PURPOSE: We aimed to externally validate the performance of new screening criteria for retinopathy of prematurity (ROP) developed in the Postnatal Growth and Retinopathy of Prematurity (G-ROP) study among a Japanese cohort. DESIGN: Validation of screening criteria. METHODS: We reviewed premature infants screened for ROP between September 2009 and May 2017 at a single institution. The G-ROP criteria, except hydrocephalus, were applied as a prediction model for infants with both a known outcome of ROP and serial measurements of weight gain. We assessed sensitivity and specificity for treatment-requiring ROP, and reduction in the number of infants who receive ROP screening and in the number of retinal examinations. RESULTS: Of 692 premature infants screened for ROP, 537 had information of ROP outcome and weight gain. In this cohort, 81 infants required treatment for ROP; in 218 infants, ROP regressed spontaneously; and 238 infants did not develop any ROP. The G-ROP model reached a sensitivity of 100% (95% confidence interval [CI], 95.4%-100%) and specificity of 28.9% (95% CI, 24.9%-33.2%). No infants required any treatment for ROP before the date of risk determination. The number of infants requiring screening and the number of examinations would have been reduced by 24.5% and 12.9%, respectively. CONCLUSIONS: This is the first validation study of the G-ROP criteria in a developed country other than North America. The criteria demonstrated high sensitivity in this Japanese cohort, even though the criterion of hydrocephalus was excluded.
PURPOSE: We aimed to externally validate the performance of new screening criteria for retinopathy of prematurity (ROP) developed in the Postnatal Growth and Retinopathy of Prematurity (G-ROP) study among a Japanese cohort. DESIGN: Validation of screening criteria. METHODS: We reviewed premature infants screened for ROP between September 2009 and May 2017 at a single institution. The G-ROP criteria, except hydrocephalus, were applied as a prediction model for infants with both a known outcome of ROP and serial measurements of weight gain. We assessed sensitivity and specificity for treatment-requiring ROP, and reduction in the number of infants who receive ROP screening and in the number of retinal examinations. RESULTS: Of 692 premature infants screened for ROP, 537 had information of ROP outcome and weight gain. In this cohort, 81 infants required treatment for ROP; in 218 infants, ROP regressed spontaneously; and 238 infants did not develop any ROP. The G-ROP model reached a sensitivity of 100% (95% confidence interval [CI], 95.4%-100%) and specificity of 28.9% (95% CI, 24.9%-33.2%). No infants required any treatment for ROP before the date of risk determination. The number of infants requiring screening and the number of examinations would have been reduced by 24.5% and 12.9%, respectively. CONCLUSIONS: This is the first validation study of the G-ROP criteria in a developed country other than North America. The criteria demonstrated high sensitivity in this Japanese cohort, even though the criterion of hydrocephalus was excluded.