Mobile Phone Tools for Field-Based Health care Workers in Low-Income Countries

MOUNT SINAI JOURNAL OF MEDICINE 78:406–418, 2011 406

Mobile Phone Tools forField-Based Health care Workers in

Low-Income CountriesBrian DeRenzi, MS,1 Gaetano Borriello, PhD,1 Jonathan Jackson, MEng,2

Vikram S. Kumar, MD,2 Tapan S. Parikh, PhD,3 Pushwaz Virk, MD, MPH,2

and Neal Lesh, PhD, MPH2

1University of Washington, Seattle, WA; 2Dimagi, Inc., Charlestown, MA; 3University of California, Berkeley,Berkeley, CA

OUTLINEOVERVIEW OF MOBILE SUPPORT FOR

HEALTH WORKERSHealth System Function 1:

Data CollectionHealth System Function 2:

Training and Access toReference Material for Health Workers

Health System Function 3:Facilitating Communication amongHealth Workers

Health System Function 4:Providing Job Aids andDecision Support

Health System Function 5:Supervision of Health Workers

Health System Function 6: PromotingHealthy Behaviors in the Population

CHALLENGES FOR MOBILE TOOLS

NEED FOR MOBILE TOOLS IN COMMUNITY HEALTH

PROGRAMS

SURVEY OF INFORMATION AND COMMUNICATION

TECHNOLOGY SYSTEMS FOR COMMUNITY

HEALTH WORKERSDiscretionary SystemsInformation and Communication

Technology–Based SystemsComputer-Supported Supervision

for Information and CommunicationTechnology–Based Systems

Lessons LearnedCONCLUSION

Address Correspondence to:

Brian DeRenziUniversity of Washington

Seattle, WAEmail: [email protected]

ABSTRACT

In low-income regions, mobile phone–based toolscan improve the scope and efficiency of fieldhealth workers. They can also address challengesin monitoring and supervising a large number ofgeographically distributed health workers. Severaltools have been built and deployed in the field,but little comparison has been done to help under-stand their effectiveness. This is largely because noframework exists in which to analyze the differentways in which the tools help strengthen exist-ing health systems. In this article we highlight 6key functions that health systems currently performwhere mobile tools can provide the most benefit.Using these 6 health system functions, we compareexisting applications for community health work-ers, an important class of field health workers whouse these technologies, and discuss common chal-lenges and lessons learned about deploying mobiletools. Mt Sinai J Med 78:406–418, 2011. 2011Mount Sinai School of Medicine

Key Words: CHW, CommCare, community healthworkers, ICT4CHW, mHealth, mobile field workers,mobile health, mobile phones.

Many personal digital assistant (PDA)- and phone-based systems have been developed for healthworkers in low-income settings. Applying mobiledevices and applications to health problems in low-and middle-income countries (LMICs) is commonlyreferred to as mHealth. Using mobile technologyto address these problems provides many advan-tages compared with traditional approaches. Forexample, mobile data collection is much fasterthan collecting data on paper and later enter-ing the results into a computer database, andhas a reduced risk of transcription error. Mobileapplications for data collection or decision sup-port ensure branching logic is correctly followed

Published online in Wiley Online Library (wileyonlinelibrary.com).DOI:10.1002/msj.20256

2011 Mount Sinai School of Medicine

MOUNT SINAI JOURNAL OF MEDICINE 407

and that only valid data values are accepted.Mobile devices allow rich forms of data to becaptured, including pictures, audio, video, or geopo-sition information. Linking mobile workers througha central server helps manage referrals to otherhealth facilities and aids interclinician communica-tion. These are just a few examples of how mobilephone–based systems can help improve health carein LMICs.

During the last few years, there has been workthat extends these advantages to community healthprograms; this is informally known as ICT4CHW(Information and Communication Technology forCommunity Health Workers).1 Despite the manyprojects in the ICT4CHW space,2–10 relatively littlehas been formally published and there has been nocomprehensive effort to describe the systems relativeto one another in an effort to determine what factorsinfluence the success or failure of each. The objec-tive of this article is to identify trends, challenges,and opportunities in this new field.

Despite the many projects inInformation and CommunicationTechnology for Community HealthWorkers (ICT4CHW), relativelylittle has been formally publishedand there has been nocomprehensive effort to describethe systems relative to one anotherin an effort to determine whatfactors influence the success orfailure of each.

In this article we first identify 6 important healthsystem functions (HSFs), or common functions that ahealth system must perform, where mobile applica-tions have been used. Second, we discuss commonchallenges for deploying mHealth tools. Third, weprovide a brief overview of community health pro-grams, illustrating the relevance of each of the6 HSFs. Fourth, we present a comprehensive reviewof mobile applications designed for use by commu-nity health workers (CHWs) that incorporate many ofthese different HSFs. We also identify a set of Infor-mation and Communication Technology (ICT)-basedsystems that are meant to be used in every visit,and target many or all of the HSFs. Finally, we elab-orate on lessons learned and potential benefits ofICT-based systems.

OVERVIEW OF MOBILE SUPPORT FORHEALTH WORKERS

Mobile phones increasingly offer many featuresbeyond voice and text messaging. Such phones arerapidly becoming cheaper, more powerful, and morecommon in remote areas of low-income regions.These phones bring increased Internet connectivelythrough general packet radio service and other cel-lular technologies, which have contributed to theincreased adoption of mobile tools for supportinghealth workers. We identify the following 6 HSFs inwhich mobile tools support health workers in LMICs:data collection, training and access to referencematerial, communication between health workers,decision support, supervision, and promoting healthybehaviors in the population.

Health System Function 1:Data Collection

Health systems require many forms of data collec-tion, such as household surveys, routine reportingfrom clinics, and supervisory reports. Typically, thesedata are collected on paper forms and later enteredinto a computer database. If high accuracy is needed,2 data clerks will digitize the forms and discrepanciesare resolved. Mobile data collection systems have thebenefit that users capture data directly into a PDAor phone and it is uploaded directly to a centralizedserver.

There are a number of PDA- and phone-basedsystems that have been developed to improve datacollection and decision support in LMICs. Table 1summarizes the advantages, as well as the chal-lenges, of mobile phone–based survey instruments.These systems offer many advantages compared withpaper-based data collection. The data are collectedmore quickly and with reduced risk of transcriptionerror. Additionally, the data-collection software run-ning on the mobile device can ensure that branchinglogic–eg, only ask the pregnancy question if thegender is female–is correctly followed and that onlyvalid data values are accepted.

Recent advances in mobile phone capabilitiesare making many innovative data-collection strategiespossible. Global positioning system (GPS) technol-ogy built into a mobile phone can be used totrack the health worker and identify the location ofhomes in the absence of mapping systems.11 Mobilephone–based applications have been used in a widevariety of surveillance activities.12 These include gath-ering data on respiratory illness,13 maternal and childhealth,14 and child nutrition.15

DOI:10.1002/MSJ

408 B. DERENZI ET AL.: MOBILE PHONE TOOLS FOR FIELD-BASED HEALTH CARE WORKERS

Table 1. Advantages and Challenges of Mobile Phone–Based Data Collection.

Advantages Challenges

Reduced errors fromtranscription

Data exchange andinteroperability issuesbetween different systems

More timely surveycompletion and accessto data

Coordination andcommunication constraintsleading to duplication offunctionality acrosssystems

Support for complexbranching logic,making surveys easierto administer

Complexity in deployingworkflow management insystems developed fordata collection

Automatic constraints onquestions (eg, notemperature >110◦F)

Privacy concerns withwireless transmission

Better adherence tosurvey questions

Loss or theft of mobiledevices

Reduced cost per surveyover long term

Poor or absent connectivity

Difficulty keeping batteriescharged

Several open-source data-collection systems forphones have emerged recently. Open source meansthat the original source code is available for down-load free of charge so that anyone can modify orexpand upon the original application. Companieswho release the source code of their products oftenmake their money by supporting the software, mak-ing custom modifications, or handling the installationand infrastructure requirements for customers.

EpiSurveyor16 from DataDyne enables its usersto create a data-collection form, download it onto aJava-enabled cell phone to collect data, and send thatdata to a central database. It is cloud-based, and at thetime of writing in early March 2011, has >3200 usersaround the world with >134,000 forms collected.EpiSurveyor allows for geo-tagging of forms and hasadvanced logic functionality. Open Data Kit (ODK)17

is a free and open-source collection of tools that helporganizations collect, aggregate, and visualize thedata. Built on the Android platform, ODK can replacepaper forms to enable capture and playback of multi-ple data types, including text, location, photos, video,audio, and barcodes. It also supports editing savedforms, question grouping, constraints, complex logic,and multiple languages. Many of the mobile-phonedata-collection projects are based on a few coreopen-source technologies, including JavaRosa andRapidSMS. For example, both EpiSurveyor and ODKare built upon the JavaRosa software platform. Theconsistent use of these platforms for data collection

has enabled communities of developers and imple-menters to collaborate and build on each other’sexperiences.

Health System Function 2:Training and Access toReference Material for Health Workers

A core activity of any health system is the initialand continuous training of its health workers. Healthworkers of every cadre need to be trained beforethey begin their work and retrained frequently asnew procedures are introduced. Technologies thathave the potential to make training more efficient,less expensive, or more effective can greatly benefita health system.

In Uganda, AED-SATELLIFE is delivering clinicalcontent to health care providers working out ofcommunity health centers in remote areas of Ugandausing handheld computers, wireless access points,and wireless communications infrastructure.18

Health System Function 3:Facilitating Communication amongHealth Workers

Health workers must often talk to each other foradvice or expertise, or to manage referrals to otherclinics. Mobile phones provide an obvious benefit forfacilitating communication among distributed work-ers in LMICs. Various studies have documented theuse of cell phones by medical residents,19 healthworkers,20 and nurse midwives21 to communicatewith peers, supervisors, or patients on health issues.World Vision’s Aceh Besar Midwives with MobilePhones Project was started to use mobile communi-cations and technology to facilitate, accelerate, andimprove the quality of health services by connectingmidwives to obstetrician-gynecologists.22

Mobile phones are used for coordination inresponding to emergencies, consulting with subspe-cialists, accessing second-line staff, obtaining consentor permission for action, and receiving and dis-cussing results from lab tests. In other instances,birth attendants coordinate care with delivery nursesand physicians in cases of complicated births. Physi-cians are also using smartphones to access sensor orimage data on their mobile phones.

Telemedicine is an important example of howhealth providers can communicate remotely. Projectshave used phones to capture images for assessmentby specialists,23,24 and have even been used to helpassess stroke victims.25 Dimagi, along with the Centre

DOI:10.1002/MSJ


for Infectious Disease Research in Zambia and oth-ers, implemented a cervical screening project wherenurses were trained to perform visual inspection withacetic acid aided by a mobile phone application thatsent images of the cervix for cervical cancer screen-ing by an expert.26 University of Pennsylvania, inpartnership with ClickDiagnostics, has implementedpilots in Botswana in dermatology, radiology, cer-vical cancer, and oral medicine. Using cell-phonecameras, images are transmitted to remote physiciansfor interpretation and diagnosis.27 The mobile healthcare services mQure in India28 and Telemedicine Ref-erence Center Ltd in Bangladesh29 provide access tohealth care providers over cell phones along withshort message service (SMS)-based reminders. Theseservices have enabled patients in rural and remoteareas access to quality medical care.

Health System Function 4:Providing Job Aids andDecision Support

Numerous studies have shown that health care canbe improved through job aids and decision support.Several mobile applications have been designed tohelp clinicians adhere to the clinical guidelines. Thisis important, because improved adherence to clinicalguidelines can reduce mortality rates. In Tanzaniait was shown that the Integrated Management ofChildhood Illness (IMCI) guideline reduced childmortality.30 Unfortunately, it is also commonlyobserved that health workers do not fully adhere tothe procedures they have been trained to perform.31

Mobile applications can guide health workersthrough complex decision-support protocols andclinical algorithms by presenting one question at atime, with a question automatically determined by theanswer to the preceding question. For example, thee-IMCI program running on a PDA was shown to sig-nificantly improve clinicians’ adherence to protocolsfor treating childhood illness in a small study in ruralTanzania compared with clinicians using traditionalmethods.32

Health System Function 5:Supervision of Health Workers

In order for a service-delivery system to runeffectively, the workers must be consistently andadequately supervised. Improved supervision hasbeen shown to have a substantial impact on theperformance of a health care system.33 Supportivesupervision has been show to ‘‘consistently [have]

moderate to large effects’’ in combination with other‘‘managerial tools.’’34

For community health programs, supervision isoften a neglected and under-resourced component.11

In most cases, supervision is performed by staffwho ‘‘may not understand the CHWs or theirown role properly and. . .may resent the additionaltask.’’35 The need for supervision also holds truein high-resource settings in order to maintain CHWperformance.36 The majority of supervision of CHWsoccurs during follow-up in the field, where super-visors often accompany CHWs on home visits andreview any data collected by the CHWs. There area number of challenges for supervisors. They aregenerally unskilled, they lack tools and transportto perform their duties, they are often burdenedwith administrative and reporting responsibilities,they may miss planned supervision visits, and peopleseem to become demoralized to the point that ‘‘noone seems to care whether supervision is done.’’34

Mobile applications have the potential to enablemore effective supervision. When health workersuse mobile tools, their activity can be monitoredin near real-time, providing supervisors with theability to more quickly react to changes in healthworker behavior and provide corrections or positivefeedback. Remote guidance can be provided fromsupervisors, as well as automatic supervision in termsof motivating messages or reminders.

Health System Function 6: PromotingHealthy Behaviors in the Population

In addition to treatment, most health systems pro-mote preventive care and other healthy behaviorsin the population, as well as provide post-treatmentsupport. Mobile phones offer an attractive way toquickly reach large populations through the phonesthey own. A number of SMS- and mobile applica-tion–based interventions, including games, incentiveprograms, and chronic disease–management tools,have been deployed for preventive health.

The Claim Mobile project in Uganda37,38 is incor-porating a mobile-payment system for health carepayments and claims adjudication to speed up theprocess of reimbursements for sexually transmit-ted infection tests. Johnson & Johnson’s BabyCenterprovides advice and information to parents before,during, and after pregnancy through weekly SMSdelivered based on the mother’s delivery date. Voxivaoffers mobile solutions designed to educate partic-ipants, encourage behavior change toward healthylifestyles, and promote adherence to their treatmentprotocol in the areas of maternal health, smoking

DOI:10.1002/MSJ


cessation, diabetes, human immunodeficiency virus(HIV)/acquired immune deficiency syndrome (AIDS),flu, and fibromyalgia. Dimagi is running a NationalInstitutes of Health–funded study at Boston Med-ical Center in which HIV/AIDS patients are sentcustomized text messages to improve their compli-ance with antiretroviral therapy (ART) regimens. Ina similar project by WelTel Kenya, patients on ARTwho received weekly SMS messages from a clinicnurse and were required to respond within 48 hoursto those messages had significantly improved ARTadherence and rates of viral suppression comparedwith individuals in a control group.39

CHALLENGES FOR MOBILE TOOLS

Table 1 lists many common challenges for deployingmobile phones. There is no single solution to theseproblems, but most of these must be addressed forany mHealth deployment. The challenges and solu-tions presented here are pulled from experiencesof the authors and those shared on the ICT4CHWmailing list.1

Some of the challenges relate to maintainingand operating the handsets, including keeping thephones charged, dealing with areas of low connec-tivity, and managing the loss or theft of the mobilephones. The extent of these operational challengesvaries depending on the country and the implement-ing program, and must be taken into account in termsof calculating the total cost of introducing a mobileapplication into a health system.

Mobile applications may introduce or alter pri-vacy concerns. Any system that captures patient dataelectronically introduces risk during transmission andstoring of the data. Some mobile applications alsopresent the challenge of having patient data avail-able on multiple handsets, and many phone softwareplatforms make it difficult to protect access beyondbasic passwords.

Another challenge for organizations deployingelectronic solutions is how to link different toolstogether and how to link them to existing databases.Even within the same health program, there mightbe different data systems developed at different timesor from different funding sources. This leads to sep-arate platforms being run by government-supported,private, and grant-funded health care infrastructure.Many open standards have been proposed to addressthis issue; however, no single set of standards hasbecome universally accepted.

Even if we ignore data exchange betweendifferent systems, organizations utilizing the same

core platform often find it hard to leverage eachother’s software modifications due to the increasedoverhead of coordination and communication. Morecommonly, organizations are able to share bestpractices and lessons learned, but end up writingtheir own new modifications no matter how similarin functionality they may be to another existing effort.

Finally, the goals of many organizations areunclear when they deploy applications. They maybegin with a focus on mobile data collection toimprove an existing paper process, but they quicklywant to support actions resulting from the data col-lected. This presents a challenge for organizationsadopting mobile data collection technologies, as thetechnologies that are easy to deploy for data col-lection may not be suitable for creating customizedworkflow support.

The goals of many organizationsare unclear when they deployapplications. They may begin witha focus on mobile data collectionto improve an existing paperprocess, but they quickly want tosupport actions resulting from thedata collected. This presents achallenge for organizationsadopting mobile data collectiontechnologies, as the technologiesthat are easy to deploy for datacollection may not be suitable forcreating customized workflowsupport.

NEED FOR MOBILE TOOLS INCOMMUNITY HEALTH PROGRAMS

Despite large differences between programs andcountries where CHWs ‘‘can be men or women,young or old, literate or illiterate,’’40 several elementsare common across community health programs. Thevast majority of the programs involve training CHWsto visit homes and promote healthy behaviors. Insome programs the CHWs visit every household intheir catchment area, whereas in others they visittarget demographics (eg, newborns or HIV+ individ-uals). Some visits are routine, and others are targetedvisits (eg, following up with lab results). But in almost

DOI:10.1002/MSJ


all cases, the CHWs themselves are mobile, and thusmobile applications represent the primary means toprovide electronic support.

Additionally, all 6 uses of mobile phones men-tioned above (HSFs) are applicable to communityhealth programs. For example, the majority of pro-grams collect data (HSF 1) during visits to providelongitudinal care, track client health over time, andfulfill monitoring and reporting requirements. How-ever, in traditional CHW programs, much of the datacollected by CHWs is never analyzed and reportedback to the community where it was collected. Thus,ICT4CHW systems have the potential to provide anew source of health surveillance data.

The CHWs must be trained appropriately andcontinuously (HSF 2). Targeted follow-up traininghas been shown to lead to CHW performanceimprovements.34 Computer-based trainings havebeen found to be potentially less expensive and‘‘as effective as traditional methods.’’33

Access to health experts and health informationis another major challenge facing all levels of healthworkers (HSF 3). They ‘‘lack access to basic, practicalinformation to enable them to deliver safe, effectivecare.’’41 Written reference materials are the traditionalway of supplying this information.

A job aid (HSF 4), such as a checklist, worksheet,communication aid, or other set of short instructions,can be useful for CHWs to perform their routine tasks.In Zambia, CHWs were able to use rapid diagnostictests more quickly with the assistance of a job aidexplaining the steps.42

Rowe et al. found that ‘‘dissemination of writtenguidelines without additional interventions wasgenerally ineffective’’ with CHWs.33 The writtenguidelines being referred to are a verbose explanationof a task, as opposed to a job aid that summarizesthe task. This reinforces the point that adequatetraining is required and that CHWs need to belinked to health expertise, as opposed to only healthinformation.

Furthermore, a review found that poor perfor-mance of CHWs was not due to only a lack of knowl-edge and skills.33 One longitudinal study of CHWperformance found that although the first follow-up training intervention was successful, performancedecreased after a second refresher training.35

As discussed in the section on supervision (HSF5), there are a number of challenges facing supportivesupervision of CHWs. Difficulties with reporting,managing geographically distributed workers, andautomation of mundane tasks are all areas to addresswhen improving supervision with mobile tools.

Finally, one primary role of CHWs is to pro-mote healthy behavior among their clients (HSF 6).

Mobile phones have the potential to increase CHWeffectiveness by providing educational tools or sup-plementing their efforts with automated outreach inbetween CHW visits to clients who own phones.

SURVEY OF INFORMATIONAND COMMUNICATION

TECHNOLOGY SYSTEMS FORCOMMUNITY HEALTH WORKERS

In this section, we review ICT projects that specificallyaim to address the needs of community health workerprograms. In Table 2, we categorize the projects dis-cussed in this section by the HSFs that they target. Wealso compare the projects to one another in Table 3,where we look at the higher-level properties thatemerge.

Discretionary Systems

One class of ICT4CHW systems are those that can beused at the discretion of the CHW and tend to targetone particular use of mobile phones within healthcare.

Medic Mobile is an example of an ICT4CHWsystem that helps facilitate communication amonghealth workers (HSF 3). Medic Mobile is built ontop of the FrontlineSMS platform43 and providesa new communication medium over SMS forhealth facilities.7 The SMS messages are generallyunstructured, freeform text. As an example, messagesfrom the CHW might contain updates about patients,double-checking drug dose and use, questions abouta patient, or a note to alert clinics of a referral. Clinicsmay send messages to check on the status of a CHWor patient or ask the CHW to mobilize the community.One lesson from Medic Mobile is the value ofallowing a flexible system that the CHWs can decidehow to use. Patterns of usage emerge as the CHWssend messages that can be acted upon by clinic staff.

HealthLine provided low-literate CHWs access tohealth expertise (HSF 2) through an automated inter-active voice recognition system that would providehealth information.8 Although it was never deployed,the system was evaluated by CHWs in the Sindhprovince of Pakistan.44 The authors ran 3 differentpilot studies with 10 users each before running amore formal user study with 18 participants. Therewere 2 main conclusions: the literacy of a userdirectly impacts his/her ability to complete a task,and a well-designed speech interface can outperforma touch-tone interface, independent of literacy.

DOI:10.1002/MSJ


Table 2. HSF Improvement for ICT4CHW Systems.

HSF 1 HSF 4

ProjectData

CollectionHSF 2

ReferenceHSF 3

CommunicationDecisionSupport

HSF 5Supervision

HSF 6Behavior

Medic Mobile Unstructured SMS expert SMSHealthLine IVR systemGuideView Multimedia

guidesFirst Days Educational

andtestimonialvideo

ChildCount Child registration Weekly printedsheet,(planned)automatedreminders

ClickDoc Multimedia datacollection

Send to remoteexpert

Yes

CommCare Caseregistration∗

Automatedreminders

Yes Automatedreminders

Abbreviations: HSF, health system function; ICT4CHW, Information and Communication Technology for CommunityHealth Workers; IVR, integrated voice response; SMS, short message system.∗Each CommCare deployment can specify cases to be registered and followed up (eg, pregnant mothers, households, orsuspected cases of tuberculosis).

Table 3. Comparing Properties of ICT4CHW Systems.

Project License StatusUsed

Every VisitStructured

DataRegisterCases Phone Type

Medic Mobile Free Pilot/development (optional plug-in) Basic∗HealthLine Unreleased† Pilot BasicGuideView Unknown Pilot � Feature,‡ smart§

First Days Unreleased† Pilot � FeatureChildCount FOSS Deployed � � � BasicClickDoc Unknown Pilot � � � AndroidCommCare FOSS Deployed � � � Feature, Android

Abbreviations: FOSS, free and open source software; ICT4CHW, Information and Communication Technology forCommunity Health Workers.∗Basic phones support only voice, SMS, and the SIM Application Toolkit (very basic menu-only applications).†Unreleased means that the code is not publicly available, but it is free if the authors of the respective project arecontacted.‡Feature phones support custom applications, often written in Java.§GuideView is available on the feature phones and the iPhone, with more smartphone platforms planned.

GuideView provides automated job aids toCHWs (HSF 4). Originally designed for astronautsneeding to provide medical support for one another,it allows practitioners to design multimedia clinicalguides for use on phones and other devices.45 In acontrolled study of 10 users, it was found that voiceadvice was considered ‘‘indispensable,’’ though textwith still pictures and video were favored as well.46

Most recently, the system has been adapted for astudy with 50 CHWs in Colombia, where it was

found that using GuideView reduced errors by 35%and increased adherence to protocols by 35%.47

The First Days project helps CHWs promotehealthy behaviors in the population (HFS 6). FirstDays aims to ‘‘mobilize health workers to improvematernal and child health in rural India.’’48 CHWsare given phones with educational and testimo-nial videos. The goal is that during a householdvisit, the videos will prompt discussion and encour-age a client to adopt healthy practices. Drawing

DOI:10.1002/MSJ


on work in persuasive technologies and tutoredvideo instruction, a pilot of the First Days projectaimed to persuade village women to adopt healthierpractices and to increase the self-efficacy of CHWs inorder to motive them to improve their performance.Researchers found that videos engaged both CHWsand clients, CHW self-efficacy showed ‘‘marginallysignificant gains (P = 0.06)’’ and there were ‘‘modestlearning gains’’ by the CHWs. It was also found thatmaking the videos was very engaging and impor-tant individuals in the community participated ‘‘to asurprising extent.’’6

Information and CommunicationTechnology–Based Systems

The second set of systems we label ICT-based CHWsystems and define them as having the followingproperties: (1) Every visit is captured electronically(no visits without the mobile phone), (2) structureddata is collected, (3) data are sent in real time, and(4) the system allows CHWs to register and trackclients.

The properties of ICT-based CHW systems arerestated in Table 4 and the benefits of such systemsare detailed. Each ICT-based system tends to supportmany of the mobile functions described above.

One ICT-based system that relies on SMS is theChildCount project.2 It was designed for use with sim-ple phones, but was also extended to support manualpaper-based data collection as well in the event that aphone is not available. The CHWs follow a worksheetto generate structured texts that are either sent viaSMS or recorded on paper entered at a later time bydata entry clerks. For example, sending ‘‘new diallofatimata f 080408 Amie’’ via SMS would register a newfemale child named Fatimata Diallo. Her mother’sname is Amie and she was born on August 4, 2008.

Table 4. Properties and Benefits of ICT-Based Systems.

Properties ofICT-Based Systems

Benefits ofICT-Based Systems

Application used forevery visit

Automated reports on CHWactivity for supervisors

Structured data iscollected

Data can be used for healthsurveillance

Data is sent in real time Reports available in realtime for monitoring andevaluation

CHWs can register andtrack cases

Automated reminders aboutfollowing up on cases

Abbreviations: CHW, community health worker; ICT,information and communication technology.

ChildCount supports registering children aged<5 years, reporting measurements from a nutritionalscreening and reporting the result of a malariarapid-diagnostic test. The system is being expandedto include capturing child deaths, pregnancies, andbirths; support for diarrheal illness through oral rehy-dration salts and zinc; drug-inventory managementfor CHWs and clinics; and an alert system for clinics todispatch CHWs to follow up on sick patients.2 Child-Count is built on top of RapidSMS49 to automaticallyreceive and parse incoming SMS from the CHWs.The server will process and respond to SMS in realtime. For example, when it receives a nutritionalscreening data form, the system will classify thechild and recommend a treatment, if any. Finally,the server generates reports on CHWs and theirclients to enable better supervision by identifyingCHWs who are struggling and more targeted inter-ventions by looking for trends in the populationdata.

One of the interesting findings from ChildCountwas the value of providing printed lists to the CHWseach week, with a list of each of the CHW’s clientsand their status. The list would include any updatesmade by the CHW that week to their clients. Thisprovided a useful overview for a CHW, allowingthem to see which clients needed further attention.More important, this feedback loop would motivatethe CHW to continue to use the system.

ClickDoc is an ICT-based system that runs onthe Android operating system.50 It allows the healthworker to fill out surveys and capture pictures sothat complex cases can be sent to remote experts. Atele-dermatology solution was built in Egypt and pre-liminary results show that remote and on-site expertscame to the same diagnosis in 75% of cases.51 InBotswana the software was extended to add cervi-cal cancer screening, with almost 77% of screeningsdiagnosed the same between remote and on-site.52

In both the Egypt and Botswana studies, the dis-agreement was the result of ‘‘insufficient history’’ ofthe patient’s situation being communicated to theremote professional. In Bangladesh, CHWs visitinghouseholds collect data on mothers and children.Results are automatically analyzed and categorizedso that doctors can review cases with high-risk pro-files. Information is then returned to the CHW (andthen the client) either immediately or the next day,depending on doctor availability.52

CommCare, which several of the authors haveworked on, runs on feature phones or Android andsupports various modules so that it can be customizedto a particular CHW program.4 For example, a Comm-Care deployment run by Pathfinder International andD-Tree International in Dar es Salaam, Tanzania,

DOI:10.1002/MSJ


Number of household visit forms received per weekWeek

2010

800N

umbe

r of

form

s re

ceiv

ed

600

400

200

0

Wee

k 36

Wee

k 37

Wee

k 38

Wee

k 39

Wee

k 40

Wee

k 41

Wee

k 42

Wee

k 43

Wee

k 44

Wee

k 45

Wee

k 46

Wee

k 47

Wee

k 48

Wee

k 49

Wee

k 50

Wee

k 51

Wee

k 2

Wee

k 3

Wee

k 4

Wee

k 5

Wee

k 6

Wee

k 7

Wee

k 8

Wee

k 9

Wee

k 10

Wee

k 11

Wee

k 12

2011

Fig 1. Number of household follow-up forms received.

supports CHWs who are registering and followingup with patients who need chronic care–the vastmajority being HIV+.

A CommCare deployment in Dodoma, Tanza-nia, supports a module for routine household visitsand a module for pregnant women. The householdvisit module allows households to be registered andtracked over time. If there is a serious illness orwound, a referral can be generated to refer thehousehold member to the health facility. These refer-rals are also tracked so that the CHW follows up withthe household to ensure care was received.

The pregnancy module allows CHWs to registerpregnant women and track them over time. Itincludes a checklist of healthy actions to completebefore the woman gives birth. Danger signs areassessed at every visit in order to identify high-riskpregnancies, and the case is closed once the womangives birth.

In Dodoma, there are 35 CHWs using Comm-Care. In the first 9 months that the deployment wasrunning, we received >12,600 home-visit follow-upforms. Figure 1 shows the number of householdfollow-up forms received, broken down by weekduring the fourth quarter of 2010 and the first twoweeks of 2011. As expected, there was a decrease inactivity during the holiday period. Out of all of thehousehold follow-up visits captured, a member of thehousehold is referred to the health facility 24% of thetime, with 90% of the referrals being about sicknessor a wound. The remaining 10% are for pregnantwomen being referred for antenatal care or for birthregistration of newborns.

The mobile phones send their data to aserver called CommCareHQ. This is a Django-basedplatform to be used by supervising staff of theimplementing partner to view the status of the CHWsand the clients. It can produce automated reportsfor staff and is being expanded to produce andsend automated reminders. Similar to FrontlineSMS,43

CommCareHQ also allows SMS messages to besent and received using RapidSMS as a backendsupport.

An important shift in CommCare has been tointroduce multimedia elements to reinforce each stepin the forms, including both audio and icon orimages. This helps CHWs with lower literacy whomight not able to use a text-based system. It alsoturns the application into a powerful educationaltool, borrowing elements from the First Days projectdescribed in the previous section. The CHWs showtheir clients the images and play the audio clips tohelp reinforce the points and also stimulate morediscussion.

Computer-Supported Supervisionfor Information and CommunicationTechnology–Based Systems

Improved supervision (HSF 5) is an area of greatpromise for ICT4CHW systems, which can addressseveral of the challenges of supervision, includingcommunication difficulties between supervisors and

DOI:10.1002/MSJ


Improved supervision (HealthSystem Function 5) is an area ofgreat promise for ICT4CHWsystems, which can address severalof the challenges of supervision,including communicationdifficulties between supervisorsand community health workers.

CHWs. These systems can provide higher-qualityinformation about the CHWs and offer a commu-nication medium, given that supervisors and CHWsrarely see one another.

Capturing data immediately from the field avoidsthe delay of using paper records that are later enteredby a data-entry clerk. This enables data to be aggre-gated in real time, allowing organizations to makeinformed decisions and respond more promptly toneeds in the field. The data from ICT-based systemcan be used to support supervisors in a variety ofways, including:

Client StatusAs a result of capturing structured data during routinevisits, the status of clients can be computed, makingit easy to construct reports showing which of thoseclients have been visited recently–or not–and whichare routinely having trouble. Communicable diseaseincidence can also be explored on an aggregate levelacross locations.

For example, if the CHWs were collectingsickness data at a household level with GPS, it wouldbe possible to map the incidence of sickness to lookfor hot spots. Even if GPS is not used, the data couldstill be mapped by CHW catchment area.

CHW StatusSimilar to client status, the performance of individualCHWs can be tracked. The number of visits per daycan be computed, and supervisors can look at themost recent visits to ensure the CHWs are visitingtheir clients regularly enough.

As an example, CommCareHQ, the backendfor the CommCare project, generates email reportsindicating the activity of each CHW using CommCare.Authorized users can receive reports as frequentlyas desired. The reports show the number of formssubmitted over each of the last several days, as wellas the total number of active clients for each CHW,and how many of those clients are overdue for a visit.

Automated RemindersIn rural areas, time to care is extremely important. InUganda in 2008, Kallander et al. looked at fatal pneu-monia cases in children and cited ‘‘delays in seekingcare’’ as one of 3 primary reasons for the deaths.53

Timeliness was also one of the metrics used byShankar et al. when measuring CHW performance.54

Automated reminders can be sent to ensurethe CHWs make timely visits. These remindersare enabled because the application captures everyvisit, allowing important events to be tracked. Forexample, a reminder could be sent to a CHW who hasoverdue visits, as defined above, requesting that heor she follow up with those clients. Similar reminderscan be sent to CHWs who have referred clients to thehealth facility to ensure the client was able to receivecare.

Activity ReportsRamachandran et al.6 reported that CHWs are oftenare not performing the required number of homevisits. Sending activity reports that briefly state theamount of work done (eg, number of householdsvisited or number of overdue clients) to a group ofCHWs may be a way to take advantage of socialpersuasion to improve CHQ performance, thoughfurther exploration and design is required.

Lessons Learned

There are a number of lessons learned from theseearly deployments of ICT4CHW systems. Similar tothe challenges listed earlier, the lessons presentedhere are pulled from experiences of the authors andthose shared on the ICT4CHW mailing list.1

An important lesson learned is not to automatebroken systems. If a community health program isnot functioning well, then it is unlikely that addingan ICT4CHW component will solve the underlyingproblems. These systems are best seen as tools thatcan strengthen a program, rather than salvage it.

Another important lesson pulled from our reviewof these systems is the gap between how a CHWprogram was originally designed to work and howit is functioning in the field. This has a significantimpact when trying to introduce automation, becauseautomating the original design will fail if what ishappening in the field deviates significantly. Asan example, a community health program is oftendesigned to include a large number of visits andmaterial to be covered at each visit. However, CHWs

DOI:10.1002/MSJ


If a community health program isnot functioning well, it is unlikelythat adding an ICT4CHWcomponent will solve theunderlying problems. Thesesystems are best seen as tools thatcan strengthen a program, ratherthan salvage it. Another importantlesson is the gap between how aCHW program was originallydesigned to work and how it isfunctioning in the field.

may find it impractical to follow the program designfully. When an automated system that provides moretools for supervision is introduced, this tension isbrought to the surface. Ultimately, this is an advan-tage because it forces the CHWs and the program toconverge. The CHWs must more closely follow theprogram and the program must become more prac-tical. However, it also entails a lot of programmaticwork to revise the overall community health programto be more practical.

A related lesson is the need to allow substantialtime to iterate on any new system with real users. Atypical CommCare deployment will schedule severalmonths of ‘‘refinement,’’ which takes place after aninitial prototype has been designed with input fromexperts and users, but before a pilot. The refinementperiod can have as few as 5 users who meet witha field implementer frequently. During this stage,the system is modified frequently and new ideasare tested. After the refinement period, the pilotmight be run with 30–40 users. During this time, theapplication is changed much less frequently.

An important, albeit common, observation is thatcontext matters. It is important to understand thedynamics of how people use phones. For example,it can create excitement as well as conflict to givea CHW a phone if her husband does not have one.Similar to context, content matters as well. One les-son that has come up repeatedly is that getting themessages, images, or audio clips right can make adifference in how well an ICT4CHW system functions.

Almost all of the lessons above are variations ona commonly stated theme: technology is not the diffi-cult part–what matters most is how the technology isused. Incentives must also be aligned properly, anda mechanism to adapt the system over time must bepresent.

CONCLUSION

Community health workers have the potential to pro-vide health information, encourage healthy practices,and be an accessible point of entry into the nationalhealth system for rural and low-income populations.However, CHW programs are difficult to maintainbecause of challenges in monitoring and supervisinga large number of geographically distributed healthworkers. Mobile applications provide an opportunityto address these challenges by strengthening CHWprograms and making them more efficient.

In this article, we identified 6 key HSFs wheremobile phone applications are providing benefit. Wesurveyed peer-reviewed and grey literature publi-cations about mobile-phone applications, discussingthem in the context of the identified HSFs. We alsodiscussed the challenges and lessons learned frommHealth systems. Finally, we identified criterion forICT-based applications CHW systems and enumer-ated the benefits of such applications.

It is clear that using mobile phones to increaseCHW performance is an important area and that thereis a lack of work evaluating such interventions. Oneway to address this is for funding agencies to requirean evaluation component in the mHealth projectsthey fund. The technology providers can also pro-mote research about health outcomes by educatingtheir clients about research design and analysis andcollaborating with the implementation partners forsuch studies. The results of studying the impact oftechnology in this setting will help organizationsdesign and implement more effective interventions.Not only will they generalize across CHW programs,but they will also benefit mobile field workers inother domains, such as agriculture, microfinance, andlivestock.

ACKNOWLEDGMENT

We would like to express gratitude to all of theCHWs we have worked with, our project partners,field staff, and funding organizations. We are gratefulto the staff of D-tree for their partnership over theyears, which has greatly deepened our understandingof ICT4HW systems. We would also like to thankthe International Development Research Centre andthe National Science Foundation Graduate ResearchFellowship Program for funding this work. Finally,we are grateful to those who have helped shapethis manuscript through their edits and comments:Michael Buettner, Yaw Anokwa, the reviewers, andthe editors.

DOI:10.1002/MSJ


DISCLOSURES

Potential conflict of interest: Jonathan Jackson,Vikram S. Kumar, Pushwaz Virk, and Neal Lesh workfor Dimagi, Inc. which developed CommCare and isseeking further funding for CommCare through grantsand contracts. They could benefit financially from apositive perception of CommCare. Brian DeRenzi hasconsulted for Dimagi and D-tree on work related toCommCare and thus could also benefit financiallyfrom a positive perception of CommCare.

REFERENCES

1. ICT4CHW Google group. http://groups.google.com/group/ict4chw. Accessed July 7, 2010.

2. Berg M, Wariero J, Modi V. Every Child Counts–The Use of SMS in Kenya to Support the CommunityBased Management of Acute Malnutrition and Malariain Children Under Five. http://www.childcount.org/reports/ChildCount Kenya InitialReport.pdf. PublishedOctober 15, 2009. Accessed June 29, 2010.

3. ClickDiagnostics. http://clickdiagnostics.com. AccessedJuly 7, 2010.

4. Mhila G, DeRenzi B, Mushi C, et al. Using mobileapplications for community-bsed social support forchronic patients. Presented at: Health Informaticsin Africa Conference (HELINA); April 16–18, 2009;Abidjan, Cao te d’Ivoire.

5. eMOCHA. http://emocha.org. Accessed July 13, 2010.6. Ramachandran D, Canny J, Das P, et al. Mobile-

izing health workers in rural India. Presented at:ACM Conference on Human Factors in ComputingSystems; April 10–15, 2010; Atlanta, GA. http://portal.acm.org/citation.cfm?id=1753326.1753610.

7. Medic Mobile. http://medicmobile.org. AccessedMarch 7, 2011.

8. Sherwani J, Ali N, Mirza S, et al. HealthLine:speech-based access to health information by low-literate users. Presented at: IEEE/ACM InternationalConference on Information and CommunicationTechnologies and Development; December 2007;Bangalore, India.

9. Chib A. The Aceh Besar midwives with mobile phonesproject: design and evaluation perspectives usingthe information and communication technologies forhealth care model. J Comput Mediated Commun 2010;15: 500–525.

10. Mobile Technology for Community Health (MoTeCH).http://www.grameenfoundation.applab.org/section/ghana-health-worker-project. Accessed June 27, 2010.

11. Krishnamurthy R, Frolov A, Wolkon A, et al.Application of pre-programmed PDA devices equippedwith global GPS to conduct paperless householdsurveys in rural Mozambique. AMIA Annu Symp Proc2006; 991.

12. Shirima K, Mukasa O, Schellenberg JA, et al. The useof personal digital assistants for data entry at the pointof collection in a large household survey in southernTanzania. Emerg Themes Epidemiol 2007; 4: 5.

13. Diero L, Rotich JK, Bii J, et al. A computer-basedmedical record system and personal digital assistants

to assess and follow patients with respiratory tractinfections visiting a rural Kenyan health centre. BMCMed Inform Decis Mak 2006; 6: 21.

14. Anantraman V, Mikkelsen T, Khilnani R, et al.Handheld computers for rural health care: Experiencesfrom research concept to global operations. Presentedat Development by Design Conference; December 1–2,2002; Bangalore, India:1–10.

15. Blaschke S, Bokenkamp K, Cosmaciuc R, et al. UsingMobile Phones to Improve Child Nutrition Surveillancein Malawi. UNICEF Malawi and UNICEF Innovations;May 2009.

16. EpiSurveyor. http://www.episurveyor.org. AccessedMarch 6, 2011.

17. Open Data Kit. http://www.opendatakit.org. AccessedJune 25, 2010.

18. AED-SATELLIFE. Empowering health workers tosave lives: Uganda Health Information Network.http://www.healthnet.org/uhin. Updated February 28,2011. Accessed March 5, 2011.

19. Scott RE, Ndumbe P, Wootton R. An e-healthneeds assessment of medical residents in Cameroon.J Telemed Telecare 2005; 11(suppl 2): S78–S80.

20. Mechael PN. Exploring Health-Related Uses of MobilePhones: An Egyptian Case Study [PhD thesis]. London,UK: London School of Hygiene and Tropical Medicine;2006.

21. Mechael PN; with the Dodowa Health ResearchCenter. MoTeCH: mHealth Ethnography Report.http://www.grameenfoundation.applab.org/uploads/Grameen Foundation Final. The Grameen Foundation;August 1, 2009.

22. Chiba A, Lwina MO, Anga J, et al. Midwives andmobiles: using ICTs to improve health care in AcehBesar, Indonesia. Asian J Commun 2008; 4: 348–364.

23. Gormley RH, Quinley KE, Shih T, et al. Use ofmobile telemedicine for cervical cancer screening inGaborone, Botswana. http://clickdiagnostics.com/wp/press/documents. Published November 2009. AccessedMarch 7, 2011.

24. Tsai HH, Pong YP, Liang CC, et al. Teleconsultationby using the mobile camera phone for remotemanagement of the extremity wound: a pilot study.Ann Plast Surg 2004; 53: 584–587.

25. Schwamm LH, Holloway RG, Amarenco P, et al. Areview of the evidence for the use of telemedicinewithin stroke systems of care: a scientific statementfrom the American Heart Association/American StrokeAssociation. Stroke 2009; 40: 2616–2634.

26. Parham GP, Mwanahamuntu MH, Pfaendler KS, et al.eC3–a modern telecommunications matrix for cervicalcancer prevention in Zambia. J Low Genit Tract Dis2010; 14: 167–173.

27. McWilliams J. Connecting cell phones with medicine inBotswana. http://www.upenn.edu/pennnews/current/features/111110-4.html. Published November 11, 2010.Accessed January 11, 2011.

28. mQure. http://www.mqure.com/about.html.AccessedMarch 5, 2011.

29. Telemedicine Reference Center Ltd. http://www.trclcare.com/products.php. Accessed March 5, 2011.

30. Schellenberg JA, Bryce J, de Savigny D, et al. Theeffect of Integrated Management of Childhood Illnesson observed quality of care of under-fives in ruralTanzania. Health Policy Plan 2004; 1: 1–10.

DOI:10.1002/MSJ


31. Kelly JM, Osamba B, Garg RM, et al. Community healthworker performance in the management of multiplechildhood illnesses: Siaya District, Kenya, 1997–2001.Am J Public Health 2001; 10: 1617–1624.

32. DeRenzi B, Lesh N, Parikh T, et al. e-IMCI: improvingpediatric health care in low-income countries.Presented at ACM Conference on Computer-HumanInteraction (CHI 08); April 2008; Florence, Italy.

33. Rowe A, de Savigny D, Lanata C, et al. How canwe achieve and maintain high-quality performance ofhealth workers in low-resource settings? Lancet 2005;9490: 1026–1035.

34. Rennert W, Koop E. Primary health care for remotevillage communities in Honduras: a model for trainingand support of community health workers. Fam Med2009; 9: 646–651.

35. Rowe SY, Olewe MA, Kleinbaum DG, et al. Longitudi-nal analysis of community health workers’ adherenceto treatment guidelines, Siaya, Kenya, 1997–2002. TropMed Int Health 2007; 5: 651–663.

36. Gilson L, Walt G, Heggenhougen K, et al. Nationalcommunity health worker programs: how can they bestrengthened? J Public Health Policy 1989; 4: 518–532.

37. Martins HM, Jones MR. What’s so different aboutmobile information communication technologies(MICTs) for clinical work practices? A review ofselected pilot studies. Health Informatics J 2005; 11:123–134.

38. Ho MR, Owusu EK, Aoki PM. Claim mobile: engagingconflicting stakeholder requirements in health care inUganda. Presented at: International Conference onInformation and Communication Technologies andDevelopment; April 17–19, 2009; Doha, Qatar.

39. Lester RT, Ritvo P, Mills EJ, et al. Effects of a mobilephone short message service on antiretroviral treatmentadherence in Kenya (WelTel Kenya1): a randomisedtrial. Lancet 2010; 376: 1838–1845.

40. Lehmann U, Sanders D. Community Health Workers:What Do We Know About Them? The State of theEvidence on Programmes, Activities, Costs and Impacton Health Outcomes of Using Community HealthWorkers. Geneva: World Health Organization; March2007.

41. Pakenham-Walsh N, Bukachi F. Information needsof health care workers in developing countries: aliterature review with a focus on Africa. Hum ResourHealth 2009; 7: 30.

42. Harvey SA, Jennings L, Chinyama M, et al. Improvingcommunity health worker use of malaria rapid

diagnostic tests in Zambia: package instructions, jobaid and job aid-plus-training. Malar J 2008; 7: 160.

43. FrontlineSMS. http://www.frontlinesms.com. AccessedJune 29, 2010.

44. Sherwani J, Palijo S, Mirza S, et al. Speech vs. touch-tone: telephony interfaces for information accessby low literate users. Presented at: InternationalConference on Information and CommunicationTechnologies and Development; April 17–19, 2009;Doha, Qatar.

45. Iyengar M, Florez-Arango J, Garcia C. GuideView:a system for developing structured, multimodal,multi-platform persuasive applications. Presentedat: Persuasive 09: Proceedings of the FourthInternational Conference on Persuasive Technology;April 26–29, 2009; Claremont, CA. http://portal.acm.org/citation.cfm?id=1541948.1541990.

46. Iyengar MS, Sarkar S, Bacal K, et al. GuideView:structured multi-modal delivery of clinical guidelines.AMIA Annu Symp Proc 2005; 992.

47. Iyengar MS. The GuideView mHealth system. http://www.slideshare.net/gueste312b0/the-guideview-mhealth-system. Accessed July 8, 2010.

48. First Days. http://www.cs.berkeley.edu/∼divya/research/firstdays. Accessed June 29, 2010.

49. RapidSMS. http://www.rapidsms.org. Accessed June27, 2010.

50. Zimic M, Coronel J, Gilman RH, et al. Can the powerof mobile phones be used to improve tuberculosisdiagnosis in developing countries? Trans R Soc TropMed Hyg 2009; 6: 638–640.

51. Gormley RH, Kovarik C. The use of mobiletelemedicine for remote diagnosis of mucocutaneousand cervical lesions. Presented at: Health Informaticsin Africa Conference (HELINA); April 16–18, 2009;Abidjan, Cao te d’Ivoire.

52. Khan R. ClickDiagnostics experiences from Bangla-desh. http://groups.google.com/group/ict4chw/browsethread/thread/96dd344a921f124e. Published May 12,2010. Accessed July 7, 2010.

53. Kallander K, Hildenwall H, Waiswa P, et al. Delayedcare seeking for fatal pneumonia in children agedunder five years in Uganda: a case-series study. BullWorld Health Organ 2008; 5: 332–338.

54. Shankar AV, Asrilla Z, Kadha JK, et al. Programmaticeffects of a large-scale multiple-micronutrient supple-mentation trial in Indonesia: using communityfacilitators as intermediaries for behavior change. FoodNutr Bull 2009; 30(2 suppl): S207–S214.

DOI:10.1002/MSJ

Documents

Mobile Phone Tools for Field-Based Health care Workers in Low-Income Countries