone short message service on antiretroviral treatment adherence inKenya (WelTel Kenya1): a randomised trial. Lancet. 2010;376(9755):1838–45.16. van der Kop ML, Ojakaa DI, Patel A, Thabane L, Kinagwi K, Ekström AM, et al.The effect of weekly short message service communication on patientretention in care in the first year after HIV diagnosis: study protocol for arandomised controlled trial (WelTel retain). BMJ Open. 2013a);3(6):e003155.17. van der Kop ML, Memetovic J, Smillie K, Coleman J, Hajek J, Natasha VanBorek MA, et al. Use of the WelTel mobile health intervention at atuberculosis clinic in British Columbia: a pilot study. J Mob Technol Med.2013b;2(3):7–14.18. Mwai L. Scaling up WelTel: an evidence-based, patient-centred mHealthintervention. Ann Glob Health. 2015;81(1):195.19. De Vera MA, Sadatsafavi M, Tsao NW, Lynd LD, Lester R, Gastonguay L, et al.Empowering pharmacists in asthma management through interactive SMS(EmPhAsIS): study protocol for a randomized controlled trial. Trials. 2014;15(1):488.20. Jongbloed K, Friedman AJ, Pearce ME, Van Der Kop ML, Thomas V,Demerais L, et al. The cedar project WelTel mHealth intervention for HIVprevention in young indigenous people who use illicit drugs: studyprotocol for a randomized controlled trial. Trials. 2016;17(1):1.21. Murray MC, O’Shaughnessy S, Smillie K, Van Borek N, Graham R, Maan EJ,et al. Health care providers’ perspectives on a weekly text-messagingintervention to engage HIV-positive persons in care (WelTel BC1). AIDSBehav. 2015;19(10):1875–87.22. Smillie K, Borek NV, Kop MLVD, Lukhwaro A, Li N, Karanja S, et al. Mobilehealth for early retention in HIV care: a qualitative study in Kenya (WelTelretain). Afr J AIDS Res. 2014;13(4):331–8.23. Hampshire K, Porter G, Owusu SA, Mariwah S, Abane A, Robson E, et al.Informal m-health: how are young people using mobile phones to bridgehealthcare gaps in sub-Saharan Africa? Soc Sci Med. 2015;142:90–9.24. Rowe AK, de Savigny D, Lanata CF, Victora CG. How can we achieve andmaintain high-quality performance of health workers in low-resourcesettings? Lancet. 2005;366(9490):1026–35.25. Yamey, G (2011). Scaling up global health interventions: a proposed frameworkfor success. PLoS Med. 8(6):e1001049. doi:10.1371/journal.pmed.1001049.26. Street, A. (2014). Biomedicine in an unstable place: Infrastructure andpersonhood in a Papua New Guinean hospital. Duke University Press:Durham, North Carolina.27. Schneider H, Hlophe H, van Rensburg D. Community health workers andthe response to HIV/AIDS in South Africa: tensions and prospects. HealthPolicy Plan. 2008;23(3):179–87.28. Hall CS, Fottrell E, Wilkinson S, Byass P. Assessing the impact of mHealthinterventions in low-and middle-income countries–what has been shownto work? Glob Health Action. 2014;7.29. Curioso WH, Mechael PN. Enhancing ‘M-health’with south-to-southcollaborations. Health Aff. 2010;29(2):264–7.30. Gagnon MP, Ngangue P, Payne-Gagnon J, Desmartis M. M-health adoptionby healthcare professionals: a systematic review. J Am Med Inform Assoc.2016;23(1):212–20.31. Bardosh, K (2015). Deadly Flies, Poor Profits and Veterinary Pharmaceuticals:Sustaining the Control of Sleeping Sickness in Uganda. Med Anthropol.34(4):338–52.•  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submit


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items