Large Database and Registry Research in Joint Arthroplasty and Orthopaedics

摘要

The digital revolution has made possible the proliferation of large databases and the mining of “big data” from many sources in orthopaedic surgery. This revolution in the compiling of, and access to, huge amounts of data has in turn led to dramatic advances in the types of studies and research that can be done in our field. A symposium/workshop was held on October 14 and 15, 2021, in Chicago, Illinois, to explore the power and potential of large databases; consider the weaknesses and risks of analyses using these databases; provide information on methods, specific attributes, and best uses of the most commonly used databases; and provide guidance on how best to select and use these databases for orthopaedic research. The JBJS Supplement on Large Database and Registry Research in Joint Arthroplasty and Orthopaedics is the product of that symposium. The idea behind the symposium was that a systematic evaluation of the state of large-database research in orthopaedic surgery could provide much valuable information to the many stakeholders who interact with these databases. These stakeholders include orthopaedic researchers, orthopaedic surgeons who read orthopaedic research, and journal reviewers and editors who evaluate research. The symposium concentrated on databases commonly used in joint arthroplasty, but much of the information presented in the JBJS Supplement has applicability across other orthopaedic disciplines. The symposium was sponsored by a National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIH/NIAMS) P30 Center Grant (Principal Investigator, Daniel Berry, MD, and Co-Principal Investigator, David Lewallen, MD), The Journal of Bone and Joint Surgery, and the Orthopaedic Research and Education Foundation (OREF). Important input in the genesis of the symposium also came from the Editor of the Journal of Arthroplasty (John Callaghan, MD). Speakers, and authors of articles in the JBJS Supplement, were recruited for their expertise in each subject area. Many thanks to them and to the other workshop participants who gave of their time and expertise to participate in the presentations and discussions. Large databases in orthopaedics include national registries, state and regional registries, institutional registries, large payor administrative databases, and specialty or procedure-specific databases1,2. A partial list of the broad variety of research questions that can be addressed with large databases includes associations between variables such as a patient characteristic, a diagnosis, a specific procedure, a category of orthopaedic implant, or a pharmacologic agent and the relationship of each variable to the subsequent outcome of surgical or nonsurgical treatment; cost-effectiveness of treatments; and descriptive research documenting trends in orthopaedic practice. While large databases and “big data” provide “real-world evidence” to aid research that can improve the care of orthopaedic patients, these data sources also have important limitations, not all of which are intuitively obvious. One of the most important is that, although large observational databases are powerful tools to identify associations between variables (and thus are good tools to form hypotheses that can be subsequently tested with more rigorous study designs), they can rarely prove a causal relationship between an exposure and an outcome due to the potential for sampling bias and unmeasured confounders. A second important reality is that studies using large databases may show significant associations between variables—because of the power of big numbers—but these differences can be of limited clinical relevance due to the small magnitude of the effect. Some limitations are highly specific to the individual database or data source. Moreover, the data contained in specific databases may differ with respect to criteria for and selectiveness of patient inclusion; availability of specific data points with