Purpose: This study analyzed the accuracy of the Inoveon Diabetic Retinopathy (DR-3DT) system (Inoveon Corp., Oklahoma City, OK), a scalable evaluation method for the management of diabetic retinopathy using high-quality digital retinal imaging.
Design: An independent, masked, cross-sectional, clinical validation study.
Participants: Two hundred ninety adult patients with diabetes from the Chickasaw Nation's Carl Albert Indian Health Facility in Ada, Oklahoma.
Methods: All participants underwent DRS7 imaging using a Zeiss FF450 fundus camera with images recorded on 35-mm film and a Kodak DCS520 digital camera back. Masked double grading with independent third reader adjudication yielded an Early Treatment Diabetic Retinopathy Study (ETDRS) Final Retinopathy Severity Scale Level (ETDRS Level) and macular edema stage for each eye. The presence of greater-than-or-equal ETDRS Level 53, questionable or definite clinically significant macular edema in either eye, or ungradeable images was defined as a threshold event requiring referral.
Main outcome measures: Accuracy (sensitivity, specificity, predictive values) of the digital system relative to the film "gold standard" on the threshold referral criteria per patient.
Results: All patients with gradeable 35-mm slides from at least one eye were included in this per patient analysis (n = 290). The prevalence of threshold events was 19.3%. The sensitivity of the digital system in detecting threshold events was 98.2% (95% confidence interval [CI], 90.5%-100.0%) and specificity 89.7% (95% CI, 85.1%-93.3%). The positive predictive value was 69.5% and negative predictive value 99.5% for this sample.
Conclusions: When compared with the "gold standard," Inoveon's DR-3DT system provides highly accurate diabetic retinopathy referral decisions. Given their inherent advantages, high-quality digital imaging systems could replace the film "gold standard" as the basis for scalable, accessible, diabetic retinopathy evaluation.