The outcome of retinopathy of prematurity: screening for retinopathy of prematurity using an outcome predictive program

PURPOSE: The purpose of this study was to compare the calculated risk of progression to threshold retinopathy of prematurity (ROP) and risk of an unfavorable structural outcome using the computer program, RM-ROP, with the observed incidence for infants born at Jackson Memorial Hospital (JMH) and to determine how many children would have been treated unnecessarily if the threshold criteria for treatment were lowered on the basis of the clinical findings and RM-ROP risk calculations. DESIGN: Noncomparative interventional case series. PARTICIPANTS: All 292 surviving premature infants weighing 1250 g or less at birth and born at JMH between January 1, 1997, and December 31, 1998, were included in the study. METHODS: Baseline demographic factors and data from sequential ophthalmic examinations were entered into the RM-ROP program for risk calculation. Infants reaching threshold disease received diode laser indirect photocoagulation of the avascular retina. Three-month follow-up was obtained for infants receiving laser treatment. MAIN OUTCOME MEASURES: The development of threshold ROP and an unfavorable structural outcome, defined as a posterior retinal fold or posterior retinal detachment occurring within 3 months of threshold disease. RESULTS: Thirty-eight eyes were diagnosed with threshold ROP, with 18 of 20 subjects having bilateral disease. Three-month posttreatment follow-up was obtained on all 20 children, with 19 having good structural outcomes. Thirty-two percent of eyes (12 of 38) reaching threshold never had a risk estimate greater than 0.10. However, only 6% of eyes (35 of 546) that did not reach threshold ever had a model predicted risk greater than 0.15. All right eyes with zone 1 prethreshold disease, 60% of those with zone 2 stage 2+ disease, and 23% with zone 2 stage 3 disease progressed to threshold ROP. CONCLUSIONS: The similarity between the risk distributions for the Miami and the Multicenter Trial of Cryotherapy for Retinopathy of Prematurity study indicates the similarity in the populations with respect to risk factors identified as important by the model. The Miami data validated the model, with eyes reaching threshold having higher risks than eyes that did not. Actual risk estimates for eyes reaching threshold can be small. Changing the threshold criteria for treatment on the basis of various clinical and computer-generated prethreshold risk levels in our population would have resulted in the unnecessary treatment of many infants who never progressed to threshold disease. In the Miami population, if the model were used to manage an individual subject, close attention would have to be paid to small differences in risk. Although the RM-ROP software program may be a useful tool for following premature infants with ROP, the clinical examination remains the "gold standard."