Quality of outcome data in knee arthroplasty

BACKGROUND AND PURPOSE Recent reports on developer bias in unicondylar knee arthroplasty led to concerns about quality of publications regarding knee implants. We therefore compared revision rates of registry and non-registry studies from the beginning of knee arthroplasty up to the present. We assessed the time interval between market introduction of an implant and emergence of reliable data in non-registry studies. MATERIAL AND METHODS We systematically reviewed registry studies (n = 6) and non-registry studies (n = 241) on knee arthroplasty published in indexed, peer-reviewed international scientific journals. The main outcome measure was revision rate per 100 observed component years. RESULTS AND INTERPRETATION For 82% of the 34 knee implants assessed, revision data from non-registry studies are either absent or poor. 91% of all studies were published in the second and third decade after market introduction. Only 5% of all studies and 1% of all revisions were published in the first decade. The first publications on revision rates of total knee arthroplasty (TKA) started 6 years after market introduction, and reliable data were found from year 12 onward in non-registry studies. However, in unicondylar knee arthroplasty (UKA) the first publications on revision rates could be found first 13 years after market introduction. Revision rates of TKA from non-registry studies were reliable after year 12 following market introduction. UKA revision rates remained below the threshold of registry indices, and failed to demonstrate adjustment towards registries. Thus, the superiority of registry data over non-registry data regarding outcome measurement was validated.