Abstract
The differential diagnosis of pain after total knee arthroplasty (TKA) should always include periprosthetic infection (PPI). The current diagnostic tools vary in sensitivity, specificity, and predictive value. Currently there is no test with an absolute accuracy for diagnosis of PPI. Furthermore, cost effectiveness of investigations has become an important issue in recent years as increased expenses can place an unnecessary heavy burden on the medical system.
We retrospectively evaluated 296 patients who underwent revision TKA at our institution during 2000–2005 and had preoperative erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) performed. The criteria used for diagnosing infection were a positive intraoperative culture on solid media, presence of an abscess or sinus tract that communicated with the joint, positive preoperative aspiration culture, and/or elevated fluid cell count and neutrophil differential of the aspirated fluid. The sensitivity, specificity, and predictive values of ESR and CRP were determined. Combinations were performed in parallel that necessitate both tests to be negative to rule out infection. The cost of each serological test was compared to that of other commonly used screening modalities.
One hundred and sixteen patients (39%) were classified as infected and 180 patients (61%) were considered non-infected. The sensitivity, specificity, positive predictive value, and negative predictive value for the ESR were: 91%, 72%, 68%, and 93%. The sensitivity, specificity, positive predictive value, and negative predictive value for the CRP were: 94%, 74%, 70%, and 95%. The sensitivity and negative predictive value for the combined studies were 96% and 95%, respectively. However, five infected patients (4%) had a normal
ESR and CRP. An organism was cultured on solid media in 4 of the 5 cases. ESR and CRP were the least costly of all the preoperative tests including radionuclide imaging and joint fluid analysis. ESR and CRP are important preoperative tests in diagnosis of PPI and their ability to clench the diagnosis in the majority of cases should not be underestimated. When combined, these simple serological tests have improved sensitivity and negative predictive value to rule out infection.
Correspondence should be addressed to Vasiliki Boukouvala at Department of Orthopaedic Surgery & Traumatology, University Hospital of Larissa, 110 Mezourlo, Larissa, GREECE. Tel: +30 2410 682722, Fax: +30 2410 670107, Email: malizos@med.uth.gr