Later postnatal weight gain patterns and risk of severe retinopathy of prematurity: a secondary analysis of Shiraz ROP study

Abstract Background Poor early postnatal weight gain is an established predictor of severe retinopathy of prematurity (ROP); however, the patterns and clinical relevance of growth beyond the first few postnatal weeks remain uncertain. Objective To compare longitudinal weight-gain patterns between infants with spontaneously regressing ROP and those who developed treatment-requiring ROP, using both chronological postnatal age and postmenstrual age. Methods This retrospective cohort study was a secondary analysis of the Shiraz ROP registry and included preterm infants screened between January 2020 and January 2024. Mean daily weight gain was calculated from serial weight measurements obtained between postnatal weeks 4 and 30. Longitudinal growth patterns were analyzed using generalized estimating equations with robust standard errors clustered by infant. Time was modeled using restricted cubic splines, and the primary models were adjusted for birth weight and gestational age. Treatment was evaluated both as a binary variable and according to modality: laser photocoagulation, intravitreal anti-VEGF therapy, or combined treatment. Analyses were conducted using both chronological postnatal week and postmenstrual age. Mixed-effects, inverse-probability-of-treatment-weighted, and additionally adjusted models were used as sensitivity analyses. Results The primary analysis included 7,746 infants with 21,086 eligible visit records. Weight-gain patterns differed significantly between infants requiring treatment and those not requiring treatment on both the postnatal-week axis (group-by-time interaction: Wald χ²=54.5, df = 3, p < 0.001) and the postmenstrual age axis (Wald χ²=45.5, df = 3, p < 0.001). Based on chronological postnatal age, treated infants had slower adjusted weight gain during approximately weeks 4–10 (week 4 difference: −8.6 g/day; 95% CI, − 13.0 to − 4.2), followed by convergence during weeks 11–17 and an apparent later reversal. However, estimates beyond week 21 were imprecise because of sparse data. When infants were aligned by postmenstrual age, treated infants showed a persistent but gradually narrowing weight-gain deficit from postmenstrual age 34 to 46, which was no longer statistically significant by weeks 48–50. Modality-specific analyses showed similar patterns in the different treatment groups. Conclusions Infants who developed treatment-requiring ROP demonstrated an independent early deficit in postnatal weight gain that diminished with increasing physiological maturity. Postmenstrual age provided a more stable and clinically interpretable representation of longitudinal growth than chronological postnatal age. Longitudinal weight monitoring may complement, but should not replace, established ophthalmologic screening criteria.