Preliminary Study Of Risk Factors And Risk Predictionmodel For Pancreatic Infection In Severe Acute Pancreatitis

Objective: To investigate the infection of severe acute pancreatitis related factors and the establishment of infection in severe acute pancreatitis risk prediction modelMethods: Retrospective analysis and Shanghai Changhai Hospital from October 2006 to December 2009 patients admitted to SAP data. Epidata software database is established (by the Second Military Medical University, Department of Statistics completed, see appendix). By reviewing medical records and telephone follow-up, etc. detailed record of general and clinical data of patients, including home address, contact details, age, sex, symptoms, signs, laboratory parameters, admission treatment (fluid resuscitation, somatostatin, inhibition acid drugs, enteral nutrition, antibiotics, etc.), complications (such as organ dysfunction, pancreatic infection, pancreatic infection), length of stay, outcomes and so on. According to pancreatic infection or pancreatic infection were divided into groups and non-pancreatic infection group. Application of SAS 9.1.3 software for statistical analysis, the data type, measurement data using the nonparametric Kruskal-Wallis test, measurement data between the two groups with Fisher exact test and multivariate analysis using Logistic regression analysis. P <0.05 defined as the difference was statistically significant. Results: A total of 298 patients with clinical data of patients, of which 168 cases of severe acute pancreatitis patients with complete data were available for analysis. Pancreatic infection in 32 cases (19.05%), 136 cases of non-pancreatic infection (80.5%); pancreatic infection group 21 males (65.63%), female 11 cases (34.38%), pancreas, male 79 cases of non-infection group (58.09%), female 57 cases (41.19%), no significant gender differences in the two groups (P = 0.549). The mean age of pancreatic infection 51.94±13.82 years (16 to 85 years), pancreatic non-infected group, the average age was 52.96±15.18 years (28 ~ 85 years), no significant age difference between the two groups (P = 0.7897). Group of patients without pancreatic infection in 136 patients, 4 died, 2.94% of total number of cases; pancreatic infection in 32 patients, 5 deaths, 15.63% of the total number of cases, pancreatic infection was significantly higher than mortality pancreatic infection group (P <0.05). The heart rate, WBC, AKP, LDH, BUN and CTSI score, SIRS, shock, acute renal dysfunction, external pancreatic infection, microbial treatment, partial pressure of oxygen, preventive antibiotics, the use of Logistic regression of these 13 variables filter, and finally there are three variables were retained in the selection of the model, namely LDH, CTSI score and pancreatic infection. Established risk prediction model prediction accuracy rate of 54.55% of pancreatic infection, the sensitivity was 54.55%, specificity of 100%, positive prediction rate was 100% correct, negative correct prediction rate of 91.23%. Conclusion: LDH, CTSI score and external pancreatic infection is a risk factor for pancreatic infection. Pancreatic infection risk prediction model to predict the accuracy of 54.55%, 54.55% sensitivity and specificity of 100%, positive prediction rate was 100% correct, negative correct prediction rate of 91.23%.