Telemedicine: barriers and opportunities in the 21st century

Authors


Benedict Stanberry, LLB (Hons), LLM (Wales), MRIN, Director, Centre for Law Ethics and Risk in Telemedicine, Secretariat, PO Box 4015, Cardiff CF14 3YF, UK (fax: + 44 (0)29 20694744; e-mail: ben.stanberry@clert.com).

Abstract

Abstract. Stanberry B (Centre for Law Ethics and Risk in Telemedicine, Cardiff, Wales, UK). Telemedicine: barriers and opportunities in the 21st century (Internal Medicine in the 21st Century). J Intern Med 2000; 247: 615–628.

This paper aims to examine how health telematics will develop in the first 10 years of the new millennium and, in particular, to assess what operational, ethical and legal barriers may lie in the way of this development. A description of the key principles and concepts involved in telemedicine and a short historical overview of telemedicine’s evolution over the past century are followed by consideration of why empirical research into ‘info-ethics’ and other deontological and legal issues relating to telemedicine is being necessarily catalysed by, amongst others, the European Commission.

Four evolving health telematics applications are examined in some detail: electronic health records; the transmission of visual media in disciplines such as teleradiology, teledermatology, telepathology and teleophthalmology; telesurgery and robotics and the use of call centres and decision-support software. These are discussed in the light of their moral, ethical and cultural implications for clinicians, patients and society at large.

The author argues that telemedicine presents unique opportunities for both patients and clinicians where it is implemented in direct response to clear clinical needs, but warns against excessive reliance upon technology to the detriment of traditional clinician–patient relationships and against complacency regarding the risks and responsibilities – many of which are as yet unknown – that distant medical intervention, consultation and diagnosis carry.

Introduction

Time will say nothing but I told you so,

Time only knows the price we have to pay;

If I could tell you I would let you know.

(W H Auden, Collected Shorter Poems, 1966)

Crystal ball gazing can be an engaging exercise, particularly when one’s gaze is directed at a subject as vital and exciting as the potential future uses of telecommunications and information technologies in medicine and health care. Who would have predicted half a century ago, as Europe struggled to rebuild itself after World War II, that within 50 years we would all be able, if we wished, to carry mobile telephones small enough to fit in our pockets or purses, let alone that we would be able to use those telephones to transmit images or text and to speak to almost anyone, anywhere on the planet? And even if we could have accurately predicted the huge advances that information and communication technologies would make in the second half of the 20th century, it would probably have taken a genius of the calibre of Einstein to identify the tremendous restructuring of the way health care is delivered that these developing technologies would catalyse. Yet even Einstein was not so foolish as to suggest that prophesising about the future was easy or even profitable. ‘Wood chopping is so popular,’ he is quoted as saying ‘because you can see the results immediately’.

Hence it is somewhat daunting to have been asked by the editors of the Journal of Internal Medicine to contribute my own personal vision of the future of one particular field: telemedicine. The former Austrian Chancellor Franz Vranitsky, although he was referring to the future of Europe rather than the future of health care, probably got it right when he said, ‘Anyone with visions needs to see a doctor.’ So I shall begin by inserting the important caveat that I am certainly no prophet. Nor do I aspire to that dubious occupation of ‘guru’. Medicine – and telemedicine in particular – has quite enough of those already. The aim of this paper is simply to consider, in the light of the developments of the last decade, how health telematics will develop in the first decade of the new millennium and, more specifically, to highlight some of the operational, legal and ethical barriers that may lie in the way of this development. The most important lesson that the past can teach us is not to try too hard to predict the future. But I do very strongly believe, nonetheless, that looking ahead is something we should all make time to do at the start of this new millennium.

The evolution of telemedicine

There have been many attempts at producing an all-encompassing definition of telemedicine. The European Commission DG XIII (which deals, amongst other things, with health telematics) has defined telemedicine as ‘rapid access to shared and remote medical expertise by means of telecommunications and information technologies, no matter where the patient or the relevant information is located’[1] and, more recently, as simply ‘the use of telecommunications for medical diagnosis and patient care’[2]. Even this definition does not do justice to the broad range of possibilities offered by telemedicine for diagnosis, treatment, health education and research. Indeed, the very word ‘telemedicine’ is itself often felt to be too limiting, and definitions such as ‘telecare’ and ‘telehealth’ are also used to refer to the availability of nursing and community support and other public health services.

Ried [3] defines telemedicine as including the use of telecommunication technology to exchange health information which provides access to health care across time, social and cultural barriers, whilst Coiera [4] has drafted his definition with slightly more particularity. He believes that:

The essence of telemedicine is the exchange of information at a distance, whether that information is voice, an image, elements of a medical record, or commands to a surgical robot. It seems reasonable to think of telemedicine as the remote communication of information to facilitate clinical care.

But however, you wish to label the practice of medicine and health care from a distance, it is by no means an invention only of the late 20th century. The vast array of health telematics applications available today have evolved from, for instance, Einthoven’s use of an ordinary analogue telephone network to transmit electrocardiograms (ECGs) and electroencephalograms (EEGs) in 1910, and the foundation in 1920 of a medical advice service for seafarers based upon Morse code and voice radio [5]. Such technologies, although very basic by modern standards, were cutting-edge for their time and it is this common historical denominator – of finding medical and health care applications for the most state-of-the-art, innovative technologies available – that pervades the evolution of telemedicine and telecare over the last 100 years. For instance, the early use of two-way closed-circuit television systems in the 1960s to facilitate both the transmission of medical images, such as radiographs, and consultations between health care practitioners and patients has now given way to low-cost, PC-based solutions for video-conferencing and image transmission [6]. Mobile telemedicine solutions no longer depend upon fixed-line telephone systems or radio transmission, but can utilize satellite and GSM mobile phone technology to bring medical care to remote locations, including underserved rural areas, sites of major emergencies and developing countries with a shortage of medical skills.

Other applications which come under the broad headings of telemedicine, telecare and telehealth include the transmission of highly visual media, such as pathology slides (‘telepathology’) [7–9], images of skin lesions, rashes, tumours and the like (‘teledermatology’) [10–12] and radiographic images (‘teleradiology’) [13–15]. Such transmission may take place in ‘real time’ or via a ‘store-and-forward’ system, and the images may subsequently become part of a patient’s electronic health record. The emerging discipline of ‘telesurgery’ encompasses both the mentoring and training of surgeons in the operating theatre via video-conferencing systems [16] and the performance of surgical procedures using robotic surgeons guided by real surgeons at a remote site using computer data of diagnostic studies and scans [17]. ‘E-health’ applications, which include PC-based video-conferencing, the Internet and e-mail, are becoming increasingly ubiquitous as a means of communication between patients and health professionals, as well as between health professionals themselves, for the purposes of consultation, education and second opinion [18]. The transmission of electrocardiographic and other physiological data is now commonplace [19], particularly to facilitate the monitoring of patients in their own homes [20] and the gathering of clinical information from patients in inaccessible sites, such as ships, aircraft and geographically remote regions [2122]. Nurse-led call centres providing telephone triage and other care services, through the use of both the nurse’s own expertise and PC-based decision-support software, are being widely implemented as a means of providing direct benefits to patients and maximizing the satisfaction of health care ‘customers’ who subscribe to private medical insurance or managed care schemes [23].

Balancing the benefits and the burdens of telemedicine

Despite this amazing diversity of technologies and uses, practically all telemedicine applications are victims of the same cliché. The well rehearsed and much hackneyed assertion that telemedicine will bring both ‘benefits and burdens’ or ‘risks and rewards’ would be a trite observation were it not for the fact that it is so true and yet so ignored in certain circles. Wootton [24] has stated:

Telemedicine is not new. It has been practised using electronic communication for decades, and using traditional forms of communication for much longer. Despite this historical background, telemedicine appears to have a polarising effect on the health care profession. People are rarely neutral about it; they are either enthusiastic proponents or vehement opponents. The proponents believe that telemedicine represents the future. It will lead to higher standards of medical care as well as reduced costs. The opponents believe that it represents a threat to the traditional doctor–patient relationship and is an intrinsically unsafe way to practise medicine. The potential legal and ethical problems associated with telemedicine are often waved as a ‘shroud’ to support the view that the possible complications of telemedicine mean that it could not be used to form the basis of a clinical service.

At the heart of these opponents’ reservations regarding telemedicine is their perception that the guidelines, standards and regulations that are needed to ensure telemedicine is practised legally and ethically in a number of different scenarios – whether they be feasibility studies or, more worryingly, fully operational services that have been integrated into mainstream practice by a health care institution – are presently either inadequate or non-existent. Is this simply an exaggeration or have the legal and human rights of the patient – who is, after all, intended to be the principal and fundamental beneficiary of the health telematics revolution – genuinely been subordinated to technological progress or improved cost-effectiveness?

This is a difficult question to answer, although significantly the European Commission has chosen to make ‘info-ethics’ one of the central themes of its Fifth Framework Research Programme in Information Society Technologies for Health, which has an overall budget of some 40 million euros. The head of that programme, Professor Jean Claude Healy, has stated, in a recent interview [25]:

It [info-ethics] is an area that I believe causes concern to many, particularly the general public…. The scope of info-ethics ranges across the private/individual and the public/institutional aspects of ethical problems, on a national, international and global basis. Concepts particularly relevant to health care include privacy, confidentiality, principle of legitimate purpose, consent, security, transparency, participation and education.

This emphasis on attracting proposals for empirical and practical research into legal and deontological aspects of telemedicine is, it is submitted, confirmation – as if any were needed – of a moral and ethical renaissance of sorts. The intrinsic legal and human rights of the patient-citizen, having spent the greater part of the last decade relegated to the sidelines by the dazzling speed of technological development, have at last been brought back under the spotlight. The great irony is, as we shall see, that these issues which were for so long ignored in health telematics research now represent some of the most significant barriers to its large-scale implementation.

The info-ethics renaissance

The risks to patient privacy and confidentiality arising from the use of telemedicine systems are easy enough to identify and the emphasis in most disseminated research on the legal and ethical aspects of telemedicine has largely been upon these issues. But this has been to the detriment of other important aspects which fall under the broad heading of ‘info-ethics’ and upon which there is very little published information available. These aspects include the nature and extent of the doctor–patient relationship in teleconsultations, the delineation of responsibility between teleconsultants, the standards of practice that those consultants should conform to, and the clinical risks associated with telemedical practice [26–29]. Intrinsically linked with the issue of confidentiality, moreover, is that of consent and the need for the telepatient to give fully informed consent, not just to the physical intrusion that may take place during examination or treatment, but also to the uses that may be made of any electronic medical record or video recording of a teleconsultation and who will have access to it.

Yet although we have now evolved methods of working whereby video-conferencing and the creation of (or addition to) an electronic health record are rapidly becoming the norm rather than the exception to the norm, we certainly have not given much thought to the appropriate protocols or ethical guidance with which to regulate telemedical practice – until now that is. For if the last few years of the 20th century were marked by the slow transformation of telemedicine from pilot projects to mainstream services, then the first few years of the new millennium will be remembered for the ‘catching up’ that the supervision of medical practice by the professional associations, colleges and medical boards had to do in order for that supervision to properly reflect the growing ubiquity of information and communication technologies in medicine and health care. And this process has, to a certain extent, already begun.

Clinicians, health care managers, insurers and other professionals working with telemedicine all now recognize that the rights and expectations of patients have for many years gone largely unaddressed, such has been the fervour with which the technological and clinical aspects of telemedicine have been developed, assessed and embraced. The work of Leif Erik Nohr [3031] reviewing, amongst other things, the passing of a Patients Rights Act by the Norwegian Parliament and its impact on measures such as the seeking of second opinions, free choice of hospital, access to medical records and consent to their use and dissemination, provides a very notable exception.

At a recent workshop hosted by the International Space University in Strasbourg, supported by Inmarsat and the European Commission and attended by over 60 high-level representatives, a special policy and legal issues review group produced a series of recommendations for action on both national and international levels. The workshop report to the European Commission stated, amongst other things, that [32]:

Of particular importance is the promotion at a European level of national action to protect the rights of the European citizen. This could include the creation of voluntary codes of practice regarding ethical issues and standards…. It is legally impossible to devolve responsibility for the supervision of medical practice and, prima facie, the issuing of licenses to practice, to one supranational medical authority for Europe that sets universal standards. However, it would be highly beneficial for both teleconsultants and telepatients if the adoption of de facto harmonized standards of telemedical practice – in the form of codes of practice to which teleconsultants may voluntarily subscribe – could be evolved from dialogue between the national professional bodies, colleges and associations, as well as other interested parties such as the defence unions and liability insurers. Such codes of practice should have, at their core, the protection of the rights of the patient to be able, for instance, to give fully informed consent to their participation in a teleconsultation, since at present consent tends to focus on physical intrusion rather than the control that the patient should naturally have over the teleconsultation, who participates in it and who bears ultimate responsibility for its conduct.

The control and ownership of electronic health records

Of course, obtaining such informed consent to the patient’s participation in a teleconsultation, including explaining who will be at ‘the other end’, for instance, of a live video-conference and what use might be made of any recording of the teleconsultation (e.g. for clinical audit or research purposes), has significant resource implications. Ideally, every telepatient would be counselled and would give informed consent prior to the creation of an electronic health record. But if one member of staff spent about 10 min explaining this information and gaining the explicit consent of each patient entering a large hospital with, say, 1000 beds, it would potentially have to recruit more than a dozen extra staff recurrently.

Yet this is exactly the course of action which seems to be inevitable within the UK following the decision of an English court in the recent case of R. vs. Department of Health, ex parte Source Informatics Ltd[33] and which might therefore be followed soon in other countries. In that case, Source Informatics, a subsidiary of a US company, had challenged the UK Department of Health guidelines entitled The Protection and Use of Patient Information[34], which state that the disclosure of details from prescriptions would constitute a breach of confidentiality which could lay doctors and pharmacists open to legal action. The company had been trying to persuade general practitioners and pharmacists to allow them to collect data as to the prescribing habits of general practitioners. They believed that this would be of commercial value to drug companies and would provide useful data for those interested in monitoring prescribing patterns. The proposal was that, with the consent of general practitioners, the pharmacists would, for a fee, and using software provided by the company, download onto a disk the name of the general practitioner and the identity and quantity of the drugs prescribed, but no information which could identify the patient.

But the judge in this case, Mr Justice Latham, held that the release for commercial research purposes by pharmacists or general practitioners of even anonymous prescription or dispensing information without the consent of patients would be an unlawful breach of confidence. The rationale for this judgement was, in the eyes of the court, that those requiring medical assistance should not be inhibited in any way from seeking or obtaining it. There might be some patients who would feel very strongly that the pharmacist should not give any information obtained from the prescription without their consent.

Interestingly, data protection and privacy issues arising under the European Data Protection Directive (95/46/EC) were not considered in any depth by the court in this case, most probably because the data disclosed, having been anonymized, could not be linked to specific individuals and so could not therefore constitute ‘personal data’. This would be necessary for the planned use of the prescribing information to come under the terms of the Directive. But if this decision, the very first of its kind, is allowed to stand, it could instantly wipe out many of the benefits of the use of electronic health records, since it throws into question the legality of a whole range of uses of anonymized patient data taken from medical and health records, including the use of that data to manage a patient’s disease, for clinical audit and training purposes and for epidemiological and public health research. The judge in this case did not address the use of anonymized patient records for anything other than commercial purposes, so it is possible that other routine uses of patient information, including for statistical and research purposes, are still permissible. In any event, if disclosure were in the public interest or if the patients themselves gave express or implied consent to the use of their records for research, the court believed that the law would not have been broken.

The Source Informatics case clearly shows that the courts are ready to intervene, not only to protect patients from the new threats to their fundamental rights to confidentiality and autonomy that the information age is bringing, but also to assert the absolute primacy of patients’ rights to make a fully informed decision about how information about them is used. So for the time being at least, the sensible course of action for health researchers wishing to ‘mine’ electronic health records is: if in doubt, get express consent.

Responsibility, risks and standards in telemedicine practice

This new desire to re-assert the rights of the patient in the field of health telematics where we may, up until now, have been so impressed with what we could technically do that we had stopped thinking about how we should ethically be doing it, is now a central theme for European research programmes over the coming years. But why stop with info-ethics when there are other equally pressing yet little explored issues surrounding the professional accountability of clinicians using telemedicine, the standards of practice to which they should conform and the new clinical risks to which patients may now be exposed? All these issues will confront the users and providers of health telematics and informatics as these technologies achieve ever greater ubiquity in European health care. Hence in the second half of this paper I have chosen to single out three of the most rapidly evolving health telematics applications for closer examination: the transmission of visual media (in specialities such as radiology, dermatology, pathology and ophthalmology); the use of telerobotics in surgery; and the growing prevalence of telephone call centres and decision-support software.

Teleradiology, PACS and other visual media

The most widely embraced form of telemedicine today is without doubt teleradiology. There have been ≈ 250 000 annual episodes in which teleradiology has been used in the USA alone [35]. A picture archiving and communication system (PACS) that uses digital data held in a single or distributed database that is accessible through a network is rapidly becoming the new paradigm for the sharing of clinical images, since it solves many of the traditional film management problems associated with the releasing of X-rays, reporting on them and then getting them back to the radiology department again.

The term ‘teleradiology’ tends to be used to refer to the process of remotely displaying radiological studies for interpretation or consultation, or both. Similar terms have been coined for the use of telemedicine systems for other highly visual mediums, principally telepathology, but also for fields such as teledermatology and teleophthalmology. Based upon standards developed jointly in the USA by the National Electronic Manufacturers Association (NEMA) and the American College of Radiology (ACR), the Digital Imaging and Communication in Medicine (DICOM) 3.0 standard was published in 1993 as a generic standard for medical informatics. This highly detailed standard defines, amongst other things, minimum requirements for network environments, communication protocols, service classes and information objects, and has enabled an open PACS standard to develop under which teleradiology has evolved from a point-to-point application using equipment and protocols that were exclusive to a particular manufacturer’s system, to a universal network closely akin to the Internet [36–38].

The American College of Radiology, the Royal Australian and New Zealand College of Radiologists and, more recently, the Royal College of Radiologists in the UK have all published comprehensive standards for teleradiology [39–41]. All three standards require the radiologists who use teleradiology and PACS to assume responsibility for ensuring the introduction of and adherence to appropriate quality standards at both the transmitting and receiving sites. These standards must encompass not just technical aspects relating to the equipment and systems used in the teleradiology process, but also professional aspects relating to the training and conduct of the technical and clinical radiology staff themselves. It is the interpreting radiologist who assumes ultimate personal responsibility for the conduct and quality of the examination and who must also recognize the potential limitations of teleradiology and the boundaries of its diagnostic accuracy.

This last requirement sounds self-explanatory, of course, and it is certainly true that a great deal of the published scientific literature on teleradiology, teledermatology, telepathology and other visual telemedicine media has paid detailed attention to the diagnostic accuracy that is achievable with various types of imaging equipment and transmission systems. But the existence of this body of scientific study does not, of itself, necessarily equate to the existence of a legally acceptable standard of care unless it can withstand logical analysis by the courts themselves. Hence our pioneering teleconsultants in fields other than radiology, where the relevant college or professional association has yet to issue official guidance on what is and is not good practice, may be drawing dangerously false comfort from the journal literature.

When is a standard not a standard? For some 40 years or so, clinicians have been judged against the practices accepted by their peers. The English case of Bolam vs. Friern Hospital Management Committee[42] set the now almost universal defence that ‘a doctor is not negligent if he is acting in accordance with such a practice merely because there is a body of opinion which takes a contrary view’. Over time, as more and more medical negligence cases have reached the courts, the views of ever smaller minorities of doctors have qualified as ‘acceptable practice’ for the purposes of the Bolam test. Faced with accusations of negligence, in other words, it is not the size of the body of medical opinion that a doctor followed that matters, but merely the existence of that body of opinion in the first place.

But the highest court in the UK has now taken a new approach to the problem of deciding when to accept that merely following a body of medical opinion, no matter how small, should be a full defence to accusations of negligence. In a case called Bolitho vs. City and Hackney[43] it was held that a court of law had to be satisfied that the exponents of the body of opinion relied upon could demonstrate that their opinions had a logical basis. So it is now arguable that, in the absence of specific guidance for teleconsultants from their college or professional association, scientific studies which conclude, for instance, that a consistent two-thirds correlation between clinical conditions diagnosed via a video link and those diagnosed in a face-to-face consultation is an acceptable level of diagnostic accuracy for a teledermatology service are not sufficient to provide legal justification for the use of such a service [44].

Likewise, where the transmission of fetal ultrasound scans at 384 kbit s–1 is shown in a scientific study to produce no perceived differences in technical quality and only marginally worse diagnostic accuracy than a scan transmitted at 1920 kbit s–1[4546], the use of the lower bandwidth must still stand up to logical analysis if it is to be legally defensible. Amongst other factors, courts will expect the number of clinical images and patients actually involved in a scientific study to be large enough for its results to be conclusive. Additionally, the courts may well refuse to accept that the ostensibly ‘high’ levels of diagnostic accuracy which appeared acceptable, even exciting, at the time of the original research were a proper basis for the use of such a system in providing a full clinical service.

Who screwed up – me or the machine? Telemedicine is an area in which it is notoriously difficult to conduct scientific assessments [47], and success with telemedicine depends as much upon the skills of the clinicians involved as it does upon the reliability and suitability of the technical components of the telemedicine system. One of the principle dangers of telemedicine is that there may be times when it is difficult or even impossible to establish if it was a misdiagnosis by a clinician or a technical failing of the system itself that was the operative cause of harm to a telepatient, as the real case below illustrates.

Mrs Cruisey was 27 years old. Her first child, Alexander, was born at St Helier Hospital in the UK in 1989. When, at the start of her pregnancy in June 1988, she attended that hospital for an appointment with the midwife, her family history was noted, including the fact that, when she was a little girl, her mother had given birth to a baby suffering from anencephaly – a condition in which a child is born with a defect of the skull and absence of a brain.

Mrs Cruisey became pregnant again in the spring of 1991 and, as before, attended for an appointment with the midwife where, she maintained, her family history was again noted. The midwife’s notes, however, did not indicate that anyone in Mrs Cruisey’s family had given birth to an abnormal child. Mrs Cruisey returned to the hospital a month later and this time was seen by a doctor. Again, she was sure that she had told the doctor about her family history of birth defects, but still no note of this appears to have been made in her medical records. A further month of the pregnancy passed and Mrs Cruisey then attended the hospital for a routine ultrasound scan in which no abnormality of her unborn child was found. But by November, complications had arisen – she was admitted to the hospital with serious bleeding and had a second ultrasound scan and then a third a week later which confirmed the presence of a meningomyelocele – a protrusion of the membranes surrounding the spinal cord through a defect in the spine – in her unborn child. By this time, though, it was too late for Mrs Cruisey’s pregnancy to be terminated (had she so wished) and she gave birth to a physically handicapped daughter on 7 January 1992.

Mrs Cruisey commenced legal action against St Helier Hospital, alleging that, but for their negligent failure to carry out a proper ultrasound scan and/or their failure to correctly interpret that scan, she would have terminated the pregnancy (although she would have tried again for a second child) and so would not have had to bear the additional costs and expenses of bringing up a severely handicapped little girl, along with the accompanying stress that this would place upon the family as a whole. Moreover, it was argued that if her family history had been noted properly from the outset, the hospital would have been alerted to the possibility of congenital malformations.

Ultimately, the hospital settled the case with Mrs Cruisey out of court for £135 000 (sterling) without ever admitting liability. But what catalysed this settlement was not, strictly speaking, the hospital’s belief that one of their clinicians may have negligently failed to take the appropriate medical history from Mrs Cruisey or failed to spot an abnormality on the initial ultrasound scan (although this possibility could not be discounted), but the contents of a letter that came to light just a few months before the trial was due to begin. The letter had been written by the consultant radiologist at St Helier to the hospital’s chief executive about 2 months before Mrs Cruisey had attended for her initial ultrasound scan. It stated:

[T]he quality of image which we are able to obtain has, over the past few months, become increasingly variable to a point where we now have considerable problems in interpreting the scan…. I do understand the very considerable difficulties which are being encountered in formulating an effective equipment replacement list, but I think it is only fair to point out to you that although I will do my best to maintain antenatal and inpatient services I cannot continue to operate at the present level with only three out of the four machines working properly.

A new machine had been ordered in September 1991, 6 weeks after the plaintiff’s routine ultrasound scan. It seems that the hospital was not, in this case, prepared to risk a finding by the court – notwithstanding the absence of any proof of negligence on the part of the radiologists involved in the performance and interpretation of Mrs Cruisey’s scan – that the ultrasound equipment itself was, on the balance of probabilities, faulty through poor maintenance or simply ordinary wear and tear. Since the servicing and ultimately the replacement of the ultrasound scanner was the responsibility of the defendant hospital, settlement of the case was the only sensible option open to them.

For those of us practising medicine from a distance using information and communication technologies, the fundamental lesson of this particular case, and of others like it, is that the insertion of those technologies into the matrix of clinical risks associated with the treatment of distant patients requires at least as much energy and attention to be focused on the responsibilities of those providing and maintaining telemedicine systems as is focused upon the training and competence of the clinicians using them. Used properly, however, safe and well maintained telemedicine equipment can and does reduce the clinical risks associated with the interpretation of clinical images.

For instance, an investigation by Vincent et al. [48] of the ability of junior doctors working in accident and emergency departments to detect radiographic abnormalities produced a somewhat alarming result. When comparing the assessment of a radiograph by a senior house officer with that of a consultant radiologist or registrar, the former showed an error rate of 39% – almost four out of 10 cases – for abnormalities which might have clinically significant consequences. There was no improvement shown during the officers’ 6 month tenure of their posts. But the paper observes, candidly, that it would be unrealistic to expect senior house officers in accident and emergency departments to acquire the difficult skills of image interpretation in that short time. Where, however, a teleradiology or PACS system is available, linking the accident and emergency department with the radiology department, there is little or no need for radiographs to be released unreported and junior doctors are able to call upon the advice and mentoring of consultant radiologists at any time.

Telesurgery and robotics

Telesurgery is surgery in which the surgeon is not at the patient’s immediate side: visualization and manipulation are performed using tele-electronic devices [17]. Ultimately, the goal of telesurgery is to make surgical expertise available to patients who, for whatever reason, are inaccessible: for example, due to distance from the surgeon (e.g. in a remote rural area or in outer space); because they are in a hazardous environment (e.g. a battlefield or the scene of a nuclear accident); because there is some other form of barrier or danger presented to the surgical team by the patients themselves (e.g. radioactive contamination or contagious disease); or because there is some danger presented by the surgical team itself (to, say, an immunodeficient patient).

The thinking behind telesurgery is by no means new. Shortly after the end of World War II, the United States Atomic Energy Commission greatly accelerated the development and construction of nuclear power stations and it became necessary to develop ‘teleoperators’ that could carry out delicate tasks such as assembling and disassembling reactor components and experimenting with and moving radioactive materials. The first such teleoperator was developed and demonstrated by Goertz and Thompson in 1954 [49]. Within a decade, telemanipulation using electrical, hydraulic and pneumatic devices capable of operating on a microscopic scale under the command of an operator who was separated from the contaminated environment by leaded walls (the manipulating arms being viewed through a video-display screen or closed-circuit camera) or thick glass was commonplace. Yet it was not until 1972 that Alexander first suggested that such technologies could be applied to surgery – in this case to the problem of providing surgical care to NASA’s astronauts [50]. But such a system would have required computer processors and telecommunications of a speed and power that simply were not available then and hence it is only in the 1990s that the technologies have become available with which to make telesurgery a reality.

Telesurgery has also been made possible by the increasing use of robotic aids, together with microcameras, echography, lasers and optical instruments in modern surgery. Some embryonic forms of telesurgery, such as minimally invasive surgery (or ‘keyhole’ surgery) already exist, and have demonstrated that complex procedures can be performed safely using a microcamera that projects images from the patient onto a video screen and surgical instruments that allow minimal tissue manipulation. Such images can be displayed for viewing by a remote observer (as in videoendoscopy) and two-way video and audio conferencing facilities can be put in place to allow remote supervision or instruction.

The challenge of telesurgery, however, is to develop the interface devices: the digital data encoders and decoders and the video, audio and haptic subsystems that are required for remote human surgeons not just simply to position retractors or laparoscopes [51–54], but actually to feel the tissue or bone that lies ‘beneath’ their hands. Current technology is limited to use on rigid, fixed structures such as bones, and the practical necessity of controlling surgical robots from a remote location in anything other than military or outer-space scenarios has not yet been sufficiently well established.

Telesurgery is, in some ways, in the same position as telemedicine was some 10–15 years ago: wonderful possibilities exist but the attitudinal barriers that need to be overcome are formidable. Surgery is still seen, by both surgeons and their patients, as a highly complex skill that is both taught and practised as a craft [55]. Undergoing surgery involves, for most patients, placing total faith in the skills and experience of the surgeon and his or her team and this is why the giving of informed consent to the surgical ‘invasion’ of one’s body still has such primacy in medical ethics. The use of robot surgeons capable of performing functions that are predetermined by their programming or the instructions given to them by a surgical team raises important questions, not just about who should bear responsibility for any harm caused by robotic surgeons, but also about when it is ethically appropriate for them to be used in the first place. Should we, perhaps, envisage a future in which ethical guidance from a ‘Royal College of Robotic Surgeons’ resembles the writings of Isaac Asimov [56]?

1 A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2 A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.

3 A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

Clinical risks in telesurgery. There are some very good arguments indeed for using robots in any surgical situation where they can perform a task better than a human surgeon, such as in preparing the femoral canal to receive an artificial hip replacement [57], in ‘keyhole’ laparoscopic surgery [5859] and in very delicate procedures such as the suturing of blood vessels [6061] or stereotactic tissue biopsy [6263]. In these types of applications, the critical part of the robot is not the mechanical part (which must perform non-repetitive actions within an accuracy of microns) but the software determining its use and the means by which the points of reference that will guide the movement of the robot are ‘mapped’. Yet it is this very mapping and the software used to facilitate it that may present a far greater danger than control of the robot itself.

Consider, for instance, the Therac-25 case in which at least six patients were killed or maimed by an X-ray machine with faulty software [64]. The patients in question were all undergoing radiotherapy in which a medical linear accelerator was used to create a high-energy electron beam that was focused onto the patient to destroy tumour tissue, leaving healthy tissue outside the focus of the beam relatively unaffected. A tungsten shield was used in the accelerator, both as a lens that focused the beam to the correct tissue depth and as a filter to ‘flatten’ the beam to therapeutic levels. In the Therac-25 accidents, the tungsten shield was not in place because of modelling errors in the system’s software and hardware, and as a result patients received a full dose of the ‘raw’ 25 MeV electron beam containing 20 000 rad, when in fact 200 rad was a sufficient therapeutic dose. A 500 rad whole-body dose causes death in 50% of cases.

In this instance, the malfunction was caused by the use of software developed for an older machine, the Therac-20, which had a physical back-up safety system which the Therac-25 did not have. Moreover, the system that measured the radiation dose given to patients became saturated when exposed to the full strength of the Therac-25’s unfiltered beam and so gave a misleadingly low reading [4].

This example alone demonstrates the dangers of displacing human beings with robots and computers or of giving them a level of autonomy that enables them to process information and make what they perceive as ‘logical’ decisions which turn out, in actual fact, to be fatally flawed. Of course, anything powered by electricity has an ‘off’ switch and a prerequisite for robotic surgery would naturally be, amongst other things, the existence of a manual override for use by the surgeon in charge at the remote site, as well as the availability of alternative communication lines for use should the initial lines fail. Standards for the security, integrity and reliability of the broad-band network connections necessary to carry the high volume of information required to command and receive feedback from a robotic surgeon as well as appropriate methods for the sterilization of the robot and all its components will have to be developed. Most importantly of all, the colleges and medical associations as well as health care liability insurers and medical defence organizations will all require telesurgery systems to conform to as yet unwritten European standards (necessary for the issuing of the essential ‘CE’ mark) before they can be approved for clinical use.

But there may still remain the possibility, as in the Therac-25 case, that it will be impossible to tell if any harm has been done until it is too late. Hence the continuing evolution of telesurgery and robotics to execute more and more complex interventions is a huge technical task, but one which will go hand in hand with the development of the features necessary for such technology to be ethically and legally acceptable in the first place as a safe and reliable means of performing surgery from a distance.

Call centres and decision-support software

Telemedicine, we should remind ourselves, need not necessarily involve using the most cutting edge and sophisticated technology available. The humble telephone has been enabling health care providers and patients to communicate for the best part of a century and, with over 90% of European households now having at least one telephone line, the use of the telephone for tasks other than mere social or business exchanges has been increasing rapidly.

Thus telephone call centres, already ubiquitous in industries such as banking and insurance, are fast becoming key features of the provision of front-line primary care services, and NHS Direct, the UK National Health Service’s new telephone helpline, has received widespread attention as a model for the provision of such services elsewhere in Europe. This helpline, trialled in three British pilot sites, has four principle objectives:

  • • to help the public understand what is a health care emergency – the majority of the general public does not understand the severity of many medical conditions;
  • • to assist the public in dealing with health care emergencies themselves whenever possible;
  • • to direct those needing assistance to the appropriate source of care; and
  • • to answer questions and provide information to the public.

NHS Direct is just one example of the telephone triage services now available in 14 countries around the world. Although these services have to cater for different types of health system or market, the common denominator that links them all is the use of trained nurses receiving telephone calls from the general public and giving information and advice with the aid of computer-based algorithms. The software package that contains these prompts the nurse to ask callers a series of questions designed to ascertain the nature and seriousness of their condition and to direct them to the most appropriate source of treatment, whether that be self-help at home, a visit to their general practitioner or – at the farthest end of the scale of seriousness – transferring them to the local ambulance service for immediate emergency admission to hospital. Follow-up calls can be made by the nurse to check on the progress of individual callers.

Call centres such as these don’t need to limit themselves to telephone triage and advice. They can be used for liaison between outpatient departments to manage routine appointments, admissions and follow-ups. They can provide out-of-hours messaging and appointment services for general practitioners as well as assisting with the monitoring of patients who have been recently discharged from hospital or others at risk. In due course, therefore, we should see these call centres taking on an ever greater role in tasks such as the monitoring of remote and implanted sensors and the management of social alarm services for the sick, elderly or handicapped who need assistance to live independently in the community [23].

Autonomy and responsibility in computer-based decisions. Call centres are still, however, an unknown quantity as far as the risks involved in using algorithm-based clinical protocols are concerned. At the most basic level, the making of a ‘decision’, in so far as someone contacting a call centre with medical symptoms is advised to pursue a certain course of treatment, may not involve the skill and training of the nurse who takes the call at all. Rather, by requiring the nurse to answer a series of questions in a logical order, the computer software used by the call centre can obviously come to its own conclusions regarding the most likely diagnosis, designed to suggest to the nurse the safest possible course of action. For example, where the possibility of meningitis could not be excluded, the nurse might be prompted to advise immediate admission to hospital and to phone ahead to advise the hospital to expect the admission.

The potential uses of such decision-support systems are by no means limited to call centres. They can assist clinicians in calculating drug dosages, warn pharmacists about the possibility of drug interactions and contraindications, help intensive care nurses to work out i.v. drip rates, enable electrocardiograms to be interpreted and dysrhythmias in hospitalized patients to be spotted, as well as, of course, enabling researchers to locate relevant medical texts [65]. The continued expansion of managed care in the USA may well lead to greater reliance on decision-support systems working in parallel with electronic health records, to identify and manage courses of treatment [66].

A number of short- and longer-term risks are inherent in the use of such systems, however. The most immediate risks are those associated with the reliability and integrity of the software itself. As a means of assisting a doctor or nurse in the making of a clinical diagnosis, decision-support software should really be treated no differently from a stethoscope – it is a tool for maximizing the effectiveness of the human clinician’s own ability to receive and process information and, ultimately, to make a diagnosis [67]. Properly trained health professionals know when to rely on such a system (which can absorb the kind of volume of medical information that it would take a human being years of study and access to a vast library to fully command) and when to rely upon their own judgement. There may well be cases in which the software gives a clearly incorrect or over-defensive solution for a patient’s treatment.

But just as human clinicians are required to stay abreast of the latest progress and developments in their field, so too must decision-support software be kept up-to-date in order to reflect the best possible levels of practice. Yet this may not be easy when massive amounts of effort are needed to build and maintain the medical knowledge content of such software [68–70]. In the future, therefore, should we expect decision-support systems to earn their continuing medical education (CME) credits in the same way as human clinicians? Moreover, where these systems are subject to long-term use, will we witness a skills degradation amongst clinicians who become over-reliant upon them and the eradication (or, at the very least, a serious erosion) of the traditional doctor–patient relationship, based as it is on compassion and trust?

The imperfect nature of human–machine interaction, the individuality of patients and the way in which they can present identical medical complaints differently, the changing vocabulary of medical science and the limitations of the information made available in a distant medical encounter all conspire to make the use of decision-support systems, and particularly their application within call centres, a legal and ethical enigma. The present immaturity of such systems requires, I believe, cautious and close supervision of their practical use.

Conclusion

The rapid and ongoing development of the social and medical uses of information and communication technologies is changing the landscape of health care practice forever. But the changes that are taking place may not necessarily be for the better. As we stand at the dawn of the 21st century, health telematics applications such as electronic patient records, the transmission of medical images, telesurgery and telephone call centres are all shrinking the world in which we live by removing the barrier of distance that so often lies between ourselves and those we look to for our health care. The fundamental question is: do we (or should we) want all the possible consequences of this?

For when we use e-mail or our telephone to stay in touch with friends or family in faraway places, we exercise a personal choice to use that medium rather than visit them in person, because it is impossible or impractical to so. The technology that enables us to do this enriches our lives by removing the boundaries which previously kept us apart. The same argument applies to the new speed with which we can trade and communicate with those with whom we work and do business. But what if e-mail or the telephone became the only way in which we could ever make such contact? We would, quite literally, become prisoners of our own technology.

Of course, we wouldn’t let this happen. No amount of technology can remove the basic need for human and social contact or for exposure to a wider and more stimulating environment than simply the inside or our own homes. So just as e-commerce will not stop us going shopping, because we love to go out, to look at new things and to try on new clothes, e-health and health telematics should not prevent us from going to see a health care provider in person should we wish to do so.

The potentially dangerous consequences of the health telematics revolution, however, will be our increasing reliance upon technology and the de-skilling this may entail for medical professionals; the heightened risks to our personal privacy and autonomy presented by the use of electronic rather than paper-based health records; the threats to the quality and integrity of professional medical services caused by the growing number of sources from which we can obtain them; and most crucially, the possible eventual loss of personal choice in how we receive those services.

The sceptics of telemedicine are right to insist that it should only be used where a clear case can be made for its implementation based upon clinical need, cost–benefit or the improvement of the quality of a health service, because the inevitable result of our overenthusiastic development of technologies for distant consultation and treatment – no matter how fast we can be treated in the future – will be the loss of the one intangible thing that we most value about traditional health care: the comfort and compassion human beings can only truly bring each other when they are face to face.

Received 22 February 2000; accepted 1 March 2000.

Ancillary