Delivering health care at a distance is a practical and moral imperative in a world where underserved populations are the rule rather than the exception. Telemedicine epitomises the potential of technology to reshape health care delivery, changing the way it is organised, improving access, and reducing the cost. This is even more so for breakthroughs in communication technologies, which, in contrast to prototypical cutting-edge technological developments, are already reaching a large proportion of the world’s population, including those in most need. It is easy to understand why researchers have been producing evidence in this field at a fast pace.
However, telemedicine remains in an early adoption stage, even in high-income countries. There has been progress in remote monitoring, and there have been some success stories, such as teleradiology. But for interactive telemedicine – real-time distant interaction between a health professional and a patient – implementation has stalled. The reasons behind slow adoption are a matter of debate, but the lack (real or perceived) of clear evidence about its benefits, costs, and acceptability plays an important role.[1,2]
Does interactive telemedicine work? For which conditions? Is it equivalent to face-to-face interaction with a health professional? Many attempts have been made to answer these questions. A search for 'telehealth' or 'telemedicine' in PDQ-Evidence, a health systems version of the Epistemonikos database, retrieves more than 170 systematic reviews, most of them dealing with interactive telemedicine for a wide variety of clinical conditions.
A newly updated Cochrane Review makes a great contribution to our knowledge about the effects of interactive telemedicine on professional practice and healthcare outcomes. If there were medals for backbreaking reviews this one would surely be on the podium. The review included 94 trials, with more than 20,000 patients, synthesised over 500 pages.
After expressing admiration for the effort of the review team, it is difficult not to be discouraged by the modest conclusion: that interactive telemedicine can lead to similar health outcomes as face-to-face delivery of care, but the cost to a health service and acceptability by patients and healthcare professionals is not clear. Moreover, in order to complete this massive review the authors had to put 63 studies under 'awaiting classification', and the last search was in 2013. A quick search for more recent relevant randomised trials retrieves more than 300 hits. Most Cochrane Review authors would agree the effort needed to summarise the evidence for this question exceeds the capacity of any team using the current approach to systematic reviews.
So, what should review authors do? The natural reaction when confronting a massive review like this one is to take pragmatic decisions, such as restricting to studies in specific conditions or choosing only some technologies. However, there are no compelling methodological reasons for splitting this question, and doing so would only fuel proliferation of systematic reviews of telemedicine and more disorganisation of the literature, as has happened with other technological interventions.
The many definitions of telemedicine highlight how it is an open and constantly evolving science, as it incorporates new advancements in technology and responds and adapts to the changing health needs and contexts of societies. So, a fundamental question we need to ask when looking for answers on this question is whether it possible to summarise the effects of an ever-changing intervention? Can studies from 1992 tell us something about contemporary telemedicine? And what about studies from 2008 (the median for included studies in the Cochrane Review)?
The task of synthesising the evidence on the benefits, costs, and acceptability of telemedicine reminds us of the task of Sisyphus, punished to roll an immense boulder up a hill, only to watch it roll back down when the task was almost completed. As soon as the review is published, it feels somewhat like sitting in a modern plane, reading about how zeppelins will transform the way we travel.
Changes in telecommunication technology also pose a challenge for trialists in this field. Testing the benefits, cost, and acceptability of a specific technology with the traditional approach will frequently produce information that is obsolete soon after it is published. Accurate information on cost and acceptability is particularly elusive given vertiginous changes in data capture and transfer cost. Without this information, it is hard to believe decision-makers will give the green flag for telemedicine.
Generating the evidence we need to make decisions in this and other rapidly evolving fields, and being able to synthesise it in a timely manner, requires a different approach. More efficient production of trials, including those in low- and middle-income settings, is starting to be discussed, and more efficient evidence synthesis articulated through technology and innovation is also gaining momentum.
Many of the tasks required to conduct a review can be automated or semi-automated. Other tasks can be distributed to a wider number of people, including lay people in models such as crowdsourcing. Even more important, data sharing between different players of the 'evidence ecosystem' can play a major role to make production of reviews more efficient. The last piece of the puzzle are platforms streamlining each of the steps to conduct a review and able to reuse data collected by other evidence producers, to integrate software for replacing or facilitating some tasks, and to allow the coordinated contribution of the crowd. The materialisation of all of these developments could finally lead to living systematic reviews, which are continuously updated as new evidence emerges.[9,10]
Systematic reviews are becoming ever more complex and challenging. The telemedicine review is a clear example. The authors rolled the boulder up the hill, but if we do not want to watch the boulder roll back, we need to change the way we are doing things. Innovation thrives in times of need. However, if we are not able to make it mainstream, our chances to keep providing reliable evidence for decision-making are scarce.