Page 93 - 《南京医科大学学报》2026年第1期
P. 93

第46卷第1期          张慧文,祁红玉,张 华,等. 儿童创伤后癫痫风险预测模型构建与验证:基于队列研究的
                  2026年1月                 Meta分析[J]. 南京医科大学学报(自然科学版),2026,46(1):82-93                  · 87   ·


                                                表2 儿童PTE风险因素的Meta分析结果
                                   Table 2 Meta⁃analysis results of risk factors for pediatric PTE in children
                                          Number of                                 Sample    Heterogeneity analysis
                       Risk factor                            References
                                                                                                2
                                           studies                                 size(n)     I(%)        P
                  Age                        3               [19,21,30]                836      75.8      0.006
                  GCS score                  8        [18,19,20,21,24,27,28,30]     126 274     92.6      0.065
                  Open head injury           5            [20,21,26,27,28]         1 244 623    47.8      0.287
                  Immediate seisure          3               [18,24,25]                922      93.7     <0.010
                  Loss of consciusness       4             [21,27,28,29]           1 369 022    98.8     <0.001
                  Neurosurgical treatment    4             [19,26,28,30]            125 132     78.5      0.003
                  Intracranial hematoma      4             [19,20,21,22]              1 100     40.7      0.167
                  Cerebral contusion         6          [19,21,22,24,27,30]        1 245 443    82.0     <0.010
                  Subdural hemorrhage        4             [19,27,28,29]           1 369 056    95.5     <0.010
                  Subarachnoid hemorrhage    3               [19,23,27]            1 244 636     0.       0.999
                                              Effect                            Pooled result
                       Risk factor
                                              model         OR            95%CI           Z score          P
                  Age                        Random         1.059       0.904-1.240        00.71        <0.476
                  GCS score                  Random         1.466       0.977-2.199        01.85        <0.065
                  Open head injury            Fixed         1.750       1.310-2.340        03.78        <0.010
                  Immediate seisure          Random         3.043       1.933-4.790        04.81        <0.010
                  Loss of consciusness       Random         3.023       1.299-7.037        02.57        <0.010
                  Neurosurgical treatment    Random         1.256       0.643-2.453        00.67        <0.504
                  Intracranial hematoma       Fixed         3.796       2.700-4.892        06.79        <0.010
                  Cerebral contusion         Random         2.708       1.379-5.321        02.89        <0.010
                  Subdural hemorrhage        Random         1.302       1.165-1.438        18.66        <0.010
                  Subarachnoid hemorrhage     Fixed         4.143       3.150-5.135        11.90        <0.010
                   Gcs:Glasgow coma scale.

                                                     表3 风险因素的敏感性分析
                                               Table 3  Sensitivity analysis of risk factors

                                                      Fixed⁃effects model               Random⁃effects model
                         Risk factor
                                                  OR             95% CI             OR              95% CI
                    Age                          1.077          1.022-1.135        1.044          0.600-1.817
                    GCS score                    0.970          0.892-1.056        1.486          0.836-2.633
                    Open head injury             1.750          1.310-2.340        1.554          0.536-5.523
                    Immediate seisure            3.043          1.933-4.790        4.540          0.622-33.119
                    Loss of consciusness         1.856          1.744-1.975        2.950          1.566-5.555
                    Neurosurgical treatment      1.493          1.312-1.670        1.173          0.439-3.138
                    Intracranial hematoma        4.216          3.200-5.555        3.723          2.459-5.639
                    Cerebral contusion           4.320          3.566-5.231        2.648          1.173-5.981
                    Subdural hemorrhage          3.676          3.206-4.214        2.587          1.089-6.148
                    Subarachnoid hemorrhage      4.183          3.305-5.294        4.183          3.305-5.294


                估模型区分度,训练队列AUC为0.80(95%CI:0.74~                   究中使用该列线图预测 PTE 的发生风险会比对所
                0.87,P < 0.05,图 3A),内部验证队列 AUC 为 0.83             有患者实施干预方案更有效,在这个范围内,预测
               (95%CI:0.73~0.93,P < 0.05)(图3B),结果表明预测             模型的净收益显著高于两个极端情况,即所有患者
                模型具有良好区分能力。DCA 评估模型的临床有                           都被给予临床干预,结果见图4。Hosmer⁃Lemeshow
                效性。如果医生和患者的阈值率>20%,则在当前研                          拟合优度检验评价模型校准度,检验结果显示在建
   88   89   90   91   92   93   94   95   96   97   98