Diabetic Retinopathy Screening in a Primary Care Setting Using Non-Mydriatic Photography and Automated Retinal Image Analysis Improves Compliance with Follow-Up Ophthalmic Care

Abstract

Purpose : Vision loss from diabetes can be effectively prevented with appropriate retinal screening, currently recommended at least once a year for most patients. However, less than half of patients with diabetes adhere to these guidelines. We hypothesized that point-of-care screening leveraging artificial intelligence-based detection methods could increase the rate of compliance with recommended retinal surveillance in patients with diabetes when administered during primary care visits.

Methods : We performed a prospective cohort study of adults ages 18 or older with diabetes seen in our medical institution’s primary care clinic from 3/1/18 to 8/31/18. Participants were screened using non-mydriatic fundus photography and automated diabetic retinopathy (DR) screening software. Those with positive or inconclusive screening results were referred for comprehensive evaluation and the rate of compliance among this group was determined. All images were over-read by trained retina specialists. Historical rates of compliance were determined by retrospective analysis of adults with diabetes and compared to the compliance rate of prospective study participants.

Results : 148 adults with diabetes were included in the prospective study cohort. 11 participants (7.4%) had a positive screening result for referable diabetic eye disease, 19 participants (12.8%) had a positive screening result for vision threatening diabetic eye disease, and 30 participants (20.3%) had an inconclusive screening result. There was 100% agreement between software-grading and human-grading among eyes with no retinopathy. The rate of compliance with follow-up eye care in those who received a positive or inconclusive screening result was 53.3% compared to a historical compliance rate of 18.7% (p<0.0001, chi-squared).

Conclusions : Advances in fundus photography and machine-learning based image analysis have eased access to DR screening. The implementation of an automated DR screening system in our institution's primary care clinic reduced referrals for those with no retinopathy and increased the rate of compliance with follow-up eye care recommendations in those who had positive or inconclusive screening results. These findings suggest that implementation of automated screening could effectively increase compliance with ophthalmic care among those with referable DR.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.