Huyen T Bui, Dirk J van Hemert, Tianli Peng, Shuhui Yang, Hai-Yen T Nguyen, Hannah Noor, Siofra C Harrington, Sander C M Kneepkens, Lene A Hagen, Craig S Walker, CREAM-Kids Consortium, Edward Mallen, Matthew Cufflin, Rigmor C Baraas, Neema Ghorbani-Mojarrad, Olavi Pärssinen, Louise Terry, Shi-Ming Li, Amanda N French, Jason C Yam, Caroline C W Klaver, Jeremy A Guggenheim
Despite moderate diagnostic performance in European samples, the current AL/CR-based models lacked sufficient accuracy for screening and performed poorly in East Asian children, suggesting that AL/CR is not a reliable non-cycloplegic marker of pseudomyopia. Further studies incorporating additional ocular biometric parameters and employing a more rigorous definition of pseudomyopia may build on these initial findings.
PURPOSE: Pseudomyopia is an accommodative disorder that can be identified by the difference between pre- and post-cycloplegic refractive error. The current study aimed to evaluate whether the axial length to corneal curvature (AL/CR) ratio, in conjunction with non-cycloplegic refractive error, can reliably identify pseudomyopia in children, thus avoiding the need for cycloplegia.
METHODS: Refractive error and biometry data from longitudinal studies based in Denmark (n = 645; age 7-17 years) and China (Anyang Childhood Eye Study; n = 2487; age 6-13 years) were used to train separate linear mixed models for children of European or East Asian ancestry, respectively, with AL/CR ratio as outcome and cycloplegic spherical equivalent refractive error (SER), age and sex as predictors. The models could be used to calculate the AL/CR 95% prediction interval in a child of known age, sex and non-cycloplegic SER; if the child's AL/CR fell below the lower limit of the AL/CR 95% prediction interval, then the child was classified as pseudomyopic. The European and East Asian models were externally validated using data from participants of the Ireland Eye Study (IES; n = 92) and Sydney Myopia Study (SMS; n = 686) cohorts, and the Hong Kong Children Eye Study (HKCES; n = 9519) and Vietnam (n = 1165) cohorts, respectively.
RESULTS: The European model demonstrated moderate detection accuracy for pseudomyopia: sensitivity = 0.82 (IES) and 0.81 (SMS); specificity = 0.75 in both cohorts. By contrast, the East Asian model performed poorly: sensitivity = 0.22 (HKCES) and 0.13 (Vietnam); specificity = 0.99 in both cohorts.
CONCLUSIONS: Despite moderate diagnostic performance in European samples, the current AL/CR-based models lacked sufficient accuracy for screening and performed poorly in East Asian children, suggesting that AL/CR is not a reliable non-cycloplegic marker of pseudomyopia. Further studies incorporating additional ocular biometric parameters and employing a more rigorous definition of pseudomyopia may build on these initial findings.