We have investigated the reliability of the American Knee Society Score (AK
S), a widely used functional outcome score for knee arthroplasty, 29 subjec
ts mere assessed in random order by 6 observers, each patient seen twice by
each observer. The interobserver reference intervals were 16 points for th
e Knee Score (0-100 points) and 21 points for the Function Score (0-100 poi
nts), The intraobserver reference intervals were smaller, 11 points and 13
points for the Knee Score and the Function Score, respectively. The presenc
e of arthritis in other joints and general debility did not affect the reli
ability of the scores. The more experienced observers had greater intraobse
rver reproducibility. Notably, we found moderate agreement between observer
s in the subjective variables, while the objective variables produced lower
levels of agreement.
The high inter- and intraobserver variations of the AKS makes estimation of
score change questionable. Reliable use of the AKS would necessitate repea
ted evaluation by an experienced observer.