Validation of the DIGIROP algorithm in identifying preterm infants at risk for developing retinopathy of prematurity in a Belgian NICU

Abstract

Purpose

Retinopathy of prematurity (ROP) is a blinding disease, however largely preventable by timely detection of severe ROP and treatment when required. Clinical use of the DIGIROP decision support tool (DIGIROP-screen and DIGIROP-birth) can help detecting neonates at high risk of sight-threatening severe ROP needing treatment and to reduce unnecessary screening exams in low-risk infants. External validation is necessary before the tool can be used in clinical decision-making.

Methods

Retrospective cohort analysis of all preterm infants who were screened for ROP at the University Hospitals Ghent Belgium from January 1, 2020, to December 31, 2022. Validation of the DIGIROP decision support tools was the primary outcome variable. In a secondary analysis pre-, peri-, and postnatal characteristics were compared in different cohorts.

Results

311 infants were eligible for routine ROP-screening. Infants with a (gestational age) GA beyond 30 weeks (N = 80) and those who died (N = 19) were excluded resulting in 212 infants eligible entered in DIGIROP-birth. 112 infants did not need to be screened according to DIGIROP-birth. This resulted in 94 infants entered in DIGIROP-screen. These infants had a median GA of 27 weeks, a median birth weight (BW) of 892.5 g, and 62.5% were boys. DIGIROP birth showed a sensitivity of 100%. Most important risk factors for ROP across all subgroups are GA, BW.

Conclusions

The DIGIROP decision support tool demonstrated very high performance in our setting. However, multicenter prospective validation studies with large cohorts should confirm our findings before the use of the model can be generalized in Western NICU's with similar levels of care.