The vision health of primary and secondary school students has become a growing concern due to increasing academic pressure, the widespread use of electronic devices, and poor eye care habits. High myopia, as an advanced stage of myopia, emerging as a global public health issue. While many existing studies have explored the risk factors for myopia, few have specifically addressed the complex non-linear relationships between these factors and the development of high myopia. This study combines a nested case-control study with restricted cubic splines to develop a risk prediction model for high myopia in primary and secondary school students. By identifying high-risk individuals early, this model aims to delay or prevent the progression of high myopia, contributing to tertiary prevention of myopia, and positively impacting the academic and life quality of students.

To investigate the prevalence and risk factors of high myopia in primary and secondary school students, and conduct a risk prediction model to provide a scientific basis for myopia prevention.

A nested case-control study was conducted in 2023, involving students with moderate myopia from 12 schools in Hangzhou to establish a cohort. Myopia status among the students was monitored in accordance with the National Monitoring and Intervention Program for Common Diseases and Health-Related Factors in Students. Students who progressed to high myopia were classified as the case group, while the others formed the control group, vision care behaviors were surveyed for both groups. Lasso regression was used to select feature variables, followed by multivariate logistic regression analysis to identify factors influencing the development of high myopia among primary and secondary school students. A Nomogram was employed to visualize the risk prediction model. The model's performance was evaluated using the Hosmer-Lemeshow test, receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA). Additionally, restricted cubic splines were used to further clarify the relationship between age and the risk of high myopia.

A total of 2468 students from 12 primary and secondary schools were enrolled. Among them, 1 293 students did not progress to high myopia, while 1 175 students with moderate myopia progressed to high myopia, resulting in a high myopia incidence rate of 47.61% (1 175/2 468). Significant differences were observed between the two groups in terms of age, grade level, BMI, daily sleep time, geographic region, outdoor activity time, electronic device usage time, after-school homework time, and household lamp usage (P<0.05). Lasso regression identified eight feature variables: grade level, BMI, daily sleep time, geographic region, outdoor activity time, electronic device usage time, after-school homework time, and household lamp usage. Multivariate Logistic regression analysis revealed that the risk factors for high myopia in primary and secondary school students included being in middle school (OR=2.612, 95%CI=2.185-3.127), being overweight or obese (OR=2.140, 95%CI=1.458-3.169), being underweight (OR=1.807, 95%CI=1.430-2.290), sleeping after 22: 00 daily (OR=1.408, 95%CI=1.188-1.670), engaging in outdoor activities for 1-2 h/d (OR=1.371, 95%CI=1.122-1.675) or <1 h/d (OR=1.648, 95%CI=1.342-2.027), using electronic devices >2 h/d (OR=1.440, 95%CI=1.1191.856), and spending 1-2 h/d (OR=1.461, 95%CI=1.126-1.899) or >2 h/d (OR=1.534, 95%CI=1.218-1.935) on after-school homework (P<0.05). In contrast, being in high school was identified as a protective factor against high myopia (OR=0.560, 95%CI=0.419-0.743, P<0.05). A risk prediction model was constructed based on six variables: grade level, BMI, daily sleep time, outdoor activity time, electronic device usage time, and after-school homework time. The model achieved an area under the ROC curve of 0.840 (95%CI=0.825-0.855), demonstrating good fit, consistency, and applicability. Additionally, restricted cubic spline analysis indicated that the age group of 13~15 years was the high-risk period for developing high myopia.

The incidence of high myopia among primary and secondary school students was notably high. The risk prediction model could provide a scientific basis for the prevention and control of high myopia. Strengthening myopia prevention and control measures in middle school, along with improving students' vision care behaviors, was essential for reducing the occurrence of high myopia.

中小学生由于学习压力、电子产品普及以及不良用眼习惯，视力健康问题日益严重，成为近视的高发群体，高度近视作为近视的严重阶段，已成为全球范围内的公共卫生问题。尽管现有诸多研究对近视的危险因素进行了探讨，但鲜有研究针对性阐明风险因素与高度近视发生的复杂非线性关系，本研究结合巢式病例对照研究和限制性立方样条，开发一个中小学生高度近视风险预测模型，通过早期识别高风险个体，延缓或阻止高度近视的发展，实现近视的三级预防，对中小学生的学业和生活质量有积极意义。

探究中小学生高度近视的流行现状及危险因素，构建风险预测模型，为中小学生高度近视防控提供科学依据。

采用巢式病例对照研究，于2023年选取杭州市12所学校中度近视的学生作为研究对象建立队列，按照全国学生常见病和健康影响因素监测与干预工作方案，对纳入研究的中小学生开展近视状况监测，研究期间进展为高度近视的中小学生作为高度近视发生组，其余未进展为高度近视的中小学生作为对照组，对两组研究对象开展视力保健行为进行调查。采用Lasso回归筛选特征变量后进行多因素Logistic回归分析探究中小学生高度近视发生的影响因素，并采用列线图对风险预测模型可视化，同时采用Hosmer-Lemeshow检验、受试者工作特征（ROC）曲线、Calibration曲线、决策曲线分析（DCA）对模型性能进行评估，最后采用限制性立方样条进一步明确年龄与高度近视发生风险的关系。

12所中小学校共纳入2 468名学生，未进展为高度近视的学生1 293名，中度近视进展为高度近视的学生1 175名，高度近视发生率为47.61%（1 175/2 468）。两组学生年龄、年级、BMI、每日入睡时间、地区、户外活动时间、电子产品使用时间、课后作业时间、家用台灯比较，差异有统计学意义（P<0.05）。Lasso回归筛选出8个特征变量：年级、BMI、每日入睡时间、地区、户外活动时间、电子产品使用时间、课后作业时间、家用台灯。多因素Logistic回归分析结果显示，初中年级（OR=2.612，95%CI=2.185~3.127），超重或肥胖（OR=2.140，95%CI=1.458~3.169）、偏瘦（OR=1.807，95%CI=1.430~2.290），每日入睡时间在22：00之后（OR=1.408，95%CI=1.188~1.670），户外活动时间1~2 h/d（OR=1.371，95%CI=1.122~1.675）、<1 h/d（OR=1.648，95%CI=1.342~2.027），电子产品使用时间>2 h/d（OR=1.440，95%CI=1.119~1.856），课后作业时间1~2 h/d（OR=1.461，95%CI=1.126~1.899）、>2 h/d（OR=1.534，95%CI=1.218~1.935）为中小学生高度近视发生的危险因素（P<0.05）；而高中年级（OR=0.560，95%CI=0.419~0.743）为中小学生高度近视发生的保护因素（P<0.05）。基于年级、BMI、每日入睡时间、户外活动时间、电子产品使用时间、课后作业时间等6个变量构建的预测模型ROC曲线下面积为0.840（95%CI=0.825~0.855），具有良好的拟合优度、一致性、应用性，限制性立方样条分析显示13~15岁为高度近视的高发年龄段。

中小学生高度近视发生率较高，风险预测模型为高度近视的预防和控制提供了科学依据，应在初中年级加强近视防控措施，改善学生的视力保健行为，降低高度近视的发生。