Large Database Studies in Orthopaedics and Sports Medicine Have Inherent Limitations and Biases That Must be Acknowledged and Managed.

Abstract

An increasing number of orthopaedic and sports medicine research studies use large databases of electronic health records and claim data aggregated from multiple health systems and corporations. While these databases benefit from study populations far larger than any single institution, they create additional risks of bias that may not exist in traditional orthopaedic clinical research designs. Researchers need to consider how the health data is transformed ("harmonized") into research data, how much data is missing, whether the length of follow-up within the database is disclosed, and whether the quality of the data input is uniform across contributing institutions. In some cases, these data may not be disclosed, especially for databases maintained by private for-profit entities. It remains unclear whether the perceived benefits of these databases truly outweigh concerns about transparency and bias, which need to be acknowledged and managed when presenting the data-supported conclusions.