Clinical Treatment Analysis And Risk Prediction Model Of Acute Respiratory Distress Syndrome

Objectives To evaluate the response of clinicians in Intensive Care Unit（ICU）of Hebei General Hospital to identify the patients who were in line with the diagnostic criteria of Acute Respiratory Distress Syndrome（ARDS）in clinical practice,ventilation management and the use of adjuvant treatment measures,as well as the incidence and outcome of ARDS.In addition to establish a model for early prediction of the risk of ARDS.Methods A retrospective study assisted by Artificial Intelligence（AI）technology.A total of 516 eligible patients in ICU of Hebei General Hospital from April 10,2017 to October 31,2021 were admitted to our study.The patients were divided into ARDS group（n=217）and non-ARDS group（n=299）according to whether they had ARDS.The baseline data,ventilation management,and adjunctive therapy measures for the two groups were collected and analyzed.The case data of the two groups were compared,and the independent risk factors of ARDS were screened out by multivariate Logistic regression analysis,and the risk prediction model was established.Results 1 To evaluate clinicians’recognition of ARDS,ventilation management,the use of adjuvant therapy measures in routine clinical practice,and the incidence and clinical outcome of ARDS.Of the 6081 patients hospitalized at ICU during the study period,258patients conformed with the Berlin definition criteria for ARDS,and 217 patients were included in this study after analysis.The proportions of mild,moderate and severe ARDS were 20.7%（45/217）,49.3%（107/217）and 30.0%（65/217）,respectively.The clinical recognition rates of mild,moderate and severe ARDS were 13.4%（29/217）,28.1%（61/217）and 20.7%（45/217）,respectively.A total of 204 patients（94.0%）received invasive mechanical ventilation during their ICU hospitalization.The tidal volume of all patients who met the diagnostic criteria for ARDS was 7.77（6.54,9.49）mL/kg of their ideal body weight（PBW）.The positive end expiratory pressure（PEEP）was 8（5,10）cm H₂O.Platform pressure of 4 cases and driving pressure of 5 cases were recorded when included.Among these patients,deep sedation was used in 111 cases（51.2%）,decruitment maneuver was used in 25 cases（11.5%）,drone position was used in 41 cases（18.9%）,high-dose hormone therapy was used in 27 cases（12.4%）,neuromuscular blockers were used in 17 cases（7.8%）,Extracorporeal Membrane Oxygenation（ECMO）was used in 6cases（2.8%）.33 cases（15.2%）were discharged automatically after family members signed.41 cases died in ICU（18.9%）.The ICU mortality rates in mild group,moderate group and severe group were 15.6%（7/45）,21.5%（23/107）and 16.9%（11/65）,respectively.The adjusted overall case mortality rate was 46.1%（100/217）.2 Univariate analysis showed that pneumonia,aspiration,pulmonary contusion,extrapulmonary sepsis,pancreatitis,non-cardiogenic shock,massive blood transfusion,actual respiratory rate>30beats/min,higher acute physiology and chronic health score system Ⅱ（APACHE Ⅱ）and sequential organ failure score（SOFA）were high risk factors for ARDS.Multivariate Logistic regression analysis showed that pneumonia,aspiration,extrapulmonary sepsis,pancreatitis,massive blood transfusion,actual respiratory rate>30 times/min and SOFA score≥7.5 were independent risk factors for ARDS.These risk factors were incorporated into the ARDS risk prediction model as independent risk warning indicators.Finally,ARDS risk prediction model is p=e^x/（1+e^x）,x=-2.68+0.593×（pneumonia=1）+1.746×（aspiration=1）+0.898×（extrapulmonary sepsis=1）+1.462×（pancreatitis）+1.593×（massive blood transfusion）+1.318×[（actual respiratory rate>30 times/min）=1]+0.180×[（SOFA score≥7.5）=1].The prediction probability of ARDS in each patient was calculated according to the model formula,and each prediction probability was taken as the test variable,and the occurrence of ARDS was taken as the state variable to plot the receiver operating characteristic curve（ROC curve）.The area under ROC curve（AUC）of the model was 0.723（95%CI:0.679-0.768,P<0.001）.The sensitivity and specificity of the model for predicting the occurrence of ARDS were 0.664 and 0.742respectively.Conclusions The overall prevalence rates of ARDS in ICU of Hebei General Hospital was 3.6%（217/6081）in the same period.Recommended lung protective ventilation strategies were highly respected to some extent,and adjuvant therapy was well implemented but not standardized enough.The ICU mortality was relatively low,but the proportion of automatic discharge was high.2 Pneumonia,aspiration,extrapulmonary sepsis,pancreatitis,massive blood transfusion,actual respiratory rate>30 times/min and SOFA score≥7.5 were the independent risk factors for predicting ARDS.The ARDS early prediction model established based on these factors could effectively predict the occurrence of ARDS.Figure 2;Table 7;Reference 158...