Risk Factors Analysis And Prediction Research Of Difficult Weaning In Children With Mechanical Ventilation

Objective:To analyze the influencing factors of weaning difficulty in children with mechanical ventilation,and the early prediction models were constructed on this basis.It is expected to provide scientific basis for ICU nursing staff to implement personalized caring in weaning and to improve the weaning success rate of children with mechanical ventilation.Methods:Part 1: The risk factors and evaluation indexes related to difficult weaning of children with mechanical ventilation were screened through literature review.This study included children who were received mechanical ventilation treatment in the pediatric intensive care unit during October1,2018 to December 31,2019.Then the R language statistical software was used to divide them into two groups: the modeling group and the validation group.According to the influence of mechanical ventilation time on children,two key timing were selected(within 48 ～ 72 hours of MV and within 1 hour before the first SBT).In the modeling group,univariate comparative analysis was conducted to screen out the risk factors with statistical differences.The screened risk factors were analyzed by Logistic regression to find the independent risk factors at different timing,and then the nomograms prediction models at two different timing were constructed based on this,and the internal test of the models was carried out in the modeling group.Part 2: According to the prediction models constructed at two different timing,the accuracy and fitting degree of the model were tested in verification group.The accuracy of the model was completed by the AUC analysis(Area Under the ROC Curve).The fitting degree of the model is completed by the K-S calibration curve.Secondly,according to the predicted probability distribution of the two models,the risk degree of difficult weaning was divided into three levels:low risk,medium risk and high risk.The clinical characteristics of children with different risk levels were compared and analyzed.Results:1.A total of 205 children were included in this study,and the incidence of weaning difficulty was47.8%.Univariate analysis results of the modeling group showed that,The median number days of tracheal intubation was 15,the median number days of ICU hospitalization was 25,the median starting time of weaning was 8 days,the median time of weaning was 7 days,and 9 patients(6.3%)failed in liberation.Compared with the simple weaning graop,the differences were statistically significant(P<0.05).2.Logistic regression results: the Lung ultrasound score(MV 48～72h)(OR=2.374,955 CI 1.729,3.420,P<0.001)and the PCIS score(OR=0.750,955 CI 0.619,0.885,P=0.001)were elected into the regression modes 1.The Lung ultrasound score(within 1 hour before the first SBT)(OR=3.420,955 CI 2.215,5.857,P<0.001)and RSBI value(SBT lasts for 30 minutes)(OR=2.178,955 CI 1.078,4.633,P=0.035)were elected into the regression model 2.3.The test result of the nomogram model in the modeling group is that,The AUC values of the two models were 0.874 and 0.919,respectively.The accuracy of the model is high.The K-S calibration curve shows that the predicted line and the ideal line do not fit closely,indicating that the model’s fitting degree needs to be further improved.4.There was no statistical difference in baseline data between the modeling group and the verification group.The model can be validated and analyzed in a validation group.The AUC values of the two models in the validation group were 0.846 and 0.919.The k-S calibration curve shows that the degree of coincidence between the predicted line and the ideal line of the two models is higher than that of the modeling group.However,it still does not fit closely,which indicates that there are some differences between the predicted results and the actual results.According to the distribution law of predicted probability,the risk degree of difficult weaning is divided into three grades: low risk,medium risk and high risk.The population distribution was different between model 2 and model 1 in the three risk levels.Model 2 has more people at high risk than Model 1.Secondly,there were significant inter-group differences in the incidence of weaning and risk factor indicators,MV days,length of stay in ICU,length of weaning,and starting time of weaning at three different risk levels.Conclusion:1.The first independent risk factors for weaning outcomes in children were the Lung ultrasound score(MV 48～72h)and the PCIS score.The second independent risk factors were the Lung ultrasound score(within 1 hour before of SBT)and RSBI value(SBT lasts for 30 minutes).2.The two predictive models at different time points can accurately identify the at-risk population with difficult weaning.By using the nomogram model,the risk probability of weaning difficulties at different time points and the main causes of the risk of weaning difficulties can be rapidly estimated.The overall accuracy of the two models is relatively high and the fitting degree is slightly poor.Combined with the stratified assessment of the risk degree of weaning difficulty,ICU nursing staff can further clarify the risk level and risk trend of different stages of weaning difficulty for children.3.The risk level of difficult weaning increased with the extension of mechanical ventilation time.This is accompanied by an increased incidence of difficult weaning,longer weaning time and longer length of ICU stay,and a higher failure rate of liberation.Model 1 has the advantage of better early prediction for mechanical ventilation within 48 ～ 72 hours.Model 2 is applied before weaning test and has the advantage of more accurate weaning readiness evaluation.The combined use of the two models can make up for the deficiency of traditional methods for predicting children’s weaning outcome and guide nursing staff to implement personalized caring in different weaning stages.