Myopia Development Among Young Schoolchildren: The Myopia Investigation Study in Taipei

Abstract

Purpose: To investigate the annual incidence of myopia and associated factors among young schoolchildren in Taipei City.

Methods: The Myopia Investigation Study in Taipei was a citywide, population-based cohort study. During the fall 2013 semester (baseline), a total of 11,590 grade 2 schoolchildren completed ocular examination and were included for further analysis. A parent-completed questionnaire was administered to collect data on risk factors for myopia development. Follow-up visits were arranged biannually over 3 years. The first-year results are reported here. Schoolchildren who were emmetropic/hyperopic at baseline and had myopia (spherical equivalent ≤ −0.5 diopters) in either eye at follow-up were identified as having incident myopia.

Results: Among 7376 baseline nonmyopic participants, 6794 (92.1%) were examined during the first-year follow-up, and 1856 (25.2%) with incident myopia were identified. The incidence density of myopia was 31.7 (95% confidence interval [CI]: 30.6–32.8) per 100 person-years. Cox hazard proportional regression analysis revealed that participants who were emmetropic at baseline (hazards ratio [HR]: 19.37; 95% CI: 4.84–77.57), who had two myopic parents (HR: 1.21; 95% CI: 1.04–1.42), and who spent ≥5 hours every week on after-school tutoring programs (HR: 1.12; 95% CI: 1.02–1.22) had greater risk for incident myopia. By contrast, protective factors included suburban area of residence (HR: 0.91; 95% CI: 0.83–1.00) and spending ≥30 minutes outdoors after school every weekday (HR: 0.90; 95% CI: 0.82–0.99).

Conclusions: This study provides population-based data on the annual incidence of myopia among Taiwanese schoolchildren, and found that baseline refractive status, parental myopia, area of residence, time outdoors after school on weekdays, and time spent on after-school tutoring programs are associated with risk of new-onset myopia.

There are several lines of evidence demonstrating that myopia, the most common refractive error, has become increasingly prevalent over the past few decades.^1–4 In the urban areas of East Asia, myopia affects more than 80% of schoolchildren and young adults.^4–8 Although refractive error problems are usually correctable, myopia is still a major public health issue worldwide. Visual impairment can be caused by uncorrected or undercorrected refractive error. Moreover, the ocular comorbidities associated with high myopia (≤−6.0 diopters [D]) may increase the risk of irreversible loss of vision later in life and carry a heavy socioeconomic burden.² 

It has been reported that myopes with a younger age of onset may show faster myopic progression and are more likely to have high myopia in adult life.^9–13 Therefore, understanding the development of myopia in young children and associated risk factors is essential to the prevention of myopia in the future. Despite an abundance of cross-sectional studies on school myopia prevalence in the literature, there is a scarcity of data on incident myopia from large-scale, population-based longitudinal cohort studies in this age group. The existing data showed a wide variation in incidence of school myopia.^14–26 In Northern Ireland, the annual incidence of myopia among Caucasian children aged 6 to 7 years was 2.2%.²³ In Singapore, the 10-month cumulative incidence was as high as 33.6% among predominantly Chinese ethnic children aged 7, 9, and 12 years.²¹ 

Several cross-sectional surveys involving schoolchildren of different ages in Taiwan have reported significant increases in myopia prevalence.^4,27 Nevertheless, no population-based, longitudinal study regarding incidence of childhood myopia and its risk factors has been reported in Taiwan so far. The mean refractive status among Taiwanese schoolchildren was found to be myopic by grade 3 (the age of 9 years).²⁸ Therefore, the schoolchildren around this age were expected to experience the highest incidence of myopia. To assess the development of myopia among these young schoolchildren, the Myopia Investigation Study in Taipei (MIT) was designed to provide a citywide eye examination program for grade 2 students of all elementary schools in Taipei City, with biannual follow-up examinations of this school cohort over 3 years.²⁹ This paper describes the first-year follow-up data from the MIT study and reports the incidence density and risk factors of new-onset myopia among the nonmyopic participants at baseline. 

Methods

Study Design

The MIT study was funded by the Taipei City Government and approved by the Institutional Review Board of the Taipei City Hospital (TCHIRB-1020501). Initiated in 2013, the MIT study recruited participants from grade 2 students of all 153 elementary schools in Taipei City for 3 consecutive years to prospectively investigate the development of myopia among school-aged children. Written informed consent was obtained from either parent of each child and the Declaration of Helsinki was adhered to throughout. 

The methodology of the MIT study has been formally described elsewhere.²⁹ In brief, all grade 2 schoolchildren in Taipei City were invited to participate in the MIT study. A total of 20 hospitals and 54 private clinics joined the citywide program, collaboratively providing eye examinations for eligible participants. To meet a consistent standard operating procedure, the MIT organizing committee formulated a standardized protocol for staff training and held a series of coordination courses for participating ophthalmologists and other registered medical practitioners. In addition, a quality control committee regularly visited the MIT-associated hospitals/clinics to evaluate eye examination procedures during the survey campaign. It was requested that all examinations be carried out according to the standardized protocol. The baseline examination of the grade 2 schoolchildren (born mainly between September 2005 and August 2006) in the 2013 school year was conducted between July 2013 and September 2013, and the first two follow-up examinations of this cohort took place approximately 6 months later (between January 2014 and March 2014) and 1 year later (between June 2014 and October 2014). 

Eye Examinations

Eye examinations focused on the measurement of visual acuity (unaided and best-corrected) and refractive status (before and after cycloplegia). Cycloplegia was achieved with two drops of a cycloplegic agent (1% cyclopentolate, 1% tropicamide, or 0.5% phenylephrine hydrochloride/0.5% tropicamide) administered 10 minutes apart to each eye. There were 20 hospitals/clinics using 1% cyclopentolate, 27 using 1% tropicamide, and 27 using 0.5% phenylephrine hydrochloride/0.5% tropicamide. These drugs were chosen based on hospital/clinic preferences. Thirty minutes after the second drop of the cycloplegic agent, a penlight was used to check the pupil light reflex. If the pupil still responded to penlight, an additional 10-minute wait was required before autorefraction. We defined myopia as spherical equivalent (SE) ≤ −0.5 D, emmetropia as −0.5 D < SE < 0.5 D, and hyperopia as SE ≥ 0.5 D. Those who were emmetropic or hyperopic in both eyes at baseline and became myopic in either eye at follow-up were identified as having incident myopia. 

Questionnaire Data

To examine the potential risk factors for myopia development, parent-completed questionnaires were collected at baseline and annual follow-ups. The questionnaire was composed of 38 questions in seven sections, which covered demographics, medical history, parental myopia, time spent on near work and time spent on outdoor activities after school (during weekdays and weekends), reading habits, and eye care–seeking behavior. The time spent on near work and time outdoors during school hours were not measured in the current study, because the curriculum timetable is similar for all early-grade schoolchildren in Taipei. The MIT parent questionnaire was composed of closed-ended questions with two-option (yes/no) responses or a list of ordered choices. Respondents were asked to check the choice they felt was most appropriate.²⁹ 

Statistical Analysis

Databases of both examination variables and questionnaire items were constructed with Microsoft Access database software (Redmond, WA, USA). Only the eye with less SE (less positive or more negative refractive error) in each child was included for myopia analysis (e.g., identifying myopes and calculating average SE at each visit) in the present study. All data were expressed as mean ± standard deviation (SD) or percentage. Statistical analysis was performed using SPSS software (Version 17.0; SPSS, Inc., Chicago, IL, USA). The present study analyzed the data from the first three iterations of the citywide refractive status examinations (baseline and two follow-up studies) to calculate the annual incidence of myopia in the 2013 cohort of grade 2 schoolchildren. The incidence rate for myopia was defined as the number of incident cases during the study period divided by the sum of the length of follow-up period for each child in the study cohort. The length of follow-up period was from baseline examination date to the last follow-up date or to the date that myopia diagnosis was made at follow-up. The SE change over the follow-up period of the eye with less SE at baseline in each child was used to calculate the SE progression rate, which was defined as SE difference divided by the length of the follow-up period. The differences between groups in terms of refractive status and questionnaire items were determined by the independent Student's t-test, ANOVA, or Pearson's χ² as appropriate. To explore the predictors for myopia onset, Cox proportional hazard regression analysis with adjustment for potential confounders at baseline (P value < 0.1 in the independent Student's t-test or Pearson's χ² test) was conducted. Further, multiple linear regression modeling of the SE progression rate was also conducted with the same set of covariates. The significance level for all statistical tests was inferred at a 2-sided P value < 0.05. 

Results

Among all 19,374 students of grade 2 in school year 2013, 11,590 (6135 [52.9%] boys) were finally recruited and analyzed in the MIT study. There was no significant difference in terms of the baseline demographic characteristics between the 11,590 analyzed participants and the whole target population.²⁹ At baseline, 7376 (3772 [51.1%] boys) were nonmyopes. A flowchart illustrating the subjects participating in each stage of the study is provided in the Figure. Of these, 6794 (3471 [51.1%] boys) were reexamined at least once, and the follow-up rate was 92.1%. Table 1 shows the baseline characteristics for participants (n = 6794) and nonparticipants (n = 582) in the follow-up examinations. The distributions of sex and baseline SE were similar between these two groups, although these participants were more likely to reside in suburban areas than nonparticipants (47.8% vs. 42.8%, P = 0.020). 

Figure

 

A flowchart illustrating the subjects participating in each stage of the study.

Table 1

 

Baseline Characteristics for Participants and Nonparticipants in Follow-Up Examinations

 

There was a shift of mean SE toward less positive refractive error among the participants at follow-up visits: 0.38 ± 0.66 D at baseline (n = 6794), 0.17 ± 0.77 D around 6 months later (n = 6368), and 0.01 ± 0.80 D around 1 year later (n = 6089). During the follow-up period, a total of 1856 subjects (919 [49.5%] boys) became myopic. Among them, the SE progression rate was −0.98 ± 0.80 D per year, which was significantly greater than that of the subjects who remained nonmyopic (−0.12 ± 0.52 D per year; P < 0.001). The incidence rate of myopia in grade 2 schoolchildren was 31.7 cases per 100 person-years from 2013 to 2014, with the 95% confidence interval (CI) ranging between 30.6 and 32.8 cases per 100 person-years. 

Table 2 shows the comparisons of baseline characteristics between the participants who became myopic (n = 1856) and those who remained nonmyopic (n = 4938) during 1-year follow-up. Compared to those who remained nonmyopic, the participants with incident myopia had significantly lower baseline SE and were more likely to reside in urban areas, have two myopic parents, spend less than 30 minutes per day outdoors after school on weekdays, and spend 5 hours or more per week on after-school tutoring programs. 

Table 2

 

Baseline Characteristics for Participants With Incident Myopia and Those Who Remained Nonmyopic

 

Table 3 shows the results of Cox proportional hazard regression analysis for incident myopia. After adjusting for baseline characteristics including area of residence, parental myopia, baseline SE distribution, time spent on near work, eye-to-object distance in doing near work, time outdoors after school on weekdays, and time spent on after-school tutoring programs, it was revealed that participants who had baseline SE between −0.5 and 0.5 D (hazards ratio [HR]: 19.37; 95% CI: 4.84–77.57), whose parents both had myopia (HR: 1.21; 95% CI: 1.04–1.42), and who spent ≥5 hours per week on after-school tutoring programs (HR: 1.12; 95% CI: 1.02–1.22) had greater risk for incident myopia. On the other hand, residing in a suburban area (HR: 0.91; 95% CI: 0.83–1.00) and ≥30 minutes per day spent outdoors after school on weekdays (HR: 0.90; 95% CI: 0.82–0.99) were protective factors for incident myopia. In the multiple linear regression analysis (F: 4.494, P < 0.001), the SE progression rate was significantly associated with having two parents with myopia (β: −0.135, P < 0.001) and spending ≥5 hours per week on after-school tutoring programs (β: −0.049, P = 0.006) after adjusting for the above-mentioned baseline characteristics. There was no fundamental difference in the mean SE progression rates among four baseline SE groups (−0.45 ± 1.11 D per year in the group with baseline SE ≥ 3.0 D; −0.43 ± 0.84 D per year in the group with baseline SE ≥ 1.5 D and < 3.0 D; −0.34 ± 0.59 D per year in the group with baseline SE ≥ 0.5 D and < 1.5 D; −0.35 ± 0.76 D per year in the group with baseline SE > −0.5 D and < 0.5 D; P = 0.150). The mean SE progression rate by baseline characteristics is provided in the Supplementary Table. 

Table 3

 

Predictors of Incident Myopia by Cox Proportional Hazard Regression Analysis

 

Discussion

In the current cohort study, our results show that the annual incidence rate of myopia was 31.7% among 7- to 8-year-old schoolchildren in Taipei City and the risk of incident myopia was associated with area of residence, parental myopia, baseline refractive status, time outdoors after school on weekdays, and time spent on after-school tutoring programs. To our knowledge, this study provides the first population-based, longitudinal data on the incidence of childhood myopia in Taiwan. 

Published studies reporting on myopia incidence for schoolchildren are not as numerous as those reporting on myopia prevalence. Table 4 summarizes the major findings of epidemiologic studies on incidence of childhood myopia. Similar to the variability in reported results for myopia prevalence, there is also significant variation in myopia incidence figures across studies, which also differ in terms of study period, region, ethnicity, age, or methodology.^13–26 According to the results of population-based studies using autorefraction, East Asian children appeared to have a higher incidence of myopia than Caucasian children.^{13–15,17,18,22–24} A 1-year longitudinal cohort study in Hong Kong revealed a mean annual incident rate of 14.41% among 3149 children aged between 5 and 16 years at baseline.¹⁸ The annual incidence rates were reported by Saw et al.¹³ to be 15.9%, 12.8%, and 10.8% for 7-, 8-, and 9-year-old Singaporean children (n = 569), respectively. In China, Zhao et al.¹⁷ reported a slightly lower annual incidence rate (7.8%) of myopia among 3899 schoolchildren (aged 5–13 years) in a rural district, and Zhou et al.²⁴ reported a figure of 10.6% among 1591 children of similar age in a district with various urbanization levels. By contrast, Caucasian children were reported to have low and similar annual incidence of myopia in the United States (4.3% for 8- to 9-year-olds),¹⁴ Australia (1.3% for 6- to 7-year-olds; 2.9% for 12- to 13-year-olds),²² and the United Kingdom (2.2% for 6-to 7-year-olds; 0.7% for 12- to 13-year-olds).²³ Of interest, the annual incidence rates for the children of East Asian ethnicity living in the Sydney Metropolitan region (6.9% for 6-to 7-year-olds; 7.3% for 12- to 13-year-olds)²² are comparable to those for rural populations of China and lower than those for children living in other metropolitan cities in East Asia, such as Hong Kong, Singapore, or Taipei, which indicates that environmental and lifestyle differences may play an important role in the development of childhood myopia. 

Table 4

 

Epidemiologic Studies on Incidence of Myopia Among Schoolchildren

 

Several lines of evidence have demonstrated that the rapid escalation of myopia prevalence is a global trend.^{1,2,4,30–32} In Taiwan, the population-based surveys found a 3.6-fold increase in myopia prevalence from 1983 (5.8%) to 2000 (21%) among the 7-year-old schoolchild population.⁴ It is considered that these changes in prevalence were related to intensive environmental pressures.³⁰ However, there is very limited knowledge about the changing pattern of myopia incidence over time among schoolchildren. Fan et al.¹⁸ conducted an initial survey in 1998 and a follow-up study 12 months later among Chinese children in Hong Kong, and the age-specific annual incidence rates of myopia were reported as 13.13% and 14.84% for 7- and 8-year-olds, respectively. These results were similar to those for age-matched Singaporean children in a 3-year cohort study implemented in 1999.¹³ Nevertheless, our study, initiated in 2013, found more than a two times greater annual incidence of myopia (31.7%) in a similar-aged group in Taipei. It seems that the annual incidence of childhood myopia has also increased in the industrialized regions of East Asia over the past few decades. Further studies are warranted to investigate the myopia incidence changes among schoolchildren of other regions and ethnicities and the driving forces behind these changes. 

Recent studies have found a negative association between children's time spent outdoors and the development of myopia.^{14,19,25,26,33} Among 514 grade 3 schoolchildren in the United States, survey-based data revealed that reduced participation in sports and outdoor activity at baseline increased the risk of having myopia by grade 8.¹⁴ In a prospective longitudinal study of 9109 children in Southwest England, greater time spent outdoors at age 8 to 9 years was associated with a reduced incidence of myopia development especially between the ages of 11 and 15 years.¹⁹ In a 5- to 6-year longitudinal follow-up study of a total of 2059 Australian schoolchildren (aged 6–7 years and 12–13 years), children with incident myopia had spent significantly less total time outdoors per week at baseline than those who remained nonmyopic.³³ In a 1-year intervention study in southern Taiwan, staying outdoors during the daily break time for 80 minutes each school day was reported to be helpful in reducing the annual incidence rate of myopia by approximately 50% among schoolchildren aged 7 to 11 years.²⁵ In a randomized clinical trial in Guangzhou, there was also a significant difference in the 3-year cumulative incidence rate of myopia among grade 1 schoolchildren for the six intervention schools (30.4%), where an additional 40-minute outdoor activity class was scheduled at the end of each school day, compared with the six control schools (39.1%).²⁶ Similarly, we found that participants with incident myopia were more likely to have decreased time spent on outdoor activities after school on weekdays than those who remained nonmyopic (<30 minutes per day: 61.0% vs. 54.5%; P < 0.001) and, though the value did not reach significance, on weekends (<2 hours per day: 68.9% vs. 67.0%; P = 0.282). 

Though the exact mechanism is still not well understood, spending more time outdoors after school seems to be a protective factor against the new onset of childhood myopia in this population-based study. However, in terms of the estimated amount of time outdoors after school, there was no significant difference between the intervention group and the control group in both of the above-mentioned school-based trials.^25,26 Wu et al.²⁵ also reported that outdoor activity after school was not associated with myopic shift, and these findings were explained by the insufficient amount of time outdoors after school (<3 hours per week) among the schoolchildren of two participating schools.²⁵ Another possible reason could be lack of adequate variation in time outdoors after school to yield a significant result in the school-based study not having a large enough sample size. 

Consistent with previous cohort studies,^15,33 we found that the baseline refractive status was the best predictor of subsequent development of myopia during the first year of follow-up. Compared to the hyperopic children with baseline SE of +3.0 D or above, emmetropic children were at more than 19-fold higher risk for developing myopia. Similarly, the Collaborative Longitudinal Evaluation of Ethnicity and Refractive Error Study in the United States has identified cycloplegic SE refractive error as the single best predictor of the onset of myopia among ethnically diverse schoolchildren aged 6 to 11 years.¹⁵ In Australia, the Sydney Adolescent Vascular and Eye Study reported that a less hyperopic refraction at baseline was the most significant risk factor of incident myopia in both younger (aged 6–7 years) and older (aged 12–13 years) schoolchildren.³³ In the present study, a myopic shift in refractive error can be observed in each baseline SE group, and the mean SE progression rate was comparable across these groups. Accordingly, this trend in the refractive development put the emmetropic children at the greatest risk of being myopic. The Refractive Error Study in Children, investigating the natural refractive error end-points of development, suggested that children with early-onset emmetropia appear to progress through emmetropia to become myopic, rather than being trapped in the emmetropic status.³⁴ These findings have some practical implications. They emphasize that future research on myopia prevention needs to focus on young children with emmetropia who are most at risk for developing myopia. 

The pathogenic mechanism of childhood myopia is thought to be related to both hereditary and environmental factors. Parental myopia may imply a hereditary factor and has been reported to be associated with the prevalence and incidence of childhood myopia.^2,14,33,35 In addition to genetic influence, parents may impose behaviors on their children similar to those they experienced as children. In the current study, children with two myopic parents were found to have a significantly but not greatly increased (21%) hazards ratio for subsequently developing myopia compared to those without any myopic parent. However, this risk did not increase with the number of myopic parents in a linear fashion. There was little difference in risk associated with no or one myopic parent. Moreover, the longitudinal study in Australia revealed that parental myopia had no effect on incident myopia among schoolchildren of East Asian ethnicity but was the significant risk factor among European Caucasian children.³³ The school-based trial in southern Taiwan also found no association between the presence of a myopic parent and myopia shift in schoolchildren.²⁵ The possible reason for this inconsistency may be that there is regional or ethnic difference in the effect of parental myopia, and the effect of parent myopia, which may not be as great as those of environmental influences, for example, a myopiogenic lifestyle, for these Asian children. 

Attending after-school tutoring programs, such as English language courses, is popular among schoolchildren in Taipei City. Based on our results, there was a slight but statistically significant increase in the chance of new-onset myopia for schoolchildren who spent ≥5 hours on after-school tutoring programs every week. These children were more likely to spend <30 minutes outdoors after school every weekday than those who spent <5 hours on this kind of program every week (62.7% vs. 54.6%, P < 0.001), which may partially explain the association between time spent on after-school tutoring programs and the development of childhood myopia. 

This study has several strengths. The MIT study is a large-scale, population-based longitudinal study with a representative sample of the entire grade 2 schoolchildren population in Taipei City. Instead of using a sampling frame design, the MIT study launched citywide refractive status surveys and invited all grade 2 schoolchildren to participate, which may minimize potential selection bias occurring in the sampling procedure. Besides, as many as 92.1% of baseline nonmyopic schoolchildren were reexamined during the first-year follow-up. However, there are several limitations in this study. First, schoolchildren lost to follow-up were more likely to reside in an urban area than those participating in follow-up examinations (57.2% vs. 52.2%, P = 0.020). Since residing in a suburban area was a protective factor for incident myopia in this study population, the incident rate of childhood myopia may therefore be underestimated. Second, the exposure to myopiogenic factors was assessed based on the parent-reported data at baseline. Accordingly, differential recall is possible and the estimation of association between these factors and incident myopia may be less accurate. However, all the participants of this incidence study were nonmyopic in the beginning, and their exposure information was collected in the same way and with similar timing, which may eliminate the degree of recall bias between study groups. Third, to provide the citywide examination service for all grade 2 schoolchildren, as many as 74 ophthalmic units were recruited in the MIT study. Although data quality and integrity have been emphasized and monitored throughout the entire study, consistency regarding the implementation of the standard examination procedure across all 74 MIT-approved examination sites is still a concern. Finally, the epidemiology of myopia may vary among different geographic regions, age groups, and ethnicities. The generalizability of our results to other populations needs to be further confirmed. 

In conclusion, the annual incidence rate for myopia derived from population-based data covering schoolchildren from a metropolitan area of northern Taiwan was found to be very high at 31.7%. The schoolchildren with emmetropia (−0.5 D < SE < 0.5 D) at baseline had the greatest risk of developing myopia during the following 1 year. In addition to baseline refractive status, area of residence, parental myopia, and after-school activities were also found to be associated with incident myopia. Spending more time outdoors after school and less time on after-school tutoring programs were protective factors against new-onset myopia. 

Acknowledgments

The authors thank Chi-Hung Lin, MD, PhD, Lin-Chung Woung, MD, PhD, Shiow-Wen Liou, MD, PhD, Pei-Yu Lin, MD, and Ching-Yao Tsai, MD, PhD, for their full administrative support for the MIT. 

Supported by Grants H10237 and P10303 from the Taipei City Government. 

Disclosure: D.-C. Tsai, None; S.-Y. Fang, None; N. Huang, None; C.-C. Hsu, None; S.-Y. Chen, None; A.W.-H. Chiu, None; C.J.-L. Liu, None