Construction of a predictive models for early postoperative infection in solid organ transplant recipients
OBJECTIVE To construct a predictive model for early postoperative infection in solid organ transplant recipients.METHOD The clinical and laboratory data of 157 organ transplant recipients at the First Affiliated Hos-pital of Naval Medical University from Jan.2020 to Jul.2023 were retrospectively analyzed,and a stepwise regres-sion model was used to construct a predictive model for early infection in solid organ transplant patients.RESULTS Among the 157 transplant recipients,31 cases developed postoperative infections,with an infection rate of 19.75％.A total of 57 strains of pathogenic bacteria were detected,with sputum being the main type of specimen(40.35％).The top three pathogenic bacteria detected were Enterobacteriaceae(24.56％),Pseudomonas(17.54％),Acinetobacter(15.79％),and Stenotrophomonas(15.79％),respectively.An infection prediction model was constructed based on stepwise regression.RESULTS Logit(P)=-2.411+0.050 X central venous cath-eterization duration+0.027 X postoperative antibiotic use duration+0.064 X catheter use duration-0.146 X organ transplantation type-0.398 X hypertension history.The area under the curve(AUC)of the infection prediction model and Bootstrap internal validation were 0.762(95％CI:0.684-0.840)and 0.764(95％CI:0.686-0.843),respectively.When the cutoff value was 0.209,the diagnostic sensitivity was 83.33％and the specificity was 66.67％.CONCLUSION Early postoperative infections after solid organ transplantation were mainly caused by Enterobacteriaceae infection,and the combination of multiple indicators improved the prediction performance of the model.
Organ transplantationEarly infectionPathogenNomogramPredictive modelsEvaluation
NETL
NSTL
万方数据
