Discussion
We analyzed the refractive errors and ocular biometric changes over time in a Turkish population at the onset age ranges of myopia over time. Regarding onset ages of cohort, there were significant differences between onset biometric parameters and age. Accordingly, the increased onset age was caused to significant increase ACD (p = 0.002) and CCT (p = 0.044) whilst it was caused to significant decrease in LT (p = 0.011). Regarding consecutive follow-ups, there were significant differences in AL (p = 0.045), ACD (p <0.001) and LT (p = 0.026). In younger onset age, the progression rate of AL and ACD were higher than older onset age whereas, the regression rate LT was lower. Interestingly, these significant differences among biometric measurements and final age were lost at final visits (p>0.05). It was noted that only myopia prevalence was increased in observation time periods. Regarding the Cox regression results, myopic SE was significantly correlated by age (HR: 0.73, 95% CI: 0.57-0.92, β = -0.32, p = 0.009), AL (HR: 4.55, 95% CI:2.87-7.24, β= 1.52, p<0.001), and Kmean (HR: 2.04, 95% CI:1.55-2.67, β = 0.71, p < 0.001). In the cohort study, factors correlating the average SE at the end of three years were evaluated by logistic regression. The onset data of SE (β=0.916, p<0.001), AL (β = -0.451, p < 0.001), ACD (β = 0.430, p = 0.005), and Kmean (β = -0.172, p<0.001) were found to be significantly associated with the mean SE at the final data and therefore, the null hypothesis was rejected.
As an output of the statistical analyses of this study, an equation was proposed using the logistic regression model for calculating the estimated SE following three years. The proposed equation or calculator uses the onset SE, ACD, AL, and Kmean parameters as input to calculate estimated SE following three years. Regarding the proposed equation, the output is positively (increasingly) impacted by the onset SE and ACD, while it is negatively (decreasingly) impacted by the onset AL and Kmean.
Physiologically, biometric parameters such as AL, ACD, LT, and corneal power could affect the steadily refractive condition of the eye.12 Correspondingly, the onset of myopia or its progression could be seen during the enlargement of AL which is not able to be tolerated naturally.4-6 More specifically, 5-15 aged children are considered as the onset age of myopia.13,14 It has been reported that observing the changeover of optic biometric values on school-age children up to adolescence is an ideal method for interpreting the initiation of myopia.9 Besides, it has been reported that the increasing onset age of young patients could be one of the significant variables correlating clinical outcomes in a previous report.13 Hence, the proposed equation in this study should be considered as specific for the children between the ages of 7-12. When it is employed to younger ages or older ages than the cohort, the progression cannot estimate via the proposed calculator due to unique developmental mechanisms.15,16
In a previous report, it has been reported that potential myopia calculators could be beneficial tools to reflect the average potential outcome.10 In addition, it has been reported that such tool built-on the data should be collected from cases examined for between 2 and 5 years.10 In agreement with the previous report, the underlying data of the proposed calculator were collected from cases examined for 3 years. For the first time, the logistic regression was employed to propose an SE calculator in 7-12 aged children. Regarding the proposed calculator, associated biometric variables were identified as the onset values of SE, ACD, AL, and Kmean. The input parameters consisted of the department database records collected from registered instruments (Topcon RM-A7000B and Lenstar LS900). Hence, data variability might be possible when the proposed calculator is validated with different optical biometry instruments due to the incompatibility phenomenon. The rationale of this phenomenon, the low in agreement status in interchangeably used biometry instruments has been also reported in a previous study.17
The environmental factors regarding the onset of myopia in children are the outcomes of close-up physical activities.18,19However, this retrospective study cannot standardize individual environmental factors. The retrospective design of this study was considered a limitation of the study. As the nature of a retrospective study, selection bias could occur unintentionally among the patients. The present study described the calculation of estimated refractive error after 3 years using baseline biometric values in primary school children. Only patients with consecutive measurements for three years were included. Consequently, we found that the 3-year estimation of the refractive error correlated regularly with initial parameters in 7-12 aged children. Regarding the proposed logistic model, it could be used for patients with developing or progressing myopia as scheduling follow-ups and treatment planning purposes. Also, in patients with an estimation of fewer refractive error, the follow-ups could be made less often those. With this motivation, further prospective studies are needed to validate the proposed logistic regression equation in follow-up patients.
In this cohort study, data from Eskisehir-city and surroundings were analyzed therefore, the outcomes might not reflect the general Turkish population or other nations. However, the primary purpose of this study was not to conduct a demographic study.
It has been reported normative values for the axial length can be used to monitor eye growth in European children at both 6 and 9 years of age (N = 6934) in a previous study.20 Similarly, Sanz Diez et al.21 have reported a clinical model for the prediction of myopia development based on the creation of percentile curves of axial length in school-aged children from Wuhan in central China (N=12,554). Very recently, Truckenbrod et al.22have reported correlation curves for axial length by spherical equivalent, age, or gender in German children between 2-18 years of age (N=1965). In addition, the authors have concluded that the percentile curves of axial length can be used as a predictive measure for the likelihood of developing as well as the progression of myopia.22 In agreement with the previous reports, the axial length was one of the biometric variables in our generated model in the present study. Sanz Diez et al .21 have used the successive measurements of 226 children to verify the predictive power of the axial length growth percentile curves. In agreement with the previous report, data validation was performed on the dataset of included children (N=197, 100%) in this study. The equation or the calculation model was firstly generated in this study. Though less sample size was used relatively from the previous reports, the model was validated with our data set.
Within the limitation of this cohort study, we can conclude as follows:
- The proposed equation with this study can provide a foresight about refractive errors three years after in school-age children aged 7-12 years based on identified optical biometric measurements. Knowing about the level of refractive errors might act as guidance to assist for this process management and creating a prevention strategy. - This study assessed the onset of myopia and its progression correlated by optical biometric measurements in 7-12 aged children. This information and the model may as a reference for future comparative studies to control myopia progression.