Telehealth for long term conditions

Almost 50 years of innovation in telehealth have seen great progress in tackling a wide range of conditions using a variety of technologies and covering a wide range of outcomes. Although this work shows great promise, it also creates challenges for interpretation. The uncertainties in defining terms like “telehealth” reflect broader difficulties in interpreting the complex interplay of technology, service designs, clinical input, and patient involvement. New studies can challenge existing findings as much as they corroborate them. For example, two recent large scale trials of telehealth for heart failure found no benefit,1 2 whereas previous meta-analyses suggested reductions in mortality.3 4 5 To this we can now add the initial findings of one of the largest telehealth and telecare studies ever conducted: the UK Whole System Demonstrator trial (doi:10.1136/bmj.e3874).6

Telehealth does not just “work” or “not work.” Particular interventions may be successful, but this depends on many factors, including the specific contributions of the type of technology and of the context, such as the willingness and ability of clinical staff to change their care processes; the disease stage and severity of disease in the patients involved, their social backgrounds, and their needs and expectations; the predictive power of any monitoring data that are collected; and, indeed, the endpoints that are used to specify success.4 7 The research agenda established by systematic reviews of telehealth consistently argues for study designs that can generate insights into the active components within the black box of telehealth interventions.8 9 Although factors that might be important for successful telehealth can be described (box), we need more clarity on how to interpret the relative contributions of these elements. This concern is intimately bound up with uncertainties about the cost, quality, and safety of telehealth interventions, their effects on patient-clinician relationships, and their scalability and sustainability.

These uncertainties must be resolved no matter how desirable it might seem to roll out telehealth. Although frustrating for patients and industry advocates, evidence based policy making is particularly needed when health systems are facing unprecedented resource pressures. The UK Department of Health is therefore to be commended for, in 2008, funding the Whole System Demonstrator project to be one of the largest telehealth (and telecare) trials ever conducted. The early findings of the telehealth component will be complemented over the following months by economic analysis and by qualitative exploration of the implementation process and participant experiences.

The highlight of the initial findings is reduced mortality in patients offered telehealth—an absolute reduction of 3.7% (4.6% v 8.3%; odds ratio 0.54, 0.39 to 0.75), or about 60 lives over a 12 month period.6 This welcome finding needs a plausible explanation of how it was achieved (and can be replicated), especially because numbers of admissions were essentially unaltered in the intervention group, and existing evidence on the impact of telehealth on mortality is either mixed or lacking.1 2 7 10

The demonstrator trial combines three conditions—diabetes, chronic obstructive pulmonary disease (COPD), and heart failure—and was powered on a pooled analysis of effect. However, when there are questions about mechanisms of action and the relative contribution of subgroups to the ultimate findings, this strategy deserves further thought. For example, in England, the number of annual hospital admissions where diabetes is recorded as the primary diagnosis is about half that of admissions for COPD, despite diabetes being more than twice as prevalent. Telehealth related changes for patients with COPD might, therefore, have a greater effect on pooled estimates of hospital activity, particularly when the relative excess of patients with COPD in the demonstrator study group (compared with population prevalence) is taken into account. Furthermore, only a third of those invited to participate did so. Is telehealth particularly attractive for patients with COPD, and if so, why? There will also be questions, particularly if the pooled economic analysis is unfavourable, about whether targeting particular groups, perhaps by disease severity, might have been advantageous. These questions will not be answered by the current study.

Does the demonstrator trial provide convincing evidence for commissioning a national roll-out of telehealth? The findings reported to date suggest not, although we recommend caution until the full data are released. Does it provide justification for initiatives like “3millionlives”—the UK Department of Health’s plan to bring telehealth and telecare to three million people with long term conditions and complex care needs—not to proceed?11 Equally not: the evidence base is essentially unchanged and uncertainties remain. If it is politically expedient and makes sense in terms of how commissioners wish to innovate to meet patient needs then perhaps that is justification enough for the 3millionlives project. But we’re not sure. The difficulty of interpreting complex studies with nuanced findings, like the demonstrator trial, does not make decision making easy, but neither does it mean that the research is unnecessary.

Some unanswered questions will need new trials, but others can be dealt with by other routes. For example, the wealth of data generated by telemonitoring combined with data from electronic health records provides an important opportunity for large scale observational analyses, and that could include 3millionlives.12 Policy makers, commissioners, and guideline developers should help ensure that the research agenda focuses on areas where telehealth shows most promise. There is great potential but also still much to be done.