Logistic Regression Analysis Of Risk Factors And Risk Predict Of Peptic Ulcer Bleeding Under Sudden Changes In Temperature (Low Temperature) In Nan Chang

Objective:This study aimed to identify the risk factors of bleeding among patients with peptic ulcers under sudden changes in temperature (low temperature) in Nan chang. A prediction model was then built on the basis of these factors and validated the model.Methods:Retrospectively investigate the clinical data of inpatients and outpatients who visited the First Affiliated Hospital of Nan Chang University under sudden changes in temperature (low temperature) during the period of December 2007 to February 2008. According to their endoscopic diagnosis and clinical manifestations, they were divided into two groups:the case group (peptic ulcer bleeding) and the control group (peptic ulcer which is not complicated with bleeding). In total, there were 151 patients in case group and 229 patients in control group. Detailed personal questionnaires were performed for those patients. Logistic regression analysis was used to investigate risk factors, such as environmental and genetic factors that influenced the differences between each group, and then establish a risk prediction model. To confirm the prediction model accuracy, new patients who integrated the research (200 cases) during the period of February to April in 2008 year, carried on the follow-up observation.Results:①Univariate Logistic Analysis showed that there were about ten factors related to peptic ulcer bleeding under sudden changes in temperature (low temperature), as below: elderly (age> 60 years) (OR=2.086,95%CI=1.204-3.614), previous history of bleeding (OR= 22.876,95%CI=12.334 -42.428), previous history of NASIDs (OR=2.631,95%CI=1.466-4.723), smoking (OR=1.519,95%CI=1.001-2.305), O-type blood (OR=1.565,95%CI=1.034-2.369), duodenal ulcer (OR=2.378, 95%CI=1.416-3.993), ulcer size larger than 0.5cm (OR=1.707,95%CI=1.020-2.857) were risk factors, upper abdominal pain (OR=0.280,95%CI=0.161-0.490), dyspepsia(bloating, acid reflux, belching)(OR=0.324,95%CI=0.211-0.496), outgoing personality (OR=0.536,95%CI= 0.353-0.812) were protective factors. However, there were no significant differences between gender, Helicobacter pylori infection, concomitant diseases, PU family history, have a sweet tooth, alcohol, tea, coffee, type of work, level of economic and cultural and peptic ulcer bleeding under sudden changes in temperature (low temperature) (P>0.05).②Make those factors (P＜0.05) into Multivariate non-conditional Logistic regression analysis, the results concluded that elderly (age> 60 years) (OR=2.274,95%CI=1.043-4.958), previous bleeding history (OR=27.249,95%CI=13.578-54.685), previous history of NASIDs (OR= 2.953,95%CI=1.339 -6.513), duodenal ulcer (OR=2.446,95%CI=1.174-5.099 were the risk factors; abdominal pain (OR=0.368,95%CI=0.176-0.769), dyspepsia (bloating, acid reflux, belching) (OR=0.334,95%CI=0.187-0.599) were the protective factors, then establish the prediction model of peptic ulcer bleeding under sudden changes in temperature (low temperature):Logit (P)=-5.851+0.822 (elderly:age> 60 years)+3.305(previous history of bleeding)+1.083(previous history of NASIDs)+ 0.895(duodenal ulcer)-0.999(upper abdominal pain)-1.095 dyspepsia (bloating, acid reflux, belching), the area under the receiver operating characteristic (ROC)curve was 0.881.③Further prospective studies on the results of previous retrospective study to verify the accuracy of prediction model, the new total 200 cases of peptic ulcer were included, follow-up to the end of February 2011, a total of 183 cases (91.50%)has been effectively completed . According to the results of following-up, evaluate the validity and authenticity of the prediction model. The sensitivity, specificity, accuracy, positive likelihood ratio and negative likelihood ratio of the model was 87.10%, 90.79%,90.16%,9.45,0.14, respectively.Conclusion:1. Filter out ten factors of peptic ulcer complicated with hemorrhage under sudden changes in temperature (low temperature), and previous history of bleeding, duodenal ulcer were risk factors, while upper abdominal pain, dyspeptic symptoms, outgoing personality were protective factors.2. Establish a prediction model with six factors which were easy accessed in clinical, also the model has better accuracy and sensitivity.