How clinicians can assess the strength of comparative real-world evidence

In this video, Martin Schoen, MD, PhD, discusses how clinicians can assess the robustness of comparative real-world evidence on androgen receptor pathway inhibitors (ARPIs) and determine when these data may meaningfully inform treatment decisions. Schoen is a medical oncologist at the VA Medical Center and a professor at Saint Louis University in St. Louis, Missouri.

Schoen explains that the strength of real-world analyses depends in part on how closely their design and patient selection approximate those of a prospective randomized trial. Analyses that incorporate individual patient-level data, particularly when patients from different studies can be appropriately matched, can provide a more reliable basis for comparison. Similarly, observational analyses that use complete patient data and appropriate matching and inclusion and exclusion criteria may offer greater confidence in their findings.

Network meta-analyses represent another approach that can strengthen comparative assessments by incorporating data from multiple similarly designed trials and increasing the available sample size and statistical power. However, Schoen emphasizes that clinicians should examine how the underlying studies were conducted and whether their patient populations and treatment timing are sufficiently comparable.

“The principles of understanding who are the patients and how are they selected is important,” he says.

As comparative real-world evidence becomes more common in settings without head-to-head trials, Schoen emphasizes that clinicians should assess whether the methodology approximates the rigor of a randomized study. This includes evaluating treatment timing, inclusion and exclusion criteria, patient selection, and whether the analysis follows a target trial framework. For example, if patients in a real-world analysis begin an ARPI substantially later than the androgen deprivation therapy initiation window used in a clinical trial, the resulting comparison may not reflect the same treatment setting. These considerations can help clinicians determine whether findings are sufficiently robust to inform treatment discussions or should instead be viewed as hypothesis-generating.