Abstract
The prospective evaluation of two hundred and seven symptomatic total knee arthroplasties presenting for revision total knee arthroplasty is reported. On univariate analysis patients who had infection differed significantly (p< .001) from those without infection with regards to: elevated ESR, CRP, positive aspiration, and history of; revision procedure less than two years since last surgery, early wound problems, ongoing pain since index procedure, and discharging wound. On multivariate analysis elevated ESR or CRP, positive aspiration, pain since index procedure and early wound complications were significant predictors of infection (p< .05). These variables were then used to formulate an evidence-based multivariate predictive algorithm to assist the clinician in decision making prior to surgery.
Differentiating septic from aseptic failure of total knee arthroplasty on the basis of clinical features and diagnostic tests can be troublesome for the clinician. The purpose of this paper is to describe significant differences between cases of septic and aseptic failure of total knee arthroplasty. The incorporation of these variables into a practical multivariate clinical prediction algorithm can provide assistance in establishing the diagnosis of infection prior to revision knee arthroplasty.
A simple clinical prediction algorithm can assist in the diagnosis of infection in patients with painful total knee arthroplasty. Patients with five of five criteria have a 99% probability of infection whereas patients with zero of five criteria have a 1% probability of infection.
This is the first multivariate evidence-based clinical prediction algorithm presented for use in decision making prior to revision total knee arthroplasty. The surgeon can use the information derived from clinical and laboratory assessment to compute an approximate pre-operative probability of infection prior to surgery (see table).
On multivariate analysis elevated ESR or CRP, positive aspiration, pain since index procedure and early wound complications were significant predictors of infection (p< .05). These variables were then used to formulate an evidence-based multivariate predictive algorithm to assist in clinical decision making.
Prospective data was collected on two hundred and seven symptomatic knee arthroplasties presenting for revision arthroplasty. A multivariate logistic regression model was used to determine the probability of infection using five significant variables. Combinations of these five variables can provide the clinician with an estimate of the probability of infection prior to revision knee arthroplasty.
Please contact author for tables and/or charts.
Correspondence should be addressed to Cynthia Vezina, Communications Manager, COA, 4150-360 Ste. Catherine St. West, Westmount, QC H3Z 2Y5, Canada
