Accuracy and reliability of teleophthalmology for diagnosing diabetic retinopathy and macular edema: a review of the literature.

Published

Journal Article (Review)

Using teleophthalmology for the delivery of routine eye care for patients with diabetes mellitus is becoming an increasingly common practice. Paramount in the consideration of any new diagnostic test is an analysis of its diagnostic accuracy and reliability and how that compares with conventional care. This review summarizes existing data on the diagnostic accuracy and reliability of teleophthalmology and conventional clinic-based eye care for detecting diabetic retinopathy and diabetes mellitus. The sensitivity of ophthalmoscopy for detecting diabetic retinopathy performed by eye care clinicians has varied widely, with point estimates ranging from 0% to 96%. Alternatively, specificity has been universally high. The sensitivity of teleophthalmology for detecting diabetic retinopathy has been shown to be comparable, if not better, than clinic-based examinations. Sensitivity values have ranged from 50% to 93%. The specificity of teleophthalmology, like clinic-based examinations, has been consistently high. High levels of diagnostic reliability, analyzed by both simple agreement and kappa values, have been found between ophthalmoscopy and teleophthalmology for detecting and classifying diabetic retinopathy. Evaluating the accuracy of macular edema detection requires the use of dual gold standards, the clinical examination using slit-lamp biomicroscopy and stereoscopic photography. Teleophthalmology, compared with both gold standards, has, overall, been a highly sensitive and specific test. Reliability studies that compared the two gold standards with one another have found moderate to substantial levels of agreement. Based on existing data, teleophthalmology appears to be an accurate and reliable test for detecting diabetic retinopathy and macular edema.

Full Text

Duke Authors

Cited Authors

  • Whited, JD

Published Date

  • February 2006

Published In

Volume / Issue

  • 8 / 1

Start / End Page

  • 102 - 111

PubMed ID

  • 16472057

Pubmed Central ID

  • 16472057

International Standard Serial Number (ISSN)

  • 1520-9156

Digital Object Identifier (DOI)

  • 10.1089/dia.2006.8.102

Language

  • eng

Conference Location

  • United States