Culture of multiple peroperative biopsies and diagnosis of infected knee arthroplasties.

Prosthetic joint infections remain difficult to diagnose. In 1981, Kamme & Lindberg described a diagnostic procedure with five peroperative biopsies in patients with total hip arthroplasty (Clin Orthop Relat Res 1981;154:201-7). Its usefulness, however, has not been ascertained for other prosthetic joints undergoing surgical revision. Therefore, we undertook a retrospective study of 120 surgical revisions in 118 patients with knee arthropasties where such biopsies had been obtained. Cases were categorized into three groups based on information available prior to revision and peroperative inspection: prosthetic joint infection (n = 26), aseptic loosening (n = 58), and mechanical problems (n = 36). Fifteen sets were positive, 13 had significant growth (i.e. > or =3 biopsies with the same microbe/s), and 2 had insignificant growth (< or =2 positive biopsies). Excluding the group with a mechanical problem, the sensitivity for infection was 12/26 (46%), the specificity 58/58 (100%), the positive predictive value 12/12 (100%), and the negative predictive value 58/72 (81%). In the group with infection there was a trend towards less exposure to antibiotics in cases with positive cultures than cases with negative cultures. The Kamme & Lindberg procedure is applicable also to knee arthroplasties, but the low sensitivity and negative predictive value underline the need for new diagnostic methods.