The main outcome measurements were timing and risk factors of ROP recurrence after IVR monotherapy. Any recurrence in one or both eyes was defined as a ROP recurrence. Descriptive statistics were used to characterize the study population, including means and SDs or median value (range) for continuous variables, frequencies and proportions for categorical variables. Time of recurrence and related fundus characteristics were analyzed according to serial retinal examinations after IVR by the RetCam wide-angle fundus imaging system and binocular indirect ophthalmoscopy. Time-varying recurrence hazard rate was estimated using the hazard function of life-table analysis. Recurrence-free survival was estimated by the Kaplan-Meier method with earliest recurrence needing a retreatment as the endpoint. Any difference in recurrence-free survival was evaluated with a log-rank test. Patients who had involution of ROP without recurrence at the last visit were censored for the purpose of data analysis. Potential risk factors obtained from clinical charts were the patient's sex, birth weight (BW), gestational age (GA), postmenstrual age (PMA) of IVR treatment, clock hours (30° sectors) of preexisting retinal neovascularization, zone of ROP, patient number of comorbidities including anemia (hemoglobin <110 g/L), sepsis (positive blood culture), congenital heart disorder, intraventricular hemorrhage, neonatal respiratory distress syndrome (NRDS), bronchopulmonary dysplasia (BPD), asphyxia, and the number of patients requiring oxygen support before IVR, and oxygen requirement after IVR treatment before involution. These clinical factors were included in univariate analysis, excluding the items occurring less frequently (n ≤ 10). A logistic multivariate regression backward stepwise model was constructed to identify independent risk factors from variables demonstrating statistically significant associations (P < 0.05) with risks of ROP recurrence by univariate analysis. Statistical analyses were performed using SPSS 22.0 for Windows (SPSS, Inc., Chicago, IL, USA). A P < 0.05 (2-sided) was considered significant for all tests.