目的：探讨飞秒激光辅助准分子激光原位角膜磨镶术(FS-LASIK)矫正中高度近视术后1 a屈光回退的危险因素并构建预测模型。

方法：回顾性分析2017-06/2018-11于西安高新医院行FS-LASIK手术矫正中高度近视患者400例800眼临床资料，将患者按照3：1的比例随机分为建模组(n=300例)和验证组(n=100例)，依照患者术后1 a屈光回退发生情况将建模组分为回退组与未回退组两个亚组，观察其角膜曲率和角膜厚度变化情况，采用Logistic回归分析筛选FS-LASIK矫正中高度近视患者术后1 a屈光回退的危险因素，基于回归系数构建预测模型，采用受试者工作特征曲线(ROC)评估模型区分度。

结果：建模组44眼、验证组15眼于术后1 a发生屈光回退。建模组中未回退组术后6、12 mo时角膜前表面曲率低于回退组(均P<0.05)； 术后1、3、6、12 mo时回退组角膜增生程度大于未回退组(均P<0.05)； FS-LASIK矫正中高度近视患者术后1 a屈光回退发生概率值回归方程为：P=1/\〖1+e^{-(-5.989+0.127×年龄+2.019×术前屈光度-0.022×术前角膜中央厚度+0.043×切削深度-1.569×切削光学区直径)}\〗，Hosmer-Lemeshow检测回归方程拟合优度(P=0.818)，利用建模组数据进行内部验证，ROC曲线下面积为0.890(95%CI 0.843-0.937)，灵敏度为81.82%，特异度为84.71%； 利用验证组数据进行外部验证，ROC曲线下面积为0.838(95%CI 0.717-0.959)，灵敏度为80.00%，特异度为87.57%。

结论：构建的风险模型判别效度良好，可用于识别中高度近视患者行FS-LASIK术后1 a屈光回退高危人群。

METHODS: A retrospective analysis was performed on the clinical data of 400 patients(800 eyes)with moderate and high myopia undergoing FS-LASIK correction in Xi'an Gaoxin Hospital from June 2017 to November 2018, and the patients were randomly divided into modeling group(n=300)and verification group(n=100)according to a ratio of 3:1. The modeling group was divided into regression group and non-regression group according to the occurrence of refractive regression at 1 a after surgery. The changes of corneal curvature and corneal thickness were observed. Logistic regression analysis was used to screen the risk factors of refractive regression in patients with moderate and high myopia at 1 a after FS-LASIK correction, and the prediction model was built based on the regression coefficient. Receiver operating curve(ROC)was used to evaluate model differentiation.

RESULTS:Refractive regression occurred in 44 eyes of the modeling group and 15 eyes of the verification group at 1 a after surgery. The anterior corneal surface curvature in the modeling group was lower than that in the regression group at 6 and 12 mo after surgery(all P<0.05). The degree of corneal hyperplasia in the regression group was greater than that in the non-regression group at 1, 3, 6 and 12 mo after surgery(all P<0.05). The regression equation of the probability value of refractive regression in patients with moderate and high myopia corrected by FS-LASIK at 1 a after surgery is as follows:P=1/\〖1+e^{-(-5.989+0.127×age+2.019×preoperative diopter-0.022×preoperative central corneal}

^{thickness+0.043×depth of cutting-1.569×diameter of cutting optical region)}\〗, Hosmer-Lemeshow detected the goodness of fit of the regression equation(P=0.818). Internal verification using the modeling group data showed that the area under ROC curve was 0.890(95%CI: 0.843-0.937), the sensitivity was 81.82%, and the specificity was 84.71%. The area under ROC curve was 0.838(95%CI: 0.717-0.959), the sensitivity was 80.00%, and the specificity was 87.57%.

CONCLUSIONS:The established risk model has good discriminating validity and can be used to identify the high-risk group of refractive regression at 1 a after FS-LASIK in patients with moderate and high myopia.