To the Editor We congratulate Binenbaum et al1 on their work on validation of postnatal growth and retinopathy of prematurity screening criteria. Safely reducing the number of infants requiring a retinopathy of prematurity (ROP) examination is challenging. We respectfully share some concerns about the statistical aspects of the article.
The Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) statement is a 22-item checklist for reporting on the development and validation of prediction models.2 In addition, important aspects for consideration of performance are discrimination (the C index) and calibration (a plot of expected outcome probabilities against observed outcome frequencies). Were all the TRIPOD recommendations used for developing3 or validating1 the model, since the report has sensitivity with confidence intervals as a performance measure?