Accuracy of Predictive Algorithms in Total Hip and Knee Arthroplasty Acute Periprosthetic Joint Infections Treated With Debridement, Antibiotics, and Implant Retention (DAIR)

Abstract

BackgroundDebridement, antibiotics, and implant retention (DAIR) failure remains high for total hip and knee arthroplasty periprosthetic joint infection (PJI). We sought to determine the predictive value of the CRIME80 and KLIC for failure of DAIR in acute hematogenous (AH) and acute postoperative (AP) PJIs, respectively. MethodsWe identified 134 patients who underwent DAIR for AH PJI with <4 weeks of symptoms after index arthroplasty and 122 patients who underwent DAIR for AP PJI <90 days from index. In the AH group, 15 patients (11%) failed at 90 days and overall, 33 (25%) had failed by 2 years. In the AP group, 39 (32%) failed at 90 days and overall, 52 (43%) failed by 2 years. Logistic regression models were used to determine the area under the curve (AUC) to establish thresholds using the Youden index. ResultsFor the AP cohort, AUCs were below 0.66 for KLIC, Charlson comorbidity index, Elixhauser comorbidity index, and McPherson host grade. For the AH cohort, 90-day AUCs were 0.70 for CRIME80 and below 0.66 for Charlson comorbidity index, Elixhauser comorbidity index, and McPherson host grade. In multivariate analysis controlling for age, sex, and body mass index, the CRIME80 AUC improved to 0.77 at 90 days. ConclusionTo the authors’ knowledge, this study represents the first external validation of the KLIC and CRIME80 for predicting DAIR failure in a North American population. The results indicate that alternative methods for predicting DAIR failure at 90 days and 2 years for acute PJI are needed. Level of EvidencePrognostic III.