Risk Analysis for Postoperative Joint Infection after Arthroscopic Anterior Cruciate Ligament Reconstruction in Chinese Population:a Retrospective Study / 中国运动医学杂志

ABSTRACT

Objective To analyze the risk factors of postoperative joint infection after arthroscopic an-terior cruciate ligament(ACL)reconstruction,so as to provide evidences for prevention.Methods This is a retrospective case control study.Among 20549 arthroscopic anterior cruciate ligament reconstruc-tion performed between year 2002 and 2018,62 was diagnosed as postoperative joint infection,includ-ing 54 males and 8 females,with an average age of 26.5 years.Another 638 without postoperative in-fection was selected using stratified sampling according to the number of operations per year.The gen-eral condition and surgery data of all patients were collected from the electronic medical record sys-tem.Then univariate Logistic regression analysis was conducted to screen risk factors and multivariate analysis was carried out to create a prediction model.Moreover,an artificial neural network(ANN)model was created to predict the probability and compared with Logistic regression model.Results The univariate Logistic regression analysis found 14 factors associated with postoperative joint infection,in-cluding gender,age,height,weight,body mass index(BMI),socioeconomic status,companioned pro-cedure,single or double bundle reconstruction,portal number,tourniquet time,drainage number,pro-phylactic antibiotics,previous knee surgery and companioned illness.The sensitivity,specificity,accura-cy and area under curve of Logistic regression model was 100%,55.6%,60%and 0.843.As for ANN model,the corresponding values were 80%,89.9%,90%and 0.917.Conclusion Risk factors associated with postoperative joint infection include gender,age,BMI,socioeconomic status,surgery date,tourni-quet time,drainage number and previous knee surgery.Both Logistic model and ANN model yield sat-isfying predicting efficacy,with ANN model showing higher accuracy.