CRP LEVELS IN INFECTED JOINT REPLACEMENTS

Abstract

Detection of infection in total joint replacements (TJR) is notoriously difficult. Ideally the diagnosis should be known before revision arthroplasty is undertaken. The level of C-reactive protein (CRP) is one readily available test. Sanzen et al. reported sensitivity of 78% and specificity of 100% for CRP in distinguishing infection in 23 infected TJRs and 33 non-infected TJRs undergoing revision, using a cut off of 2mg/dl¹. However, they used only intra-operative cultures as the standard to compare the CRP against. We have analysed the reliability of CRP to diagnose infection pre-operatively in a group of patients undergoing revision arthroplasty, using the following criteria as the reference standard for infection: 2 or more intra-operative cultures positive for the same organism; presence of acute inflammatory response on histology; presence of pus in the joint at revision (1/3 positive indicates true infection), as described by Hanssen et al.²

The results of CRP and the operative investigations of 26 patients undergoing revision arthroplasty (15 hips and 11 knees) were studied prospectively. In our unit CRP is assayed in mg/dl serum by an automated machine. During revision arthroplasty, multiple specimens were taken from around the joint for microbiological and histological examination. Microbiological cultures were carried out on solid media and broth in aerobic and anaerobic conditions. Histological analysis assessed the level of neutrophils present in the tissue. The presence or absence of pus was noted. The results were analysed graphically and a cut off level of CRP was then chosen for analysis of reliability.

Thirteen patients were infected and 13 were not. Eleven of the 13 infected patients had a CRP greater than 2 mg/dl, and 10 of the 13 non-infected patients had a CRP less than 2 mg/dl. Using 2 mg/dl as a cut off, CRP had a sensitivity of 85% and a specificity of 77%. If 4mg/dl is taken as the threshold for infection, then CRP is 100% specific but only 61% sensitive.

CRP is a useful investigation in the diagnosis of infection in joint replacements. However we have shown that a cut off of 2mg/dl is not 100% specific for non-infected patients. Increasing the threshold improves the specificity, but reduces the sensitivity. Unfortunately there is no single investigation that is 100% accurate in this setting. CRP results must be interpreted in the light of the clinical picture and other investigations. These patients are part of an ongoing study to identify the most reliable criteria for diagnosing the presence of infection in total joint replacement.