Abstract
Purpose :
To test feasibility and accuracy of the “remote diagnosis model” of tele-ophthalmic retinal imaging (remote diagnosis) as a clinical screening method to identify referable macular pathology compared to the “gold standard”.
Methods :
Prospective, non-randomized study on 159 patients (318 eyes) conducted in Duke endocrinology clinic with high prevalence of diabetic retinopathy, and Duke assisted living centers with high prevalence of age-related macular degeneration. All patients underwent remote screening (un-dilated imaging using iFusion by Optovue, performed by trained, non-expert imager) and gold standard exam. Remote images were graded by masked readers for interpretability and presence of predetermined referable retinal pathology, with each eye graded independently and being its own control. Retinal pathology was defined as referable to retina specialist, requiring further intervention or follow up; and incidental findings referable to general ophthalmologist or optometrist for annual follow up
Results :
22% of eyes were determined to require referral to the retinal specialist by gold standard examination. Remote diagnosis image interpretability was significantly better by OCT relative to CFP (97% vs. 73% respectively). Remote diagnosis imaging had high diagnostic accuracy in identifying referable macular pathology: OCT was 94% sensitivity and 94% specificity; CFP was 94% sensitivity and 63% specificity. There was substantial agreement in referable vs. non-referable retinal pathology between the gold standard and OCT (kappa=0.76- 0.047, p<0.001) and with CFP (kappa=0.73-0.063, p<0.001). However, remote diagnosis imaging demonstrated low sensitivity in identifying routine incidental findings (22%). Patient satisfaction surveys revealed that majority of participants (77%) preferred remote imaging over the standard of care examination.
Conclusions :
Our data show that novel, remote ophthalmic technologies at point of service, harbor diagnostic information concurrent to the one obtained by the gold standard and could promote accurate referral and timely treatment. Remote diagnosis imaging demonstrates very high level of agreement with the retinal specialist in terms of identifying referable macular pathology requiring urgent treatment or close monitoring, but is not sensitive enough at identifying routine abnormalities, and this approach cannot replace the general ophthalmic examination.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.