Periprosthetic joint infection often raises diagnostic challenges, as the published criteria are heterogeneous. New markers for predicting periprosthetic infection have been evaluated. Here, we assessed one of these markers, C-reactive protein (CRP), in joint fluid.

We hypothesised that intra-articular CRP levels would perform better than serum CRP concentrations in diagnosing knee prosthesis infection.

We prospectively included 30 patients including 10 with native-knee effusions, 11 with prosthetic-knee aseptic effusions, and 11 with prosthetic-knee infection defined using 2011 Musculoskeletal Society criteria. Serum CRP was assayed using turbidimetry or nephelometry and intra-articular CRP using nephelometry. Appropriate statistical tests were performed to compare the three groups; P values < 0.05 were considered significant.

Serum and intra-articular CRP levels were 5- to 16-fold higher in the group with periprosthetic infection than in the other two groups. Although the areas under the ROC curves were not significantly different, the likelihood ratios associated with the selected cut-offs suggested superiority of intra-articular CRP: a value > 2.78 mg/L suggested possible infection (100% sensitivity and 82% specificity) and a value > 5.37 mg/L probable infection (90% sensitivity and 91% specificity).

Our findings suggest a possible role for intra-articular CRP assay in diagnosing knee prosthesis infection and perhaps periprosthetic infection at any site.

Level III, diagnostic study, development of a diagnostic criterion in consecutive patients comparatively to a reference standard.