We aimed to answer following questions: (1) What is the most accurate preoperative marker for the diagnosis of PJI in hip and knee arthroplasty? (2) What is the diagnostic accuracy at current thresholds and are there better diagnostic thresholds for serum C-reactive protein (CRP), synovial white blood cell (WBC) count, and synovial polymorphonuclear neutrophil percentage (PMN%) and (3) What is the combined diagnostic performance at current International Consensus Meeting (ICM) thresholds?

We retrospectively reviewed 260 patients that underwent a revision knee or hip arthroplasty at our institution between 2015 and 2017. All major and minor 2018 ICM criteria (except ESR, D-Dimer) were available for all patients included. Cases with at least 1 major criterion were considered as infected. Receiver operative characteristic (ROC) curve analysis was performed for preoperative minor criteria.

The diagnostic performance of the preoperative minor criteria ranked according to area AUC was: PMN% (0.926), AD (0.922), WBC count (0.916), LE (0.861), and serum CRP (0.860). Increasing the PMN % cut-off from 70% to 77.8% improves the diagnostic accuracy (86.5% versus 90.8%). The highest diagnostic performance was achieved by combining all 5 preoperative parameters and at current ICM thresholds the diagnostic accuracy, sensitivity (SE), Specificity (SP), positive predictive value (PPV), and negative predictive (NPV) was 93.5%, 95.4%, 92.1%, 89.7%, and 96.5%.

The diagnostic performance of preoperative minor criteria were outstanding (PMN%, AD, WBC count) or excellent (LE, serum CRP). PMN% showed the best diagnostic utility (AUC), should have an increased weight-adjusted score in the ICM scoring system.