The diagnostic accuracy of blood C-reactive protein and erythrocyte sedimentation rate in periprosthetic joint infections - A 10-year analysis of 1510 revision hip and knee arthroplasties from a single orthopaedic center.

Abstract

Background: Despite the availability of many highly accurate biomarkers and novel criteria, serum C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) are still the basis for diagnosing periprosthetic joint infection (PJI). This study aims to analyze the influence of different demographical and clinical factors on the cut-off values and accuracy of CRP and ESR in diagnosing chronic PJI.

Methods: A total number of 4757 patients (with ICD-10 codes T84.0 and T84.5) operated on between January 2014 to December 2023 in a single orthopaedic center were screened in terms of the inclusion and exclusion criteria. Finally, 1510 patients (1032 aseptic revisions and 478 periprosthetic joint infections [PJI]) were included in the analysis. The best cut-off values, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for both CRP and ESR for all cohort and for subgroups divided depending on the demographical (gender, joint and BMI) and clinical factors (prosthesis fixation, specific diagnostic criteria, and virulence of the bacteria) were calculated.

Results: For all cohort, the best cut-off value for CRP was 9.6 mg/L with an area under the curve (AUC) of 0.93 and for ESR was 29 mm/h with the AUC of 0.891. For CRP the sensitivity was higher (84.9%) than for ESR (75.1%), with the same values of specificity (90.5% and 90.8%, respectively). According to the specific subgroups, for CRP higher sensitivity was observed for males (89.6%) than for females (82.6%) if lower thresholds were used. Similarly, when the higher cut-off value for CRP was applied, better specificity for high-virulent (94.8%) than for low-virulent pathogens (88.9%) was observed. For ESR, superior sensitivity values were observed if a fistula was observed, for lower BMI thresholds and for infections caused by high-virulent pathogens. Higher optimal threshold and better specificity were observed for knees than for hips, without the appearance of fistula, when minor criteria were used and infection was caused by high-virulent pathogens.

Conclusions: Our study indicates better sensitivity for CRP than ESR and similar specificity values for diagnosing chronic PJI. If results oscillate close to 10 mg/L for CRP and between 25 and 30 mm/h for ESR we propose to use different cut-off values depending on the demographic and/or clinical factors to increase diagnostic accuracy.