Trends in myopia prevalence among late adolescents in South Korea: a population-level study and future projections up to 2050

Abstract

Objective

This study aims to provide updated prevalence estimates of myopia and high myopia among late adolescent men in Seoul, South Korea, and predict future trends up to 2050.

Methods and analysis

This cross-sectional and population-level study includes late adolescent men of the same age who underwent a series of medical examinations at the Seoul Regional Military Manpower Administration between 2013 and 2022. The population with myopia and high myopia was estimated, and the prevalence for 2050 was forecasted. Associated risk factors and ocular disease status of the high myopia population were investigated.

Results

Over the 10-year period, the prevalence of myopia and high myopia among late adolescent men in Seoul was 70.67% and 20.29%, respectively. Between 2013 and 2022, the prevalence of the myopia and high myopia has increased significantly (p<0.001 and 0.006, respectively). The annual growth rate for the prevalence of myopia and high myopia was 0.61% and 0.33%, respectively. Regression analyses predicted that by 2050, myopia and high myopia prevalence will reach 90.90% and 31.26% by linear regression, and 90.75% and 31.17% by non-linear regression, respectively. Risk factor analysis identified that a high education level was associated with a higher prevalence of high myopia. Retinal detachment was significantly more common among those with high myopia, while retinal dystrophy was less common.

Conclusion

This study highlights a concerning trend of increasing myopia and high myopia prevalence, which is likely to reach 90.90% and 31.26% by 2050, with significant future burden for public health and society.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• The prevalence of myopia and high myopia is growing rapidly worldwide. South Korea has emerged as one of the leading countries experiencing in a steep rise in myopia prevalence, as validated by numerous studies.

WHAT THIS STUDY ADDS

• This study predicted that by 2050, the prevalence of myopia and high myopia in South Korea will escalate to 90.90% and 31.26%, respectively.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study suggests that by 2050, most urban Korean late adolescent men will be myopic, and this will be accompanied by an increase in the prevalence of retinal detachment. Therefore, this study raises the need to actively perform ophthalmologic screening at a younger age in the future.

Introduction

The prevalence of myopia and high myopia is growing rapidly worldwide.¹ By 2050, it is estimated that half of the global population will be affected by myopia, and 9.8% of the global population will suffer from high myopia.² Past studies have reported that the prevalence of myopia is increasing rapidly in the modern era, especially among those currently in the younger age group.³ This growing trend of myopia has important clinical implications because degenerative changes in the sclera, choroid and retinal tissues can lead to a wide range of serious eye diseases.⁴,⁷ South Korea has emerged as one of the leading countries experiencing a rise in myopia prevalence, as validated by numerous studies.⁸,¹² Investigating the trajectory of myopia prevalence in South Korea provides valuable insights into future global trends.

It is believed that the factors influencing myopic changes may vary by age. During childhood and adolescence, a period characterised by significant bodily growth and numerous developmental changes, it is essential to clearly identify the risk factors for myopia specific to this age group, distinguishing them from those affecting adults. In South Korea, a conscription system mandates that all late adolescent men of Korean nationality who were born in the same year undergo a physical examination in a specific year. As part of this examination, an ophthalmologic assessment is also performed, which allows for the observation of the prevalence of myopia in this demographic at the population level. Therefore, data from the Military Manpower Administration (MMA) were chosen for investigation. However, the data have one major limitation: according to MMA guidelines, a refraction test was only performed if the uncorrected visual ability (UCVA) of an eye was 0.3 in decimal or worse. This means that if both eyes had a UCVA better than 0.3, the refraction test was not performed. This limitation is not significant for estimating the prevalence of high myopia, as high myopia is rarely associated with a UCVA greater than 0.3. However, this is a significant limitation for estimating the prevalence of myopia. Previous studies have underestimated myopia prevalence by not counting myopic individuals without measured refractive error or have overestimated it by including only those with refractive error data in the study population, thus misreporting the majority as myopic.⁸¹²,¹⁴ In this study, a formula was derived to estimate the prevalence of myopia in a population without a refraction test, based on statistical results from Korea National Health and Nutrition Examination Survey (KNHANES) 2019–2021. This formula was then applied to MMA data, enabling a more accurate determination of myopia prevalence. Furthermore, since it is impossible to accurately assess risk factors related to myopia using this data, the risk factor analysis was conducted only for high myopia.

High myopia typically refers to a refractive state where the spherical equivalent (SE) is −5.0 or −6.0 diopters or lower.¹⁵ Multiple studies suggest that although the definitions of high myopia and pathological myopia vary, pathological symptoms are uncommon in mild-to-moderate myopia but increase significantly in cases of myopia greater than –5.0 to –6.0 diopters.^{6 7 16} Notably, clinically significant pathological changes can be observed in middle-aged patients or even in those who are younger.¹⁷ Considering this, the MMA records were analysed to investigate the association between ocular disease diagnosis status and high myopia among the participants.

Therefore, this study aims to update the current prevalence based on the most recent data available and to predict the prevalence of myopia in Korea by 2050. Accordingly, a study was undertaken to determine the prevalence of myopia among individuals at the Seoul Regional MMA, focusing on a demographic of late adolescent men conscripts. In parallel, associated risk factors and ocular diseases were explored by analysing a range of available data, including height, weight, body mass index (BMI), education level and results from blood tests.

Materials and methods

Study design

This cross-sectional study includes late adolescent male participants of the same age who underwent a series of medical examinations at the Seoul Regional MMA between 1 January 2013 and 31 December 2022. The data were collected through formal request based on Freedom of Information Act.

Data collection and selection of research participants

At the MMA, non-cycloplegic autorefraction is mandatory for individuals with an UCVA of 0.3 or less. For other individuals, non-cycloplegic autorefraction is optional. For instance, if a participant with a UCVA better than 0.3 requests the test, it will be performed. As a result, the MMA data includes refractive measurements for individuals with a UCVA of 0.3 or less, as well as for those who specifically requested it, even if their UCVA was better than 0.3.

The inclusion criteria were restricted to individuals undergoing military health examinations for the first time at the institution. This criterion ensures that all participants are of the same age. In cases where a participant underwent multiple tests, only the initial test was considered to avoid repetition and maintain consistency in the sample and data. The study encompassed both individuals who underwent non-cycloplegic autorefraction and those who did not, to accurately estimate the prevalence of myopia at the population level.

For risk factor analysis, all measurement data available from the study population were collected, including physical measurements and blood test results; height, weight, BMI, serum glucose, serum triglycerides (TG), serum total cholesterol, aspartate aminotransferase (AST), alanine aminotransferase (ALT), blood urea nitrogen (BUN), serum creatinine (Cr), glomerular filtration rate (GFR), white blood cell count (WBC), red blood cell count (RBC), haemoglobin (Hb), platelets (PLT) and haematocrit (HCT). Furthermore, the data for the year of the test and the education level of these individuals were collected. The education level was categorised based on the type of school the participants ultimately attended: high school graduate or less, students in 2-year to 3-year colleges and students in 4-year to 6-year universities.

Determination of prevalence

Based on the data collected over a 10-year period, the prevalence of myopia and high myopia for each year was calculated. A person was classified as having myopia if the person had an SE status of −0.5 or more diopters of myopia in either eye, or as having high myopia if the person had an SE status of −6.0 or more diopters of myopia in either eye. While this criterion was adequate for identifying the prevalence of high myopia, there was still a challenge in determining the prevalence of myopia. As mentioned above, non-cycloplegic refraction was performed on individuals with a decimal UCVA of 0.3 or less in either eye. Therefore, unless requested, individuals with UCVA better than 0.3 did not undergo refraction even if myopia was present.

Past research indicates that UCVA is highly sensitive and specific for detecting myopia, and the relationship between increasing SE of myopia and decreasing UCVA is quite strong.¹⁸ To estimate the myopic population based on decimal UCVA in cases where refractive values were not investigated, the following formula was applied. This formula was derived from statistical results of KNHANES 2019–2021. The relationship between UCVA and refractive error was investigated in adolescents aged between 10 and 19 (online supplemental figure 1). The proportion of myopia for each UCVA category was calculated (online supplemental table 1), and the following formula was subsequently derived.

$${\displaystyle Myopiapopulation=\begin{array}{l}\frac{3}{5}\left(populationwithdecimalUCVAof0.4to0.5\right)\\ +\frac{1}{4}\left(populationwithdecimalUCVAof0.6to0.7\right)\\ +\frac{1}{5}\left(populationwithdecimalUCVAof0.8to0.9\right)\\ +\frac{1}{10}\left(populationwithdecimalUCVAof1.0\right)\end{array}}$$

Based on the above formula, the prevalence of myopia can be roughly estimated based on the UCVA values in a subset of individuals whose refractive error was not measured. On the other hand, this adjustment was not necessary for estimating the high myopia population, as it is highly unlikely that a person with UCVA of better than 0.3 would have high myopia.¹⁹

From 2013 to 2022, the myopia population can be calculated using the following process: first, determine the confirmed myopia cases using available refraction data. Next, perform a UCVA-based adjustment for individuals without refraction data to estimate additional myopia cases. The final myopia population is obtained by summing these confirmed and estimated cases, as shown in the equations below.

$${\displaystyle Myopiapopulation=\begin{array}{l}Myopiapopulationwithrefractiondata\\ +Myopiapopulationwithoutrefraction\\ data\left(UCVA-basedadjustment\right)\end{array}}$$

Regression analysis for prevalence prediction

Based on over a decade of prevalence data derived from this process, the best fit linear and non-linear regression models to predict the prevalence of myopia and high myopia in 2025, 2030, 2035, 2040, 2045 and 2050 were explored. Through regression analyses, the prevalence curve to equations were fitted with the following structures.

$${\displaystyle Linearregression:Prevalence=a\times Year+b}$$

$${\displaystyle Non-linearregression:Prevalence=a\times \log \left(Year\right)+b}$$

The parameters, a and b, were derived from each regression analysis. Similar to other biological prevalence rates, it is assumed that the increase in myopia prevalence would decay as it approaches 100%, and it was therefore concluded that logarithmic regression is the most appropriate model among non-linear regression approaches. The regression model parameters were selected only if the statistical significance of the model and the statistical significance of the parameters were satisfied (online supplemental table 2).

We consider the myopia population, including the UCVA-based adjustment, to be a more accurate representation of the true myopia prevalence. However, considering the inaccuracy of the UCVA-based adjustment, we also predicted future prevalence based solely on the myopia population with available refraction data.

Associated risk factors and ocular disease diagnosis status

The study population was divided into two groups: those with high myopia and those without high myopia. Education levels, physical measurements and blood test results were considered as independent variables, and a statistical analysis was conducted to determine their association with the prevalence of high myopia. In addition, it was investigated whether the rates of ocular disease conditions, including retinal detachment (RD), surgically removed cataract, nystagmus, glaucoma, retinal dystrophy and non-glaucomatous visual field defect, differed between those two groups.

Statistical analysis

Statistical analyses were performed using R V.4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). The annual growth estimate of prevalence and the significance of increasing pattern were evaluated by linear regression analysis. The differences in high myopia prevalence values among the years and among the education levels were evaluated with the χ² test for equality of proportions. For comparisons of median values of measurements between high myopia group and non-high myopia group, Mann-Whitney U test was performed. To analyse the risk factors of high myopia, the entire study population was divided into two groups based on the median value of each variable: at/above median group and below median group. Simple univariate logistic regression analysis for high myopia was performed to assess risk factors. Risk factors with a p value of less than 0.1 were identified as potential risk factors. Subsequently, these selected candidates underwent multiple logistic regression analysis for high myopia. During this analysis, variables exhibiting a variance inflation factor (VIF) greater than 5 were excluded. The OR and 95% CI values for high myopia were reported. To avoid false positives from big data analysis, ORs of either 0.9 or higher or 1.1 or lower were considered clinically insignificant, even if they were statistically significant. The difference in the rates of ocular disease diagnosis between the high myopia group and the non-high myopia group was evaluated with a χ² test for equality of proportions. A p value<0.05 was considered statistically significant.

Patient and public involvement

The study participants did not participate in the design or conduct of this study; however, their clinical information was used. After publication, the findings regarding the prevalence of myopia in South Korea will be shared with the MMA to help guide future screening and diagnosis.

Results

Estimated prevalence of myopia and prevalence of high myopia in late adolescent male South Korean population

This study included a total of 519,703 participants of the same age. Of these, 367,287 (70.67%) were estimated to have myopia, and 105 439 (20.29%) had high myopia. Between 2013 and 2022, the estimated prevalence of myopia and the prevalence of high myopia significantly increased (p<0.001 and p=0.006), with annual growth rates of 0.61% and 0.33%, respectively. Among the 10-year total population, 58.93% had UCVA of 0.3 or less, and this significantly increased over the ten years with an annual growth rate of 0.75% (table 1 and figure 1).

Table 1. Prevalence of myopia and high myopia in late adolescent male South Korean population and annual prevalence growth.

	Total	UCVA of 0.3 or less (%)	Myopia (%)			High myopia (%)
			With available refraction data	UCVA-based adjustment	Adjusted prevalence
Year
2013	64 652	36 003 (55.69)	36 971 (57.18)	7144 (11.05)	44 115 (68.23)	12 414 (19.20)
2014	62 407	34 675 (55.56)	35 629 (57.09)	6647 (10.65)	42 276 (67.74)	12 038 (19.29)
2015	58 663	32 775 (55.87)	35 116 (59.86)	5476 (9.33)	40 592 (69.20)	10 544 (17.97)
2016	56 747	33 108 (58.34)	35 065 (61.79)	5092 (8.97)	40 157 (70.76)	11 213 (19.76)
2017	52 615	32 229 (61.25)	33 107 (62.92)	4680 (8.89)	37 787 (71.82)	11 331 (21.54)
2018	50 488	30 838 (61.08)	31 633 (62.65)	4511 (8.93)	36 144 (71.59)	10 627 (21.05)
2019	51 918	31 968 (61.57)	32 639 (62.87)	4691 (9.04)	37 330 (71.90)	10 995 (21.18)
2020	43 957	26 948 (61.31)	27 600 (62.79)	3887 (8.84)	31 487 (71.63)	9234 (21.01)
2021	38 904	23 621 (60.72)	25 340 (65.13)	3061 (7.87)	28 401 (73.00)	8368 (21.51)
2022	39 352	24 120 (61.29)	25 922 (65.87)	3076 (7.82)	28 998 (73.69)	8675 (22.04)
Total	519 703	306 285 (58.93)	319 022 (61.39)	48 265 (9.29)	367 287 (70.67)	105 439 (20.29)
Annual growth		0.75%	0.92%	−0.31%	0.61%	0.33%
Adjusted R2		0.71	0.87	0.80	0.87	0.63
P trend		0.001	<0.001	<0.001	<0.001	0.006

Adjustments to the prevalence were made using UCVA data where refraction data were lacking.

UCVA, uncorrected visual ability

Figure 1. Prevalence of myopia and high myopia for each year, along with projections up to 2050. The predicted prevalence points include 95% CIs. (A) Linear regression analysis for predicting myopia and high myopia prevalence based on a decade of prevalence data from 2013 to 2022. (B) Non-linear regression analysis for predicting myopia and high myopia prevalence based on a decade of prevalence data from 2013 to 2022.

Prevalence of myopia with available refraction data and UCVA-based adjustment

The prevalence of myopia, based on available refraction data from 2013 to 2022, was 61.39%. An additional 9.29% of the total population was estimated to be myopic based on UCVA data. The prevalence of myopia confirmed through refraction data increased by 0.92% annually, while the estimated myopic population derived from our formula decreased by 0.31% annually (table 1).

Projection of prevalence through 2050

Based on a decade of data, regression analyses were performed, and the most suitable model coefficients for predicting future prevalence were determined. Equations from linear regression analyses are as follows.

$${\displaystyle \begin{array}{l}PredictedPrevalenceofMyopia\\ \left(withavailablerefractiondata\right)\end{array}=-1797.8+0.9217\times Year}$$

$${\displaystyle \begin{array}{l}PredictedPrevalenceofMyopia\\ \left(afterUCVAbasedadjustment\right)\end{array}=-1167.0+0.6138\times Year}$$

$${\displaystyle PrevalenceofHighMyopia=-646.41746+0.33057\times Year}$$

Equations derived from non-linear regression analyses are as follows.

$${\displaystyle \begin{array}{l}PrevalenceofMyopia\\ \left(withavailablerefractiondata\right)\end{array}=-14090.5+1859.8\times log\left(Year\right)}$$

$${\displaystyle \begin{array}{l}PrevalenceofMyopia\\ \left(afterUCVAbasedadjustment\right)\end{array}=-9352.8+1238.4\times log\left(Year\right)}$$

$${\displaystyle PrevalenceofHighMyopia=-5054.5+666.9\times log\left(Year\right)}$$

According to linear regression analysis, the predicted prevalence of myopia by 2050 is 90.90%, and for high myopia, it is 31.26%. (table 2 and figure 1A) Non-linear regression analysis predicts a myopia prevalence of 90.75% and a high myopia prevalence of 31.17%. (table 2 and figure 1B) The predicted prevalence of myopia in 2050, using only available refraction data, is even higher in both linear (91.77%) and non-linear models (91.54%). Both linear and non-linear models exhibited good model fitness and high explanatory power (online supplemental table 2).

Table 2. Projected prevalence rates and their 95% CIs of myopia and high myopia from 2025 through 2050 using linear and non-linear regression analysis.

Year	Myopia with available refraction data		Myopia after UCVA-based adjustment		High myopia
	Linear (%)	Non-linear (%)	Linear (%)	Non-linear (%)	Linear (%)	Non-linear (%)
2025	68.73 (66.57 to 70.89)	68.72 (66.57 to 70.87)	75.56 (74.13 to 76.99)	75.55 (74.12 to 76.98)	23.00 (21.33 to 24.66)	22.99 (21.33 to 24.66)
2030	73.34 (69.89 to 76.78)	73.31 (69.87 to 76.74)	78.63 (76.34 to 80.92)	78.61 (76.33 to 80.89)	24.65 (21.99 to 27.31)	24.64 (21.98 to 27.29)
2035	77.95 (73.19 to 82.71)	77.88 (73.14 to 82.62)	81.70 (78.54 to 84.86)	81.65 (78.51 to 84.80)	26.30 (22.62 to 29.98)	26.28 (22.61 to 29.94)
2040	82.56 (76.47 to 88.65)	82.45 (76.39 to 88.50)	84.77 (80.72 to 88.81)	84.69 (80.67 to 88.71)	27.95 (23.25 to 32.66)	27.91 (23.23 to 32.59)
2045	87.16 (79.74 to 94.59)	87.00 (79.63 to 94.37)	87.84 (82.91 to 92.76)	87.72 (82.83 to 92.62)	29.61 (23.87 to 35.34)	29.55 (23.85 to 35.24)
2050	91.77 (83.01 to 100.53)	91.54 (82.85 to 100.22)	90.90 (85.09 to 96.72)	90.75 (84.98 to 96.51)	31.26 (24.49 to 38.03)	31.17 (24.46 to 37.89)

The projection was made based on decadal prevalence data from 2013 to 2022.

UCVA, uncorrected visual ability

Education level and prevalence of high myopia

The prevalence of high myopia was significantly different between education levels (p<0.001). The prevalence of high myopia was 15.09% for high school graduate or less education, 17.68% for students in 2-year to 3-year colleges and 23.69% for students in 4-year to 6-year universities (figure 2).

Figure 2. Comparison of high myopia prevalence by education level. Each bar represents the prevalence of high myopia for each education level, with an asterisk (*) indicating a statistically significant difference between groups (p<0.001).

Potential risk factors of high myopia

The median values of physical measurements and blood test results for the high myopia and non-high myopia groups are shown in online supplemental table 3. There were no significant differences in ALT (160 IU/L vs 160 IU/L, p=0.058) and Hb (161 g/dL vs 161 g/dL, p=0.050) between the two groups. In all other cases, except for BUN (11.8 mg/dL vs 12 mg/dL, p<0.001) and Cr (0.88 mg/dL vs 0.89 mg/dL, p<0.001), the high myopia group had higher median values. Specifically, significant differences were found in height (174 cm vs 174 cm, p<0.001), weight (68.4 kg vs 68.1 kg, p<0.001), BMI (22.5 kg/m² vs 22.5 kg/m², p=0.005), glucose (18 mg/dL vs 18 mg/dL, p=0.046), TG (17 mg/dL vs 16 mg/dL, p<0.001), cholesterol (96 mg/dL vs 96 mg/dL, p<0.001), AST (94 IU/L vs 93 IU/L, p<0.001), GFR (111.56 mL/min vs 111.33 mL/min, p<0.001), WBC (6.82×10⁹/L vs 6.8×10⁹/L, p<0.001), RBC (5.35×10¹²/L vs 5.33×10¹²/L, p<0.001), PLT (47.2×10⁹/L vs 47.1×10⁹/L, p<0.001), and HCT (239% vs 237%, p<0.001).

The ORs and 95% CIs for high myopia based on median values of physical measurements and blood test results derived from simple and multiple univariate logistic regression analyses are shown in online supplemental table 4. In a simple regression analysis, all variables were potential factors associated with the occurrence of high myopia (p<0.1) except for AST (p=0.963). In a subsequent multiple regression analysis, AST was excluded for this reason. To address the multicollinearity issue, height, weight, BMI, BUN and Cr were also excluded (VIF>5). According to this final multiple regression model, glucose (OR: 1.060, 95% CI: 1.040 to 1.079), TG (OR: 1.038, 95% CI: 1.018 to 1.058), ALT (OR: 1.065, 95% CI: 1.045 to 1.085), GFR (OR: 1.097, 95% CI: 1.077 to 1.118), WBC (OR: 1.025, 95% CI: 1.006 to 1.045), RBC (OR: 1.094, 95% CI: 1.069 to 1.119) and PLT (OR: 1.022, 95% CI: 1.003 to 1.042) were positively associated with the occurrence of high myopia. In contrast, cholesterol (OR: 0.949, 95% CI: 0.931 to 0.967) and Hb (OR: 0.966, 95% CI: 0.943 to 0.989) showed an inverse association with high myopia. HCT (OR: 1.018, 95% CI: 0.994 to 1.042) was not significantly associated with high myopia.

High myopia and ocular diseases

The rates of ocular disease diagnosis in the high myopia group and the non-high myopia group are shown in online supplemental table 5. RD was more frequently diagnosed in the high myopia group (0.096% vs 0.028%, p<0.001), and retinal dystrophy was less frequently diagnosed (0.001% vs 0.010%, p=0.008). There was no difference in the diagnosis of surgically removed cataract, glaucoma, nystagmus and non-glaucomatous visual field defect.

Discussion

Our study found that 70.67% of late adolescent men had myopia over the past decade, with a yearly increase of 0.61%. Additionally, 20.29% of these individuals experienced high myopia, which increased annually by 0.33%. By the year 2050, it is anticipated that myopia prevalence will reach 90.90%, with high myopia reaching 31.26%. These projected rates notably exceed the global average. Interestingly, our non-linear regression results were similar to those of linear regression, with the regression line appearing almost linear (figure 1 and online supplemental table 2). This suggests that the trends in myopia and high myopia prevalence have followed a linear pattern thus far. Notably, a recent study on global myopia trends through 2050 also demonstrates a nearly linear increase in both myopia and high myopia prevalence.²

Several studies have highlighted a rising trend in myopia prevalence, particularly in East and Southeast Asia, regions identified as myopia hotspots.²⁰,²⁶ Since this study incorporates the most recent data, it reveals one of the highest myopia prevalence rates compared with earlier studies conducted in these areas.

In this study, a clear association was identified between education level and high myopia (figure 2). There have been consistent reports of links between myopia and the number of years spent in education, as well as academic performance.^{8 11 13 27 28} It has been reported that about half of the relationship between education and myopia may be mediated by children not spending time outdoors during schooling.^{29 30}

The results in online supplemental tables 3 and 4 show that there are statistically significant differences in physical measurements and blood test results between high myopia group and non-high myopia group, although the clinical significance is questionable. To accurately interpret the results of this study, the clinical significance of the difference should be evaluated. Even when a statistical difference exists between two groups, if the magnitude of this difference is clinically insignificant and falls within the range of measurement error, it cannot be concluded that there is a meaningful difference between the groups based on these findings.

The results in online supplemental table 4 show that the ORs and 95% CIs obtained after multiple regression analysis are all in the range of 0.9 and 1.1. Given the characteristics of biological measurements, the median values were used as a reference point, resulting in the division of groups that may exhibit more distinct clinical differences than in any other scenario. However, despite these clear clinical differences between the two groups, the fact that the ORs and 95% CIs identified were all between 0.9 and 1.1 suggests that they actually play little role as risk factors for high myopia. Additionally, the OR tends to overestimate its likelihood more than the relative risk, especially since high myopia is not a rare occurrence.³¹ Therefore, the results of this study did not identify any risk factors with clear clinical significance among the physical measurements and blood test results.

The relationship between myopia and various diseases has been studied extensively.³⁰ Myopia is commonly known to be associated with RD and glaucoma in previous research.³²,³⁴ In this study, it was found that late adolescent men with high myopia have a 1 in 1000 chance of developing RD, which is about 3.4 times higher than those without high myopia (online supplemental table 5). This suggests that RD may increase significantly as the prevalence of high myopia increases.

On the other hand, this study could not identify a clear difference in the occurrence of glaucoma between the high myopia group and the non-high myopia group (p=0.070). Given the age of our study population, most of the glaucoma identified in this study was juvenile open-angle glaucoma (JOAG). Based on previous studies of JOAG and myopia, myopia seems to be a clear correlate of JOAG.^{35 36} However, this study used high myopia rather than myopia as the criterion, which could be one of the reasons that this study could not identify the same results.

In Korea, many studies have investigated the prevalence of myopia using KNHANES data.³⁹,¹¹ Studies based on KNHANES data were conducted through sample extraction across various age groups, whereas this study differs in that it analyses the entire population within the same age group. There have also been previous efforts to estimate the prevalence of myopia from the MMA in South Korea.⁸¹²,¹⁴ However, those previous studies had the following limitations. First, previous studies did not adequately account for populations lacking refraction data. Second, previous studies included redundant data from revisitors. However, this study accounts for the myopic population in cases lacking refractive data, thereby preventing the underestimation of myopia prevalence. Moreover, this study maintained the consistency in the sample and data by exclusively focusing on the first-time examinees.

One limitation of this study is the potential inaccuracy in estimating myopia proportions based on UCVA. Additionally, the approximation involved in converting from logMAR to decimal may introduce further inaccuracies. However, compared with previous studies that did not address these aspects, this approach facilitates a more accurate prediction of myopia prevalence.

There are several other limitations in this study. First, this study is a cross-sectional study and is primarily focused on identifying association rather than causation. Second, the participants of this study are entirely male sex and exclusively from Seoul, the most urbanised city in South Korea, making the results not fully representative of the entire South Korea population. Previous studies have shown that myopia prevalence varies with the residential area,¹⁴ suggesting that the prevalence rates found in the study may be slightly higher than the national average.

Third, although the MMA requires individuals who undergo orthokeratology to be tested only after the effect has worn off, other myopia suppression tools and the impact of refractive surgeries were not considered. Last, since this study used only non-cycloplegic autorefraction data, it may overestimate myopia compared with a study with cycloplegic refraction data. According to research by the International Myopia Institute, it is recommended to use a slightly more myopic threshold, such as −0.75 diopters, particularly in studies that use non-cycloplegic refraction, to avoid including non-myopic individuals.³⁷ However, to ensure comparability with previous studies, we used the commonly accepted −0.50 diopters threshold for myopia, which may have included some non-myopic individuals.

Despite the above limitations, this study provides insights into the future impact of myopia on our society using 10-year population data based on the most recent data available. This extensive dataset enhances the statistical power of the study and offers clues about future myopia trends.

In conclusion, the prevalence of myopia and high myopia among late adolescent men in Korea is currently estimated at 70.67% and 20.29%, respectively. These rates are on a steady rise and are projected to reach 90.90% and 31.26% by 2050. A high education level was identified as a clear risk factor for high myopia, while other physical measurements and blood test results were inconclusive. RD was significantly more common among those with high myopia, while retinal dystrophy was less common. The rising prevalence of myopia is likely to result in a substantial burden on society in the future because of its associated conditions. Therefore, greater efforts to actively research and prevent the risk factors associated with myopia are needed.

Data availability statement

Data are available on reasonable request.