**Purpose**:
Moderate to high hyperopia is associated with visual deficits. Currently, to our knowledge no study has reported its longitudinal refraction change in a large sample of schoolchildren. We investigated the longitudinal changes in spherical equivalent (*SE*) refractive error among schoolchildren with moderate to high hyperopia.

**Methods**:
Medical records of patients seeking refractions at Zhongshan Ophthalmic Center between 2009 and 2017 were reviewed retrospectively. Eligible criteria included hyperopia ≥+2.00 diopters (D) at an initial age of 6 to 8 years, at least three visits, and at least a 2-year follow-up. Individual pattern of refraction development was evaluated based on the mean rate of change in *SE*. Mixed-effect regression analysis was used to explore factors associated with the rate of change.

**Results**:
A total of 1769 cases were identified. Median initial age was 6.4 (interquartile range [IQR], 5.9 to 7.1) years and median age at the final visit was 10.1 (IQR, 8.9 to 11.5) years. Median initial *SE* was +3.13 (IQR, +2.38 to +5.25) D. On average, participants experienced a myopic shift of −0.35 ± 0.27 D/year. A considerable number of eyes (721, 40.8%) demonstrated a longitudinal change of less than ±0.25 D/year and approximately 1 of 3 (611/1769) eyes demonstrated a change of >−0.50 and ≤−0.25 D/year. Children with greater initial hyperopia (β = −0.02, *P* < 0.001) experienced significantly faster reduction in hyperopic refraction. Age and sex had statistically significant but clinically insignificant impacts on the rate of hyperopia reduction.

**Conclusions**:
Variation exists in the refraction development of schoolchildren with moderate to high hyperopia. A considerable percentage of eyes demonstrates longitudinally stable refraction.

*SE*) refractive error ≥+2.00 diopters (D), the prevalence of hyperopia has been reported to vary from 2.4% to 13.2% among children aged 6 years, depending on the populations studied.

^{1–6}Increasing evidence suggests that moderate to high hyperopia is associated with greater risks of visual deficits, including reduced visual acuity

^{7,8}and stereoacuity.

^{1,8,9}Prevalence of other ocular conditions, such as amblyopia,

^{1,9,10}strabismus,

^{1,9,11}anisometropia,

^{9}astigmatism,

^{9}reduced accommodative response,

^{8}and abnormal convergence,

^{1}has been reported to be greater among hyperopic children compared to those with emmetropia. Moreover, impaired performance on academic-related tasks, including literacy,

^{12,13}reading,

^{14,15}and visual information processing,

^{14}has been proved associated with hyperopic refractive errors in children.

^{1–4,16–21}A few longitudinal studies have reported the amount of reduction in hyperopic refraction over time. In the Sydney Adolescent Vascular and Eye Study (SAVES), the 5- to 6-year mean rate of change in

*SE*in 80 Australian schoolchildren with baseline

*SE*≥+2.00 D and age 6 years was −0.19 D/year and that of 26 children aged 12 years at baseline was −0.26 D/year.

^{17}In a clinical sample of 86 Korean children initially aged 3 to 9 years and having hyperopic

*SE*>+0.75 D, the rate of change in

*SE*over a mean follow-up of 11.2 years was −0.26 D/year.

^{22}Longitudinal refraction evolution and ocular component growth curves of hyperopic eyes have been described using data of the Orinda Longitudinal Study of Myopia (OLSM; 43 eyes with persistent and 253 with emmetropizing hyperopia)

^{23}and the Singapore Cohort Study of the Risk Factors for Myopia (SCORM; 47 eyes with persistent and 142 with emmetropizing hyperopia).

^{24}In contrast to the plausible argument that eyes with persistent hyperopia would demonstrate an absence of growth, the studies found that these eyes grew with time at all and the growth patterns of hyperopic eyes shared many similarities with those of eyes with persistent emmetropia.

^{23,24}

*SE*of a group of Chinese schoolchildren with moderate to high hyperopia and explored factors associated with the changes.

^{25}

*SE*(calculated as spherical power plus 1/2 cylindrical power) ≥+2.00 D and interocular

*SE*difference <1.00 D at the initial visit; being followed for a minimum of 2 years since the initial visit and having at least three records on cycloplegic refractions; absence of any other ocular conditions, including aphakia, strabismus, nystagmus, and amblyopia (for the age range in our study, satisfying the following: presence or a history of at least one amblyogenic factor; presenting BCVA ≤ 20/40 in either eye or an interocular difference in BCVA of two or more lines after wearing optimal optical correction for a minimum of three months; and no cause for reduced visual acuity suspected other than amblyopia); and absence of any concurrent systemic disorders, such as developmental delay. The study was approved by the ethics committee of Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou and conducted in accordance with the Declaration of Helsinki.

^{26}Cycloplegia was achieved using three successive 1% tropicamide or 1% cyclopentolate eye drops instilled 5 minutes apart. Refraction was measured after complete cycloplegia (determined by the absence of light reflex and a dilated pupil at least 6 mm in diameter) was achieved and at least 30 minutes after the last administration of cycloplegic drugs. Cycloplegic retinoscopy was performed before measurements of subjective refractions and retinoscopy results were used as starting points of subjective refraction assessments. The spherical endpoint of subjective refraction was determined to be the most hyperopic spherical power providing best visual acuity. The Jackson cross cylinder technique was applied to determine the astigmatism among children aged 8 years or older. For children younger than 8 years, the power and axis of astigmatism were taken from the results of retinoscopy. Refractions were recorded in increments of 0.25 D. When the examining optometrist was uncertain about the result, refraction was checked and determined by one of two senior optometrists in charge. At follow-up visits, examining optometrists were masked to patient previous refractions during retinoscopy and subjective refraction measurement.

*SE*≥+2.00 D or astigmatism >1.00 D were recommended to be prescribed. Other considerations, including symptoms and school performance, also were taken into account. Spectacles were prescribed based on cycloplegic subjective refractions, providing full correction of astigmatism and either full correction or symmetrical undercorrection of hyperopia by no more than +1.50 D.

*SE*and astigmatism in vector notation (J

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

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

^{27}were calculated using records of subjective refractions. The primary outcome of our study was the rate of change in

*SE*. The mean rate of change in

*SE*was calculated as the difference between the final and initial

*SE*refractive errors divided by the follow-up duration in years. Patterns of refraction development were defined according to the mean rate of change, including refraction stability (mean rate of change in

*SE*, <±0.25 D/year), mildly myopic shift (mean rate of change in

*SE*, >−0.50 and ≤−0.25 D/year), moderately myopic shift (mean rate of change in

*SE*, >−1.00 and ≤−0.50 D/year), and rapidly myopic shift (mean rate of change in

*SE*, ≤−1.00 D/year).

*SE*s of the same individual (

*r*= 0.9825,

*P*< 0.0001, pairwise Pearson correlation test), only data from the right eyes were presented. Initial

*SE*was classified into three categories: ≥+2.00 and <+4.00 D, ≥+4.00 and <+6.00 D, and ≥+6.00 D. Initial J

_{0}and J

_{45}were classified into ≤−0.50 D, >−0.50 and <+0.50 D, and ≥+0.50 D categories. Descriptive statistics were applied. The Shapiro-Wilk test was performed to examine whether variables were normally distributed. Means ± SDs were reported for normally distributed variables and medians (interquartile ranges [IQRs]) were reported for skewed data. Numbers and percentages were reported where appropriate.

*SE*was examined by fitting the following mixed-effect model:

*i*refers to the

*i*th child,

*j*refers to the

*j*th follow-up visit;

*SE*of the

*i*th child at the

*j*th follow-up visit;

*SE*of the

*i*th child at the initial visit;

*i*th child at the

*j*th follow-up visit;

*i*th child at the initial visit;

*i*th child received 1% tropicamide at the

*j*th follow-up visit and 1 if 1% cyclopentolate was used;

*β*

_{1},

*β*

_{2},

*β*

_{3},

*β*

_{4},

*β*

_{5},

*β*

_{6}are fixed effects;

*SE*(D/year) and the fixed effect,

*SE*.

*SE*was calculated by categories of individual initial characteristics. Sex difference in the mean rate of change was examined using the unpaired

*t-*test. Trend analysis was performed to detect any differences among different initial

*SE*categories and ANOVA was conducted to detect differences among initial J

_{0}and J

_{45}categories. Individual patient profiles, as well as linear mixed-effect regression lines, relating

*SE*refractive error with age, were plotted to visualize the longitudinal changes in

*SE*in different sex and initial refraction categories. The associations of rate of change in

*SE*with sex and initial

*SE*, J

_{0}, and J

_{45}were assessed using multivariate mixed-effect regression models. Similar to Model 1 established to examine the effect of age on the rate of change in

*SE*, terms of

*SE*.

*SE*, J

_{0}, and J

_{45}categories. The distributions are presented as stacked bar graphs and were analyzed by the χ

^{2}test.

^{23,24,28,29}a secondary analysis was performed, limiting study population solely to those who were nonmyopic (defined as

*SE*>−0.50 D) throughout the follow-up period. Mean rate of change in

*SE*, as well as numbers and percentages of eyes demonstrating refraction stability, mildly, moderately, and rapidly myopic shift, were calculated and compared between sex and initial refraction categories. Multivariate mixed-effect regression analysis was performed to assess the associations of rate of change with sex,

*SE*, J

_{0}, and J

_{45}at the initial visit.

*P*< 0.05 was considered statistically significant. All statistical analyses were performed using STATA version 12.0 (Stata Corporation, College Station, TX, USA) or R (available in the public domain at https://www.r-project.org/).

*SE*was +3.13 D (IQR, +2.38 to +5.25 D) and that of initial J

_{0}and J

_{45}was +0.75 D (IQR, +0.47 to +1.13 D) and 0 D (IQR, −0.18 to +0.13 D), respectively. Distribution of the initial

*SE*is demonstrated in Figure 1A. Detailed characteristics by participant initial age are provided in Table 1.

*SE*at the last follow-up visit. Eighty-six (4.9%) children developed myopia (

*SE*≤ −0.50 D) during the follow-up period. Their median age at the initial visit was 6.5 years (IQR, 6.1 to 7.3 years) and 49 (57.0%) were boys. Compared to the remaining 1683 children who were non-myopic throughout the follow-up period, children who developed myopia demonstrated lower initial hyperopic

*SE*(median [IQR], +2.31 [+2.13, +2.75] D,

*P*< 0.0001), more positive initial J

_{0}(median [IQR], +0.87 [+0.59, +1.35] D,

*P*= 0.0290), and were followed up for a longer period of time (median [IQR], 5.9 [4.7, 6.9] years,

*P*< 0.0001). Initial age, sex, and J

_{45}were not statistically significantly different between the two groups (all

*P*> 0.1).

*SE*With Age

*SE*was −0.35 ± 0.27 D/year. The rate of the 1683 children who remained non-myopic throughout the follow-up period was −0.33 ± 0.26 D/year. Among all participants, mixed-effect regression analysis found a small effect of age (β = −0.006,

*P*< 0.001) on the rate of change in

*SE*. The association became statistically insignificant (

*P*> 0.1) when analyzed in the group of children who remained nonmyopic throughout the follow-up period.

*SE*by participant initial characteristics. Individual

*SE*evolution and linear mixed-effect regressions by sex and initial

*SE*, J

_{0}, and J

_{45}categories are illustrated in Figures 2 to 4. Girls (β = −0.04,

*P*< 0.001) and children with greater initial hyperopic

*SE*(β = −0.02,

*P*< 0.001) were associated with more rapid decrease in hyperopia in multivariate mixed-effect regression analysis of all children (terms of

*P*< 0.0001). Similar associations were found when analyzed among 1683 children who were persistently nonmyopic (for girls and greater levels of hyperopia, both β = −0.04 and

*P*< 0.001; log likelihood and

*P*value of the model were −5204.5 and < 0.0001, respectively). The associations of initial J

_{0}and J

_{45}with the rate of change in

*SE*were of borderline significance (for initial J

_{0}, both β = 0.005,

*P*= 0.078 and 0.045 in all and persistently nonmyopic children, respectively; for initial J

_{45}, β = 0.013 and 0.009,

*P*= 0.021 and 0.072 in all and persistently nonmyopic children, respectively).

*P*= 0.256), nor among initial J

_{45}categories (

*P*= 0.171). Greater proportions of eyes in lower initial hyperopic

*SE*categories demonstrated patterns of slower hyperopia reduction (

*P*< 0.001, Fig. 5). Eyes in the ≥+2.00 and <+4.00 D initial

*SE*category predominantly (533/1104, 48.3%) demonstrated refraction stability over the follow-up period, whereas in the ≥+4.00 and <+6.00 D and ≥+6.00 D categories, the eyes predominantly (37.2% [116/312] for the ≥+4.00 and <+6.00 D category and 38.5% [136/353] for the ≥+6.00 D category) demonstrated mildly myopic shift in the long term. Percentage of eyes demonstrating longitudinally stable refractions in the ≥+4.00 and <+6.00 D category was 33.0% (103/312) and that in the ≥+6.00 D category was 24.1% (85/353). Of seven children with initial J

_{0}≤−0.50 D, six demonstrated refraction stability and the other one presented mildly myopic shift. Percentage of eyes demonstrating refraction stability was 31.0% (162/522) in the >−0.50 and <+0.50 D initial J

_{0}category and 44.6% (553/1240) in the ≥+0.50 D category.

*SE*category were predominantly stable in refraction and those in the two higher initial

*SE*categories predominantly demonstrated mildly myopic shift. Approximately 52.2% (533/1022) of eyes in the ≥+2.00 and <+4.00 D category demonstrated longitudinal refraction stability. The percentages of eyes in the ≥+4.00 and <+6.00 and ≥+6.00 D categories were 33.4% (103/308) and 24.1% (85/353), respectively. Regarding initial J

_{0}, percentage of eyes demonstrating refraction stability was 85.7% (6/7) in the ≤−0.50 D category, 32.1% (162/505) in the >−0.50 and <+0.50 D category, and 47.2% (553/1171) in the ≥+0.50 D category.

*SE*refraction changes with age among Chinese schoolchildren with moderate to high hyperopia. In the study, a wide range of initial hyperopic refractive errors was examined, with a considerable sample size and adequate periods of follow-up. Rates of change in

*SE*and distributions of patterns of refraction development were reported by participant initial refractions. The effects of age and sex on hyperopic refractive error evolution were determined.

*SE*in our sample of schoolchildren with initial

*SE*≥+2.00 and <+4.00 D was −0.30 D/year and that among children with initial

*SE*≥+4.00 and <+6.00 D and ≥+6.00 D was −0.38 and −0.45 D/year, respectively. In the Sydney Myopia Study, the 5-year mean change in

*SE*was −0.87 D among children with baseline age of 6 years and

*SE*≥+2.00 and <+3.50 D, and −0.57 D among those with baseline

*SE*≥+3.50 D (Ilango M, et al.

*IOVS*2018;59:ARVO E-Abstract 3376). Although the figures of the two studies may not be exactly comparable, it seems likely that children in the Sydney Myopia Study underwent a much slower reduction in hyperopia compared to our Chinese hyperopic sample. It should be acknowledged our data were obtained from clinical settings and the potential of bias could not be excluded. Data from the Refractive Error Study in Children (RESC) series demonstrate divergent patterns of refraction development among schoolchildren with different ethnicities and at different study sites.

^{16}Regarding hyperopia (

*SE*refractive error ≥+2.00 D), specifically, there were substantial differences in the prevalence in young age groups among study sites; while by the age of 15, the prevalence cleared to insignificant levels at most sites.

^{16}These data suggest that considerable variation in refraction development exists among hyperopic children of different ethnic origins and cultural settings.

*SE*. It seems that during this early period of school age, the pace of reduction in hyperopic refractive error is relatively constant across time and would not present great variation between boys and girls. However, limited to the relatively short follow-up of our study, full characterization of the evolution of hyperopia must be explored by further studies with longer follow-up extending into adolescence and early adulthood.

*SE*. During the first year or two of human life, the rate of hyperopia reduction also positively relates to the initial level of hyperopia.

^{30,31}The exact mechanisms underlying the associations are not yet illuminated. It is likely that the human eyes can respond to some features related to refractive errors and converge toward a low hyperopic value over early life and the school age. “Regression to the mean” could be a contributor to the phenomenon of greater reduction in hyperopic

*SE*among children with initial greater hyperopia. When analyzed by the duration of follow-up, the estimated amount of regression to the mean of this dataset (calculated as [1 − correlation between individual initial

*SE*and

*SE*at the last follow-up visit] × 100%) ranged from 4% to 24%.

*SE*, it must be noted that at the initial visit, children at older ages demonstrated greater hyperopic refractive errors (Table 1). It could be possible that higher levels of hyperopia create clinical symptoms more often among older children, making them appear at older ages in the clinic even as hyperopia becomes lower on an individual basis over time.

*SE*categories. It is suggested that the presence of significant hyperopia at age 6 years results from a failure of effective emmetropization during early life: having an initial hyperopia too great to be corrected and/or being very slow in longitudinal refraction change.

^{32}Little change in refraction has been observed in some highly hyperopic infants in previous longitudinal studies.

^{31,33,34}We speculated that among our study participants, at least some of those who demonstrated refraction stability have persistent deficiencies in refraction change from birth through to their school age.

^{35}(GOALs;

*n*= 97; baseline age, 6.6 ± 0.4 years; 43 [44.3%] boys; and baseline

*SE*, +2.42 ± 0.43 D [range, +2.00 to +3.63 D]), the mean rate of change in

*SE*in children of our clinical sample with comparable initial refractive errors (≥+2.00 and <+4.00 D initial

*SE*group) was significantly slower (0.30 ± 0.25 vs. 0.39 ± 0.27,

*P*= 0.0005,

*t*-test). There was a probability that children with persistent hyperopia were more inclined to revisit the clinic and those with emmetropization might have missed their follow-up visits, leading to an underestimation of the rate of hyperopia reduction in this study. Optical correction can have an important impact on the growth of hyperopic eyes. However, we failed to collect and record children's compliance with spectacle prescription and the effect of optical correction remains uninvestigated. Ocular biometric data were not available. Thus, the precise mechanisms coordinating the optical and structural development of the eye cannot be completely understood. As a general limitation, the study results may not necessarily generalize to all hyperopic schoolchildren. There may be variations, depending on the sample ethnicity and cultural setting.

*SE*among Chinese schoolchildren with moderate to high hyperopia. On average, a reduction in the hyperopic refractive error was observed. Age and sex did not have clinically significant impacts on the rate of hyperopia reduction. Children with greater initial hyperopic refractive errors demonstrated faster rates of hyperopia reduction. Considerable percentage of eyes, even among children with high hyperopia, demonstrated longitudinal refraction stability. Further studies are warranted to examine the potential discrepancies among children of different ethnicities and cultural settings, to determine the impact of optical correction on the refraction evolution, as well as to illuminate the longitudinal changes in ocular biometric parameters.

**Y. Hu**, None;

**X. Ding**, None;

**J. Zeng**, None;

**D. Cui**, None;

**C. Li**, None;

**M. He**, None;

**X. Yang**, None

*Ophthalmology*. 2008; 115: 678–685.

*Invest Ophthalmol Vis Sci*. 2004; 45: 793–799.

*Invest Ophthalmol Vis Sci*. 2002; 43: 623–631.

*Invest Ophthalmol Vis Sci*. 2003; 44: 3764–3770.

*Ophthalmology*. 2017; 124: 1826–1838.

*Ophthalmology*. 2012; 119: 2021–2027.

*Ophthalmology*. 2011; 118: 2262–2273.

*Am J Ophthalmol*. 2016; 170: 143–152.

*Optom Vis Sci*. 2014; 91: 383–389.

*Ophthalmology*. 2014; 121: 622–629.

*Ophthalmology*. 2011; 118: 2251–2261.

*Ophthalmology*. 2016; 123: 681–689.

*Arch Dis Child*. 2005; 90: 150–153.

*Optom Vis Sci*. 2015; 92: 227–236.

*Dev Med Child Neurol*. 1985; 27: 504–513.

*Acta Ophthalmol*. 2010; 88: 877–884.

*Ophthalmology*. 2013; 120: 1482–1491.

*Ophthalmology*. 2005; 112: 678–685.

*Am J Ophthalmol*. 2000; 129: 427–435.

*Am J Ophthalmol*. 2000; 129: 436–444.

*Am J Ophthalmol*. 2000; 129: 445–454.

*J Pediatr Ophthalmol Strabismus*. 2017; 54: 43–51.

*Invest Ophthalmol Vis Sci*. 2005; 46: 2317–2327.

*Invest Ophthalmol Vis Sci*. 2010; 51: 1341–1347.

*Optom Vis Sci*. 1993; 70: 1019–1026.

*Invest Ophthalmol Vis Sci*. 2019; 60: 154–160.

*Optom Vis Sci*. 2009; 86: 599–602.

*Invest Ophthalmol Vis Sci*. 2007; 48: 2510–2519.

*Ophthalmology*. 2012; 119: 1478–1484.

*Vision Res*. 1995; 35: 1325–1328.

*Invest Ophthalmol Vis Sci*. 2000; 41: 3726–3731.

*Eye (Lond)*. 2014; 28: 169–179.

*Vision Res*. 2001; 41: 2799–2810.

*Invest Ophthalmol Vis Sci*. 2005; 46: 3074–3080.

*JAMA*. 2015; 314: 1142–1148.