External Validation of a Model to Predict Postoperative Globe Axial Length in Children After Bilateral Cataract Surgery

Abstract

PURPOSE: To perform the external validation of a model to predict postoperative axial length (AL) in children over 2 years of age who were undergoing bilateral cataract surgery with primary intraocular lens (IOL) implantation. DESIGN: Validation study using a retrospective case series. METHODS: Using a population different from the one that created the model, but with the same characteristics regarding age, bilateral cataract, primary IOL implantation, and follow-up assessment, AL was estimated. The AL values estimated by the model were compared with the AL measured in the follow-ups. RESULTS: In all, 55 eyes of 30 children were selected for this study; in 5 children with bilateral cataracts, only 1 eye was included. The median age at the time of surgery was 5.01 years. Follow-up AL measurements were obtained for 179 visits. The median age at the final follow-up visit was 10.15 years. The median AL measured and estimated by the model in all visits were 22.37 mm and 22.16 mm, respectively (Pearson coefficient: 0.9534; Lin correlation: 0.9258). In the Bland–Altman analysis, the 95% limit of agreement between the 2 methods (measured and estimated AL) was 0.71 to −1.19. In 3 eyes (1.68%) with AL shorter than 21.2 mm, the difference was >0.71, and in 9 eyes with AL longer than 22.5 (5.03%), it was less than −1.19. The median AL measured and estimated at the final visit were 22.69 mm and 22.43 mm, respectively. CONCLUSION: Our previously developed prediction model for globe AL growth demonstrated good external validity by accurately predicting measured AL changes with growth in the validation cohort.