Introduction

Uncorrected refractive error, especially myopia, is the most common cause of vision impairment worldwide and the second most common cause of reversible blindness1. Therefore, myopia has attracted increasing attention. However, myopia is an increasingly widespread condition around the world, particularly in east Asia2, and east Asia shows by far the greatest productivity loss resulting from myopia in total dollar terms in the world3. In China, the prevalence of myopia among teenagers has shown a concerning trend, escalating from 48.4% prior to 2008 to 84.8% post-2013 in individuals aged 16 to 184. Expected increases in myopia prevalence are likely to cause greatly increased public health and economic problems in the future3. Hence, strategies are warranted to reduce the prevalence and the progression of myopia. A comprehensive understanding of the factors contributing to myopia development is imperative for devising effective prevention strategies. Numerous factors have been proposed to play a role in myopia onset, ranging from non-modifiable genetic predispositions5 and physical characteristics6, to modifiable environmental influences7, educational attainment8, outdoor exposure9, and digital device usage10, among others. A review by the International Myopia Institute highlighted the significant associations of time spent outdoors, educational attainment, and parental myopia with myopia development11. In recent years, attention has increasingly turned towards housing and living environments as potential contributors to the myopia epidemic. Reduced exposure to green spaces has been identified as a potential risk factor for myopia12,13. Disparities in myopia prevalence have been observed between urban and suburban regions and across different housing types, as evidenced by a study conducted in Sydney14. In another study in Hong Kong15, small home size was supposed to be an environmental threat for myopia development.

Over the past three decades, China has undergone rapid urbanization on an unprecedented scale. It is projected that by 2030, China’s urbanization rate will reach 70%, reflecting the migration of nearly 300 million individuals from rural to urban areas16. Accompanying this urbanization, there has been a notable shift in housing types, transitioning predominantly from low-rise multistory flats to high-rise apartment complexes. However, the potential impact of these changes in residential environments on myopia development remains largely unexplored.

In this study, we aimed to investigate the influence of residential transformations on myopia development by collecting questionnaires from a homogeneous sample in a major city in mid-west China. Additionally, we utilized a commercial digital satellite mapping tool to quantify the extent of building shadows within the relevant communities, serving as a proxy measure for residential changes. The merits of this metric are (1) both density and height of the constructions within a living community are taken into account, (2) it is readily available since the measurements are based on an open satellite map. Through this investigation, we endeavor to contribute to the understanding of how urbanization and housing alterations may shape the prevalence of myopia in urban environments.

Materials and methods

Study design and population

We conducted a general survey of 10th-grade students enrolled in the International Course Center at Xi’an Gao Xin No. 1 High School, situated in Xi’an, a major city characterized by a blend of tradition and modernization in mid-west China. In May 2023, following a popular science lecture on myopia delivered by an ophthalmologist and optometrist (Z.L.), we administered a questionnaire, as detailed in Table 1, to each student anonymously. Non-myopic eyes were defined as those with an equivalent spherical lens (ESL) greater than − 0.5D, while other categories included mild myopia (ESL between − 3.0D to -0.5D), moderate myopia (ESL between − 6.0D to -3.0D), and high myopia (ESL ≤ -6.0D).

Table 1 Myopia questionnaire in the international course center from Xi’an Gao Xin No. 1 high school.
Full size table

Of primary interest was the residential address where each survey participant predominantly resided within the past six years. Based on the review of risk factors for myopia from the International Myopia Institute11, the confounding factors we assessed in the questionnaire included parental myopia status, time spent outdoors, time dedicated to near-work activities, and the utilization of orthokeratology lenses which have been widely used in China to control the progress of adolescent myopia. Notably, the inquiry into the use of low-dose atropine was omitted from the questionnaire as it has yet to receive official approval from the Chinese Food and Drug Administration (CFDA). To ensure the accuracy of responses, the questionnaires were not distributed and collected on the same day, allowing students to engage in thorough discussions with their parents. Questionnaires with invalid living address or unspecified range of myopia should be excluded from the study. The questionnaires were distributed to all the 6 classes (150 students) in grade 10, except 30 ones who were absent at the time. Therefore, 120 students (73 females and 47 males, all of Asian descent) finally participated in the questionnaire with the mean age of 15.9±0.614,15,16,17,18 years old. All students were residents of Xi’an city, except one who hails from Bao Ji city, which lies at approximately the same latitude as Xi’an.

The studies involving human participants followed the tenets of Declaration of Helsinki and were reviewed and approved by the Ethics Committee and Institutional Review Board of The First Affiliated Hospital of Xi’an Jiaotong University. We confirmed that all experiments were performed in accordance with relevant guidelines and regulations. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

Image processing and measurements

A commercially available digital satellite map (Amap Software Co., Ltd, (2023) No.4047, Beijing, China. Image source: DigitalGlobe&spaceview. URL link: https://www.amap.com/) was utilized to geolocate the residential addresses, and a 500 m×500 m area of the image centered on each automatic label was extracted from the map (as depicted in Fig. 1). Subsequently, the full-color image was converted into a histogram image (Fig. 2) using the “Color Inspector 3D” plugin within the ImageJ software (version 1.53t, developed by Wayne Rasband and contributors at the National Institutes of Health, USA). We developed a novel metric, the proportion of shadow area (PSA), to approximate the residential environment of a community by quantifying the shadows cast by buildings in remote sensing digital images, given that the sun-synchronous orbit satellite passes a ___location on earth at the same time of a day, thus the angle of sunlight should be approximately the same within a certain district and a period of season. PSA was determined by summing the frequencies of RGB values within the shadowed regions (refer to Fig. 2). A detailed description of the measurements was included in the supplementary material 1. Similarly, the proportion of greenness area (PGA) was calculated using the same methodology.

Fig. 1
figure 1

A cropped satellite image showing an area with 500 m×500 m in size centered on a yellow star which is automatically labeled as a specific living community on the map. (Satellite map from Amap Software Co., Ltd, (2023) No.4047, Beijing, China. Image source: DigitalGlobe&spaceview. URL link: https://www.amap.com/).

Full size image
Fig. 2
figure 2

The histogram of the color image shown in Fig. 1. The dark gray color with RGB Value of (68, 68, 68) approximates the shadow area, which composes 21.2% (the frequency of this color is displayed on the top right) of the whole image. (Satellite map from Amap Software Co., Ltd, (2023) No.4047, Beijing, China. Image source: DigitalGlobe&spaceview. URL link: https://www.amap.com/).

Full size image

In addition to residential areas, consideration was given to schools attended during grades 4 through 9. PSA within school was not assessed due to regulations stipulated by the Chinese government, which mandate that school buildings for the nine-year compulsory education program should not exceed five floors in height. The PGA within the official boundaries of each school, as delineated on the satellite map, was determined using the aforementioned methodology.

Statistics

The univariate correlation between the degree of myopia and each candidate factor was assessed using Spearman’s rho in SPSS 20.0 (SPSS Inc., Chicago, IL). A graphical representation illustrating the relationship between PSA and the odds ratio of myopia was generated using the open-source R program (version 4.3.1) and RStudio (version 2023.09.1 + 494). Identification of a cut-off point for PSA from the graph allowed for the stratification of the sample into two groups. Subsequently, a comparison of the degree of myopia, along with confounding factors, between these two groups was conducted using either the chi-square test or independent samples T-test (Mann-Whitney test should be used if a normal distribution is not satisfied) in SPSS 20.0 (SPSS Inc., Chicago, IL). A significance level of P ≤ 0.05 was employed to determine statistical significance.

Results

Among the 120 students surveyed, the prevalence of myopia in at least one eye was 80.0%, with 8.3% exhibiting high myopia. Mild and moderate myopia accounted for 39.2% and 32.5% of the sample, respectively. Over 90% of students reported spending more than 6 h per day engaged in near-work activities, while less than 20% reported rare use of electronic devices. Additionally, more than half of the students reported spending less than 3 h engaged in outdoor activities per week. Orthokeratology lens usage for over one year was documented in 29.2% of the students.

Out of the 120 participants, 109 were successfully geolocated to their living addresses (from 93 distinct communities) on the satellite map. The mean PSA and PGA within a 500 square meter radius of the addresses were (20.24 ± 8.59)(2.7–43.6)% and (20.53 ± 8.78)(7.1–39.6)%, respectively. Furthermore, the 120 participants originated from 38 elementary schools (grades 4 to 6) and 29 middle schools (grades 7 to 9), with mean PGA values of (10.78 ± 7.16)(2.0-34.5)% and (12.26 ± 5.59)(3.8–35.0)%, respectively.

Correlations between the level of myopia and each factor are summarized in Table 2. Notably, residential PSA exhibited a negative correlation with the level of myopia, albeit with marginal significance (r=-0.189, P = 0.05). The restricted cubic spline analysis (Fig. 3) revealed a cut-off point of 20% for PSA in terms of the odds ratio for moderate and high myopia. Beyond this threshold, PSA appeared to confer a protective effect against myopia, and vice versa. Subsequently, the sample was stratified into two groups based on the PSA cut-off point (20%). A chi-square test demonstrated a significant difference in the level of myopia between the two groups (P = 0.012), while all other confounding factors exhibited no statistically significant differences (refer to Table 3). The scatter plot of the relationship between myopia level and residential PSA was shown in Fig. 4.

Table 2 Correlations between the level of myopia and each factor.
Full size table
Fig. 3
figure 3

Dose-response relationship between the proportion of shadow area (PSA) and myopia. The reference is non-myopia and mild myopia.

Full size image
Table 3 Comparison between two groups exploiting the cut point (20%) of PSA.
Full size table
Fig. 4
figure 4

The scatter plot of the relationship between myopia level and residential PSA. (1-no myopia, 2-mild myopia, 3-moderate myopia, 4-high myopia).

Full size image

Discussions

In this cross-sectional study, we aimed to investigate the relationship between varied urban living environments and the prevalence of myopia among adolescents. And PSA was developed to approximate the residential environment of a community. Surprisingly, our findings revealed a negative correlation between shadow proportion and myopia development, with shadow proportions below 20% posing a risk, while proportions beyond 20% potentially exhibiting a protective effect (illustrated in Fig. 3).

Through a comprehensive review of the satellite images, we observed that communities primarily comprised of high-rise apartments tended to have higher shadow area ratios. These areas typically featured larger inter-architectural distances compared to those dominated by multistory flats or houses in our study city. Examples showcasing high and low proportions of shadows are depicted in Fig. 5. We postulate that taller floors and greater inter-architectural spacing may contribute to reduced myopia progression. However, it’s noteworthy that our analysis only considered community-wide environments. Future research should explore the effects of specific living floor levels on myopia development. In addition, there are studies linking violet and blue light to less progression of myopia17,18. It is also intriguing to explore the impact of color distribution within a living environment on the progress of myopia in future studies.

Fig. 5
figure 5

Examples showing two communities with a low (left) and a high (right) shadow area, with the shadow proportion of 5.4% and 40.2%, respectively. In the left image, the community mainly composed of townhouses and multistory flats, while in the right image, the community mainly composed of high-rise apartments commonly with a larger inter-structural space. (Satellite map from Amap Software Co., Ltd, (2023) No.4047, Beijing, China. Image source: DigitalGlobe&spaceview. URL link: https://www.amap.com/).

Full size image

Prior research has suggested conflicting findings regarding the influence of living conditions on myopia. A large-scale survey in China suggested that lower living floors may be protective against myopia19, and a recent study in South India indicated a higher prevalence of myopia among children living in apartments compared to other types of housing20. However, these studies did not account for rural-urban or socioeconomic disparities, which may influence housing type differences and the development of myopia. Future investigations should adjust for these disparities or adopt a more uniform survey population to elucidate the physical impact of city constructions on myopia.

Our study sample is notably homogenous, comprising solely 10th-grade students with an average age of approximately 16 years. These students predominantly hail from middle-class or affluent families, as they are enrolled in the international course class of a private high school, aiming to pursue higher education abroad. Additionally, the vast majority of participants originate from Xi’an city, providing a microcosmic representation of China’s urbanization trends over the past two decades. Consequently, participants share similar educational and socioeconomic backgrounds, which are known to influence myopic development21,22. Therefore, the study sample should be a typical one in assessing urban conditions and meanwhile minimize possible confounding effects from factors such as education and social stratum.

The prevalence of myopia and high myopia of the sample was 80% and 8.3% each, which is comparable to the large-sample studies in China4,23. Some factor such as near work, parental myopia and time outdoors were not significantly correlated with myopia in this study, probably due to the limited sample and small variations of the factors. We also calculated the green space within both the residentials and the schools using the green color coding from the histograms of satellite maps, which was not associated with myopia either. One disadvantage of this method is that the green space within the shadows could not be calculated. In previous studies suggesting a protective effect of green space12,13, the normalized difference vegetation index was used as an indicator, although it could also be affected by the shadows24. Still, there were several limitations of this study. Firstly, our study population is limited in scope and the limited scope may increase the likelihood of selection bias. However, the homogeneity of the sample may mitigate bias. Moreover, despite the small sample size, a significant difference in myopia levels was observed between higher and lower shadow area groups, with adjustments made for other confounding factors. Secondly, our study relies primarily on questionnaire data rather than performing rigorous measurements, which may introduce inaccuracies. One study have pointed out the tendency of questionnaire to overestimate the outdoor time while underestimate near-work time25. To enhance reliability, a lecture was conducted by an optometrist prior to questionnaire distribution. Thirdly, the measurement of building shadows may not be universally applicable, cause it varies with different latitude and season. Fourthly, the green space could be underestimated due to the overlap of the shadows and the shadow area could also be miscalculated resulting from its projections on the neighbor structures which composed of only a very small fraction. And we consider them as the systematic errors in this study.

In summary, our findings demonstrate a negative correlation between myopia levels and shadow areas within living communities among a sample of 10th-grade students in a megacity. These results alleviate concerns that high-rise constructions accompanying urban modernization may adversely affect adolescent eye health. On the contrary, high-rise apartments may offer a solution for megacities by maximizing ground space while potentially exerting a protective effect against myopia progression. Nevertheless, a prospective and large sample study is warranted to validate the effect of PSA on myopia. Further, a coefficient might be introduced to overcome the geographical variations of PSA.