Binocular Ophthalmoscopy versus Smartphone images for Retinopathy of Prematurity (ROP) Telemedicine in Latinamerica: a preliminary report of diagnostic accuracy.

Abstract

Purpose : To evaluate the diagnostic accuracy of the Binocular Indirect Ophthalmoscopy (BIO) versus Smartphone Telemedicine in ROP babies.

Methods : A cross-sectional study was designed for a comparative evaluation between clinical and telemedical diagnosis accuracy, performed in eyes with ROP, evaluating in different centers of Buenos Aires (Argentina), from March to December 2020, analyzing the first 100 results from a validation protocol of 400 cases.
Two Latinamerican Regional Teams were built. Team 1: Clinical BIO Screening (4 pediatric ophthalmologists), who performed clinical evaluation and imaging. Team 2: Tele-ophthalmologists Team (3 pediatric ophthalmologists, 2 pediatric retinal experts). Unit of analysis: clinical funduscopy and smartphone images of the retina evaluation by telemedicine.
BIO screening was performed as usual. For smartphone imaging, the imagers used the helmet device for hands-free digital indirect fundoscopy through a smartphone, simultaneously performing scleral depression or globe rotation in order to reach the far periphery of the ocular fundus (Figure 1). The SP ROP (Panamerican Society of Retinopathy of Prematurity) algorithm for decision making for Referral and Treatment Warranted ROP was followed. A standard consensus was agreed, to compare ophthalmoscopy and telemedicine results, using percentage agreement. Separated Google Forms for each participant were used. Team 1: perinatal data and ROP screening evaluation for each ID; Team 2: videos and pictures of funduscopy were presented to participants who completed another Form with their data evaluation.

Results : Among all the first 100 eyes registered, the average gestational age was 30.9 weeks; and the average birth-weight was 1517.1 grams. Regarding the SP ROP algorithm that evaluates NO PLUS (n:80) / PLUS (n:20), corresponding to consensual “Posterior Vascular Changes'', we found a preliminar accuracy of NO PLUS: 85.4% (n:24) and PRE PLUS: 58.3% (n:15). For ROP stage 3: 84.3% (n:8); Type 1 ROP: 83.7% (n:20); TYPE 2 ROP: 81.2% (n:37). The complete results were presented in Table 1.

Conclusions : A preliminary high accuracy was found using the SP ROP algorithm and smartphone images in ROP telemedicine triage. A more complete data analysis is needed for the final conclusions of this regional study.

This is a 2021 Imaging in the Eye Conference abstract.