Impact of implant developers on published outcome and reproducibility of cohort-based clinical studies in arthroplasty.

BACKGROUND: The reproducibility of the results of cohort-based clinical studies of arthroplasty procedures by the average orthopaedic surgeon is a major issue involving the quality of the scientific literature. We compared the results of a comprehensive literature analysis with data from national arthroplasty registries to examine the influence of arthroplasty implant inventors on the outcomes published in peer-reviewed journals.
METHODS: A structured review of the literature published in MEDLINE-listed journals was performed. A comparison of the average revision rates in the identified studies (adjusted for the number of arthroplasty cases and the duration of follow-up) with the registry data was then conducted.
RESULTS: Seventeen of the analyzed arthroplasty implants were developed in the United States. Studies by the developers of these implants often had a substantial influence on the published outcome. For approximately 50% of the implant systems analyzed, the average revision rate derived by combining all published studies showed a statistically significant and clinically relevant deviation from the revision rate derived from the arthroplasty registry data, which reflected the outcome in the average patient. For the majority of implants for which the revision rate calculated from the published clinical studies was very low compared with the rate calculated from the registry data, the developing institution accounted for 39% to 100% of the published outcome data for the implant. In contrast, the published results were usually reproducible in clinical practice if <25% of the published data were reported by the developers. Three of the nine arthroplasty implants developed in Europe showed a significant and clinically relevant difference between the revision rate derived from only the studies published by the developers and the rate calculated from the registry data. However, because of the considerably greater amount of data from independent studies that was typically available for the European implants than for the American implants, studies by the developer that deviated significantly from the registry data could usually also be identified as outliers by a meta-analysis of all published studies. A high proportion of the published data involving three of the European implants was reported by the inventors, but comparison with the registry data revealed that the average published revision rates for two of these three implants were reproducible in clinical practice.
CONCLUSIONS: The published results of the clinical studies involving many of the arthroplasty implants, especially implants developed in the United States, were highly influenced by reports from the center that developed the implant. This often had a substantial effect on the reproducibility of the outcome data. There appeared to be relevant differences between the medical research systems in Europe and the United States that also affected the reproducibility and applicability of the results for the average surgeon. Registry data can contribute substantial added value to an informed discussion of arthroplasty outcomes.