Diabetic Eye Exams Increased in Community Clinics After Introduction of AI-Based Testing
An analysis of the STATUS program suggests the percentage of patients that received an annual eye exam increased from 65.2% to 72.8% in the clinics utilizing the IDx-DR system.
Austen N Knapp, MD
Credit: American Society of Retina Specialists
Diabetic eye exams increased in community clinics utilizing an artificial intelligence-based screening system, according to new research presented at the 2023 Association for Research in Vision and Ophthalmology (ARVO) Annual Meeting in New Orleans, Louisiana.1
As a result, the investigative team, led by Austen N Knapp, MD, Byers Eye Institute, Stanford Medicine, suggests in-person follow-up may see an increase with the use of artificial intelligence in remote diabetic retinopathy detection.
“Incorporating artificial intelligence into a remote diabetic retinopathy detection workflow may increase in-person follow-up compared to teleophthalmology alone,” investigators wrote.
Artificial intelligence-based testing platforms for diabetic retinopathy have recently been introduced in clinical use, but there is a paucity of outcomes data on their use. Knapp and colleagues performed a retrospective observational study to report real-world outcomes of follow-up and accessibility to ophthalmologic care in patients with diabetic retinopathy using a community, artificial intelligence-based testing program.
The program called the Stanford Teleophthalmology Autonomous Testing and Universal Screening (STATUS) utilizes the FDA-cleared autonomous AI technology, IDx-DR, to detect diabetic retinopathy. It is used in 7 primary-care and endocrinology clinics in the San Francisco Bay area. Adults with type 1 diabetes (T1D) or type 2 diabetes (T2D) who did not have a prior diagnosis of diabetic retinopathy and dilated fundus exam in the prior 12 months were offered inclusion in the program.
A medical assistant in the clinic collected 2 nonmydriatic fundus images of each eye – one 45-degree image centered on the macula and one centered on the optic disc. Investigators graded sufficient images on the presence or absence of referral-warranted more-than-mild diabetic retinopathy (MTMDR), defined as Early Treatment Diabetic Retinopathy Study (ETDRS) level ≥35, and/or diabetic macular edema (DME), in ≥1 eye. Those with MTMDR or an ungradable image were referred for an ophthalmologic exam.
Knapp and colleagues collected patient demographics and characteristics from those with referrable diseases. Additionally, a random sample of patients was given a survey regarding their experience and follow-up.
Over a 12-month period, a total of 1222 patients were screened with IDx-DR and 145 patients (12%) were diagnosed with MTMDR and referred for in-person ophthalmologic evaluation. Of this population, 93 patients (64.1%) saw an ophthalmologist at a median of 60 days after the screening visit.
Data showed 30.3% of patients were referred after an artificial intelligence-based exam followed up at the university-based eye institute, compared with 11.5% of people referred for follow-up after teleophthalmology screening.
Moreover, Knapp and colleagues indicated the percentage of patients receiving an annual eye exam increased from 65.2% to 72.8% in the clinics utilizing IDx-DR. They noted this exceeded the 90th percentile national benchmark (67.89%) for the Healthcare Effectiveness Data and Information Set (HEDIS) quality measure.
References
- Knapp A, Dow E, Chen K, Khan N, Do D, Mahajan V, Mruthyunjaya P, Leng T, Myung D. Real world outcomes from artificial intelligence to detect diabetic retinopathy in the primary care setting: 12-mont experience. Poster presented at The Association for Research in Vision and Ophthalmology Annual Meeting; April 2023; New Orleans, LA.
