As this review is interested only in the UK-based experience of teleconsultations, the study’s search strategy was restricted to the United Kingdom or NHS affiliated authors. Databases searched were PubMed, Scopus, Cochrane library, and the Cumulative Index of Nursing and Allied Health Literature. Studies up to the end of 2018 were included with no predetermined lower range. The search strategy was developed in PubMed and translated into other databases. This is outlined in Table 1. Search results were exported to the Mendeley reference manager (Elsevier) and duplicated results were removed.

This review included all participants that used teleconsultations in a health care setting. The only restriction was geography, as the area of the interest is specifically the NHS in the United Kingdom. All studies which used teleconsultations for direct patient-facing care were included. Any studies that used teleconsultations in a nonpatient facing capacity (eg, professional to professional teleconsultations for multidisciplinary meetings) were excluded.

This review was limited to studies that were conducted in the United Kingdom, and there was no restriction on the specialty or type of professionals involved in the consultations. Studies up to the end of 2018 were eligible for inclusion. All studies up to December 31, 2018, were included, and the date of the last search was on February 7, 2019. Study titles and abstracts were independently screened by two reviewers (MOC and MAS) based on predetermined inclusion and exclusion criteria, which are outlined in Table 2. Where abstracts were not available, these articles were excluded.

There was no restriction on the study type eligible for inclusion.

All studies up to December 31, 2018, were included, and the date of the last search was on February 7, 2019. Study titles and abstracts were independently screened by two reviewers (MC and MS) based on predetermined inclusion and exclusion criteria, which are outlined in Table 2. Where abstracts were not available, these articles were excluded. If the study suitability was not clear from the abstract, the full paper was requested for review. Disagreements between reviewers were resolved through consensus. The reasons for exclusion were only recorded at the full-text stage.

Technical feasibility relates to practical issues with using the technology used for teleconsultations, for example, where a study reports difficulties with the audiovisual link. We identified 18 articles, which reported aspects related to technical feasibility of which there were 8 pilots [14,32,37,39,43,56,82,86], 2 case reports/series [17,36], 1 descriptive study [46], 1 RCT [61], 2 systematic reviews [73,75], 2 nonsystematic reviews [54,67], 1 mixed method study [105], and 1 study protocol [38].

In total, 11 studies, including both systematic reviews, reported acceptable quality and reliability of the teleconsultation link. In total, 3 studies reported problems with audio or video quality, and this tended to follow the perceived importance of that deficiency. For instance, one study reported that the image quality of a video link was so poor such that 46% of dermatologists felt the diagnostic accuracy would be adversely affected [14]. In another, among peritoneal dialysis patients, poor image quality affected its utility in assessing Tenckhoff catheter sites [61]. A teletherapy study found that sound lag affected the flow of therapy at times, although it concluded that this was not prohibitive to continuing the session [43].

In a study designed to assess the feasibility of a dedicated teleconsultation link in a police college, the authors reported significant reliability issues, where only half of the intended patients were able to be seen by teleconsultation due to unspecified technological failures [86]. A more recent study, using commercial Web-based video calling technology, found that technical issues were minor but often prohibitive to proceeding with the consultation if not rectified. Workarounds by the clinician rectified these; loss of sound on two occasions was overcome by using a telephone for audio, and lack of video was found to be due to the patient forgetting to turn on the Web camera [105].

Satisfaction and user experience with teleconsultations was reported in 43 articles of which there were 17 pilots [9,14-16,23,39,43,45,50,52,56,70,77,81,85,97,108], 11 qualitative studies [33,51,57,76,89,93,98,100,101,107,109], 4 RCTs [21,40,59,61], 2 mixed method studies [102,105], 3 trial protocols [38,68,106], 2 case-control studies [42,69], 1 cohort [18], 1 systematic review [31], 1 nonsystematic review [54], and 1 descriptive study [47].

Satisfaction was assessed using feedback questionnaires in 23 articles, including 3 RCTs. Of these, 19 reported high levels of satisfaction with the medium. In 2 RCTs, satisfaction in the teleconsultation arm was actually greater than the face-to-face group [21,59], whereas another found no difference [40]. By contrast, only 1 small RCT found that patients were less satisfied with teleconsultations due to poor image and audio quality [61]. One pilot found that, while patients were satisfied, health care staff were uncomfortable with it; citing that they felt more on show to senior colleagues and families than would be normal in a face-to-face appointment [70].

A systematic review of patient satisfaction with teleconsultations concluded that although the published evidence suggests that teleconsultations appear to have high satisfaction rates in a variety of settings, we should be cautious about interpreting that as a true reflection of real life [31]. The authors suggest that most studies conducted tried to minimize the inconvenience for those taking part, and often, patients were seen both in person and by teleconsultation.

Patients and staff may be satisfied with teleconsultations, but that is not to say that they are preferable to face-to-face consultations. Several studies found that patients were satisfied with teleconsultations but also that they would still want the option to attend in person as they believe it to be the gold standard [77,107,109].

Qualitative studies exploring users’ experiences of teleconsultations find that the main benefits commonly reported by patients are convenience, reduced travel, and greater accessibility to specialist care and improved flexibility of appointments, allowing minimal disruption to daily life [102,107]. Several studies found that the medium allowed patients to open up more than face-to-face consultations and that they felt empowered to ask more questions [57,107]. Among staff, a greater sense of job satisfaction and a reduced burden of travel have been reported [101]. Among a cancer population, participants reported a preference for receiving bad news in the comfort of their homes rather than in hospital [105].

By contrast, among a teenaged population being treated for chronic fatigue, participants raised concerns about privacy, fearing that they might be overheard by family. Their parents worried that the connection might not be secure enough to ensure privacy, while some health professionals thought it was an invasion of patients’ personal space [107]. There was the awareness that teleconsultations had certain physical limitations; the qualitative analysis from the large RCT by Wallace et al [59] found, either due to patient expectation or physician need, that the inability to perform physical examinations limited its usefulness [76]. A recent study found that physicians often restricted who was offered teleconsultations based on preconceived impracticalities, or they simply refused to participate in them [105].

One study, in which teleconsultations between patients and hospital specialists were facilitated by general practitioners (GPs), concluded that teleconsultations had a different dialogue flow than traditional face-to-face appointments. In particular, the opening phase of the consultation was found to be unfamiliar, leading to interruptions and apologies on both sides while a dialogue flow was established [89]. Morris et al [102] reported that patients and staff could find the medium awkward and uncomfortable when there was no previous relationship built up. The authors concluded that when there were staff changes in service or new-patient appointments, teleconsultations would not be appropriate. Haig-Ferguson et al [107] found that some participants felt teleconsultations were less personal and that the therapist was less real over a video link, with the screen acting as a physical and emotional barrier. Paradoxically, the same study found that being physically removed from the therapist allowed other participants to open up more easily. Due to these potential social difficulties, authors have suggested that teleconsultations are more appropriate for follow up appointments [102,107].

A telestroke study found that the utility of teleconsultations in facilitating timely care was acknowledged by families of acute stroke sufferers but that the clinical expertise of the on-site team was important for them to have confidence in the process [98].

Where articles commented on the efficacy, safety or other clinical outcomes, these were categorized as assessing the clinical effectiveness of teleconsultations in delivering health care. There were 48 articles of which there were 12 pilots [14,15,30,34,43,48,62,63,65,86,88,108], 5 case reports/series [19,22,36,72,99], 7 case-controls [28,42,44,49,53,69,78], 1 qualitative study [10], 2 mixed method [102,105], 4 retrospective studies [35,58,80,92], 3 audits [83,94,95], 7 RCTs [26,40,59,64,74,84,87], 1 single-clustered, blinded crossover design study [79], 1 cohort [66], 2 systematic reviews [73,75], 1 nonsystematic review [54], 1 descriptive study [90], and 1 service report [96].

Within psychiatry, a single-cluster balanced crossover, blind study (where each patient had both a face-to-face and teleconsultation with a different researcher and each researcher was blind to the psychiatric assessment of the other) concluded that there was significant intermethod concordance, confirming its accuracy in psychiatric assessment [79]. This confirms the findings of an earlier systematic review [75].

Teleconsultations in acute stroke management networks are now widespread in the United Kingdom. In total, 3 retrospective studies of a combined 287 patients conducted in the United Kingdom confirm that its implementation has been safe; door-to-needle time, morbidity, mortality, and discharge rates were comparable to national standards for acute stroke management [92,94,95]. A novel study exploiting an inherent advantage of teleconsultations describes an international telestroke service between Scotland and New Zealand. In this small case series, there were no negative patient outcomes, and the authors suggest that utilization of the time difference would avoid doctor fatigue [99].

In total, 2 case-control studies in neurology assessed the concordance of diagnosis in both an inpatient and outpatient setting and found 96%-100% of cases were accurately diagnosed and managed via teleconsultation [28,29]. An RCT in a neurology outpatient setting compared face-to-face consultations with teleconsultations and found that the teleconsultation arm generated more investigations despite no difference in the diagnostic category of the cases seen. The authors conclude that this reflected a lack of confidence in their teleconsultation diagnosis [40]. A cohort study of 111 inpatients assessed by video link found no difference in 3-month mortality compared with all other hospital admissions during that time. On follow-up, no patient had their diagnosis or management changed when seen face-to-face, and no difference was seen in the use of hospital services in the following 3 months after discharge [66].

A large multispecialty RCT, by Wallace et al [59], enrolled over 2000 patients. They measured the number of investigations per patient and follow-up rates and, in contrast to the previously mentioned RCT, found that teleconsultations actually resulted in fewer investigations, at a rate of 0.79 per patient. However, this figure is offset by a higher rate of subsequent follow up seen in this group.

In an A&E setting, an RCT found no significant differences in diagnostic accuracy or management when teleconsultations were compared with the traditional model of care [64]. In minor injury units, the use of teleconsultations, connecting with a regional A&E center, allowed the majority of patients to be managed locally, with continued improvements seen with increasing technological familiarity [48,58,62,65].

In rheumatology, 2 studies reported conflicting findings. Graham et al [30] found rheumatologists—using a junior doctor as a proxy—were only 40% accurate in assessments via teleconsultation with physicians missing subtle but clinically important signs of inflammation. A year later, Leggett et al [42] concluded that teleconsultations—using a 3-way consultation between the patient, GP, and specialist—were 97% accurate in diagnosing fibromyalgia, degenerative arthritis, rheumatoid arthritis, and soft tissue disease.

In ophthalmology, some eye conditions such as simple ptosis and strabismus could be accurately assessed in up to 97% of cases. However, more complex eye conditions such as socket problems in patients who had a previous enucleation or those with nonspecific ocular pain were better assessed in a face-to-face consultation [44,49].

In Airedale NHS trust, providing a teleconsultation link between care homes and hospitals reduced nonelective admissions by 1731—a 37% reduction—compared with the same period before the intervention [96]. In a hospital diabetic clinic setting, over 4 years, appointment did not attend (DNA) rates were lower (13% vs 28%) in patients choosing to attend by teleconsultation with improved hemoglobin A_1c control [102]. In a prison inmate setting, the use of teleconsultations, coupled with other interventions, improved clinical outcomes for those being screened and treated for hepatitis C compared with controls [108].

Real-world data on teleconsultation appointments as a proportion of clinical activity has been rarely reported and ranges from 2% among a diabetic cohort to 22% among postoperative hepatobiliary cancer patients [105].

In total, 19 articles looked at health care provider cost, patient cost, or costs incurred by both. This comprised 5 RCTs [25,26,41,60,71], 3 systematic reviews [73,75,104], 2 nonsystematic reviews [54,91], 1 retrospective study [27], 1 case-control study [78], 1 service report [96], 3 pilots [12,43,55], and 3 protocols [38,68,106]. Nearly all of the studies reported higher costs for health care providers, including all RCTs. The issue of cost is closely related to the technology used, which has changed greatly over the period of this review—from expensive audiovisual systems to the use of smartphones and computers.

Early studies found that the initial cost of suitable videoconferencing equipment was prohibitively expensive. One early study quotes a figure of £48,000 (US $61,439) to establish a teleconsultation link, including videoconferencing unit and integrated services digital network connection charges [26]. Loane et al published the results of an RCT in 2 papers [25,26]. They found that real-time teleconsultations were 5 times more expensive to run for health providers than store-and-forward teleconsultation models; £132.10 (US $169) vs £29.60 (US $37.90). Although patients saved time and money due to reduced traveling, health care-associated costs were higher in the real-time arm as they took up more physician time than the store and forward model.

Direct comparison with face-to-face appointments in an outpatient setting has found teleconsultations to be more expensive for health care providers. In total, 3 RCTs and 1 case-control study set in pediatrics, neurology, secondary care outpatients, and A&E concluded that teleconsultations were between 15% and 100% more costly to run [41,60,71,78]. A systematic review from an A&E setting found only 23% of studies reported that teleconsultations were cost-effective [73].

Costs could be saved by improving access to specialist care in areas with limited local access to services. A systematic review of telepsychiatry concluded that cost savings would be made by doing just that, and they speculate that technology would become cheaper in the future [75]. A pilot study of teleconsultations in a rural dentistry setting found that up to £270 per patient appointment could be saved by the health service if it adopted teleconsultations to allow rural patients access specialist services [55]. In Scotland, a report on telehealth services found teleconsultations for a 10-week rehabilitation course could be delivered for 3% to 10% of the cost associated with an outreach model (where the therapist travels) or a centralized model (where the patient travels), with the savings primarily being delivered through reduced travel costs [91].

A service report from a well-established teleconsultation service in Airedale, which links an acute hospital with several care homes, reported setup and maintenance costs of £175,000 (US $223,938). However, factoring in costs from avoided A&E attendances and reduced nonelective admissions, the project is estimated to have saved £1,194,083 (US $1,529,939)—a saving of £6.82 (US $8.73) for £1 (US $1.28) invested [110].

In all, 2 RCTs, in different settings, comparing patient costs reported conflicting results. Jacklin et al [60], assessing teleconsultations in multiple outpatient specialties found those who took part in teleconsultations, saved an average of £19 (US $24.34) compared with face-to-face appointments. However, Noble et al [71], assessing its use in a minor injuries unit setting, found patient costs were nearly £15 (US $19) more.

There were 16 studies that either assessed or commented on aspects relating to logistics or operational challenges. These included 3 RCTs [26,59,64], 1 case-control study [53], 2 qualitative studies [57,101], 2 mixed method studies of real-world teleconsultation services [102,105], 4 pilots [34,62,70,108], 1 cohort [66], 1 audit [95], 1 nonsystematic review [54], and 1 report [103]. There is considerable variation in the extent to which this is assessed and exactly what logistical element of interest was described.

Several studies make reference to consultation length, in which teleconsultations are as much as 4 times as long as their face-to-face equivalent [53,57,64]. However, several others found them to be shorter in length [34,102,105]. Williams et al [103] found no difference in consultation length but reports that by avoiding travel to peripheral clinics, clinicians were able to provide more emergency care with the time saved, thus maximizing their clinical efficiency. In older studies, there was often an intermediary, either a GP or another health care professional who would sit with the patient and establish the teleconsultation link [26,59,66]. The extent to which this disruption affects the service provision of the health care professional that is acting in this capacity is not described. In addition to normal duties, clinical staff have reported that they are often needed to triage those who might be suitable for teleconsultations [105]. Furthermore, clinicians may not have dedicated time to do teleconsultations, having to fit them around their normal outpatient schedules instead [59]. Benger et al [64] found that waiting times to access A&E advice were shorter for patients seen by teleconsultation than face-to-face consultation as they bypassed the normal admission processes—in essence skipping the queue.

Altering the way patients are seen can lead to improved operational efficiencies; Ditchburn et al [101] describe a service established to support patients undergoing peritoneal dialysis at home. By avoiding the need to travel to individual patient homes, staff reported that their time was used more efficiently as they were able to do other work on their computer while monitoring the patient via the video link. Nonattendance at hospital appointments is a source of lost revenue for health care providers and results in inefficient use of clinician time. By selectively choosing patient populations with high DNA rates, it is possible to achieve more operational efficiency. Morris et al [102], among a diabetic cohort, improved the DNA rate from 28% to 13%. Morey et al [108], among a prison population, describe a complete overhaul of a hepatitis C screening program pathway (along with other measures, teleconsultations were introduced), which led to a significant fall in DNA rates. These examples demonstrate clinical staff as drivers of change, but they can also be barriers to wider implementation [70].

Teleconsultations, as described by many of the studies included, can be seen as supporting a hub and spoke model of care, with district general hospital (DGH) spokes using teleconsultations to connect with more specialized hub hospitals. Agarwal et al [95] describe a telestroke mesh network of DGHs without a central hub, where out of hours stroke thrombolysis support was provided using telestroke rota shared across the region, thus reducing the frequency of a stroke physician’s on-call nights. Furthermore, such a model meant that thrombolysis care could continue without significant investment in staff and reorganization of thrombolysis care into a hub and spoke model.

The extent to which these services have been integrated into routine practice has been largely superficial. This means that small scale services were often provided with ad hoc support from IT departments rather than formal arrangements [101]. To provide a wider rollout of teleconsultations would require dedicated support from IT. Greenhalgh et al [105], through key stakeholder interviews, reported that NHS IT processes would require major changes to speed up the introduction of new technologies into practice.

In the NHS, hospital trusts are reimbursed through tariffs, often based on a per-patient seen basis, with different tariffs in place for face-to-face consultations and phone consultations. No such tariff existed for teleconsultations, which means that managers are often unwilling or unable to justify diverting the cost of such services from increasingly stretched clinical budgets [70,105].