Mobile phone text-messaging to support treatment for people with type 2 diabetes in sub-Saharan Africa: a pragmatic individually randomised trial

Treatment for patients with type 2 diabetes could be substantially improved in sub-Saharan Africa. Amongst the problems identified, failure to take medicines to treat diabetes regularly, is a major problem. Resources to identify and support patients who are not making best use of medicine in low and middle-income settings are scarce. Mobile phones are widely available in these settings including among people with diabetes and linked technologies such as SMS-text messaging, have shown promise in delivering low cost interventions efficiently.. However, evidence that these interventions will work when carried out at a larger scale, and of the extent to which they will improve health outcomes when added to usual care is limited. Despite wide policy encouragement for using this type of technology to improve outcomes for non-communicable diseases such as diabetes, there are only a small number of studies evaluating this type of system. There are none, as far as we are aware evaluating systems sending SMS text-messages to individual patients with diabetes that have embedded their system within existing community based diabetes care. In addition, we are also not aware of studies that are measuring the costs and benefits of such systems.
We plan to carry out a study (a clinical trial) that will provide information about the overall benefits and harms of using carefully developed SMS text-messages to inform people about the benefits of their diabetes treatment and, motivate and prompt them to take it regularly. We also plan to collect detailed information about how the systems were set up and used in three different health care settings to guide future attempts to implement similar systems more widely in other settings and for other long-term conditions, and to carry out a detailed study of the costs of wider implementation and the potential value for money of such a system.
The knowledge gained from carrying out this study will have wide application. The technology being evaluated has been developed in-line with the direction of health, and information and communications technology policy in the countries where the research is based. We will use open source software that will intercommunicate with government systems to register and send individualised messages to patients. The same system also allows us to allocate patients to receive either informative messages, or non-health related messages. We will allocate patients in this way using a randomly generated assignment plan so that the benefits of the system can be tested in a carefully controlled trial. We will follow-up people for twelve months and measure important risk factors for the development of complications in diabetes including blood glucose control, blood pressure control and the levels of cholesterol so we can estimate potential health benefits. Before starting the trial, we will work with our clinical collaborators and patients with diabetes to ensure that the technology is embedded in clinical care, and that the message content is appropriate for each of the planned sites. We will extend this qualitative work during and after the main study to see how the technology was used and how people felt about it.
This study will provide data, not otherwise available, that will inform health policy and guide future implementation of this type of technology. We will work with a wide network that includes patients, clinicians, academics, industry and policy makers to help us identify opportunities for informing people about the work and raise awareness of what is being developed and studied.

Non-communicable long-term conditions and associated failure to use effective medicines place a major burden on the health of patients and health care systems. Current strategies to address problems of motivation and volition in taking medicines as intended are costly and of limited effectiveness. Mobile phones and other forms of digital health technology offer promise in delivering more effective support to patients around medicines adherence, but studies have not generally explored broader implementation within health care, comparative effectiveness and costs in low-resource settings. We have carried out a study using a novel system to send individualised text messages linked to available health care data for people with hypertension. We now plan to extend this work to people with diabetes, in different health-care settings and include formal qualitative methods and evaluation of costs.
We propose to (i) test the impact of educational and motivational messages, and prompts delivered through a system of SMS text-messaging on diabetes risk biomarkers, health status and satisfaction with care in a randomized trial; (ii) determine the factors associated with successful implementation of the system using a mixed-methods approach; (iii) study the costs of implementing the system and of its cost-effectiveness. The primary trial outcome is HbA1c with secondary measure including blood pressure and the ratio of total-cholesterol to HDL-cholesterol. The SMS text-message system has already been used with hypertensive patients in a low-income setting with 87% follow up of patients over one year. We are an international collaboration with expertise in diabetes research in sub-Sarahan Africa. We have statistical support from a UK registered primary care clinical trials unit, engineering support from a world-leading biomedical engineering based mHealth group, able to provide support for clinical measurement at the proposed research sites.

We anticipate that the health impacts from this work will come from setting up the infrastructure based around mobile phones, from the knowledge gained about effectively communicating through text messages, and from the lessons learned about the contextual factors that need to be addressed in wider implementation of technology.

In the first instance, setting up the technology will offer direct benefits to patients and health care staff. In our previous work at a single site, we have noted that patients see the text messages as a sign of more personal care, and thus improved satisfaction with the health care offered. Looking more closely at processes within the clinic, and the way that information is stored also has immediate application. Even at its most basic, the ability to store a small amount of data, linked to the identifier of a mobile phone number, will allow information about patients (for example the use of specific health facilities) to be stored and made accessible to health workers. The ability to track use of medications also has immediate application in identifying the possible impact when medications are unavailable, or identifying specific individuals who may be affected by "stock-outs". The wider integration of data would enable vulnerable groups of patients to be identified and better managed.

The specific knowledge that we might gain about effectively communicating through text messages, and how best to address contextual factors has potential to inform the content of text messages to support medication adherence widely. Many of the messages used in our previous work are likely to require little adaptation, and we will test whether this is an accurate perception in the different settings. Some messages are likely to be facility or process specific, and by observing the development of the new messages at the different sites, we will be able to learn how to facilitate the efficient processes of implementation and local ownership at different levels. The evidence this work will provide about the clinical effectiveness of the specific approach to delivering text messages and their content will ensure that the work has wide application and directly engages policy makers. For example, proven impact on health care outcomes, quality of life, satisfaction with health care, and careful recording of costs (health care and technology) will add to the case for wider use of the technology and may also encourage individual facilities to implement the technology.

We also aim to demonstrate the potential for open data and open source software to deliver greater integration of data to support delivery of healthcare. With our group's continuing involvement in developing and evaluating Internet linked mobile devices for long-term conditions in high as well as low-income setting, we are well placed to link to international work to establish information technology standards for health data.

In addition to impact for the early career investigators supported through this initiative, we intend to develop local capacity through the work. We will encourage appropriate staff to undertake educational courses alongside their work, and also encourage secondment of PhD and MSc students.

We will establish and be guided by an advisory board and also review recommendations for leveraging health innovations (e.g. the WHO ExpandNet programme) with the aim of ensuring that clear benefits are realised within three to five years of the end of a successful programme. The basic infrastructure (mobile phones) is already in place and we anticipate that within this time-frame, many health workers, will be using smart-phones for themselves with potential for leveraging our work through the wider availability of compatible technology.