Commentary: Telemedicine trials—clinical pull or technology push?BMJ 1996; 313 doi: http://dx.doi.org/10.1136/bmj.313.7069.1380 (Published 30 November 1996) Cite this as: BMJ 1996;313:1380
- Jeremy C Wyatt, senior research fellowa
Telemedicine—remote consultation via an electronic link—is now feasible and increasingly easy to realise. But is it necessary for patients, doctors, or the health service? Randomised trials of telemedicine were first performed two decades ago,1 2 but some recent studies seem to have been driven by technology push rather than clinical pull.3 4 5 These studies give inadequate attention to three fundamental aspects of any trial: defining what is done, to whom it is done, and what is measured.6
Defining what is done
Doctors conducted remote consultations long before the advent of telemedicine by diagnosing a sick patient over the telephone or reviewing mailed histology slides. The unique feature of telemedicine is the fast two way electronic network that allows interactive communication such as video conferencing.5 To show that telemedicine confers benefit we must compare electronic means for transferring information with the most appropriate alternative.2 This means comparing teledermatology with a telephone discussion of mailed photographs or videoconferenced Doppler scanning with discussion about a video delivered by courier. Only then can we disentangle the unique effects of telemedicine from effects due to exchange of high resolution colour or moving images, which can easily be achieved without telemedicine.
To whom is it done?
Turning to the trial subjects, we should be suspicious of studies that rely on enthusiasts or volunteers, who may tolerate technical eccentricities that most doctors would not. Although evaluation by enthusiasts is a necessary first step, to estimate the general benefit of telemedicine we need randomly selected doctors and patients and statements of the overall recruitment and success rates.7
What is measured?
Telemedicine has many potential benefits and side effects, so trialists must make appropriate measurements. Consider a conventional specialist consultation. Apart from allowing more precise diagnosis because of access to hospital-only investigations or treatments, it also generates a detailed written report for the referring doctor. The meeting between specialist and patient informs the former about the patient's personality and the latter about the condition's aetiology, prognosis, and relevant treatments—leading to enhanced patient participation in decisions8 and a placebo effect, accompanied by greater compliance.9
Some of this will be reduced by telemedicine, so we need to measure patient satisfaction, compliance, and outcomes. We must ask participating doctors if they feel less able to consider patients' preferences and measure what proportion of their time is taken up with useful interaction as opposed to managing technology. Since the patient is party to all communication, this may reduce the usefulness of the consultation in terms of changed clinical actions, which need to be compared with those after a conventional consultation.
Making general practitioners reliant on expensive telemedicine may reduce the range of specialists to whom they can refer, leading to technology “lock in”; this needs to be checked. Telemedicine may also change the profile of referrals, leading to some patients receiving only the more economical teleconsultation; so the referral profile should be monitored. We also need to measure the ancillary resources required to install and maintain the equipment and train doctors in its use—resources readily available to enthusiasts but scarce elsewhere. Logistical problems in getting doctors and patient present at the same time may inhibit success and should be logged.10
On the positive side, telemedicine may enhance the exchange of clinical knowledge compared with conventional continuing medical education,11 so educational benefit should be measured. The turnaround time is clearly improved, but the quality of written records may be reduced; both should be measured. The two way flow of information during the consultation will probably lead to improved detection of clinical signs, which should be compared with the normal consultation process.12 One problem is that, if telemedicine leads to more accurate staging of disease or detection of complications, this may cause an apparent worsening in outcomes for telemedicine patients compared with less rigorously investigated controls.6 Equally, because doctors often fail to record a precise diagnosis, the obligation to record one in teleconsultations may lead to a spurious increase in diagnostic accuracy.13
Trials of telemedicine need to be conducted on representative cases and subjects to ensure that results can be generalised. The control intervention must be the best that can be achieved without telemedicine, as otherwise it is hard to credit any benefit to telemedicine itself. Since telemedicine is simply another kind of medical technology, the same principles of rigorous evaluation of costs and benefits apply.6 14 However, investigators may need to strive harder to maintain their clinical perspective and scientific rigour, since these trials are often driven by high technology and sponsored by those who provide it.