This study protocol was reviewed and approved by the Institutional Review Board of Korea University Anam Hospital and adhered to the tenets of the Declaration of Helsinki. Among the first-grade elementary school children (6-7 years old), those whose parents agreed for them to participate in this study were enrolled. Informed consent was obtained from all children and their parents, and ophthalmologic examinations were performed.

The classroom and outdoor illuminance values were obtained in 50 elementary schools in South Korea (Seoul, Gyunggido, Daejeon, and Chungchungdo) which permitted measurements of illuminance. The outdoor and classroom illuminance were measured at 10:30 to 11:00 a.m. and 2:30 to 3:00 p.m. considering that the sun reaches the meridian transit altitude at 12:35 p.m. The incidence angle of sunlight was the same between these two time periods. Classroom illuminance was measured at nine points on the desk plane (Fig. 1) of all classrooms, and the mean value was obtained. All indoor and outdoor illuminance values were measured by Korea Institute of Lighting Technology using an illuminometer (UA-002-64; Onset Hobo, Bourne, MA, USA). The daylight factor, which was defined as the ratio of the illuminance at a point on the working plane and outdoor illuminance, was calculated (mean value of indoor illuminance [lux] at nine points on the desk plane / outdoor illuminance [lux] × 100%). Indoor light level in the classroom is strongly affected by sunlight through windows because sunlight is a far more intense light source than artificial lighting. So, classroom illuminance varies according to the outdoor light level (time and weather). However, the daylight factor is relatively constant and reflects indoor light level regardless of outdoor light intensity because it is the ratio of indoor illuminance through the window to outdoor illuminance. When outdoor illuminance increases, indoor illuminance through the window also increases, so the ratio remains relatively constant. Among the 50 elementary schools, the school with the lowest daylight (LD) factor (LD school; daylight factor, 0.51%) and that with the highest daylight (HD) factor (HD school; daylight factor, 13.35%) were selected. Both schools were located in metropolitan cities with populations more than a million. The north latitude of LD school is 37° and that of HD school is 36°.

Presenting visual acuity was measured with a Snellen acuity chart, and the anterior segment was examined with a slit-lamp (Kowa Optimed, Torrance, CA, USA). Subjects with anterior segment abnormalities, such as corneal opacity or cataracts, were excluded. Refractive error was measured with an autokeratorefractometer (KR-800; Topcon, Tokyo, Japan), and the spherical equivalent value (SE) was obtained. Axial length (AL) was measured with IOL Master (Carl Zeiss Meditec, Jena, Germany). A survey was conducted to evaluate children’s history for ophthalmologic care, parental myopia, and the amount of near-work and outdoor activity (Table 1).

Trivial diseases that did not affect refractive error or visual acuity, such as temporary conjunctivitis, hordeolum, and chalazion, were not included. The ophthalmologic examination was performed 1 to 2 months after school admission (April to May) and was repeated 6 months after the first examination (October to November). There was summer vacation between the two measurements. The values obtained from the right eye were used for analysis. Myopia was defined as SE ≤−0.50 diopters (D). The SE, prevalence of myopia, and AL were compared between the two schools at each time point and the amount of change was compared. The ophthalmologic history, parental myopia, and the amount or near-work and outdoor activities were compared between the two schools using the survey results. Initial AL of emmetropic children whose initial SE was −0.25 to +1.00 D (n = 155) was 22.7 ± 0.63 mm. We compared changes in ophthalmologic parameters in children with AL values <22.7 mm and ≥22.7 mm separately to determine whether illuminance can affect children’s refractive error differently according to AL [16,23].

The statistical analysis was performed using IBM SPSS ver. 21.0 (IBM Corp., Armonk, NY, USA). Student t-test and Mann-Whitney U-test were performed to compared SE, AL, the amount of change, and the amount of near-work and outdoor activity between the two schools as appropriate. Changes after 6 months were compared using the paired t-test and Wilcoxon signed-rank test. Ophthalmologic history and parental myopia were compared using the chi-square test. A p-value <0.05 was considered significant.

At the initial examination, 162 children from the LD school and 153 children from the HD school participated. Among them, the ophthalmologic examination was successfully performed in 145 from the LD school and 147 from the HD school after 6 months. The data of these 292 children were used for analysis. There were 75 boys (51.7%) in the LD school, and 78 (53.1%) in the HD school, and no difference in the sex distribution was detected between the two schools (p = 0.981).

Nine children (two with amblyopia, four with strabismus, two with nasolacrimal duct obstruction, and one with a dermoid) at the LD school had an ophthalmologic history. Only one had undergone surgery (dermoid excision). Two children at the HD school had amblyopia, one had strabismus, and three had epiblepharon (two were surgically corrected) (p = 0.411). All children with either strabismus or amblyopia showed presenting visual acuity ≥0.7. Parents of 284 children answered the survey about parental myopia. The numbers of children with both parents without myopia, one parent with myopia, and both parents with myopia were 43, 44, and 54, respectively in the LD school, and 52, 40, and 52 in the HD school (p = 0.539) (Table 2). The amount of near-work, including reading books and using a computer or smart phone was not different between the two schools (Table 2). More time was spent outdoors per week by children at the HD school (p = 0.036), but the difference was only 0.1 times per week, which was too small to induce any change.

Table 3 shows the changes in SE, AL, and prevalence of myopia. At the initial examination, SE was more myopic at the HD school (p = 0.002), but no difference in AL values were observed between the two schools. The amount of change in SE and AL after 6 months was not different between the schools. The initial prevalence of myopia was higher in the HD school than that in the LD school. However, the prevalence increased in the LD school, and it became similar after 6 months (p = 0.227), suggesting that development of new myopia was higher in the LD school among previously nonmyopic children with a shorter AL (Table 3).

Thus, using mean AL of emmetropic children (155 children, SE >−0.25 D and <+1.00 D), which was 22.7 ± 0.63 mm, we analyzed children with AL <22.7 mm and ≥22.7 mm separately. No difference in the change in AL was observed between the two schools in children with AL ≥22.7 mm (Table 4). However, the amount of change in AL after 6 months was significantly larger in the LD school (0.19 mm) than that in the HD school (0.15 mm) in children with AL <22.7 mm (p = 0.049) (Table 4). The number of children whose AL increased more than the mean (0.17 mm) was significantly larger in the LD school (31 children, 46.3%) than in the HD school (15 children, 26.8%) among children with initial AL <22.7 mm (p = 0.026). No difference in children with initial AL ≥22.7 mm was detected between the two schools (p = 0.273).

This study showed changes in SE and AL in two schools with different indoor illuminance and other known factors related with myopia. Children with myopic parents have a greater chance of developing myopia [13]. Many studies showed a relationship between myopia and near-work. Saw et al. [12] reported children who read more than two books per week have a higher degree of myopia than those who do not. Lee et al. [11] noted that increased time spent reading induces more myopia. Outdoor activity is thought to play a protective role against myopia [10,11,15,17,24]. Rose et al. [15] reported that the amount of time spent outdoors is more important than the activity itself. Wu et al. [16] documented that outdoor activity during class recess decreased the development of myopia in a prospective study. We investigated the presence of parental myopia, amount of near-work and outdoor activity in addition to indoor illuminance to adjust the effect on myopia. According to our study, children in classrooms with low illuminance showed a greater increase in AL when their initial AL was <22.7 mm (mean for emmetropic children). Classroom illuminance did not affect myopia in children with a longer initial AL. Factors other than classroom illuminance, such as the presence of parental myopia, the amount of near-work and outdoor activity, were not different between the two schools.

Wu et al. [16] noted that the effect of outdoor activity differs according to the initial refractive status of children. Outdoor activity during class recess decreases the shift in refractive error toward myopia in nonmyopic children, but does not retard progression of myopia in children who already are myopic. In our study, we could not perform cycloplegic refraction because many parents did not give consent. Although manifest refraction has been used in studies on myopia, it can be inaccurate in some children with various degrees of accommodation. We decided to use AL as a reference and divided the children into longer and shorter AL groups. In our study population, mean AL of emmetropic children was 22.7 mm. Mutti et al. [25] reported that AL of emmetropic children (SE, −0.25 to 1.00 D) in their study population (age, 6 to 14 years) was also 22.7 mm. We used AL of 22.7 mm as a reference point and divided the children according to this value. Our results correspond well with Wu et al. [16] in which light reduced axial elongation of the eye in nonmyopic school children.

Hua et al. [23] also investigated the effect of classroom illuminance on myopia. They selected two schools in which the classrooms did not fulfill the required classroom illuminance of 300 lux. Artificial lighting was used to increase the illuminance to 558 lux, and the effect of lighting was compared between brighter and darker classrooms. They reported that nonmyopic children from the brighter classroom showed less refractive shift toward myopia. Axial elongation was 0.13 mm/yr in the high illuminance classroom and 0.18 mm/yr in low illuminance classroom among nonmyopic children, which were 0.15 mm/0.5 yr and 0.19 mm/0.5 yr in our study. However, the study setting was different. Hua et al. [23] increased classroom illuminance from 74 to 558 lux using a lighting system, and illuminance was measured at 8:00 to 9:30 p.m. However, school children are at school during the daytime (about 9:00 a.m. to 2:00 p.m. for the first-grade elementary students in South Korea), and the sun is the most powerful light source during the day. Classroom illuminance is more strongly affected by sunlight than a lighting system during the day. Consequently, classroom illuminance is affected by weather and time and changes constantly. We used a daylight factor rather than daytime classroom illuminance, which tends to be constant regardless of outdoor brightness. The daylight factor reflects daytime classroom illuminance in a more practical way. Additionally, the schools enrolled in our study were built to the Korean Standard (KS A 3011) in which the lowest illumination at the desk plane is 300 lux. Thus, our study reflected practical daytime classroom illuminance.

The mechanism of the decrease in myopic shift due to increased indoor light level is not clear. In our study, the classrooms had windows, and a large portion of classroom illuminance originated from sunlight. Thus, increased classroom illuminance would share the mechanism of decreased axial elongation by outdoor activity. In addition, hyperopic retinal blur due to a high lag of accommodation during near-viewing activities has been proposed as a cause of juvenile-onset myopia progression in humans [17,19] A lower classroom illuminance level may disturb pupil constriction and cause a lag of accommodation, subsequently increasing peripheral retinal defocus. Additional studies on the mechanism of the protective effect of indoor illumination on refractive error should be performed.

Some limitations in this study should be mentioned. Cycloplegic refraction was not performed, and the follow-up period was short. Only two schools with highest and LD factor were included. The presence of parental myopia, the amount of outdoor activity and near-work were evaluated by survey, which may have produced subjective results. Other factors related with myopia, such as parent education level and family income, were not investigated in this study [3]. Further study assessing children’s genetic and environmental factors objectively with a longer follow-up period will be necessary.

In conclusion, high classroom illuminance including sunlight from windows during the daytime reduced axial elongation of the eye in the children with shorter AL. The increase in classroom light level by permitting more sunlight to enter might be protective against the development of myopia in addition to increasing outdoor activity.