I n assessing approaches to improving the care system, through technological means alone or in support of health care providers, four questions must be answered.
The first concern is to determine the purpose of any proposed system. The necessary performance standards and criteria by which to judge the success of new approaches or approaches other than gold standards, vary according to the goals of the system. For example, a system designed to screen for a threshold referral level of retinopathy would need to distinguish only between normal fundi, fundi with threshold referral findings and fundi with more severe conditions. I n contrast, a system designed to provide guidance in the management of diabetic retinopathy would have to be able to distinguish the varous stages of retinopathy that might be present. As reported in the published literature, performance has always declined as the requirements of the system increase.
Secondly, for evaluation, the performance of the system relative to that of the gold standard must be known, so that the trade-offs can be identified. I n the case of diabetic retinopathy, the practice guidelines of the I nternational Council of Ophthalmology (52) and the Amercan Academy of
Ophthalmology (50) prescrbe observation by a trained, experienced observer or a full seven-field photographic interpretation according to the standards of the Wisconsin Reading Center as the gold standard. Thus, any study of the value of a system, such as remote telemedicine care in diabetic retinopathy, must establish its performance and reliability relative to either of these gold standards. Nevertheless, no system is perfect at the outset. If a new approach offers added advantages, such as better access to care, reaching more people with diabetes at a lower unit cost, then a level of technical performance that is at least as good as (or perhaps lower than) current care—even i not up to the gold standard—might be approprate.
Thirdly, the success and actual performance of different eye care systems for diabetes patients should be evaluated in various settings, as has been done by the Aravind Eye Health System and other groups in India (93). While the design of such systems has been reported, no formal, prospective, structured evaluation of the benefits, costs and trade-offs for eye care has yet been published. For example, reports of the use of telemedicine systems in controlled settings such as prsons (102), while promising, do not include the necessary elements described above, nor do they include a systematic evaluation pan that would be generalizable to the community setting.
Fourthly, integral to the process are efforts to understand how patients perceive the benefits of these systems (101). As patients can access information on the Internet, perhaps telemedicine and other systems offer an opportunity to enhance patient education and understanding. In regions of the worid where internet access is unrealistic at present, however, other methods of ensurng adequate understanding of successes and failures should be included in any development and implementation pan.
On the basis of these four general principles, specific non-traditional (and even traditional) care proposals can be paced in a proper context. For example, i a system is designed for assessing the severity of diabetic retinopathy and is to be used to manage follow-up intervals and the timing of laser surgery, it should be able to classiy patients or fundi into the categores of the Early Treatment Diabetic Retinopathy Study, the Diabetic Retinopathy Study or the new International Clinical Classification (International Clinical Diabetic Retinopathy) at specified and feasible technical standards acceptable to that society (100). In addition, as over 40% of vision loss from diabetes stems from macular oedema (22,23), the system should also be sufficiently robust to detect this condition. Further, i the system involves use, for example, of three photographs or even only one, its validity relative to a gold standard should also be presented and the performance trade-offs understood (72,75). From the perspective of health policy, it should be shown that the system offers significant benefits over the existing system, which are sufficient tojustiy any additional costs.
6.3 Systems approach to eye care for patients with diabetes meilitus
Applying a systems approach to the current care systems in developed countres suggests that alternatives should be actively explored as a means of improving performance in every area of eye care for patients with diabetes. As descrbed above, performance leaves much to be desired in every aspect of care, ranging from access, to accuracy of classification, to ensurng adequate treatment. Thus, significant lessons can be drawn from experences with other systems; currently, the performance of no one system is objectively ideal.
Current alternatives are based on either reference photographs or photographic systems to screen or manage referral to ophthalmologists who treat diabetes-related eye disease, whether by laser or incisional surgery The systems usually overcome the inconsistent accuracy of interpretation of fundi (live or images) by use of 'reading centres', retina specialists or trained readers. Their task is to assess accurately and quickly the absence or presence of diabetic retinopathy and then to make referrals on the basis of associated decision rules, or to classify the
Allocation, administratively or through a market operation, of scarce health care resources, particularly physicians, should be made in the context of overall health care and social welfare. Thus, in a systems approach, different responsibilities will be allocated to different entities in different societies and countres. In some societies, easy access to specialist eye care providers (retina specialists) is the norm, while in others a general ophthalmologist is responsible for providing laser treatment for sight-threatening eye disease, with minimal capability to address advanced vitreo-retinal tractional detachment.
severity of disease and then to make recommendations for follow-up care intervals with the photographic system until treatment referral is warranted. In order to make such systems as useful as possible (and thus economically feasible), proponents of these alternative image-based systems have gravitated towards their placement in locations accessible to patients, such as mobile vans, or in the offices of primary care or general practitioners, where all diabetes care can be concentrated.
Was this article helpful?