To the extent that diabetic patients receive most of their health care in primary care physicians' offices, a telemedicine system has the potential to provide an accurate and cost-effective means of improving screening rates. Retinal photographs can be taken by non-ophthalmologists (e.g., nurses, primary care physicians, optometrists, or trained technicians), and the photographs can then be converted to digital images and sent electronically for evaluation to a qualified ophthalmologist with specialty training in diabetic retinopathy assessment.
The main objective of this study is to evaluate the technical quality of a telemedicine system for assessing the retinopathy status of patients with diabetes. The main hypothesis to be addressed is that the use of a telemedicine system results in the same diagnosis as does the process of assessing retinopathy status using seven-field, 35 mm, mydriatic photographs, generally considered to be the "gold-standard" in retinal assessments.
This pilot study was conducted at the Ann Arbor VA Medical Center and at the University of Michigan Kellogg Eye Center. The research consisted of a prospective, cohort design, comparing the diagnostic evaluation resulting from a trained retinal grader's assessment of the retinopathy status of diabetic patients using mydriatics (pupil dilation) and seven-field, 35 mm photographs (the current gold standard for diagnostic assessments) with two alternative telemedicine assessments: 1) that from the grader's and an ophthalmologist's review of mydriatic digital photographs; and 2) that from the grader's and an ophthalmologist's review of non-mydriatic digital photographs. Analysis of accuracy will be performed by calculating sensitivity and specificity of overall level of diabetic retinopathy, as well as of the components that make up the overall diagnosis, including microaneurysms, retinal hemorrhage, hard exudates, soft exudates, etc.
249 patients from the Ann Arbor VAMC were enrolled with photos taken. Agreement between 35-mm slide and digital assessments was higher for the trained grader (kappa of 0.57 for mydriatic assessments) than for the two fellows (kappas of 0.42 and 0.40). Non-mydriatic digital assessments had slightly lower kappas. Specificity was high (over 0.99) for all three readers, but sensitivities were low-0.51 for assessments by the trained grader and 0.37 and 0.38 for those by the fellows.
Telemedicine is being increasingly used, both within and outside VHA, to improve patients' access to retinopathy screening services, which, in turn, can help to initiate early treatment and improve outcomes. However, the findings from this study suggest limitations with the technology, which must be recognized when designing and implementing telemedicine screening systems.
- Lowery JC, Kim HM. Assessment of diabetic retinopathy via telemedicine. Poster session presented at: American Telemedicine Association Annual Meeting; 2005 Apr 1; Denver, CO.
- Kim HM, Lowery JC, Kurtz R. Assessment of retinopathy via telemedicine. Paper presented at: VA HSR&D National Meeting; 2005 Feb 1; Baltimore, MD.