Development And Validation Of Risk Prediction Model For Intensive Care Unit Acquired Weakness

Objective:1、To systematically evaluate the risk prediction models of ICU acquired weakness and analyze the advantages and disadvantages of the existing models;2、Identify the risk factors of ICU acquired weakness through meta-analysis;3、To construct a risk prediction model of ICU acquired weakness,to provide a useful tool for clinical workers to early find the high-risk people of ICU acquired weakness,improving the management of ICU acquired weakness,reducing the clinical incidence of ICU acquired weakness.Methods:1、Literature analysis: extensively search and read the literature at home and abroad to understand the research status of ICU acquired weakness risk prediction model.Retrieve the literature related to risk factors of ICU acquired weakness and analyze the publication year,literature type,and research content of the literature.2、Systematic review: systematically search the research related to ICU acquired weakness prediction models in domestic and foreign databases such as The Cochrane library,Pub Med,Embase,Web of Science,Wanfang database,CNKI,etc.,and trace the relevant references by snowballing.Extract literature data and use the Prediction model Risk Of Bias ASsessment Tool to evaluate the methodology of the included model and integrate the content.3、Computer search of domestic and foreign databases such as The Cochrane library,Pub Med,Embase,Web of Science,Wanfang database,CNKI,and systematic search of cohort studies or case-control studies related to risk factors of ICU acquired weakness,and through snowballing,trace the relevant references;Extract the literature materials that meet the inclusion criteria;Use the Newcastle-Ottawa scale to evaluate the quality of the included literature,and use the Rev Man 5.3 software to conduct a meta-analysis of the risk factors of the included study.4、Logistic modeling and verification: Substituting the combined risk values of ICU acquired weakness obtained by Meta analysis into the Logistic model to establish a risk prediction model;Cases from 2 tertiary A hospitals were collected and SPSS25.0software were used to draw Receiver Operating Characteristic curve to evaluate the predictive performance of the model.The consistency of the two diagnostic methods(predictive model and MRC)was calculated by Kappa coefficient.Results:1、Results of systematic review:(1)A total of 8 models were included,separately from France,Czech Republic,China,Argentina,United States,Netherlands and Spain;(2)Among the eight models,seven models had high risk of bias,and all models had good applicability.2、Results of Meta-Analysis:(1)A total of 20 articles with good quality were included,and the NOS score was between 6 and 7;(2)The results of meta-analysis showed that Age OR(1.03 [1.02,1.04]),APACHE OR(1.57 [1.20,2.06]),Shock OR(2.28 [1.55,3.36]),Mechanical Ventilation OR(1.47 [1.22,1.76]),MODS OR(3.04[1.94,4.77]),Sepsis OR(2.86 [1.49,1.59]),Neuromuscular Blocker OR(2.04 [1.47,2.84]),Vasoactive Medications OR(4.93 [2.64,9.20]),Steroid OR(4.22 [1.81,9.87]),Aminoglycoside OR(3.21 [1.94,5.31]),Hyperlactate OR(4.76 [2.50,9.07]),Hyperglycemia OR(1.76 [1.39,2.23]),and Renal Replacement Therapy OR(7.19[3.88,13.32])are risk factors of ICU acquired weakness.3、Logistic modeling and validation results:(1)The final logarithmic function model of ICU acquired weakness based on meta-analysis was Logit(P)=-0.781 + 0.030 Age + 0.451 APACHE + 0.824 Shock + 0.385 Mechanical Ventilation + 1.051 Sepsis+ 0.713 NMBA + 1.595 Vasoactive Medications + 1.440 Steroids + 1.560 Hyperlactate + 0.565 Hyperglycemia + 1.973 Renal Replacement Therapy;(2)A total of 137 patient data from 2 tertiary A hospitals were collected.43 of them had ICU acquired weakness and the incidence was 31.39%.;(3)The Area Under the Receiver Operating Characteristic Curve of the model was 0.787,95% Confidence Interval(0.702,0.872),the cut-off value was 0.907,the sensitivity was 0.698,and the specificity was 0.798.The kappa coefficient between predictive model and MRC was 0.477.Conclusion:1、ICU acquired weakness prediction models have better performance,but there is a certain bias in the method of development and report.Future research should standardize the whole process of model development,and focus on model validation and updating,so as to provide high-quality and practical models for clinical.2、The general condition of critically ill patients,the severity of the disease,and treatment factors can all increase the risk of ICU acquired weakness.The evaluation and prevention of patients with ICU acquired weakness should be considered from many aspects.3、The risk prediction model of this study has been verified to have good discrimination,and can be used as a tool for early clinical assessment of ICU acquired weakness.When the predicted incidence of ICU acquired weakness is greater than or equal to 90.70%,it can be considered as high-risk population of ICU acquired weakness.