Can Standard Preoperative Serum Laboratory Tests Predict Bacterial Presence at the Time of Revision Shoulder Arthroplasty?

INTRODUCTION: The utility and optimal threshold for serum laboratory tests taken prior to revision shoulder arthroplasty are unknown. We performed a multi-institutional study of consecutive revision shoulder arthroplasties to define the optimal thresholds of serum laboratory samples to predict bacterial presence in intraoperative cultures with and without Definite Periprosthetic Joint Infection (PJI). METHODS: Multicenter data was collected on 579 revision shoulder arthroplasties. Preoperative serum erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complete blood count differentials, and ratios were reported. Analysis was stratified based on International Consensus Meeting (ICM) Definite and Non-Definite PJI and two different thresholds of culture positivity (≥2 or ≥3). Receiver operating characteristic (ROC) curves were constructed, and area under the curve (AUC), optimal thresholds, and diagnostic utility for each test were calculated. RESULTS: Eighty-three patients (14%) had Definite PJI per ICM criteria while 496 (86%) had Non-Definite PJI. Cutibacterium was the most common bacteria recovered in both Definite PJI and Non-Definite PJI. For Definite PJI using a threshold of ≥3 positive cultures, AUC values were 0.543 for ESR and 0.659 for CRP. Optimal thresholds based on ROC curves were 50.5 mm/hr for ESR and 5.9 mg/L for CRP. Sensitivity and specificity were 0.422 and 0.714, respectively, for ESR and 0.818 and 0.536, respectively, for CRP. For Non-Definite PJI, the AUC value was below 0.5 for ESR indicating the test was worse than chance. AUC values for CRP were 0.533 and 0.540 using thresholds of ≥2 and ≥3 positive cultures, respectively. Optimal threshold based on ROC curves were 9.1. CONCLUSION: This is the first large-scale, multicenter study of consecutive revision shoulder arthroplasties analyzing the utility of preoperative serum laboratory values in predicting positive intraoperative cultures. Overall diagnostic utility of these tests in predicting bacterial presence is low in the setting of both Definite and Non-Definite PJI, particularly serum ESR. In approaching patients with more obvious clinical symptoms (Definite PJI), optimal cutoff values are 50 mm/hr for ESR and 10 mg/L for CRP. When approaching workup of a patient without obvious signs of infection (Non-Definite PJI), serum ESR and CRP have limited value in predicting presence of bacteria at the time of revision arthroplasty.