首页|医院感染预测模型研究方法学与报告质量评价

医院感染预测模型研究方法学与报告质量评价

扫码查看
原文链接
  • 万方数据
  • 维普
目的 评价当前我国医院感染预测模型研究方法学质量和报告质量,以期为规范研究过程和报告方法提供参考.方法 系统检索中英文数据库截至2022年底的国内医院感染相关预测模型研究,研究团队按纳入、排除标准独立筛选文献、交叉核对提取资料后,应用临床预测模型偏倚风险和适用性的评估工具(prediction model risk of bias assessment tool,PROBAST)评价方法学质量,使用个体预后与诊断预测模型研究报告规范(transparent reporting of a multivariable pre-diction model for individual prognosis or diagnosis,TRIPOD)声明评价研究报告质量.结果 共纳入81篇医院感染预测模型研究,PROBAST偏倚风险评估的平均完成度为58.11%±13.88%,TRIPOD声明报告质量的平均完成度为56.11%±16.35%.方法学缺陷主要存在于数据来源、未考虑数据复杂性与采用单因素分析筛选预测因子等,报告质量中风险分层、样本量估算与补充信息等条目未报告或报告不全面.结论 国内医院感染预测模型研究存在方法学缺陷与不完整的报告情况,限制了研究结果的可靠性与适用性,有较大的改进空间.
Assessment of Methodological and Reporting Quality of Hospital Infections Prediction Model
Objective To evaluate the quality of prediction model on healthcare-associated infections in China,so as to standardize research process and reporting methods.Methods It performed a literature search for healthcare-associated infections prediction model studies published using the following databases by the end of 2022.After independently screening the literature and cross-checking the extracted data according to the inclusion and exclusion criteria,the research team applied the prediction model risk of bias assessment tool(PROBAST)to evaluate the methodological quality,and the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis(TRIPOD)statement to evaluate the quality of study reporting.Results A total of 81 healthcare-associated infections prediction studies were identified.Their median PROBAST overall adherence were 58.11%±13.88%,median TRIPOD adherence were 56.11%±16.35%.The main methodological flaws involved participants defined,ignored complexities in data,and omitted missing data.The reporting flaws lay in the items of risk groups,sample size,and supplementary information.Conclusion There are methodological deficiencies and incomplete reporting of domestic hospital infection prediction modelling studies,which limit the reliability and applicability of the results and leave much room for improvement.

hospital infectionsprediction modelprediction model risk of bias assessment tooltransparent reporting of a multivariable prediction model for individual prognosis or diagnosis

单娇、包小源、龚志忠、曹煜隆

展开 >

首都医科大学附属北京积水潭医院 北京 100035

北京大学医学部医学信息学中心 北京 100191

首都医科大学附属北京中医医院 北京 100010

北京大学人民医院 北京 100044

展开 >

医院感染 预测模型 临床预测模型偏倚风险和适用性的评估工具 个体预后与诊断预测模型研究报告规范

国家卫生健康委医院管理研究所医疗机构感染预防与控制科研项目

GY2023044

2024

中国医院管理
中国医院管理杂志社

中国医院管理

CSTPCD北大核心
影响因子：2.15
ISSN：1001-5329
年,卷(期)：2024.44(11)