Study examines the use of AI–based detection of diabetic retinopathy in the US

News
Article

A recent cohort study revealed low adoption of FDA-approved AI-based diabetic retinopathy detection, with less than 5% of diabetic patients receiving ophthalmic imaging. Researchers emphasize the need for improved awareness, cost-effectiveness, and integration to increase diabetic retinopathy screening rates.

Image credit: AdobeStock/jeremyculpdesign

(Image credit: AdobeStock/jeremyculpdesign)

A team of researchers recently conducted a cohort study to investigate the adoption of artificial intelligence (AI)-based diabetic retinopathy (DR) detection in US healthcare settings, highlighting both the promise and slow uptake of this technology.

In the study, the researchers traced the use of Current Procedural Terminology code 92229, an AI-based screening reimbursement code instituted in January 2021, to evaluate national trends in AI-based DR detection.1

According to the study, published in JAMA Ophthalmology, while FDA-approved systems like LumineticsCore and EyeArt achieve high sensitivity and specificity, less than 5% of diabetic patients in the study received ophthalmic imaging for DR from 2019 to 2023.1 Among these patients, only a small fraction underwent AI-based screening, despite its higher referral rate for OCT imaging compared with traditional methods.

The findings reveal regional and demographic patterns in AI imaging use and underscore the need for increased awareness, cost-effectiveness, and workflow integration to enhance diabetic eye care. Although vision loss from DR is preventable, less than two-thirds of patients with type 1 or type 2 diabetes undergo an annual eye examination.2

The researchers noted that FDA-approved systems such as LumineticsCore (formerly IDx-DR, Digital Diagnostics) and EyeArt (Eyenuk) analyze retinal fundus images for more than mild DR with high sensitivity (87.2% and 96.0%, respectively) and specificity (90.7% and 88.0%, respectively).3,4 Moreover, they found that the use of these systems can help increase detection in the primary care setting among patients with diabetes, while optimizing ophthalmic examinations for those with vision-threatening DR.

“We tracked usage of Current Procedural Terminology code 92229, an AI-based screening reimbursement code instituted in January 2021, to evaluate national trends of AI-based DR detection,” the researchers wrote.

The researchers conducted a retrospective cohort study using the TriNetX federated database, which encompasses more than 107 million patients across 62 healthcare organizations in the US. Records for patients with diabetes were examined from January 2019 to December 2023, with data analysis completed in May 2024.3,4

According to the study, use rates (measured per 100,000 patients with diabetes) of code 92229 were compared with traditional codes for remote eye imaging (92227 and 92228) and imaging modalities from secondary referrals, including fundus photography (92250) and optical coherence tomography (OCT, 92134), with no adjustment to P values for multiple analyses.5,6

The study was exempt from approval by the Stanford University institutional review board, as TriNetX deidentifies all patient information, and it followed STROBE reporting guidelines.7 The researchers conducted statistical analysis in Python version 3.8 (Python Software Foundation).7

Of 4,959,890 patients with diabetes in the TriNetX system from January 2019 to December 2023, 209,673 unique patients (4.2%) received at least one of the targeted codes. The mean (SD) patient age was 64 (16) years, and 2,380,747 patients (48.0%) were female. Among the patients with at least one targeted code, OCT imaging (92134) was used in 168,382 patients (80.3%), fundus photography (92250) in 73,363 patients (35%), and traditional remote imaging (92227 and 92228) in 2135 patients (1%) and 5232 patients (2.5%), respectively. Since 2021, AI imaging (92229) was used in 3440 of 154,136 cases (2.2%), meaning that only 0.09% of all patients with diabetes received this new modality for DR detection.5

The researchers found that in 2021, AI imaging (92229) was used 58 times per 100,000 patients, which increased slightly by 1% to 58.6 times per 100,000 patients by 2023. Although use of traditional remote imaging increased by 185.4% between 2021 and 2023, AI imaging had a higher referral rate (7.74%) to OCT imaging than traditional remote imaging (5.53%). Use of all remote imaging modalities increased by 90.16% from 2021 to 2023 (95% CI, 69.60%-110.72%; P < .001). The study found that more than 80% of the patients who received AI imaging were from the South, a region that made up only 40% of other imaging modalities, and almost half of the patients who received AI imaging were Black, compared with approximately a quarter in other imaging modalities.5

“The FDA’s approval of AI-based systems may be a step forward in DR detection, although adoption is nascent and traditional remote monitoring methods for DR remain more prevalent,” the researchers wrote.

The researchers noted that the cohort study wasn’t population based, but it did reveal that just 4.2% of diabetic patients received ophthalmic imaging for DR over 5 years.

“Imaging rates may be artificially low due to eye care professionals not routinely performing ancillary ophthalmic imaging during general diabetic eye examinations,” they wrote.

While AI-based imaging is leading to more OCT referrals compared with traditional methods, hurdles remain, including cost, awareness, integration, and FDA approval of AI software for imaging devices.6

The researchers noted the potential underuse of AI for DR screening may be further pronounced, given that OCT was conducted more frequently.

“Broader adoption may require support to help physicians and organizations integrate these systems into existing workflows,” they wrote.

Programs such as the Stanford Teleophthalmology Autonomous Testing and Universal Screening program highlight AI’s potential in improving DR detection and the importance of streamlined workflows, close collaboration between primary care and ophthalmology, and patient-friendly appointment scheduling.

“These findings support further evaluation of imaging practices to develop targeted strategies for improving diabetic eye imaging rates and patient outcomes,” they concluded.

References
1. Shah SA, Sokol JT, Wai KM, et al. Use of Artificial Intelligence–Based Detection of Diabetic Retinopathy in the US. JAMA Ophthalmol. Published online October 31, 2024. doi:10.1001/jamaophthalmol.2024.4493
2. Diabetes Workgroup. Increase the proportion of adults with diabetes who have a yearly eye exam — D-04. Healthy People 2030, Office of Disease Prevention and Health Promotion. Accessed February 24, 2024. https://health.gov/healthypeople/objectives-and-data/browse-objectives/diabetes/increase-proportion-adults-diabetes-who-have-yearly-eye-exam-d-04
3. Abràmoff MD, Lavin PT, Birch M, Shah N, Folk JC. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. NPJ Digit Med. Published online August 28, 2018. doi:
4. Lim JI, Regillo CD, Sadda SR, et al. Artificial intelligence detection of diabetic retinopathy: subgroup comparison of the EyeArt system with ophthalmologists’ dilated examinations. Ophthalmol Sci. 2022;3(1):100228. doi:10.1016/j.xops.2022.100228
5. AMA releases 2021 CPT code set. American Medical Association. Accessed February 25, 2024. https://www.ama-assn.org/press-center/press-releases/ama-releases-2021-cpt-code-set
6. Chen EM, Chen D, Chilakamarri P, Lopez R, Parikh R. Economic challenges of artificial intelligence adoption for diabetic retinopathy. Ophthalmology. 2021;128(3):475-477. doi:10.1016/j.ophtha.2020.07.043
7. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies | EQUATOR Network. Equator-network.org. Published 2014. Accessed November 11, 2024. http://www.equator-network.org/reporting-guidelines/strobe/
Recent Videos
© 2024 MJH Life Sciences

All rights reserved.