1. There is not enough evidence to support or refute the claim that mobile phones "work" as a healthcare intervention
With regard to Tables and , perhaps we should not be surprised that the effects of telephone interventions on various clinical and other outcomes are mixed. To conclude that such interventions probably work some
of the time is a trivial response. More significantly, and particularly with respect to improving medication adherence in important chronic non-communicable conditions that are increasingly prevalent in less developed countries (hypertension, diabetes, depression), any realistic intervention to improve adherence must be both dynamic and sustainable over time as patients' lives and circumstances will change. Adherence interventions must be temporally flexible and creative to track changes in the patients' relationship to the healthcare system. Indeed, such interventions as summarized in Tables and might in principle be effective most
of the time provided we can understand how to give the appropriate message in a way that becomes an integral part of the recipients' life. This is clearly true whether or not phones are used as the intervention. This long-term contextual view of adherence to medicines is particularly germane to the chronic conditions mentioned previously. A health-related message must be understood consistently over time and be culturally and socially appropriate to the indication and to the real-time needs of the patient. This is a daunting challenge for whatever medium is used. A recent review [61
] of the varied health-related uses of SMS applications suggests that it " deliver [s] both efficiency savings and improvements in the health of individuals and public health." However, many of these uses have not yet been subjected to clinical trials and none have been systematically extended on a large scale. The overall lack of well designed, randomized clinical trials with economic evaluation to confirm or refute clinical and economic benefits with mobile phone/healthcare interventions is an evidence gap that should be addressed in a systematic way.
The physical components of a telephone, i.e., the handset or headset and the network, are not isolated but are part of an entire system that includes pricing plans and other incentives which can provide leverage employed by healthcare professionals and policymakers. Notwithstanding any impact on health outcomes by the message itself, the effect of mobile phones, the particular payment plan and related components. i.e., the medium itself, on delivery of the "intervention" is not well understood either. Indeed, the medium that delivers an intervention may have a neutral, positive, or even negative impact on the health intervention it is delivering. This aspect of the debate about use of telecommunications as a healthcare intervention has hardly been addressed at all, in any environment.
2. A developed world model of mobile phones may not be appropriate in developing countries
Inter-country comparisons of aggregate statistics for 73 countries derived from the International Telecommunications Union [62
] are shown in Figure , below and in additional File 1
: Spreadsheet.xls of summary statistics of GDP per capita and mobile subscriptions per capita for various countries.
The Relationship of GDP/capita (US$-2003) and Mobile phone subscriptions/capita (2003) for Various Countries. Data obtained directly from reference  as reproduced in additional File 1.xls.
In Figure , the relationship between GDP/capita and mobile phone subscriptions per capita suggests that small changes in "wealth" will result in large changes in mobile phone penetration in poorer countries at GDP/capita less than about $3–4,000. Whether or not this inference really holds for resource-poor countries that lie at the lower end of this graph is an open question. The non-linear nature of Figure also suggests that income has less of an effect on mobile phone penetration per capita in the more affluent countries. It is worth noting that the nature of Figure is similar to the relationship between "wealth" and health indicators such as life expectancy. The ramifications of this latter relationship are still subject to continuing debate. It is possible that the health of individuals in a society also depends on the degree of income inequality in that society and that the effect of distribution of income on health, and possibly on many other things including mobile phone penetration, is more important than absolute income. Aggregate-level analyses of "developed" and "developing" countries will not illuminate issues about determinants of individual health, or mobile-phone use as related to health. The question as to whether computer/web/phone communications technology can solve development/health problems should be shifted from a discussion about 'developing vs. developed" countries to whether use of telecommunications, and mobile telephones in particular, in healthcare is appropriate to the specific national and local context.
In Africa, mobile penetration rates are low by developed country standards but use of pre-paid calling cards and the informal sharing of mobile phones between people all increase accessibility, even in rural communities. The impact of mobile extends well beyond what might be suggested by measuring the aggregate number of subscriptions. Shared use in some locations could be an important constraint if mobile phones are to be used to convey health information since two-way communication in a shared system is difficult as a non-owning user can make outgoing calls but cannot receive spontaneous calls [4
]. SMS text messages, if not deleted, can be observed by subsequent users. These informal arrangements that extend the reach of telecommunications beyond the individual user seem very powerful. Policy debates on information technology policy generally and health policy in particular are not sufficiently informed by evidence of this type [5
3. Creating a sustainable, large-scale mobile phone/healthcare model requires agreement among different stakeholders with different agendas
The work summarized in Tables and are almost invariably small, academic pilot or feasibility studies. A major unresolved issue when approached from the point of view of "who is doing the intervention" relates to whether these studies can be scaled-up in the community and whether they can have an impact on individual and, ultimately, on public health. Table summarizes the different perspectives of some of the major stakeholders who might be expected to use mobile phone technology in a large-scale health intervention.
Stakeholders' Positions regarding Mobile Phones as a Healthcare Intervention
Patients are looking at an intervention using telecommunications broadly, and mobile telephones in particular, to eliminate or at least ameliorate suffering and reduce their financial burden during the illness and healing process. With respect to aspects of healthcare counselling, some patients may prefer face-to-face contact rather than phone or text message contact. For some persons, communication of almost any type using SMS messages will lack nuance and individual "tailoring" so that synchronous, real-time voice communication between patient and healthcare provider will be preferred. Real-time communication can clearly be realized using mobile phone technology. A consideration with respect to asynchronous communication, i.e., with a time lag between sending and receiving, is that such communication may have to be secured or otherwise encrypted, especially with shared and/or stolen mobile phones.
From the viewpoint of a patient with TB or HIV or epilepsy, the ease of use of mobile devices could be a potential problem since, unless encrypted in some way, an e-mail/text message opened because of a theft or viewed inadvertently will increase the risk of being stigmatized. It is not clear if this issue is important in actual practice. "Privacy" can be seen as an aspect of security – one in which trade-offs between the interests of one group and another can become particularly clear [63
]. Security services (e.g. that based on digital signatures) probably do not come without transaction costs to the end-user as well as society since supportive law would need to be implemented in many countries. Nonetheless, in mobile infrastructure in developing countries, privacy/security and authentication services can be based on certificates and secret keys implemented in SIM (Subscriber Identity Module) cards. Here the patients and healthcare professionals may sign and prove digitally, and if needed, encrypt all their communications. This is a subject well beyond the scope of this paper but see, for example [64
Healthcare providers are also looking for treatment that will eliminate or at least ameliorate suffering and improve communication of health-related issues between themselves and patients. Providers in managed care settings utilizing telecommunication/mobile structure as an intervention nonetheless might share the same concern, albeit based in easing their own financial burden and improving their bottom line. From this viewpoint, voice counselling may be time and money- intensive so providers may actually prefer automated interactions. Although a provider's first priority might be to proactively transmit information via mobile phone to the patient (i.e., "We notice that your blood sugar has gotten low... do this..."), the ability of this to make a clinical difference will be a function of whether the patient can understand the information and act upon it. This is therefore a function of the mobile phone context, i.e., its intrusiveness, timing, quality, clarity.
It is worth noting that with respect to using mobile phones to monitor diagnostic indices, any chemical, biological or physical marker must be easily determined and easily sent via the mobile phone. Blood glucose, spirometry, adherence (e.g., number of cigarettes/pills), blood pressure, weight, physical activity, mental state, side effects can all be transferred with relative ease. For HIV there is no simple diagnostic useful in this context as a patient cannot now simply phone in their CD4 or viral load count. Weight loss and known side effects are more likely markers for "wireless" monitoring of HIV status. The great potential advantage of mobile phone technology in managing chronic conditions is that it can collect small amounts of data rapidly, efficiently and with minimum intrusion. A healthcare intervention that requires communication of relatively simple information (e.g. weight or a spirometry result or a blood glucose value) may be preferable to content that demands more sophisticated modalities like video. Even with the relatively simple interventions under review here, the mobile phone company must be aware of possibly unique legal issues relating to security, privacy authentication, theft of identity, liability for harm due to unauthorized/negligent transmission of health information and the like [64
From a business point of view, mobile telephone companies make their profit in the private sector. They are only likely to invest in such technology in the public research sector for reasons of – for want of a better term- "corporate responsibility". Clearly, however, the more realistic priority in scaling-up mobile phone infrastructure to support a phone-based healthcare intervention will be to keep their existing clients and attract new ones. Monitoring the cost of the content (the message) as opposed to mere connectivity (the medium) is important. An additional consideration is their attempt to manage their way through a changing regulatory environment, especially with state-owned telecom networks [52
]. Creating a sustainable business model among the stakeholders, as well as insurers and pharmacists will be needed and is a challenge A supportive legal, governmental and business infrastructure for such a model is no less a challenge in a developed country.
New modalities such as broadband access technologies (e.g. WiMAX, Flash-OFDM, VoIP and so on) are being created all the time. Within these infrastructures, not only data (e.g. web, e-mail), but also voice over internet (VoIP) services will be widely possible in many places. With these new wireless access technologies, transmission speeds of 500–1000 kilobit/s, even higher, are possible. When framed in the present context, the question of whether or not these are suitable modalities for improving health outcomes, must be informed by the particular social and behavioral health context at several levels, i.e., country-level down to patient-level.
The larger debate about communications technology as a barrier or spur to development may not be resolved for some time. The communications and services infrastructure to support large-scale use of telecommunications as a health intervention exist in some parts of Africa and in much of Asia. At present, one would hope that healthcare applications such as accessing medical self care, receiving medication adherence reminders (e.g., all the applications used in developed countries), facilitating case management of chronic conditions (e.g., diabetes, TB) are more suitable for the majority of the poor in developing countries [55
], than receiving mortgage information or buying concert tickets.
Notwithstanding the fact that large-scale supportive infrastructure exists, a top priority goal for all governments should be to (re)-align the regulatory and pricing policy of the telecommunications sector with health policy goals. Use of various information technologies (including mobile telephones) to less developed countries and communities has been ongoing for some time, mostly via the many specific initiatives, led by communities, development, donor and business organizations. Evidence on the effectiveness of these initiatives with particular regard to their use as healthcare interventions is mostly in the form of anecdotal material. More rigorous evidence is needed for drawing conclusions.
○ The developed world model of personal ownership of a phone may not be appropriate, and may even be irrelevant, to the developing world where telephones are often shared.
○ Convincing evidence regarding the cost-effectiveness of mobile phones as a " telemedicine" intervention is limited and good-quality studies are rare in less developed countries.
○ Evidence of the cost effectiveness of fixed or mobile telephones as such an intervention to improve adherence to medicines was difficult to identify. Given the rapid expansion of chronic disease management (TB, HIV, non-communicable chronic conditions) in less developed countries, the ability of mobile telephone interventions to improve long-term adherence to medicines in chronic disease is unknown but could be of major benefit. Such interventions must be part of a repertoire of interventions to be used in a changing way over the lifetime of a patient. One advantage of telephones to manage chronic disease is its ability to create a two-way interaction between patient and provider(s) and thus facilitate the dynamic nature of the relationship and accompanying interventions.
○ A framework for debate among telecommunications, development and public heath experts about the use and value of mobile phones as health intervention in developing countries will have to account for the different primary perspectives of the relevant stakeholders, the value-added of each stakeholder in a sustainable business model, as well as the context-specific nature of information technology systems in general. For a mobile telephone system to be successful, whether or not as a healthcare intervention, it has been shown that the local context is understood.
○ Regulatory reforms required for proper operation of basic and value-added telecommunications services are a priority if mobile telecommunications are to be used for healthcare initiatives.