Research On Construction Of Risk Pre-warning Model Of Nosocomial Infection In The Elderly

Objective:(1)Understand the basic situation of the nosocomial infection in elderly patients.(2)Guided by Logistic regression model to establish anosocomial infection risk prediction model,quantitative analysis the correlation of nosocomial infection and each index,and verify the feasibility of the model.(3)Based on the predictive value to evaluate the level of nosocomial infection risk.Methods:The occurrence of nosocomial infection among elderly patients(?60 years)who were admitted to hospital more than 48 hours from January 2015 to December 2015 were investigated restrospectively.The main monitoring programs were medical records and data from NI surveillance network.Data were used to identify possible risk factors as the indicators,quantizing the indicators and constructing the risk predict model by using Logistic regression.Using decision tree to evaluate the level of nosocomial infection risk in elderly patients.Results:(1)Among 18204 elderly patients investigated,414 patients and 517 cases suffered from nosocomial infection with an infection rate of 2.27% and the infection case-rate was2.84%.(2)Distribution of nosocomial infection sites:the top three sites of nosocomial infection is lower respiratory tract(214cases,41.39%),urinary tract infection(64cases,12.38%)and catheter-related urinary tract infection(50cases,9.67%)(3)Catheter indwelling and associated infection:414 cases of length of stay were10740 d.Using times of ventilator,central venous catheter,urinary catheter were 1256 d,6351d,4909 d respectively,lien rate were 11.69%,59.13%,45.71%,the incidence of ventilator associated infection,central venous catheter related infections,urinary related infection were 12.54‰,3.46‰,and 10.19‰ respectively.(4)Distribution of pathogenic bacteria: among 414 cases,a total of 779 strains of pathogens were isolated,including 420 strains of gram-negative bacteria,accounted for53.92%,dominated by Klebsiella pneumoniae,followed by Acinetobacter baumannii and Escherichia coli;259 strains of gram-positive bacteria,accounted for 33.25%,with coagulase-negative staphylococci top the list;and 100 strains of fungi,accounted for12.84%.(5)Mortality in elderly patients with hospital infection:mortality in elderly patients was 1.01%,mortality in patients who acquired noosocomial infection was 5.07%,whereas mortality of uninfected patients was 0.91%.(6)Univariate Logistic regression analysis revealed 16 factors associated with nosocomial infections in elderly patients,namely using time of antibiotics,combination of antiiotics,ICU,surgery,the usage of central venous catheterization,the usage of urinary catheterization,the usage of ventilator,gender,age,length of stay,diagnosis,Primary pulmonary infection,level of Serum albumin(P<0.05).(7)Further multivariate Logistic regression analysis indicate that 8 fators were independent risk factors of nosocomial infection in elderly patients,which include age,length of stay,combination of antibiotics,the time of central venous catheterization,the time of urinary catheterization,diagnosis,Primary pulmonary infection,level of Serum albumin.Besides,level of Serum albumin was the protective factors for nosocomial infection in the elderly.(8)Using Logistic regression methods,the predict model obtained:Logistic(P)=-6.169+0.183X1(age)+0.608X2(length of stay)+0.608X3(combination of antibiotics)+0.428X4(times of central vein catheter)+0.358X5(times of urinary catheter)+coefficient X6(diagnose)+0.497X7(Primary pulmonary infection)-0.582X8(serum a lbumin level).The model displayed good calibration(Hosmer-Lemeshow c2=11.382,D F=8,P=0.181)and moderate power of prediction(Au ROC:0.884;95% confidence inte rval:0.866~0.902).Choose 0.04 as cut-off points,the accuracy of modeling samples a nd testing samples were 88.3% and 77.56% respectively.(9)Risk grade: The evaluate of grade was set up based on decision tree,the ranking of risk are following: 0?Pr<0.654,0.654?Pr<0.786,0.786?Pr<0.83,0.83?Pr<0.875,Pr?0.875.AUC of decision tree classification model was 0.895.Conclusion:1.Elderly patients had higher nosocomial infection incidence,the prevention and control of nosocomial infection can benefit from strengthening objective monitoring of nosocomial infection in elderly patients.2.The occurance of nosocomial infection in elderly has been influenced by many factors.The necessary measures to prevent nosocomial infection in elderly patients are shorten the length of hospitalization,use antibiotics logically,pay attention to indwelling and maintenance of urinary catheter,central vein catheter and ventilator,improve the nutritional status.3.The Logistic predict model and nosocomial infection risk rating can achieve good effect on prediction and grade classification of nosocomial infection in elderly patients.Early warning can be achieved through qualitative and quantitative evaluation.