不同预测工具对心肺复苏后神经功能预后预测能力的比较:诊断试验网状Meta分析OA
Comparison of the predictive ability of different types of prediction tools for neurological function after cardiopulmonary resuscitation:a network Meta analysis of diagnostic tests
目的 通过诊断试验网状Meta分析整合直接及间接证据,系统比较不同预测工具对心脏骤停后神经功能结局的预测效能.方法 遵循PRISMA-DTA指南,系统检索 Medline、Embase、Web of Science、Cochrane Library及Scopus数据库2020年1月至2025年12月间发表的文献.纳入以脑功能分类(CPC)为金标准、可提供四格表数据的前瞻性队列或回顾性研究.采用贝叶斯分层随机效应模型合并敏感度、特异度及诊断比值比(DOR),通过累积排序概率(SUCRA)对预测工具进行效力排序,并运用贝叶斯选择模型评估发表偏倚.结果 共纳入13项研究(总样本量n=6 927),评估了 5种预测工具(AASSEP、MRI、NSE、EEG、CT)及其亚组.主分析显示,AASSEP与MRI具有最优综合诊断效能,DOR分别为13.06(95％CI:7.74～21.12)和 7.82(4.58～12.47),SUCRA 值分别为 0.97 和 0.66.AASSEP 敏感度较高(0.82,95％CI:0.75～0.88),MRI特异度更优(0.82,0.75～0.87).CT呈现极端准确性分布,特异度达0.93(0.89～0.96),但敏感度仅0.19(0.14～0.26).NSE 与 EEG 总体 DOR 分别为 5.41(3.92～7.46)和 6.65(4.70～9.24).亚组分析显示,NSE high risk(＞60 μg/L)DOR高达 38.30(19.01～77.18),特异度 0.71(0.59～0.81);NSE low risk(＜33 ng/mL)敏感度达 0.89(0.81～0.94),适用于早期筛查;EEG良性模式DOR为2.34(1.53～3.60).贝叶斯选择模型未提示显著发表偏倚(所有工具β_se-lection的95％CI均包含0).结论 AASSEP与MRI对心脏骤停后神经功能结局具有最优综合预测效力,前者更适用于筛查,后者更适用于确诊.NSE high risk对不良预后确诊价值极高,而lowrisk适合早期排除.临床实践中应避免单独依赖中等NSE值或恶性EEG模式进行决策,推荐采用多模式、分阶段评估策略.
Objective To systematically compare the predictive performance of different diagnostic tools for neu-rological outcome after cardiac arrest by integrating direct and indirect evidence through diagnostic test accuracy network Meta-analysis(DNMA).Methods Following PRISMA-DTA guidelines,we systematically searched Med-line,Embase,Web of Science,Cochrane Library,and Scopus from January 2020 to December 2025.Prospec-tive cohorts and retrospective studies providing 2×2 contingency table data with Cerebral Performance Category(CPC)as the reference standard were included.Bayesian hierarchical random-effects models were constructed into pool sensitivity,specificity,and diagnostic odds ratio(DOR).The surface under the cumulative ranking curve(SUCRA)was used to rank the predictive utility of each tool,and Bayesian selection models were applied to assess publication bias.Results A total of 13 studies(n=6 927)evaluating five predictive tools(AASSEP,MRI,NSE,EEG,and CT)and their subgroups were included.In the main analysis,AASSEP and MRI demon-strated the highest comprehensive diagnostic performance with DORs of 13.06(95％CI:7.74-21.12)and 7.82(4.58-12.47),and SUCRA values of 0.97 and 0.66,respectively.AASSEP showed high sensitivity(0.82,95％CI:0.75-0.88),while MRI exhibited superior specificity(0.82,0.75-0.87).CT displayed an extreme accuracy profile with specificity of 0.93(0.89-0.96)but sensitivity of only 0.19(0.14-0.26).The overall DORs for NSE and EEG were 5.41(3.92-7.46)and 6.65(4.70-9.24),respectively.Subgroup analyses revealed that NSE high-risk group(＞60 ng/ml)had a DOR of 38.30(19.01-77.18)with specificity of 0.71(0.59-0.81),whereas NSE low-risk group(＜33 ng/ml)showed high sensitivity of 0.89(0.81-0.94),suitable for early screening.The benign EEG pattern demonstrated a DOR of 2.34(1.53-3.60).Bayesian selection models indicated no significant publication bias(95％CI of β_selection for all tests included 0).Conclusion AASSEP and MRI possess the optimal comprehensive predictive power for neurological outcome after cardiac arrest;the former is more suitable for screening and the latter for confirmation.NSE high-risk group of-fers high confirmatory value for poor outcome,while low-risk group is appropriate for early exclusion.Clinical practice should avoid relying solely on medium-range NSE values or malignant EEG patterns,and adopt a multi-modal,staged assessment strategy.
杨国盛;蔡亚林;叶青青;蒋雪梅;宋仁杰;喻安永;段海真
遵义医科大学附属医院急诊科,贵州遵义 563000联勤保障部队第九二五医院急诊科,贵州贵阳 550000遵义医科大学附属医院急诊科,贵州遵义 563000遵义医科大学附属医院急诊科,贵州遵义 563000遵义医科大学附属医院急诊科,贵州遵义 563000遵义医科大学附属医院急诊科,贵州遵义 563000遵义医科大学附属医院急诊科,贵州遵义 563000
医药卫生
心脏骤停自主循环恢复网状Meta分析预测价值神经功能预后
cardiac arrestrestoration of spontaneous circulationnetwork Meta analysispredictive valueneurological prognosis
《遵义医科大学学报》 2026 (5)
532-544,13
国家自然科学基金资助项目(NO:82260385)贵州省科技厅科技计划项目[NO:黔科合基础-ZK(2023)580]贵州省卫生健康委科学技术基金资助项目(NO:gzwkj2023-103).
评论