The Evolution of Teleophthalmology Programs in the United Kingdom

Abstract

Modern ophthalmic practice in the United Kingdom is faced by the challenges of an aging population, increasing prevalence of systemic pathologies with ophthalmic manifestations, and emergent treatments that are revolutionary but dependent on timely monitoring and diagnosis. This represents a huge strain not only on diagnostic services but also outpatient management and surveillance capacity. There is an urgent need for newer means of managing this surge in demand and the socioeconomic burden it places on the health care system. Concurrently, there have been exponential increases in computing power, expansions in the strength and ubiquity of communications technologies, and developments in imaging capabilities. Advances in imaging have been not only in terms of resolution, but also in terms of anatomical coverage, allowing new inferences to be made. In spite of this, image analysis techniques are still currently superseded by expert ophthalmologist interpretation. Teleophthalmology is therefore currently perfectly placed to face this urgent and immediate challenge of provision of optimal and expert care to remote and multiple patients over widespread geographical areas. This article reviews teleophthalmology programs currently deployed in the United Kingdom, focusing on diabetic eye care but also discussing glaucoma, emergency eye care, and other retinal diseases. We examined current programs and levels of evidence for their utility, and explored the relationships between screening, teleophthalmology, disease detection, and monitoring before discussing aspects of health economics pertinent to diabetic eye care. The use of teleophthalmology presents an immense opportunity to manage the steadily increasing demand for eye care, but challenges remain in the delivery of practical, viable, and clinically proven solutions.

The burden of disease in diabetes eye care has reached gargantuan proportions in the 21st century. There are an estimated 552 million people globally who will live with diabetes by the year 2030, with half of this population eventually developing diabetic retinopathy.¹ As such, we are approaching a landmark in history where radical changes in the way we practice ophthalmology will be required to maintain delivery of high quality care to these patients.

The examination of the human retina was only made possible in 1851, when Hermann von Helmholtz first held up a lens to a patient’s eye using a naked candle as a source of illumination.² To this day, ophthalmologists round the world use the same physical principles to perform a retinal examination. The evolution of the ophthalmic examination has occurred in a gradual step-wise fashion, following the developments of its time; for example, the invention of electricity allowing for more powerful light sources, and modern optical lenses that offer a wider field-of-view of the retina. However, it has been the more recent exponential leaps in digital technology that have fuelled innovation in ophthalmology—in the form of digital ophthalmic imaging. This includes optical coherence tomography (OCT), a retinal imaging modality that uses the light scattering properties of tissue to give a cross-sectional image of the eye of with resolutions better than 10 µm^{3, 4} as well as ultra wide field fundus photography, which enables imaging of 80% of the entire retina in a single 200-degree frame.⁵ However, at present, standard 30-degree color photography remains a widely used modality in retinal screening.

Currently, digital retinal imaging technology is largely used as an adjunct to the ophthalmology consult, integrated into existing patient pathways where the patient is seen by a nurse, ophthalmic technician, photographer, before being examined by an ophthalmologist. If digital retinal imaging technology continues to be used solely in traditional models of care, its advancement is in danger of stagnating. For digital retinal imaging to maintain its exponential advancements, it must build on core technological advancements such as computing power, storage, bandwidth, and integrate with similarly growing platforms. Perhaps most important, it must remain relevant to the needs of the population. An example outside of telemedicine is the field of telephony; with its powerful mobile devices and wireless networks, this existing technology has harnessed the unprecedented pace of technological growth, combined with other successful industries (music, movies, gaming), remained relevant to its users, and has been further amplified as a result. Another field of development, which shows a lot of potential, is that of image processing and analysis. Increasing access to processing power has lead to a steady stream of research in ophthalmic image processing and analysis.⁶ However, current systems tend to be compared to human specialist interpretation as the gold standard and derive focused measures that are limited in scope. For example, a method used to assist in the diagnosis of early diabetic retinopathy is that of automated image processing applied to detection of microaneurysms (one of many features of diabetic retinopathy) in fundus images. They do not yet provide true independent machine vision decision algorithms that would provide holistic management decisions negating the need for specialist input.

Teleophthalmology is a branch of telemedicine that delivers eye care at a distance, which is then transferred via telecommunications technology to remote eye specialists. It provides the ideal platform in which digital retinal imaging may interact with mutual technological advancements, combined with similar achievements made in the field of pharmaceuticals to create the next wave of ophthalmic innovations. Telemedicine programs have traditionally been used for the purposes of distributing medical care to places with limited access such as developing countries.⁷ However, current estimates suggest that a wide disparity exists even in the first-world setting, where only 60 to 90% of patients with diabetes have access to early detection, evaluation, and prompt treatment of diabetic retinopathy.⁸ In the United Kingdom, between 20 and 50% of patients develop complications before their diagnosis of diabetes.⁹ In type 2 diabetes, approximately 35% of newly diagnosed patients have some evidence of diabetic eye disease at the time of diagnosis.¹⁰ This places a large burden on the UK National Health Service (NHS): capacity in eye clinics and the costs of providing such a service. Further compounding the problem are the therapeutic advances achieved for diabetic retinopathy in the past decade; though highly effective in preventing visual loss, these require more frequent visits to the eye clinic.^11-14 One example is anti–vascular endothelial growth factor (VEGF) injections for the treatment of diabetic macular edema that are not only administered monthly, but studies have also shown that patients may require long-term treatment for more than 3 years.¹² Now more than ever, diabetic eye clinics are bursting at their seams. This, and the disruptive effects of rapid advancements of technology, digital retinal imaging, and therapeutics have challenged our traditional models for the delivery of eye care.

This review evaluates the current teleophthalmology programs deployed in the United Kingdom, in particular the different approaches used in diabetic retinopathy, its response to increasing demand for eye care, and digital diagnostics and therapeutic developments in recent years.

The Evidence for Teleophthalmology

Despite the apparent benefits of teleophthalmology, there has been limited evidence from randomized controlled trials (RCTs) demonstrating its effectiveness. However, some authors have suggested that such trials may not be the method of choice to evaluate their efficacy.¹⁵ Table 1 briefly summarizes the available RCT data to date in the field of teleophthalmology. Two studies examining diabetic retinopathy demonstrated that initial examination with digital fundus images improved attendance and adherence to follow-up with clinicians.^16,17 Additional meta-analyses of clinical studies have demonstrated that the pooled sensitivity and specificity of telemedicine for diabetic retinopathy classification is greater than 80% and 90%, respectively.¹⁸ To date, 1 RCT comparing OCT and nonmydriatic fundal images for detection of age-related macular degeneration (AMD) compared to standard care demonstrated no difference in delay to treatment or final visual outcome (Table 1).¹⁹

Table 1.

A Summary of Randomized Controlled Trials in Teleophthalmology.

Study	Disease	Intervention	N	Outcome	Results	Conclusion
Davis et al 2003	Diabetes	Teleophthalmology program using a nonmydriatic retinal camera versus usual care in a rural poorly served setting	59	Frequency of eye examinations	Relative risk was 5.56 (95% CI 2.19-14.10)	Patients who received their eye exam via telemedicine at the primary care site were 5 times more likely to obtain a screening eye examination then those who were asked to make an appointment with an eye care specialist
Conlin et al 2006	Diabetes	Teleophthalmology program using a nonmydriatic retinal camera (Topcon TRC-NW5S) versus usual care	448	Adherence to follow-up dilated exams by health care professionals in 12 months	Adherence rate was 87% vs 77% in teleophthalmology group (P < .01)	Teleophthalmology improves attendance at diabetic assessment clinics
Li et al 2015	AMD	Teleophthalmology program using an OCT and nonmydriatic retinal camera versus usual care	169	Delay to treatment, diagnostic accuracy, and visual outcome	No significant difference in delay to treatment or visual outcome between groups; diagnostic accuracy in the teleophthalmology group was 42.3%	There is no difference in delay to treatment or final visual outcome; referral accuracy to hospital services can be improved

Abbreviations: AMD, age-related macular degeneration; OCT, optical coherence tomography.

An Overview of Teleophthalmology Programs Currently Deployed in the United Kingdom

The United Kingdom has an aging population, and with it an associated increase in prevalence of ocular diseases such as diabetes, age related macular degeneration, and glaucoma and consequent growth in the number of patients attending hospital care services on a regular basis. In the last few decades, developments in imaging, such as nonmydriatic fundus photography and the widespread adoption of OCT have resulted in a large volume of digital information that can be rapidly acquired. Assimilation and analysis of this information is a common rate-limiting step in clinical decision-making. Parallel to developments in imaging, information technology systems have similarly progressed and many challenges related to storage and transmission of digital data have been overcome.^20,21 As a result, teleophthalmology is currently burgeoning as a way to improve clinical decision making efficiency and is likely to become a key factor in the provision of modern eye care.²²

Glaucoma

Teleophthalmology programs were initially introduced to the United Kingdom by glaucoma specialists in community-based, optometrist-led clinics, driven by the increasing prevalence of disease in an aging population and the requirement for life-long monitoring in this chronic disease. The prevalence of glaucoma is 2% in persons aged above 40 years but rises to 10% in those over 80 years.^23-25 Furthermore, population-based surveys estimate that the number of patients with primary open angle glaucoma will increase by one-third in England and Wales by 2021²⁶ and consequently, hospital services are seeing an influx of new glaucoma referrals. Several region-based programs incorporating telemedicine principles have developed in an attempt to resolve both current and predicted capacity issues from these new cases of glaucoma, as well as manage the existing cohort of patients requiring follow-up visits.

At the Portsmouth-based glaucoma refinement program, referral pathways from optometrist to hospital eye services for new glaucoma patients were evaluated.²⁷ There, an optometrist examines patients with a standard Humphrey 24-2 visual field, applanation tonometry and a digital disc photograph. All the information is passed digitally to a consultant in a “virtual clinic,” who then decides whether the patient should be referred to the hospital or followed up in the community by an optometrist. An audit of this program demonstrated that only 11% of the referrals needed an appointment in the hospital. The positive predictive rate was double the previous referral system and represented an annual saving of approximately £250 000 to the Portsmouth Hospital Trust.

As well as improving efficiency in referral pathways, teleophthalmology has been used to follow-up glaucoma patients in the community. A study in Peterborough examined low-risk glaucoma patients who were followed up in the community by a designated optometrist.²⁸ Visual fields, intraocular pressure measurements and optic disc photography were performed at a local optometrist with good agreement between the optometrist and the consultant ophthalmologist virtual review, allowing many low-risk patients to be effectively monitored in the community. In Bristol, a series of mobile clinics were set up in 2007.²⁹ These were run by optometrists and technicians and have examined over 100 000 individuals with Humphrey 24-2 visual fields testing with optic disc imaging in a store and forward model, reviewed by consultant ophthalmologists in a ‘virtual clinic.’ In this model, it was observed that the use of teleophthalmology allowed a more efficient assessment and senior clinical review of patients resulting in a 2.4% reduction in the number of follow-up appointments. Similar community-based programs in Cambridge have demonstrated improved diagnostic accuracy and a reduction in the need for hospital follow-up with the establishment of virtual image-based triage clinics.³⁰ A more recent report has focused on establishing a national glaucoma referral refinement scheme (CHANGES) where a stratified risk categorization is made based on senior optometrist or consultant review of applanation tonometry, Humphrey visual field testing, and fundus photography. While the false negative rate of this optometrist led system was 15%, no diagnoses of glaucoma were missed suggesting that although inaccuracies and disagreements exist, virtual assessment is a safe adjunct in glaucoma care.³¹

Retinal Disease

The store and forward model has also been applied in diagnosis and management of macular (central retina) diseases.³² In a study performed by Kelly et al, fundus photographs and OCT scans performed by a community optometrist were transferred by secure file transfer email for review by a consultant ophthalmologist.³³ In 48 of the 50 of cases referred, images were reviewed within 48 hours and one-third of patients were successfully managed in the community without needing to referral to hospital services. Similar optometry led retinal referral programs in Scotland based on digital fundus photography (Topcon, NW6 nonmydriatic cameras) have resulted in a reduction in over one-third of inappropriate referrals to hospital eye services.^32,34 A further example of community-based management of retinal disease is the monitoring of choroidal naevi; a common incidental finding on routine fundal examination that may be associated with an increased risk of choroidal melanoma.^35,36 Although practice differs across the United Kingdom, the Royal College of Ophthalmologists suggests benign typical nevi can safely be photographed and monitored in the community given the low risk of malignant transformation.³⁷

Emergency Eye Care

In Wales, a pilot teleophthalmology program has been deployed in the emergency eye services where a smartphone was used to capture and send images to a consultant ophthalmologist. This innovation was designed to overcome the geographic difficulties in obtaining an expert opinion, as patients requiring a senior clinical opinion needed to travel 3 hours to be seen at a larger hospital by a consultant.³⁸ Implementation of this program meant that a nurse practitioner could examine these patients in the local eye unit, obtain timely and appropriate advice from a consultant and treat the patient accordingly. Over a 1-year period, all patients demonstrated an improvement in their acuity and a resolution of their symptoms. No patients needed to be physically transferred to diagnose and manage their condition. This model is similar to established mobile health or “mHealth” models used in developing countries’ and other resource poor settings where access to eye care services in remote populations is logistically challenging. One of the largest uses of teleophthalmology in this way was recently performed in rural India, where nearly 20 000 patients attended “virtual eye camps” where fundus images were obtained and sent to Ophthalmologists in the nearest city.³⁹ Patients needing surgical intervention were efficiently selected and referred directly to the nearest treatment centre. There are many other examples of the use of teleophthalmology in similar settings providing efficiency in diagnosis and high patient satisfaction.^40-42 Table 2 highlights the current teleophthalmology programs that have been successfully deployed in the United Kingdom.

Table 2.

Successfully Deployed Teleophthalmology Programs in the United Kingdom by Geographic Location and Disease Entity.

Nondiabetic teleophthalmology programs in the United Kingdom
Location	Disease entity
Portsmouth28	Glaucoma detectiona
Peterborough29	Glaucoma detectiona
Bristol30	Glaucoma detectiona
Cambridge31	Glaucoma detectiona
Bolton34	Macula disease
Tywyn36	Emergency eye disease

^aCHANGES,³² a national glaucoma referral refinement scheme, was suggested in 2015.

Retinopathy of Prematurity

A further important development has been in the diagnosis of retinopathy of prematurity (ROP). The use of Retcam images in screening of premature or low-birth-weight babies for ROP represents an important advance in the application of teleophthalmology.⁴³ Digital image capture in this setting where examination is challenging, can provide more objective information for disease detection, thereby facilitating Internet consultation and retrospective analysis. In Germany, a 6-year study of over 1000 at-risk babies did not identify any missed cases when comparing wide field imaging versus indirect ophthalmoscope examination.⁴³ Similar community-based studies have also demonstrated its safety with no children with treatable disease being missed.^43-45

A systematic review of screening strategies using digital imaging for retinopathy of prematurity, however, concluded that the current evidence base is not sufficient to recommend the routine use of retinal imaging by neonatal units to identify infants who have severe retinopathy.^46,47 Despite this, the role of telemedicine in ROP screening in the United Kingdom is continuing to expand and is likely to assume increasing importance in the delivery of health care to at-risk babies.

Approaches to Teleophthalmology in Diabetic Retinopathy

The United Kingdom has not been immune to the global epidemic of diabetes. The Health Survey for England in 2013 showed that over the past decade, the prevalence of diabetes has more than doubled (2.9-6.9% in men, and 1.9-5.6% among women), with around a quarter of all adults in the population classified as obese (26% of men and 24% of women).⁴⁸ In England, there are currently 2.6 million people aged 12 or over diagnosed with diabetes.⁴⁸ Of these, 1.9 million (76%) were screened at least annually under the NHS Diabetic Eye Screening Programme (DESP).⁴⁹ Similarly, in the Scottish Diabetic Retinal Screening Program, 79% of the eligible population were screened.⁵⁰ Patients found with potentially sight-threatening retinopathy within the DESP were referred to hospital eye clinics. The current threshold for referral is defined by the National Screening Committee, and is equivalent to the Early Treatment Diabetic Retinopathy Study (ETDRS) definition of moderate nonproliferative diabetic retinopathy or any sign of diabetic maculopathy.^51,52 This constitutes approximately 4.5% of the screening population.⁵³ The majority of patients who receive diabetes eye care in the NHS eye clinics are referred via the DESP pathway.

Diabetic Retinopathy Screening

In the United Kingdom, the main implementation of teleophthalmology lies in its national diabetic retinopathy screening program. The success of the national screening service is best reflected the changes in blindness registration since the inception of the DESP in 2003. Diabetic retinopathy has for the first time in the past 50 years been replaced by inherited retinal disease as the leading cause of blindness in working aged adults in the United Kingdom.⁵⁴ The Four Nations Diabetic Retinopathy Screening Study Group, which comprises England, Wales, Scotland, and Northern Ireland, observed an elevated rate of developing referable retinopathy in patients with type 2 diabetes who were not screened promptly, and a further a 4-fold increase in risk of developing proliferative diabetic retinopathy was identified in patients whose screening is delayed for 3 years or more.⁵⁵ Furthermore, an association between nonattendance at screening, poor glycemic control, and blindness registration has been observed. In fact, a single missed screening appointment is associated with a 3-fold increase in needing retinal laser therapy.⁵⁶ In the Scottish DESP, Leese et al observed that factors such as a younger age, longer diabetes duration, poor glycemic control, and social deprivation rather than distance/time taken to travel for retinal screening were associated with poor attendance at screening.⁵⁶ The authors make an important point that screening for diabetic retinopathy differs from most population screening programs, which involves a healthy population. Diabetic retinopathy screening is performed on a population with a predefined illness and its associated health care demands. It is therefore unfortunate that patients who are most likely to benefit from screening are least likely to attend. However, it may be that this population provides the greatest opportunity for innovations in teleophthalmology.

Teleophthalmology Versus Screening

What is the difference between teleophthalmology and screening? The American Telemedicine Association defines telemedicine as the use of medical information exchanged from one site to another via electronic communications to improve a patient’s clinical health status.⁵⁷ In contrast, population screening is defined as identifying people in a population that are at risk of a disease. Although screening encompasses traits of telemedicine such as information gathering, transfer, and interpretation separated by both space and time, its main purpose is to detect rather than treat disease. As such, the approach of teleophthalmology in the context of diabetic eye disease is allowed to be more flexible, with a broader scope than that of screening. More specifically, different digital retinal imaging modalities may be applied or combined to detect features that indicate progression toward the 2 main sight threatening but treatable consequences, namely diabetic macular edema and proliferative diabetic retinopathy. Unlike other ophthalmic conditions such as glaucoma where a diagnosis cannot be made consistently using a single-test modality, for example, intraocular pressure measurements, optic nerve head imaging, or visual field tests, diabetic eye disease lends itself to predefined, binary outcomes, measurable using digital retinal imaging. The threshold of these outcomes may set according to the purpose of the teleophthalmology assessment. For disease detection, the threshold may be set at “present” or “absent,” or in the case of monitoring established disease, the thresholds may be set as “treatable” or “stable.”

Disease Detection

In the United States, teleophthalmology programs such as the Joslin Vision Network (JVN) provide opportunistic screening for federal health care agencies have been observed to be a similarly successful model, both less costly and a more effective strategy compared to conventional clinic-based ophthalmoscopy.^58,59 The JVN program has been implemented in primary care setting within a population at a high risk for visual impairment secondary to diabetic eye disease. An example is the Indian Health Service where eligible American Indian and Alaska Native people receive health care without direct out-of-pocket expense. Over a period of 5 years, this resulted in a 50% increase in diabetic retinopathy surveillance rate and a corresponding 50% increase in the rate of laser treatment.

Does opportunistic screening have a place in the United Kingdom where a national screening service already exists? The JVN program demonstrated that by making an existing teleophthalmology infrastructure accessible to a high-risk population, it was able to increase surveillance and laser treatment rates dramatically within a short space of time. In the United Kingdom, opportunistic “teleophthalmology” rather than “screening”; by engaging primary care services and optometrists using this approach has already been used in glaucoma, and may be useful for detection of diabetic eye disease in the population who do not attend screening.⁶⁰ It may also detect other retinal conditions such as AMD or vein occlusions that could otherwise be asymptomatic.³³ However, we know that the distance travelled to receive eye screening is not associated with attendance.⁵⁶ Furthermore, formidable barriers to implementing such a scheme include the lack of robust evidence for cost-effectiveness, high costs of digital retinal imaging equipment required, and the current cold financial climate in the NHS.⁶¹ Purchasing expensive imaging devices for placement in primary care services and optometry practices across the country is unrealistic. A solution to this may lie in hospital eye departments where such equipment already exists. A primary care referral or self-referral to “Direct Access Diagnostics” within an eye department, and a store and forward model where images are assessed in a virtual clinic, may be a cost-effective strategy in which to provide improved disease detection in patients with diabetes.

Stable Disease Monitoring

Diabetic eye disease is chronic disease with the requirement of life-long retinal surveillance. It has been calculated that in 20 years’ time, for each person diagnosed with diabetes to have an annual retinal examination, ophthalmologists will have to attend to 2.7 million eyes per day, worldwide.⁶² There are not enough ophthalmologists in the world to make a dent in this colossal volume of patients requiring attendance.^63-65 As mentioned previously, there is an additional subset of patients who are either receiving anti-VEGF therapy or having recently received treatment that will require monthly monitoring of disease. Furthermore, diabetic eye disease comprises 2 main phenotypes: retinopathy, vascular anomalies of the peripheral retina; and maculopathy, microvasculopathy of the central retina. Both can coexist in different levels of severity and rarely occurs in isolation in an individual patient. Presently, no single retinal imaging technique is 100% effective in monitoring both phenotypes. It is likely that a combination of retinal imaging devices (multimodal imaging), which have a high sensitivity of detection of either phenotype that would increase the detection rate of treatable disease in patients. There has been no clinical trial to date that has examined “multimodal” retinal imaging in field of teleophthalmology for diabetic eye disease.

Ultra-wide field retinal photography (Optos PLC, Dunfermline, Scotland, UK) lends itself well for the purpose of monitoring progression of diabetic retinopathy to proliferative disease where new abnormal blood vessels grow on the surface of the optic nerve or retina. It compares favorably to standard-field photography and is able to detect potentially sight-threatening lesions in the peripheral retina.^66,67 These lesions have further been observed to increase the grade of retinopathy severity in up to 10% of eyes examined. Complementary to this, as previously discussed, OCT retinal imaging, by displaying the cross-section of the macula is able to detect diabetic macular edema encroaching on the visually significant fovea (center of the macula) with greater sensitivity than a 2-dimensional color photographs.⁶⁸ The performance of utilizing multimodal retinal imaging to monitor detect and monitor visually threatening phenotypes of diabetic eye disease warrant further study.

Health Economic Analysis of Teleophthalmology in Diabetic Retinopathy

Over the last 15 years, there have been a number of studies of the economic aspects of teleophthalmology for diabetic eye care. Most of these have focused on the cost-effectiveness of diabetic retinopathy screening, with the aim of determining the answers to several critical issues:

• How cost-effective is eye care in diabetic patients?

• What evidence is available to justify systematic screening for diabetic eye disease?

• Is telescreening more cost-effective than a retinal specialist?

• What is the ideal frequency of screening?

• What factors determine the cost-effectiveness of a screening program?

A systematic review of the economic evidence for diabetic retinopathy screening has been conducted by Jones and Edwards.⁶⁹ Diabetic eye disease fulfills the Wilson-Jungner criteria for appraising the validity of the screening program (World Health Organization, 1968)⁷⁰. Notably, such screening programs often generate additional clinical workload where extra health care resources are required. The complete costs therefore needs to be balanced against the benefit, and a threshold of US$50 000 per quality-adjusted life year (QALY) has previously been used.^{71, 72} Almost universally, economic analyses using mathematical modeling have confirmed that not only is telescreening for diabetic eye disease cost-effective relative to this arbitrary QALY threshold, but also in many cases, provides cost savings where the disease is detected early (prompt diagnosis and treatment).

Several economic analyses have confirmed that eye care provided following identification of abnormalities on screening is highly cost-effective. Since it would be unethical to conduct a clinical trial of screening versus no screening,⁷² these studies have used computer simulation to mathematically model the effects of diabetic retinopathy screening programs. In a Swiss study, modeling the introduction of diabetic retinopathy screening for type I diabetics was found to be both more effective and less costly—regarded in health economic terms as “dominant”—than conventional insulin therapy alone.⁷³ Two studies based in the Netherlands supported these findings, and further found that screening in young patients was far more cost-effective than in older patients. Polak et al found no additional benefit from diabetic retinopathy screening in patients over 65 at the time of diabetic diagnosis.⁷³ Crijns et al found that no benefit from screening in patients who developed diabetes after age 75.⁷² A more recent study has found that DR screening is cost-effective up to the age of 80, but not beyond.⁷⁴

Early studies in the United Kingdom compared the benefits of systematic telescreening with the existent method of screening at the time, termed “opportunistic,” which was performed by optometrists, GPs or diabetologists using a direct ophthalmoscope. The latter system had almost no additional administrative cost because there was no call-recall system, audit, training or central coordination. A study by James et al found that the systematic telescreening was marginally more expensive, but identified 157 more cases of sight threatening retinopathy, at a cost of just £32 per case.⁷⁵ Assuming 6000 screening “events,” the sensitivity analysis showed that as long as compliance with systematic screening was greater than 54%, systematic screening was more cost-effective than opportunistic screening.⁷⁵

While the above studies have identified that systematic screening is beneficial and that the resulting treatment is highly effective, these findings alone do not show greater cost-effectiveness of teleophthalmology over the conventional method of screening by a retinal specialist. Screening models of rural communities in Canada⁷⁶ and Norway⁷⁷ both confirmed that teleophthalmology screening is more cost-effective than an ambulatory clinician, although the Norwegian study identified that this was only the case if more than 110 patients were to be screened annually.⁷⁷ In contrast to these findings however, Tu et al compared optometry screening with teleophthalmology screening, but concluded that neither model was cost-effective.⁷⁸

Optimum screening interval, in terms of cost-effectiveness, remains a subject of debate. In a US study, Vijan et al⁷⁹ used cost data from Medicare to model cost-effectiveness for varying screening intervals, and stratified for age and glycemic control. The authors found that for low- and medium-risk groups (as defined by glycemic control), 2-yearly screening was as cost-effective as annual screening, but that annual screening is required for high-risk groups. It is recognized, however, that patients with the poorest glycemic control are often those least likely to attend screening, and this remains a challenge for policy makers.⁷⁹ Most recently, a UK-based study has concluded that in terms of cost-effectiveness, 3 yearly screening is more beneficial than annual screening, but patients should ideally be stratified into high-risk and low-risk categories, allowing personalization of the screening interval to 2-yearly and 5-yearly, respectively.⁸⁰

The emerging evidence supports the use of teleophthalmology for diabetic retinopathy screening. As Kroenke has noted, teleophthalmology is not only useful for providing health care access to patients living in rural, remote areas, but can also be cost-effective in highly urban areas also.⁸¹ Critically, the degree of cost-effectiveness is related to several other factors including the size of the population eligible for screening,⁷⁷ the prevalence of diabetic retinopathy in that population,⁷⁴ age,^72-74 glycemic control,⁷⁹ and patient concordance with screening.⁷⁸ Further analyses of these factors provides the opportunity for identifying subgroups in whom teleophthalmology is both more effective and more cost-effective, enabling policymakers to better plan future services.

Conclusions

We have arrived in an era where the “tricorder” used by Dr Leonard McCoy in Star Trek, appears clunky and primitive. Wearable devices, smartphone sensors, and artificial intelligence have all but replaced stethoscopes and doctors. In spite of this, much of the available evidence for telemedicine in ophthalmology has been narrowly confined to domains such as population screening, and preventative medicine for specific conditions, for example, glaucoma or diabetic retinopathy. More robust RCTs and health economic analyses will be required to extend the application of teleophthalmology toward broad-spectrum ocular disease detection as well as monitoring known, treated, stable eye conditions.

In the United Kingdom, key challenges to effective adoption and systemic diffusion of innovation in telehealth have been identified as the lack of access to data, commissioning services lacking tools and capabilities to drive change, and the lack of a leadership culture and organizational infrastructure to support such a change.⁸² In teleophthalmology, barriers to moving beyond diabetic retinopathy screening include the lack of a nationally accessible electronic health record system and a lack of good evidence for patient and provider satisfaction with its use. Available evidence indicate that there are exceptionally high levels of perceived satisfaction in telemedicine, compared traditional forms of health delivery. However, there is a lack of uniformity in these studies that exhibit design and methodological weaknesses.⁸³ Potential downsides to teleophthalmology such as decreased human interaction underline the need for well controlled studies before programs are broadly implemented. Local legal and administrative and regulatory requirements also need to be addressed before widespread use. Even once these issues have been resolved it is imperative that patients are engaged in steering planning and assessing satisfaction from early deployment of teleophthalmology programs. It is imperative that patients are engaged in steering planning and early deployment of teleophthalmology programs. This fits nicely into the current climate of the NHS, which supports the policy of “patient-led care” This in turn may drive a cultural change in those commissioning and providing care. A more specific barrier in teleophthalmology, particular in retinal diseases, will be how digital retinal images are read. Reading retinal images is a highly skilled process and trained retinal image readers are expensive and in limited supply worldwide. It will be necessary to automate the process of image evaluation at some point in the workflow pathway, to keep up with the volume of retinal images generated, while maintaining cost-effectiveness and accuracy.⁸⁴