Abstract
Purpose :
Wilmer Eye Institute is a tertiary ophthalmology referral center and part of Johns Hopkins Medicine (JHM), an integrated healthcare system with 30+ primary care sites. Autonomous AI testing for DED was deployed at several JHM primary care clinics starting in August 2020. We aim to compare the characteristics of patients referred to Wilmer via the standard of care (SOC) and AI pathways.
Methods :
Retrospective analysis of patients with diabetes mellitus who were referred to and examined at Wilmer between August 2020 and September 2022. SOC group: referral for diabetic eye exam ordered by a primary care provider. AI group: referable diabetic retinopathy or non-diagnostic result diagnosed by autonomous AI (IDx-DR, Digital Diagnostics, Coralville, IA). Patient variables included race, best corrected visual acuity (BCVA) of each eye at presentation and insurance coverage. Systemic health variables included hypertension (HTN) status and chronic kidney disease (CKD) status. Variables were compared between the SOC and AI groups using Mann-Whitney U, Student's T, chi-squared, and Fisher's exact tests. Multivariable linear regression analyzed the difference in presenting BCVA between the two groups while controlling for covariates. A P value of <0.05 was considered statistically significant.
Results :
A total of 3745 patients (52.4% F; median age=63.2 years) was included, 3352 in SOC and 393 in AI group. The AI group was more likely to be Black (64.9% vs. 44.4%, p<0.001), have higher Medicare coverage (39.4% vs. 30.9%, p<0.001) and have higher systemic disease burden: HTN (p=0.001) and CKD (p=0.017). Both groups presented with good vision: median BCVA of the better-seeing eye was 0 (Snellen 20/20) for SOC and 0.1 (Snellen 20/25) for AI. The median BCVA of the worse-seeing eye was 0.1 (Snellen 20/25) for SOC and 0.1 (Snellen 20/25) for AI. However, referral group (SOC vs. AI) was not associated with better or worse-seeing eye BCVA.
Conclusions :
Patients who underwent autonomous AI DED testing had higher systemic disease burden and corresponding higher risk for DED. However, there was no difference in presenting BCVA between referral groups. This may indicate that autonomous AI increases health equity for patients who are more at risk for poor visual outcomes by facilitating evaluation before visual damage has occurred.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.