A generalized additive model analysis of school classroom environment and students' refractive state progression Authors: Meng Qin, Wei-xing Shi, Shan-xi Gao, Yue Niu, Jing-wei Zheng Published: 2024 ABSTRACT AbstractBackgroundThe study, undertaken in Fengxian District, Shanghai, seeks to investigate the correlation between the sanitary conditions of high school classrooms and the changes in students' uncorrected visual acuity and refractive status, given their predominantly classroom-centric lifestyle and the considerable academic pressures they encounter.MethodsUtilizing a cluster random sampling approach, we randomly selected 48 classrooms for monitoring from 8 high schools within the jurisdiction, taking into account factors such as classroom orientation and floor level. A cohort comprising 1087 students was enrolled and followed for a year, with their refractive data being collected. A generalized additive model was employed to examine the associations between classroom environmental factors and students' visual acuity as well as refractive status.ResultsSignificant increases were observed in the spherical equivalent, cylinder, and corneal astigmatism for both the right and left eyes from 2022 to 2023 (P < 0.001). However, no significant change was noted in uncorrected visual acuity (P > 0.05). The compliance rate for blackboard size (height) was 100%, with an average of (1.16 ± 0.06) m and a minimum of 1.06 m. None of the classrooms met the standards for the distance between the multimedia display and the front seat, with an average of (177.26 ± 17.78) m. Classroom per capita area, blackboard size (height), and the distance between the multimedia display and the front seat influenced changes in uncorrected visual acuity. Additionally, classroom per capita area affected changes in the spherical equivalent. The window-to-ground area ratio and blackboard reflectance ratio impacted changes in astigmatism values at 90°-180° (J180) and 45°-135° (J45). Blackboard size (width) and the window-to-ground area ratio influenced changes in students' corneal astigmatism. The classroom with the least impact on refractive status had a per capita area of 1.6 m2, a blackboard size (width) of approximately 4.2 m, and a window-to-floor area ratio of around 0.27.ConclusionsSchools should adhere to national standards to enhance the classroom visual environment and further improve the compliance rate.Keywords: Refractive state, Abnormal vision, Classroom environmental factors Introduction Good vision is crucial for adolescents' physical development, academic success, and overall well-being. The primary factors influencing adolescents' visual health encompass genetics, environmental conditions, lifestyle habits, and more [1–10]. In 2019, the nationwide prevalence of myopia among children and adolescents stood at 50.2%. Subsequently, the pandemic's impact saw this figure rise to 52.7% in 2020. Although there was a slight decrease to 52.6% in 2021 compared to the previous year, by 2022, the overall myopia rate among Chinese youths continued to show a minor downward trend, reaching 51.9%, yet it remained at a notably high level. Consequently, ongoing vigilance and the implementation of effective measures are imperative to reduce and control vision abnormalities in adolescents. The school environment plays a pivotal role in students' learning and is intimately linked to their health. The Guidelines for Appropriate Techniques for the Prevention and Control of Myopia in Children and Adolescents stipulate that schools must provide a teaching environment that meets ocular hygiene standards. In recent years, researchers have increasingly focused on the issue of vision abnormalities in children and adolescents, resulting in a surge of related research literature [4–6, 11–13]. These studies suggest that improving lighting conditions can effectively reduce the incidence and progression of vision abnormalities. However, most prior research in this field has concentrated on outdoor environmental factors and employed cross-sectional designs [14–17], which do not facilitate longitudinal monitoring of vision changes within the same population over time. Therefore, there is an urgent need for a longitudinal study to gain a more precise and in-depth understanding of the risk factors for vision abnormalities, in order to provide a scientific foundation for their prevention and control among adolescents. Fengxian District, situated in the outer suburbs of Shanghai, is home to high school students who reside on campus. Given the substantial academic workload faced by these students, vision abnormalities represent a significant health concern. As they spend the majority of their time in classrooms, optimizing the classroom visual environment is paramount to safeguarding their visual health. Thus, this study focused on high school students, longitudinally tracking their visual acuity and monitoring classroom environmental factors, in order to explore the relationship between these factors and the progression of students' uncorrected vision and refractive status. Methods Subjects and study design The study was carried out in Fengxian District, Shanghai, China, between September 2022 and September 2023. All eight high schools in the district were encompassed in this research. From each school, six representative classrooms were randomly chosen based on factors such as classroom orientation and floor level, resulting in a total of 48 classrooms included. High school students from these selected classrooms were recruited for the study and followed for a duration of one year (Fig. 1). Each participant underwent vision assessments at two time points: at baseline and one year later. Additionally, environmental factors within their classrooms were measured at the baseline. Prior to the commencement of the study, all enrolled students and their guardians were fully informed and provided signed informed consent. This study followed the tenets of the Declaration of Helsinki and was approved by Shanghai Municipal Center for Disease Control and Prevention Ethical Review Committee (approval number: KY-2024–10). Study subject inclusion process Vision measurements Vision assessments on the students were conducted between September and November of both 2022 and 2023. These assessments encompassed evaluations of uncorrected visual acuity and refractive status. Specifically, uncorrected visual acuity was assessed using a standardized visual acuity examination form that adhered to national standards (GB 11533–1989). Students with uncorrected visual acuity of less than 5.0 in either eye were classified as having abnormal vision. Throughout the measurements, we adhered to lighting standards outlined in GB7793-2010 and the 2024 Myopia Prevention Guidelines to ensure an optimal testing environment. This study employed four indicators to evaluate the refractive status of the students: spherical equivalent, corneal astigmatism, and astigmatism values at 90°−180° (J180) and 45°−135° (J45). These parameters were determined through refraction measurements using standardized automated computer refractometers (TOPCON KR-800) under non-cycloplegic conditions. The refractometers displayed refractive values in the format S/C × θ, and vector decomposition was performed based on Fourier analysis using the following formulas: M = S + C/2, J180 = -Ccos(2θ)/2, and J45 = -Csin(2θ)/2. Here, M, J180, and J45 represent the spherical equivalent, astigmatism at 90°−180°, and astigmatism at 45°−135°, respectively. S and C denote spherical and cylindrical powers. Corneal astigmatism was calculated by determining the absolute refractive power difference between the steepest and flattest corneal meridians. During the vision examinations, each eye was measured three times, and the average value, rounded to two decimal places, was recorded. If there was a difference of 0.50 diopters (D) or more in the spherical power measurements between any two readings, additional measurements were conducted, and a new average value was calculated. In the event that the display screen showed "E" or "ERR," indicating a measurement reliability of less than 70% (often due to factors such as irregular astigmatism, cataracts, or blinking during the examination), the cause (e.g., blinking, nystagmus, lens opacity, or excessively high refractive power) was promptly identified and documented. If a student was unable to complete the examination for any reason, the reason was also noted. Furthermore, for students with particularly strong accommodative abilities and significant variations in multiple measurement values, a note was made on the record sheet. Classroom environment measurements Classroom environmental factors were measured for 14 parameters from September to November in 2022, including classroom per capita area, desk and chair allocation compliance rate, blackboard size (width), blackboard size (height), blackboard hanging height, blackboard reflectance ratio, blackboard surface average illumination, blackboard surface illumination uniformity, classroom desk average illumination, classroom desk illumination uniformity, lighting coefficient, window-to-ground area ratio, rear wall reflection ratio, multimedia display to front seat distance. The conformity of these classroom environment monitoring indicators was assessed based on the relevant national standards for classroom environments. Statistical analysis Data analysis and processing were conducted using R4.3.2 software. Quantitative indicators were described using means, 95% confidence intervals (95% CI), and quantiles. Paired t-tests were employed to compare changes in refractive status indicators. When constructing the Generalized Additive Model (GAM), we applied spline smoothing functions and assumed that the refractive status data followed a Gaussian distribution. Baseline refractive data were included as a baseline in the model, with adjustments made for variables such as gender, age, ethnicity, and grade. Subsequently, the model was used to individually assess classroom environmental factors, screening for those with statistically significant effects on refractive status. After performing collinearity diagnostics, collinear or statistically insignificant factors were excluded, leaving only the relevant factors for multivariate analysis. When multiple influencing factors were identified, their interactions were further analyzed and visually presented. All data analyzed in the model were based on right-eye measurements, with statistical significance defined as P < 0.05. Results Descriptive analyses In 2022, a total of 1,650 students from 48 classes were recruited. However, by 2023, 563 former senior high school students were lost to follow-up due to college entrance or personal reasons. Ultimately, 1,087 students from 33 classes were included in our analyses. These students comprised 676 first-year and 411 s-year high school students, with 584 boys and 503 girls. In 2022, the proportion of abnormal vision was 88.18% for boys and 90.26% for girls. Overall, statistically significant differences were observed in spherical equivalent, astigmatism values at 90°−180° (J180), and corneal astigmatism between 2022 and 2023, but not in astigmatism values at 45°−135° (J45) or uncorrected visual acuity (Table 1). Specifically, for right eyes, the spherical equivalent shifted from −3.49 D in 2022 to −3.56 D in 2023, astigmatism values at 90°−180° (J180) changed from 0.414 D to 0.016 D, and corneal astigmatism varied from 1.39 D to 1.47 D. Table 1Changes in uncorrected visual acuity, spherical equivalence, cylinder, and corneal astigmatism among 1087 high school students from 2022 to 2023In 2022In 2023tPM95% CIM95% CIThe uncorrected visual acuity of the right eye4.374.35 ~ 4.404.364.33 ~ 4.391.590.111Spherical equivalent in the right eye−3.49−3.63 ~ −3.34−3.56−3.71 ~ −3.402.710.007Cylinder of right eye−0.98−1.04 ~ −0.92−1.02−1.08 ~ −0.963.62 < 0.001*Astigmatism values at 90°−180° (J180)0.4140.384 ~ 0.4430.0160.005 ~ 0.02724.16 < 0.001*Astigmatism values at 45°−135° (J45)0.002−0.011 ~ 0.0140.004−0.000 ~ 0.009−0.400.688Corneal astigmatism in the right eye1.391.33 ~ 1.441.471.42 ~ 1.52−4.48 < 0.001**p < 0.05 Changes in uncorrected visual acuity, spherical equivalence, cylinder, and corneal astigmatism among 1087 high school students from 2022 to 2023 *p < 0.05 A total of 33 classrooms were monitored, with an average of 33 students per classroom. The blackboard size (height) was 1.16 ± 0.06 m, ranging from 1.06 m to 1.17 m. The blackboard size (width) was 4.10 ± 0.14 m, with a minimum value of 3.81 m and a maximum value of 4.2 m. Multimedia display to front seat distance was 177.26 ± 17.78 cm, ranging from 145cm to 190 cm. The classroom per capita area was 1.67 ± 0.34 m2, with a range of 1.16 m2 to 1.82 m2.The average window-to-ground area ratio was 0.24 ± 0.06, ranging from 0.14 to 0.29 (Table 2). Table 2Monitoring results of environmental factors in students' classrooms (weighted n = 1087)Classroom environmental factorsNo. of passesRate/%X ± sMinp25Mp75MaxClassroom per capita area/ m286379.391.67 ± 0.341.161.421.701.821.82Desk and chair allocation coincidence rate/ %19918.310.36 ± 0.260.000.200.300.400.40Blackboard size (width)/ m90483.164.10 ± 0.143.814.054.164.204.20Blackboard size (height)/ m1087100.001.16 ± 0.061.061.111.151.171.17Blackboard hanging height/ m51647.470.96 ± 0.090.830.890.921.011.01Blackboard reflectance ratio44641.030.16 ± 0.050.090.130.150.170.17Blackboard surface average illumination/ lx78572.22676.42 ± 288.29217.81491.06623.06797.31797.31Blackboard surface illumination uniformity105697.150.90 ± 0.060.690.880.920.950.95Classroom desk average illumination/ lx94887.21622.58 ± 189.26236.50524.81663.59779.37779.37Classroom desk illumination uniformity93385.830.78 ± 0.130.400.720.830.880.88Lighting coefficient/ %46943.150.02 ± 0.010.000.010.020.030.03Window-to-ground area ratio67462.010.24 ± 0.060.140.190.220.290.29Rear wall reflection ratio71365.590.78 ± 0.060.660.730.790.820.82Multimedia display to front seat distance (cm)00.00177.26 ± 17.78145.00162.00185.00190.00190.00 Monitoring results of environmental factors in students' classrooms (weighted n = 1087) Results of the single-environmental factor models Generally, the majority of classroom environmental factors monitored in this study exhibited significant impacts on uncorrected visual acuity, including classroom per capita area, desk and chair allocation compliance rate, blackboard size (width), blackboard size (height), blackboard hanging height, blackboard reflectance ratio, blackboard surface average illumination, classroom desk illumination uniformity, rear wall reflection ratio, multimedia display to front seat distance. The variation in astigmatism values within the range of 90°−180° was influenced by blackboard size (height), blackboard surface illumination uniformity, lighting coefficient, and window-to-ground area ratio. Meanwhile, the variation in astigmatism values within the range of 45°−135° was significantly affected by blackboard reflectance ratio. Additionally, changes in corneal astigmatism were notably influenced by the desk and chair allocation compliance rate, blackboard size (width), blackboard size (height), blackboard reflectance ratio, blackboard surface illumination uniformity, classroom desk average illumination, window-to-ground area ratio, rear wall reflection ratio, and multimedia display to front seat distance (Table 3). Table 3Analysis of the impact of classroom environmental factors on uncorrected visual acuity, spherical equivalent, variations in 90°−180° astigmatism, variations in 45°−135° astigmatism, and corneal astigmatism changes by individual factorsClassroom environmental factorsUncorrected visual acuity changesSpherical equivalent changesVariations in 90°−180° astigmatismVariations in 45°−135° astigmatismCorneal astigmatism changesdfFPdfFPdfFPdfFPdfFPClassroom per capita area1.98728.430 < 0.0011.99822.950 < 0.0011.9081.0570.3171.0000.1940.6601.0001.3210.251Desk and chair allocation compliance rate1.8957.0800.0061.0000.7160.3981.8531.1180.3881.0000.1950.6591.0003.967 < 0.001Blackboard size (width)1.98238.420 < 0.0011.7689.3450.0011.8790.9430.4061.0000.4880.4851.9958.300 < 0.001Blackboard size (height)1.61532.520 < 0.0011.7502.0630.2171.6104.3230.03861.0000.0360.8491.93110.170 < 0.001Blackboard hanging height1.9937.5410.0011.0006.0210.0141.1690.1100.7641.0000.2710.6031.9572.6580.057Blackboard reflectance ratio1.9925.3430.0051.0000.6580.4181.0000.9650.3261.0004.2080.04051.99711.850 < 0.001Blackboard surface average illumination1.98511.290 < 0.0011.8271.8890.0981.9702.3540.08921.1210.1340.7311.8650.7690.396Blackboard surface illumination uniformity1.0003.8590.0501.0000.1040.7471.4104.210.01941.8821.7730.1151.0007.9490.005Classroom desk average illumination1.4240.6800.5861.0007.0650.0081.1930.0320.9631.0000.1210.7281.0006.9530.008Classroom desk illumination uniformity1.00014.090 < 0.0011.0003.710.0541.0090.0310.8771.5840.3090.7691.0000.2830.595Lighting coefficient1.2630.0290.8971.7031.6180.3171.0004.4200.03581.0000.6720.4121.2360.1880.669Window-to-ground area ratio1.9722.4550.0841.9697.0720.0031.9957.637 < 0.0011.4820.3220.5481.99718.100 < 0.001Rear wall reflection ratio1.9774.1160.0151.6050.2690.7741.0000.7530.3861.0000.3820.5361.8385.1640.014Multimedia display to front seat distance1.90814.28 < 0.0011.0000.1440.7041.0000.6670.4141.0000.2510.6171.9446.6000.005 Analysis of the impact of classroom environmental factors on uncorrected visual acuity, spherical equivalent, variations in 90°−180° astigmatism, variations in 45°−135° astigmatism, and corneal astigmatism changes by individual factors Results of the multi-environmental factor models Excluding co-curve variables and non-statistically significant factors, the results of multivariate analysis showed that classroom per capita area (Ref df = 1.873, F = 14.436, p < 0.001), blackboard size (height) (Ref df = 1.770, F = 20.422, p < 0.001), the multimedia display to front seat distance (Ref df = 1.947, F = 6.338, p = 0.004) had an effect on the change of uncorrected visual acuity. Classroom per capita area (Ref df = 1.998, F = 22.950, p < 0.001) had an effect on the change of spherical equivalent. The blackboard reflectance ratio has an impact on the variation of astigmatism values within the range of 45°−135°,while the reflectance of rear wall reflection ratio affects the variation of astigmatism values within the range of 90°−180°.Window-to-ground area ratio (Ref df = 1.982, F = 4.759, p = 0.006) had an effect on the change of total astigmatism. Blackboard size (width) (Ref df = 1.913, F = 5.390, p = 0.003) and window-to-ground area ratio (Ref df = 1.997, F = 17.890, p < 0.001) had an effect on corneal astigmatism. And the reference degrees of freedom were all greater than 1, indicating that the influence of classroom environmental factors on refractive status were nonlinear. Further analysis did not reveal any interaction. The influence of environmental factors on changes in uncorrected visual acuity The classroom per capita area increase correlated with rising uncorrected visual acuity, especially exceeding 1.6 m2, showing positive improvement (Fig. 2(a)). Blackboard size (height) < 1.15 m had little effect, but > 1.15 m led to a decline in visual acuity (Fig. 2(b)). With increasing distance from the multimedia display to the front seat, the change in uncorrected visual acuity first decreased (< 180 cm) and then increased (> 180 cm) with a wider confidence interval (Fig. 2(c)). Relationship between classroom per capita area, blackboard size (height), multimedia display to front seat distance and changes of uncorrected visual acuity When the per capita area of the classroom was ≥ 1.6 m2, the absolute value of the changes in uncorrected visual acuity increased and then fell with increasing distance from the multimedia display to the front seats. When the per capita area of the classroom was < 1.6 m2, the absolute value of the changes first decreased then increased. For a fixed display-seat distance, the absolute value of the changes in uncorrected visual acuity increased with classroom per capita area (Fig. 3(d)). Three-dimensional diagram showing the relationship between classroom per capita area, blackboard size (height), distance of multimedia display to front seat and changes in uncorrected visual acuity When the per capita area of the classroom was ≥ 1.6 m2, the absolute value of the changes in uncorrected visual acuity decreased with blackboard size (height). When the per capita area of the classroom was < 1.6 m2, it increased. Fixing blackboard size (height), if the blackboard size (height) was ≥ 1.15 m, the absolute value of the changes in uncorrected visual acuity first decreased, then increased with the per capita area. If the blackboard size (height) was < 1.15 m, the absolute value of the changes in uncorrected visual acuity increased with the per capita area (Fig. 3(e)). When the distance from the multimedia display to the front seats was < 160 cm or was ≥ 200 cm, the absolute changes in uncorrected visual acuity increased with the blackboard size (height) increasing. Between 160cm and 200 cm, the absolute changes first decreased then increased. With a fixed blackboard size (height), if the blackboard size (height) was ≥ 1.15 m, the absolute changes in uncorrected visual acuity decreased first then increased with the display-to-seat distance increasing. If the height was < 1.15 m, the absolute changes increased first then decreased (Fig. 3(f)). The influence of environmental factors on changes in the refractive state When the per capita area of the classroom was ≤ 1.6 m2, the changes in spherical equivalent were not significant. When the per capita area of the classroom was > 1.6 m2, as the per capita area of the classroom increased, the rate of changes in spherical equivalent also increased, with a change greater than 0, resulting in a continuous decrease in the equivalent spherical power (Fig. 4). Relationship between classroom per capita area and changes of spherical equivalent When the window-to-ground area ratio increased, the 90–180-degree astigmatism variation first decreased and then increased; specifically, when the ratio reached 0.25, the variation of astigmatism was minimal (Fig. 5(g)). When blackboard reflectance ratio increased, the 45–135-degree astigmatism variation exhibited an inverse relationship, tending to decrease accordingly (Fig. 5(h)). Relationship between blackboard reflectance, window-to-floor area ratio and the vector changes in astigmatism When the blackboard size (width) was < 4 m, the corneal astigmatism change rate remained stable; when the blackboard size (width) was > 4 m, the astigmatism change accelerated and increased; When the blackboard size (width) was > 4.1 m, the changes in corneal astigmatism were < 0, indicating an increase in corneal astigmatism (Fig. 6(i)). When the window-to-ground area ratio was < 1:4, the rate of changes in corneal astigmatism decreased rapidly with an increase in the ratio; When the window-to-ground area ratio was > 1:5, the changes in corneal astigmatism were also < 0, showing a tendency for corneal astigmatism to increase. Once the window-to-ground area ratio was > 1:4, the changes in corneal astigmatism shifted to a positive increase, resulting in a continuous decrease in corneal astigmatism (Fig. 6(l)). With a fixed blackboard width, the absolute value of the changes in corneal astigmatism first decreased and then increased with the window-to-ground area ratio increasing, reaching its minimum at a ratio of about 0.27. At a constant window-to-ground area ratio, the absolute value of the changes in corneal astigmatism decreased with blackboard size (width) increasing, reaching its minimum at approximately 4.2 m (Fig. 6(m)). Relationship between blackboard size (width), window-to-ground area ratio and changes of corneal astigmatism Discussion The survey results indicated that numerous issues persisted in the teaching environment of high school classrooms in Fengxian District. Model analysis revealed that classroom per capita area, the blackboard size (height) and multimedia display to front seat distance influenced changes in uncorrected visual acuity. However, due to the instability of uncorrected visual acuity, which is easily affected by environmental and subjective factors, additional parameters such as spherical equivalent, corneal astigmatism, and astigmatism values at 90°−180°(J180) and 45°−135°(J45) were included in this study to further explore the relationship between classroom environmental factors and students' refractive status. The analysis demonstrated that classroom per capita area affected the change in spherical equivalent, while window-to-ground area ratio and blackboard reflectance ratio influenced the change in astigmatism values at J180 and J45. Furthermore, the blackboard size (width) and window-to-ground area ratio impacted the change in students' corneal astigmatism. Given that the study's target population was high school students, most of whom were myopic, and had a more relaxed accommodation compared to younger children, the autorefraction results were considered relatively more reliable. Many scientific studies [18, 19] and large-scale epidemiological investigations [2, 19–23] also used autorefraction results, therefore, this study utilized autorefraction results after non-cycloplegic refraction for high school students. Based on the nonlinear results of model analysis and comparisons with national standards for classroom environment-related indicators, it was evident that China's classroom environmental standards generally met the needs for controlling refractive state changes. Despite focusing on high school students, the study showed that adjusting certain environmental indicators could improve refractive state changes. For instance, increasing the classroom per capita area to over 1.6 m2 improved visual acuity, and a window-to-ground area ratio of 1:4 minimized astigmatism variation. Strict adherence to current standards for blackboard size, window-to-ground area ratio, and multimedia display to the front seat distance was crucial, with continuous efforts to improve compliance rates. These findings provide new perspectives on understanding the relationship between classroom environments and abnormal vision among adolescents, and emphasize the importance of considering both classroom environments and adolescent visual health in future public health strategies. Notably, only classroom per capita area influenced refractive progression, possibly due to reduced distractions, enhanced focus, and diverse teaching methods in a more spacious environment. Additionally, good ventilation and lighting were also speculated to positively impact vision. But further research is needed to elucidate the specific mechanisms. The survey found that although the qualification rates of blackboard surface average illumination (72.22%), blackboard surface illumination uniformity (97.15%), classroom desk average illumination (87.21%), and classroom desk illumination uniformity (85.83%) were higher than the 2020 Shanghai monitoring results, they still did not meet the requirement for a 100% compliance rate with school classroom lighting hygiene standards. The classroom lighting environment required urgent improvement. The weighted qualification rates for various classroom environmental factors,such as classroom per capita area, desk and chair allocation compliance rate, blackboard hanging height, blackboard reflectance ratio, lighting coefficient, window-to-ground area ratio, rear wall reflection ratio(79.39%, 18.31%, 47.47%, 41.03%, 43.15%, 62.01%, 65.59%) were all lower than the 2020 Shanghai monitoring results. Extensive research have shown the severe disease burden associated with abnormal vision among adolescents [10, 24] and the close relationship between classroom environment and student myopia [12, 22, 25–27]. Relevant studies [28, 29] on school classrooms and students' visual needs have emphasized the public health significances of preventing abnormal vision among adolescents [30, 31]. This study's monitoring results aligned with previous findings [32–34], highlighting the importance of improving the classroom visual environment. A generalized additive model was used to analyze the relationship between classroom environmental factors and students' uncorrected visual acuity and refractive status changes, a method widely applied in medical research [20, 21, 35–37]. Although numerous studies have demonstrated that environmental factors impact refractive status, the nature of this relationship—whether it follows a linear pattern or exhibits threshold effects—remains unclear. For instance, does an increase in light brightness lead to a continuous decrease in spherical equivalent, or is there a point where the spherical equivalent stabilizes or even increases? Traditional approaches, such as linear regression and mixed-effects models, struggle to capture such nonlinear associations. Therefore, we utilized GAM to investigate the nonlinear relationship between classroom environmental factors and refractive status, along with the potential presence of threshold effects. The limitations of this study are outlined as follows. First, given its observational nature, establishing causal relationships presents inherent challenges. To further validate the observed correlations and delve into underlying mechanisms, future research should adopt more rigorous designs, such as randomized controlled trials. In our subsequent work, we plan to incorporate behavioral factors, such as time spent on near-work activities, outdoor exposure, and screen time, to obtain a more holistic understanding of the relationship between classroom environmental factors and students' vision. Second, the current measurements of classroom environmental factors reflect macro-level assessments. Exploring the potential use of personal lighting devices for individual-level measurements could enhance the precision of exposure assessments in future studies. Third, to enable more in-depth analyses, future research should incorporate additional environmental monitoring indicators, including glare, flicker (strobe), color rendering index, and spatial apparent brightness. Lastly, this study utilized non-cycloplegic autorefraction results, it is important to note that this method may lead to an overestimation of myopia prevalence [38]. However, considering that the majority of the participants were myopic and previous research has indicated the relative reliability of autorefraction in myopic patients [39], as well as the feasibility of conducting this large-scale study [40], we opted to use non-cycloplegic autorefraction results instead of cycloplegic refraction. In conclusion, schools can improve the classroom visual environment by adjusting the layout, desk and chair allocation, lectern position, blackboard material, window area, and lighting. Health education for school teachers and health workers should be strengthened to promote correct methods and standards for selecting and adjusting these elements. Relevant departments should also enhance supervision and guidance of unqualified schools. Acknowledgements Not applicable. Authors' contributions Meng Qin (First author): Research design, obtain research grants, article writing and manuscript revision. Wei-xing Shi (Co-first author: Critical review and administrative support for the intellectual content of the article. Shan-xi Gao (Co-first author): Conduct research, technical or material support. Yue Niu (Co-author): Guide. Jing-wei Zheng (Corresponding author): Data Analysis, manuscript revision. All authors reviewed the manuscript. Funding Fengxian District (Social) Science and Technology Development Fund Project (Project No.:20221403). Data availability Data is provided within the manuscript and supplementary information files. Declarations Ethics approval and consent to participate Prior to the commencement of the study, all enrolled students and their guardians were fully informed and provided signed informed consent. This study followed the tenets of the Declaration of Helsinki and was approved by Shanghai Municipal Center for Disease Control and Prevention Ethical Review Committee (approval number: KY-2024–10). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Meng Qin, Wei-xing Shi and Shan-xi Gao have contributed equally to this work and share first authorship. References Associated Data Data Availability Statement Data is provided within the manuscript and supplementary information files.