Overview
Real-world data is used to understand how cancer treatments work outside of controlled clinical trials, but it comes with a major challenge: "unmeasured confounding." This occurs when factors not recorded in a patient’s medical chart influence treatment choices and make it difficult to determine if a drug is truly the cause of a better outcome. This is especially problematic when comparing a new, modern therapy to an older one that was used before the new drug was available.
In this study, researchers tested a quasi-experimental statistical method called "regression discontinuity in time" (RDiT) to see if it could produce results more consistent with trials than traditional mathematical approaches. Using data from nearly 2,000 lung cancer patients, they compared the effectiveness of the immunotherapy pembrolizumab to an older chemotherapy drug, docetaxel. The researchers found that the RDiT method produced survival estimates that were much closer to the results of gold-standard clinical trials, whereas traditional methods tended to overestimate the benefits of the newer treatment.
Why this matters
This work demonstrates that using more sophisticated statistical "tools" can make real-world evidence much more reliable and trustworthy. By showing that the RDiT method can successfully mirror clinical trial results while accounting for the complexities of real-world care, this research paves the way for higher-quality evidence generation. Better evidence leads to more accurate insights for regulators, insurance payers, and doctors, eventually helping to ensure that the treatments patients receive are proven to be truly effective.