Hyperglycemia, insulin and slower growth velocity may increase the risk of retinopathy of prematurity.

Objective: Strategies to reduce Retinopathy of Prematurity (ROP) have focused primarily on respiratory management. Hyperglycemia (HG) and insulin use, risk factors for adult diabetic retinopathy, as well as growth rates may be modifiable variables useful to reduce ROP.

Study Design: This was a retrospective chart review of all infants born at <30 weeks gestation from 2003 to 2007 who survived to discharge in our neonatal intensive care unit (NICU). All whole-blood glucose values (BG in mg dl(-1)) done in the first 29 days of life were collected for analysis.

Result: BGs were done at least every 3 to 6 h for the first 48 to 96 h of life, then every 6 to 24 h thereafter, as long as infants remained on hyperalimentation. Hyperglycemia was defined as mild (BG 151 to 180), moderate (181 to 210) or severe (>210). Insulin use (given if BG>180 to 210) was also noted for each simultaneous BG. ROP was classified as none, mild (stage 1 to 2) or severe (stage 3 to 4). Growth velocity (g kg(-1) per day), length and head circumference were also analyzed. In all, 372 infants mean (s.d.) gestational age 27.6 (1.4) weeks, mean (s.d.) birth weight 994 (242)g had 18,649 BGs analyzed. 103 (28%) of the infants had mild ROP and 29 (8%) had severe ROP. 137 (37%) of the infants received at least 1 day of exogenous insulin (median days 9, range 1 to 26). Higher cumulative mean BG, more episodes of HG, and more insulin exposure were associated with an increased incidence and severity of ROP. Ordinal logistic regression identified lower gestational age, male gender, fetal growth restriction, slower NICU growth velocity, and higher BG as predictors for severity of ROP. However, insulin use was a stronger predictor than BG, and replaced it in the risk model.

Conclusion: After adjusting for important risk factors, HG and especially insulin use in premature infants may increase the risk of ROP. In addition, slower NICU growth velocity, but not rates of head or length growth, was predictive of ROP.