A group of statisticians and clinical trial methodologists published a new study introducing a method called EQPS-rMAP designed to integrate real-world data from local populations together with trial data from other regions. This hybrid Bayesian framework aims to help in bridging trials and multi-regional clinical trials, especially when there are differences in baseline patient characteristics across regions.

The proposed method works by using propensity score stratification to adjust for baseline differences, then building stratum-specific predictive priors from the external data. It also introduces what the authors call “equivalence probability weights” to quantify how much conflict there is between the external data and the trial data, so that any risk of bias is assessed dynamically. In simulation studies the method showed better bias control, good control of type I error, and in many settings smaller sample sizes were needed while keeping statistical power. Real-world application with retrospective data in psoriasis (testing risankizumab) confirmed its advantages over older methods in reducing bias and improving precision in treatment effect estimation.

This design is relevant because many trials, particularly for rare diseases or trials spanning multiple countries, struggle with enrolling enough local patients to meet regulatory or statistical requirements. External data could help fill gaps, but regulators often worry about heterogeneity between populations, different measurement standards, or unmeasured confounders. The EQPS-rMAP method tries to address those concerns.

Why this matters

Traditional clinical trial designs require fresh data only from participants enrolled in the trial itself, often leading to long recruitment periods, high costs, and sometimes underpowered studies when local enrolment is low. Using external real-world data sensibly could reduce those burdens. If the new method gains acceptance among investigators and regulators, we may see faster approvals and more trials in regions that currently get excluded because of logistical or demographic challenges. Diseases with small patient populations or with trials that span very different geographical regions stand to benefit a lot.

Also important is that this could encourage more collaboration between institutions that collect real-world data (clinics, registries, etc.) and trial sponsors. Standardising data collection and ensuring transparency in external datasets will be critical so that these hybrid designs can be trusted.

Things to watch

Even though the simulations and retrospective analyses are promising, real-time trials using EQPS-rMAP will be needed to truly test its performance under conditions where data are being collected prospectively. Regulators will need to be comfortable with the assumptions (for example, that baseline variables adequately capture the source of heterogeneity). Also, real-world data often suffer from missing data, varying measurement methods, or selection bias; the method helps, but can’t always fix all these issues. How this methodology is received by the FDA, EMA or other agencies will also shape its adoption.

Larger scale adoption will depend on how easy it is for trial statisticians and sponsors to implement the approach, including computational tools, software, and training.