Risk analysis of prethreshold retinopathy of prematurity.

OBJECTIVE: To present a new multifactorial algorithm to integrate important risk factors for unfavorable retinal outcome in eyes that reached prethreshold retinopathy of prematurity (ROP) in the Cryotherapy for Retinopathy of Prematurity (CRYO-ROP) study. A computerized risk model (RM-ROP2) was developed from this algorithm to identify high-risk prethreshold eyes for enrollment in the Early Treatment for Retinopathy of Prematurity randomized trial.

METHODS: Data were analyzed from 613 eyes (1 eye per infant) in the natural history cohort of the Multicenter Trial for Cryotherapy for Retinopathy of Prematurity. These eyes were selected from infants in whom 1 or both eyes progressed to prethreshold ROP. Eyes that progressed to threshold ROP and were randomized to cryotherapy were excluded from this study, but control eyes that reached threshold ROP were included. The course of ROP for 1 prethreshold eye for each infant was tracked until the evaluation of its structural outcome at 3 months' postterm. Tables present structural outcome by selected risk characteristics. A multiple logistic risk model is used to summarize the combined effect of all of these known prognostic risk factors as they relate to structural outcome.

RESULTS: Eyes were classified by predicted outcome into 10 risk categories, lowest to highest. Both the observed and predicted outcomes in each category showed an increasingly unfavorable outcome when viewed from lowest to highest risk. Prethreshold ROP eyes were then divided into 2 groups, high-risk, eyes (risk, 0.15-1.0) and low-risk eyes (risk, 0.15). High-risk eyes had an unfavorable outcome of 36% at 3 months' postterm; whereas, only 5% of the low-risk eyes had an unfavorable outcome.

CONCLUSION: The model effectively identifies prethreshold ROP eyes that have a relatively high risk and eyes that have a lower risk of an unfavorable structural outcome at 3 months.