| Background:Acute mesenteric ischemia(AMI)refers to the thrombosis of the acute mesenteric artery or venous,or decreased circulatory pressure,resulting in decreased blood flow in the mesenteric and difficult to meet the metabolic needs of its corresponding organs,including occlusive AMI(OAMI)and nonocclusive mesenteric ischemia(NOMI),while OAMI includes arterial occlusive mesenteric ischemia(AOMI)and mesenteric venous thrombosis(MVT).AMI does not include isolated intestinal ischemia,or focal and segmental ischemia caused by external pressure factors such as adhesive intestinal obstruction and hernia.Chronic mesenteric ischemia(CMI)and ischemic colitis are also separate cases and are not included in this class.A large study from Malm?,Sweden,from 1970 to 1982(87% autopsy rate,approximately 250,000 in total)showed that the annual incidence of AMI was about12/100000.Besides,among all emergency department patients,its proportion was0.09%-0.20%.The aging of the population and the prevalence of cardiovascular disease risk factors in developed countries make the incidence of AMI continue to increase.Although the application of enhanced CT allows AMI patients to be diagnosed clearly after admission,a large proportion of patients still have secondary peritonitis before the intervention of vascular manipulation.The key point of treatment,in this case,is to remove the necrotic intestine,restore blood supply,and supportive intensive care.Although the mortality rate of AMI patients has dropped significantly in the past two decades,its value is still as high as 50%-70%,and the overall postoperative mortality rate is still between 30%-50%.A study compiled the AMI cases reported in China in the past 20 years and found that the fatality rate of AMI was 37.3%.In addition,survivors of AMI are also at risk of severe short bowel syndrome and late death.Improving the understanding of AMI,early diagnosis,and timely intervention are the keys to save the ischemic intestine and improve the prognosis.Early assessment of the death risk of AMI patients can help implement more active and targeted interventions.At present,there are almost no prospective randomized studies to guide the treatment of AMI,because most studies on the prognosis of AMI are retrospective design with a small sample size in a single center.In addition,there is no effective model for predicting the death risk of AMI.A comprehensive grasp of the death risk factors of AMI will help doctors,patients and their families make decisions and make treatment plans.Local hospital-based surveys will also provide unique insights for the management of AMI.In addition,building a predictive model that can accurately identify the risk of death for AMI at an early stage will also contribute to more active early intervention of AMI to achieve multi-modal individual intervention.Objective:1.Through systematic reviews and meta-analysis,to analyze the preoperative risk factors related to short-term death of AMI after exploratory laparotomy.2.Through retrospective research,analyze the clinical characteristics of OAMI patients and the risk factors related to postoperative death.3.Using MIMIC-Ⅲ(Medical Information Mart for Intensive Care)database as the data source,through classical statistical methods and machine learning,to establish and verify the mortality risk prediction model for AMI.Methods:1.Use the strategy of combining Pubmed Mesh subject terms and free words to determine search terms.In Pubmed(Medline),Embase,and Google scholar databases,retrieved from January 2000 to March 2020,search randomized controlled trials,cohort studies,and case-control studies about preoperative risk factors of short-term deaths of AMI after laparotomy.The Newcastle-Ottawa Scale(NOS)was used to evaluate the quality of the literature.Construct a data collection table to extract clinical variables related to the short-term postoperative death of the patient,including the patient’s demographic characteristics,medical history,and preoperative laboratory examination,and conduct a systematic review and meta-analysis.For the categorical variables,the odds ratio(OR)was used as the test effect indicator.For the continuous variables,the standardized mean difference(SMD)was used as the test effect indicator.The GRADE evidence level scoring system is used to evaluate the evidence level of the meta-analysis results.2.This study continuously included patients with OAMI diagnosed in West China Hospital of Sichuan University from 2011 to 2018.Collect the patient’s clinical data,including the patient’s gender,age,current medical history,past medical history,routine laboratory examinations(blood routine,coagulation routine,and routine biochemical examinations)upon admission,medical imaging examinations,and postoperative pathological reports of surgical patients,the patient’s survival were identified through the hospitalization records and family members’ telephone communication.The clinical characteristics of patients with different subtypes of OAMI were described and compared.The clinical variables in the patient’s clinical data that may be related to the 30 days postoperative death were selected for univariate statistical analysis,and the variables with statistical differences in the univariate analysis were further analyzed by multivariate logistic regression.3.The patients admitted through emergency department for AMI in MIMIC-Ⅲdatabase were included in this study.A total of 338 patients met the inclusion and exclusion criteria,and they were randomly divided into training set(238 cases)and validation set(100 cases).According to the patient’s hospitalization outcome,all patients were divided into survival group and hospital death group.First,in the training set,univariate and multivariate logistic regression analysis was used to determine the independent risk factors related to hospital deaths in AMI,and on this basis,the nomogram was developed.The variable selection methods in machine learning(Lasso,Boruta)were used to determine valuable predictors in the training set.The support vector machine(SVM),XGBoost(extreme gradient boosting),and extreme learning machine(ELM)were used to build prediction model.Finally,in the validation set,receiver operating characteristic(ROC)curve and area under curve(AUC),calibration curve and Hosmer-Lemeshow test,and decision curve analysis were used to evaluate the performance of the predictive model based on classical statistics(nomogram)and the optimal machine learning model in discriminative ability,calibration ability and clinical utility.Results:1.Based on the inclusion criteria and exclusion criteria,this study finally included 20 articles with a total of 5,011 patients.Studies were of high quality,with a median Newcastle-Ottawa Scale Score of 7.Summary short-term postoperative mortality was 44.38%(range,18.80%–67.80%).Across included studies,49 potential risk factors were examined,at least two studies.The meta-analysis based on at least four studies identified the following preoperative risk factors related to short-term postoperative mortality of AMI after exploratory laparotomy: The short-term postoperative mortality rate was higher in elderly patients(OR = 1.90,95%CI [1.57,2.30],P < 0.0001),GRADE evidence grade was low;compared with MVT,AOMI patients had higher postoperative short-term mortality(OR = 2.45,95%CI [1.12,5.33],P = 0.04),GRADE evidence grade was moderate;patients with a history of heart failure had higher short-term postoperative mortality(OR = 1.33,95%CI [1.03,1.72],P = 0.03),GRADE evidence grade was low;patients with a history of chronic kidney disease had higher short-term postoperative mortality(OR = 1.61,95%CI [1.24,2.07],P = 0.0003),GRADE evidence grade was low;patients with a history of peripheral vascular disease had higher short-term postoperative mortality(OR = 1.38,95%CI[1.00,1.91],P = 0.05),GRADE evidence grade was low;patients in the short-term postoperative death group were older(SMD = 0.32,95%CI [0.24,0.40],P < 0.0001),GRADE evidence was low;Patients in the short-term postoperative death group had higher serum creatinine at admission(SMD = 0.50,95%CI [0.25,0.75],P < 0.0001),GRADE evidence is moderate;The platelet count of patients in the short-term postoperative death group was lower at admission(SMD =-0.32,95%CI [-0.50,-0.14],P = 0.0004),and the GRADE evidence was low.2.A total of 108 patients of OAMI were included in the study,including 58 AOMI and 50 MVT,with an overall average age of 57.1 years.Compared with MVT patients,AOMI patients were older(64.1±14.3 vs.49.1±16.1,P < 0.001).The proportion of men in the MVT group was higher(78.0% vs.55.2%,P = 0.013).This study found that compared with the MVT group,the AOMI group experienced heart disease(37.9% vs.8.0%,P < 0.001),hypertension(34.5% vs.16.0%,P = 0.029)and diabetes(15.5% vs.4.0%,P = 0.048)often,and there are significant statistical differences.The proportion of MVT patients with liver disease history was significantly higher than that of AOMI patients(22.0% vs.5.2%,P = 0.009).In terms of symptoms(abdominal pain,vomiting,abdominal distension,diarrhea,and bloody stools)and physical signs(abdominal tenderness,abdominal rebound pain,and muscle tension),there was no significant difference between the AOMI patient group and the MVT patient group,with a P value>0.05.Compared with MVT patients,plasma high-density lipoprotein(1.3±0.6 mmol/L vs.0.9±0.4 mmol/L,P = 0.001)and fibrinogen(4.5±1.9g/L vs.3.6±1.5g/L,P = 0.011)in AOMI patients,were higher with significant statistical difference.In this study,77 patients underwent exploratory laparotomy,and the 30-day mortality rate was 29.9%.Multivariate logistic regression analysis showed that the time interval from admission to surgery(OR = 1.19;95%CI [1.07,1.34],P = 0.005),platelet count(OR = 0.98;95%CI [0.97,0.99],P = 0.008)and AOMI(OR = 5.55;95%CI [1.36,22.55],P = 0.017)were independent predictors for 30-day mortality of OAMI after exploratory laparotomy.Further analysis of the 45 patients of AOMI showed that the time interval from admission to surgery(OR = 1.22;95%CI [1.01,1.47],P = 0.036)and platelet count(OR = 0.98;95%CI [0.97,0.99],P = 0.020)were independent prognostic factors for 30-day mortality of AOMI after exploratory laparotomy.3.In the training set,univariate and multivariate logistic regression analysis identified the following independent prognostic factors related to hospital death of AMI,including diastolic blood pressure(OR = 0.955,95%CI [0.934,0.976],P <0.001),blood lactate(OR = 1.407,95%CI [1.185,1.671],P < 0.001),blood p H(OR= 0.009,95%CI [0.001,0.339],P = 0.011),blood creatinine(OR = 1.524,95%CI[1.210,1.919],P < 0.001),red blood cell distribution width(OR = 1.431,95%CI[1.190,1.720],P < 0.001)and age(OR = 1.048,95%CI [1.019,1.077],P = 0.001).Similarly,in the training set,the Lasso variable selection method identified seven clinical variables with potential predictive value,including systolic blood pressure,blood lactic acid,blood p H,anion gap,blood creatinine,red blood cell distribution width and age;Boruta variable selection method determined nine clinical variables with potential predictive value,including blood lactate,anion gap,blood creatinine,systolic blood pressure,red blood cell distribution width,diastolic blood pressure,blood p H,age and blood urea nitrogen.In the validation set,the optimal machine learning model is the XGBoost model(AUC = 82.9%,95%CI [74.9%,91.0%])constructed based on the variables selected by the Boruta variable selection method.The model and the nomogram(AUC =77.0%,95%CI [67.9%,86.1%])both had good discrimination ability.The model calibration curve and Hosmer-Lemeshow test show that the nomogram(P = 0.076)and the optimal machine learning model(P = 0.876)were well calibrated.The decision curve analysis showed that when the threshold probability was between 0.22 and 0.85,the optimal machine learning model had higher clinical utility than that of nomogram.Conclusion:1.Advanced age,AOMI,history of heart failure,history of chronic kidney disease,and history of peripheral vascular disease are preoperative risk factors for short-term mortality of AMI after exploratory laparotomy.Patients in the short-term postoperative death group are older and have higher blood creatinine at admission,but lower platelet counts.2.AOMI,low platelet count,and long time interval from admission to surgery are independent risk factors for 30-day mortality for obstructive AMI after exploratory laparotomy.Low platelet count and a long time interval from admission to surgery are also independent risk factors for AOMI after exploratory laparotomy,and the time interval from admission to surgery is more important for AOMI patients.3.The nomogram developed in this research achieves a relatively accurate prediction of AMI hospital death risk in a concise form,while the machine learning model developed seems to have better discriminative ability and clinical utility.The nomogram machine learning(Nomo-ML)predictive model combination may help improve medical management and help clinicians make medical decisions related to AMI management. |
PDF Full Text Request