The use of datasets that combine patients’ clinical information with their genomic data are increasingly being used to understand the value of precision therapies in cancer. Recent studies have shown though that patients receive next-generation sequencing (NGS) at varying points in their cancer care, a phenomenon known as delayed entry, which can create survivorship bias by over-representing patients who live long enough to receive NGS. Researchers used the Flatiron Health-Foundation Medicine Clinico-Genomic Database to understand the impact of this type of bias by comparing overall survival (OS) of patients with non-small cell lung cancer regardless of their NGS testing status.
Why this matters
As real-world data (RWD) comes of age, and is deployed for increasingly complex and consequential purposes (such as regulatory approvals), it is important to develop methodologies that measure up to the sophisticated analytical challenges in RWD-based research. Observational data where cohort eligibility criteria inherently cause delayed entry, such as genetic or genomic testing, may be particularly vulnerable to bias, which in turn, may diminish their perceived robustness. It is the responsibility of investigators to craft the analytical tools that overcome those issues, in order to move the RWD research field forward towards meeting its full potential.