**Purpose**:
Preschool myopia generally indicated a high risk of progression to high myopia. However, no previous study has reported its longitudinal evolution. This study aimed to investigate the longitudinal changes in preschool myopia and explore the associated key determinants.

**Methods**:
Medical records of patients seeking refractions at Zhongshan Ophthalmic Center between 2009 and 2017 were retrospectively reviewed. Mean rates of change in spherical equivalent (SE) refractive errors were evaluated in patients with preschool myopia. Association between the rate of change in SE and patient characteristics at the initial visit were examined using linear mixed-effect regression models.

**Results**:
A total of 495 cases (median initial age: 5.12 years, interquartile range [IQR], 4.12–5.76 years) were assessed with at least 2-year follow-up. The initial median SE was −3.00 D (IQR, −5.25 to −1.75 D) and the median duration of follow-up was 3.69 years (IQR, 2.89–4.99 years). On average, myopia progressed by −0.59 ± 0.47 D/year. A total of 312 (63.0%) children demonstrated myopia progression (mean rate of change in SE ≤ −0.50 D/year in either eye) and 177 (35.8%) children demonstrated refraction stability (mean rate of change < ±0.50 D/year in both eyes). Older age (β = −0.06, *P* = 0.003), female sex (β = −0.09, *P* = 0.035), and initial lower myopic SE (β = −0.07, *P* < 0.001) were associated with faster myopia progression.

**Conclusions**:
Preschool myopia on average progresses, although considerable proportion of subjects demonstrates longitudinal refraction stability. The rate of myopia progression is associated with initial patient characteristics.

^{1}while the prevalence of myopia is relatively low in children aged less than 6 years.

^{2–5}Myopia that occurs before 6 years of age is defined as “preschool myopia” to differentiate it from “school myopia”. Apart from the difference in prevalence, myopia in preschool

^{3,4,6–8}and school-aged children

^{9–12}also shows distinct risk factors.

^{13–17}

^{18}The results indicated that the clinical course of preschool myopia and its related factors may be somewhat different from that of myopia in school-aged children. However, considering the high refractive errors of study participants and the relatively small sample size, the conclusions of that study cannot be fully generalized to preschool myopia.

_{0}[Jackson cross-cylinder with axes at 180° and 90°] and J

_{45}[Jackson cross-cylinder with axes at 45° and 135°])

^{19}were calculated.

*P*< 0.0001, pairwise Pearson correlation test). Therefore, refractive error data were reported for the right eyes only, with the exception in reporting the interocular SE difference and proportions of children demonstrating myopia progression, refraction stability, and myopia regression. Descriptive statistics were applied, with means and standard deviations (SDs), medians and interquartile ranges (IQRs), or numbers and percentages reported where appropriate. Considering the confounding impact of the use of different cycloplegic agents on the assessments of longitudinal SE changes, additional analyses were conducted in a subgroup of 233 patients who used the same cycloplegic agents at the initial and final visits. The mean rate of change in SE and percentages of children demonstrating myopia regression, stability, and progression were calculated in this group of patients and compared with the corresponding values for the remaining 262 patients by using the unpaired

*t*-test and chi-squared test where appropriate.

_{0}, J

_{45}, and interocular SE difference by using unpaired

*t*-test or trend analysis. To visualize differences in longitudinal evolution among children with these different initial characteristics, linear mixed-effect models were further applied. Individual patient profiles, as well as the regression lines, relating SE refractive errors with the time since the initial visit, were plotted. The time factor used in the modeling was the time since the initial visit.

_{0}, J

_{45}, and interocular SE difference at the initial visit affected the rate of change in SE, the interaction terms of these characteristics with time since the initial visit were added to the linear mixed-effect models. A statistically significant coefficient of the interactions would indicate that the characteristic had an effect on the rate of change in SE. The method of cycloplegia at each visit was also included in regression models to adjust for its confounding impact on refraction measurements.

*P*value < 0.05 was considered statistically significant.

_{0}, J

_{45}, and interocular SE difference were +0.57 D (IQR, +0.19 to +1.13 D), 0.00 D (IQR, −0.16 to +0.29 D), and 0.50 D (IQR, 0.25–1.13 D), respectively.

*P*> 0.1).

_{45}smaller than +0.50 D, and interocular SE difference <1 D at the initial visit showed greater rates of myopia progression. Individual SE change with time and linear mixed-effect regressions based on the participant characteristics at the initial visit are demonstrated in Figures 1 through 5.

*P*= 0.003) and lower myopic SE (β = −0.07,

*P*< 0.001) at the initial visit were associated with a greater rate of myopia progression. The association between female sex and faster myopia progression (β = −0.09,

*P*= 0.035) was of borderline significance.

^{9,14,20}While most cases of myopia in school-aged children showed myopia progression after disease onset, about 37% of our preschool participants demonstrated longitudinally stable myopia or even reductions in their myopia. A previous study reported that the proportion of long-term stable refraction (longitudinal SE change equal to or less than ±0.50 D) or myopia regression (longitudinal SE change ≥ +0.75 D) was about 55% in a group of preschool children with myopia of −5.0 D or more.

^{18}These results may not be exactly comparable to our study findings because the age and refraction distributions of the study populations were different. However, the results do indicate that preschool myopia is likely to demonstrate its own course of evolution.

^{18}Thus, it is most likely that the decreasing lens thickness and power during early childhood would compensate for the growing globe and maintain ocular refraction or cause myopia regression.

^{2,21,22}Further studies investigating longitudinal changes in the ocular biometric parameters, as well as refraction, are warranted to comprehensively understand the mechanisms.

^{18}have also reported a relationship between lower initial myopia and greater progression among preschool children with high myopia. These findings are opposite to those found in school myopia, in which older age and lower initial myopic refractive errors are found to be associated with slower rate of myopia progression.

^{14–17}These differences indicate that the factors related to myopia progression are different in preschool and school-aged children. The underlying mechanisms need to be explored in more detail.

**Y. Hu**, None;

**X. Ding**, None;

**W. Long**, None;

**M. He**, None;

**X. Yang**, None

*. 2012; 379: 1739–1748.*

*Lancet**. 2017; 124: 1826–1838.*

*Ophthalmology**. 2009; 116: 739–746.e4.*

*Ophthalmology**. 2010; 117: 140–147.*

*Ophthalmology**. 2010; 51: 1348–1355.*

*Invest Ophthalmol Vis Sci**. 2011; 118: 1966–1973.*

*Ophthalmology**. 2015; 56: 8101–8107.*

*Invest Ophthalmol Vis Sci**. 2010; 94: 1012–1016.*

*Br J Ophthalmol**. 2010; 29: 520–542.*

*Prog Retin Eye Res**. 2008; 49: 2903–2910.*

*Invest Ophthalmol Vis Sci**. 2015; 314: 1142–1148.*

*JAMA**. 2008; 115: 1279–1285.*

*Ophthalmology**. 2005; 46: 51–57.*

*Invest Ophthalmol Vis Sci**. 2000; 77: 549–554.*

*Optom Vis Sci**. 1990; 67: 631–636.*

*Optom Vis Sci**. 1996; 3: 13–21.*

*Ophthalmic Epidemiol**. 1999; 19: 22–29.*

*Ophthalmic Physiol Opt**. 2006; 20: 888–892.*

*Eye (Lond)**. 2009; 86: 599–602.*

*Optom Vis Sci**. 2012; 53: 7169–7175.*

*Invest Ophthalmol Vis Sci**. 2004; 81: 819–828.*

*Optom Vis Sci**. 2004; 33: 39–43.*

*Ann Acad Med Singapore*