Purchase this article with an account.
Jennifer Lindsey, Aaron Noll, Amy Chomsky; Efficacy of Teleretinal Imaging for Detection of Nondiabetic Eye Disease. Invest. Ophthalmol. Vis. Sci. 2018;59(9):5237.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Teleretinal Imaging (TRI) has been used successfully by the Veterans Health Administration (VHA) and other groups worldwide to screen for diabetic retinopathy (DR). The ability of TRI to detect other eye diseases has not been well validated. We performed a retrospective study to evaluate the detection of nondiabetic eye disease by TRI at the Veterans Affairs, Tennessee Valley Health System (VA TVHS).
Retrospective chart review was performed for 1755 patients at VA TVHS who underwent TRI. De-identified data were collected including, but not limited to: demographic factors (age, sex, race), date of TRI, findings on screening exam, and findings on face-to-face clinical examination following screening. Descriptive analysis was performed, in addition to specificity, sensitivity, and positive and negative predictive value for certain data points.
Of the 1675 patients who underwent TRI and had adequate images for assessment, 748 (44.7%) had a future visit to the Nashville VA Eye Clinic, allowing for confirmatory diagnoses to be made. A posterior segment finding other than DR was noted in the screening result in 162 (9.7%) patients. Of these, 95 patients had a face-to-face exam to allow correlation of physical and screening findings. Commonly identified TRI findings in this cohort were glaucomatous optic nerve changes (GC) (25 or 15.4% of TRI patients with non-DR findings), macular degeneration (AMD) (18, 11.1%), and epiretinal membrane (ERM) (7, 4.3%). Among the cohort with screening findings and a follow-up exam, specificity was highest for ERM (94.6%), was 82.5% for AMD, and 75.9% for GC. Negative predictive value was over 85% for each disease category. Sensitivity of TRI was low for GC (37.5%), ERM (33.3%), and AMD (33.3%). Positive predictive value was also low (below 25%) for all three categories.
In our study, TRI was found to have high specificity and negative predictive value, but poor sensitivity for detecting nondiabetic posterior segment eye disease including glaucomatous optic nerve changes, AMD, and ERM. Limitations of the study include its retrospective nature and its restriction to a single VA facility. Other institutions have reported greater success with TRI in screening for nondiabetic eye disease and greater agreement between TRI and clinical examination findings. Further study is needed to determine how TRI can be optimized to detect nondiabetic pathology.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.
This PDF is available to Subscribers Only