A risk scoring model to predict progression of retinopathy of prematurity for Indonesia

Johanes Edy Siswanto, Asri C. Adisasmita, Sudarto Ronoatmodjo, Peter H. Dijk, Arend F. Bos, Florence Manurung, Pieter J.J. Sauer

Research output: Contribution to journalArticlepeer-review


Introduction Retinopathy of prematurity (ROP) is a serious eye disease in preterm infants. Generally, the progression of this disease can be detected by screening infants regularly. In case of progression, treatment can be instituted to stop the progression. In Indonesia, however, not all infants are screened because the number of pediatric ophthalmologists trained to screen for ROP and provide treatment is limited. Therefore, other methods are required to identify infants at risk of developing severe ROP. Objective To assess a scoring model’s internal and external validity to predict ROP progression in Indonesia. Method To develop a scoring model and determine its internal validity, we used data on 98 preterm infants with ROP who had undergone one or more serial eye examinations between 2009 and 2014. For external validation, we analyzed data on 62 infants diagnosed with ROP irrespective of the stage between 2017 and 2020. Patients stemmed from one neonatal unit and three eye clinics in Jakarta, Indonesia. Results We identified the duration of oxygen supplementation, gestational age, socio-economic status, place of birth, and oxygen saturation monitor setting as risk factors for developing ROP. We developed two models—one based on the duration of supplemental oxygen and one on the setting of the oxygen saturation monitor. The ROP risk and probabilistic models obtained the same sensitivity and specificity for progression to Type 1 ROP. The agreement, determined with the Kappa statistic, between the ROP risk model’s suitability and the probabilistic model was excellent. The external validity of the ROP risk model showed 100% sensitivity, 73% specificity, 76% positive predictive value, 100% negative predictive value, positive LR +3.7, negative LR 0, 47% pre-test probability, and 77% post-test probability. Conclusion The ROP risk scoring model can help to predict which infants with first-stage ROP might show progression to severe ROP and may identify infants who require referral to a pediatric ophthalmologist for treatment.

Original languageEnglish
Article numbere0281284
JournalPloS one
Issue number2 February
Publication statusPublished - Feb 2023


Dive into the research topics of 'A risk scoring model to predict progression of retinopathy of prematurity for Indonesia'. Together they form a unique fingerprint.

Cite this