目的：探讨早产儿视网膜病变(ROP)的临床特征及患病的影响因素分析，构建预测早产儿ROP的列线图模型。

方法：回顾性研究，选择2022年1月至2025年9月在我院进行眼底检查的早产儿进行分析，采用RetCam Ⅲ系统进行眼底检查，记录ROP发生情况。按7：3比例将数据拆分为训练集和验证集，对训练集数据采用卡方检验进行单因素分析，采用二分类多因素Logistic回归进行多因素分析，多因素筛选出的变量构建列线图并进行验证。

结果：早产儿3 841例中ROP 428例，发生率11.43%，Ⅰ期138例(32.24%)，Ⅱ期151例(35.28%)，Ⅲ期103例(24.07%)，Ⅳ期33例(7.71%)，Ⅴ期3例(0.70%)。训练集和验证集临床资料比较无差异(均P>0.05)。多因素分析结果显示，新生儿败血症、机械通气、输血治疗、凝血功能异常、支气管肺发育不良(BPD)、新生儿呼吸窘迫综合征(NRDS)、奶粉喂养、无创吸氧时间>1 wk是早产儿ROP的风险因素(均P<0.05)，出生体质量(1 500-2 499 g、≥2 500 g)、出生胎龄(32-34 wk、35-36 wk)、体质量增长≥20 g/d、5 min Apgar≥8分是早产儿ROP的保护性因素(均P<0.05)。列线图预测模型在训练集和验证集曲线下面积(AUC)分别为0.890和0.907，灵敏度分别为80.67%和82.81%，特异度分别为83.18%和85.14%。训练集和验证集中校准曲线均趋于理想曲线，Hosmer-Lemeshow拟合优度检验显示模型的预测值与实际观测值间拟合度较好(χ²=12.918，P=0.115； χ²=4.047，P=0.853)。临床决策曲线表明训练集与验证集中均具有较高的净收益。

结论：早产儿ROP发生率为11.43%，基于多因素的Logistic回归构建的列线图模型整合了出生体质量、胎龄、败血症、机械通气等关键风险与保护因素，具有较高的预测价值、良好的校准度及较高的净收益，可为早产儿ROP早期个体化风险评估提供直观有效的工具。

METHODS: This retrospective study enrolled premature infants who underwent fundus examinations in the hospital from January 2022 to September 2025 for analysis. Fundus examinations were performed using the RetCam III system, and the occurrence of ROP was recorded. The data were split into a training set and a validation set at a ratio of 7:3. Univariate analysis was conducted using the Chi-square test and multivariate analysis was performed using binary Logistic regression on the training set data. Variables identified in the multivariate analysis were used to construct a nomogram, which was subsequently validated.

RESULTS: The incidence of ROP(428 cases)among the 3 841 premature infants was 11.43%, with 138 cases(32.24%)in stage I, 151 cases(35.28%)in stage II, 103 cases(24.07%)in stage III, 33 cases(7.71%)in stage IV, and 3 cases(0.70%)in stage V. No statistically significant differences were found in the clinical data between the training and validation sets(all P>0.05). Multivariate analysis identified neonatal sepsis, mechanical ventilation, transfusion therapy, coagulation dysfunction, bronchopulmonary dysplasia(BPD), neonatal respiratory distress syndrome(NRDS), formula feeding, and non-invasive respiratory support duration >1 wk as risk factors for ROP(all P<0.05). Birth weight(1 500-2 499 g, ≥2 500 g), gestational age(32-34 wk, 35-36 wk), weight gain rate ≥20 g/d, and 5-minute Apgar score ≥8 were identified as protective factors(all P<0.05). The area under curve(AUC)of the nomogram prediction model was 0.890 in the training set and 0.907 in the validation set, with sensitivity of 80.67% and 82.81%, and specificity of 83.18% and 85.14%, respectively. The calibration curves in both sets approached the ideal curve, and the Hosmer-Lemeshow goodness-of-fit test indicated good agreement between the predicted and observed values(χ²=12.918, P=0.115; χ²=4.047, P=0.853). The decision curve analysis demonstrated high net benefits in both the training and validation sets.

CONCLUSION: The incidence of ROP in premature infants was 11.43%. The nomogram model, constructed based on multivariate Logistic regression and integrating key risk and protective factors such as birth weight, gestational age, sepsis, and mechanical ventilation, demonstrates high predictive value, good calibration, and high net benefit. It can serve as an intuitive and effective tool for early individualized risk assessment of ROP in premature infants.