What does the evidence tell us – handling potentially conflicting conclusions in diabetes trials

Mike Clark


There are a number of challenges in developing the evidence base for telecare and telehealth.

Research in this area has increased significantly over the past five years, and results are being published on a weekly basis in a wide range of specialist journals, not solely those associated with technology. Different terminology can sometimes be confusing, particularly to commissioners and policy-makers from non-specialist backgrounds. Some organisational features in a study may not be easily transferable for UK audiences. There is also a wide range of approaches in studies, and this can make it difficult to draw conclusions without understanding the inherent structural limitations.

As the WSDAN evidence database increases in size (currently around 230 articles), it becomes more important for people to be able to draw conclusions quickly from a range of available articles. Most people will have access only to an online abstract and not the full article. Some expertise is needed in understanding trial approaches and results, particularly when statistical significance is applied. It takes time (which we often do not have enough of) to work through articles and draw overall conclusions that are relevant to our own implementation plans.

For instance, what does the range of recent articles on telehealth and heart failure tell us about the possibility of reducing hospital admissions? Is telehealth likely to be effective in supporting people with type 2 diabetes? Does the evidence tell us that we are likely to get better results if we concentrate our resources lower down the long-term conditions triangle or should we work in areas where we have expertise such as COPD? What do you do when studies appear to draw conflicting conclusions, show a poor return on investment or do not confirm a hypothesis?

Many trials are conducted in the USA, which has a significantly different health care system; however, there are aspects of health provision and care involving Veterans Health, Medicare, Medicaid and some not-for-profit organisations that have a similarity to the NHS.

The large-scale implementation of telehealth in the real world can raise issues very different to those of a small scale pilot programme. Moving from a desktop analysis of research findings and learning from the experience of others into an ‘industrial strength’ implementation involving thousands of service users and patients is a big step even for innovators and early adopters. Understanding the available evidence is an important part of making that step change in services.

Sometimes there are examples where the outcome for users has improved slightly but costs remain the same. Because of the complexity, structure and cost of trials, there are relatively few current examples where outcomes have improved significantly whilst costs have reduced substantially. This may be affected by the current costs of equipment and follow-up service per patient. A trial that shows that a telehealth intervention is more expensive than the usual care could be quite alarming for commissioners, who may be searching primarily for cost reduction as a driver for telehealth implementation. However,  often means studying the detailed methodology and results of trials may help to identify key learning points.

The following recently published short case studies demonstrate some of these points for telehealth and diabetes.

Telemedicine impact on Medicare costs for diabetes in New York

In a randomised trial over the period 2000 to 2007, case management in the form of televisits were compared with usual care for type 2 diabetes. Although there were some small improvements in clinical outcomes, the costs for the equipment and services were considerably higher than other forms of service provision.

In this case, computers were used, which at the start of the trial may have been expensive. The technology is changing quickly, and units costs could be reduced over time,  making this intervention more cost effective. If service costs are lower compared with other forms of care then even a modest improvement in diabetes control could still make telemedicine a viable proposition to reduce the longer-term cost of treatment for and the impact on quality of life of complications such as blindness and amputations.

The impact of electronic messaging on quality of care and hospital visits for people with diabetes

In this 2004/5 analysis, researchers set out to see if an electronic messaging system would lead to higher quality of care and fewer hospital visits for people with diabetes. Patients using electronic messaging had better glycemic control but had more outpatient visits.

If we constantly focus on reductions in hospital visits and an economic return on investment, we may lose sight of the benefits for individuals who may feel more empowered to manage their conditions. Improved quality of life is much more difficult to quantify and cost but can be as important as ‘efficiency’ savings. Even if it is possible to reduce hospital visits, it is often difficult to realise these savings. 

The effect of telemedicine on outcome and quality of life in pregnant women with diabetes

In a study of 276 pregnant women, the use of telemedicine for glucose monitoring improved pregnancy outcome in women with gestational diabetes and improved quality of life in all diabetic pregnancies.

This study looks particularly at quality of life and suggests that there are some gains to be made in using a telehealth approach. It does not draw conclusions around cost effectiveness. Organisations implementing telehealth may decide to implement telehealth on the basis of quality and outcome improvements as long as costs are not significantly higher. However, the challenge is making large-scale organisational changes while maintaining an existing service.

Telemedicine influence on the follow-up of patients with type 2 diabetes

A one-year study compared results of people with type 2 diabetes using real-time transmission of blood glucose results  via a teleassistance system with an option for telephone consultations with those of a control group who were followed up at a health care centre. HbA1c reductions appeared in both groups but was not statistically significant. There was a significant decrease in blood pressure, total cholesterol, LDL-cholesterol, and body mass index in the intervention group. In the control group, the only significant reduction was in low-density lipoprotein cholesterol.

This study is looking at the feasibility of using telehealth to follow up type 2 diabetes. Changes in lifestyle (eg, exercise, healthy diet) can have a significant impact on maintaining control and reducing the later risks of complications. However, changes in behaviour can be difficult to maintain in the longer term without coaching, prompts and reminders. It may not be necessary for these contacts to be face to face. Short-term improvements can be seen in the intervention group, but what happens if the intervention is removed? Again, this study is less about cost and more about behaviour change and improved quality of life in the longer term.


Even though trials and studies may not provide conclusions about financial return on investment, there is much to be learned from reviewing the detailed research. The development of the searchable evidence database for telecare and telehealth together with plans to summarise key articles through our features and briefing papers should go a long way to providing robust summaries for commissioners and policy makers in organisations implementing telecare and telehealth. Register for regular updates

Mike Clark is co-project lead for WSDAN