Diabetic retinopathy (DR), a major cause of adult blindness globally, is prevalent in approximately one-third of individuals with diabetes. Early detection and intervention are pivotal in preventing blindness, thereby making regular eye examinations indispensable for patients with diabetes.1,2 The Liverpool Declaration recommends that European nations should aim for a minimum of 80 % screening coverage for retinopathy in their diabetic populations.3

However, the limited availability of ophthalmologists and eye care professionals, particularly in underprivileged and rural regions, poses significant barriers to accessing these essential services. This gap highlights the necessity for an efficient, effective screening tool to pinpoint diabetic individuals at heightened risk for DR to enable patients with diabetes to gain the opportunity to be regularly screened for diabetic retinopathy. Emerging technologies, such as the IDx-DR software, offer promising solutions to enhance screening accessibility and alleviate healthcare system burdens.

In established screening markets like the United Kingdom, The Netherlands, and Scandinavia, the utilization of image-based DR screening has met this objective.4 The automation of image analysis could extend the reach of DR screening to areas currently lacking such facilities, without compromising the quality of care.5 Automated grading systems are now a part of routine clinical practice in The Netherlands and the UK and are under evaluation in various other regions.5,6

The IDx-DR software, an AI-based diagnostic tool endorsed by the US FDA in 2018, identifies DR by analyzing retinal images. This tool, significant for being the first AI-driven medical device to make diagnostic decisions independently of clinician interpretation, has demonstrated accuracy and reliability in detecting DR across multiple settings, including primary care and endocrinology clinics.6,7 Having received CE marking in April 2013, it has been integrated into clinical practice within the EU since then.6,7

Currently, Austria does not employ digital fundus-image-based DR screening as the standard of care, although appointments with ophthalmologists are difficult to obtain without a long waiting time. This scenario presents an opportunity to enhance DR screening's efficiency and accessibility, introducing an automated screening model potentially more scalable and cost-effective than traditional “reading center” models, which rely on human graders.9 Such a shift could significantly reduce the workload on eye care professionals and may improve patient outcomes. Furthermore, it might be a possibility to select patients who need an urgent appointment with an ophthalmologist in order to avoid eyesight threatening consequences.

This study aims to assess the feasibility and efficacy of the IDx-DR software as a DR screening tool in a regular diabetes outpatient setting. We performed a prospectively planned analysis of patients with diabetes who underwent retinal imaging with IDx-DR software during routine visits at the Diabetes outpatient clinic, Klinik Landstraße, Vienna, Austria, from March 2021 to October 2022. The study compared the software's diagnostic accuracy with that of comprehensive eye exams conducted by trained ophthalmologists. Additionally, the safety and acceptability, of the IDx-DR software as a screening tool were evaluated.