Hyperopic reserve, early myopia risk and screening strategies in children: A multi-regional analysis from the CREAM-KIDS consortium

Summary

Reduced hyperopic reserve (HR) in early childhood is a strong predictor of future myopia. A harmonized dataset from the CREAM-Kids Consortium was utilized to assess regional differences in HR. Among 7-year olds, mean (SD) HR was 1.44 (1.17) D in Europe, 1.32 (0.80) D in Australia, 0.78 (0.89) D in Hong Kong, and 0.59 (0.66) D in Singapore (p < 0.001). The annualized decline in HR was −0.12 D/year among European-Australian followed from age 6.73 (0.64) to 11.64 (1.49), and −0.35 D/year among Asian followed from age 7.71 (0.77) to 10.71 (0.77). In European-Australian aged 6–7, the sensitivity and specificity of myopia screening was 67%–75% and 98%, and in Asian aged 4–9, 92%–99% and 70%–77%, respectively. Asian children demonstrated substantially lower HR and a faster consumption than their European and Australian peers, placing a larger proportion of Asian children at higher risk for early-onset myopia.

Graphical abstract

Highlights

• •Asia children showed less age-specific hyperopic reserve to Europe and Australia
• •Asia children showed faster hyperopic reserve consumption to Europe and Australia
• •“Adequate” hyperopic reserve if cycloplegic SER ≥ 0.75D between 6 and 9 years of age
• •Myopia prevention shall shift earlier, to help preserve hyperopic reserve

Ophthalmology; Eye observable entity

Introduction

Myopia, also known as nearsightedness, is primarily caused by excessive elongation of the axial length of the eye. For many years, myopia was considered merely a refractive error condition, but due to its high prevalence and significant health consequences, it is increasingly recognized as a disease.¹ Globally, the prevalence of myopia is projected to reach approximately 50% by 2050.² The highest prevalence rates are observed in East and Southeast Asian countries, such as China, Japan, Korea, and Singapore, where myopia prevalence is 80%–90% among young adults and a substantial proportion of school-aged children.³ In Europe, the prevalence is estimated to be approximately 23.5% in the general population.⁴ In Australia, reported myopia prevalence is 14.4% at age 12 years, 29.6% at age 17 years, and 25.8% at age 20 years^5,6

The concept of hyperopic reserve has recently emerged as a useful way to describe the refractive “buffer” that protects a non-myopic child from progressing to myopia.^{7,8,9,10,11,12,13} In simple terms, hyperopic reserve refers to the amount of age-appropriate physiological hyperopia present before the eye crosses into emmetropia and, ultimately, myopia. Although the idea has been widely used in Chinese clinical and public health practice, formal definitions have only recently begun to appear in the literature. Consensus documents and cohort studies from China describe hyperopic reserve as a physiological hyperopic refractive status that precedes emmetropia and myopia, and use age-specific reference intervals to classify children as having adequate or reduced reserve.^9,10,11 However, outside China the term is rarely used, and there is no widely adopted operational definition that links hyperopic reserve to subsequent myopia risk or to practical screening thresholds.

Closely related is the concept of pre-myopia, which the International Myopia Institute defines as a refractive state of ≤ +0.75 D and > −0.50 D, combined with additional risk factors such as parental myopia, axial length and environmental exposures.¹⁴ Pre-myopia is therefore a composite risk construct that integrates both refractive status and secondary risk factors.⁸ In contrast, hyperopic reserve is a purely refractive concept: it describes the magnitude of physiological hyperopia at a given age, independent of risk modifiers. When refractive error in early childhood falls below a safe level of hyperopic reserve (for example <+0.75 D between 6 and 9 years), the child enters a higher-risk zone where the probability of imminent myopia onset increases sharply.^7,15,16 In this context, hyperopic reserve can be viewed as the refractive substrate on which pre-myopic risk factors act.

From a public health perspective, hyperopic reserve provides a useful lens through which to understand global differences in myopia risk. Neonatal refractive distributions in east Asian and Western populations are remarkably similar, with mean hyperopia around +3.00 D and standard deviations of 1.5–2.0 D, suggesting that babies are born with high level of hyperopic reserve in refractive terms.^17,18,19,20 The marked divergence in myopia prevalence by later childhood may, in part, reflect differences in the rate at which hyperopic reserve is depleted in the preschool and early school years. Large cohort data from Asia show that once hyperopic reserve falls below approximately +1.00 D in children aged 6–9 years, myopic shift accelerates and the risk of incident myopia rises steeply.²¹ In effect, children who enter school with little or no hyperopic reserve are already on a short trajectory to myopia within the next few years.

Against this background, hyperopic reserve offers a potentially powerful tool for early myopia prevention and for designing age-appropriate screening strategies. If Asian and European infants start from similar refractive baselines but differ in their early consumption of hyperopic reserve, this would imply that primary prevention efforts—especially in high-risk regions—must begin before school entry. It would also suggest that screening programs should monitor hyperopic reserve as well as the presence of myopia itself. However, there are few data directly comparing the distributions and trajectories of hyperopic reserve across regions.

Data from the Consortium for Refractive Error and Myopia (CREAM)-Kids, comprising 20 pediatric cohort studies from Europe, Australia and Asia, were used to address three questions. First, how do age-specific distributions of hyperopic reserve differ between these regions? Second, how rapidly is hyperopic reserve consumed over time, and does this differ between regions? Third, how well do the existing, largely non-cycloplegic, myopia screening approaches perform in settings where children enter school with high versus low hyperopic reserve?

Results

A total of 28,505 children aged 4–9 years with cycloplegic autorefraction data were included in the analysis. Among them, 22,144 non-myopic participants at baseline were analyzed for hyperopic reserve, 2,813 participants with follow-up data were analyzed for depletion in hyperopic reserve over time, and 24,021 participants with both cycloplegic and non-cycloplegic measurements were included in the myopia screening analysis (Figure 1).

Figure 1.

Flow chart

The demographic characteristics of the study population are summarized in Table S2. The overall mean (SD) age was 6.94 (1.12) years, and 14,914 (52.33%) participants were male. By region, 1,379 (4.84%) were from Europe, 1,725 (6.05%) from Australia, and 25,401 (89.11%) from Asia.

Hyperopic reserve

A total of 22,144 participants were included in this analysis, among whom 1,339 (6.05%) were from European studies, 1,669 (7.55%) were from Australian studies, and 19,106 (86.40%) were from Asian studies. As shown in Table S3; Figure 2, at ages 4 and 5, children in China’s Hong Kong had a mean (SD) hyperopic reserve of 1.06 (0.90) D and 1.02 (0.84) D, respectively. Among 6-year-old children, European participants had a slightly higher mean hyperopic reserve (1.47 [1.32] D) than Australian participants (1.28 [0.80] D) (difference [95% CI] = 0.19 [0.04, 0.33] D; p = 0.01), and both groups had significantly higher values than Hong Kong children (0.91 [0.85] D; ANOVA p < 0.001) (Europe vs. Hong Kong: difference [95% CI] = 0.56 [0.43, 0.69] D; p < 0.001; Australia vs. Hong Kong: difference [95% CI] = 0.37 [0.31, 0.45] D; p < 0.001). At ages 7 and 8, differences between European and Australian children became smaller (age 7: 1.44 [1.17] D vs. 1.32 [0.80] D, difference [95% CI] = 0.12 [0.03, 0.22] D, p = 0.01; age 8: 1.31 [1.07] D vs. 1.36 [1.22] D, difference [95% CI] = -0.05 [−0.52, 0.43] D, p = 0.84). Singaporean children consistently had the lowest hyperopic reserve, significantly lower than that of children from Hong Kong (age 7: 0.59 [0.66] D vs. 0.78 [0.89] D, difference of [Hong Kong-Singapore] [95% CI] = 0.19 [0.13, 0.24] D, p < 0.001; age 8: 0.55 [0.60] D vs. 0.70 [0.88] D, difference of [Hong Kong - Singapore] [95% CI] = 0.15 [0.09, 0.22] D, p < 0.001). This trend persisted at age 9 (Europe: 1.05 [0.95] D; Hong Kong: 0.60 [0.84] D; Singapore: 0.44 [0.50] D; ANOVA p < 0.001) (Europe vs. Hong Kong: difference [95% CI] = 0.45 [0.23, 0.68] D; p < 0.001; Europe vs. Singapore: difference [95% CI] = 0.61 [0.39, 0.84] D; p < 0.001; Hong Kong vs. Singapore: difference [95% CI] = 0.16 [0.08, 0.24] D; p < 0.001).

Figure 2.

Mean cycloplegic spherical equivalent (D) among non-myopic children by age in region or countries

Error bars indicate 95% confidence interval. Data are derived from cross-sectional analyses; each point reflects the mean value for children at a specific age. Lines connecting the points are included to illustrate overall trends and do not represent longitudinal changes within the same individuals.

Figure 3 shows the distributions of hyperopic reserve among 7-year-old non-myopic children by region or country. All four distributions were unimodal and approximately normal but skewed to the right due to the −0.50 D myopia cutoff. Although European and Australian children had similar mean values (1.44 D and 1.32 D, respectively), the Australian distribution was more symmetrical and centered around the mean, whereas the European distribution was more right-skewed, with a difference of 0.43 D between the mode and mean. The distributions for Hong Kong and Singapore were also similar, though Singaporean children presented lower level of hyperopic reserve.

Figure 3.

Histograms of hyperopic reserve (D) distribution by region or countries in age 7 children

(A) Europe (N = 791).

(B) Australia (N = 1,068).

(C) China’s Hong Kong (N = 5,324).

(D) Singapore (N = 627). Dashed lines indicate the mean of each distribution, and the kernel density curve represents the estimated probability distribution. Kurtosis reflects the degree to which a distribution is peaked or has heavy tails, and hence deviates from a normal distribution. Although the European and Australian distributions had similar mean values, the Australian distribution was more concentrated around the mean, whereas the European distribution showed heavier tails, indicating a higher proportion of children with high hyperopia and pre-myopia.

Consumption of hyperopic reserve over time

A total of 2,813 participants were included in this analysis, among whom 327 (11.62%) were from European studies, 855 (30.39%) were from Australian studies, and 1,631 (57.98%) were from Asian studies. Consumption of hyperopic reserve over time by region or country is shown in Figure S1. Among participants of the combined European-Australian group, the mean (SD) baseline age was 6.73 (0.64) years, and their mean hyperopic reserve was 1.35 (0.95) D. By the time their mean (SD) age increased to 11.64 (1.49) years, the mean hyperopic reserve had declined to 0.74 (1.21) D, the annualized consumption rate was −0.12 D/year. In Asian children, the baseline hyperopic reserve was lower, at 0.55 (0.62) D at a mean (SD) age of 7.71 (0.77) years. At follow-up (mean [SD] age: 10.71 [0.77] years), the mean spherical equivalent decreased to −0.51 (1.26) D, indicating a greater loss of hyperopic reserve, with annualized consumption rate of −0.35 D/year.

As shown in Table 1, baseline age, sex, baseline spherical equivalent refraction (SER), and region were all significantly associated with depletion of hyperopic reserve per year in both univariable and multivariable analyses. After adjustment for baseline age, sex, and baseline SER, European and Australian children showed 0.23 D (standard error = 0.01 D) less depletion of hyperopic reserve per year compared with Asian children. To note that, large samples increase the likelihood of detecting small effects. Though sex was statistically significant risk factor in the model, the difference between males and females was approximately −0.05 D, which is not clinically significant.

Table 1.

Univariable and multivariable linear regression of hyperopic reserve consumption per year (D/year)

Variables	Univariable		Multivariable
	Beta (SE)	p value	Beta (SE)	p value
Baseline age (years)	−0.04 (0.01)	<0.001	0.05 (0.01)	<0.001
Sex	–	–	–	<0.001
Male	Reference	<0.001	Reference	–
Female	−0.05 (0.01)	–	−0.05 (0.01)	–
Baseline SER (D)	0.09 (0.01)	<0.001	0.05 (0.01)	<0.001
Region	–	<0.001	–	<0.001
Asia	Reference	–	Reference	–
Europe and Australia	0.23 (0.01)	–	0.23 (0.02)	–

p value < 0.05 indicate statistical significance, and are shown in Bold fond in the table.

Among children who developed myopia during follow-up, the rate of depletion of hyperopic reserve was faster in Asian children than in European and Australian children, the difference of annualized consumption rate was −0.19 D/year (Figure S2). For participants who remained non-myopic at follow-up (Figure S3), the rate of hyperopic reserve decline was more similar between the regional groups, with a difference of −0.14 D/year. Among European and Australian children, mean hyperopic reserve decreased from 1.43 (0.95) D at age 6.73 (0.61) years to 0.97 (0.98) D at age 11.66 (1.51) years. Among Asian children, mean hyperopic reserve decreased from 0.85 (0.62) D at age 7.77 (0.77) years to 0.42 (0.61) D at age 10.77 (0.77) years.

Myopia screening

A total of 24,021 participants were included in this analysis, among whom 778 (3.24%) were from European studies, and 23,243 (96.76%) were from Asian studies. The sensitivity and specificity of non-cycloplegic SER for detecting myopia by region or country are shown in Table 2. In the combined European-Australia group, the prevalence of myopia was very low among children aged 6–7 years (1.4%–2.3%), resulting in low sensitivity (67%–75%) but high specificity (98%). In Asia, myopia prevalence increased substantially with age, from 6.8% at age 4 to 44.5% at age 9. Correspondingly, sensitivity remained high (92%–99%), whereas specificity ranged from 70% to 77%.

Table 2.

Sensitivity and specificity of non-cycloplegic SER in screening myopia, by region

Age (years)	Europe and Australia			Asia
	Prevalence (%)	Sensitivitya TP/(TP + FN) (%)	Specificityb TN/(TN + FP) (%)	Prevalence (%)	Sensitivitya TP/(TP + FN) (%)	Specificityb TN/(TN + FP) (%)
4	N/A	N/A	N/A	6.8%	25/27 (93%)	263/340 (77%)
5	N/A	N/A	N/A	7.0%	150/163 (92%)	1,443/1,986 (73%)
6	1.4%	3/4 (75%)	204/208 (98%)	13.7%	884/920 (96%)	4,073/5,510 (74%)
7	2.3%	12/18 (67%)	499/508 (98%)	23.9%	1,668/1,708 (98%)	3,865/5,281 (73%)
8	0%	0/1 (0%)	37/39 (95%)	34.9%	1,873/1,909 (98%)	2,497/3,461 (72%)
9	N/A	N/A	N/A	44.5%	862/871 (99%)	742/1,067 (70%)

^aSensitivity is represented as TP/(TP + FN) (%), where TP stands for true positive, FN stands for false negative.

^bSpecificity is represented as TN/(TN + FP) (%), where TN stands for true negative, FP stands for false positive.

Discussion

Main findings

In this multi-regional analysis of non-myopic children aged 4–9 years, we used the concept of hyperopic reserve to describe early refractive development and myopia risk. We propose an operational definition of hyperopic reserve as the physiological hyperopia present in non-myopic eyes, and consider reserve to be “adequate” when cycloplegic SER is ≥ +0.75 D between six and nine years of age. This threshold aligns with pre-myopia risk definitions, where SER ≤ +0.75 D marks entry into a higher-risk refractive zone, while retaining the simplicity of a purely refractive construct. Using this framework, we found that European and Australian children had substantially greater hyperopic reserve than their peers in Hong Kong and Singapore at all ages, and that Asian children not only started with lower reserve but also consumed it more rapidly over time.

Regional differences of hyperopic reserve

Studies reporting hyperopic reserve in European and Australian children are relatively limited. A PubMed search using the terms (“refractive error” OR “Australia”) and (“Europe” OR “Australia”) identified only a few relevant studies, most of which were already included in the CREAM-Kids consortium (e.g., NICER and SMS studies). Nevertheless, our findings on hyperopic reserve among European, Australian, and Asian children are consistent with previous reports.^11,22,23,24 In the present study, the mean (SD) hyperopic reserve among 7-year-old European children was 1.44 (1.17) D. Similarly, Harrington et al. reported a mean (SD) cycloplegic SER of 1.46 (1.29) D among 621 Irish children aged 6–7 years, with a myopia prevalence of 3.2%.²² Rose et al. reported a mean (SD) SER of 1.35 (1.22) among children aged 3–6 years, with a myopia prevalence of 1.8% from the Sydney Paediatric Eye Disease Study.²⁴ Among 4–6-year-old children in China’s Hong Kong, the mean hyperopic reserve was 0.91–1.06 D. Align with Pu et al.’s finding, the mean SER was 1.03–1.09 D among aged 4–6 years children in Beijing, China.¹¹

These findings sit in striking contrast to neonatal refractive data, which show broadly similar distributions in East Asian and European infants, with mean refractions around +3.00 D and standard deviations of 1.5–2.0 D.^17,18,19,20 This suggests that children in high- and lower-risk regions present high level of hyperopic reserve when born in refractive terms, and that the marked divergence in myopia prevalence by late childhood arises from differences in the rate at which hyperopic reserve is depleted in the preschool and early school years. By the time children in high-risk Asian settings start school, many have already lost most of their hyperopic reserve, entering the pre-myopic refractive range and setting up a short time-horizon for myopia onset. Our data therefore support the view that the critical window for primary myopia prevention in East and Southeast Asia, lies before or around school entry rather than later in childhood.

Consumption of hyperopic reserve

When examining the consumption of hyperopic reserve, Asian children not only started with a lower baseline hyperopic reserve but also exhibited a faster rate of consumption compared with their European and Australian counterparts. This pattern aligns with the substantially higher incidence and prevalence of childhood myopia observed in Asia. Consistent with our findings, Naduvilath and colleagues reported that axial elongation and refractive error progression were slower in European children than in East Asian children, in both myopic and non-myopic groups.²⁵ The annualized difference in refractive error progression between the two regions was −0.10 D/year in non-myopic eyes,²⁵ comparable to the −0.14 D/year observed among participants in our study who remained non-myopic throughout follow-up. For children who remained non-myopic throughout follow-up, baseline hyperopic reserve was substantially higher than in those who later developed myopia. In Europe and Australia, children aged 6–7 years with a hyperopic reserve of approximately +1.43 D were less likely to develop myopia than those with only +0.53 D (Figures S2 and S3). Supporting this, McCullough and colleagues followed 390 European children aged 6–7 years for nine years and found that those with a hyperopic reserve greater than +1.00 D were unlikely to develop myopia over time.²⁶ In Asia, the “safe” hyperopic reserve threshold may be around +0.85 D at ages 7–8 (the level of hyperopic reserve of children who didn’t become myopic at follow-up in our sample, Figure S3). However, Chen and colleagues reported that myopic shift accelerates once hyperopic reserve falls below +1.00 D in Asian children aged 6–9 years.²¹

Conceptually, our findings also help to clarify the relationship between hyperopic reserve and pre-myopia. Pre-myopia can be viewed as a high-risk state in which hyperopic reserve has already been eroded to ≤ +0.75 D and additional risk factors, such as parental myopia, longer axial length and intensive near work, further increase the likelihood of incident myopia.^8,14 Hyperopic reserve on its own does not capture these secondary factors, but it does indicate how close a child is, refractively, to the point at which myopia becomes likely in the near term. In regions where many children enter school with low hyperopic reserve, screening programs that focus solely on detecting manifest myopia will miss an important opportunity to identify high-risk children earlier. Conversely, in regions where hyperopic reserve remains high in early primary school, screening at ages 6–7 years is less efficient for myopia detection and may be better targeted toward hyperopia, amblyopia, and strabismus, with a shift toward myopia detection at older ages when reserve has declined. Importantly, related constructs—such as axial length, myopia prevalence, and myopia incidence—reflect different dimensions of refractive development and should be interpreted complementarily rather than interchangeably. Hyperopic reserve describes refractive status, axial length describes ocular growth, and prevalence and incidence describe population-level burden and onset rates. Importantly, these findings reflect not only differences in myopia prevalence between regions, but also differences in how refractive distributions are positioned relative to the diagnostic threshold. In populations where many children cluster close to the myopia cut-off, small accommodative fluctuations under non-cycloplegic conditions may increase the likelihood of misclassification. Thus, distributional positioning of hyperopic reserve, rather than prevalence alone, may influence screening performance.

Taken together, these data suggest that although children across regions may start life with broadly similar refractive profiles, many children in high-risk Asian settings have already lost most of their hyperopic reserve by school entry, placing them on a short trajectory to myopia. This underscores the need to shift myopia prevention upstream into the preschool years, particularly in Asia, through strategies that protect hyperopic reserve—such as increased outdoor time and balanced near-work demands—before children reach the pre-myopic zone. Integrating routine monitoring of hyperopic reserve into pediatric eye care and school health systems could provide a simple, scalable way to identify children at highest risk and to tailor screening and intervention strategies by region.

Limitations of the study

This study has several limitations. First, the years during which the individual cohort studies were conducted do not fully align across regions, which might lead to cohort effect; however, they still collectively provide a general overview of refractive development patterns. Second, the sample sizes for the European and Australian cohorts were relatively small, which may limit the precision of estimates for these regions. Third, for the comparison of myopia progression, it was not possible to perfectly match participants from Europe, Australia, and Asia by age, the comparisons should be interpreted with caution, as they are not fully comparable. In addition, “hyperopic reserve annualized consumption rates” was based on linear interpolation between two points (baseline and consecutive follow-up). In reality, change of refractive error may not be linear over time.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Weizhong Lan (lanweizhong@aierchina.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The data used for analysis is not publicly available. For reasonable requests, please contact the lead contact.
• •This study does not report original code.
• •Any additional information required to reanalyze the data reported in this study is available from the lead contact upon request.

Acknowledgments

We gratefully acknowledge all participants, their families, teachers, and school administrations across all contributing studies for their invaluable participation. Core support and funding were provided by multiple national and international research councils, universities, foundations, and institutes. Specific financial backing came from a range of grant bodies and research organizations, whose support was essential to this work. Funding bodies include: Tampere Region Support Foundation for the Visually Impaired, Eye Foundation Finland; The SNOW study was funded by the University of South-Eastern Norway; Australian National Health and Medical Research Council (NHMRC) Project Grants 512530 and 253732; UK Medical Research Council Clinical Research Training Fellowship and University of London Chadburn Clinical Lectureship; The Northern Ireland Childhood Errors of Refraction (NICER) study was funded by the College of Optometrists, UK; the Science and Technology Innovation Program of Hunan Province, China (2023RC1079, 2024RC5002); European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Programme (grant 101098324, CCWK), CORR (2023-8.2.0), Landelijke Stichting Blindenbelangen, Oogfonds, and Algemene Nederlandse Vereniging ter voorkoming van Blindheid (Uitzicht grant 2024-30). The authors alone are responsible for the content and writing of the study.

Author contributions

Conceptualization, W.P., I.F., and W.L.; study design, W.P., S.H., G.L., M.M., O.P., R.C.B., L.A.H., A.F., and W.L.; data preparation, W.P., S.C.M.K., and D.J.v.H.; data analysis, W.P.; manuscript drafting, W.P.; providing resources, K.W., S.C.M.K., S.M., O.P., R.C.B., A.F., S.-M.S., J.Y., C.K., I.F., and W.L.; manuscript criticizing and revising, W.P., S.H., G.L., M.M., L.L.F., S.C.M.K., S.M., K.W., O.P., R.C.B., L.A.H., A.F., S.-M.S., J.Y., C.K., I.F., and W.L.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Children	Cream-Kids Consortium Dataset including 20 pediatric cohort studies from Europe, Australia and Asia	Not applicable
Software and algorithms
R Statistical Software (v4.3.2; R Core Team 2021, r-project.org)	R Foundation for Statistical Computing	https://www.r-project.org/

Experimental model and study participant details

The CREAM-KIDS Consortium is an international collaboration combining ophthalmologic and lifestyle data from 20 school-based or population-based pediatric cohort studies conducted across Asia, Europe, and Australia. In this study, we included children aged 4 to 9 years at baseline (Note that the majority (90.1%) of participants had only one visit from cross-sectional designed original pediatric studies), who had undergone cycloplegic auto-refraction. Participants were excluded if they had any ocular disease other than refractive error (e.g., glaucoma, amblyopia, strabismus), a history of ocular surgery, or missing or extreme refraction values (i.e., spherical or cylindrical power greater than +10.00 D or less than −10.00 D, which were considered likely measurement errors). For the analysis of hyperopic reserve distributions, children who were myopic at baseline were further excluded. For the analysis of hyperopic reserve consumption, only participants with follow-up data were included. To evaluate myopia screening programs, all eligible subjects with both cycloplegic and non-cycloplegic auto-refraction measurements were analyzed. The overall mean (SD) age was 6.94 (1.12) years, 52.33% of participants were male, and 47.67% were female.

Ethics statement

De-identified data from the individual studies were harmonized into the CREAM-KIDS dataset, which was used for the analyses in the present study. Each pediatric cohort study within the CREAM-KIDS Consortium was conducted in accordance with the tenets of the Declaration of Helsinki and complied to its IRB requirement. Specific IRB included: The Generation R study protocol was approved by the Medical Ethical Committee of the Erasmus Medical Center, Rotterdam (MEC 217.595/2002/20); the TU Dublin Research Ethics Committee; Ulster University’s Research Ethics committee (REC/18/0102); the Human Research Ethics Committee of the University of Sydney, the New South Wales Department of Education and Training and the Catholic Education Office; the Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethic Committee (CREC Ref. No.: 2015.033); the Regional Committees for Medical Research Ethics in South East Norway; the Singapore Eye Research Institute Ethics Committee; Kings College London Research Ethics Committee (References: PNM/09/10–104 and HR/DP-20/21–22060).

Method details

This study used harmonized data from the CREAM-KIDS Consortium, which were merged, cleaned, and curated by S.K. and D.H. The data were analyzed cross-sectionally to describe the distribution of hyperopic reserve and evaluate myopia screening performance, and longitudinally to examine changes in hyperopic reserve over time.

Individual study designs and sampling methods are provided in the Supplementary Information (Table S1).^{27,28,29,30,31,32,33,34} The included individual studies and their recruitment were summarized as: 1) Generation R (Gen R, 2008–2024), is a population-based prospective cohort study in Rotterdam the Netherlands designed to examine environmental and genetic factors influencing health, growth, and development from fetal life through young adulthood. The cohort includes 9,778 mothers with delivery dates between April 2002 and January 2006, and then following the parents and their children until the children reached 18-year-old. 2) Ireland Eye Study (IES, 2015–2018) is a cross-sectional, population-based epidemiological study designed to assess the prevalence of refractive error and visual function in school-aged children across Ireland. The study recruited 1,626 children aged 6–7 and 12–13 years using stratified random cluster sampling, with schools serving as clusters. 3) Northern Ireland Childhood Errors of Refraction Study (NICER, 2006–2016) is a prospective population-based cohort study designed to investigate the prevalence, progression and determinants of refractive error among children and young adults in Northern Ireland. The study used stratified, random cluster sampling to recruit a representative sample of school children aged 6–7 years (younger cohort) and 12–13 years (older cohort) between 2006 and 2008 (baseline) from both rural/urban and deprived/non-deprived areas of Northern Ireland. 4) Southeast Norway Vision and Visuomotor Study (SNOW, 2015–2024) is a school-based prospective cohort study designed to examine development of refractive errors, vision and motor skills, and how these interact with health and learning in children and adolescents aged 7 to 16 years. All children in 2^nd (7–8 years), 5^th (10–11 years) and 10^th (15–16 years) grade in all community schools of the Kongsberg municipality in Norway were recruited. 5) Twins Early Development Study (TEDS, 1994–2019) is a large-scale longitudinal study of twins from early childhood through adolescence. More than 15,000 pairs of twins were enrolled in TEDS identified from birth records of twins born in the United Kingdom between 1994 and 1996. 6) Sydney Myopia Study (SMS, 2003–2012) is a population-based random cluster-sample of children in two age cohorts, 6 years (n = 1765) and 12 years (n = 2353) from 55 schools across the Sydney metropolitan area. 7) Hong Kong Children Eye Study (HKCES, 2014–2021) is an on-going population-based cohort study, started from March 2015, designed to assess environmental and genetic factors associated with pediatric ocular diseases, particularly myopia, among school aged children. Participants were recruited across Hong Kong, using a stratified random cluster sampling method from all primary schools registered under the Education Bureau. 8) Singapore Cohort of Risk factors for Myopia (SCORM, 1999–2004) is a prospective cohort where children from grades 1 to 3, aged 7 to 9 years, were recruited from three Singapore schools.

Outcome and measurements

The primary outcome of this study was hyperopic reserve, expressed in diopters (D), defined as the spherical equivalent refraction (SER) among children who were non-myopic at baseline. Myopia was defined as SER ≤ −0.50 D. SER was calculated as the spherical power plus half of the cylindrical power. Refraction was performed with auto-refractor, with or without cycloplegia (cyclopentolate 1.0%). For the first two analyses (hyperopic reserve distributions and hyperopic reserve consumption), SER referred to cycloplegic SER. For the myopia screening analysis, non-cycloplegic SER was used for screening, while cycloplegic SER served as the diagnostic (gold-standard) measure.

Quantification and statistical analysis

This study comprised three main analytical components. In the first part, we compared age-specific mean hyperopic reserve (D) by age across groups defined by region/country using line plots and described the distributions of hyperopic reserve among 7-year-old children in each group using histograms. In the second part, we examined consumption of hyperopic reserve by age and region using line plots. Hyperopic reserve consumption per year was further analyzed with univariable and multivariable linear regression models including baseline age, sex, baseline cycloplegic SER, and region as covariates. Hyperopic reserve in the first and second part of analyses was based on cycloplegic measurements. In the third part, we compared the sensitivity and specificity of non-cycloplegic SER measurements to identify myopic children against the gold standard, cycloplegic SER, stratified by age and region.

When analyzing hyperopic reserve distributions, both China’s Hong Kong and Singapore datasets had sufficiently large sample sizes, allowing us to examine them separately. However, the Hong Kong dataset did not include follow-up data, and the Singapore dataset lacked non-cycloplegic measurements required for evaluating screening performance. Therefore, only Singapore data were used to represent Asia in the analysis of hyperopic reserve consumption, while Hong Kong data were used to represent Asia in the myopia screening analysis. In addition, we found that European and Australian children showed similar hyperopic reserve distributions, so these two regions were combined into a single group for later analyses.

In the present study, the unit of analysis was the right eye (Pearson correlation of cycloplegic SER between left and right eye was 0.93). Results of the continuous variables were summarized using the mean and standard deviation (SD). Comparisons between two groups were conducted using t test, while comparisons among three or more groups were performed using ANOVA test. Categorical variables were described as frequencies (N) and percentages (%), and comparisons were made using the chi-squared test. All analyses were conducted using R Statistical Software (v4.3.2; R Core Team 2021, r-project.org), Two-sided p-values <0.05 were considered statistically significant.