Abstract
Purpose :
Screening rates for diabetic retinopathy (DR) in the United States are lower than recommended standards, especially among low-income adults. We performed a cost-effectiveness analysis of Automated Retinal Image Analysis Systems (ARIAS)-based DR screening in a low-income, primary care patient population.
Methods :
We utilized data from a recent prospective study of 148 adults with diabetes at our institution’s primary care clinic in which ARIAS-based DR screening was implemented using non-mydriatic, two-field digital retinal photography (CR-2, Canon) coupled to EyeArt 2.0 software (Eyenuk, Inc.). A retrospective chart review of 974 adults with diabetes from the same clinic who were seen in the 9 months just prior to ARIAS-implementation served as the control group. Cost-effectiveness modeling compared the net Medicare reimbursements minus costs for screening, diagnosing, and treating DR in a simulated cohort of primary care patients using data from the prospective and retrospective groups, data from the Centers for Medicare and Medicaid Services, precedent from the published literature, and expert opinion.
Results :
The prospective and retrospective groups were comparable in terms of age, sex, race, and most recent hemoglobin A1C. Implementing ARIAS led to a 60% reduction in total referrals to the eye clinic and a nearly 3-fold increase in percentage of patients completing the referral. In cost-effectiveness modeling, accounting for costs and reimbursements associated with screening, diagnosing, and treating DR, ARIAS DR screening generates a net-positive $194,850 in the first year, compared to net-positive $1,732,440 by the fifth year. In contrast, referral of all diabetic patients for eye screening generates net-positive $90,8901 in year one compared to net-positive $684,147 by year five.
Conclusions :
Primary care-based ARIAS DR screening increases selective referral of patients with DR requiring treatment, reduces referrals for patients with minimal or no DR, and results in increased compliance rates. Cost-effectiveness modeling showed an increase in reimbursements minus costs with ARIAS implementation compared to current practice. Utilizing these advantages results in better allocation of healthcare resources for screening, diagnosis, and treatment of DR in low-income patients.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.