1810-LB: Diagnostic Accuracy of Automated Pupillometry for Diabetic Retinopathy Screening

Introduction: Diabetic retinopathy (DR) is a major and preventable cause of blindness. Although regular fundoscopic examination (FE) is recommended to detect DR, screening coverage remains low in Latin America, the USA, and other regions, supporting the need for new screening tools. Because pupillary reactivity to a short light stimulus is decreased in patients with DR, this study aimed to assess the diagnostic accuracy of automated pupillometry (AP), an inexpensive tool, for DR screening in type diabetes mellitus (T2DM) patients.

Methods: Diagnostic accuracy study performed prospectively in 93 adults with T2DM (186 eyes) referred for FE from three public primary care health centers in Santiago, Chile. Patients underwent a clinical evaluation, AP assessment, and FE during the same visit. AP was performed to obtain three valid trials by eye with a clinical pupilometer. For DR diagnosis, pupil-dilated eye fundus images were taken with a retinal camera and later informed by an ophthalmologist. Patients and AP/FE operators were blind to FE/AP data. The diagnostic accuracy of AP for DR was assessed by the area under the curve (AUC), sensitivity, specificity, and predictive values (PV).

Results: 50% females, 64±10 years old, 68% hypertension, 34% insulin users. Frequency of DR was 22%. All AP-derived parameters were lower in those with DR (p-value<0.001). Of all eye assessments, 75% had increased the inverse of basal pupillary diameter (BPD^-1), 12% had a decreased pupillary constriction (PC) and 51% had both parameters altered. The AUC, sensitivity, specificity, negative PV, and positive PV for DR were, respectively: 0.778, 95%, 32%, 27%, and 96% for BPD^-1 and 0.765, 43%, 96%, 76%, and 86% for PC. Only 4% of patients with a conserved BPD^-1 and 7% of those with a conserved BPD^-1 or PC had DR.

Conclusion: This study proves for the first time that AP is a viable screening tool for DR, with a sensitivity and negative PV of about 95%. Further studies are needed to confirm if AP avoids FE or helps triage diabetics waiting for FE.