16 X R D S • w i n t e r 2 0 1 2 • V o l . 1 9 • n o . 2

feature

seven factors for Designing successful mHealth projects Although mobile technology has the power to vastly improve healthcare delivery in developing regions, many issues can affect the success of mHealth systems.

By Atanu GaraiDOI: 10.1145/2382856.2382865

T he use of mobile technology for communicating information about medicine and public health is known as mHealth. Developed countries have been using mobile telephones, miniature laptops, and diagnostic devices in the health sector for many years. With the rapid proliferation of mobile telephony in developing countries,

health managers quickly recognized the potential for mHealth to strengthen their healthcare delivery systems. In fact, mHealth has created a unique opportunity to strengthen healthcare administration, care delivery, and communication among providers and

beneficiaries throughout the entire healthcare system.

Yet despite the promise of mHealth tools for development, there are many issues that hamper the provision of mHealth systems in developing con-texts. For example, resource con-straints typically mean that installa-tion of computer components, power backups, and broadband connections can be problematic, and providing training to health providers in using such devices can be expensive. More-over, the absence of reliable computing infrastructure in many rural areas fur-ther compounds the issue, especially as more and more health workers be-gin to use mHealth for healthcare pro-vision in these communities.

Despite these problems, mHealth systems can have many benefits and can be considered a standout among

ICT for development. In health ad-ministration, the management of health systems relies upon timely and accurate data on care delivery, along-side proper management of financial, human, and material resources. How-ever the collection of accurate data can often be problematic, although mobile systems can sometimes help to overcome this. For example, one evaluation study in a pilot project for nutritional surveillance reported a discard rate of 14 percent for paper-based reports because of issues like illegible handwriting, missing deci-mals, or outliers in the forms. Yet, when forms were submitted via SMS, error rate dropped to just three per-cent [1]. A number of projects have used SMS in various aspects of pro-gram management: ChildCount in Kenya to enroll beneficiaries [2], In-

tegrated Nutrition and Food Security Surveillance in Malawi to monitor nutrition programs [3], TRACnet in Rwanda to manage care for HIV/AIDS patients [4], and BloodBank SMS in Kenya to report the status of blood stocks for transfusions [5].

mHealth has also proven effec-tive in motivating the practice of certain healthy behaviors. In 2011, Catholic Relief Services, an interna-tional non-profit organization, col-laborated with mHealth technology provider Dimagi to implement the ReMiND (Reducing Maternal and New-born Deaths) Pilot Project. The ReMiND project uses a mobile-based educational and counseling tool for community health workers serving around 1,000 people in villages in the Kaushambi District of Uttar Pradesh, India. The tool includes a mobile

17X R D S • w i n t e r 2 0 1 2 • V o l . 1 9 • n o . 2

ences with a high degree of interest in the advertised product. However, users of mobile phones in developing coun-tries tend to overlook SMS from sources other than the ones from which they typically receive services. This means health information disseminated with-out the proper due care and attention stands to be untrusted or, worse still, ignored. A generalized message ran-domly distributed every so often could lack the seriousness and attention it de-serves by the intended beneficiary.

One way mass messaging can work is for advertising events like local polio or HIV/AIDS camps. For example, Proj-ect Masiluleke sent some 300 million “Please call me” messages using the un-structured supplementary service data (USSD) protocol to South Africans en-couraging them to undergo HIV/AIDS test. The average daily call volume to the National AIDS Helpline tripled within a few months of this campaign [7].

SMS campaigns tend to work best for promoting awareness of specific events, products, and services. Pro-motion through mobile communi-cation is undertaken mostly by me-dia agencies as part of their media campaigns. Since these campaigns are one-off, mHealth initiatives can

phone-based pregnancy checklist with audio and visual prompts that help workers systematically counsel and assess pregnant women during regular home visits. The experiment with 10 healthcare workers suggests mobile multimedia content is help-ful for managing tasks and provid-ing information to beneficiaries. Other projects demonstrated the viability of mobile video in counsel-ing, including mSakhi, organized by IntraHealth International in Uttar Pradesh, India, and ASHA Assist, a project in Odisha, India, organized by the University of California-Berke-ley. Since multimedia appears to be more impactful than textual or graphical information alone [6], pre-recorded mobile videos displayed on the mobile phones of health workers have the potential to impart knowl-edge on natal issues among pregnant women in these contexts.

As more and more mHealth proj-ects are implemented across the globe, they offer us strategic lessons and insights that can help us define factors that enable mHealth projects to achieve their objectives. Learning from these projects can help program managers, investors, healthcare sec-

tor workers, and the telecommunica-tions industry at large to develop and implement projects. But what are the issues that influence the success of a particular mHealth project, and how can service providers design to accom-modate these issues? Listed are seven factors that will affect the success of mHealth implementations. Of course, any list of success factors is non-static and evolving, especially since more and more projects offer unique les-sons from their individual successes or failures. The list here is based on some strategic examples and high-lights some salient issues that have come to the fore in past projects.

fAcToR 1: mHeAlTH is noT mAss meDiAMany organizations are still dissemi-nating SMS or calls in the way that television and radio broadcast their content. This process essentially en-tails partnering with a mobile service provider or third-party SMS gateway to disseminate SMS or calls to all the num-bers registered with the mobile opera-tor. For health campaigns, bombarding the population with SMS or calls works as advertisements and this method works best for a limited number of audi-

members of the community in Chhattisgarh, India, undergoing training on the accredited nurse and midwife (aNm) mhealth application, (right). a healthcare worker using the mother and child tracking system (mCTs) mobile mhealth application, (top).

Pho

togr

aphs

cou

rtes

y of

vir

tuos

os S

olut

ions

Pri

vate

lim

ited

18 X R D S • w i n t e r 2 0 1 2 • V o l . 1 9 • n o . 2

feature

seldom run for long, primarily be-cause many such campaigns are of limited duration. In the area of ma-ternal and child health, pregnant women will require messages on dieting, danger signs, baby growth, and other topics based on the gesta-tional period. This means messages can be matched to target pregnant

women based on menstrual cycle or their child’s gestational period. Such targeting is useful because beneficiaries may ignore healthcare messages if the system does not send the message based on the profiles of the intended recipients.

fAcToR 2: DeliVeR mUlTiple seRVices THRoUgH mHeAlTH Many health managers think of mHealth as yet another tool for data collection and communication about health behavior. They can hardly be faulted for this biased perception as many mHealth projects have concen-trated only on those two aspects. How-ever, mHealth has the potential to pro-vide instruction and guidance to steer the patients toward attaining certain behavioral goals. In doing so, mHealth can provide guidance to access and uti-lize health services.

Since research suggests the impact of audio-visual communication is far greater than textual or audio commu-nication alone [6], mHealth services using second generation, 2-G mobile telephony are likely to have a limited role in raising awareness about key health behaviors. But mobile com-munication can improve self-effica-cy, social support, and interactions between patients and doctors, while the media itself can increase aware-ness and knowledge on specific health issues. Mobile communica-tion can allow providers to tailor the delivery of information according to time, geography, desired frequency of messages, and other orientations demanded by the health and social, cultural, and behavioral patterns of the patient. As such, mHealth prac-tices can cover more health issues, as compared to mass media. For exam-ple, mass media has been successful-ly exploited to raise awareness about polio in India and knowledge about polio is now very high among citizens. However, there are many other vac-cines that were not covered in media campaigns, and, as a result, knowl-edge about those vaccinations is very low. Media campaigns and even inter-personal communication will require more resources for providing appro-priate information and guidance, as compared to mobile communications.

fAcToR 3: eVolVe BUsiness moDels foR cURATiVe AnD pReVenTiVe HeAlTH Program managers are increasingly looking for ways to ensure financial sustainability, both in profit-making and not-for-profit projects. While most mHealth projects do seem to be finan-cially sustainable in the medium- to long-term, financial and other con-straints mean that beneficiaries may be less likely to pay for SMS or phone calls that aim to promote healthy be-haviors. This means projects target-ing preventive health segments may not be able to charge their customers for services, making financial sustain-ability problematic.

On the other hand, beneficiaries may pay for services related to emer-gency and curative health issues, espe-cially as requirement for treatment is immediate and often visiting a doctor or treatment facility can be difficult. A number of mobile operators have started providing counseling on mo-bile phones for a fee, which the opera-tor deducts from the talk time balance of the caller [8]. Projects in preven-tive health areas like smoking cessa-tion and maternal and child health typically depend on public healthcare budgets that are supported by taxes or other financial sources. One issue is governments in developing countries may be reluctant to support mHealth projects targeted at preventive health-care due to contractual obligations or software-as-service models. However, governments may become convinced as more and more evidence emerges promoting the benefits of mHealth projects in areas like maternal and child health. Investors should recog-nize the growth potential of mHealth for preventive healthcare, before re-jecting those ideas for their dependen-cy on government contracts.

fAcToR 4: consiDeR moBile DATA collecTion As THe meAns To An enD Until now, most mHealth projects have focused on data collection. Of course, data collection projects helped demonstrate the efficacy of mobile devices as instruments for data col-lection with fewer errors and lower cost. But, once the data is collected

Tools and FrameworksOpenMRS, Open Medical Record System http://Openmrs.org

eMOCHA, electronic Mobile Open-source Comprehensive Health Application http://main.ccghe.net/content/emocha

Datadyne EPISurveyor www.datadyne.org/episurveyor

FrontlineSMS www.frontlinesms.com

RapidSMS www.rapidsms.org

Useful WebsitesKnowledge for Health Program www.k4health.org/toolkits/mhealth

mHealth in Low-Resource Settings www.mhealthinfo.org

MobileActive.org www.mobileactive.org/areaofpractice/Health

mHealth Alliance www.mhealthalliance.org/news/publications

Health Unbound www.healthunbound.org/content/about

Mobile Health Information and Resources www.fic.nih.gov/RESEARCHTOPICS/Pages/MobileHealth.aspx

WHO Global Observatory for eHealth www.who.int/goe

Healthcare Information for All by 2015 www.hifa2015.org

19X R D S • w i n t e r 2 0 1 2 • V o l . 1 9 • n o . 2

fessionals often do not understand the language of computer scientists, who in turn may not have total appre-ciation of user behaviors as they occur “in-the-wild.” On top of this, many factors contribute to the resistance of mHealth tools in the health sector and technology is not necessarily the most important factor. Appreciating the potential concerns among users and mitigating them in the design stage is the real challenge faced by practitioners of ICT for development. The good news is that students, re-searchers, and practitioners inter-ested in mHealth can all play a role in enhancing healthcare for people in developing and underserved popula-tions around the globe.

References

[1] Blaschke, S., Bokenkamp, K., Cosmaciuc, R., Denby, M., Hailu, B., and Short, R. Using mobile phones to improve child nutrition surveillance in Malawi. UnICEF Malawi. UnICEF Innovations, Mobile Development Solutions, 2009.

[2] Childcount. http://www.childcount.org

[3] RapidSMS. Malawi – nutritional Surveillance. http://www.rapidsms.org/case-studies/malawi-nutritional-surviellence

[4] Center for Health Market Innovations. TRACnet. http://healthmarketinnovations.org/program/tracnet

[5] Entrepreneurial Programming and Research on Mobiles. http://www.media.mit.edu/ventures/EPRoM/research.html

[6] Xu, H., oh, l-B., and Teo, H.-H. Perceived Effectiveness of Text vs. Multimedia location-Based Advertising messaging. International Journal of Mobile Communications 7, 2 (2009), 154–177.

[7] Project Masiluleke. PopTech. http://www.poptech.org/project_m

[8] vodafone launches Mobile Health Website. Economic Times. 4 July 2012. http://articles.economictimes.indiatimes.com/2012-07-04/news/32537015_1_vodafone-india-vodafone-subscribers-website

[9] Mother and Child Tracking System. http://nrhm-mcts.nic.in/MCH/

[10] IFC, Gates Foundation works with India’s Bihar State to improve health payments system. http://www.ifc.org

[11] Government of Zambia Central Board of Health, 2003. The HMIS in Zambia: a trace on the implementation steps. http://www.cboh.gov.zm

Biography

Atanu Garai designs systems for improving rural employment service and public food distribution in the Indian state of odisha as part of the odisha Modernizing Economy, Government and Administration program. He advises leading organizations across the globe on the design and implementation of ICTs in the health sector. He has a graduate degree in electronic governance from École Polytechnique Fédérale de lausanne and is currently pursuing a master’s degree in computing. Garai is a student member of the ACM.

© 2012 ACM 1528-4972/12/12 $15.00

and stored, it does not—in and of it-self—act as a trigger for action among the intended stakeholders. There will be little change on the ground unless the data is translated into meaningful information and communicated to relevant stakeholders in a timely and acceptable manner.

Most randomized controlled trials demonstrate the change among pa-tients who receive personalized infor-mation through SMS. However, some mHealth projects have yet to show any meaningful change because projects have failed to convert data into useful information for users. In addition to data collection, researchers have ad-vocated the use of mHealth to evalu-ate performance of community health workers, make mobile payments, track supplies, and assist clinical care by nurses and doctors. To make this happen, the International Finance Corporation is currently working on a project in the Indian state of Bihar to use data reported to Mother and Child Tracking System [9]—a system to track delivery of scheduled services—as the basis for payment of compensation to rural health workers [10]. Considering that health workers sometimes wait for months to receive their compensation, they are likely to be interested in deliv-ering services on time with the prom-ise of regular payment.

fAcToR 5: pARTiAl AUTomATion DeTeRs mHeAlTH ADopTion in HeAlTH sysTem Donors ask for mHealth projects that work in a health issue of their choice. Examples include child vaccination and conditional cash transfer, among others. Partial automation is automat-ing a tiny component of the service de-livery—such as one polio vaccination out of five vaccines for children. Em-phasis on polio vaccination is one of several services that child health pro-grams deliver and automating data col-lection only for vaccination works as a lesser incentive for health workers. It is best to avoid such structural drawbacks during the program design phase.

Partial automation like this only places additional burdens on health workers by requiring they maintain records in manual and electronic for-mats. Such burdens can discourage

them from adopting mHealth. Zam-bia, for example, started deploying a Health Management Information System in 1997 and initially only two modules of the system were made available to providers. A later evalua-tion found that the unfinished com-ponents increased pressure among health staff [11].

fAcToR 6: mHeAlTH is inTegRATiVe mHealth technologies have the capac-ity to integrate otherwise disparate functions or processes within the tra-ditional health system. For example, traditional Web-based health manage-ment information systems, which col-lect routine service delivery data, can be used for monitoring and supervi-sion of health workers on the ground, for evaluating their performance, and for incentivizing use. This stands in stark contrast to typical performance evaluation, which is usually considered as a domain within human resource management. Additionally, traditional media has promoted target health be-haviors without any connection to pro-viders. mHealth, on the other hand, promotes healthy behaviors with ac-tive support from providers.

fAcToR 7: mHeAlTH is mUlTiDisciplinARy Since mHealth integrates various do-mains within the health sector, it is multidisciplinary by nature. As a re-sult, professionals from diverse disci-plines collaborate to develop products and services. In the case of maternal and child health, for example, medical doctors can help computer program-mers in defining algorithms for de-veloping decisions based on datasets on beneficiaries. Management profes-sionals can often shed light on the best way to process incentives for the health workers, while behavioral scientists can aid in designing messages for the providers and beneficiaries.

To conclUDeThere is a need to have further dis-cussion and debate on these factors to avoid common pitfalls in this new practice area. In particular, the mul-tidisciplinary nature of mHealth has been a great challenge for many practitioners—communication pro-