Analysis of risk factors for periprosthetic joint infection after primary total knee arthroplasty and construction of a nomogram prediction model / 中华创伤骨科杂志

ABSTRACT

Objective:To investigate the risk factors for periprosthetic joint infection (PJI) after primary total knee arthroplasty (TKA) and construct a nomogram model for prediction of such risks.Methods:In this retrospective study, we enrolled 69 patients with PJI after primary TKA (the infection group, n=69) who had been admitted to Department of Orthopedics, Nanjing Jinling Hospital, The First School of Clinical Medicine, Southern Medical University from January 2010 to December 2019. The non-infection group included the patients of the same kind but without postoperative infection during the same period who were matched according to time of admission, age, and gender in a ratio of 1∶3 ( n=207). The data on body mass index, anesthesia method, operation time, preoperative C-reactive protein, preoperative albumin, and comorbid medical conditions were collected from both groups to screen the risk factors for postoperative development of PJI using univariate and multivariate conditional logistic regression analyses. After a nomogram of the risk factors was plotted using R software, the consistency index (C-index) was calculated. The receiver operating characteristic curve, calibration curve, and clinical decision curve were drawn. Results:Multivariate conditional logistic regression analysis showed that preoperative albumin <35 g/L ( OR=7.166, 95% CI: 3.427 to 14.983, P<0.001), operation time >90 min ( OR=3.163, 95% CI: 1.476 to 6.779, P=0.003), diabetes mellitus ( OR=3.966, 95% CI: 1.833 to 8.578, P<0.001), rheumatic diseases ( OR=3.531, 95% CI: 1.362 to 9.156, P=0.009), and chronic lung diseases ( OR=4.734, 95% CI: 1.790 to 12.521, P=0.002) were risk factors for development of PJI after primary TKA. The nomogram constructed with R software visualized the model. The C-index of the nomogram was 0.809 (95% CI: 0.751 to 0.867), indicating a good predictive capability of the model. The calibration curves of the model showed that the nomogram was in good agreement with the actual observations. The decision curves showed that the threshold probabilities of the model ranged from 0.08 to 0.75, providing a good net clinical benefit. Conclusions:Preoperative low albumin, prolonged operation time, diabetes, rheumatic diseases, and chronic lung diseases may be the risk factors for PJI after primary TKA. The nomogram prediction model based on these factors can provide a reference for clinicians to prevent PJI.