摘要
目的 探讨重症监护病房(intensive care unit,ICU)患者亚谵妄(subsyndromal delirium,SSD)风险预测模型的偏倚风险及其适用性。方法 选取中国知网、万方数据库、维普中文科技期刊全文数据库、中国生物医学文献服务系统、Pubmed、Embase、CINAHL、Psycinfo、Web of Science、Wiley及Medline等数据库建库至2025年3月公开发表的关于ICU患者SSD预测模型的队列研究、病例对照研究。由2名研究者独立筛选文献、提取资料,依据预测模型研究的偏倚风险评价工具对纳入研究的偏倚风险和适用性进行评价。结果 共检索1192篇文献,最终纳入10篇文献。纳入的10个风险预测模型AUC/C-index为0.710~0.956,3个模型进行了外部验证。10个模型中出现频率最高(出现次数≥3次)的预测因子包括年龄、APECHEⅡ评分、约束、疼痛评分。10个模型的偏倚风险均为高风险,主要原因为样本量不足、未报告缺失值及缺失值的处理、模型采用单因素分析法筛选变量、缺少外部验证;9个模型适用性评价高,1个模型适用性不清楚。结论 ICU患者SSD风险预测模型整体预测性能表现良好,但整体偏倚风险高,建议集合临床医生、统计学家及方法 学专家开展大样本多中心研究,选择合适的变量筛选方法 ,增加外部验证,开发适合ICU患者SSD的风险预测模型。
Objective To explore the risk of bias and applicability of a risk prediction model for subsyndromal delirium(SSD)in intensive care unit(ICU)patients.Methods The cohort study and randomized controlled trial studies on ICU patient SSD prediction models published in databases such as China National Knowledge Infrastructure(CNKI),Wanfang Database,VIP Chinese Science and Technology Journal Full text Database,China Biomedical Literature Service System,Pubmed,Embase,CINAHL,Psycinfo,Web of Science,Wiley,and Medline until from the establishment of the database to March 2025 were selected.Two researchers independently screened literature,extracted data,and evaluated the bias risk and applicability of the included studies based on the bias risk assessment tool of the predictive model research.Results A total of 1192 articles were retrieved,and 10 articles were ultimately included.The AUC/C-index of the 10 risk prediction models were 0.710-0.956,and three models underwent external validation.The most frequently occurring predictors(occurring≥3 times)in the 10 models were age,APACHEⅡscore,restraints,and COPT score.The bias risk of all 10 models were high,mainly due to insufficient sample sizes,lack of reporting on missing values and their handling,use of univariate analysis for variable selection,and absence of external validation.Nine models had high applicability evaluations,while one model has unclear applicability.Conclusions The overall predictive performance of the SSD risk prediction model for ICU patients is good,but the overall risk of bias is high.It is recommended to conduct a large-sample,multi-center study involving clinicians,statisticians,and methodologists,select appropriate variable screening methods,increase external validation,and develop a risk prediction model suitable for SSD in ICU patients.
作者
夏玲丽
刘春生
史广玲
Xia Lingli;Liu Chunsheng;Shi Guangling(Department of Intensive Care Unit,Nanjing Brain Hospital Affiliated to Nanjing Medical University,Nanjing 210029,China)
出处
《北京医学》
2025年第6期483-491,共9页
Beijing Medical Journal
基金
南京市医学科技发展项目计划[宁卫规财(2017)5号]。
关键词
重症监护病房
亚谵妄
风险预测模型
系统评价
intensive care unit(ICU)
subsyndromal delirium(SSD)
risk prediction model
systematic review
