Construction And Validation Of Predicting Models And Scales For Critical And Severe Trauma Patients’ Outcomes In The Emergency Department

Objective1.To screen the predictive factors that can predict the prognosis of critically ill patients in the Emergency Department（ED）,and to construct a prediction model.Transforming the prediction model into a simple scoring tool,namely,the critical patients’outcome scale for emergency department（CPOS-ED）,and comparing the CPOS-ED with modified early warning score system（MEWS）、rapid acute physiology score（RAPS）,rapid emergency medicine score（REMS）which are commonly used in the ED to evaluate its efficacy.2.To screen the predictive factors that can predict the prognosis of severe trauma patients in the ED,and to construct a prediction model.Transforming the prediction model into a simple scoring tool,namely,the severe trauma patient’outcome scale in ED,and comparing it with injury severity score（ISS）and revised trauma score（RTS）which are commonly used in ED trauma patients,to evaluate its efficacy.3.Embedding the CPOS-ED and the severe trauma patient’outcome scale in the electronic triage system in the ED to verify their clinic application effects.At the same time,calculation of their time needed,calculation accuracy,and user satisfaction for manual application by emergency nurses and compared with scoring tools commonly used in the ED.Method1.Constructing a prognosis prediction model for critically ill patients in the ED.This is a single-center retrospective study,critically ill patients who were triaged at gradeⅠand gradeⅡadmitted to the ED of the First Affiliated Hospital of Soochow University from September 2019 to August 2020 were selected to collect their general information,physiological indicators when admitted,and prognosis in the ED.Patients were randomly stratified based on their prognosis at the time of leaving the ED（survival or death）in a 7:3 ratio between the modeling group and the validation group.Univariate analysis and multivariate Logistic regression analysis were performed to screen out independent prognostic factors and to construct a prognosis predication model for critically ill patients’outcomes in the ED.2.Transforming the prediction model of critical emergency patients in the ED into a simple scoring tool（CPOS-ED）.According to the results of multivariate Logistic regression analysis and theβvalues of each factor in the regression equation,the prediction model was transformed into a simple scoring tool,CPOS-ED,and compared its application effect with MEWS,REMS,and RAPS scores which are commonly used in the ED.3.Constructing a prognosis prediction model for severe trauma patients in the ED.This is a single-center retrospective study,severe trauma patients admitted to the First Affiliated Hospital of Soochow University ED from September 2019 to November 2020 were selected to collect their general information,physiological indicators when admitted,and emergency prognosis.Patients were randomly stratified based on their prognosis at the time of leaving the ED（survival or death）in a 7:3 ratio between the modeling group and the validation group.Univariate and multivariate Logistic regression analyses were used to screen out independent prognostic factors and to construct a prognosis predication model for severe trauma patients in the ED.4.Transforming the prediction model of severe trauma in the ED into a simple scoring tool.According to the results of multivariate Logistic regression analysis and theβvalues of each factor in the regression equation,the prediction model was transformed into a quick and simple scoring tool,and compared its application effect with ISS and RTS scores which are commonly used in trauma patients.5.A prospective validation of CPOS-ED and severe trauma patients’outcome scale was conducted.Using modern technology,the emergency prognosis scoring tool for critically ill patients and severe trauma patients constructed in this study was made into application software and embedded into the electronic triage system of the ED,and prospectively verified its clinical application effect for two months.At the same time,the effect of manual application of the two scoring tools constructed in this study in emergency nurses was investigated,that is,the time spent by nurses in a manual calculation,the accuracy of calculation,the satisfaction degree of the user,and compare with the scoring tools which are commonly used in the ED.Results1.Constructing a prognosis prediction model for critically ill patients in the ED.From September 2019 to August 2020,there were 5,338 patients who met the inclusion and exclusion criteria,including 150 patients who died in the ED.Three thousand seven hundred thirty-seven patients were included in the modeling group,and 1601 patients were included in the validation group.Univariate analysis showed that there were statistically significant differences in age,systolic blood pressure,pulse rate,respiration rate,body temperature,peripheral oxygen saturation（Sp O₂）,AVPU score,and admission method between the two groups（P<0.05）.Multivariate Logistic regression analysis showed that age,systolic blood pressure,pulse rate,respiration rate,body temperature,Sp O₂,AVPU,and admission method were independent predictors of critically ill patients’outcomes in the ED（P<0.05）.Meanwhile,the area under receiver operating curve（AUC）of the prediction model was 0.882,the sensitivity was 0.848,and the specificity was 0.755.Transforming the prediction model of critical emergency patients in the ED into a simple scoring tool（CPOS-ED）.The AUC of CPOS-ED in predicting critically ill patients’outcomes in the ED was 0.876,the cut-off was 14,the sensitivity was 0.810,the specificity was 0.775;The AUC of MEWS was 0.811,the cut-off was 4,the sensitivity was 0.705,and the specificity was 0.712;The AUC of RAPS was 0.737,the cut-off was 5,the sensitivity was 0.505,and the specificity was 0.872;The AUC of REMS was 0.785,the cut-off was 9,the sensitivity was 0.619,and the specificity was 0.825.The predictive efficiency of CPOS-ED was better than MEWS,RAPS,and REMS（P<0.05）,while the predictive efficiency of MEWS and REMS were better than RAPS（P<0.05）.2.Prospective validation of CPOS-ED.Investigated the manual application of CPOS-ED and compared it with MEWS,RAPS,and REMS.There were 997 patients who met the inclusion and exclusion criteria from October to November 2021,including 27 patients who died in the ED.The results showed that the AUC of CPOS-ED was 0.889,the sensitivity was 0.963,the specificity was 0.695.The mean time for each calculation of CPOS-ED was 46.8 seconds,the calculation accuracy was95.0%,and the user satisfaction was 93.8%;The mean time for MEWS was 30.2seconds,the calculation accuracy was 96.3%,and the user satisfaction was 100.0%;The mean time for RAPS was 55.3 seconds,the calculation accuracy was 87.5%,and the user satisfaction was 68.8%;The mean time for REMS was 68.8 seconds,the calculation accuracy was 75.0%,and the user satisfaction was 62.5%.CPOS-ED showed no significant difference in the mean time,calculation accuracy,and user satisfaction（P>0.05）.The mean time of CPOS-ED was less than that of RAPS and REMS,and its calculation accuracy and user satisfaction were higher than that of REMS,with statistical significance（P<0.05）.3.Constructing a prognosis prediction model for severe trauma patients in the ED.There were 863 severe trauma patients who met the inclusion and exclusion criteria from September 2019 to November 2020.There were 21 severe trauma patients who died in the ED.There were 603 cases in the modeling group and 259cases in the validation group.Univariate analysis showed that there were statistically significant differences in age,systolic blood pressure,Sp O_2,and AVPU scores between the survival group and the death group（P<0.05）.Multivariate Logistic regression analysis showed that systolic blood pressure,Sp O₂,and AVPU scores were independent predictors for severe trauma patients’outcomes in the ED（P<0.05）.The AUC of the model was 0.938,the sensitivity was 0.867,and the specificity was 0.949.Transforming the prediction model of severe trauma in the ED into a simple scoring tool.The AUC of severe trauma patients’outcome scale predicting emergency mortality was 0.933,and the cut-off was 5,with a sensitivity of 0.867 and specificity of 0.942;The AUC of RTS was 0.800,and the cut-off was 11,with a sensitivity of 0.868 and specificity of 0.733;The AUC of ISS was 0.833,the cut-off was 15,the sensitivity was 0.909,and the specificity was 0.796.The AUC was not significant difference between the severe trauma patient’outcome scale and ISS（P>0.05）.The AUC of the severe trauma patient’outcome scale was significantly higher than that of RTS（P<0.05）.4.Prospective validation of the severe trauma patient’outcome scale.There were 123 patients who met the inclusion and exclusion criteria from October to November 2021,including 3 patients who died in the ED.Investigated the manual application of severe trauma patient’outcome scale and compared it with RTS.The results showed that the AUC of severe trauma patient’outcome scale was 0.919,the sensitivity was 1.000,the specificity was 0.767.The mean time for each calculation of severe trauma patient’outcome scale was 15.2 seconds,the calculation accuracy was 98.9%,and the user satisfaction was 100.0%;The mean time for RTS was 19.0seconds,the calculation accuracy was 97.5%,and the user satisfaction was100.0%.Compared with RTS,the mean time of severe trauma patient’outcome scale was 4 seconds less than RTS,and the difference was statistically significant（P<0.05）.The calculation accuracy was 98.8%,and the user satisfaction was 100.0%,which was the same as RTS.Conclusion1.The mortality rate of critically ill patients in the ED was about 2.8%.The age,systolic blood pressure,pulse rate,respiration rate,body temperature,Sp O₂,AVPU score,and admission method were independent predictors for critically ill patients’outcomes in the ED.The predictive efficacy of CPOS-ED is superior to MEWS,REMS,and RAPS which are commonly used in the ED.2.The mortality rate of severe trauma patients in the ED was about 2.4%.The systolic blood pressure,Sp O_2,and AVPU scores were independent predictors for severe trauma patients’outcomes in the ED.The severe trauma patient’outcome scale is superior to RTS.3.The results of prospective validation showed that both CPOS-ED and severe trauma patient’outcome scales had good prediction effects,easy to use,better calculation accuracy,and higher user satisfaction.