DIGIROP efficacy for detecting treatment-requiring retinopathy of prematurity in a Portuguese cohort

Abstract

Background/Objectives: To determine the efficacy of the DIGIROP model in detecting treatment-requiring retinopathy of prematurity (TR-ROP) in a Portuguese cohort. Subjects/Methods: Multicentre, retrospective cohort study of all consecutive preterm infants who underwent ROP screening from April 2012 to May 2019 in two neonatal units. Gestational age (GA), birth weight (BW) and sex were inserted in the DIGIROP platform. The optimal cut-off point to achieve 100% sensitivity was calculated. Area under the receiver operating characteristic curve (AUC) was calculated. Results: Of the 431 infants who underwent ROP screening, 257 were eligible for DIGIROP analysis and 174 infants were excluded for having a GA outside the range 24–30 weeks imposed by the DIGIROP algorithm. Median GA was 29 weeks (range 24–30) and BW was 1060 g (range 408–2080). Twenty-tree infants (8.9%) developed TR-ROP. The highest risk obtained for TR-ROP was 0.5404 (95% CI 0.4343–0.6616) with a median achieved risk of 0.0938 (range 0.0016–0.5404). The optimal cut-off point to achieve 100% sensitivity on TR-ROP was 0.0016. The number of infants receiving ROP examinations would have been reduced from 257 to 187 infants (−27.2%) if the model was applied. Conclusions: In our cohort, of 257 infants, the optimal cut-off point to achieve 100% sensitivity for TR-ROP was 0.0016 with moderate accuracy in the AUC (0.70). The number of infants requiring screening would have decreased 27.2% if the model was applied. It is essential that algorithms continue to be tested in different populations, especially in cohorts that include both younger and older GA infants.