| Objective:To analyze the medical records of elderly inpatients with non-variceal upper gastrointestinal bleeding(NVUGIB),to find out the independent risk factors of NVUGIB in elderly inpatients.Then we build a risk prediction model and develop a nomogram model initially.Based on the data of the elderly patients admitted to hospital and the nomogram model,the incidence rate of NVUGIB in elderly inpatients is predicted,so that clinicians could detect and prevent NVUGIB in the early stage and reduce the incidence rate of NVUGIB in the hospital.Methods:A retrospective study was conducted to collect the medical records of elderly patients in a top-three hospital in Qingdao from January 1st,2018 to January 1st,2019.According to the inclusion and exclusion criteria,a total of 1085 patients were included.According to whether the elderly patients had NVUGIB during hospitalization,they were divided into bleeding group(173 cases)and control group(912 cases).T test,Mann-Whitney U test and pearsonχ~2test were used for statistical analysis.Medical records included general situation,disease situation,medication situation and laboratory test.Univariate and multivariate logistic regression analysis were used to screen out the independent risk factors of NVUGIB in elderly inpatients.After these steps,prediction model was constructed initially.The nomogram was developed with R language.Receiver operating characteristic curve(ROC),area under curve(AUC)and Hosmer-Leme show goodness of fit test were used to verify and evaluate the discrimination and calibration of the model,and the calibration curve was drawn.Result:1.The differences of medical records between the two groups are as follows:(1)The general situation:smoking history,drinking history;disease history:peptic ulcer,tumor history,coronary atherosclerotic heart disease,atrial fibrillation,stroke,helicobacter pylori(HP)infection history,acute respiratory failure and other diseases;medication:anti-coagulant and anti-platelet drugs,non-steroidal anti-inflammatory drugs,glucocorticoid drugs;and laboratory tests:leukocyte count,hemoglobin(HB),albumin,C-reactive protein,procalcitonin,international normalized ratio,D-dimer,triglycerides levels of the two groups are compared.The differences are statistically significant(P<0.05).(2)There is no significant difference in age,gender,history of hypertension,heart failure,diabetes and low-density lipoprotein cholesterol between the two groups(P>0.05).2.Univariate and multivariate logistic regression analysis screen out the risk factors of NVUGIB in elderly inpatients:tumor history(P=0.036,OR=1.552),peptic ulcer(P<0.01,OR=4.797),HP infection(P=0.005,OR=1.139),acute respiratory failure(P=0.009,OR=1.336),use of anti-coagulant and anti-platelet drugs(P<0.01,OR=2.715),increased leukocyte count(P<0.01,OR=10.370),prolonged INR(P=0.007,OR=21.314)and hypoproteinemia(P=0.001,OR=1.970)are independent risk factors for NVUGIB in elderly inpatients.3.The logistic regression model is established as P=1/(1+exp[-(-3.125+0.44×tumor history+1.568×peptic ulcer history+1.974×HP infection+1.091×acute respiratory failure+0.999×anti-coagulant and anti-platelet drugs+3.059×prolonged INR+2.339×increased leukocyte count+0.678×hypoproteinemia)].4.R language is used to develop the nomogram prediction model,and the effect of the model is further verified and evaluated:AUC=0.837(95%CI=0.8007～0.8737),Hosmer-Leme show goodness of fit test P=0.65(P>0.05).Conclusion:1.Tumor history,peptic ulcer history,HP infection and acute respiratory failure history,use of anti-coagulant and anti-platelet drugs,increased leukocyte count,prolonged INR and hypoproteinemia may be independent risk factors for NVUGIB in elderly inpatients.2.This study initially establishes the risk prediction model of NVUGIB in elderly inpatients and develops nomogram.The model has good differentiation ability and consistency,which can provide certain theoretical reference for clinicians and has certain clinical practicability. |
PDF Full Text Request