患者结肠镜检查前肠道准备失败风险预测模型的系统评价

Risk Prediction Model for Inadequate Bowel Preparation before Colonoscopy:A Systematic Review

目的系统地评价患者结肠镜检查前肠道准备失败的风险预测模型。方法计算机检索中国知网(CNKI)、中国生物医学文献数据库(SinoMed)、万方数据知识服务平台、维普中文科技期刊数据库(VIP)、PubMed MeDILINE、Cochrane Library、Embase、CINAHL、Web of Science数据库,检索出符合本研究的文献,由2名研究者独立提取信息,并使用PROBAST工具进行质量评价。结果共纳入8篇研究,其中6篇为前瞻性队列研究,2篇为回顾性病例对照研究。3篇研究开展了内部验证,1篇研究开展了外部验证。8篇研究的ROC曲线下面积为0.63~0.87,预测性能良好,但仍存在偏倚风险,主要是在数据分析部分未报告缺失数据的处理方法以及数据复杂性,部分研究筛选预测因子的方法不恰当,并且未对模型进行内外部验证。Meta分析结果显示,慢性便秘[OR=2.51,95%CI(2.15,2.92),Z=11.81,P<0.001]、糖尿病[OR=2.14,95%CI(1.79,2.56),Z=8.26,P<0.05]、服用抗抑郁药物[OR=5.46,95%CI(3.48,8.58),Z=7.37,P<0.001]是患者结肠镜检查前肠道准备失败的独立危险因素。结论肠道准备失败预测模型还处于发展阶段,各项模型均存在一定的偏倚风险,未来可开展模型的验证研究或开发本土化预测模型,并将其运用于临床,检验其有效性。医护人员应重点关注慢性便秘、糖尿病、服用抗抑郁药物患者的肠道准备情况,提前给予干预策略。

Objective To systematically evaluate the risk prediction model of patients with bowel preparation failure before colonoscopy.Methods We searched databases including CNKI,SinoMed,Wan fang Databases,VIP,PubMed,MeDLINE,Cochrane Library,Embase,CINAHL,and Web of Science databases for literature that met the criteria.Two researchers independently extracted the information,and used the PROBAST tool for quality evaluation.Results Eight studies were included,6 of which were prospective cohort studies and 2 retrospective case-control studies.Three studies carried out internal verification and one study external verification.The area under the ROC curve of the eight studies was 0.63~0.87.The prediction performance was good,but there was still a risk of bias.The main reason was that processing method of missing data and the complexity of the data are not reported in the data analysis part.The method of screening predictors in some studies was not appropriate.And the model had not been verified internally and externally.Meta-analysis showed that chronic constipation[OR=2.51,95%CI(2.15,2.92),Z=11.81,P<0.001],diabetes[OR=2.14,95%CI(1.79,2.56),Z=8.26,P<0.05],antidepressant use[OR=5.46,95%CI(3.48,8.58),Z=7.37,P<0.001].All of them were independent risk factors for inadequate bowel preparation before colonoscopy.Conclusion Bowel preparation failure prediction model is still in the development stage,and each model has a certain risk of bias.In the future,model verification studies or the development of localized prediction models can be carried out,and they can be used in clinical practice to test their effectiveness.Medical staff should focus on the patients with chronic constipation,diabetes or those taking antidepressant medication,and give intervention strategies in advance.