Risk analysis and an alternative protocol for reduction of screening for retinopathy of prematurity.

PURPOSE To determine whether a multivariate risk model can select infants with low-risk eyes for an alternative protocol that reduces retinopathy of prematurity (ROP) screening without loss of effectiveness. METHODS This was a retrospective, nonrandomized, comparative study. We assigned 712 eyes of 357 premature infants of 401-1,250 g birth weights as high or low risk for prethreshold or threshold ROP using a risk model with variables of birth weight, gestational age, multiple birth, race, and gender. Using simulations, infants with high-risk eyes (p > or = 0.15) were screened conventionally, while those with low-risk eyes were screened with the 35q3 protocol (initial examination at 35 weeks postmenstrual age followed by screening every 3 weeks, with increased or decreased intervals based on ROP severity). The resultant reduction in ROP screening and the delay to detection of severe ROP were calculated. RESULTS The 35q3 protocol reduced the mean number of eye examinations per infant in the overall population by 13.4% (p = 0.0051). No eyes had a delay to the detection of threshold or type 1 zone 1 ROP. Of a total of 57 type 1 ROP eyes, 2 eyes with type 1 zone 2 ROP were delayed to detection by 1 week. As our study was done prior to the Early Treatment of ROP study, the 2 eyes were observed and regressed without progression to threshold ROP. CONCLUSIONS A risk-based alternative screening protocol increased the efficiency of ROP screening for infants of 401-1,250 g birth weight without an apparent loss of effectiveness. This investigational approach requires further validation by multicenter studies.