Knee scoring systems in gonarthrosis - Evaluation of interobserver variability and the envelope of bias

10 experienced orthopedic surgeons assessed 15 patients using 3 commonly used composite scoring systems and by some simple variables to evaluate knee replacements. Statistical evaluation showed that the scores were valid and reflected the disease process with a reasonable reproducibility. In the individual case, however, considerable changes of the total scores and the simple variables are needed to represent a true difference at the 95% confidence limit. The coefficient of repeatability varied from 45 to 72 for the scores. Our study, which is suggested to represent any clinical investigation, showed that clinical measurements are not robust and meticulous efforts in terms of study design must be made to protect an investigation against the action of bias. Knee scores are exceedingly unreliable.