DAIR OUTCOMES IN ACUTE HIP AND KNEE PROSTHETIC JOINT INFECTIONS: PREDICTIVE FACTORS FOR FAILURE AND PERFORMANCE OF KLIC AND CRIME80

Introduction & Aim Selection of patients for debridement, antibiotics, and implant retention (DAIR) in hip and knee prosthetic joint infection (PJI) remains challenging. Several scores have been proposed to improve decision-making, including KLIC and CRIME80. This study aimed to assess the performance of these scores to predict DAIR failure and to identify risk factors associated with DAIR failure. Methods We retrospectively included all consecutive adult patients with early and late acute hip and knee PJI between 1 May 2021 and 1 April 2025, with a minimum 1-year follow-up. PJI was defined according to the EBJIS criteria and was classified as early acute or late acute. Our primary outcome was DAIR failure, defined as the need for reintervention with 1 year due to infection, mortality due to infection or a clinical diagnosis of relapsing or progressive infection without possibility of reintervention. We assessed comorbidities, implant and infection characteristics, infection timing and symptoms, antibiotic therapy, laboratory parameters, RBC transfusions, medical complications and outcome. Eligible patients were identified. Data was collected through manual patient file review. The KLIC score was calculated for early acute PJI, and CRIME80 score for late acute infections. Results A total of 72 patients were included (87.5% acute post-operative, 12.5% late acute). The median age was 75.1 years (IQR, 66.8–81.7), 56.9% were female, 59.7% had hip PJI, and 40.3% had knee PJI. The overall 1-year DAIR failure rate was 25.0% (18/72), including 9 repeat DAIRs (12.5%) and 10 implant removals (13.9%). For prediction of DAIR failure, KLIC showed a sensitivity of 60.0%, specificity of 62.5%, and ROC AUC of 0.630; whereas CRIME80 showed a sensitivity of 66.8%, specificity of 66.8%, and ROC AUC 0.833. The factors associated with increased risk of DAIR failure was: ASA score (in the DAIR success group, ASA ≤ III was 96.3%, whereas in DAIR failure group, ASA ≥ III was 77.8%; p=0.014); and preoperative leukocyte count (8.0 ×10 3 /mm 3 , IQR 6.3–10.6, vs. 10.0 ×10 3 /mm 3 , IQR 8.3–14.6; p=0.02). Conclusions In this cohort, KLIC and CRIME80 showed a reasonable but limited discriminatory ability for predicting DAIR failure, with ROC AUCs of 0.630 and 0.833, respectively. Our findings suggest that patient-specific factors remain central to decision-making, particularly ASA score, which reflects comorbidity burden, and preoperative leukocyte count. Our ability to identify clear risk factors was limited due to the small sample size, which suggest the need for future multicentric studies on this subject.