Abstract:

Objective To systematically analyze and evaluate the prediction model of fall risk in convalescent stroke patients,in order to provide references for clinical practice. Methods Original articles in Chinese and English were systematically searched from Chinese databases（CNKI and Wanfang） and English databases（PubMed,Embase,Cochrane） published prior to June 2021. There were 2 researchers who independently screened the literature,extracted information,and analyzed and compared the characteristics of the population,study types,predictive factors,model construction methods and model prediction results. Results This systematic review included 12 pieces of the literature on the development of predictive models,including 15 models. In these models,10 models reported the area under the curve and 4 models simultaneously reported the confidence interval of the area under the curve;4 models showed good differentiation in the internal verification population. In addition,only 6 models were tested for goodness of fit test,4 of them reported P value,and the results showed that the consistency was good. All included studies have low concerns regarding applicability,but with a certain bias,which mainly due to insufficient number of dependent events in the field of analysis,neglect of missing data,screening predictors based on univariate analysis,lack of model performance evaluation and over-simulation of the model. Conclusion There are still some shortcomings in the fall risk prediction model of convalescent stroke patients. Future research should improve the research design and research report,and carry out internal and external verification.

Key words: Stroke, Falls, Risk factors, Forecasting, Systematic Review, Nursing Care

摘要：

目的 系统评价康复期脑卒中患者跌倒风险预测模型,以期为临床实践提供参考依据。 方法 系统检索中国知网、万方数据知识服务平台、PubMed、Cochrane和Embase数据库截至2021年6月发表的康复期脑卒中患者跌倒风险预测模型相关文献,语种限定为中文和英文。由2名研究者独立筛选文献和提取数据,对纳入研究的人群特征、研究类型、预测因素、模型构建方法及模型预测结果等进行分析和比较。 结果 共纳入12项研究,包含15个模型,其中10个模型报告了模型的曲线下面积,有4个模型同时报告了曲线下面积的可信区间,4个模型表现出良好的区分度。仅有6个模型进行了拟合优度检验,其中4个模型报告了P值,结果显示一致性良好。12项研究的适用性较好,但存在较高的偏倚风险,主要是因为分析领域的应变量事件数不足、忽略缺失数据、基于单变量分析筛选预测因子、缺乏模型性能评估以及模型过度拟合。 结论 康复期脑卒中患者跌倒风险预测模型尚存在一些不足,未来研究应完善研究设计和研究报告,并进行内部验证和外部验证。

关键词: 脑卒中, 跌倒, 危险因素, 预测, 系统评价, 护理