This article examines the efficacy of using the DIGIROP online prediction model in estimating the risk of retinopathy of prematurity (ROP) requiring treatment in two Portuguese hospitals. Portuguese screening criteria for ROP mirrors that of the UK (<32 weeks gestation and under 1500g birth weight). As ROP is a leading cause of sight loss, to achieve a high sensitivity many neonates are screened that would have never required treatment. Various prediction models have been used to try and more accurately predict which neonates will go onto develop ROP that requires treatment. DIGIROP uses only gestational age (GA), birth weight and sex to calculate this risk.
The authors report that using DIGIROP risk outcomes to achieve 100% sensitivity meant they would screen children who had a risk of TR-ROP if they scored higher than 0.00016. Of the 431 infants who underwent screening, 174 were excluded as they fell out of the GA range of 24-30 weeks. Therefore, 257 were eligible. Using DIGIROP, 257 would have fallen to 187 infants requiring screening instead. Limitations of using DIGIROP is the aforementioned exclusion of very young neonates and also the older neonates. Two children in the excluded cohort developed TR-ROP highlighting this factor. The authors conclude prediction models have to be tested in the intended cohort as neonatal and ROP risk factors change depending on how advanced the centres are. Portugal, similar to the UK, has an advanced neonatal system.
