Retinopathy of Prematurity Scoring System Stratifies Risk

Premature Baby Retinopathy
Retinopathy In Premature Infant. Retinal Fluorescein Angiogram Showing Hypoxia Of The Retinal Periphery. Recolorized, Black And White Image. (Photo By BSIP/UIG Via Getty Images)
Researchers created a prediction model based on gestational age and birth weight alone.

A prediction model for type 1 retinopathy of prematurity (ROP) has high sensitivity and specificity in Asian infants, and can be used to help identify those at risk for the condition, according to a study published in the British Journal of Ophthalmology.

To develop the prematurity-birth weight ROP (PW-ROP) prediction model, investigators conducted a retrospective cohort study that included 1043 infants born premature (530 boys, 513 girls) who underwent ROP screening in a tertiary hospital in Hong Kong between 2006 and 2018. 

The investigators conducted a multivariate logistic regression analysis on type 1 ROP using only parameters available at birth; specifically gestational age (GA) and birth weight (BW). Model performance was evaluated by receiver operating characteristic curve analyses. The team also used an external validation group of 353 infants assessed at another tertiary hospital in Hong Kong between 2014 and 2017 to validate the PW-ROP model.

For the 1043 infants in the study group, the median gestational age was 30 weeks and 1 day, and median birth weight was 1286 g. The predication model achieved an area under curve of 0.902 (sensitivity, 87.4%; specificity, 79.3%). 

The risk was stratified into 3 categories; low (when the PW-ROP score was <0), moderate (when it was between 0 and <300), and high (when it was ≥300). The prediction model requires clinicians to add details of their patients to a formula; 1670-7×GA in days-0.3×BW in grams. 

The researchers developed this after finding that the beta-coefficients of GA and BW in the multivariate logistic regression analysis were -0.0663 and -0.0026, respectively, weights of -0.07 and -0.003 were assigned for GA and BW respectively. ROC analysis shows that -16.7 is the best cut-off volume.

Within the study group, 0.9%, 17.4%, and 50% of infants with PW-ROP scores <0, between 0 and <300, and ≥300, respectively, developed type 1 ROP (P <.001). In the external validation group, the PW-ROP model accurately predicted 95.8% of type 1 ROP (sensitivity, 95.8%; specificity, 74.8%).

Limitations of the study included the retrospective design and inability to evaluate other potential predictive factors, a relatively old mean gestational age in the study group, missing data from infants who died or were transferred to other hospitals, and involvement of only local centers only.


Iu LPL, Yip WWK, Lok JYC, et al. Prediction model to predict type 1 retinopathy of prematurity using gestational age and birth weight (PW-ROP). Br J Ophthalmol. Published online February 12, 2022. doi:10.1136/bjophthalmol-2021-320670