Abstract
Background: Two-stage revision arthroplasty is widely accepted as the reference standard for chronic periprosthetic joint infection (PJI) after total knee arthroplasty (TKA); however, reinfection or persistent infection occurs in a substantial subset of patients. We aimed to identify predictors of uncontrolled PJI following two-stage revision using an extended, single-center cohort. Methods: We retrospectively reviewed 177 knees with PJI after TKA treated with a uniform two-stage revision protocol between September 2011 and February 2022. Patients were classified as controlled (no further infection-related surgery or antimicrobial therapy ≥ 2 years after reimplantation) or uncontrolled (persistent infection after the first stage or reinfection after reimplantation). Demographics, comorbidities, laboratory parameters, perioperative factors, and microbiological characteristics were compared. Multivariable logistic regression with penalized estimation was used to identify independent predictors of uncontrolled infection. Results: Among 177 knees, 145 (81.9%) achieved infection control and 32 (18.1%) were classified as uncontrolled. On multivariable analysis using Firth penalized logistic regression, liver cirrhosis (odds ratio [OR] 9.87; 95% confidence interval [CI] 1.01-96.23; p = 0.049), the presence of a sinus tract at the first infection-control surgery (OR 3.47; 95% CI 1.43-8.40; p = 0.006), and fungal pathogens (OR 8.92; 95% CI 2.46-32.32; p = 0.001) were independently associated with uncontrolled PJI. Pre-reimplantation erythrocyte sedimentation rate (ESR) was significantly higher in the uncontrolled group on univariate analysis but was evaluated as a supportive marker due to limited availability in patients who did not undergo reimplantation. Demographic variables and most routine preoperative laboratory markers, including C-reactive protein before the first stage, were not independently associated with treatment failure. Conclusions: Liver cirrhosis, the presence of a sinus tract, and fungal infection are independent risk factors for uncontrolled PJI after two-stage revision TKA. Preoperative risk stratification incorporating host, local, and microbiological factors may assist in optimizing reimplantation timing, tailoring antimicrobial strategies, and improving patient counseling.